design, synthesis and biological utility of polysaccharide
TRANSCRIPT
Design, Synthesis and Biological Utility of Polysaccharide
Chain-Terminating Glycosides
By
Christopher D. Brown
A dissertation submitted in partial fulfillment of
the requirements for the degree of
Doctor of Philosophy
(Chemistry)
At the
University of Wisconsin—Madison
2013
Final Oral Defense: July 19, 2012
This dissertation is approved by the following members of the Final Oral Committee
Laura L. Kiessling, Professor, Chemistry
Samuel H. Gellman, Professor, Chemistry
Deane F. Mosher, Professor, Biomolecular Chemistry
Eric Strieter, Assistant Professor, Chemistry
Jennifer M. Schomaker, Assistant Professor, Chemistry
i
Design, Synthesis and Biological Utility of Polysaccharide Chain-Terminating Glycosides
Christopher D. Brown
Under the supervision of Laura L. Kiessling At the University of Wisconsin—Madison
Polysaccharides are ubiquitous in nature, yet their assembly is poorly understood at the
molecular level. Unlike their nucleic or amino acid counterparts, carbohydrate polymers are
assembled via a template independent process. Glycosyltransferases, the enzymes required for
polymer assembly, are able to control both the length and patterning of a polysaccharide chain.
Structural contributions from the enzyme itself, as well as the substrates it recognizes, play a
central role in the regulation of assembly. From this, it follows that perturbation to the
structure of the enzyme or substrates would result in concomitant perturbation of the activity
of the system. This approach was used to develop a new method for assessing patterning
fidelity of one such carbohydrate polymerase.
The enzyme GlfT2 from Mycobacterium tuberculosis constructs a linear polysaccharide
of galactofuranose (Galf) residues termed the galactan from the donor sugar UDP-Galf. The
galactan is an integral part of the Mycobacterial cell wall and aids in the organism’s innate
ability to resist and evade its human host immune response. A better understanding of this
biosynthetic process may aid in the development of new treatments for this global disease. A
closer examination of galactan structure reveals that the monosaccharides are linked via an
alternating pattern of β-(1,5) and β-(1,6) glycosidic linkages. For this bifunctional enzyme, we
ii
wondered if it was faithful to the alternating pattern and devised a method to challenge this
fidelity.
The pivotal hydroxyl groups at the C5 or C6 positions of Galf are the requisite
nucleophiles for chain extension by GlfT2. We reasoned that a chemical substitution of either
hydroxyl group for a non-nucleophilic functionality such as fluorine or hydrogen would block
extension at that position. We hypothesized that this block would force the enzyme to continue
polymerization via the remaining (unsubstituted) hydroxyl group or induce chain termination,
depending on its fidelity to alternating linkage deposition. A divergent synthesis of fluorinated
UDP-Galf analogs was developed, and the donors induced chain termination in a sequence-
specific manner. This suggested GlfT2 operates with high fidelity for constructing the
alternating pattern. Further studies with chain-terminating probes promise to yield insight into
the mechanism of galactan assembly.
iii
Acknowledgements
A PhD is a community effort. No one produces a new body of work in a vacuum. Many
people supported me professionally and privately and I reserve this space to thank them. While
it feels almost like bringing the cast out before the performance to take a bow, this is as it
should be: Without these people, there would be no show, no spectacle, and no mystery
unraveled.
Laura Kiessling is an advisor who values unwavering rigor and the excitement born only
of scientific inquiry. I stand today on a firm scientific foundation because I chose to emulate her
drive for discovery. She has been a mentor to me in how to approach scientific thinking and
scientific writing. I have enjoyed learning from her a great deal.
The Kiessling Group operates with a level of achievement and excellence that has me in
awe on a regular basis. All of my coworkers are intelligent, hard-working and clever. The results
of these traits are apparent in the work culture: It is collaborative and fun, engaging and an
easy place to learn. I want to thank Kenzo Yamatsugu, Rebecca Splain, Matt Levengood, Raja
Annamali, Darryl Wesener and John May in particular, all of whom have worked on GlfT2 and
improved our understanding of its function in fundamental ways. I have learned laboratory
skills from each of them and benefited from materials they have generated for my own studies.
I also want to thank my office mates Joe Grim, Mario Martinez and Jonathan Hudon for sharing
the workspace. Laughter is my favorite de-stressor and we had more than our fair share in
5128.
iv
For two years I had the privilege of working with an undergraduate chemistry major,
Max Rusek. I feel lucky to have had the chance to mentor him. He worked furiously to supply
me with important chemical intermediates for many of the studies performed in this thesis. He
accelerated the work we accomplished and did it with a smile every day. I wish him well in his
post-bachelor degree endeavors.
The staff members who keep the chemistry building running are top rate. In particular
the facilities directors Charlie Fry, Monika Ivancic and Martha Vestling have been excellent
resources for NMR and mass spectrometry issues. We have top notch facilities because of the
top notch people that run them. I want to thank Kat Myhre, my go-to person for logistical help
regarding event planning as well as a fellow gardening enthusiast. Theresa, Lauren and
Stephanie were always willing to stop and ask about how my day was going. I will miss our
talks, friends.
Lastly, but most importantly, I want to thank my friends and family. My best friend Kelly
Kraft kept me sane by insisting on annual backpacking trips which were always great
adventures. Deb Heilert was great at finding fun things to do in Madison which kept me
balanced. My partner for life, my boyfriend, Peter Gruett, cannot be thanked sufficiently in an
acknowledgement. You are the best part of every single day, and have been since our first date.
I love you so much. I thank my uncles Dick and John and my aunt Bobbi for reminding me where
I come from and what my values are. Finally, my mom and my brother have been my core, my
family, my driving force. Mom, I wrote the words one thing at a time on my lab bench because
v
when I was most stressed by graduate school your long-ago words of counsel were my source
of strength, as they have always been. We did it.
vi
Table of Contents
Abstract i
Acknowledgements iii
Table of Contents vi
List of Figures ix
List of Abbreviations xii
Chapter 1: The Regulation of Bacterial Polysaccharide Assembly 1
1.1 Introduction 2
1.2 Biosynthesis 5
1.2.1 Genomic organization 5
1.2.2 The location of assembly 6
1.3 Substrate Recognition 7
1.3.1 The Glycosyltransferases 7
1.3.2 Lipid Binding 9
1.3.3 Acceptor binding and the role of primases 13
1.3.4 Donor Binding 14
1.3.5 Determining the order of binding 16
1.4 Polymerization 17
1.4.1 Directionality 17
1.4.2 Processivity 18
1.4.3 Subsites 23
vii
1.4.4 Length Control 24
1.4.4.1 Intrinsic factors 24
1.4.4.2 Extrinsic factors 26
1.4.4.3 Chemical chain termination 29
1.4.5 Pattern control 30
1.5 Conclusions 32
Chapter 2: Design and Synthesis of Fluorinated UDP-Galf Analogs 34
2.1 Introduction 35
2.2 GlfT2: A bifunctional polymerase from Mycobacterium tuberculosis 37
2.3 Synthetic probes 39
2.3.1 Synthetic acceptors 39
2.3.2 Synthetic donors 39
2.4 Chemical probes of pattern fidelity 40
2.5 Synthesis of fluorinated Galf analogs 43
2.5.1 Accessing vinyl Galf 43
2.5.2 Setting the stereochemistry at C5 45
2.5.3 Installation of fluorine 46
2.5.4 Installation of phosphate 47
2.5.5 Phosphate coupling 48
2.6 Experimental procedures 51
viii
Chapter 3: Evaluation of Fluorosugars as Probes of Carbohydrate Polymerase Activity 69
3.1 Introduction 70
3.2 GlfT2 operates by a sequence specific mechanism 72
3.3 The sequence specificity of GlfT2 is general 74
3.4 A mistake on initiation? 74
3.5 GlfT2 struggles to recognize the terminal linkage of short acceptors 77
3.6 Fluorinated acceptors are dead-end substrates 82
3.7 Experimentals 85
Chapter 4: Expanding the Utility of Polysaccharide Chain-Terminating Glycosides 89
4.1 Introduction 90
4.2 A comparison of fluorinated and deoxygenated Galf donors 93
4.2.1 As sequence specificity probes 93
4.2.2 As Inhibitors 94
4.3 Fluorinated acceptors 97
4.3.1 Synthesis 97
4.3.2 Evaluation as inhibitors 99
4.4 GlfT2 makes two linkages with one active site 100
4.5 Experimentals 102
ix
List of Figures
Figure 1.01 Bacterial polysaccharides 3
Figure 1.02: Location of polymerization 7
Figure 1.03: Glycosyltransferase mechanism 8
Figure 1.04: Glycosyltranserfase crystal structures 10
Figure 1.05: Lipid specificity 12
Figure 1.06: Donor binding 15
Figure 1.07: Common Bi Bi mechanisms 16
Figure 1.08: Determining reducing versus non-reducing chain elongation 19
Figure 1.09: Distraction assay for polymerization 22
Figure 1.10: TagF is distributive 22
Figure 1.11: Length control 28
Figure 1.12: Pattern control 32
Table 1.01 4
Figure 2.01: Homopolymers of glucose 35
Figure 2.02: Polysaccharides with alternating linkages 36
Figure 2.03: Galactan biosynthesis 38
Figure 2.04: Synthetic acceptors 40
Figure 2.05: Synthetic donors 41
Figure 2.06: Fidelity assessment assay 41
Figure 2.07: Retrosynthesis of fluorinated UDP-Galf analogs 43
Figure 2.08: Determining the stereochemistry of the 5F furanose 48
Scheme 2.01: Accessing vinyl Galf 44
x
Scheme 2.02: A kinetic separation of diastereomers 45
Scheme 2.03: Installation of fluorine 46
Scheme 2.04: Installation of phosphate for 6F-Galf 49
Scheme 2.05: Installation of phosphate for 5F-Galf 49
Scheme 2.06: Phosphate coupling 50
Table 2.01: A comparison of physical parameters for C-F and C-OH bonds 42
Figure 3.01: A mass spectrometry method to observe polymerization by GlfT2 in vitro 71
Figure 3.02: Chain termination is sequence specific with a hexasaccharide acceptor 73
Figure 3.03: Chain termination is seuqnece specific with a tetrasaccharide acceptor 75
Figure 3.04: Synthetic tetrasaccharide behaves the same as chemoenzymatic 76
tetrasaccharide
Figure 3.05: Model for carbohydrate-directed binding 78
Figure 3.06: Disaccharides are poor substrates for extension with F-Galfs 81
Figure 3.07: Trisaccharides reveal importance of lipid in acceptor orientation 83
Figure 3.08: Fluorinated acceptors are dead-end substrates 84
Scheme 3.01: Allyl acceptor synthesis 79
Scheme 3.02: Synthesis of disaccharide acceptors 80
Figure 4.01: Panel of UDP-Galf analogs 90
Figure 4.02: Competition experiment to evaluate fluorinated donors as inhibitors of 94
polymerization
Figure 4.03: Coupled assay reveals drop in UDP production in the presence of 96
fluorinated donors
Figure 4.04: Assessing the polymerization competency of fluorinated acceptors 100
Figure 4.05: Mutant GlfT2 polymerization assays 101
xi
Figure 4.06: GlfT2 D371E retains high sequence fidelity 102
Scheme 4.01: Synthesis of deoxy and azido Galf 92
Scheme 4.02: Fluorinated acceptor synthesis 97
Table 4.01: Summary of chain termination results 93
xii
List of Abbreviations
ABC ATP binding cassette Ac Acetyl Aq aqueous ATP adenosine triphosphate B. subtilis Bacillus subtilis BSIT benzenesulfonylimidazolium triflate CAN Acetonitrile AcOH acetic acid Bn benzyl Bz benzoyl C. jejuni Campylobacter jejuni CMP cytosine monophosphate CSA camphor sulfonic acid d doublet DAST diethylaminosulfur trifluoride DCM dichloromethane dd doublet of doublets d.r. diastereomeric ratio DIPEA diisopropylethylamine DMF dimethylformamide DNA deoxyribonucleic acid E. coli Escherichia coli EMM exact mass measurement ESI Electrospray ionization Et ethyl Galf galactofuranose GalNAc N-acetylgalactosamine GDP guanine diphosphate Glc glucose GlcNAc N-acetylglucosamine GlfT2 Galactofuranosyltransferase 2 G. stearothermophilus Geobacilus stearothermophilus GT glycosyltransferase HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid HPLC high performance liquid chromatography IPTG isopropylthiogalactopyranoside K4CP chondroitin polymerase from E. coli K4 LDH lactate dehydrogenase
xiii
m multiplet MALDI-TOF matrix-assisted laser desorption ionization time of flight Man mannose ManNAc N-acetylmannosamine mCPBA meta-chloroperoxybenzoic acid Me methyl MeOH methanol MRSA methicillin resistant staphylococcus aureus NADH nicotinamide adenine dinucleotide NMR nuclear magnetic resonance P. aeruginosa Pseudomonas aeruginosa PBP penicillin binding protein PDB protein data bank PET positron emission tomography PK pyruvate kinase q quartet RT room temperature s singlet S. aureus Staphylococcus aureus SDS-PAGE sodium dodecyl sulfate polyacrylamide gel electrophoresis S. enterica Salmonella enterica S. equisimilis Streptococcus equisimilis S. pneumonia Steptococcus pneumoniae STD NMR saturation transfer difference nuclear magnetic resonance STP standard temperature and pressure TEA triethylamine TFA trifluoroacetic acid TFAA trifluoroacetic anhydride THF tetrahydrofuran TLC thin layer chromatography UDP uridine diphosphate UMP uridine monophosphate UMP-Im. uridine monophosphate imidazolium UV ultraviolet
1
Chapter 1
The Regulation of Bacterial Polysaccharide Assembly
This chapter is intended for submission to the Annual Review of Biochemistry
Brown, Christopher D. and Kiessling, Laura L., The Regulation of Bacterial Polysaccharide Assembly. 2012
2 1.1 Introduction
Polysaccharides are pervasive in nature. They exist in intracellular, pericellular and extracellular
environments and participate in a wealth of biological processes ranging from energy storage to cell
structural integrity to pathogenesis. Their range spans from short oligomeric sequences to long chains
thousands of monomers in length. These carbohydrate chains can likewise be linear or branched,
charged or neutral, viscous or slippery, rigid or flexible. Indeed, polysaccharides are so diverse in size,
shape and physical characteristics that research into their biosynthesis has demanded an equally
multifaceted approach.
While both eukaryotic and prokaryotic organisms produce polysaccharides, the bacteria are the
true masters of the art. Bacterial polysaccharides are can be covalently bound to the cell wall, anchored
to the membrane, attached to proteins, or secreted into the extracellular matrix. Structures include
peptidoglycan,1 capsular polysaccharide,2 exopolysaccharide,3 O-antigen,4,5 teichoic acid,6 N- and O-
linked glycans7,8 and various other unique structures that decorate the exterior of bacterial cells (Figure
1). They project outward into the environment and serve several important functions. They preserve
cellular integrity by regulating osmotic pressure and cell shape by providing a rigid structural coat. They
serve as docking sites for bacteriophage9 and are involved in colonization, pathogenesis and virulence in
a range of hosts. Many excellent reviews have been written on aspects of polysaccharide biosynthesis,
usually in the context of a specific organism or type of polymer. This review describes the regulation of
polysaccharide assembly, with a focus on the chemical structure of the enzyme and substrates it utilizes
as well as the functional characterization of the polymerization event itself.
Many bacterial polysaccharides are of great medical importance. They are causal elements
underlying much of the morbidity and mortality of bacterial infections. Lipopolysaccharide (LPS) is
involved in septic shock and multiple organ failure. Capsular polysaccharides in streptococcal
3
Figure 1. Representative examples of bacterial polysaccharides exhibit great structural diversity. They are found anchored to cellular structures, attached to proteins, lipids, other carbohydrates, or are free in the extracellular matrix. Carbohydrate polymers exposed to the external environment are the point of first contact for many important biological interactions.
4 bacteria facilitate colonization of the nasopharyngeal or meningeal epithelium.2 Some have been
identified as highly immunogenic and co-opted for vaccine development.10 Teichoic acid polymers from
Gram-positive organisms including methicillin-resistant Staphylococcus aureus (MRSA) are thought to
facilitate septation during cell division by harboring a reserve of divalent cations for glycosyltransferase
activity.11 The polyanionic polymer alginate from Pseudomonas aeruginosa is integral to formation of
biofilms that render its infection in the lungs of cystoic fibrosis patients so recalcitrant to antibiotics.3 A
current push to generate large quantities of chemically-defined heparin has given rise to the possibility
of using bacterial enzymes involved in capsular polysaccharide assembly for chemoenzymatic
production.12
Table 1.01. Representative bacterial polysaccharides vary in sequence and length.
Variation of polysaccharide sequences and lengths in bacteria compounds the complexity
observed in Nature. Some bacteria employ homopolymers with simple repeating patterns wherease
others construct complicated and rare glycans taylor-made for the required actions of the organism.
Length also varies from short oligomeric sequences, such as the N-linked glycan of Campylobacter jejuni,
to the very large polymers that mimic glycosaminoglycans such as the hyaluronic acid capsular
5 polysaccharide of Streptococcus pneumoniae. Length and sequence are so fundamental to
polysaccharide function that most research involving carbohydrate polymer biosynthesis has focused on
understanding the regulation of these two structural features (Table 1.01).
An effort to understand what factors influence glycan localization, length and patterning has
illuminated the features of polysaccharide biosynthesis. The polymerization event is at the heart of this
assembly process, in which activated monosaccharides are appended iteratively end to end to generate
the polysaccharide chain. Elongation of a repeating sequence of sugars occurs in a variety of contexts.
Sometimes the process is the result of the actions of a single polymerase while other systems employ a
division of labor strategy, utilizing the specific activities of several enzymes in sequence or in concert.
This review highlights the biochemical strategies used by bacteria to construct these carbohydrate
polymers and the methods researchers have used to define them. It focuses on the biochemical
regulation of polymerization and highlights concepts used to better define the complicated assembly
process. The review will also highlight techniques that have proven effective to study polymerizations
and identify areas that are in need of further method development.
1.2.1 Biosynthesis: Genomic organization
The methods used to reveal aspects of assembly have been cross-disciplinary. Polysaccharides
isolated from nature are sequenced using degradation and chemical modification techniques that date
back a century. Older methods have been supplemented with newer techniques such as X-ray
crystallography and NMR spectroscopy and mass spectrometry to elucidate structure of enzyme and
polymer alike. High resolution techniques like electron microscopy, radiation inactivation and atomic
force microscopy have facilitated our understanding of higher order organization of these cell surface
structures. A continued drive to sequence biologically important genomes has yielded rich information
about bacterial genomic organization. In bacteria, polysaccharide biosynthesis is organized into gene
6 clusters. Genes involved in polysaccharide assembly include sugar nucleotide biosynthetic enzymes,
glycosyltransferases, proteins involved in transport of polymers or precursors across the plasma
membrane, and other regulatory elements. Mutagenesis and deletion studies revealed the importance
of polysaccharides on bacterial behavior, ultrastructure and physiology. Assays for polymerase activity
were generated. Newer techniques that combine genetics, molecular biology, biochemistry, chemical
biology and bioengineering have marshaled in a new era of study into carbohydrate structure and
function.
1.2.2 Biosynthesis : The location of assembly
Bacterial polysaccharide chains are formed in many cellular environments. Consider the three
predominant methods for assembly of O-antigen (Figure 1.02).13 Researchers have classified O-antigen
biosynthesis based on the location of assembly and mechanism of export. The three categories are
Wzx/Wzy, the ATP binding cassette (ABC) and the synthase pathways. Polymers assembled via the
Wzx/Wzy pathway are built up from undecaprenyl phosphate (Und-P) carrier lipids bearing the
fundamental repeat sequence. These monomers are assembled on the cytoplasmic face of the
membrane, exported to the periplasm via the Wzx protein, polymerized by Wzy, ligated to the lipid A
core and transported to the outer membrane. Polymerization in the ABC pathway also occurs on Und-P
carriers, but on the cytosolic face, prior to transport via an ATP Binding Cassette transporter protein for
which the pathway is named. The synthase pathway generates the polysaccharide coupling
polymerization to export. (Figure 1.02) Capsular polysaccharide biosynthesis uses a similar set of
polymerization/export strategies.2
The location of assembly determines the resources available to build the polymer. How
glycosyltransferases utilize these resources to build polymers requires an understanding of enzyme
structure, substrate availability and the key recognition events that bring together the various
7 precursors. The structural features of enzyme, donor and acceptor dictate the substrate recognition
events necessary for efficient biosynthesis.
Figure 1.02. Polymerization can take place on the cytosolic, periplasmic or intramembrane portions of the cell. E. coli O9 is assembled via the ABC pathway on the cytosolic side of the membrane (A). E. coli O86 uses the Wzx/Wzy pathway and polymerizes modified Und-P lipids on the periplasmic face of the inner membrane (B). Synthases such as WbbF from S. enterica couple polymerization to export (C). 1.3.1 Substrate Recognition: The Glycosyltransferases
All glycosyltransferases must bring together an acceptor substrate and activated donor to
mediate catalysis. Efforts to understand how these enzymes bind their substrates have given us a
molecular picture of polymer assembly. Glycosyltransferases are classified according to sequence
homology into families numbered GT-1, GT-2 up to GT-110 at present.14 A second method of
classification is based on predicted and solved structural domains; glycosyltransferases predominantly
have either GT-A or GT-B domains. A third classification defines glycosyltransferases mechanistically as
either retaining or inverting. (Figure 1.03) Inverting transferases generate a glycosidic linkage that has
the opposite stereochemistry from the stereochemistry of the anomeric position of the donor. This
mechanism usually involves deprotonation of the acceptor via a general base mechanism followed by
direct displacement of the donor leaving group. For retaining glycosyltransferases, an enzyme-donor
covalent intermediate is formed in a two-step double-inversion process resulting in a glycosidic bond
8 with the same stereochemistry as the donor. A comprehensive review on these classifications and
implications the structures have a glycosyltransferase function was published recently.14
Figure 1.03. Glycosyltransferases are either inverting or retaining, depending on whether the stereochemistry at the donor sugar switches or is maintained, respectively. For example, a donor with an axial (red) NDP group can be used to generate a glycosidic bond with either equatorial (blue) or axial stereochemistry.
The most direct technique for understanding how glycosyltransferases recognize their donor
and acceptor substrates is X-ray crystallography.(Figure 1.04) Few structures of polymerases have been
solved, precluding generalizations about regulation of polysaccharide assembly, however the structures
that are reported provide a framework for illuminating the molecular mechanisms at work.(Refs) The
number of polymerases with solved crystal structures is still too small to make broad generalizations
about the regulation of polysaccharide assembly, but the structures that are reported go a long way
toward the illumination of the key molecular mechanisms. For instance, K4CP—a bifunctional
polymerase that generates a chondroitin polymer—is a member of the GT-A family in which two
Rossmann-like folds create a continuous twisting beta sheet surrounded by alpha helices.15
Glycosyltransferases of this family bind a nucleotide donor sugar and an acceptor substrate to engage
polymerization. The crystal structure of the staphylococcal polymerase LtaS, which is responsible for
polyglycerol lipoteichoic acid biosynthesis,16 was solved as a cocrystal with glycerol phosphate and a
9 Mn2+ atom.17 The enzyme was found to adopt a sulfatase-like fold in which the conserved polar and
charged residues bind the metal cation. The cation in turn interacts with the phosphate moiety of the
glycerol phosphate group.17 Crystal structures of peptidoglycan transglycosylases, which catalyze the
transfer of sequential disaccharide units to build up the carbohydrate backbone of peptidoglycan
strands, reveal large, multifunctional enzymes with transmembrane, transglycosylation and
transpeptidase domains.18,19 A co-crystal of E.coli transglycosylase bound to the inhibitor moenomycin
(A mimic of the endogenous substrate lipid IV) demonstrates how the enzyme may bind two molecules
of lipid II, condense them to generate lipid IV, and then extend the polymerization via elongation at the
reducing end of lipid IV, maintaining contact with the growing chain as it slips along a groove on the
enzyme’s surface.19
The structures of a variety of carbohydrate polymerases gives an image of how the enzyme
preorganizes the ternary complex to generate a new bond and what structural features of the enzyme
and substrates are essential for binding and catalysis. What imparts an enzyme with specific or broad
reactivity and can that be exploited? Two thematic areas of study have emerged. The first is concerned
with substrate recognition, the second with measuring the resulting activity. A discussion regarding
these two areas is framed around specific examples from the literature and focused on the techniques
used to investigate the structural and functional underpinnings of polysaccharide biosynthesis.
1.3.2 Substrate Recognition: Lipid binding
Most bacterial glycans are assembled on a lipid phosphate carrier, undecaprenyl phosphate
(Und-P) in virtually all cases. Und-P consists of 7 cis, 3 trans and one terminal isoprene units. O-antigens,
capsular polysaccharides and bacterial N- and O-glycans are assembled on this carrier. Mycobacteria use
the related lipid decaprenyl phosphate to assemble the arabinogalactan, a cell wall component.
10
Figure 1.04. Active site of chondroitin polymerase K4CP with UDP-GalNAc and Mn2+ (A). LtaS from Bacillus subtilis with glycerol phosphate and Mg2+ (B). Ribbon (C) and surface (D) projections of E. coli PBP1b transglycosylase bound to a lipid IV analog. PDB codes: 2Z87 (K4CP),2W5S (LtaS) 3FWL (PBP1b)
Membrane anchored glycans such as the lipoteichoic acids and some streptococcal capsular
polysaccharides are assembled directly onto phosphatidylglycerol lipids.
Bacterial polysaccharides are appended to a wide range of structures including lipids in the
cellular membrane, the peptidoglycan and proteins. The ultimate ligation site is dictated not by the
polysaccharide chain, but by the lipid carrier on which it was built. To ensure that polysaccharide
assembly results in ligation to the appropriate cellular structure, the process is regulated at some point
by a glycosyltransferase with high specificity for the carrier lipid. In this sense, the specificity of an
enzyme for a lipid carrier acts essentially as a targeting sequence, allowing bacteria to synthesize
complicated glycan structures and deliver them en bloc to their appropriate location. Strict adherence to
11 lipid identity is not observed at every step of polysaccharide assembly and different organisms have
adopted specificity at unique points in the biosynthetic sequence (Figure 1.05). Thus,
glycosyltransferases that recognize and act on a range of structural variants have been exploited by
chemists who have designed simplified glycolipid acceptors for enzymatic studies.20,21 This has prompted
scientists interested in glycan engineering to explore the potential for tailored glycan displays
constructed with promiscuous biosynthetic machinery (Figure 1.05).22
The Wang group evaluated the lipid dependence of the polymerase Wzy in the construction of
the E. coli O86 antigen and found that unsaturation or trans geometry in the isoprene unit closest to the
glycan all but abolished the enzymatic activity of Wzy.23 Interestingly, the lipid specificity is relaxed for
the transferases responsible for assembling the repeating unit. The ligase WaaL which catalyzes the
transfer of the fully polymerized O-antigen to Lipid A core, exhibits the highest degree of promiscuity in
the system.24
The Imperiali Group undertook a thorough investigation of the lipid dependence of the pgl
glycosyltransferases responsible for N-glycan assembly in Campylobacter jejuni.25 Biosynthesis of the N-
glycan produced in Campylobacter jejuni is general and organized into a gene cluster containing
nucleotide donor sugar synthases, glycosyltransferases and gene products involved in export and
ligation. An undecaprenyl phosphate lipid is modified sequentially by the enzymes PglC and PglA which
add a 2,4-diacetamido-2,4,6-trideoxyglucose and an α-(1,3) linked GalNAc, respectively.26 The primase
PglJ adds an α-(1,4) linked GalNAc to generate the trisaccharide primer substrate recognized by the
polymerase PglH.27 The three priming enzymes exhibit a strong preference for lipids bearing alpha
unsaturation with cis configuration.25 Saturation of the proximal bond reduced the enzyme reaction rate
of PglC, PglJ and PglB to approximately 25%, 37% and 39% of normal, respectively. Switching from cis to
trans geometry reduced the primase activities to 2%, 30% and 13% respectively. If the proximal
12 geometry was preserved, but the length reduced, the enzymes only exhibited mild reductions in
reaction rate. The lipid specificity of the polymerase PglH has not been evaluated. Likewise, the ligase
PglB has not been investigated for lipid specificity, but it has been noted that it exhibits relaxed
specificity for both the glycan and acceptor substrates.22,28,29 In fact, if PglB is expressed in E. coli ΔWaaL,
the enzyme will transfer the accumulated O-antigen to proteins.30,31 The possibility for engineering
glycoproteins using the pgl gene cluster is of broad interest.22,30,31
Figure 1.05. The polymerase Wzy from E. coli has high specificity for the undecaprenyl phosphate carrier lipid whereas the preceding primases exhibit higher promiscuity of the lipid. The reverse is true for N-glycan assembly in C. jejuni. The promiscuity of the ligase PglB has been exploited to generate non-natural proteoglycans.
Exopolysaccharides are a class of polysaccharides which do not require priming of a lipid carrier.
Polymerization of these structures occurs directly from the nucleotide sugar precursors and oligomeric
acceptors, and is often coupled to export. P. aeruginosa produces the exopolysaccharide alginate which
is present in the extracellular matrix of biofilm-forming infections like those in the lungs of cystic fibrosis
patients.3,32 The membrane protein Alg8 polymerizes a homopolymer of mannuronic acid and couples
13 elongation to transport of the nascent chain to the periplasm.33,34 The epimerase AlgG present in the
periplasm randomly converts some mannuronic acids to the C5 epimer, guluronic acid.35 The guluronic
acids protect nascent alginate chains from degradation by a lyase AlgL.36 Additional post-polymerization
acylations are added to the 2-O and 3-O positions of ManA. Export across the outer membrane is
mediated in part by additional chaperones like Alg44,37,38 AlgK39 and AlgX.40 Without a lipid anchor to
target the polymer for ligation to the membrane, alginate is released into the extracellular matrix where
it exerts its effects through regulation of the biofilm. In this sense, it is its lack of targeting sequence
which directs it to the appropriate extracellular location.
1.3.3 Substrate recognition: Acceptor binding and the role of primases
In addition to lipid requirements, the acceptor is also recognized by the polymerase at the
carbohydrate terminus. Almost all carbohydrate polymerases recognize a short primer of sugar residues
constructed by one or more glycosyltransferases acting upstream.41,42 These primases have a conserved
role in glycan biosynthesis. They act as gatekeepers for their promiscuous polymerizing counterparts by
synthesizing the carbohydrate primer required for substrate recognition. One intriguing example of the
importance of priming glycosyltransferases involves the biosynthesis of wall teichoic acids (WTAs) in
Staphylococcus aureus and Bacillus subtilis.43 The glycosyltransferases which assemble teichoic acid exist
in large multienzyme complexes and produce polyphosphate chains on undecaprenyl carriers on the
cytosolic face of the membrane.44 The WTAs are polyribitol or polyglycerol phosphate polymers
anchored to the peptidoglycan of gram positive bacteria. Disruption of WTA biosynthesis results in
abnormal growth and septation defects.45 Because teichoic acids act as divalent cation reservoirs, their
absence may cause the altered phenotype via decreased glycosyltransferase activity. Since most
glycosyltransferases require a divalent cation for catalytic activity, having teichoic acids in close
proximity to sites of cell wall biosynthesis may be beneficial.46,47 Deficient cells also exhibit increased
14 sensitivity to β-lactam antibiotics, implicating WTA polymers in the regulation of peptidoglycan
crosslinking.48
The teichoic acids of S. aureus and B. subtilis come in two varieties: the wall teichoic acids
attached to the peptidoglycan and the lipoteichoic acids, which are membrane anchored via a
phospholipid. Biosynthesis of WTAs in S. aureus occurs on an Und-P carrier on the cytoplasmic face of
the membrane.6 After a short disaccharide sequence is assembled on the lipid, the enzyme TagB
transfers a glycerol phosphate residue from CDP-glycerol to the C6-OH of the disaccharide. The enzyme
TarF then acts by transferring a second glycerol phosphate, and polymerization follows by the actions of
TarL.49
In B. subtilis the nomenclature for the WTA biosynthetic machinery is similar to S. aureus, but
the roles are somewhat shifted. After TagB transfers a glycerol phosphate to the same disaccharide
acceptor, the transferase TarK transfers a ribitol phosphate to the chain. Polymerization then
commences through the actions of B. subtilis’ TarL analog. Interestingly, the Walker Group identified an
analog of TarF in B. subtilis that was capable of generating oligosaccharides of glycerol phosphate in
vitro, however attempts to isolate WTAs with polyglycerol phosphate backbones were unsuccessful.49
1.3.4 Substrate recognition: Donor binding
By localizing polysaccharide synthesis to specific intra- or extracellular environments the
availability of component parts for polymer assembly is altered. The activated monosaccharide building
blocks found on the cytosolic side of the membrane are soluble nucleotide diphosphate (NDP) sugars
such as UDP-Glc or GDP-Man or CDP-glycerol (CDP-Gro). These NDP donors are recognized by
glycosyltransferases inside the cell. For catalysis of glycosylation, cytosolic glycosyltransferases have
evolved two strategies for binding and orienting the donor sugars. The first strategy uses acidic or polar
residues to coordinate a divalent metal cation like magnesium or manganese in the donor binding site.
15 (Figure 1.06) The pyrophosphate moiety also coordinates to the metal, positioning the donor for
nucleophilic attack. The mycobacterial galactan, gram negative O-antigens in the ABC pathway and wall
teichoic acids are examples of polymers synthesized from NDP precursors. The second strategy is to use
a basic residue to position the donor pyrophosphate. On the cell exterior, NDP sugars are not present in
controlled amounts. Glycosyltransferases operating extracellularly favor a strategy of coupling two lipid-
linked glycans, using one as an acceptor and the other as a donor. Peptidoglycan synthesis, O-antigens
assembled via the Wzx/Wzy pathway and lipoteichoic acids favor this extracellular assembly strategy.
Figure 1.06. Glycosyltransferases bind NDP donor sugars by coordinating the pyrophosphate group, positioning it for nucleophilic displacement. In order to traffic polymers with the same composition to different destinations, organisms will
exploit the intracellular and extracellular assembly pathways. Consider the biosynthesis of lipoteichoic
acid and wall teichoic acid from B. subtilis. Both are linear repeating sequences of glycerol phosphate.
Lipoteichoic acids are anchored to the cell membrane via a phosphatidyl glycerol lipid. Wall teichoic
acids are ligated to the C6-OH of MurNAc residues on peptidoglycan. Lipoteichoic acids are assembled
on the extracellular side of the membrane using phosphatidyl glycerol lipids as donors for
polymerization. Wall teichoic acids are assembled in the cytosol on Und-P carrier lipids. This physical
16 separation of biosynthesis is necessary for targeting to the appropriate structure and effected via the
utilization of chemically unique donors.
1.3.5 Substrate recognition: Determining the order of binding
In generic terms, a glycosyltransferase binds an acceptor and a donor and converts those two substrates
into two products: the glycosylated acceptor and the byproduct of donor hydrolysis. This reaction type is
termed a Bi Bi reaction and it can have several distinct kinetic mechanisms, the most common of which
are ping pong, sequential ordered or sequential random (Figure 1.07). In ping pong enzyme kinetics, the
enzyme binds one substrate followed by generation of the first product, then the second substrate binds
and the second product is made. In sequential ordered reactions, the binding of one substrate is
contingent on the binding of the second substrate and the release of the products follows a similarly
ordered pattern. Sequential random enzyme mechanisms are such that each substrate binds
independently of the other and the products are released in a similarly random order.
Figure 1.07. Common Bi Bi mechanisms proceed via a ping pong (top left), sequential ordered (bottom left) or random ordered (top right) binding mechanism. A fourth mechanism, Theorell-Chance, is a variant of ordered binding in which the kinetic parameters indicate turnover is rapid upon assembly of the ternary complex.
17
Understanding the binding order of substrates for glycosylation is important for several reasons.
Crystals for x-ray crystallography, high throughput assays for inhibitors, or information regarding binding
surfaces can all be aided by this mechanistic inquiry. Enzyme kinetics can reveal which mechanism is
operational for a specific glycosyltransferase. This strategy has been successfully employed to determine
the mechanism of many glycosyltransferases.50-52 Many polymerizing glycosyltransferases employ a
sequential ordered mechanism.50-52 Attempts at isolating co-crystals and developing high throughput
inhibitor screens have also been aided by knowing substrate binding order.
1.4.1 Polymerization: Directionality
In polysaccharide synthesis, the polymerization event itself is mediated by one or more enzymes
that add a repeating sequence of residues to an acceptor. The sequence can be a simple homopolymer
of identical monosaccharides or a heteropolymer of repeating monosaccharides of varying complexity.
Elongation can occur on the cytoplasmic face of the membrane, on the extracytoplasmic face, or as a
nascent chain passages through the bilayer in the pore of a transmembrane-bound polymerase. Many
features of enzyme-specific polymerization events have been defined, including the direction of
elongation, processivity, length and patterning control. By defining the mechanism of polymer assembly
within these categories, researchers are able to focus on how to enhance, augment, re-engineer or
inhibit production of a polysaccharide. Complete understanding of the biosynthetic process promises to
provide tremendous leverage in our efforts to elucidate polysaccharide function.
A polysaccharide chain can be extended from its monomeric building blocks at either the
reducing or non-reducing terminus. The two mechanisms differ in which component acts as nucleophile
and which acts as electrophile, or in other words, which acts as donor and which as acceptor. (Figure
1.08) Experiments which determine elongation direction either rely on chemical block at one direction
or differentially label the reducing and non-reducing termini. The Walker Group determined the
18 direction of elongation in peptidoglycan transglycosylation reactions by developing a compound which
blocked elongation at the non-reducing end.53 By swapping the native glucose residue for a galactose, a
subtle switch which preserved substrate binding, they generated a substrate that could only be
elongated from one direction. Observation of elongated products proved the directionality of growth to
be from the reducing end.53
The Yother group developed a strategy to distinguish between elongation directions of capsular
polysaccharide production in S. pneumoniae. In a variation on a pulse-chase experiment, the
bifunctional polymerase Cps3S was incubated with tritiated UDP-Galp and then with 14C-labeled UDP-
GalA.54 By treating the elongated polymers with a depolymerase known to act at the non-reducing
terminus of cellubiuronan polymers, they were able to isolate short oligomers from the reducing end of
depolymerized chains. They found that the reducing end was enriched in tritium indicating elongation
occurrs from the non-reducing terminus.54 Both examples highlight how careful selection of the
appropriate chemical label can yield insight into the mechanism of polymerization.
1.4.2 Polymerization : Processivity
A glycosyltransferase capable of polymerization can do so in a processive or distributive fashion. In
processive polymerization, the enzyme maintains contact with the nascent chain between individual
elongation events. In distributive polymerization, each intermediate-length chain dissociates from the
enzyme and must re-bind prior to subsequent elongation. The mode of elongation has consequences
for the regulatory features of polymer synthesis: length and pattern control. Distinguishing between
these mechanisms involves biasing the experimental conditions to statistically favor or disfavor binding
of an acceptor to the enzyme. Several strategies have been devised to elucidate processivity in
carbohydrate polymerases. Many of the techniques used to assess processivity employ isotopically-
19
Figure 1.08. Polymer elongation can occur at the reducing or non-reducing end of the growing chain. In general, when both donor and substrate are lipid-linked glycans, elongation occurs at the reducing end. Enzymes which utilize NDP-sugar donors favor elongation at the non-reducing terminus. One method for determining elongation direction of cellubiuronan biosynthesis involved a double labeling experiment followed by treatment with a depolymerase known to act at the non-reducing terminus. The radioisotope enriched to a further degree on isolated oligomers suggested elongation occurred via the non-reducing terminus.
20 labeled substrates. Classically, these experiments are similar to those performed by Messelson and Stahl
in which “heavy” and “light” DNA could be tracked to elucidate the mechanism of DNA replication.55,56
The Imperali Group investigated the processivity of PglH, a polymerase involved in N-glycan
assembly.51 PglH transfers three GalNAc residues to a primed Und-P carrier. PglH was incubated with
high concentrations of tritiated acceptor and then the +1, +2 and +3 adducts were separated and
quantified. If the enzyme-substrate complex dissociated after each glycosylation, then the chance of re-
binding the same acceptor molecule would be low. Thus, in this “single hit” assay, the observation of
long polymers at short time points indicated the enzyme was processive.51 Under conditions where the
starting acceptor concentration was high relative to enzyme concentration, the predominant product
was the +3 adduct at early time points, and the +1 adduct at late time points. Early accumulation of +3
product suggested that the enzyme is processive, but the number of sequential glycotransfers is
controlled through a product inhibition mechanism: As the concentration of +3 adduct increased, the
accumulation of +1 and +2 intermediates was afforded via competitive inhibition.51
An alternative strategy for assessing processivity was employed by the Yother Group to examine
the type 3 capsular polysaccharide from Streptococcus pneumoniae.2 The polysaccharide is a
cellubiuronan polymer anchored to the membrane by a phosphatidyl glycerol lipid.57 Isolated
membranes containing the membrane-bound polymerase Cps3S were treated in a pulse chase
experiment with 14C-labeled UDP-GlcA donor.54 The reaction products were separated by size exclusion
chromatography and each fraction was measured for radioactivity. Rapid increases in large molecular
weight polymer appeared at short time points (5 and 10 minutes). Identification of long polymers at
early time points suggested Cps3S was highly processive.
The Walker Group used a similar strategy to assess the processivity of peptidoglycan
glycosyltransferase PBP1A from Aquifex aeolicus.53 The enzyme condenses two molecules of lipid II to
21 generate lipid IV and then continues to append disaccharides to the reducing end of the growing chain.
By incubating PBP1A with 14C-labeled lipid IV and a 100-fold excess of unlabeled lipid II, reaction
products could be observed using SDS-PAGE, which separates the polysaccharides by molecular weight.
The observation that large polymers were appearing at short time points indicated that PBP1A operates
via a processive mechanism. The crystal structure has been solved, and the authors interpreted a mobile
loop near the active site to be responsible for maintaining contact with the elongating substrate
between rounds of glycotransfer.18
The mycobacterial galactan is assembled on a decaprenyl carrier lipid and polymerized by the
glycosyltransferase GlfT2.58-61 The Kiessling Group has investigated the processivity of the enzyme using
several techniques. First, by incubating the enzyme with synthetic disaccharide acceptors at 1,000-fold
excess, polymer length was measured at very short time points.62 To quantify the degree of processivity,
an isotope labeling strategy akin to the classic Messelson-Stahl experiment was performed. In this assay,
GlfT2 is preincubated with normal acceptor for a time t1 before treatment with a heavy pentadeuterated
acceptor.56 If the enzyme is processive, the heavy acceptor will not have access to enzyme occupied by
the light acceptor. After a time t2, the product distribution for heavy and light acceptors can be
compared directly. (Figure 1.09) For each polymer elongated by n residues, a larger intensity peak for
the light isotope indicated that GlfT2 was highly processive. The processivity could be quantified by
measuring the ratio of heavy to light peak intensities.
Not every carbohydrate polymerase is processive. The Brown Group analyzed the polymerase
TagF, which generates the polyglycerol phosphate wall teichoic acid in B. subtilis.63,64 Under initial rate
conditions a 14C-labeled acceptor was found to be elongated by TagF in the presence of the donor, CDP-
22
Figure 1.09. A distraction assay using isotopes analogous to the Messelson-Stahl experiments allowed quatitative assessment of GlfT2’s processivity.
Figure 1.10. TagF was shown to be perfectly distributive by demonstrating that saturating amounts of labeled acceptor generated the +1 product in an amount identical to the amount of hydrolyzed donor. glycerol. (Figure 1.10) Surprisingly, the amount of CMP produced was found to be equal to the amount
of +1 product. Because the starting acceptor was in 100-fold excess of the enzyme, the system was
biased against observing a rebinding event. Thus TagF operates by a distributive mechanism.52 The
question is raised concerning how a distributive polymerase can regulate length control without
23 maintaining contact with the nascent strand and what evolutionary advantage is born out of a
distributive mechanism of action. Such advantages and discussed in this system below (vide infra).
1.4.3 Polymerization: Subsites and the role of primases
Processive enzymes must make transient contacts with the saccharide residue acted upon
chemically, as well as with several saccharide residues upstream of the action as a way to maintain
contact as the growing polymer slips over the surface. This is true for all processive enzymes, including
for example DNA and RNA polymerases and ribosomes. Efficient processive polymerization appears to
depend on subsite binding.62 This suggests an important role for primases in the assembly of a
polysaccharide. Perhaps these priming glycosyltransferases produce short oligosaccharide primer
sequences that fill the glycosyltransferase subsites. Acceptor substrates with appropriate priming
eliminate any kinetic lag phase and give rise to rapid polymerization.
The cellubiuronan capsular polysaccharide from Streptococcus pneumoniae is important for
colonization and survival of the pathogen. The polymerase Cps3S exhibits a transition from
oligosaccharide to polysaccharide synthesis triggered by generation of and binding to an oligosaccharide
approximately 6 – 8 residues in length.65 By incubating the enzyme and unlabeled acceptor with 14C
UDP-Glc, separating the products by paper chromatography and then quantifying by liquid scintillation
they observed oligomers rapidly reach a steady state levels while polysaccharide production lags at first
and then exhibits an exponential increase. The polymerase becomes highly processive once it binds an
acceptor approximately 8 monosaccharides in length.65,66 This length dependence suggests that
acceptors capable of filling subsites on the enzyme surface exhibit the highest level of processivity
because they are able to interact with – and thus remain bound to – the enzyme most efficiently.
24 1.4.4 Polymerization: Length control
Regulation of polymer length is a central feature of polymer synthesis. Unlike their nucleic acid
and amino acid counterparts, carbohydrate polymerases act without a template to direct them. This
template-independent feature of polysaccharide synthesis raises the questions of just how the
polymerization regulates length. Polymer length imparts biological relevance in several systems. Chain
length is proportional to cation buffering capacity in teichoic acids.67 Differences in signal transduction
pathway activation are observed depending on the length of glycosaminoglycan present. The degree of
polymerization and crosslinking of the peptidoglycan renders cells resilient or sensitive to osmotic
pressures. How then, is length controlled? The structure of the polymerase itself always exerts some
level of control. There are features intrinsic to the enzyme, the substrates and the ternary complex they
form that dictates the resultant polymer length. Likewise there are extrinsic factors that impart a
regulatory function over polymer length. Finally, regulation at the gene expression level has an influence
as well.
1.4.4.1 Intrinsic factors
The Kiessling group recently put forth a model for length control based on a lipid tethering
mechanism in galactan assembly.62 Curiously, longer polymers were observed following incubation with
a lower concentration of acceptor. Moreover, increasing lipid length also resulted in longer
polymers.21,62 Inhibition with the geranylgeraniol lipid further supported a model wherein the enzyme
binds not only the carbohydrate portion of the acceptor, but the lipid itself in a bivalent manner.
Bivalent binding explains how template independent polymerases are able to regulate length by
increasing the likelihood of rebinding after passive dissociation of the glycan, particularly when the Galf
oligosaccharide is short. At longer galactan lengths, the efficacy of the tether would be reduced and
resemble monovalent binding, at which point a passive dissociation would terminate polymerization.
25 Recently, a crystal structure of GlfT2 was solved.68 The structure has the classic Rossmann-type fold that
glycosyltransferases of the GT-A family share, but has no distinct lipid binding site. To reconcile the
structure with the lipid-dependence of the enzyme, it may be necessary to evoke some interaction with
the lipid bilayer. Indeed the membrane itself could serve as the lipid tethering site.
The Walker group discovered that the lipid tethering model is also valid for PBP1A, the
peptidoglycan glycosyltransferase which catalyzes the construction of the carbohydrate backbone of
peptidoglycan.53 To monitor the effect that lipid length would have on polymerization, the group
synthesized a galactosylated lipid II analog that was incompetent as an acceptor, but still bound the
enzyme as a donor. This allowed them to vary the chemical features of the donor and assess the effect
the changes had on polymerization. The lipid length of the donor, but not the acceptor, determines the
degree of polymerization. They postulate that the lipid tether enhances the processivity of the system.
Because PBP1A elongates from the reducing end, the nascent chain must translocate after each round
of glycosylation to accommodate a new lipid II acceptor. When the lipid length becomes too short,
polymer length decreases to oligomers only 4 – 8 carbohydrates in length.
The capsular polysaccharides of gram-positive bacteria are bound to the peptidoglycan or
anchored to the plasma membrane. Many of these polysaccharides resemble or are identical to the
glycosaminoglycan structures produced in mammalian cells.69 The hyaluronan capsular polysaccharide
from Streptococcus equisimilis is produced by a single enzyme, hyaluronan synthase (HAS). The
hyaluronan polymer has a repeating alternating sequence GlcNAc-β-(1,3)-GlcA-β-(1,4) which is
assembled from UDP donor sugars onto a short hyaluronan oligomeric primer. HAS is an integral
membrane protein that has a strong dependence on phospholipids for its activity.70 Radiation
inactivation studies were performed to determine the monomer is in complex with 16 cardiolipin
phospholipids in its active state.71,72 Polymer size, which can range from 0.2 to greater than 2 MDa, can
26 be measured by a combination of size exclusion chromatography and laser light scattering.73 A western
blot based enzyme capture assay using a biotinylated hyaluronan binding protein in conjunction with
streptavidin-agarose allowed measurement of the percentage of active synthases present in the
membrane at any one time.74
The Weigel Group investigated the role of length control using a combination of sequence
analysis and mutagenesis.75 The predicted topology of HAS included two conserved polar residues, K48
and E327, within two transmembrane domains. Mutation at these conserved residues resulted in
severely reduced enzymatic activity and polymer length. Surprisingly, the double mutant K48E/E327K
restored activity, yet still produced shorter polymers. This led the authors to suggest that perhaps these
residues form a salt bridge that creates a ratcheting mechanism to regulate passage of the growing
chain through the channel during export.
1.4.4.2 Extrinsic factors
A second approach to length control can be found in the O-antigen assembly process, which is
part of the wzy/wzz system. The Wang Group has established biochemically that the wzz gene product
imparts a modality to length. Similarly, lipoteichoic acid length in B. subtilis is regulated by differential
gene expression of four synthases .76 Shigella flexneri uses variable expression of two Wzz proteins,
WzzB and Wzzp-HS2, to generate O-antigens with 17 or 90 repeating units, respectively.77 By using a dual
inducible promoter system, researchers were able to observe that production of the very long O-antigen
required concomitant high levels of Wzy polymerase expression. Low expression of Wzy favored the
shorter repeat. Cross-linking experiments were unable to identify a Wzy-WzzB binding interaction,
suggesting that length may be regulated by an indirect mechanism.77
Two hypotheses are discussed for how Wzz proteins modulate chain length: the molecular
ruler78 and the molecular stopwatch.79 The molecular ruler hypothesis is supported by the observation
27 that Wzz oligomerizes into higher order structures on the membrane surface. These structures mark out
a predetermined polymer length by virtue of the quaternary structure, though supportive biochemical
evidence is lacking. The Whitfield group used full-length Wzz in lipid membranes and observed regular
hexamers by electron microscopy.80 This regularity may support the molecular ruler hypothesis by
negating earlier studies that demonstrated high variability in quaternary structure.
The molecular stopwatch hypothesis put forward by the Reeves Group suggests that Wzz
regulates length by binding either directly to the polymerase or to the polymer for a discrete amount of
time.79 Thus, the kinetics of the complex would directly impart a variation of length control, either
through stabilization or destabilization of the polymerization reaction. The authors expressed skepticism
that a single class of proteins could somehow measure the chain lengths to such variable degrees and
favor the stopwatch hypothesis, but acknowledge that experimental evidence that identifies the binding
interaction and correlates binding affinity to chain length would be necessary to confirm the stopwatch
model.
For the enzyme TagF from B. subtilis, an association with the membrane is necessary for length
control.63,81 Solubilized TagF was found to exhibit a constant Km and reaction velocity regardless of the
degree of polymerization of the acceptor substrate, generating very long polymers of glycerol
phosphate. Membranes from B. subtilis pre-treated with a general proteinase were found to mitigate
length control in TagF polymerization. This suggests that TagF interacts with membrane lipids, although
the specific interaction is not known at this time.
Finally, LTA biosynthesis in B. subtilis is less well understood.11,82 Genetic studies appear to
indicate that there are as many as four synthases expressed.76 One synthase, YvgJ, appears to act as a
primase, exhibiting low hydrolytic activity with a PG substrate and no detectable LTA production in vitro.
The three remaining synthases – YqgS, YfnI and LtaSBS— generate short, medium and long LTAs,
28 respectively, but do so independently of the primase. In systems with all synthases expressed, medium
length LTAs predominate, suggesting that LtaSBS is the major active polymerase. Thus, polymer length
may be determined in part by differential expression. The Grundling Group hypothesizes that different
length LTAs may be beneficial to the cell depending on the different environmental stresses acting on it.
Figure 1.11. Intrinsic and extrinsic factors contribute to polysaccharide length control. Extrinsic factors include interactions with membrane lipids or protein cofactors. Intrinsic factors include conserved salt bridges or lipid binding sites.
29 1.4.4.3 Length control by chemical chain termination
Termination of polymerization is often passive, requiring only a failure to elongate. This kinetic
model is defined by conditions in which the ternary complex becomes unfavorable. Often, termination
is a function of donor sugar concentration. When the concentration falls, elongation is aborted due to
unfavorable kinetics of bringing the acceptor and donor together in a productive turnover. For capsular
polysaccharide type 3, a nascent chain is ejected from the enzyme when the concentration of UDP-GlcA
or UDP-Glc is low. The Yother Group has demonstrated in detail how this dual donor specificity is
leveraged to finely tune the percentage of acceptor cellubiuronan that is elongated.83 Cps3S is
exquisitely sensitive to UDP-GlcA concentration and can be used to predict average chain length.
Chemical modifications to the polysaccharide are a second mechanism used to terminate
polymerization. For the E. coli O9a antigen, a competition between WbdB and WbdD determines
whether the chain is elongated (WbdB) or terminated with a dual phosphorylation-methylation event
(WbdD).84 The Whitfield Group used a fluorescent acceptor in an in vitro assay with purified WbdD to
demonstrate that chemical modification of a non-reducing mannose at the 3-position terminated
polymeriztaion. NMR studies definitively revealed a methylphosphate modification.85 The mechanism by
which this competition between elongation and termination is regulated is not understood, but it has
been shown that WbdA binds to WbdD.86 The product of WbdA is the substrate for WbdD. Thus,
sequestration of the chain terminating enzyme in the proximity of the elongating chain likely plays a part
in biasing the system toward chain termination.
The prokaryotic surface layer (S-layer) is a semi-crystalline array of protein that self assembles to
coat the surface of some bacterial cells.87 These proteins are subsequently modified with an O-glycan. In
Geobacillus stearothermophilus, a gram-positive bacterium, the O-glycan is polymer of rhamnose with a
Rha-α-(1,2)-Rha-α-(1,3)-Rha- β-(1,2) repeating sequence. Assembly of the polysaccharide takes place on
30 the primed carrier lipid Rha-α-(1,3)-Rha-α-(1,3)-Gal-α-1-PP-Und on the cytoplasmic face of the lipid
bilayer.20 The rhamnan is terminated by a 2-OMe group on nonreducing Rha-α-(1,3)-Rha linkage. The
modularity of glycan biosynthesis coupled with the regularity of an S-layer array promises to be an
excellent platform for developing high throughput glycan interrogation technologies.88
The Messner and Schäffer Groups examined the putative glycosyltransferases present in the slg
gene cluster and assigned their functional roles.89 The undecaprenyl lipid is modified by WsaP, WsaD
and WsaC in turn to generate the Rha-Rha-Gal primer. A pair of enzymes, WsaE and WsaF, then act to
generate the repeating trisaccharide sequence. Incubation of the expressed proteins with synthetic octyl
glycoside acceptors revealed that WsaE is a multifunctional enzyme capable of generating the
regiochemical α-(1,2) and α-(1,3) linkages as well as the 2-O-methylation. Thus, WsaE must choose
between elongating a non-reducing Rha-α-(1,3)-Rha via α-(1,2) glycotransfer or chain termination by
methylation. The conditions under which this decision is made have yet to be explored.
1.4.5 Polymerization: Pattern control
One question to consider during polysaccharide assembly is what factors regulate the repeating
patterning during elongation. In template-dependent polymerizations, the sequence fidelity is dictated
by the template; when DNA polymerase encounters cytosine on the template, then guanine is added to
the nascent chain, and when the ribsosome encounters the codon GCC, an alanyl-tRNA binds and
alanine is added. For template-independent polymerases like polysaccharide synthases, the fidelity of
pattern deposition has not been evaluated.
The Kido Group has been studying a polymannose polymer present in the E. coli O9 antigen,
consisting of a repeating sequence of mannose composed of 3 α-(1,2) linkages followed by 2 α-(1,3)
bonds.90 The enzyme WbdA catalyzes the addition of the α-(1,2) linkages while WbdB installs the α-(1,3)
linked sugars. However, a point mutation in WbdA, R55C, confers a specificity slip in which the mutant
31 can only transfer 2 residues with α-(1,3) regiochemistry.90 This suggests that the tertiary structure of
glycosyltransferases is carefully tuned to deliver a certain number of residues.
The polymerase GlfT2 is a bifunctional enzyme with a single active site.91 Bifunctional
glycosyltransferases are capable of generating more than one chemical linkage. This bifunctionality can
arise from transfer of two different monosaccharides, or transfer of the same monosaccharide but with
different regio- or stereoselectivity. The Kiessling group developed a method to monitor sequence
fidelity for bifunctional enzymes that alternate between regiochemical linkages using a chain
termination agent strategy.92 The strategy requires synthetic donors in which one of the hydroxyl groups
that participate in glycosidic bond formation is substituted with a fluorine atom while the other hydroxyl
group is left intact. The fluorine substitution is a close stereoelctronic mimic of the hydroxyl group, so
perturbations to substrate binding are minimized. Incubation of these mimics with a bifunctional
enzyme followed by mass analysis for chain termination can reveal the enzyme’s linkage fidelity: Chain
termination would suggest that the enzyme cannot deviate from the alternating linkage pattern,
whereas polymer detection would suggest that it possesses low fidelity.
Other systems with alternating linkages and bifunctional polymerases could be analyzed using
similar chemistry.93-95 The polysialyltransferase from E. coli capsular polysaccharide assembly alternates
between α-(2,8) and α-(2,9) NeuAc residues. When incubated with 9-azido CMP-NeuAc donors,
polymerization has been shown to terminate.10,96 The Withers Group has recently developed a synthesis
for 8- and 9-fluoro CMP-Sialic acid donors suggesting the technology is present to interrogate the fidelity
of the polysialyltransferase.97 Another capsular polysaccharide, O54 serovar Boreze from Salmonella
enterica, has alternating β-(1,3) and β-(1,4) ManNAc linkages. Polymerization by the enzyme WbbF is
another intriguing target for this approach.93
32
Figure 1.12. The fidelity of GLfT2 was assessed using fluorinated chain termination agents. Analysis of terminated products suggested GlfT2 is chaste with respect to alternating linkage formation. A point mutation in WbdA was found to cause a fidelity slip in the construction of a polymannan O-antigen suggesting that small intrinsic structural features on the glycosyltransferases dictate their ability to orient a nascent chain for appropriate patterning.
1.5 Conclusions
The structural and functional requirements of the biosynthetic machinery engaged in
polysaccharide assembly are varied and complex. Inquiry into the mechanistic detail of many such
systems has yielded an enzyme by enzyme picture of the process. Decades of groundwork has allowed
us to line up the events like movie stills and infer the machinations of these polymerases. An effort has
been made here to categorize the mechanistic underpinnings of polymer construction. Binding order,
processivity, length control, pattern control and termination are important features of biosynthesis
33 because they help us identify discrete events surrounding polymerization that are not yet understood in
toto. Our ability to piece together the molecular mechanisms that regulate polymerization will depend
on leveraging our increasing level of comfort with these technically and chemically challenging systems.
The research frontier should have many fronts: we still have only a partial understanding of the
biological role of polysaccharides, partially because we have yet to connect biosynthesis to structure and
structure to function. With our increasing ability to manipulate and synthesize non-natural variants, we
should be able to delve deeply into the importance of carbohydrate polymer structure and relate it to
function. Chemical biology methods which combine synthetic compounds with well-defined biochemical
systems have proven to be a powerful technique for understanding the step by step assembly process.
Continuing this process and applying it to multiprotein complexes and systems as well as to glycan
engineering projects will be essential.
The future promises to herald some important developments for bacterial polysaccharide
research and technology. Perhaps the most exciting is that our understanding of the requirements for
polysaccharide biosynthesis paves the way for glycan engineering. Our ability to install the glycan of our
choosing onto the protein or array of our choosing will allow for the generation of high throughput
platforms. These platforms are suited to the “omics” world where large amounts of data generated
rapidly and cheaply can be interrogated for discovery. The medical community will benefit from this on
two fronts. Our ability to screen for and identify glycans of medical importance will give rise to new
vaccination strategies and novel approaches to antibiotic treatment.98 The effects polysaccharides have
on virulence and pathogenesis will be better understood through a better understanding of their
biosynthesis. Lastly, research that sheds light onto how multiprotein complexes create and transport
these biomacromolecules will come as new strategies and technologies for studying complex systems
are developed.
34
Chapter 2
Design and Synthesis of Fluorinated UDP-Galf Analogs
Portions of this work are published:
Brown, Christopher D., Rusek, Max S. and Kiessling, Laura L. Fluorosugar Chain Termination Agents as Probes of the Sequence-Specificity of a Carbohydrate Polymerase. J. Am. Chem. Soc. 2012, 6552-6555
Contributions
Max Rusek synthesized provided scale-up quantities of vinyl Galf intermediates
Kenzo Yamatsugu provided all trisaccharide acceptors and the synthetic tetrasaccharide
35 2.1 Introduction
The diversity of carbohydrate polymers in Nature is vast. Polysaccharide structure is defined by
the basic repeating series of sugars that comprise its sequence. Polysaccharide sequences ultimately
determine the functional properties of the polymer. Consider, for instance, the simple case of glucose
(Glc) homopolymers. Cellulose, a structural polymer from plants, is a homopolymer of Glc with β-(1,4)
linkages. Glycogen, a polymer utilized by cells for energy reserves, is also a glucan homopolymer, but
with α-(1,4) linkages. Zymosan is a glucan from yeast that is routinely used as a TLR2 agonist, in which
the glucose monomers are linked via β-(1,3) bonds. Finally, chitin, a structural component of insect
exoskeletons, is a homopolymer of N-acetyl glucose (GlcNAc) with β-(1,4) glycosidic bonds. Each subtle
structural change dramatically changes the function of the polymer. (Figure 2.01)
Figure 2.01: Homopolymers of glucose. Minor structural variations in the polymer impart dramatically different functions to a sugar chain.
At the other extreme, polysaccharides such as bacterial O-antigens can have complicated and
exotic structure with upwards of eight monosaccharides linked together to generate the basic repeating
pattern. Polysaccharides with alternating linkage motifs are common. Most glycosaminoglycans, like
hylauronan for instance, are built on an alternating pattern of N-acetyl and uronic acid functionalized
36 hexoses. The capsular polysaccharide from E. coli K92, which has been investigated for meningococcal
vaccine development, alternates between α-(2,8) and α-(2,9) sialic acid linkages. Agarose, an industrially
important gelling agent, is a polymer of alternating D- and L-galactose derivatives. The O-antigen from S.
enterica O54 serovar Boreeze is a polymer of ManNAc sugars with alternating α-(1,3) and α-(1,4) bonds.
The glycoprotein α-dystroglycan requires an oligomer of alternating xylose and GlcA residues to function
properly. (Figure 2.02)
Figure 2.02: Heteropolysaccharides with alternating sugars or linkages.
For a long time, the field of carbohydrate research held a paradigm: one enzyme, one linkage.
The thinking went that for each glycosylation, a unique glycosyltransferase recognizes the cognate
substrate, binds the appropriate donor sugar and appends that sugar to the acceptor substrate with
regio- and stereochemical control. For each glycosidic bond found in a glycan, a different enzyme would
be responsible for its construction. This rule is broken for glycosyltransferases which are able to act
upon more than one substrate or bind and utilize more than one donor sugar. Some
heteropolysaccharide chains which alternate between two monosaccharides or linkage types are
synthesized by these bifunctional glycosyltransferases. For these enzymes, questions are raised as to
how the two catalytic activities are used in tandem to generate the alternating pattern.
37 2.2 GlfT2: A bifunctional polymerase from Mycobacterium tuberculosis
To study the patterning fidelity of polymerizing bifunctional glycosyltransferases we used the
enzyme GlfT2 as a model system. GlfT2 is a galactofuranosyltransferase from Mycobacterium
tuberculosis.61 It polymerizes a portion of the mycobacterial cell wall termed the galactan. The galactan
is a linear polysaccharide comprised of alternating β-(1,5) and β-(1,6) linked galactofuranose (Galf)
residues. It is built on a primed decaprenyl carrier lipid on the cytosolic side of the membrane, exported
via a putative ABC transport mechanism, modified with additional carbohydrate and lipid structures and
ultimately ligated to the peptidoglycan.58 (Figure 2.03) Besides tethering the mycolic acids to the
peptidoglycan, the functional significance of the galactan is unknown. Deletion studies identify it as an
essential gene: deletion mutants are not viable.99,100 Small molecule inhibitors of the mutase UGM,
which generates UDP-Galf, were shown to be bactericidal.101 Thus, understanding more about the
molecular mechanisms of galactan polymerization is the first step toward designing high throughput
screens and ultimately inhibitors which could serve as new antibiotics.
The galactan was identified as part of a serologically active compound with non-reducing
arabinose residues in the 1930’s. Not until the early 90’s though did GC-MS and FAB-MS experiments
help identify the structure of the polymer as exclusively furanose, the thermodynamically less stable
isomer of galactose. Early biochemical studies using overexpressed enzymes in membrane preparations
indicated that the gene product of Rv3808c (GlfT2) was a glycosyltransferase that utilized UDP-Galf to
modify a membrane phospholipid acceptor.102 Later studies would reveal that synthetic acceptors with
terminal β-(1,5) or β-(1,6) linkages were elongated by up to 4 Galf sugars. Chemical analysis of the +1
products for a β-(1,6) linked disaccharide suggested that the new linkage was β-(1,5). Analysis of the +1
38
Figure 2.03: The galactan is synthesized on a decaprenyl pyrophosphate carrier lipid on the cytosolic face on the membrane. Following polymerization, the polymer is transported to the extracellular face and modified with arabinan and mycolic acids.
product for a β-(1,5) linked disaccharide revealed it to be a β-(1,6) linkage suggesting that there was an
intrinsic ability of the enzyme to set the alternating linkage pattern.59 Expression and purification of
GlfT2 in the Lowary Group confirmed these activities were specific to GlfT2.41 The primase GlfT1
was discovered through careful experiments with isolated natural acceptors which showed that that
native decaprenyl pyrophosphate linked disaccharides and trisaccharides were not substrates for GlfT2,
but were elongated by Rv3782 (GlfT1).61,103 The product tetrasaccharide was found to be elongated to a
pentasaccharide by GlfT2.61 (Figure 2.03) Overall, the biosynthetic sequence of events has been
accepted, however the mechanisms that underlie the regulation of galactan biosynthesis remain
somewhat more mysterious.
39 2.3.1 Synthetic acceptors
The putative natural acceptor of the polymerase GlfT2 is a tetrasaccharide with the structure
Galf-β-(1,5)-Galf-β-(1,4)-Rhap-α-(1,3)-GlcNAc-α-decaprenyl pyrophosphate. The efforts to isolate this
natural product from biological sources have not been high yielding. Additionally, the compound is likely
somewhat unstable and susceptible to hydrolysis at the pyrophosphate bond. For reasons of synthetic
ease and acceptor chemical robustness, efforts in many labs have yielded simplified synthetic GlfT2
acceptors for activity-based study. Generally, GlfT2 activity has been assessed using disaccharides,21,59
trisaccharides61,104,105 and tetrasaccharides106 bearing octyl, decenyl or phenoxy-capped dodecenyl lipids.
This variety of viable structures that GlfT2 acts upon is a convenient way to leverage the promiscuity of
the enzyme for rapid structure-activity relationships. (Figure 2.04)
2.3.2 Synthetic donors
The first synthesis of UDP-Galf was reported by Tsvetkov in 2000 and employs a stereospecific
phosphorylation-phosphate coupling sequence to install the UDP group.107 The strategy for generating
the donor has been improved synthetically108,109 and chemoenzymatically110,111 since then, but the
overall approach has remained similar to the original report. The de Lederkremer lab reported a
synthesis of a C6-3H-labeled UDP-Galf synthesis which has seen moderate use as a measureable probe in
activity assays.112 Similarly, a 14C-labeled UDP-Galf is routinely generated chemoenzymatically by
incubating UDP-14C-Galp with UGM. We noted the paucity of examples with substitution at the C5
position. Fewer reports have been published on the permissiveness of GlfT2 toward donor modification,
but it seemed clear that the synthetic accessibility of acceptor and donor substrates makes possible the
study of galactan biosynthesis using a chemical approach. (Figure 2.05)
40
Figure 2.04. Representative disaccharide and trisaccharide synthetic acceptors recognized by GlfT2 for elongation. Synthetic acceptors with simplified chemical structure have allowed interrogation of GlfT2 activity.
2.4 Chemical probes of pattern fidelity in a bifunctional polymerase
Considering the synthetic feasibility of generating donors and acceptors with specific functional groups,
we returned to the question of pattern fidelity in a bifunctional carbohydrate polymerase. For an
enzyme that can generate two distinct glycosidic linkages in an alternating and repeating pattern, what
regulates the deposition of the sequence, and is polymerization dependent on appropriate patterning?
If an activated glycoside with a bond-blocking functional group was recognized by the enzyme and
41
Figure 2.05. UDP-Galf analogs reported in the literature. Isotopically labeled sugars and substitutions at the C2, C3, C5 and C6 positions have been reported sporadically. Reports of substitution at C5 are lacking.
incorporated into a nascent chain, would the enzyme polymerize a galactan chain with an “incorrect”
pattern? We envisioned such an assay would serve as a measurement of the enzyme’s patterning
fidelity by observing the product profile of a reaction with a non-natural donor. Short polymers would
indicate the enzyme was chaste with respect to pattern deposition whereas long polymers would reveal
its linkage promiscuity. (Figure 2.07)
Figure 2.06. Fidelity assessment assay.
42
Table 2.01. Fluorine is an excellent mimic of the hydroxyl group. Size, shape and electronic contributions to molecular structure and conformation are well watched.113
In determining what the X-group should be, we considered several features of the mimic that
would improve our chances to successfully assay polymerization. Our X group had to adequately
substitute for a hydroxyl to maximize the chances that GlfT2 would recognize the synthetic donor as a
substrate. The substitution of fluorine in place of a hydroxyl group has several advantages. (Table 2.01)
First, fluorine is small relative to a hydroxyl group, so it is a minimal perturbation from a sterics
perspective. In addition, the bond lengths for C-F and C-O are identical. Perhaps the most promising
feature of a fluorine substitution is the stereoelectronic properties. Both C-O and C-F bonds are highly
polarized, as evidenced from Pauling electronegativity values of 3.5 and 4.0, respectively. Strongly
electron withdrawing groups vicinal to hydroxyl groups favor gauche interactions, so it is likely that the
conformational preference of a fluorosugar would be similar to that of a hydroxyl group. Further
evidence in support of this comes from modeling and spectroscopy of 2-fluoroethanol and ethylene
glycol which demonstrates that the conformation and even the hydroxyl proton orientation are
identical.114-115The inductive effects on the vicinal hydroxyl group are not only conformational: they also
43 influence hydroxyl pKA. Since hydroxyl group pKa is thought to be important to the enzyme’s
mechanism of action91 it seemed prudent to select an electron withdrawing functional group that would
mimic the properties of a vicinal diol.
Fluorosugars have been used as mechanistic probes to study glycosyltransferases. The location
of the fluorine on the sugar can have a dramatic impact on the glycosyltransferase activity. The Withers
Group investigated the effect a C2 or C5-Fluorinated glucopyranose donor would have on the
oxocarbenium transition states of glycosylation reactions and discovered that inductive effects were
destabilizing.116 This physical property rendered fluorosugars useful as competitive inhibitors. Metabolic
precursors of fluorosugars can block glycan formation by disrupting nucleotide sugar biosynthesis.
Fluorinated carbohydrates have been used in PET and 19F NMR studies to understand protein-
carbohydrate interactions.117 The evidence of biological compatibility suggested that fluorinated Galf
donors would serve as ideal substrates for GlfT2 interrogation.
2.5 Synthesis of fluorinated Galf analogs
Figure 2.08. Retrosynthesis of UDP-6F-Galf and UDP-5F-Galf to a common intermediate, a vinyl Galf.
2.5.1 Accessing vinyl Galf
We required access to both C6 and C5 fluorinated Galf donor structures 2.01 and 2.02. A 6F-Galf
substrate was accessed by generating a cyclic sulfate from a Galf diol using a published method.
(Ferrieres) The route is useful for substitution at the C6 position, but is difficult to access C5-modified
44 Galf structures. To our knowledge, there are no reports of a 5-fluorinated UDP-Galf. Thus, we devised a
divergent synthetic route that could generate both 5- and 6- fluorosubstituted regioisomers 2.03 and
2.04 from a common intermediate, vinyl Galf 2.05. (Figure 2.08) Accessing the common intermediate
involved a short synthesis from commercially available α-methylgalactopyranose 2.06. Benzylidene
protection of the 4- and 6-OH groups followed by benzyl protection afforded fully protected pyranose
intermediate 2.08. Selective removal of the benzylidene acetal in the presence of the methyl acetal
using iodine in refluxing methanol afforded diol 2.09. Regioselective iodination of the primary alcohol
generated the key intermediate 2.10 for accessing the vinyl furanose scaffold. By treating the
Scheme 2.01. Synthesis of the vinyl Galf intermediate.
iodopyranose with n-butyl lithium and carefully controlling the temperature of the reaction, the alkyl
lithium underwent a self-elimination reaction in excellent yield. Separation of the anomers 2.12 and
2.13 was accomplished by treating the cyclic hemiacetal mixture 2.11 with methyl iodide under phase
transfer conditions (Scheme 2.01). Their isolation allowed access to the vinyl Galf intermediate that we
anticipated could be acted upon by a variety of chemical transformation and allow us to generate non-
natural furanosides by functionalizing the double bond.
45 2.5.2 Setting the stereochemistry at C5
Selective chemical transformations at a specific position are the central challenge for
carbohydrate chemists. While it is possible to differentiate the C6 position based on sterics and the C1
position on oxidation state, it is challenging to selectively operate on the secondary alcohols at C2, C3
and C5. Strategies to circumvent this reactivity problem include either a de novo approach118,119 or tying
the reactivity of the either the C6 or C1 position to one of the undifferentiated positions. The
benzylidene acetal protecting group, which ties together the C6 and C4 positions, is one example of this
approach. The Danishefsky group has pioneered a similar strategy for glycals, which allows selective
functionalization at C2. Other strategies are pervasive in the literature.109,120,121 The vinyl Galf is a
powerful addition to this strategy by allowing selective chemical transformation at the C5 and C6
positions of hexofuranosides.
We recognized a key challenge of this divergent synthesis would be the need to develop
diastereoselective reactions to both C5 epimers. Because nucleophilic fluorination operates via the SN2
mechanism, a synthetic intermediate with the opposite C5 stereochemistry is required for C5
fluorination. The asymmetric fluorination is still a transformation at the synthetic frontier.122 To
Scheme 2.02. Kinetic separation of C5 epimers.
46 circumvent this limitation, we adapted the classic hydrolytic kinetic resolution to effect a kinetically
controlled separation of diastereomers.123 The resolution would have two purposes: It would allow us to
avoid stereoselective additions, and it also provided a chromatographic way to separate the two.
Epoxidation of common intermediate 2.12 using mCPBA generated 2.14 as an inseparable 1:1
mixture of diastereomers. Treatment of this mixture with Jacobsen’s cobalt salen catalyst and 0.5
equivalents of water resulted in clean conversion of the S-diastereomer to the diol (2.16), leaving the R-
diastereomer (2.15) unreacted. Diastereoselectivity varied from 10:1 to > 20:1. The diol and epoxide
were easily separated by silica chromatography. This kinetically controlled separation of diastereomers
allowed for isolation of both stereodefined intermediates necessary to complete the synthesis without
necessitating a large screen for two diastereoselective transformations. (Scheme 2.02)
Scheme 2.03. Installation of fluorine at C5 and C6.
2.5.3 Installation of fluorine
To install fluorine at the C6 position, R-epoxide 2.15 was treated with HF-TEA complex in a
completely regioselective transformation to access fluorinated Galf derivative 2.17. (Figure 2.03)
Previous reports on sugar epoxides had indicated that this acid-catalyzed epoxide opening proceeds via
nucleophilic attack at the less hindered carbon. Hydrogenolysis afforded 6F-methyl-Galf 2.03. The
47 installation of fluorine at the C5 position of Galf proceeded from the diol 2.16. Treatment of the diol
with dibutyl tin oxide in refluxing toluene generated the nucleophilic cyclic tin acetal which was trapped
with benzyl bromide to generate a key intermediate with only the C5-hydroxyl free (2.18). Treatment of
2.18 with diethylamino sulfur trifluoride (DAST) generated a 5-fluoro furanose (2.19). To confirm the
stereochemistry at the C5 position, a benzoylated derivative of 2.04 was prepared and compared to the
hydroxylated analog our group had made previously.21 (Figure 2.09) The 3JHH coupling constant between
H4 and H5 was identical. Because the magnitude of a coupling constant is proportional to the dihedral
angle between nuclei, this suggested that the fluorofuranose had Galf stereochemistry at the C5
position.
2.5.4 Installation of phosphate
Fluorosugar intermediates in hand, we turned to the synthetic endgame: installation of the UDP group.
The two main challenges for this part of the synthesis are controlling the anomeric stereochemistry and
working with the hydrolytic lability of sugar phosphates and pyrophosphates. 6F-methyl-Galf 2.03 was
perbenzoylated using benzoyl chloride in pyridine to generate protected intermediate 2.20. Acetolysis of
2.20 removed the stable methyl acetal in favor of the more activated acetyl anomer 2.21. Treatment
with HBr generated a Galf bromide 2.22. The bromide was displaced with dibenzyl phosphate to form a
protected Galf monophosphate with 4:1 selectivity for the α-anomer (2.23) Hydrogenolysis and mild
saponification yielded 6F-Galf monophosphate 2.25 in 78% yield over two steps. (Scheme 2.06) An
analogous sequence was performed on 5-fluoro Galf intermediate 2.04 to generate 5F-Galf
monophosphate 2.31 (Scheme 2.07)
48
Figure 2.09. A comparison of 1H NMR spectra between bezoylated Galf (left) and 5-fluorofuranose 2.xx (right) revealed both H4 signals had identical coupling constants to H5. Since the magnitude of the coupling constant is proportional to the dihedral angle, the equivalent values suggest the stereochemistry at C5 is also equivalent.
2.5.5 Phosphate coupling
Coupling of Galf phosphates to UMP completed the synthesis of UDP-Galf derivatives. The Kiessling
group had recently published an improved coupling strategy over the original UDP-Galf synthesis that
used an N-methylimidazolium-activated UMP (2.35).108 To generate the N- methylimidazolium, UMP
(2.34) was treated with freshly distilled trifluoroacetic anhydride followed by N-methylimidazole. This
activated UMP solution was added to a solution of Galf phosphate and stirred at room temperature for
4 hours. For the reaction to proceed, it was necessary to add 4+ equivalents of UMP. Unfortunately,
purification by HPLC was hindered by a UMP byproduct that coeluted with the UDP-6F-Galf product. To
circumvent this problem, a different approach to generate activated UMP was persued using
49
Scheme 2.04. Stereoselective phosphorylation of 6F-Galf.
Scheme 2.05. Stereoselective phosphorylation of 5F-Galf
50 benzenesulfonylimidazolium triflate (BSIT).124 BSIT is advantageous over the traditional activation
because it does not use trifluoroacetic anhydride and allowed the equivalents of UMP to be dropped
from 4 to 2. This allowed separation of the UDP-Galf products (2.01 and 2.02) in ~ 10% yield, a 4-fold
improvement. (Scheme 2.08)
Scheme 2.06. Phosphate coupling to generate UDP-Galf analogs.
The yield for generating UDP-Galf using this coupling strategy is typically around 35%. The lower
isolated yields for the fluorinated analogs may be a result of inductive effects on the nucleophilicity of
the Galf phosphates. In future, it may be prudent to abandon the synthetic methodology in favor of a
chemoenzymatic approach. The Field Group reported that an uridyltransferase could generate UDP-Galf
from Galf monophosphate in 70% yield110 and others have used the system to generate UDP-6F-
Galf.111,125 Another potential chemoenzymatic synthesis involves generating a Galf aryl glycoside that
51 may be accepted as a substrate for the uridyltransferase.126 For this purpose however, the material
isolated was used in subsequent biological studies.
In summary, we have developed a divergent synthesis to install fluorine atoms as hydroxyl
group mimics. This strategy allowed access to both the C6 and C5-fluorinated donor sugar analogs of
UDP-Galf. Central to the success of the synthetic route was our use of a kinetic separation of epoxide
diastereomers using the Jacobsen cobalt salen catalyst in a reaction analogous to a hydrolytic kinetic
resolution. This allowed isolation of stereodefined intermediates that could be elaborated in parallel to
the fluorosugar targets. Such a strategy promises to be general for the synthesis of non-natural
furanoside sugars.
2.6 Experimental Procedures
I. General Procedures and Materials
All compounds were purchased from Sigma Aldrich (Milwaukee, WI) or Fischer Scientific (Pittsburgh,
PA). Tetrahydrofuran (THF) and toluene were distilled from sodium/benzophenone ketyl.
Diisopropylethylamine (DIEA), triethylamine (TEA) and dichloromethane (CH2Cl2) were distilled from
calcium hydride. All reactions were run under nitrogen atmosphere unless otherwise stated. Analytical
thin layer chromatography (TLC) was carried out on E. Merck (Darmstadt) TLC plates pre-coated with
silica gel 60 F254 (250 μm layer thickness). Analyte visualization was accomplished using a UV lamp and
charring with p-anisaldehyde solution. Flash chromatography was performed on Scientific Adsorbents
Incorporated silica gel (32-63 μm, 60 Å pore size) using distilled reagent grade hexanes and ACS grade
ethyl acetate (EtOAc) or methanol, CH2Cl2 and acetic acid. UDP sugars were purified by semi-preparative
HPLC using a Dionex carbopak column eluting in a 300mM triethylammonium acetate buffer at neutral
pH.1H. 19F, 31P and 13C nuclear magnetic resonance (NMR) spectra were recorded on Bruker AC-300 or
52 Varian Inova-500 spectrometers, and chemical shifts are reported relative to tetramethylsilane or
residual solvent peaks in parts per million (CHCl3: 1H: 7.26, 13C: 77.0; MeOH: 1H: 3.31, 13C: 49.15; D2O: 1H:
4.79, 13C: referenced to 1H. 19F and 31P referenced to 1H) Peak multiplicity is reported as singlet (s),
doublet (d), doublet of doublets (dd), triplet (t), doublet of triplets (dt), etc. High resolution electrospray
ionization mass spectra (HRESI-MS) were obtained on a Micromass LCT.
II. Experimental Procedures
4,6-O-benzylidene-α-methylgalactopyranose (2.07): α-methylgalactopyranose (10g, 50.5 mmol) was
dissolved in DMF (100mL) in a 1L flame dried RBF and treated with camphor sulfonic acid (1.2g, 5.05
mmol) under N2. Benzaldehyde dimethyl acetal (9.5mL, 60.6 mmol) was added dropwise and the
reaction was stirred at room temperature overnight. The first reaction can be quenched with methanol
and the product precipitated upon addition of IPA to characterize the product. Alternatively, the DMF
solution can be treated with the reagents for the subsequent reaction directly in a two-step one pot
procedure.
2,3-di-O-benzyl-4,6-O-benzylidene-α-methylgalactopyranose (2.08): The solution of benzyldiene acetal
in DMF from the previous reaction was chilled to 0 °C and treated with an NaH mineral oil suspension
(12.6g, 328 mmol). Addition was paced so that the bubbly emulsion did not overflow the container and
stirring remained possible (the reaction gets thick). Benzyl bromide (46mL, 328 mmol) was added
dropwise, the reaction was allowed to warm to room temperature and stirred overnight. The reaction
was quenched with methanol and place in a freezer overnight. White crystals formed. Filtration and
washing with cold methanol yielded 13.8g of pure product. (59% yield over 2 steps) 1H NMR (300 MHz,
CDCl3, δ (ppm), 13C NMR (75 MHz, CDCl3, δ (ppm))
53 2,3-di-O-benzyl-α-D-methylgalactopyranose (2.09): 2,3-di-O-benzyl-4,6-O-benzylidene-α-D-
methylgalactopyranose (15.6g, 33.73 mmol) was dissolved in freshly distilled methanol (250mL) under
N2. Iodine (3.5g, 13.5 mmol) was added and the contents were raised to reflux and stirred overnight. The
contents were cooled to RT and quenched with saturated Na2S2O3 solution until the brown color
dissipated. Methanol was removed by rotovap and the crude mixture was extracted 3x with DCM, dried
over MgSO4, filtered and concentrated in vacuo. Purification by silica chromatography (15% 60%
100% EtOAc in hexane) gave 10.2g of white solid. (81% yield) 1H NMR (300 MHz, CDCl3, δ (ppm), 13C
NMR (75 MHz, CDCl3, δ (ppm))
2,3-di-O-benzyl-6-deoxy-6-iodo-β-D-methylgalactopyranose (2.10): 2,3-di-O-benzyl-β-D-
methylgalactopyranose (4.4 g, 11.8 mmol) was dissolved in dry THF (100 mL) under N2 atmosphere.
Triphenylphosphine (4.6 g, 17.6 mmol) and imidazole (2.5 g, 36.6 mmol). The flask was fitted with a
flame dried condenser and the contents warmed to 50 ˚C. A solution of iodine (4.5 g, 17.6 mmol) in dry
THF (20 mL) was added dropwise over 2h. Once the color of the reaction was a persistent yellow the
contents were cooled to RT, quenched with saturated Na2S2O3 solution until the yellow color dissipated.
THF was removed in vacuo then the contents were diluted with water, extracted 2x into DCM, dried
over MgSO4, filtered and concentrated in vacuo. Purification by silica chromatography (gradient 10%
40% EtOAc in Hex) gave 4.61 g of a colorless oil (81%). 1H NMR (300 MHz, CDCl3, δ (ppm)) 7.42 – 7.22 (m,
10H, Ar-H), 4.89 (d, J= 11.1, 1H, OCH2Ph), 4.73 (m, 3H, OCH2Ph, OCH2Ph), 4.26 (d, J= 7.7, 1H, H-1), 4.11
(dd, J= 3.3, 0.9, 1H, H-4), 3.59 (s, 3H, OCH3), 3.61 – 3.46 (m, 2H, H-2, H-5), 3.50 (dd, J= 9.4, 3.5, 1H, H-3),
3.41 (m, 2H, H-6a, H-6b); 13C NMR (75 MHz, CDCl3, δ (ppm)) 138.75, 137.88, 128.73, 128.53, 128.24,
128.07, 127.85, 124.71, 80.59, 78.70, 75.24, 74.87, 72.96, 67.41, 57.28.
54
2,3-di-O-benzyl-5-eno-β-D-methylgalactofuranose (2.12): Iodogalactopyanose 2.10 (4.61 g, 9.52 mmol)
was dissolved in anhydrous THF (25 mL) under N2 atmosphere and chilled to -78 ˚C. A 2.5M solution of n-
butyllithium in hexanes (8.4 mL, 20.9 mmol) was added and the reaction was warmed to -40 ˚C over 1hr
and then stirred for another 2h. The remaining n-butyllithium was quenched with a 10% w/v solution of
NH4Cl, the product was extracted 3x into ether, the organics dried over MgSO4, filtered and
concentrated in vacuo. The resulting oil was dissolved in DCM (16mL) and treated with 12M NaOH (2
mL). A large stir bar was used to vigorously mix the biphasic solution. Methyl iodide (1.5 mL, 24.1 mmol)
and tetrabutylammonium iodide (358 mg, .97 mmol) were added and the reaction stirred at RT
overnight. The reaction was diluted with DCM, washed 1x with 5% citric acid, 1x with H2O, and 1x with
brine, dried over MgSO4, filtered and concentrated in vacuo. Purification by silica gel chromatography
(gradient 5% 20% EtOAc) allowed isolation of the β anomer (1.54 g, 47%). Characterization of β
anomer: 1H NMR (300 MHz, CDCl3, δ (ppm)) 7.41 – 7.21 (m, 10H, Ar-H), 5.92 (ddd, J= 17.2, 10.3, 7.1, 1H,
H-5), 5.41 (dt, J= 17.2, 1.0, 1H, H-6 trans), 5.24 (dt, J= 10.3, 1.0, 1H, H-6 cis), 4.92 (s, 1H, H-1), 4.56 (s, 2H,
OCH2Ph), 4.53 (ABq, J=11.7, 2H, OCH2Ph), 4.43 (t, J= 6.9, 1H. H-4), 4.00 (dd, J=3.4, 1.1, 1H, H-2), 3.77 (dd,
J= 7.0, 3.4, 1H, H-3), 3.39 (s, 3H, OCH3); 13C NMR (75 MHz, CDCl3, δ (ppm)) 138.05, 137.81, 136.23,
128.69, 128.62, 128.13, 128.03, 118.32, 107.27, 88.88, 87.67, 82.36, 72.56, 72.32, 55.13
5,6-anhydro-2,3-di-O-benzyl-β-D-methylgalactofuranose and 5,6-anhydro-2,3-di-O-benzyl-β-L-
methylaltrofuranose (2.14): 5-eno-β-D-methylgalactofuranoside 2.12(1.1 g, 3.23 mmol) was dissolved in
reagent grade DCM (32 mL) and chilled to 0 ˚C. mCPBA (2.17 g, 9.69 mmol) was added and the contents
stirred at RT for 2 days. The reaction was quenched with saturated Na2S2O3 and NaHCO3 solution and
55 extracted 1x into DCM. The organics were dried over MgSO4, filtered and concentrated in vacuo.
Purification by silica gel chromatography (gradient 5% 20% EtOAc in hexane) gave 947 mg of colorless
oil as a 1:1 diastereomeric mixture of epoxides (97%). 1H NMR (300 MHz, CDCl3, δ (ppm)) 7.41 – 7.21 (m,
20H, Ar-H), 4.95 (d, J= 3.7, 2H, H-1g, H-1a), 4.63 – 4.44 (m, 8H, 2x OCH2Phg, 2x OCH2Pha), 4.03 – 3.97 (m,
2H, H-2g, H-2a), 3.96 – 3.88 (m, 2H,H-3g, H-3a), 3.80 (t, J= 6.6, 2H, H-4g, H-4a), 3.39 (s, 3H, OCH3), 3.38
(s, 3H, OCH3), 3.11 (m, 2H, H-5g, H-5a), 2.75 (m, 4H, H-6g, H-6a); 13C NMR (75 MHz, CDCl3, δ (ppm))
137.79, 137.64, 128.70, 128.69, 128.50, 128.26, 128.15, 128.10, 128.06, 107.69, 107.57, 101.31, 88.09,
87.91, 84.62, 83.82, 82.14, 82.02, 72.65, 72.34, 72.22, 72.15, 55.28, 55.21, 52.17, 51.65, 45.65, 44.03
5,6-anhydro-2,3-di-O-benzyl-beta-D-methylgalactofuranose (2.15) and 2,3-di-O-benzyl-beta-L-
methylaltrofuranose (2.16): (S,S)-N,N’-Bis(3,5-di-tert-butylsalicylidene)-1,2-cyclohexanediaminocobalt
(II) (334mg, .5528 mmol) was dissolved in 200 μL of reagent grade DCM and treated with glacial acetic
acid (25 μL, .442 mmol). The solution stirred open to air for 30 minutes. DCM was removed in vacuo. A
solution of epoxide 2.14 (1.97g, 5.528 mmol) in THF (8mL, 0.7M) and water (50 μL, 2.763 mmol) was
added to the brown residue and the reaction was capped and stirred overnight. The reaction was
stopped when TLC indicated 50% conversion to diol. Quenched with 5 mL of 2M NH3 in MeOH and
stirred for 30 minutes. Concentrated in vacuo. Purified by silica gel chromatography (DCM 60%
EtOAc in hexane, 10% stepwise gradient). Isolated 1.083 g (55%) of epoxide and 958mg (46%) of diol.
Average isolated yields for three separate experiments were 48% for epoxide, 45% for diol.
Characterization of epoxide 8: 1H NMR (300 MHz, CDCl3, δ (ppm)) 7.4-7.2 (m, 10H, Ar-H), 4.93 (s, 1H, H-
1), 4.54 (ABq, 2H, J = 12.1, -OCH2Ph), 4.53 (ABq, 2H, J = 11.6 Hz, -OCH2Ph), 3.99 (dd, J = 3.1, 1.1, 1H, H-2),
3.93 (dd, J = 6.8, 2.9, 1H, H-3), 3.79 (dd, J = 6.7, 6.0, 1H, H-4), 3.37 (s, 3H, OCH3), 3.09 (ddd, J = 5.9, 4.1,
56 2.7, 1H, H-5), 2.74 (AB part of an ABX, JAX = 4.1, JBX = 2.7, 2H, H-6a, H-6b); 13C NMR (75 MHz, CDCl3, δ
(ppm)) 137.80, 137.64, 128.70, 128.69, 128.15, 128.05, 107.57, 88.11, 84.63, 82.02, 72.64, 72.33, 55.21,
52.16, 44.02; MS (EMM)+: m/z: 379.1503 (M+Na)+ (M+Na+ calcd 379.1516) Characterization of diol (#):
1H NMR (300 MHz, CDCl3, δ (ppm)) 7.4 – 7.2 (m, 10H, Ar-H), 4.93 (s, 1H, H-1), 4.53 (ABq, J = 11.9, 2H,
OCH2Ph) 4.50 (ABq, J = 11.8, 2H, OCH2Ph), 4.07 (m, 2H, H-2, H-3), 3.99 (broad s, 1H, H-4), 3.92 (m, 1H, H-
5), 3.64 (m, 2H, H-6a, H-6b), 3.37 (2, 3H, OCH3), 2.56 (d, J = 3.9, 1H, 5-OH), 2.20 (t, J = 6.0, 1H, 6-OH); 13C
NMR (75 MHz, CDCl3, δ (ppm)) 137.56, 137.38, 128.75, 128.73, 128.36, 128.28, 128.26, 128.22, 107.39,
87.56, 82.76, 82.51, 72.49, 72.18, 71.51, 63.57, 55.15; MS (EMM)+: m/z: 397.1634 (M+Na)+ (M+Na+
calcd. 397.1622)
2,3-di-O-benzyl-6-deoxy-6-fluoro-β-D-methylgalactofuranose (2.17): Epoxide 2.15 (45 mg, .126 mmol)
was placed in a 3mL Eppendorf tube and dissolved in HF-triethylamine (180 μL, .7M). The tube was
capped, warmed to 70 °C using a sand bath and incubated for 4 days. The contents were cooled to room
temperature, diluted with acetone and carefully filtered through a silica plug. Concentrated in vacuo.
Purified by silica gel chromatography (5% 20% EtOAc in hexane gradient). Isolated 44 mg of
fluoroalcohol as a pale yellow oil (94%). 1H NMR (300 MHz, CDCl3, δ (ppm)) 7.4 – 7.2 (m, 10H, Ar-H), 4.54
(m, 5H, 2x OCH2Ph, H-6a), 4.34 (1/2 of the AB part of an ABMX, 1H, H-6b), 4.14 (dd, J = 5.8, 2.9, 1H, H-4),
4.05 ( dd, J = 5.7, 2.0, 1H, H-3), 3.98 (d, J = 1.5, 1H, H-2) 3.90 (dtd, J = 14.3, 5.9, 3.1, 1H, H-5), 3.37 (s, 3H,
OCH3), 2.23 (broad s, 1H, 5-OH); 19F NMR (280 MHz, CDCl3, δ (ppm)) -233.7 (td, 48.7, 14.1); 13C NMR (75
MHz, CDCl3, δ (ppm)) 137.71, 137.33, 128.74, 128.29, 128.23, 128.19, 128.16, 107.72, 87.11, 85.25,
83.17, 82.99, 81.31, 81.24, 72.62, 72.16, 70.01, 69.74, 55.18; MS (EMM)+: m/z: 399.1579 (M+Na)+
(M+Na+ calcd. 399.1579)
57
6-deoxy-6-fluoro-beta-D-methylgalactofuranose (2.03): 2,3-di-O-benzyl-6-deoxy-6-fluoro-β-D-
methylgalactofuranose (44 mg, .117 mmol) was dissolved in reagent grade methanol (4.5 mL, .025M)
and treated with 20 wt% palladium hydroxide on carbon (44mg). The flask was evacuated and backfilled
with H2 from a balloon three times. Stirred at STP overnight. Filtered through celite, washed with
methanol and concentrated in vacuo. Isolated 22 mg of a colorless oil (96%). 1H NMR (300 MHz, CDCl3, δ
(ppm)) 4.91 (s, 1H, H-1), 4.55 (AB part of an ABMX, J = , 2H, H-6a, H-6b), 4.5 (m, 3H, H-2, H-3, H-5), 4.1 (s,
1H, H-4), 3.40 (s, 3H, OCH3); 19F NMR (280 MHz, CDCl3, δ (ppm)) -233.5 (td, 48.1, 14.0) 13C NMR (75 MHz,
CD3OD, δ (ppm)) 110.72, 86.51, 84.27, 83.84 (d, J = 27), 83.30, 78.68, 70.48 (d, J = 89), 55.45; MS
(EMM)+: m/z: 219.0630 (M+Na)+ (M+Na+ calcd. 219.0640)
2,3,5-tri-O-benzoyl-6-deoxy-6-fluoro-β-D-methylgalactofuranose (2.20): Fluoro Galf 2.03 (250mg, 1.274
mmol) was dissolved in dry pyridine (12 mL) under N2 atmosphere and chilled to 0 °C. Benzoyl chloride
(565 μL) was added dropwise over 5 minutes. The contents were warmed to RT and stirred for 3 hours.
The reaction was diluted with toluene and azeotroped 3x . Placed on high vacuum line to dry. Purified by
silica chromatography (5% 20% EtOAc in hexane). Isolated 560 mg of white crystalline powder (86%).
1H NMR (300 MHz, CDCl3, δ (ppm)) 8.1 (dd, J = 9.1, 7.2, 4H, Ar-H), 7.90 (d, J = 7.2, 2H, Ar-H), 7.58 (m, 3H,
Ar-H), 7.42 (t, J = 7.8, 2H, Ar-H), 7.30 (t, J = 7.5, 4H, Ar-H), 5.85 (dq, J = 16.0, 5.2, 1H, H-5), 5.59 (d, J = 5.4,
1H, H-3), 5.45 (d, J = 0.9, 1H, H-2), 5.20 (s, 1H, H-1), 4.80 (ddq, J = 46.7, 9.8, 5.9, 2H, H-6a, H-6b), 4.63
(dd, J = 5.2, 3.9, 1H, H-4), 3.47 (s, 3H, OCH3); 19F NMR (280 MHz, CDCl3, δ (ppm)) -235.5 (td, 46.6, 16.0)
13C NMR (75 MHz, CDCl3, δ (ppm)) 165.92, 165.87, 165.71, 133.72, 133.61, 133.56, 130.22, 130.18,
58 130.05, 129.54, 129.20, 128.64, 107.01, 82.50, 82.38, 80.51, 80.44, 80.20, 77.75, 70.99, 70.71, 55.18; MS
(EMM)+: m/z: 531.1450 (M+Na+ calcd. 531.1426)
1-O-acetyl-2,3,5-tri-O-benzoyl-6-deoxy-6-fluoro-D-galactofuranose (2.21): 2,3,5-tri-O-benzoyl-6-deoxy-
6-fluoro-β-D-methylgalactofuranose (100 mg, .197 mmol) was placed in a flame-dried round bottom
flask and treated with a 1.4% v/v solution of concentrated H2SO4 in acetic anhydride (2 mL, .1M). The
reaction stirred at RT for 2 hours. Quenched with saturated NaHCO3 solution, diluted with brine and
extracted into DCM 3x. Dried organic layer over MgSO4, filtered and concentrated in vacuo. Azeotroped
remaining acetic acid with toluene 2x. Purified by silica gel chromatography (5% 20% EtOAc in
hexane) to give 97 mg of white crystalline solid (92%, 5:1 mix of β:α). 1H NMR (300 MHz, CDCl3, δ (ppm))
8.10 (m, 6.5H, Ar-H), 7.90 (d, J = 7.5, 2H, Ar-H), 7.6 – 7.2 (m, 13H, Ar-H), 6.61 (d, J = 4.8, .2H, H-1α), 6.50
(s, 1H, H-1β), 6.17 (t, J = 6.6, .2H, H-2α), 5.89 (dq, J = 16.2, 5.0, 1H, H-5β), 5.75 (m, .4H, H-3α, H-5α), 5.63
(d, J = 4.7, 1H, H-3β), 5.59 (s, 1H, H-2β), 5.0 – 4.8 (m, ~4H, H-4α, H-4-β, H-6aα, H-6aβ, H-6bα, H-6bβ),
2.20 (s, 3H, C(O)CH3 β), 2.13 (s, .6H, C(O)CH3 α); 19F NMR (280 MHz, CDCl3, δ (ppm)) -235.5 (td, J = 46.3,
15.7, β anomer), -237.5 (td, J = 46.1, 19.3, α anomer); 13C NMR (75 MHz, CDCl3, δ (ppm)) 169.35, 165.84,
165.72, 165.50, 133.92, 133.80, 133.65, 130.24, 130.10, 129.34, 129.07, 128.89, 128.77, 128.68, 128.65,
99.55, 93.65, 82.85, 82.79, 82.27, 81.64, 79.97, 77.68, 76.49, 70.74, 70.46, 21.29, 21.23; MS (EMM)+:
m/z: 559.1380 (M+Na+ calcd. 559.1375)
Dibenzyl 2,3,5-tri-O-benzoyl-6-deoxy-6-fluoro-α-D-galactofuranose monophosphate (2.23): 1-O-
acetyl-2,3,5-tri-O-benzoyl-6-deoxy-6-fluoro-D-galactofuranose (100 mg, .186 mmol) was placed in a
flame dried round bottom flask and dissolved in dry DCM (2.0 mL) under N2 atmosphere. The solution
59 was chilled to 0 °C and treated with a 33 wt% HBr in AcOH solution (120 μL). After stirring for 20 minutes
the reaction was warmed to RT and stirred for 4 hours. Azeotroped using freshly distilled toluene 3x to
isolate the crude glycosyl bromide as an orange oil. The oil was dissolved in dry toluene (1.0 mL) under
N2. The solution was added dropwise over 30 minutes to a stirring solution of dibenzyl phosphate (208
mg) in dry toluene (1.0 mL). The contents stirred at RT for 2.5 hours. Concentration in vacuo followed by
silica gel chromatography (12% EtOAc in toluene) allowed isolation of 97 mg of the α-anomer as a
slightly cloudy viscous oil (69%). 1H NMR (300 MHz, CDCl3, δ (ppm)) 8.14 (d, J = 7.0, 2H, Ar-H), 8.0 (t, J =
7.1, 4H, Ar-H), 7.60 – 7.00 (m, 19, Ar-H), 6.32 (dd, J = 5.6, 4.7, 1H, H-1), 6.13 (t, J = 7.2, 1H, H-3), 5.70 (m,
2H, H-2, H-5), 5.05 – 4.55 (m, 7H, 2x OCH2Ph, H-6a, H-6b, H-4); 19F NMR (280 MHz, CDCl3, δ (ppm)) -
237.2 (td, J = 46.8, 17.3); 31P NMR (122 MHz, CDCl3, δ (ppm)) -1.88 (m); 13C NMR (75 MHz, CDCl3, δ
(ppm)) 165.80, 165.76, 165.72, 135.65, 135.54, 133.92, 133.84, 133.58, 130.33, 130.24, 130.15, 130.04,
129.34, 128.87, 128.73, 128.65, 128.00, 127.89, 97.85, 97.79, 81.69, 79.37, 79.29, 76.68, 73.43, 71.53,
71.25, 69.62, 69.55, 69.45; MS (EMM)+: m/z: 777.1890 (M+Na+ calcd. 777.1872)
Triethylammonium 2,3,5-tri-O-benzoyl-6-deoxy-6-fluoro-α-D-galactofuranose monophosphate (2.24):
Dibenzyl 2,3,5-tri-O-benzoyl-6-deoxy-6-fluoro-α-D-galactofuranose monophosphate (40 mg, .053 mmol)
was dissolved in ethyl acetate (500 μL, .1M) and triethylamine (50 μL). 10% Pd on carbon (20 mg) was
added. The stoppered flask was evacuated and back filled with H2 from a balloon. The contents stirred at
STP for 90 minutes. The reaction was filtered through celite, washed with ethyl acetate and
concentrated in vacuo. Isolated 33 mg of cloudy, viscous oil (90%). 1H NMR (300 MHz, CDCl3, δ (ppm))
8.10 (t, J = 7.2, 4H, Ar-H), 7.90 (d, J = 7.2, 2H, Ar-H), 7.5 – 7.2 (m, 9H, Ar-H), 6,21 (t, J = 7.2, 1H, H-1), 6.13
(dd, J = 7.4, 4.5, 1H, H-3), 5.69 (dq, J = 18.7, 4.8, 1H, H-5), 5.60 (ddd, J = 7.7, 4.5, 1.4), 5.00 (m, 2H, H-6a,
60 H-6b), 4.54 (t, J = 5.4, 1H, H-4), 2.85 (q, J = 7.2, NCH2CH3), 1.2 (t, J = 7.3, NCH2CH3);
19F NMR (280 MHz,
CDCl3, δ (ppm)) -238.1 (td, J = 48.7, 18.9); 31P NMR (122 MHz, CDCl3, δ (ppm)) -1.88 (d, J = 7.9); 13C NMR
(75 MHz, CDCl3, δ (ppm)) 174.93, 66.03, 165.70, 133.43, 133.23, 133.16, 130.41, 130.03, 129.83, 129.54,
129.40, 128.50, 128.44, 128.26, 95.98, 82.61, 80.34, 74.16, 72.44, 72.19, 45.46, 8.57; MS (EMM)-: m/z:
573.0941 ([M-H]-, calcd. 573.0967)
Tributylammonium 6-deoxy-6-fluoro-α-D-galactofuranose monophosphate (2.25): Benzoylated galf
monophosphate (#) (42 mg, .06 mmol) was dissolved in a 5:2:1 mixture of methanol:water:triethylamine
(2.3 mL) and warmed to 30 °C. The reaction stirred for 4 days. Concentrated in vacuo then azeotroped
3x with toluene. Purified by anion exchange chromatography (Biorad AG1-X8 resin, gradient elution 250
mM 500 mM NH4HCO3 solution) using a peristaltic pump. Column fractions were combined, frozen
on dry ice and lyophilized. The residue was treated with ~200 μL Dowex-50WX8-200 resin (pH 7,
pretreated with 2.0M HNBu3Cl) for 4 hours. The resin was removed by filtration through a cotton plug
and the filtrate was frozen on dry ice and lyophilized. Isolated 18 mg of fluffy white solid (56%, 1.1
equivalents of tributylammonium). 1H NMR (300 MHz, D2O, δ (ppm)) 5.45 (t, J = 4.8, 1H, H-1), 4.65 – 4.40
(AB part of an ABMX, J = 47.0, 5.7, 3.7, 2H, H-6a, H-6b), 4.21 (dd, J = 8.6, 7.2, 1H, H-3), 4.06 (ddd, J = 8.4,
4.3, 2.3, 1H, H-2), 3.88 (dtd, 20.8, 5.4, 3.9, 1H, H-5), 3.83 (dd, J = 7.2, 5.6, 1H, H-4), 3.10 (m, 7H), 1.65 (m,
7H), 1.38 (sextet, J = 7.3, 7H), 0.9 (t, J = 7.3, 10H); 19F NMR (280 MHz, D2O, δ (ppm)) -235.6 (td, J = 46.9,
21.1); 31P NMR (122 MHz, D2O, δ (ppm)) -0.3 (d, J = 3.5); 13C NMR (75 MHz, D2O, δ (ppm)) 97.14 (d, J =
4.9), 85.20, 82.98, 81.07 (d, J = 7.0), 76.63 (d, J = 7.5), 73.59, 70.76 (d, J = 19), 52.85, 25.36, 19.43, 12.92;
MS (EMM)-: m/z: 261.0176 ([M-H]-, calcd. 261.0181)
61 Uridine diphosphate 6-deoxy-6-fluoro-α-D-galactofuranose (2.01): A solution of tributylammonium
UMP (16mg, .03 mmol) in anhydrous DMF (470 μL) was stirred with activated 4 Å molecular sieves under
N2 for 1 hour. The molecular sieves were removed and the solution was treated with
diisopropylethylamine (17 μL, .1 mmol) and 1-methyl-3-benzenesulfonylimidazolium triflate (12.4 mg,
0.4 mmol) and stirred for 1 minute. The solution was added dropwise over 30 seconds to a solution of
tributylammonium 6-deoxy-6-fluoro-α-D-galactofuranose monophosphate (10mg, .02 mmol) and MgCl2
(3mg, .03 mmol) in anhydrous DMF (220 μL) at 0 °C. The contents were raised to room temperature and
stirred for 2 hours. The reaction was quenched with 1.0 mL of 100 mM ammonium acetate and washed
1x with 1.0 mL of DCM. The DCM was extracted with 1.0 mL of 100 mM ammonium acetate, the
aqueous layers were combined, frozen on dry ice and stored at -20 °C until purification by HPLC. Purified
by HPLC using a semiprep carbopak dionex column, 300mM tirthylammonium acetate, pH = 6.0-7.0,
flow rate = 1mL/min. Collected peak eluting at 40 – 45 minutes, froze on dry ice and lyophilized to give a
light brown solid. 1H NMR (500 MHz, D2O, δ (ppm)) 7.98 (d, J = 8.3, 1H, H-5u), 5.99 (m, 2H, H-1r, H-6u)
5.67 (t, J = 4.2, 1H, H-1g) 4.62 (m, 2H, H-6ag, H-6bg) 4.15 – 4.42 (m, 7H, H-2g, H-2r, H-3g, H-3r, H-4r, H-
5ar, H-5br) 4.01 (dq, J = 20.5, 5.0, 1H, H-5g) 3.91 (t, J = 7.5, 1H, H-4g); 19F NMR (280 MHz, D2O, δ (ppm)) -
234.8; 13C NMR (125 MHz, D2O, δ (ppm)) 183.52, 168.86, 154.46, 144.27, 105.28, 100.30, 90.95, 87.08,
85.93, 85.85, 85.75, 83.82, 83.77, 79.33, 79.27, 76.38, 76.19, 73.52, 73.38, 72.28, 67.50, 49.28, 10.81;
MS (EMM)+: m/z: 591.0406 ([M+Na]+, calcd. 591.0400)
2,3,6-tri-O-benzyl-β-L-methylaltrofuranose (2.18): Dibutyltin oxide (244 mg, .979 mmol) was placed in a
flame-dried round bottom flask flushed with N2. A solution of dibenzyl Altf 2.16 (282 mg, .753 mmol) in
hot toluene (2.5 mL) was added and the flask was equipped with a claisen adapter, thermometer and
62 reflux condenser. The ensemble was tilted 45° such that the bend in the claisen adapter could collect
refluxing solvent. A spatula tip of MgSO4 was placed in the adapter and the reaction was heated to
reflux. After 5 hours, the claisen adapter was removed and the condenser was fitted directly to the
reaction flask. Benzyl bromide (116 uL, .979 mmol) and tetrabutylammonium bromide (24 mg, .075
mmol) were added the reflux continued overnight. The contents were cooled to RT and quenched with
aqueous KF. Toluene was isolated, washed with brine, dried over MgSO4, filtered and concentrated in
vacuo. Purification by silica gel chromatography (5% 20% EtOAc in hexane) gave 237 mg of colorless
oil (70%). 1H NMR (300 MHz, CDCl3, δ (ppm)) 7.4 – 7.2 (m, 15H, Ar-H), 4.94 (s, 1H, H-1), 4.50 (m, 6H, 3x
OCH2Ph), 4.22 (t, J = 4.9, 1H, H-4), 4.15 (dd, J = 4.8, 1.1, 1H, H-3), 3.99 (m, 1H, H-5), 3.95 (d, J = 1.2, 1H,
H-2), 3.55 (AB part of an ABX, JAX = 3.9, JBX = 6.7, JAB = 9.9, 2H, H-6a, H-6b), 3.37 (s, 3H, OCH3), 2.50 (d, J =
3.5, 1H, 5-OH); ); 13C NMR (75 MHz, CDCl3, δ (ppm)) 138.02, 137.55, 128.69, 128.64, 128.61, 128.30,
128.16, 128.10, 128.02, 127.96, 107.46, 87.32, 83.31, 82.65, 73.68, 72.22, 71.90, 71.13, 70.82, 55.07; MS
(EMM)+: m/z: 487.2116 ([M+Na]+, calcd. 487.2092)
2,3,6-tri-O-benzyl-5-deoxy-5-fluoro-β-D-methylgalactofuranose (2.19): 2,3,6-tri-O-benzyl-β-L-
methylaltrofuranose (26 mg, .056 mmol) was dissolved in dry DCM (500 μL) under N2 and chilled to -10
°C. DAST (11 μL, .084 mmol) was added dropwise. After stirring for 30 minutes the reaction was warmed
to RT and stirred for an additional hour. The contents were chilled to 0 °C and stirred with aqueous
NaHCO3 (2.0 mL) for 30 minutes. Extracted into chloroform, washed with water, dried over MgSO4,
filtered and concentrated in vacuo. Purification by silica gel chromatography (5% 10% EtOAc in
hexane) gave 12 mg of a colorless oil (46%). 1H NMR (300 MHz, CDCl3, δ (ppm)) 7.4 – 7.2 (m, 15H, Ar-H)
4.93 (s, 1H, H-1), 4.75 (ddt, J = 48.4, 6.7, 3.7, 1H, H-5), 4.51 (m, 6H, 3x OCH2Ph), 4.12 (ddd, J = 22.6, 6.9,
63 3.5, 1H, H-4), 4.05 (m, 1H, H-3), 3.98 (m, 1H, H-2), 3.7 (m, 2H, H-6a, H-6b), 3.37 (s, 3H, OCH3);
13C NMR
(75 MHz, CDCl3, δ (ppm)) 138.03, 137.74, 137.63, 128.67, 128.14, 127.96, 127.91, 107.51, 92.03, 89.65,
87.91, 82.78 (d, J = 6.5), 80.21 (d, J = 17.9), 73.74, 72.62, 72.25, 69.69, 69.37, 55.20, 29.92; 19F NMR (280
MHz, CD3OD, δ (ppm)) -204.9 (dq, J = 44.9, 23.7); MS (EMM)+: m/z: 489.2036 ([M+Na]+, calcd. 489.2048)
5-deoxy-5-fluoro-beta-D-methylgalactofuranose (2.04): 2,3,6-tri-O-benzyl-5-deoxy-5-fluoro-β-D-
methylgalactofuranose (51 mg, .109 mmol) was dissolved in reagent grade methanol (4.5 mL, .025M)
and treated with 10 wt% palladium hydroxide on carbon (25mg). The flask was evacuated and backfilled
with H2 from a balloon three times. Stirred at STP overnight. Filtered through celite, washed with
methanol and concentrated in vacuo. Isolated 22 mg of a colorless oil (100%). 1H NMR (300 MHz,
CD3OD, δ (ppm)) 4.71 (d, J = 1.6, 1H, H-1), 4.58 (ddt, J = 48.4, 6.9, 3.8, 1H, H-5), 4.0 – 3.6 (m, 5H, H-2, H-
3, H-4, H-6a, H-6b), 3.33 (s, 3H, OCH3); 19F NMR (280 MHz, CD3OD, δ (ppm)) -209.3 (dq, J = 48.6, 20.1);
13C NMR (75 MHz, CD3OD, δ (ppm)) 109.25, 93.37, 91.01, 82.35, 81.30, 81.05, 76.92, 76.85, 61.49, 61.17,
54.22; MS (EMM)+: m/z: 219.0638 ([M+Na]+, calcd. 219.0640)
2,3,6-tri-O-benzoyl-5-deoxy-5-fluoro-β-D-methylgalactofuranose (2.26): Fluoro Galf 2.04 (20mg, .099
mmol) was dissolved in dry pyridine (1.0 mL) under N2 atmosphere and chilled to 0 °C. Two crystals of
dimethylamino pyridine were added. Benzoyl chloride (60 μL) was added dropwise over 5 minutes. The
contents were warmed to RT and stirred for 1 hour. The reaction was diluted with toluene and
azeotroped 3x . Placed on high vacuum line to dry. Purified by silica chromatography (5% 10% EtOAc
in hexane). Isolated 35 mg of colorless oil (70%). 1H NMR (300 MHz, CDCl3, δ (ppm)) 8.08 (m, 6H, Ar-H),
7.58 (m, 3H, Ar-H), 7.45 (m, 6H, Ar-H), 5.63 (d, J = 4.5, 1H, H-3), 5.54 (s, 1H, H-2), 5.38 (m, 1H, H-5), 5.20
64 (s, 1H, H-1), 4.70 (m, 2H, H-6a, H-6b), 4.43 (ddd, J = 27.0, 4.7, 2.1), 3.48 (s, 3H, OCH3);
19F NMR (280 MHz,
CDCl3, δ (ppm)) -208.0 (dq, J = 48.6, 27.0); 13C NMR (75 MHz, CDCl3, δ (ppm))166.42, 166.22, 165.70,
139.90, 133.83, 133.77, 133.48, 130.40, 130.19, 130.01, 129.79, 129.31, 129.17, 128.73, 128.64, 107.40,
90.59, 88.17, 82.65, 82.41, 81.34, 77.89, 77.81, 64.30, 63.98, 55.35; MS (EMM)+: m/z: 531.1399
([M+Na]+, calcd. 531.1426)
1-O-acetyl-2,3,6-tri-O-benzoyl-5-deoxy-5-fluoro-D-galactofuranose (2.27): 2,3,6-tri-O-benzoyl-5-deoxy-
5-fluoro-β-D-methylgalactofuranose (84 mg, .165 mmol) was placed in a flame-dried round bottom flask
and treated with a 1.4% v/v solution of concentrated H2SO4 in acetic anhydride (2.0 mL). The reaction
stirred at RT for 1.5 hours. Quenched with 2.0 mL of saturated NaHCO3 solution, diluted with brine and
extracted into DCM 3x. Dried organic layer over MgSO4, filtered and concentrated in vacuo. Azeotroped
remaining acetic acid with toluene 2x. Purified by silica gel chromatography (5% 20% EtOAc in
hexane) to give 81 mg of white crystalline solid (91%). 1H NMR (300 MHz, CDCl3, δ (ppm)) 8.08 (m, 6H,
Ar-H), 7.58 (m, 3H, Ar-H), 7.45 (m, 6H, Ar-H), 6.60 (d, J = 4.5, .2H, H-1α), 6.52 (s, 1H, H-1β), 6.14 (t, J =
6.9, .2H, H-3α), 5.77 (dd, J = 7.4, 4.6, .2H, H-2α), 5.69 (m, 2H, H-2β, H-3β), 5.36 (m, 1H, H-5β), 5.20 (m,
.2H, H-5α), 4.65 (m, 3.4H, H-6aα, H-6-aβ, H-6bα, H-6bβ, H-4β), 4.43 (ddd, J = 26.2, 6.4, 2.8, 1H, H-4α),
2.20 (s, 3H, C(O)CH3 β), 2.09 (s, .6H, C(O)CH3 α); 19F NMR (280 MHz, CDCl3, δ (ppm)) -206.1 (dq, J = 51.6,
21.2, α) -207.8 (dq, J = 48.7, 24.5, β); 13C NMR (75 MHz, CDCl3, δ (ppm)) 169.69, 169.20, 166.34, 166.28,
166.15, 165.99, 165.44, 134.05, 133.96, 133.49, 130.26, 130.16, 130.09, 130.01, 129.81, 129.74, 129.04,
128.96, 128.86, 128.80, 128.64, 99.80, 93.14, 90.49, 90.40, 88.07, 87.96, 85.02, 84.77, 80.73, 80.46,
77.71, 77.63, 75.70, 74.29, 74.22, 64.16, 63.83, 63.58, 21.27, 21.17; MS (EMM)+: m/z: 559.1364
([M+Na]+, calcd. 559.1375)
65
Dibenzyl 2,3,6-tri-O-benzoyl-5-deoxy-5-fluoro-α-D-galactofuranose monophosphate (2.29): 1-O-
acetyl-2,3,6-tri-O-benzoyl-5-deoxy-5-fluoro-D-galactofuranose (22 mg, .041 mmol) was placed in a flame
dried round bottom flask and dissolved in dry DCM (450 μL) under N2 atmosphere. The solution was
chilled to 0 °C and treated with a 33 wt% HBr in AcOH solution (28 μL). After stirring for 20 minutes the
reaction was warmed to RT and stirred for 4 hours. Azeotroped using freshly distilled toluene 3x to
isolate the crude glycosyl bromide as an orange oil. The oil was dissolved in dry toluene (450 μL) under
N2. The solution was added dropwise over 30 minutes to a stirring solution of dibenzyl phosphate (49
mg, .176 mmol) in dry toluene (500 μL). The contents stirred at RT for 2.5 hours. Concentration in vacuo
followed by silica gel chromatography (12% EtOAc in toluene) allowed isolation of 28 mg of the α-
anomer as a slightly cloudy viscous oil (90%). 1H NMR (300 MHz, CDCl3, δ (ppm)) 8.00 (m, 6H, Ar-H), 7.55
(m, 3H, Ar-H), 7.4 – 7.2 (m, 16H, Ar-H), 6.31 (t, J = 4.7, 1H, H-1), 6.14 (t, J = 6.9, 1H, H-3), 5.70 (ddd, J =
7.1, 4.3, 2.2, 1H, H-2), 5.17 (dtd, J = 47.1, 6.4, 3.2, 1H, H-5), 5.07 (AB part of an ABX, J = 6.5, 3.5, 2H,
POCH2Ph), 4.90 (AB part of an ABX, J = 11.6, 6.9, 2H, POCH2Ph), 4.61 (m, 2H, H-6a, H-6b), 4.50 (ddd, J =
24.4, 6.4, 3.5, 1H , H-4); 19F NMR (280 MHz, CDCl3, δ (ppm)) -203.8 (dq, J = 56.9, 28.9); 31P NMR (122
MHz, CDCl3, δ (ppm)) -2.4 (m); 13C NMR (75 MHz, CDCl3, δ (ppm)) 166.13, 165.96, 165.76, 134.00,
133.89, 133.46, 130.26, 130.15, 129.96, 129.60, 128.76, 128.70, 128.63, 128.58, 128.05, 127.85, 101.02,
97.61, 97.54, 90.59, 88.15, 80.86, 80.60, 76.54, 76.44, 73.56, 73.47, 69.56, 69.47, 63.32; MS (EMM)+:
m/z: 777.1898 ([M+Na]+, calcd. 777.1872)
Triethylammonium 2,3,6-tri-O-benzoyl-5-deoxy-5-fluoro-α-D-galactofuranose monophosphate (2.30):
Dibenzyl 2,3,6-tri-O-benzoyl-5-deoxy-5-fluoro-α-D-galactofuranose monophosphate (29 mg, .038 mmol)
66 was dissolved in ethyl acetate (1.0 mL, .05M) and triethylamine (38 μL). 10% Pd on carbon (15 mg) was
added. The stoppered flask was evacuated and back filled with H2 from a balloon. The contents stirred at
STP overnight. The reaction was filtered through celite, washed with ethyl acetate and concentrated in
vacuo. Isolated 14 mg of cloudy, viscous oil (54%).1H NMR (300 MHz, CDCl3, δ (ppm)) 8.1 (d, J = 7.5, 2H,
Ar-H), 7.9 (m, 3H, Ar-H), 7.5 – 7.2 (m, 9H, Ar-H), 6.2 (m, 2H, H-1, H-3), 5.62 (m, 1H, H-2), 5.30 (m, 1H, H-
5), 4.7 (m, 2H, H-6a, H-6b), 4.5 (dt, J = 19.3, 6.2, 1H, H-4), 3.00 (q, J = 7.3, 3.8H, NCH2CH3), 1.2 (t, J = 7.5,
10.7H, NCH2CH3); 19F NMR (280 MHz, CDCl3, δ (ppm)) -199.7 (dq, J = 47.1, 25.1); 31P NMR (122 MHz,
CDCl3, δ (ppm)) 0.0 (d, J = 3.5); 13C NMR (75 MHz, CDCl3, δ (ppm)) 164.83, 164.74, 164.58, 132.26,
132.06, 131.90, 129.14, 128.93, 128.81, 128.58, 128.47, 128.44, 127.92, 127.25, 127.1795.17, 95.11,
91.02, 78.54, 78.24, 75.80, 75.70, 73.09, 72.98, 62.71, 62.43, 44.38, 7.37; MS (EMM)-: m/z: 573.0966
([M]-, calcd. 573.0967)
Tributylammonium 5-deoxy-5-fluoro-α-D-galactofuranose monophosphate (2.31): Triethylammonium
2,3,6-tri-O-benzoyl-5-deoxy-5-fluoro-α-D-galactofuranose monophosphate (14 mg, .021 mmol) was
dissolved in a 5:2:1 mixture of methanol:water:triethylamine (1.0 mL) and warmed to 30 °C. The
reaction stirred for 4 days. Concentrated in vacuo then azeotroped 3x with toluene. Purified by anion
exchange chromatography (Biorad AG1-X8 resin, gradient elution 250 mM 500 mM NH4HCO3
solution) using a peristaltic pump. Column fractions were combined, frozen on dry ice and lyophilized.
The residue was treated with ~200 μL Dowex-50WX8-200 resin (pH 7, pretreated with 2.0M HNBu3Cl)
for 4 hours. The resin was removed by filtration through a cotton plug and the filtrate was frozen on dry
ice and lyophilized. Isolated 10 mg of fluffy white solid (82%, 1.5eq. of HNBu3); 1H NMR (300 MHz, D2O,
δ (ppm)) 5.51 (dd, J = 6.2, 4.2, 1H, H-1), 4.69 (dtd, J = ~50, 5.8, 3.9, 1H, H-5), 4.19 (t, J = 7.8, 1H , H-3),
67 4.15 (ddd, J = 8.0, 4.4, 2.0, 1H, H-2), 3.95 (dt, J = 19.5, 6.3, 1H, H-4), 3.84 (m, 2H, H-6a, H-6b), 3.10 (m,
9H, NBu), 1.62 (m, 9H, NBu), 1.33 (sextet, J = 7.3, 10H, NBu), 0.9 (t, J = 7.2, 13H); 19F NMR (280 MHz,
D2O, δ (ppm)) -199.7 (dq, J = 46.4, 23.4); 31P NMR (122 MHz, D2O, δ (ppm)) -0.4 (broad s); 13C NMR (75
MHz, D2O, δ (ppm)) 96.91, 96.85, 95.41, 93.10, 79.82, 79.57, 76.74, 76.69, 73.59, 73.50, 60.71, 60.43,
52.85, 25.36, 19.44, 12.93; MS (EMM)-: m/z: 261.0182 ([M]-, calcd. 261.0181)
5-deoxy-5-fluoro-α-D-galactofuranose uridine diphosphate (2.02 ): A solution of tributylammonium
UMP (5mg, .01 mmol) in anhydrous DMF (150 μL) was stirred with activated 4 Å molecular sieves under
N2 for 1 hour. The molecular sieves were removed and the solution was treated with
diisopropylethylamine (5 μL, .03 mmol) and 1-methyl-3-benzenesulfonylimidazolium triflate (4 mg, 0.01
mmol) and stirred for 1 minute. The solution was added dropwise over 30 seconds to a solution of
tributylammonium 5-deoxy-5-fluoro-α-D-galactofuranose monophosphate (3mg, .007 mmol) and MgCl2
(1 3mg, .01 mmol) in anhydrous DMF (75 μL) at 0 °C. The contents were raised to room temperature and
stirred for 2 hours. The reaction was quenched with 1.0 mL of 100 mM ammonium acetate and washed
1x with 1.0 mL of DCM. The DCM was extracted with 1.0 mL of 100 mM ammonium acetate, the
aqueous layers were combined, frozen on dry ice and stored at -20 °C until purification by HPLC. Purified
by HPLC using a semiprep carbopak dionex column, 300mM tirthylammonium acetate, pH = 6.0-7.0,
flow rate = 1mL/min. Collected peak eluting at 40 – 45 minutes, froze on dry ice and lyophilized to give a
light brown solid. 1H NMR (500 MHz, D2O, δ (ppm)) 7.83 (d, J = 8.0, 1H, H-5u), 5.86 (m, 2H, H-1r, H-6u)
5.55 (t, J = 4.8, 1H, H-1g) 4.56 (m, 1H, H-5g) 4.00 – 4.25 (m, 7H, H-2g, H-2r, H-3g, H-3r, H-4r, H-5ar, H-
5br) 3.89 (dt, J = 18.9, 6.5, 1H, H-4g) 3.71- 3.80 (m, 2H, H-6ag, H-6bg); 19F NMR (280 MHz, D2O, δ (ppm)) -
68 232.0; 13C NMR (125 MHz, D2O, δ (ppm)) 144.25, 105.25, 102.84, 100.32, 90.97, 85.90, 85.83, 79.02,
76.41, 75.63, 72.20, 67.42, 49.28, 10.81; MS (EMM)+: m/z: 591.0406 ([M+Na]+, calcd. 591.0400).
69
Chapter 3
Evaluation of Fluorosugars as Probes of Carbohydrate Polymerase Activity
Portions of this work are published:
Brown, Christopher D., Rusek, Max S. and Kiessling, Laura L. Fluorosugar Chain Termination Agents as Probes of the Sequence-Specificity of a Carbohydrate Polymerase. J. Am. Chem. Soc. 2012, 6552-6555
Contributions
Max Rusek Synthesized disaccharide acceptors
Kenzo Yamatsugu provided all trisaccharide acceptors and the synthetic tetrasaccharide
70 3.1 Introduction
GlfT2, the polymerase that constructs the mycobacterial galactan polymer, has seen an increase
in attention since its discovery over ten years ago. Its intriguing ability to generate a polymer of
appropriate length and alternating Galf-β-(1,5)-Galf-β-(1,6) linkages has caught the interest of many
laboratories. How does a single glycosyltransferase regulate the length, regiochemistry and
stereochemistry of carbohydrate polymer synthesis? Other polymerases such as DNA polymerase or the
ribosome construct their polymers off of a template strand. Glycosyltransferases which polymerize
carbohydrates into polysaccharide chains have no such template. While some studies have shed light
onto length control and other elements of polysaccharide assembly (chapter 1), essentially no work has
been done on understanding pattern control. We intended to use our fluorinated UDP-Galf analogs in an
in vitro assay of GlfT2 previously developed in this laboratory to shed light on this aspect of polymer
biosynthesis.
We have previously published a report that details an in vitro assay of GlfT2 activity using
recombinant His6-GlfT2, synthetic acceptors and UDP-Galf.62 A magnesium-dependent polymerization
takes place when enzyme is incubated with acceptor and donor. The polymeric products are observed
by matrix assisted laser desorption ionization time of flight (MALDI-TOF) mass spectrometry. (Figure
3.01) This previous report provided the first evidence that GlfT2 was an active polymerase intrinsically
capable of generating galactan polymers of physiological length. This activity assay of polymerization
proved tremendously useful for understanding the molecular underpinnings of polysaccharide chain
synthesis in this system. By systematically varying the chemical structure of the acceptor, our lab was
able to deduce some of the requirements for efficient polymerization.
71
Figure 3.01. Incubation of GlfT2 with disaccharide acceptor 3.xx and UDP-Galf generates full length galactan polymers. The crude mixture can be analyzed by MALDI-TOF mass spectrometry.
For template independent polymerases like GlfT2, there is an outstanding question for how the
enzyme regulates the length of polymer chain it produces. By varying the length of lipid aglycone
appended to a Galf disaccharide motif and observing longer polymer chains in the mass spectrum, the
group was able to put forth a new mechanism for length control.62 Specifically, the observation that
longer lipid chains promoted synthesis of longer galactan chains suggested that the acceptors were
binding in a bivalent manner and that this “tethering” of the acceptor controlled the length of the
72 polymer by increasing the chances of rebinding after passive dissociation of the non-reducing
terminus.62 The assay was adapted again using isotopically labeled acceptor to quantify the degree of
processivity using a molecular distraction approach.56 For these reasons, the assay was considered to be
a robust readout of the enzyme’s polymerizing activity and was used as the basis for assessing pattern
control and fidelity.
3.2 GlfT2 operates by a sequence specific mechanism
To determine if fluorinated Galf donors are substrates for GlfT2, the enzyme was incubated with
hexasaccharide acceptor 3.04 and the natural donor UDP-Galf (3.02) and a typical polymer distribution
was observed. (Figure 3.02A) Next, the enzyme was incubated with acceptor 3.04 and 6-fluorinated
donor UDP-6F-Galf 2.01 for 18 hours. The hexasaccharide acceptor contains a non-reducing Galf-β-(1,6)-
Galf motif. Perpetuation of the alternating linkage pattern would begin with the generation of a new β-
(1,5) linkage in which the C5 hydroxyl group of the acceptor attacks the anomeric carbon of the UDP-
sugar. If the electrophile was the 6-fluoro donor, the resulting +1 Galf acceptor heptasaccharide 3.06
would form. This new acceptor is unable to generate a β-(1,6) linkage, however the C5 hydroxyl group
remains available. If the enzyme is able to use the C5 hydroxyl, it would generate a galactan chain with
consecutive β-(1,5) linkages: a mistake. If, however, the enzyme is faithful to the alternating linkage
pattern, chain termination should occur. Analysis of the crude reaction products by MALDI-TOF mass
spectrometry following the 18 hour incubation showed a strong signal for the +1 product 3.06 and no
other significant product peaks. (Figure 3.02B, top)
To determine if chain termination by 6F-Galf was a consequence of GlfT2’s sequence specificity
or just a result of passive dissociation of a fluorinated acceptor, the analogous experiment with GlfT2,
hexasaccharide 3.04 and UDP-5F-Galf (2.02) was performed. Again, the first linkage formed is a-β-(1,5)
73 bond generated by nucleophilic attack of the hexasaccharide C5-OH on the anomeric carbon of UDP-5F-
Galf. This heptasaccharide has a non-reducing β-(1,5) linkage and a C6-OH. Incorporation of a second
fluoro-Galf would generate a β-(1,6) linkage. Only after this +2 product forms would the enzyme
encounter a fluorine atom blocking the position where the next linkage is formed. Analysis of the
Figure 3.02. When GlfT2 is incubated with a hexasaccharide acceptor and UDP-Galf, a normal distribution of polymers is generated (A). When GlfT2, hexasaccharide and UDP-6F-Galf are incubated together, a single fluorinated Galf residue is added and then polymerization terminates abruptly. (B, top) When the same hexasaccharide is acted on in the presence of UDP-5F-Galf, two fluorinated Galf residues are added. (B, bottom) The chain termination occurs prior to generation of a mistaken linkage in both cases.
74 reaction products from the incubation of UDP-5F-Galf showed a signal corresponding to the +2 F-Galf
octasaccharide 3.07. The presence of this peak indicates that GlfT2 is able to recognize and elongate a
fluorinated acceptor, but only to the extent that the alternating linkage pattern is maintained. (Figure
3.02B, bottom)
3.3 Sequence-specific chain termination is general
That GlfT2 so carefully maintains the alternating linkage pattern is remarkable considering it
recognizes and acts on a variety of acceptors including monosaccharide Galf lipids, disaccahrides,
trisaccharides and tetrasaccharides . (Beccas paper) We wondered if this result was general for galactan
acceptor fragments with a different number of starting Galf residues. In other words, was this sequence
specific termination a general observation, or just happenstance for this specific hexasaccharide
acceptor? We performed a similar experiment using fluorinated donors and tetrasaccharide acceptor
3.08. This acceptor, like the hexasaccharide, is elongated with Galf residues by GlfT2. (Figure 3.03A)
Sequence specific chain termination would be expected to follow the same pattern on this shorter
acceptor because it also has a non-reducing β-(1,6) Galf disaccharide motif. Results of MALDI-TOF
analysis for incubation with the 6F and 5F donors again showed +1 and +2 product peaks, respectively.
This suggested that the termination was a mechanism-specific result. (Figure 3.03B)
3.4 A mistake on initiation?
The polymerization assays depicted in Figures 3.02 and 3.03 indicate the enzyme is forced to
chain terminate polymerization when incorporation of a fluorinated Galf blocks the alternating linkage
pattern. In every case, however, a residual peak with mass corresponding to a first mistake appears to
grown in as well (+2 for 6F-Galf, +3 for 5F-Galf). Two explanations are possible to explain these peaks.
75
Figure 3.03. Incubation of GlfT2 with tetrasaccharide and UD-Galf gave the expected polymer distribution (A). Repeating the experiment with UDP-6F-Galf resulted in glycotransfer of one 6F-Galf residue. (B, top) The same tetrasaccharide was elongated by two 5F-Galf residues. (B, bottom) Chain termination appears to be sequence-specific and general.
First, it may represent a small amount of polymer in which a fluorinated Galf was able to add in and
generate a mistake. Alternatively, the tetrasaccharide and hexasaccharide acceptors- both of which are
generated chemoenzymatically- may have a subpopulation with terminal β-(1,5) linkages. This is
possible considering the high promiscuity of the enzyme may pre-dispose it to misread the terminal
linkage and make a mistake upon initiation, and the chemoenzymatic method used to generate the
starting material. A subpopulation with the wrong linkage could be contributing to the residual mistaken
76 peak. To distinguish between these two possibilities, a tetrasaccharide acceptor was accessed
synthetically (courtesy of Kenzo Yamatsugu). This synthetic tetrasaccharide has the same linkage pattern
as the one used in figure 3.03, but because it was not generated chemoenzymatically, there is no
subpopulation of acceptors with the wrong terminal linkage. If the residual peaks were a result of such a
subpopulation, then incubation with the synthetic acceptor would not give a mass spectrum with those
Figure 3.04. Synthetic tetrasaccharide behaves in an identical manner to tetrasaccharide isolated from chemoenzymatic sources. The “doublet” appearance in the 6F-Galf MALDI spectrum is due to a significant amount of M+K product peaks, an occasional occurrence.
77 peaks. If the residual peaks are a result of some intrinsic low-level ability of the enzyme to generate a
galactan fragment with consecutive β-(1,5) or β-(1,6) linkages, then the peaks should still be present in
an experiment with the synthetic tetrasaccharide. Incubation of GlfT2 with this synthetic tetrasaccharide
acceptor and UDP-6F-Galf gave the same product profile: A dominant +1 F-Galf peak and a residual +2
peak. Likewise, the incubation with UDP-5F-Galf produced a dominant +2 peak and a residual +3 peak
(very faint). This finding suggests the enzyme may be able to generate a small but detectable amount of
galactan with consecutive β-(1,6) or β-(1,5) bonds, but cannot sustain polymerization in a meaningful
way. These data support a model in which the alternating linkage pattern is necessary for
polymerization, in which ideal conditions for polymerization are unable to generate even short polymers
with mistaken linkages. (Figure 3.04)
3.5 GlfT2 struggles to identify the terminal linkage of short acceptors
How, from a molecular standpoint, does GlfT2 ensure alternating linkage deposition. Recently, a
crystal structure of GlfT2 was published and the authors put forward a model for bifunctional pattern
control based on acceptor binding.127 The enzyme was crystallized as a tetramer with a molecule of UDP
bound. No evidence was provided for the functional significance of the tetramer, but the monomers do
exhibit a classic GT-A Rossmann-like fold. In the model, an acceptor binds to the enzyme active site
shallowly or deeply, depending on the non-reducing glycosidic bond geometry. This, the authors claim,
aligns the acceptor hydroxyl groups for deprotonation at either C5 or C6, respectively. The model does
not take into account the effect of the lipid on acceptor orientation. This is unusual because reports
have shown that the nature of the lipid has a profound effect on polymer length control.62 If the lipid
had no effect on acceptor orientation, then all acceptors with identical non-reducing glycosidic linkages
should behave the same in our chain termination assay.
78
Figure 3.05. Proposed carbohydrate-directed binding model for bifunctional linkage control. A fluorinated acceptor with fluorine positioned to disrupt the alternating pattern would interact unfavorably with the carboxylate base. The red line defines the latitude where the catalytic carboxylate is located.
The synthesis of disaccharide acceptors bearing either β-(1,6) or β-(1,5) glycosidic bonds was
completed following a published procedure. (Schemes 3.01, 3.02) Briefly, a diisopropylidene
galactopyranose 3.16 was protected as benzyl ether at C6. Reflux of 3.17 in acidic methanol allowed
isolation of a C6-protected furanose scaffold 3.18 which could be ester-protected with benzoyl chloride.
Treatment of 3.19 with tin tetrachloride yielded bicyclic intermediate 3.20 which was opened by
79 acetolysis to generate 1,6-bis acetylated Galf 3.21. This sequence provides a way to distinguish the C5
and C6 positions of a galactofuranoside that is capable at first of participating in a glycosylation reaction
as a donor, then selective unmasking to access an acceptor scaffold. Glycosylation with allyl alcohol gave
differentially protected Galf 3.22 in moderate yield. Deprotection of the C6-acyl group with workup
conditions that allowed benzoyl group migration afforded both the C6-OH (3.23) and C5-OH (3.24)
products. Both migrated and unmigrated products could be used as acceptors in a glycosylation with
thioglycoside 3.25. (Scheme 3.02) The allyl disaccharides were subsequently cross-metathesized with
11-phenoxy-undecene and deprotected to afford GlfT2 acceptors 3.28 and 3.31.
Scheme 3.01. Disaccharide acceptors were assembled using a published procedure. Both regiochemical linkages are accessible using this strategy. Synthesis completed by Max Rusek.
Incubation of GlfT2 with β-(1,6) disaccharide acceptor 3.26 and UDP-Galf produced a typical
distribution of polymer peaks. A similar distribution was seen with β-(1,5) disaccharide acceptor 3.31. To
80 test whether acceptor binding was directed by the non-reducing glycosidic linkage the enzyme was
incubated with 3.28 or 3.31 and UDP-6F-Galf. After 18 hours, a small +1 product peak formed for the β-
(1,6) acceptor, but no product formed for the β-(1,5) acceptor. (Figure 3.06) When the experiment was
repeated with UDP-5F-Galf, no product for either acceptor was observed. (Figure 3.06) Apparently, the
combination of short acceptors and fluorinated donors is too high a kinetic barrier to efficient
glycosylation. (Figure 3.06) These results suggest the carbohydrate-directing model for pattern control
Scheme 3.02. Silver triflate-mediated glycosylations to generate β-(1,5) and β-(1,6) linked Galf disaccharides. Synthesis completed by Max Rusek.
81 should be amended: An acceptor binds in a lipid-directed manner such that the active site aspartate is
poised to deprotonate the acceptor hydroxyl group to perpetuate the alternating linkage pattern
provided the carbohydrate primer is long enough.
Figure 3.06. Disaccharide acceptors are poor substrates for glycotransfer with fluorinated donors.
82
We then looked to trisaccahrides 3.32 and 3.33 (synthesized by Kenzo Yamatsugu). These longer
acceptors, like their cognate disaccharides, are elongated by GlfT2 (unpublished results). This is
unsurprising since 3.32 is the +1 Galf product of 3.31 and 3.33 the +1 Galf product of 3.28. If the enzyme
binds acceptors by recognizing the non-reducing glycosidic linkage and positioning it appropriately for
subsequent nucleophilic attack, then the fluorinated donors should terminate sequence-specifically.
Incubation of GlfT2 and 3.32 with UDP-6F-Galf gave a +1 product. A +1 product also was generated upon
incubation with UDP-5F-Galf. (Figure 3.07) These results are not indicative of sequence-specific chain
termination. Surprisingly, neither UDP-6F-Galf nor UDP-5F-Galf were capable of adding into the β-(1,5)
trisaccharide at all. (Figure 3.07) This suggests the enzyme does not bind the acceptor through
recognition of the non-reducing glycosidic bond and positioning it for attack. A more favorable model is
one in which the lipid directs binding of the acceptor. Since chain terminators act in a sequence specific
manner for longer acceptors, their deviation from this behavior with short acceptors implies that
initiation is the time when a mistake is most likely, when promiscuity is at its highest. Fidelity in linkage
construction is likely imparted to the system via the processivity of the enzyme. In other words, the
enzyme does not generate alternating linkages simply by binding the acceptor with appropriate pre-
organization for attack, but by remaining bound to the acceptor after the initial glycosylation. This may
partially explain why the enzyme has evolved a processive mechanism for polymerization.
3.6 Fluorinated products are dead-end substrates
Fluorinated carbohydrates have been known to bind with lower affinity than their fully
hydroxylated counterparts, and it is possible that the reaction between a fluorinated acceptor and a
fluorinated donor is kinetically disfavored because of this potential loss of affinity. Thus, we wondered if
the fluorinated products could recover from the chain-terminating pattern block if they were exposed to
83
Figure 3.07. Trisaccharide acceptors with β-(1,6) linkages at the reducing end are substrates for fluorinated donors, (top two lines) but chain termination is not sequence-specific. Acceptors with β-(1,5) reducing end stereochemistry are not substrates for fluorinated donors. (bottom two lines) The enzyme is not binding the acceptors at their non-reducing end.
fresh enzyme and UDP-Galf. Solutions of the fluorinated acceptors generated in the polymerization with
hexasaccharide acceptor were treated with fresh GlfT2 and UDP-Galf. After incubation for 18 hours, the
84 contents were analyzed by MALDI. Although the typical lipid-less galactan series did arise, no additional
Galf residues were added to the fluorinated acceptors. This result may indicate that chain terminators
can be used as inhibitors. (Figure 3.08)
Figure 3.08. Isolated Fluoro-Galf capped acceptors do not elongate when incubated with GlfT2 and very high UDP-Galf concentrations. (red peaks) A byproduct polymer series of lipid-less non-fluorinated galactan forms. (black) Fluorinated acceptors appear to be dead end substrates.
Inhibition of galactan polymerization is a long term goal of researchers in the field. A chain
terminating UDP-Galf analog would potentially have mixed modes of substrate inhibition (both acceptor
and donor inhibition). These studies using fluorinated donors have created new questions concerning
the properties of an ideal chain termination agent and whether a chain termination approach would be
effective at inhibiting polymerization. These areas of study are an exciting new direction for the field and
are addressed in the next chapter.
We have devised a strategy for assessing sequence specificity for a carbohydrate polymerase
using fluorosugars as chain termination agents, expanding the chemical glycobiologist’s toolbox. This
85 approach invokes the work of Frederick Sanger, in which chain-terminating nucleosides were used to
rapidly and accurately sequence nucleic acid polymers, and reimagines the concept to gain insight into
another class of biopolymer: the polysaccharide.128 In polymerizations such as DNA replication,
sequence specificity is determined by a template. In contrast, polysaccharide polymerization has no
obvious mechanism for ensuring sequence fidelity. Synthetic probes which challenge a bifunctional
carbohydrate polymerase’s patterning specificity were designed (chapter 2). Both probes exhibited
chain termination capability of GlfT2-catalyzed galactan polymerization. Moreover, the chain
termination occurred only when GlfT2 was challenged to deviate from the normal alternating pattern,
suggesting that polymerization occurs with sequence specific dependence. We anticipate that this
method can be generalized to study repeating motif specificity in other carbohydrate polymerases.
3.7 Experimental
General procedure for GlfT2 activity assay
Polymerization reactions consisted of 10 μL total volume containing final concentrations of 0.2 μM His6-
GlfT2, approximately 200 μM acceptor, 1.0 mM UDP donor sugar in 50 mM Hepes, pH 7.0, 25 mM
MgCl2, and 100 mM NaCl. Reactions were incubated at room temperature for 18 h, then quenched with
30 μL of a 1:1 mixture of CHCl3/MeOH. Quenched reaction mixtures were evaporated to dryness under
vacuum in a SpeedVac SC100 (Varian) then resuspended in 10 μL of 50% MeCN in MQ water for MALDI
MS analysis. Samples for MALDI MS analysis were spotted as a 1:3 mixture with α-cyano-4-
hydroxycinnamic acid matrix and spectra were recorded in positive linear mode using a Bruker Ultraflex
III mass spectrometer.
86 Isolation of tetrasaccharide and hexasaccharide acceptors
A polymerization reaction was run in the presence of GlfT2, β-(1,6) disaccharide and UDP-Galf. The
reaction was quenched after 45 minutes and suspended in 50% ACN/H2O. The solution was filtered
through a milipore syringe filter and purified by RP-HPLC (10% - 80% ACN in 0.1% AcOH in H2O).
Monitored absorbance at 195nm and collected peaks at retention times 18 – 25. Analysis by MALDI-TOF
identified pure tetrasaccharide and hexasaccharide fractions. Used in polymerizations assays without
further purification.
Synthesis of (1,6) and (1,5)-disaccharide acceptors
2,3,5,6-tetra-O-acetyl-β-D-Galfuranosyl-(1,5)-2,3,6-tri-O-benzoyl-β-O-allylgalactofuranose (3.26):
2,3,5,6-tetra-O-acetyl -1-(thioethyl)-β-D-galactofuranose (20 mg, .037 mmol) and allyl 2,3,6-tri-O-
benzoyl-β-D-galactofuranose (15.2 mg, .028 mmol) were dissolved in 1 mL dry DCM, treated with
activated molecular sieves and allowed to stir under N2 for 15 minutes at RT. The solution was chilled to
0 °C and silver triflate (1.4 mg, .006 mmol) and N-iodosuccinimide (8.8 mg, .039 mmol) were added. A
red-brown color developed. After 30 minutes of stirring at 0 °C the reaction was filtered through celite
and washed with DCM until all the pink color was washed into the filtrate. The filtrate was quenched
with 5 mL of Na2S2O3 solution (saturated), 5 mL of saturated NaHCO3 solution, 5 mL of brine. The organic
layer was dried over MgSO4, filtered and concentrated in vacuo. Purification by silica gel
chromatography gave 21 mg of a colorless oil. (70% yield) ); 1H and 13C NMR spectra matched those
previously reported.
2,3,5,6-tetra-O-acetyl-β-D-galactofuranosyl-(1,5)-2,3,6-tri-O-benzoyl-β-D-(12-phenoxy-2-
undecenyl)galactofuranose (3.27): 2,3,5,6-tetra-O-acetyl-β-D-galfuranosyl-(1,5)-2,3,6-tri-O-benzoyl-β-O-
allylgalactofuranose (18 mg, .017 mmol) was dissolved in a solution of 11-phenoxy-undecene (19 mg,
87 .077 mmol) in DCM (200 μL). The solution was placed under N2 atmosphere, treated with Grubbs’ 1st
generation catalyst (1 mg, .0019 mmol) and stirred at RT overnight. Added 1 more mg of catalyst in the
morning then concentrated on rotovap. Purification by silica gel chromatography (5% 10% 20%
EtOAc in hexane) gave 7 mg of cross metathesis product (33% yield). 1H and 13C NMR spectra matched
those previously reported.
β-D-galactofuranosyl-(1,5) -β-D-(12-phenoxy-2-undecenyl)galactofuranose (3.28): 2,3,4,5-tetra-O-
acetyl-β-D-galactofuranosyl-(1,5)-2,3,6-tri-O-benzoyl-β-D-(12-phenoxy-2-undecenyl)galactofuranose (5
mg, .004 mmol) was treated with 500 μL of 0.5M NaOMe in methanol and stirred for 3h at RT under N2.
The solution was neutralized with acidic resin, filtered off through a cotton plug, washed with methanol
and concentrated in vacuo. Purification by silica gel chromatography (10% MeOH in DCm) gave 2.4mg of
colorless film (100% yield). 1H and 13C NMR spectra matched those previously reported.
2,3,5,6-tetra-O-acetyl -1-(thioethyl)-β-D-galactofuranose (3.25): 1-O-acetyl-2,3,5,6-tri-O-benzoyl -D-
galactofuranose (20 mg, ..037 mmol) was dissolved in dry DCM (100 μL) under N2 and chilled to 0 °C.
trimethylsilyl ethyl thioether (9 μL, .056 mmol) was added via syringe. Zinc iodide (6 mg, .018 mmol) was
quickly weighed and added to the reaction. The reaction was warmed to RT and stirred 2h. TEA (2 drops)
was added to quench the reaction. A precipitate was removed by filtering through a cotton plug,
washing with DCM and the filtrate was concentrated in vacuo. Purification by silica gel chromatography
(5% 10% EtOAc in hexane) gave an oily white residue, 15 mg (75% yield); 1H and 13C NMR spectra
matched those previously reported.
2,3,5,6-tetra-O-acetyl-β-D-Galfuranosyl-(1,6)-2,3,5-tri-O-benzoyl-β-O-allylgalactofuranose (3.29):
2,3,5,6-tetra-O-acetyl -1-(thioethyl)-β-D-galactofuranose (15 mg, .027 mmol) and allyl 2,3,5-tri-O-
benzoyl-β-D-galactofuranose (11 mg, .021 mmol) were dissolved in 1 mL dry DCM (1.0 mL), treated with
88 activated molecular sieves and allowed to stir under N2 for 15 minutes at RT. The solution was chilled to
0 °C and silver triflate (1.3 mg, .005 mmol) and N-iodosuccinimide (7 mg, .029 mmol) were added. A red-
brown color developed. After 30 minutes of stirring at 0 °C the reaction was filtered through celite and
washed with DCM until all the pink color was washed into the filtrate. The filtrate was quenched with 5
mL of Na2S2O3 solution (saturated), 5 mL of saturated NaHCO3 solution, 5 mL of brine. The organic layer
was dried over MgSO4, filtered and concentrated in vacuo. Purification by silica gel chromatography
gave 10 mg of a colorless oil. (48% yield) ); 1H and 13C NMR spectra matched those previously reported.
2,3,5,6-tetra-O-acetyl-β-D-galactofuranosyl-(1,6)-2,3,5-tri-O-benzoyl-β-D-(12-phenoxy-2-
undecenyl)galactofuranose (3.30): 2,3,5,6-tetra-O-acetyl-β-D-Galfuranosyl-(1,6)-2,3,5-tri-O-benzoyl-β-
O-allylgalactofuranose (10 mg, .010 mmol) was dissolved in a solution of 11-phenoxy-undecene (11 mg,
.045 mmol) in DCM (100 μL). The solution was placed under N2 atmosphere, treated with Grubbs’ 1st
generation catalyst (1 mg, .001 mmol) and stirred at RT overnight. Added 1 more mg of catalyst in the
morning then concentrated on rotovap. Purification by silica gel chromatography (5% 10% 20%
EtOAc in hexane) gave 5.8 mg of cross metathesis product (40% yield). 1H and 13C NMR spectra matched
those previously reported.
β-D-galactofuranosyl-(1,6) -β-D-(12-phenoxy-2-undecenyl)galactofuranose (3.31): 2,3,5,6-tetra-O-
acetyl-β-D-galactofuranosyl-(1,6)-2,3,5-tri-O-benzoyl-β-D-(12-phenoxy-2-undecenyl)galactofuranose (5.8
mg, .004 mmol) was treated with 600 μL of 0.5M NaOMe in methanol and stirred for 3h at RT under N2.
The solution was neutralized with acidic resin, filtered off through a cotton plug, washed with methanol
and concentrated in vacuo. Purification by silica gel chromatography (10% MeOH in DCm) gave 2.2 mg of
colorless film (92% yield). 1H and 13C NMR spectra matched those previously reported.
89
Chapter 4
Expanding the Utility of Chain-Terminating Glycosides
Portions of this work were published:
May, J.F., Levengood, M. R., Splain, R.A., Brown, C.D., Kiessling, L.L. A processive carbohydrate polymerase that mediates bifunctional catalysis using a single active site, Biochemisry, 2012, 1148-1159
Contributions
Max Rusek and Kenzo Yamatsugu aided in the preparation of 5-deoxy UDP-Galf
John May and Rebecca Splain constructed the mutant GlfT2 plasmids
90 4.1 Introduction
Reports of chain terminating glycosyl donors are appearing with rising frequency in the
literature. For example, the polysialyltransferase from E.coli K12 is a carbohydrate polymerase that
generates a capsular polysaccharide polysialic acid with alternating α-(2,8) and α-(2,9) alternating
linkages.10,96 A 9-azido analog of the natural donor, CMP-sialic acid, was synthesized and found to
incorporate into acceptor strands in a chain-terminating manner.96 Deoxy Galf donors have been
reported to chain terminate the galactan. In some cases, a fluorinated donor sugar was thought to act as
a chain termination agent in cell-based assays, but later discovered to act as a competitive inhibitor of
components of the Leloir pathway instead.129,130
Figure 4.01. Chain terminator donor sugar panel.
We realized there was a need to comprehensively address the nature of chain terminating
glycosyl donors. For GlfT2 in particular, understanding the influence that a bond-blocking functional
group has on the nascent chain is important. The physical properties that led us to design and synthesize
fluorinated probes would be different in many respects from analogous deoxy or azido probes. For
instance, the inductive effects on vicinal hydroxyl pKa would vary considerably from a polarized C-F
bond to a non-polar C-H or C-N3 bond. This pKa is relevant for Galf transfer as GlfT2 operates via a
general base mechanism that deprotonates either the C5 or C6 hydroxyl group. This in turn may affect
91 the behavior of the chain terminator, and would be an important consideration depending on the
potential use of the compound as either an inhibitor or as a sequence fidelity probe. (Figure 4.01)
Sequence fidelity is discussed at length in chapter 3. An inhibitor could potentially be useful in several
ways. Chain terminators such as AZT, which disrupts the activity of the HIV enzyme reverse transcriptase
are used directly as antibiotics.131 Alternatively, inhibitors can be used as probes in enzyme kinetics
investigations to yield information such as substrate binding order. More generally, it would be
important for research on the mycobacterial cell wall to know what sort of chemical probes can be
incorporated into various structures, as this would be useful for cell-based studies as a potential imaging
tool. Thus, a broad comparison of fluorinated, deoxygenated and azido Galf donor would be beneficial
to a wide range of future study.
In order to compare the fluorinated, deoxygenated and azido donor behavior as patterning
fidelity probes and inhibitors we sought to synthesize the panel of inhibitors from Figure 4.01. We
returned to the vinyl Galf intermediate and found that beta vinyl methyl-Galf 2.12 could be modified
under standard hydroboration conditions to yield the 5-deoxy intermediate 4.04. It was possible to
access the 6-deoxy scaffold be treatment of resolved epoxide 2.15 with sodium borohydride, but it was
easier to access it directly from fucose (4.11). 6-azido Galf can be synthesized from epoxide 2.15
(deprotected). Longer reaction times and lower temperatures improved the regioselectivity of the ring
opening. Cognizant that the azide functionality reduces under standard hydrogenation/hydrogenolysis
conditions, we modified the stereoselective phosphorylation from dibenzyl to diallyl phosphate.
Treatment of protected 6-N3-Galf phosphate with palladium tetrakis triphenylphosphine should remove
the allyl protecting groups easily. These approaches highlight the versatility of the vinyl group to
carbohydrate mimic synthesis. (Scheme 4.01)
92
Scheme 4.01. Synthesis of deoxygenated and azido UDP-Galf analogs.
93 4.2 A comparison of fluorinated and deoxygenated Galf donors.
4.2.1 As sequence specificity probes
We first wanted to investigate whether or not fluorinated and deoxygenated donors were
equally well equipped to report on sequence fidelity for GlfT2. As discussed in chapter 3, the
tetrasaccharide acceptor 3.12 was incubated with GlfT2 and the fluorinated donors (2.01 or 2.02) to give
the characteristic +1/+2 sequence-specific chain termination result. When the 6-deoxy donor was
incubated with the same tetrasaccharide acceptor and GlfT2, a strong signal corresponding to the +1
deoxy-Galf intermediate was observed. This result is in agreement with sequence-specific chain
termination observed with UDP-6F-Galf. To confirm the chain termination is a result of pattern blocking
rather than passive diffusion of the resulting acceptor, a similar experiment with UDP-5H-Galf is
planned. Reports in the literature suggest that 5-deoxy compounds add a single 5H-Galf residue to
trisaccharides. This result is similar for our trisaccharide study, but trisaccharides are difficult acceptor
substrates to interpret: they behave idiosyncratically, potentially because they are too short to fill the
enzyme subsites. (See discussion chapter 3, Table 4.01)
Table 4.01. Summary of chain termination results for various acceptors with fluorinated and deoxygenated donors. *Results reported in Reference68
94 4.2.2 As inhibitors
We assessed the ability of our chain terminators to inhibit galactan formation in two ways. First,
we ran polymerization assays in which our fluorinated or deoxy donors were forced to compete for
incorporation with the natural donor, UDP-Galf. These polymerizations were run under the same
conditions as the single-donor experiments from chapter 3. A chain terminator with strong inhibition
would be expected to incorporate into a polymerizing chain rapidly, impairing polymerization
completely, or nearly so. Evaluation of such chain terminated products by MALDI-TOF would be
expected to reveal abruptly terminated galactan oligomers. Alternatively, a chain terminator that was
unable to effectively compete with UDP-Galf to inhibit polymerization would reveal a normal polymer
Figure 4.02. When GlfT2 is incubated with tetrasaccharide, UDP-Galf and UDP-6F-Galf, the polymerization is not inhibited, even at 10:1 fluorinated donor concentrations. Close inspection shows fluorinated Galf residues are able to incorporate into the growing chains. (top) The same experiment with UDP-5F-Galf again fails to inhibit polymerization. No fluorinated Galf incorporation was observed. (bottom)
95 distribution in our assay. The results of the competition experiments between UDP-Galf and the
fluorinated donors suggested they were weak inhibitors of polymerization. Interestingly, the 6F donor
was able to incorporate into the galactan chains, but not inhibit polymerization completely even at 10
fold excess relative to UDP-Galf. The 5F-Galf donor was unable to compete effectively even at 10-fold
excess and no incorporation was observed. (Figure 4.02)
The second method we used to demonstrate inhibition involved monitoring the rate of UDP
production in a continuous coupled assay our lab has used earlier.(May) The fluorinated donor must
bind the same active site as UDP-Galf and thus, it is likely that there is some competition for biding
between the two. Likewise, a dead-end fluorinated acceptor would compete with non-fluorinated
acceptors as well. The combined effect should be a reduction in the initial rate of turnover by GlfT2. The
coupled assay allows us to monitor that turnover and observe whether or not a reduction in
polymerization is occurring in a more quantitative sense. In this assay UDP is consumed by the enzyme
pyruvate kinase to generate the phosphopyruvate substrate for lactate dehydrogenase. In the presence
of NADH, this relay of enzymatic transformations can be observed by measuring the absorbance of
NADH over time. Absorbance is directly related to UDP consumption and so measurement of UDP can
serve as readout of enzyme activity.
GlfT2 was incubated with tetrasaccharide acceptor 3.12, UDP-Galf and the cocktail of coupled
assay enzymes and substrates. To observe inhibition, NADH absorbance was followed over time at UDP-
6F-Galf donor concentrations of 0μM, 500μM and 5 mM, which corresponded to a 0, 1, 10-fold ratio
relative to the natural donor. The rate of UDP production was extrapolated from the absorbance
change. This measurement serves as a crude readout of enzyme activity. Of course, one problem with
monitoring inhibition in this way is that UDP-6F-Galf is also a substrate for the enzyme, so the UDP
96
Figure 4.03. GlfT2-catalyzed polymerization of tetrasaccharide in the presence of the inhibitor UDP-6F-Galf. The rate of reaction decreases as the ratio of UDP-6F-Galf to UDP-Galf increases.
produced from 6F-Galf incorporation is also measured. (Figure 4.04) We observed a reduction in UDP-
production that was dose-dependent. There are potentially multiple modes of inhibition operating in
97 this system as the fluorodonor probably acts as a competitive inhibitor, a chain terminator and the
products of chain termination themselves act as competitive inhibitors of the acceptor. This mixed
activity makes it difficult to measure specific kinetic parameters such as Ki. However the reduced
turnover does suggest a degree of inhibition is taking place. A comparison to the remaining panel of
donors is planned.
4.3 Fluorinated Galf can be used to generate fluorinated acceptors
All glycosyltransferases must bring together a donor and acceptor in a ternary complex. The
mechanism of substrate binding typically involves one of three pathways: ping pong, sequential ordered
and random ordered. (Figure 1.07) Binding order is an important consideration for researchers
interested in growing x-ray crystallography quality protein crystals because it determines the order of
substrate soaking that should be trialed. High throughput screens also must consider binding order
when they are designed or the screen risks identifying hits that are inactive in a biologically relevant
context. Classic enzymology experiments can be used to distinguish between these mechanisms,
provided a competitive inhibitor is available.
Fluorinated acceptors have been used as competitive inhibitors to determine binding order for a
glycosyltransferase before. For example, to elucidate the binding order for the glycosyltransferase
MurG,50 the Walker Group synthesized a fluorinated lipid II analog that was found to competitively
inhibit the natural substrate. Another substrate analog-based approach was used to study the binding
order for the polymerase PglH.51 In that system product inhibition was found to regulate chain length.
Because the product was acting as a competitive inhibitor it was used to determine the binding order of
the polymerase. We recognized that the fluoro-capped dead-end acceptors generated in our chain
termination assays might be competitive inhibitors so we devised a synthesis of a fluorinated
98 disaccharide based on our acceptor synthesis. The 1-O-acetyl-6F-Galf intermediate 2.21 could readily be
converted to the thioethyl glycoside donor 4.22. Similarly, the 1-O-acetyl-5F-Galf intermediate 2.27 was
converted to thioglycoside 4.26. (Scheme 4.02) These thioglycosides could be used in glycosylation
reactions analogous to the ones used to generate disaccharide acceptors 3.28 and 3.31 Cross metathesis
and deprotection afforded 5F-Galf-β-(1,6)-Galf and 6F-Galf-β-(1,5)-Galf disaccharides 4.25 and 4.29,
respectively.
Scheme 4.02. Synthesis of fluorinated Galf disaccharides.
99
These analogs were designed so that the fluorine atoms were positioned to interrupt the
alternating linkages. STD NMR studies on discaccharides and trisaccharides suggest that the non-
reducing terminus has the fewest contacts with the protein, making substitution at this end the least
perturbing.105,132 The cognate non-fluorinated substrates 3.28 and 3.31 were previously found to be
elongated. In order to use the fluorinated derivative as inhibitors, we needed to confirm that these
fluorinated acceptors were incompetent for elongation by the polymerase. Incubation of GlfT2 with
fluorinated acceptor 4.25 and UDP-Galf in our standard polymerization assay surprisingly yielded
polymeric products. (Figure 4.04) Because the fluorine atom was positioned to block elongation at this
position, the only way polymeric products could form was if the enzyme initiated polymerization with a
mistaken consecutive β-(1,6) linkage. This result confirms that the enzyme is capable of making a
mistake on initiation and recovering from such an event. Additionally, the data also support the
assertion that the enzyme controls the alternating linkages during processive elongation and not via
differential binding on initiation (vida supra). When the polymerization experiment was repeated with
5F acceptor 4.29, no polymerization was observed. Studies to further assess 4.29 as a potential
competitive inhibitor are planned. To determine if the fluorinated acceptor was acting as a competitive
inhibitor, our coupled assay will be used under initial rate conditions at various concentrations of
fluorinated acceptor. If the compound is acting as a competitive inhibitor then plotting the double
reciprocal graph (1/[S] vs 1/v) for each inhibitor concentration should yield a series of lines that
converge at the Y-axis. Once a competitive inhibitor is identified, the kinetic analysis to determine
substrate binding mechanism is straightforward.
100
Figure 4.04. Assessing polymerization competency with fluorinated acceptors.
4.4 One active site and 2 linkages
One longstanding question concerning GlfT2’s bifuncionality is whether the enzyme utilizes one
active site for both linkages, or two distinct sites, one for each linkage. Inverting glycosyltransferases
typically use a so-called “DXD” motif in their active sites. This motif acts as a general base to
deprotonate the acceptor hydroxyl proton rendering it nucleophilic for attack on the donor. In
particular, GlfT2 has two such DXD motifs. Homology modeling to a related glycosyltransferase, SpsA,
indicated that 256DXD258 was positioned to coordinate the magnesium ion and bind UDP-Galf whereas
371DDA373 was positioned to act as the base.
We reasoned that mutation of this catalytic base would help us discern the number of active
sites. If Glft2 had a single active site, mutagenesis should abolish all activity. If GlfT2 possesses two
active sites, one for each linkage, then mutagenesis should abolish one, but not the other bond-forming
activity. By incubating mutant enzymes with acceptor substrates and observing the product profile, we
101 would be able to distinguish between these possibilities. Thus, alanine and glutamate mutants were
constructed and expressed in the same E. coli expression system used for the wild type enzyme.
Incubation of D371A, D371E, D372A or D372E with tetrasaccharide acceptors either abolished
polymerization entirely (D371A, D372A, D372E), or was permissive (D371E) to polymerization. These
results strongly indicate that the enzyme has a single active site. (Figure 4.05)
Figure 4.05. Mutant GlfT2 polymerization assays. Only D371E was active. These results suggest GlfT2 uses a single active site for bifunctional catalysis.
Recently, a crystal structure of GlfT2 was published.127 Satisfyingly, the enzyme does indeed
possess a single active site and D371 appears to be positioned to act as catalytic base. Interestingly, the
enzyme crystalized as a tetramer with C4 symmetry, forming a central pore. The authors were unable to
102 demonstrate that the tetramer was the functionally active state of the enzyme. The report also suggests
that bifunctionality arises from how far the acceptor extends into the binding pocket. Since the D371E
mutant is active, but contains an additional CH2 group, the “ruler” would be altered. We wondered if the
change in length would have an effect on the enzyme’s linkage fidelity so we incubated D371E with
tetrasaccharide and the fluorinated Galf donors. The same chain termination pattern resulted: 6F-Galf
added once and 5F-Galf, twice. This may not be consistent with the bifunctional ruler model. (Figure
4.06) Modeling of this change using the crystal structure coordinates is planned once the structure is
deposited in the PDB.
Figure 4.06. GlfT2 D371E maintains high sequence fidelity. The altered length arising from an aspartate to glutamate mutation appears to have little effect on patterning regulation.
4.5 Experimentals
Synthesis of UDP-5-deoxy-Galf (UDP-5H-Galf)
2,3-di-O-benzyl-5-deoxy-β-D-methylgalactofuranose (4.04): 2,3-di-O-benzyl-5-vinyl-β-D-
methylgalactofuranose (122mg) was dissolved in anhydrous THF (2.44mL) under N2. A solution of
103 borane in THF (1.0M, 1.1 mL) was added dropwise and the contents stirred overnight. The contents
were heated to 50 °C then 367μL of 3.0M NaOH solution and a solution of peroxide (30 wt % in H2O)
were aded. Waited 5 minutes, then stirred at RT for 45 minutes. The product was extracted 2x into
DCM, dried over MgSO4, filtered and concentrated in vacuo. Purification by silica gel chromatography
(10% 20% 50% EtOAc in hexane) yielded 35mg of product (27% yield). 1H NMR (300 MHz, CDCl3, δ
(ppm)); ); 13C NMR (75 MHz, CDCl3, δ (ppm))
5-deoxy-β-D-methylgalactopyranose (4.05): 2,3-di-O-benzyl-5-deoxy-β-D-methylgalactofuranose
(49mg, .137 mmol) was dissolved in reagent grade methanol (1,4 mL) and treated with Pd/C catalyst
(25mg, 50% by wt). The atmosphere was evacuated and back filled with H2 from a balloon 3x and stirred
for 2 hours. The catalyst was removed by filtering through celite and washing with methanol. The filtrate
was concentrated in vacuo to give a colorless oil, 18mg (74% yield). 1H NMR (300 MHz, CDCl3, δ (ppm)) ;
13C NMR (75 MHz, CDCl3, δ (ppm))
2,3,6-tri-O-benzoyl-5-deoxy-β-D-methylgalactofuranose (4.06): 5-deoxy-β-D-methylgalactopyranose
(18mg, .101 mmol) was dissolved in anhydrous pyridine (1.0 mL) under N2 atmosphere and treated
dropwise with benzoyl chloride (41 μL, .354 mmol) and stirred at RT for 2h. The reaction stalled so
another 3.5 equivalents of benzoyl chloride were added and the reaction was complete instantaneously.
Azeotroped 2x with toluene then purified by silica chromatography (5% 10% 20% EtOAc in
hexane). Isolated 50mg of colorless oil (100% yield). 1H NMR (300 MHz, CDCl3, δ (ppm)); 13C NMR (75
MHz, CDCl3, δ (ppm))
1-O-acetyl-2,3,6-tri-O-benzoyl-5-deoxy-β-D-methylgalactofuranose (4.07): 2,3,6-tri-O-benzoyl-5-deoxy-
β-D-methylgalactofuranose was dissolved in a 1.4% v/v solution of H2SO4 in Ac2O and stirred for 2h at
RT. The reaction was quenched with a 1:1 mixture of saturated NaHCO3 and brine, extracted 3x into
104 DCM, dried over MgSO4, filtered and concentrated in vacuo. Residual acetic acid was removed by
azeotrope with toluene. Purification by silica chromatography (15% 10% 20% EtOAc in hexane)
gave a colorless oil, 54mg (57% yield). 1H NMR (300 MHz, CDCl3, δ (ppm)); 13C NMR (75 MHz, CDCl3, δ
(ppm))
Dibenzyl 2,3,6-tri-O-benzoyl-5-deoxy -α-D-galactofuranose phosphate (4.09): 1-O-acetyl-2,3,6-tri-O-
benzoyl-5-deoxy-β-D-methylgalactofuranose (26mg, .05 mmol) was dissolved in dry DCM (500μL) under
N2 and chilled to 0 °C. A solution of HBr in AcOH (33 wt.%, 33 μL, .125 mmol) was added dropwise. After
30 minutes at 0 °C, the reaction was warmed to RT and stirred for 2 hours. The contents were
azeotroped 3x using anhydrous toluene. Separately, a solution of dibenzyl phosphate (56mg, .2mmol) in
dry toluene (500 μL) and TEA (28μL, .2 mmol) was prepared under N2. The crude oil of Galf bromide was
dissolved in dry toluene and added dropwise to the stirring solution of dibenzyl phosphate. The contents
were stirred at RT and monitored by TLC. Complete at 1h, concentrated in vacuo. Purification by silica
chromatography (10% EtOAc in toluene) gave 24mg of pure product as a cloudy oil (66% yield, 2 steps).
1H NMR (300 MHz, CDCl3, δ (ppm)); 13C NMR (75 MHz, CDCl3, δ (ppm)); 31P NMR (122 MHz, CDCl3, δ
(ppm))
Triethylammonium 2,3,6-tri-O-benzoyl-5-deoxy-α-D-galactofuranose phosphate(4. 9 ¾): Dibenzyl
2,3,6-tri-O-benzoyl-5-deoxy-α-D-galactofuranose phosphate (20mg, .027 mmol) was dissolved in reagent
grade ethyl acetate (300 μL) and TEA (30 μL) and treated with Pd/C catalyst (10% by weight, 10mg). The
flask was evacuated under vacuum and backfilled with H2 from a balloon. The reaction stirred at RT
overnight. The contents were filtered through celite, washed with ethyl acetate and concentrated in
vacuo to give a white residue, 11mg (65% yield). 1H NMR (300 MHz, CDCl3, δ (ppm)); 13C NMR (75 MHz,
CDCl3, δ (ppm)); 31P NMR (122 MHz, CDCl3, δ (ppm))
105 Tributylammonium 5-deoxy-α-D-galactofuranose phosphate (4.10): Triethylammonium 2,3,6-tri-O-
benzoyl-5-deoxy-α-D-galactofuranose phosphate (11mg, .014 mmol) was dissolved in a 5:2:1 smixture of
MeOH:H2O:TEA, raised to 30 °C and stirred for 7d. Concentrated on rotovap to remove solvents and
purified by anion exchange chromatography (Biorad AG1-X8 resin, gradient elution 250 mM 500 mM
NH4HCO3 solution) using a peristaltic pump. Column fractions were combined, frozen on dry ice and
lyophilized. The residue was treated with ~200 μL Dowex-50WX8-200 resin (pH 7, pretreated with 2.0M
HNBu3Cl) for 4 hours. The resin was removed by filtration through a cotton plug and the filtrate was
frozen on dry ice and lyophilized. Isolated 5mg of a white solid (72%, 2 steps); 1H NMR (300 MHz, CDCl3,
δ (ppm)); 13C NMR (75 MHz, CDCl3, δ (ppm)); 31P NMR (122 MHz, CDCl3, δ (ppm))
UDP-5-deoxy-Galf (4.02): Trifluoroacetic anhydride was distilled from P2O5 (BP = 36 °C). UMP
tributylammonium was placed in an oven-dried pear flask with stir bar and placed under N2 and chilled
to 0°C. ACN (1.2mL), N,N-dimethylanaline (183 μL) and TEA (50 μL) were added. A solution of
trifluoroacetic anhydride (254 μL) in CAN (450 μL) was prepared at 0 °C under N2 then cannulated to the
flask with UMP, which turned yellow. After 15 minutes, the excess TFAA was evacuated using a needle
equipped to a vacuum line with N2 backfilling from a balloon. A solution of 1-methylimidazole (86 μL) in
ACN (900 μL) and TEA (250 μL) was prepared under N2 and at 0 °C. This solution was cannulated to the
activated UMP solution and stirred at 0 °C for 30 minutes. The tributylammonium 5-deoxy-Galf
monophosphate (5mg, .008 mmol) was suspended in ACN (100 μL). Activated molecular sieves were
added. The UMP-imidazolium solution (200 μL) was added dropwise over 30 minutes. The reaction
stirred cold for 1.5h then at RT for 3.5h. The coupling was quenched with 500 μL of 100mM NH4OAc
solution, washed with 500 μL DCM, back extracted into another 500μL 100mM NH4OAc, the aqueous
layers were combined, frozen on dry ice and stored at -20 °C until purification by HPLC. HPLC purification
using a Dionex carbopak column, 300mM NH4OAc, 100μL injections gave 2.7mg of product (63% yield).
106 1H NMR (300 MHz, CDCl3, δ (ppm)); 13C NMR (75 MHz, CDCl3, δ (ppm)); 31P NMR (122 MHz, CDCl3, δ
(ppm))
D-methylfucofuranose: Fucose (1g, 6.09 mmol) was placed in a 100 mL flame-dried RBF and suspended
in dry methanol (12.5 mL) and chilled to 0 °C under N2 atmosphere. A solution of HCl in methanol was
prepared by adding acetyl chloride (348μL, 4.87 mmol) to a second flask of cold methanol (12.5mL). A
slow cannulation under positive N2 pressure was performed to transfer the HCl to the solution of sugar.
By the end of the transfer the reaction had become homogenous. Stirred at RT overnight. Added 500mg
of NaHCO3(s) and let stir 5 minutes. Some bubbles formed. Concentrated in vacuo and used immediately
in the next reaction without purification.
2,3,5-tri-O-benzoyl-D-methylfucofuranose: Methylfucofuranose (~1.1g, 6.09 mmol) was dissolved in dry
pyridine (30mL) and chilled to 0 °C under N2. Benzoyl chloride (2.7mL, 23.5 mmol) was added dropwise
and the reaction stirred at RT overnight. Warmed to 30 °C, added 20 μL of benzoyl chloride and stirred
overnight again. Filtered through fluted filter paper, azeotroped filtrate with toluene until all pyridine
was removed. Purified by silica gel chromatography (100% DCM) gave 239mg of product as a mix of
anomers (10% yield). 1H NMR (300 MHz, CDCl3, δ (ppm)); 13C NMR (75 MHz, CDCl3, δ (ppm));
1-O-acetyl-2,3,5-tri-O-benzoyl-D-fucofuranose (4.12): Tri-O-benzoyl methylfucofuranose (239mg, .487
mmol) was dissolved in a 1.4% v/v H2SO4 in Ac2O (4.9mL) and stirred at RT for 1h. The reaction was
chilled to 0 °C, quenched with a 1:1 saturated NaHCO3:brine solution, extracted 3x into DCM, dried over
MgSO4, filtered and concentrated in vacuo. Remaining AcOH was removed by azeotrope with toluene.
Purified by silica gel chromatography (5% 10% 20% EtOAc in hexane) to give 151mg of white solid
(60% yield); 1H NMR (300 MHz, CDCl3, δ (ppm)); 13C NMR (75 MHz, CDCl3, δ (ppm));
107 2,3,5-tri-O-benzoyl-1-bromo -β-D-fucofuranose (4.13): 1-O-acetyl-2,3,5-tri-O-benzoyl-D-fucofuranose
(161 mg, .311 mmol) was dissolved in dry DCM (3.1 mL) and chilled to 0 °C under N2. A solution of HBr in
AcOH (30 wt.%, 200 μL, .778 mmol) was added dropwise. After stirring cold for 30 minutes, the contents
were warmed to RT and stirred for 4h. The reaction was azeotroped with dry toluene and used in the
next reaction without further purification or characterization.
Dibenzyl 2,3,5-tri-O-benzoyl-α-D-fucofuranose phosphate (4.14): 2,3,5-tri-O-benzoyl-1-bromo -D-
fucofuranose (~168 mg, .311 mmol) was dissolved in dry toluene (1.0 mL). Separately, a solution of
dibenzyl phosphate (346mg, 1.244 mmol) in dry toluene (2.1 mL) and TEA (172 μL, 1.244 mmol) was
prepared. The bromoglycoside solution was added dropwise over 10 minutes and the reaction stirred at
RT for 2h then was concentrated on rotovap. Purification by silica gel chromatography (10% EtOAc in
toluene) allowed isolation of 77mg of pure α-anomer (34% yield, 2 steps). 1H NMR (300 MHz, CDCl3, δ
(ppm)); 13C NMR (75 MHz, CDCl3, δ (ppm)); 31P NMR (122 MHz, CDCl3, δ (ppm))
Triethylammonium 2,3,5-tri-O-benzoyl-α-D-fucofuranose phosphate: Dibenzyl 2,3,5-tri-O-benzoyl-α-D-
fucofuranose phosphate (77 mg, .104 mmol) was dissolved in ethyl acetate (1.0 mL) and TEA (100 μL)
and treated with Pd/C catalyst (10 wt.%, 35mg). The flask was evacuated and backfilled with H2 from a
balloon 3x and the reaction stirred overnight. The solution was filtered through celite, washed with ethyl
acetate and concentrated on rotovap to give 66mg of cloudy oil (96% yield). 1H NMR (300 MHz, CDCl3, δ
(ppm)); 13C NMR (75 MHz, CDCl3, δ (ppm)); 31P NMR (122 MHz, CDCl3, δ (ppm))
Tributylammonium α-D-fucofuranose phosphate (4.15): Triethylammonium 2,3,5-tri-O-benzoyl-α-D-
fucofuranose phosphate (66 mg, .1 mmol) was dissolved in a 5:2:1 MeOH:H2O:TEA mixture and stirred at
30 °C for 6d. Removed solvent on rotovap and purified by anion exchange chromatography (Biorad AG1-
X8 resin, gradient elution 250 mM 500 mM NH4HCO3 solution) using a peristaltic pump. Column
108 fractions were combined, frozen on dry ice and lyophilized. The residue was treated with ~200 μL
Dowex-50WX8-200 resin (pH 7, pretreated with 2.0M HNBu3Cl) for 4 hours. The resin was removed by
filtration through a cotton plug and the filtrate was frozen on dry ice and lyophilized. Isolated 20 mg of a
white solid (58%, 2 steps); 1H NMR (300 MHz, CDCl3, δ (ppm)); 13C NMR (75 MHz, CDCl3, δ (ppm)); 31P
NMR (122 MHz, CDCl3, δ (ppm))
UDP-6-deoxy-Galf (4.01): Trifluoroacetic anhydride was distilled from P2O5 (BP = 36 °C). UMP
tributylammonium was placed in an oven-dried pear flask with stir bar and placed under N2 and chilled
to 0°C. ACN (1.2mL), N,N-dimethylanaline (183 μL) and TEA (50 μL) were added. A solution of
trifluoroacetic anhydride (254 μL) in CAN (450 μL) was prepared at 0 °C under N2 then cannulated to the
flask with UMP, which turned yellow. After 15 minutes, the excess TFAA was evacuated using a needle
equipped to a vacuum line with N2 backfilling from a balloon. A solution of 1-methylimidazole (86 μL) in
ACN (900 μL) and TEA (250 μL) was prepared under N2 and at 0 °C. This solution was cannulated to the
activated UMP solution and stirred at 0 °C for 30 minutes. The tributylammonium 6-deoxy-Galf
monophosphate (20 mg, .027 mmol) was suspended in ACN (250 μL). Activated molecular sieves were
added. The UMP-imidazolium solution (900 μL) was added dropwise over 30 minutes. The reaction
stirred cold for 1.5h then at RT for 3.5h. The coupling was quenched with 1 mL of 100mM NH4OAc
solution, washed with 1 mL DCM, back extracted into another 1 mL 100mM NH4OAc, the aqueous layers
were combined, frozen on dry ice and stored at -20 °C until purification by HPLC. HPLC purification using
a Dionex carbopak column, 300mM NH4OAc, 100μL injections gave 21 mg of product (46% yield). 1H
NMR (300 MHz, CDCl3, δ (ppm)); 13C NMR (75 MHz, CDCl3, δ (ppm)); 31P NMR (122 MHz, CDCl3, δ (ppm))
5,6-anhydro-β-D-methylgalactofuranose: 2,3-di-O-benzyl-5,6-anhydro-β-D-methylgalactofuranose (50
mg, .140 mmol) was dissolved in reagent grade methanol and treated with Pd(OH)2/C (20wt.%, 25 mg)
109 catalyst. The flask was evacuated with H2 backfilling from a balloon 3x and stirred overnight. The
contents were filtered through celite, washed with methanol and concentrated to give 31 mg of a faint
gray oil (> 100% - catalyst contaminating); 1H NMR (300 MHz, CDCl3, δ (ppm)); 13C NMR (75 MHz, CDCl3,
δ (ppm));
6-azido-β-D-methylgalactofuranose (4.16): 5,6-anhydro-β-D-methylgalactofuranose (31 mg, ~
.140mmol) was dissolved in dry DMF under N2. Sodium azide (23 mg, .35 mmol) was added, the contents
were capped and parafilmed and heated to 100 °C to stir overnight. The reaction was cooled to RT and
concentrated on rotovap. Purification by silica gel chromatography (5% 10% DCM in methanol) 12mg
of pure product (39% yield). ); 1H NMR (300 MHz, CDCl3, δ (ppm)); 13C NMR (75 MHz, CDCl3, δ (ppm));
6-azido-2,3,5-tri-O-benzoyl-β-D-methylgalactofuranose (4.17): 6-azido-β-D-methylgalactofuranose (12
mg, .055 mmol) was dissolved in anhydrous pyridine (550 μL) and treated with benzoyl chloride (22 μL,
.191 mmol). The contents stirred overnight. Pyridine was removed by toluene azeotrope and the
product was purified by silica gel chromatography (5% 10% 20% EtOAc in hexane) to give a
colorless oil, 3mg (5% yield); 1H NMR (300 MHz, CDCl3, δ (ppm)); 13C NMR (75 MHz, CDCl3, δ (ppm));
1-acetyl-6-azido-2,3,5-tri-O-benzoyl-D-galactofuranose (4.18): 6-azido-2,3,5-tri-O-benzoyl-β-D-
methylgalactofuranose (3 mg, .007 mmol) was dissolved in a 1.4% v/v solution of H2SO4 in Ac2O (100 μL)
and stirred at RT for 2h. Quenched with a 1:1 NaHCO3:NaCl solution, extracted 3x into DCM, dried over
MgSO4, filtered and concentrated in vacuo. Purification by silica gel chromatography (5% 10% 20%
EtOAc in hexane) gave 4mg of a colorless oil (100% yield). 1H NMR (300 MHz, CDCl3, δ (ppm)); 13C NMR
(75 MHz, CDCl3, δ (ppm));
6-azido-2,3,5-tri-O-benzoyl-1-bromo-β-D-galactofuranose (4.18): 1-acetyl-6-azido-2,3,5-tri-O-benzoyl-
D-galactofuranose (4 mg, .007 mmol) was dissolved in dry DCM, chilled to 0 °C and treated dropwise
110 with a solution of HBr in AcOH (33 wt.%, 5 μL, .018 mmol). After 30 minutes of stirring cold, the reaction
was warmed to RT and stirred for 2 hours. Azeotrope with dry toluene gave a crude orange oil used in
the next reaction without further purification or characterization.
Diallyl 6-azido-2,3,5-tri-O-benzoyl-α-D-galactofuranose phosphate (4.19): 6-azido-2,3,5-tri-O-benzoyl-
1-bromo-β-D-galactofuranose (~ 4 mg, .007 mmol) was dissolved in dry toluene (100 μL). Separately, a
solution of diallyl phosphate (5 mg, .028 mmol) in dry toluene (50 μL) and TEA (4 μL, .028 mmol) was
prepared. The Galf bromide solution was added dropwise and the reaction stirred at RT for 2 hours. The
contents were concentrated on rotovap and purified by silica gel chromatography (30% 40% EtOAc in
toluene) to give pure α-anomer as a cloudy oil, 1mg (~20% yield, 2 steps). ). 1H NMR (300 MHz, CDCl3, δ
(ppm)); 13C NMR (75 MHz, CDCl3, δ (ppm)); 31P NMR (122 MHz, CDCl3, δ (ppm))
2,3,5-tri-O-benzoyl-6-fluoro-1-(thioethyl)-β-D-galactofuranose (4.22): 1-O-acetyl-2,3,5-tri-O-benzoyl-6-
fluoro-D-galactofuranose (85 mg, .158 mmol) was dissolved in dry DCM under N2 and chilled to 0 °C.
trimethylsilyl ethyl thioether (50 μL, .29 mmol) was added via syringe. Zinc iodide (30 mg, .095 mmol)
was quickly weighed and added to the reaction. The reaction was warmed to RT and stirred 2h. TEA (50
μL) was added to quench the reaction. A precipitate was removed by filtering through a cotton plug,
washing with DCM and the filtrate was concentrated in vacuo. Purification by silica gel chromatography
(5% 10% EtOAc in hexane) gave an oily white residue, 73 mg (86% yield); 1H NMR (300 MHz, CD3OD, δ
(ppm)); 19F NMR (280 MHz, CD3OD, δ (ppm)) ; 13C NMR (75 MHz, CD3OD, δ (ppm))
6-fluoro -2,3,5-tri-O-benzoyl-β-D-Galfuranosyl-(1,5)-2,3,6-tri-O-benzoyl-β-O-allylgalactofuranose
(4.23): 2,3,5-tri-O-benzoyl-6-fluoro-1-(thioethyl)-β-D-galactofuranose (20 mg, .037 mmol) and allyl 2,3,6-
tri-O-benzoyl-β-D-galactofuranose (15.2 mg, .028 mmol) were dissolved in 1 mL dry DCM, treated with
activated molecular sieves and allowed to stir under N2 for 15 minutes at RT. The solution was chilled to
111 0 °C and silver triflate (1.4 mg, .006 mmol) and N-iodosuccinimide (8.8 mg, .039 mmol) were added. A
red-brown color developed. After 30 minutes of stirring at 0 °C the reaction was filtered through celite
and washed with DCM until all the pink color was washed into the filtrate. The filtrate was quenched
with 5 mL of Na2S2O3 solution (saturated), 5 mL of saturated NaHCO3 solution, 5 mL of brine. The organic
layer was dried over MgSO4, filtered and concentrated in vacuo. Purification by silica gel
chromatography gave 21 mg of a colorless oil. (70% yield) ); 1H NMR (300 MHz, CD3OD, δ (ppm)); 19F
NMR (280 MHz, CD3OD, δ (ppm)) ; 13C NMR (75 MHz, CD3OD, δ (ppm))
6-fluoro-2,3,5-tri-O-benzoyl-β-D-galactofuranosyl-(1,5)-2,3,6-tri-O-benzoyl-β-D-(12-phenoxy-2-
undecenyl)galactofuranose (4.24): 6-fluoro -2,3,5-tri-O-benzoyl-β-D-Galfuranosyl-(1,5)-2,3,6-tri-O-
benzoyl-β-O-allylgalactofuranose (18 mg, .017 mmol) was dissolved in a solution of 11-phenoxy-
undecene (19 mg, .077 mmol) in DCM (200 μL). The solution was placed under N2 atmosphere, treated
with Grubbs’ 1st generation catalyst (1 mg, .0019 mmol) and stirred at RT overnight. Added 1 more mg of
catalyst in the morning then concentrated on rotovap. Purification by silica gel chromatography (5%
10% 20% EtOAc in hexane) gave 7 mg of cross metathesis product (33% yield). 1H NMR (300 MHz,
CD3OD, δ (ppm)); 19F NMR (280 MHz, CD3OD, δ (ppm)) ; 13C NMR (75 MHz, CD3OD, δ (ppm))
6-fluoro -β-D-galactofuranosyl-(1,5) -β-D-(12-phenoxy-2-undecenyl)galactofuranose (4.25): 6-fluoro-
2,3,5-tri-O-benzoyl-β-D-galactofuranosyl-(1,5)-2,3,6-tri-O-benzoyl-β-D-(12-phenoxy-2-
undecenyl)galactofuranose (5 mg, .004 mmol) was treated with 500 μL of 0.5M NaOMe in methanol and
stirred for 3h at RT under N2. The solution was neutralized with acidic resin, filtered off through a cotton
plug, washed with methanol and concentrated in vacuo. Purification by silica gel chromatography (10%
MeOH in DCM) gave 2.4mg of colorless film (100% yield). 1H NMR (300 MHz, CD3OD, δ (ppm)); 19F NMR
(280 MHz, CD3OD, δ (ppm)) ; 13C NMR (75 MHz, CD3OD, δ (ppm))
112 2,3,6-tri-O-benzoyl-5-fluoro-1-(thioethyl)-β-D-galactofuranose (4.26): 1-O-acetyl-2,3,6-tri-O-benzoyl-5-
fluoro-D-galactofuranose (20 mg, ..037 mmol) was dissolved in dry DCM (100 μL) under N2 and chilled
to 0 °C. trimethylsilyl ethyl thioether (9 μL, .056 mmol) was added via syringe. Zinc iodide (6 mg, .018
mmol) was quickly weighed and added to the reaction. The reaction was warmed to RT and stirred 2h.
TEA (2 drops) was added to quench the reaction. A precipitate was removed by filtering through a
cotton plug, washing with DCM and the filtrate was concentrated in vacuo. Purification by silica gel
chromatography (5% 10% EtOAc in hexane) gave an oily white residue, 15 mg (75% yield); 1H NMR
(300 MHz, CD3OD, δ (ppm)); 19F NMR (280 MHz, CD3OD, δ (ppm)) ; 13C NMR (75 MHz, CD3OD, δ (ppm))
5-fluoro -2,3,6-tri-O-benzoyl-β-D-Galfuranosyl-(1,6)-2,3,5-tri-O-benzoyl-β-O-allylgalactofuranose
(4.27): 2,3,6-tri-O-benzoyl-5-fluoro-1-(thioethyl)-β-D-galactofuranose (15 mg, .027 mmol) and allyl 2,3,5-
tri-O-benzoyl-β-D-galactofuranose (11 mg, .021 mmol) were dissolved in 1 mL dry DCM (1.0 mL), treated
with activated molecular sieves and allowed to stir under N2 for 15 minutes at RT. The solution was
chilled to 0 °C and silver triflate (1.3 mg, .005 mmol) and N-iodosuccinimide (7 mg, .029 mmol) were
added. A red-brown color developed. After 30 minutes of stirring at 0 °C the reaction was filtered
through celite and washed with DCM until all the pink color was washed into the filtrate. The filtrate was
quenched with 5 mL of Na2S2O3 solution (saturated), 5 mL of saturated NaHCO3 solution, 5 mL of brine.
The organic layer was dried over MgSO4, filtered and concentrated in vacuo. Purification by silica gel
chromatography gave 10 mg of a colorless oil. (48% yield) ); 1H NMR (300 MHz, CD3OD, δ (ppm)); 19F
NMR (280 MHz, CD3OD, δ (ppm)) ; 13C NMR (75 MHz, CD3OD, δ (ppm))
5-fluoro-2,3,6-tri-O-benzoyl-β-D-galactofuranosyl-(1,6)-2,3,5-tri-O-benzoyl-β-D-(12-phenoxy-2-
undecenyl)galactofuranose (4.28): 5-fluoro -2,3,6-tri-O-benzoyl-β-D-Galfuranosyl-(1,6)-2,3,5-tri-O-
benzoyl-β-O-allylgalactofuranose (10 mg, .010 mmol) was dissolved in a solution of 11-phenoxy-
113 undecene (11 mg, .045 mmol) in DCM (100 μL). The solution was placed under N2 atmosphere, treated
with Grubbs’ 1st generation catalyst (1 mg, .001 mmol) and stirred at RT overnight. Added 1 more mg of
catalyst in the morning then concentrated on rotovap. Purification by silica gel chromatography (5%
10% 20% EtOAc in hexane) gave 5.8 mg of cross metathesis product (40% yield). 1H NMR (300 MHz,
CD3OD, δ (ppm)); 19F NMR (280 MHz, CD3OD, δ (ppm)) ; 13C NMR (75 MHz, CD3OD, δ (ppm))
5-fluoro -β-D-galactofuranosyl-(1,6) -β-D-(12-phenoxy-2-undecenyl)galactofuranose (4.29): 5-fluoro-
2,3,6-tri-O-benzoyl-β-D-galactofuranosyl-(1,6)-2,3,5-tri-O-benzoyl-β-D-(12-phenoxy-2-
undecenyl)galactofuranose (5.8 mg, .004 mmol) was treated with 600 μL of 0.5M NaOMe in methanol
and stirred for 3h at RT under N2. The solution was neutralized with acidic resin, filtered off through a
cotton plug, washed with methanol and concentrated in vacuo. Purification by silica gel chromatography
(10% MeOH in DCM) gave 2.2 mg of colorless film (92% yield). 1H NMR (300 MHz, CD3OD, δ (ppm)); 19F
NMR (280 MHz, CD3OD, δ (ppm)) ; 13C NMR (75 MHz, CD3OD, δ (ppm))
114
References
(1) Sauvage, E.; Kerff, F.; Terrak, M.; Ayala, J. A.; Charlier, P. FEMS Microbiol. Rev. 2008, 32, 234. (2) Yother, J. Annu. Rev. Microbiol. 2011, 65, 563. (3) Ramsey, D. M.; Wozniak, D. J. Molecular Microbiology 2005, 56, 309. (4) Liu, B.; Knirel, Y. A.; Feng, L.; Perepelov, A. V.; Senchenkova, S. y. N.; Wang, Q.; Reeves, P. R.; Wang, L. FEMS Microbiol. Rev. 2008, 32, 627. (5) Samuel, G.; Reeves, P. Carbohydrate Research 2003, 338, 2503. (6) Swoboda, J. G.; Campbell, J.; Meredith, T. C.; Walker, S. ChemBioChem 2010, 11, 35. (7) Nothaft, H.; Szymanski, C. M. Nat Rev Micro 2010, 8, 765. (8) Guerry, P.; Szymanski, C. M. Trends in Microbiology 2008, 16, 428. (9) Eugster, M. R.; Haug, M. C.; Huwiler, S. G.; Loessner, M. J. Molecular Microbiology 2011, 81, 1419. (10) Shen, G.-J.; Datta, A. K.; Izumi, M.; Koeller, K. M.; Wong, C.-H. J. Biol. Chem. 1999, 274, 35139. (11) Schmidt, R. R.; Pedersen, C. M.; Qiao, Y.; Zahringer, U. Org. Biomol. Chem. 2011, 9, 2040. (12) Zhang, Z.; McCallum, S. A.; Xie, J.; Nieto, L.; Corzana, F.; Jim nez-Barbero, J. s.; Chen, M.; Liu, J.; Linhardt, R. J. J. Am. Chem. Soc. 2008, 130, 12998. (13) Raetz, C. R. H.; Whitfield, C. Annu. Rev. Biochem. 2002, 71, 635. (14) Lairson, L. L.; Henrissat, B.; Davies, G. J.; Withers, S. G. Annu. Rev. Biochem. 2008, 77, 521. (15) Osawa, T.; Sugiura, N.; Shimada, H.; Hirooka, R.; Tsuji, A.; Shirakawa, T.; Fukuyama, K.; Kimura, M.; Kimata, K.; Kakuta, Y. Biochemical and Biophysical Research Communications 2009, 378, 10. (16) Gründling, A.; Schneewind, O. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 8478. (17) Lu, D.; Wörmann, M. E.; Zhang, X.; Schneewind, O.; Gründling, A.; Freemont, P. S. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 1584. (18) Yuan, Y.; Barrett, D.; Zhang, Y.; Kahne, D.; Sliz, P.; Walker, S. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 5348. (19) Huang, C.-Y.; Shih, H.-W.; Lin, L.-Y.; Tien, Y.-W.; Cheng, T.-J. R.; Cheng, W.-C.; Wong, C.-H.; Ma, C. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 6496. (20) Steiner, K.; Novotny, R.; Werz, D. B.; Zarschler, K.; Seeberger, P. H.; Hofinger, A.; Kosma, P.; Schäffer, C.; Messner, P. J. Biol. Chem. 2008, 283, 21120. (21) Splain, R. A.; Kiessling, L. L. Bioorg. Med. Chem. 2010, 18, 3753. (22) Schwarz, F.; Lizak, C.; Fan, Y.-Y.; Fleurkens, S.; Kowarik, M.; Aebi, M. Glycobiology 2011, 21, 45. (23) Woodward, R.; Yi, W.; Li, L.; Zhao, G.; Eguchi, H.; Sridhar, P. R.; Guo, H.; Song, J. K.; Motari, E.; Cai, L.; Kelleher, P.; Liu, X.; Han, W.; Zhang, W.; Ding, Y.; Li, M.; Wang, P. G. Nat. Chem. Biol. 2010, 6, 418. (24) Han, W.; Wu, B.; Li, L.; Zhao, G.; Woodward, R.; Pettit, N.; Cai, L.; Thon, V.; Wang, P. G. J. Biol. Chem. 2012, 287, 5357. (25) Chen, M. M.; Weerapana, E.; Ciepichal, E.; Stupak, J.; Reid, C. W.; Swiezewska, E.; Imperiali, B. Biochemistry 2007, 46, 14342.
115 (26) Morrison, J. P.; Troutman, J. M.; Imperiali, B. Bioorganic & Medicinal Chemistry 2010, 18, 8167. (27) Glover, K. J.; Weerapana, E.; Imperiali, B. Proceedings of the National Academy of Sciences of the United States of America 2005, 102, 14255. (28) Li, L.; Woodward, R.; Ding, Y.; Liu, X.-w.; Yi, W.; Bhatt, V. S.; Chen, M.; Zhang, L.-w.; Wang, P. G. Biochemical and Biophysical Research Communications 2010, 394, 1069. (29) Kelly, J.; Jarrell, H.; Millar, L.; Tessier, L.; Fiori, L. M.; Lau, P. C.; Allan, B.; Szymanski, C. M. Journal of Bacteriology 2006, 188, 2427. (30) Feldman, M. F.; Wacker, M.; Hernandez, M.; Hitchen, P. G.; Marolda, C. L.; Kowarik, M.; Morris, H. R.; Dell, A.; Valvano, M. A.; Aebi, M. Proceedings of the National Academy of Sciences of the United States of America 2005, 102, 3016. (31) Wacker, M.; Linton, D.; Hitchen, P. G.; Nita-Lazar, M.; Haslam, S. M.; North, S. J.; Panico, M.; Morris, H. R.; Dell, A.; Wren, B. W.; Aebi, M. Science 2002, 298, 1790. (32) Ryder, C.; Byrd, M.; Wozniak, D. J. Current Opinion in Microbiology 2007, 10, 644. (33) Remminghorst, U.; Rehm, B. H. A. Applied and Environmental Microbiology 2006, 72, 298. (34) Remminghorst, U.; Hay, I. D.; Rehm, B. H. A. Journal of Biotechnology 2009, 140, 176. (35) Franklin, M. J.; Chitnis, C. E.; Gacesa, P.; Sonesson, A.; White, D. C.; Ohman, D. E. Journal of Bacteriology 1994, 176, 1821. (36) Schiller, N. L.; Monday, S. R.; Boyd, C. M.; Keen, N. T.; Ohman, D. E. Journal of Bacteriology 1993, 175, 4780. (37) Remminghorst, U.; Rehm, B. H. A. FEBS Letters 2006, 580, 3883. (38) Oglesby, L. L.; Jain, S.; Ohman, D. E. Microbiology 2008, 154, 1605. (39) Aarons, S. J.; Sutherland, I. W.; Chakrabarty, A. M.; Gallagher, M. P. Microbiology 1997, 143, 641. (40) Robles-Price, A.; Wong, T. Y.; Sletta, H.; Valla, S.; Schiller, N. L. Journal of Bacteriology 2004, 186, 7369. (41) Rose, N. L.; Completo, G. C.; Lin, S. J.; McNeil, M.; Palcic, M. M.; Lowary, T. L. J. Am. Chem. Soc. 2006, 128, 6721. (42) Webb, A. J.; Karatsa-Dodgson, M.; Gründling, A. Molecular Microbiology 2009, 74, 299. (43) Brown, S.; Zhang, Y.-H.; Walker, S. Chemistry & Biology 2008, 15, 12. (44) Formstone, A.; Carballido-López, R.; Noirot, P.; Errington, J.; Scheffers, D.-J. Journal of Bacteriology 2008, 190, 1812. (45) Holland, L. M.; Conlon, B.; O'Gara, J. P. Microbiology 2011, 157, 408. (46) Allison, S. E.; D'Elia, M. A.; Arar, S.; Monteiro, M. A.; Brown, E. D. J. Biol. Chem. 2011, 286, 23708. (47) Atilano, M. L.; Pereira, P. M.; Yates, J.; Reed, P.; Veiga, H.; Pinho, M. G.; Filipe, S. R. Proc. Natl. Acad. Sci. U. S. A. 2010, 107, 18991. (48) Campbell, J.; Singh, A. K.; Santa Maria, J. P.; Kim, Y.; Brown, S.; Swoboda, J. G.; Mylonakis, E.; Wilkinson, B. J.; Walker, S. ACS Chem. Biol. 2010, 6, 106. (49) Brown, S.; Meredith, T.; Swoboda, J.; Walker, S. Chemistry & Biology 2010, 17, 1101. (50) Chen, L.; Men, H.; Ha, S.; Ye, X.-Y.; Brunner, L.; Hu, Y.; Walker, S. Biochemistry 2002, 41, 6824. (51) Troutman, J. M.; Imperiali, B. Biochemistry 2009, 48, 2807. (52) Sewell, E. W. C.; Pereira, M. P.; Brown, E. D. J. Biol. Chem. 2009, 284, 21132.
116 (53) Perlstein, D. L.; Andrew Wang, T.-S.; Doud, E. H.; Kahne, D.; Walker, S. J. Am. Chem. Soc. 2009, 132, 48. (54) Cartee, R. T.; Forsee, W. T.; Schutzbach, J. S.; Yother, J. J. Biol. Chem. 2000, 275, 3907. (55) Meselson, M.; Stahl, F. W. Proc. Natl. Acad. Sci. U. S. A. 1958, 44, 671. (56) Levengood, M. R.; Splain, R. A.; Kiessling, L. L. J. Am. Chem. Soc. 2011, 133, 12758. (57) Cartee, R. T.; Forsee, W. T.; Yother, J. Journal of Bacteriology 2005, 187, 4470. (58) Besra, G. S.; Khoo, K. H.; McNeil, M. R.; Dell, A.; Morris, H. R.; Brennan, P. J. Biochemistry 1995, 34, 4257. (59) Kremer, L.; Dover, L. G.; Morehouse, C.; Hitchin, P.; Everett, M.; Morris, H. R.; Dell, A.; Brennan, P. J.; McNeil, M. R.; Flaherty, C.; Duncan, K.; Besra, G. S. J. Biol. Chem. 2001, 276, 26430. (60) Berg, S.; Kaur, D.; Jackson, M.; Brennan, P. J. Glycobiology 2007, 17, 35R. (61) Beláňová, M.; Dianišková, P.; Brennan, P. J.; Completo, G. C.; Rose, N. L.; Lowary, T. L.; Mikušová, K. Journal of Bacteriology 2008, 190, 1141. (62) May, J. F.; Splain, R. A.; Brotschi, C.; Kiessling, L. L. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 11851. (63) Schertzer, J. W.; Brown, E. D. J. Biol. Chem. 2003, 278, 18002. (64) Pereira, M. P.; Schertzer, J. W.; D'Elia, M. A.; Koteva, K. P.; Hughes, D. W.; Wright, G. D.; Brown, E. D. ChemBioChem 2008, 9, 1385. (65) Forsee, W. T.; Cartee, R. T.; Yother, J. J. Biol. Chem. 2006, 281, 6283. (66) Forsee, W. T.; Cartee, R. T.; Yother, J. J. Biol. Chem. 2009, 284, 11826. (67) Schirner, K.; Marles-Wright, J.; Lewis, R. J.; Errington, J. EMBO J 2009, 28, 830. (68) Poulin, M. B.; Zhou, R.; Lowary, T. L. Org. Biomol. Chem. 2012, 10, 4074. (69) Weigel, P. H.; DeAngelis, P. L. J. Biol. Chem. 2007, 282, 36777. (70) Weigel, P. H.; Kyossev, Z.; Torres, L. C. J. Biol. Chem. 2006, 281, 36542. (71) Tlapak-Simmons, V. L.; Kempner, E. S.; Baggenstoss, B. A.; Weigel, P. H. J. Biol. Chem. 1998, 273, 26100. (72) Tlapak-Simmons, V. L.; Baggenstoss, B. A.; Kumari, K.; Heldermon, C.; Weigel, P. H. J. Biol. Chem. 1999, 274, 4246. (73) Baggenstoss, B. A.; Weigel, P. H. Analytical Biochemistry 2006, 352, 243. (74) Kyossev, Z.; Weigel, P. H. Analytical Biochemistry 2007, 371, 62. (75) Kumari, K.; Baggenstoss, B. A.; Parker, A. L.; Weigel, P. H. J. Biol. Chem. 2006, 281, 11755. (76) Wörmann, M. E.; Corrigan, R. M.; Simpson, P. J.; Matthews, S. J.; Gründling, A. Molecular Microbiology 2011, 79, 566. (77) Carter, J. A.; Jiménez, J. C.; Zaldívar, M.; Álvarez, S. A.; Marolda, C. L.; Valvano, M. A.; Contreras, I. Microbiology 2009, 155, 3260. (78) Morona, R.; van den Bosch, L.; Manning, P. A. Journal of Bacteriology 1995, 177, 1059. (79) Bastin, D. A.; Stevenson, G.; Brown, P. K.; Haase, A.; Reeves, P. R. Molecular Microbiology 1993, 7, 725. (80) Larue, K.; Kimber, M. S.; Ford, R.; Whitfield, C. J. Biol. Chem. 2009, 284, 7395. (81) Schertzer, J. W.; Brown, E. D. Journal of Bacteriology 2008, 190, 6940. (82) Sutcliffe, I. C. Molecular Microbiology 2011, 79, 553. (83) Forsee, W. T.; Cartee, R. T.; Yother, J. J. Biol. Chem. 2009, 284, 11836. (84) Clarke, B. R.; Cuthbertson, L.; Whitfield, C. J. Biol. Chem. 2004, 279, 35709. (85) Clarke, B. R.; Richards, M. R.; Greenfield, L. K.; Hou, D.; Lowary, T. L.; Whitfield, C. J. Biol. Chem. 2011, 286, 41391.
117 (86) Clarke, B. R.; Greenfield, L. K.; Bouwman, C.; Whitfield, C. J. Biol. Chem. 2009, 284, 30662. (87) Messner, P.; Steiner, K.; Zarschler, K.; Schäffer, C. Carbohydrate Research 2008, 343, 1934. (88) Kiessling, L. L.; Splain, R. A. Annu. Rev. Biochem. 2010, 79, 619. (89) Steiner, K.; Hagelueken, G.; Messner, P.; Schäffer, C.; Naismith, J. H. Journal of Molecular Biology 2010, 397, 436. (90) Kido, N.; Kobayashi, H. Journal of Bacteriology 2000, 182, 2567. (91) May, J. F.; Levengood, M. R.; Splain, R. A.; Brown, C. D.; Kiessling, L. L. Biochemistry 2012. (92) Brown, C. D.; Rusek, M. S.; Kiessling, L. L. J. Am. Chem. Soc. 2012, 134, 6552. (93) Keenleyside, W. J.; Perry, M.; Maclean, L.; Poppe, C.; Whitfield, C. Molecular Microbiology 1994, 11, 437. (94) Keenleyside, W. J.; Whitfield, C. J. Biol. Chem. 1996, 271, 28581. (95) Keenleyside, W. J.; Clarke, A. J.; Whitfield, C. Journal of Bacteriology 2001, 183, 77. (96) Vionnet, J.; Vann, W. F. Glycobiology 2007, 17, 735. (97) Morley, T. J.; Withers, S. G. J. Am. Chem. Soc. 2010, 132, 9430. (98) Campbell, J.; Singh, A. K.; Swoboda, J. G.; Gilmore, M. S.; Wilkinson, B. J.; Walker, S. Antimicrobial Agents and Chemotherapy 2012, 56, 1810. (99) Pan, F.; Jackson, M.; Ma, Y.; McNeil, M. J. Bacteriol. 2001, 183, 3991. (100) Sassetti, C. M.; Rubin, E. J. Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 12989. (101) Dykhuizen, E. C.; May, J. F.; Tongpenyai, A.; Kiessling, L. L. J. Am. Chem. Soc. 2008, 130, 6706. (102) Mikusova, K.; Yagi, T.; Stern, R.; McNeil, M. R.; Besra, G. S.; Crick, D. C.; Brennan, P. J. J. Biol. Chem. 2000, 275, 33890. (103) Mikušová, K.; Beláňová, M.; Korduláková, J.; Honda, K.; McNeil, M. R.; Mahapatra, S.; Crick, D. C.; Brennan, P. J. Journal of Bacteriology 2006, 188, 6592. (104) Gandolfi-Donadío, L.; Gallo-Rodriguez, C.; de Lederkremer, R. M. J. Org. Chem. 2003, 68, 6928. (105) Szczepina, M. G.; Zheng, R. B.; Completo, G. C.; Lowary, T. L.; Pinto, B. M. ChemBioChem 2009, 10, 2052. (106) Completo, G. C.; Lowary, T. L. J. Org. Chem. 2008, 73, 4513. (107) Tsvetkov, Y. E.; Nikolaev, A. V. J. Chem. Soc. 2000, 889. (108) Marlow, A. L.; Kiessling, L. L. Org. Lett. 2001, 3, 2517. (109) Peltier, P.; Daniellou, R.; Nugier-Chauvin, C.; Ferrières, V. Org. Lett. 2007, 9, 5227. (110) Errey, J. C.; Mukhopadhyay, B.; Kartha, K. P. R.; Field, R. A. Chem. Commun. 2004, 2706. (111) Peltier, P.; BeláHová, M.; Dianiaková, P.; Zhou, R.; Zheng, R. B.; Pearcey, J. A.; Joe, M.; Brennan, P. J.; Nugier-Chauvin, C.; Ferrières, V.; Lowary, T. L.; Daniellou, R.; Mikuaová, K. Chem. Biol. 2010, 17, 1356. (112) Bordoni, A.; Lima, C.; Mariño, K.; de Lederkremer, R. M.; Marino, C. Carbohydrate Research 2008, 343, 1863. (113) Anslyn, E. a. D., D. Modern Physical Organic Chemistry; University Science Books: Sausalito, 2006. (114) Hagen, K.; Hedberg, K. J. Am. Chem. Soc. 1973, 95, 8263. (115) Guvench, O.; MacKerell, A. D. J. Phys. Chem. A 2006, 110, 9934. (116) Withers, S. G.; Street, I. P.; Bird, P.; Dolphin, D. H. J. Am. Chem. Soc. 1987, 109, 7530.
118 (117) Allman, S. A.; Jensen, H. H.; Vijayakrishnan, B.; Garnett, J. A.; Leon, E.; Liu, Y.; Anthony, D. C.; Sibson, N. R.; Feizi, T.; Matthews, S.; Davis, B. G. ChemBioChem 2009, 10, 2522. (118) Northrup, A. B.; MacMillan, D. W. C. Science 2004, 305, 1752. (119) Covell, D. J.; Vermeulen, N. A.; Labenz, N. A.; White, M. C. Angewandte Chemie International Edition 2006, 45, 8217. (120) Halcomb, R. L.; Danishefsky, S. J. J. Am. Chem. Soc. 1989, 111, 6661. (121) Sanders, W. J.; Kiessling, L. L. Tet. Lett. 1994, 35, 7335. (122) Furuya, T.; Kamlet, A. S.; Ritter, T. Nature 2011, 473, 470. (123) Schaus, S. E.; Brandes, B. D.; Larrow, J. F.; Tokunaga, M.; Hansen, K. B.; Gould, A. E.; Furrow, M. E.; Jacobsen, E. N. J. Am. Chem. Soc. 2002, 124, 1307. (124) Mohamady, S.; Desoky, A.; Taylor, S. D. Org. Lett. 2011, 14, 402. (125) Peltier, P.; Guégan, J.-P.; Daniellou, R.; Nugier-Chauvin, C.; Ferrières, V. Eur. J. Org. Chem. 2008, 2008, 5988. (126) Moretti, R.; Thorson, J. S. J. Biol. Chem. 2007, 282, 16942. (127) Wheatley, R. W.; Zheng, R. B.; Richards, M. R.; Lowary, T. L.; Ng, K. K. S. J. Biol. Chem. 2012. (128) Sanger, F.; Nicklen, S.; Coulson, A. R. Proc. Natl. Acad. Sci. U. S. A. 1977, 74, 5463. (129) Pouilly, S.; Bourgeaux, V.; Piller, F.; Piller, V. ACS Chem. Biol. 2012. (130) Barthel, S. R.; Antonopoulos, A.; Cedeno-Laurent, F.; Schaffer, L.; Hernandez, G.; Patil, S. A.; North, S. J.; Dell, A.; Matta, K. L.; Neelamegham, S.; Haslam, S. M.; Dimitroff, C. J. J. Biol. Chem. 2011, 286, 21717. (131) Ruprecht, R. M.; O'Brien, L. G.; Rossoni, L. D.; Nusinoff-Lehrman, S. Nature 1986, 323, 467. (132) Szczepina, M. G.; Zheng, R. B.; Completo, G. C.; Lowary, T. L.; Pinto, B. M. Bioorganic & Medicinal Chemistry 2010, 18, 5123.
119
Appendix I
NMR Spectra
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262