A Direct Fluorescent Activity Assay for Glycosyltransferases Enables Convenient High-Throughput Screening: Application to O-GlcNAc Transferase

Article abstract of DOI:10.1002/anie.202000621

Glycosyltransferases carry out important cellular functions in species ranging from bacteria to humans. Despite their essential roles in biology, simple and robust activity assays that can be easily applied to high-throughput screening for inhibitors of t

Full text of DOI:10.1002/anie.202000621

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: A direct fluorescent activity assay for glycosyltransferases
enables convenient high-throughput screening: Application to O-
GlcNAc Transferase
Authors: David Jaro Vocadlo, Matthew G. Alteen, Christina Gros,
Richard Meek, David A. Cardoso, Jil Busmann, Gontran
Sangouard, Matthew C. Deen, Hong-Yee Tan, David L. Shen,
Cecilia C. Russell, Gideon J. Davies, Phillip J. Robinson, and
Adam McCluskey
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.202000621
Angew. Chem. 10.1002/ange.202000621
Link to VoR: http://dx.doi.org/10.1002/anie.202000621
http://dx.doi.org/10.1002/ange.202000621

10.1002/anie.202000621
Angewandte Chemie International Edition
A direct fluorescent activity assay for glycosyltransferases enables
convenient high-throughput screening: application to O-GlcNAc
Transferase
Matthew G. Alteen,[a] Christina Gros,[a] Richard W. Meek,[b] David A. Cardoso,[c] Jil A.
Busmann,[d] Gontran Sangouard,[a ]Matthew C. Deen,[a] Hong-Yee Tan,[a] David L. Shen,[d] Cecilia
C. Russell,[e] Gideon J. Davies,[b] Phillip J. Robinson,[c] Adam McCluskey [e] and David J.
Vocadlo*[a],[b]
[a] Department of Chemistry, Simon Fraser University, Burnaby, BC, V5A 1S6, Canada
[b] York Structural Biology Laboratory, Department of Chemistry, University of York, York, YO10
5DD, UK
[c] Children’s Medical Research Institute, The University of Sydney, Sydney, NSW, 2145,
Australia
[d] Department of Molecular Biology and Biochemistry, Simon Fraser University, Burnaby, BC,
V5A 1S6, Canada
[e] Chemistry, School of Environmental and Life Sciences, The University of Newcastle,
University Drive, Callaghan, NSW, 2308, Australia
*Corresponding author: David J. Vocadlo (dvocadlo@sfu.ca)
Abstract: Glycosyltransferases carry out important cellular functions in species ranging from
bacteria to humans. Despite their essential roles in biology, simple and robust activity assays that
can be easily applied to high-throughput screening for inhibitors of these enzymes have been
challenging to develop. Here we report a bead-based strategy to sensitively measure the group
transfer activity of glycosyltransferases using simple fluorescence measurements, without the
need for coupled enzymes or secondary reactions. We validate the performance and accuracy of
the assay using O-GlcNAc Transferase (OGT) as a model system through detailed Michaelis-
Menten kinetic analysis of various substrates and inhibitors. Optimization of this assay and
application to high-throughput screening enabled screening for inhibitors of OGT, leading to a
novel inhibitory scaffold. We believe this assay will prove valuable not only for the study of OGT,
but also more widely as a general approach for the screening of glycosyltransferases and other
group transfer enzymes.
Keywords: Glycosyltransferase, enzyme assay, high-throughput screening, O-GlcNAc,
glycoprotein, fluorescent substrates
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
Table of contents figure
OH
O
OH
O
+
HO
HO
OGT
HO
HO
O
HN
O
NH
OUDP
HO
O
OH
O
HO
HO
Biotinylated
Fluorophore
Fluorophore
HN
peptide acceptor
O
OUDP
(excess)
Fluorophore
1) Streptavidin immobilization
2) Wash cycle
Fluorescence
scan
Kinetic
assays
Screening
A robust fluorescence-based assay accurately reports on the sugar transfer activity of
glycosyltransferases. This assay is amenable to high-throughput screening and should be
applicable to a diverse set of group transfer enzymes. We validate this assay through screening
of a library of known bioactive molecules to identify a new O-GlcNAc Transferase antagonist.
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
Introduction
The large superfamily of glycosyltransferases (GTs) are responsible for the assembly of
the diverse glycoconjugates found throughout every kingdom of life[1,2]. The increasing recognition
of the roles played by glycans, including in regulating human physiology, are raising interest in
these enzymes and the associated development of chemical approaches to monitor and modulate
their activities[3–6]. One area of emerging interest is the development of reliable
glycosyltransferase enzyme assays for characterizing their activities[7,8], alongside with their use
in high throughput screening efforts to identify small molecule inhibitors[4,9]. The standard activity
assay that directly measures glycosyl transfer relies on the enzyme catalyzed transfer of
radiolabeled sugar units from nucleotide sugar donor substrates to a suitable acceptor substrate,
which is then followed by liquid scintillation detection[9–11]. The requirement for using radioactive
materials, however, detracts from the wide uptake and applicability of such assays and has
spurred interest in developing new assays for various GTs.
Among the various strategies that have been pursued are coupled enzyme activity assays
that, for example, include measuring the release of nucleoside diphosphate leaving groups via
spectrophotometric detection of reduced nicotinamide adenine dinucleotide (NADH) or the
consumption of nucleoside diphosphates (NDPs) to drive formation of adenosine triphosphate
(ATP) that is then detected by bioluminescence[12,13]. ELISA-based luminescent approaches have
also been developed, whereby transfer of a monosaccharide unit is detected by a tagged pan-
specific antibody[14]. Additionally, a coupled fucosyltransferase assay, in which transfer of a fucose
to a fluorogenic oligosaccharide prevents a glycoside hydrolase from acting to release the
fluorophore reporter was developed recently[15]. Various fluorescence polarization (FP) assays
have also been advanced. Walker and colleagues, who pioneered work on screening for inhibitors
of O-GlcNAc Transferase (OGT), used a fluorescein-linked UDP-GlcNAc analogue that served as
a probe for an FP binding assay[16]. Paulson and colleagues expanded further on this approach
by developing fluorescently-labeled nucleotide sugar donors that are substrates for a sialyl and
fucosyltransferase, resulting in increased polarization when transferred to a glycoprotein
substrate[4]. Withers and co-workers extended this approach to create an “FP-Tag” assay which
uses an azide-labeled glycosyl donor that is transferred to a fluorescent oligosaccharide and
conjugated to biotin via click chemistry, permitting pull-down with streptavidin and subsequent FP
detection[17]
.
While these methods represent notable advancements, the use of coupled assays
requiring a secondary enzyme or chemical transfer step is not ideal as this introduces an
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
additional step at which test materials can interfere. In HTS assays in particular, such interference
can yield a high number of false positives resulting from inhibition of the secondary processes
instead of the primary target[18]. FP assays also do not directly measure enzymatic activity and
compounds that disrupt binding through non-specific means, such as binding or displacing the
fluorophore non-specifically from the surface of a protein, can similarly yield false positives[19,20]
.
Additionally, FP assays require high concentrations of protein and are sometimes incapable of
detecting weak interactions[20]. Given these considerations, we aimed to develop a simple and
direct quantitative fluorescence-based assay that would provide high-quality kinetic data using
only small amounts of enzyme without the need for downstream coupling enzymes, bioorthogonal
reactions, or detection reagents. This concept has been applied to binding processes, notably for
fluorescently-labeled aptamers[21], but surprisingly has never, to our knowledge, been applied to
group transfer enzymes such as GTs. However, we expected that such an assay might prove
useful for ongoing efforts to study the enzymatic activity of GTs toward various substrates as well
as to characterize molecules that could modulate the activity of these enzymes. Moreover, we
envisioned that adaptation of such an assay to a format compatible with high-throughput
screening would facilitate identifying new leads for GTs that could be suitable for development as
tool compounds.
As a model system of high interest with which we have experience, we focused on the
enzyme O-GlcNAc Transferase (OGT), which catalyzes the installation of a single O-linked N-
acetylglucosamine (O-GlcNAc) to the hydroxyl groups of serine and threonine residues of nuclear
and cytosolic proteins[22][23]. This reversible modification has been identified on hundreds of protein
substrates[24] and regulates central cellular processes such as transcription[25], protein stability
cellular stress response[26], and cell cycle regulation[27]. Moreover, dysregulated OGT activity has
been implicated in many signalling pathways related to cancer progression and tumor growth.
Indeed, mounting evidence suggests that OGT is a promising therapeutic target for various
cancers[28–31]. Furthermore, consistent with the high interest in this enzyme, previous efforts to
screen OGT have been pursued and yielded various leads[16,32–34]. Improvement of these leads
has, however, proven challenging. Walker and co-workers were the first to undertake this,
carrying out two separate screens for OGT inhibitors using their FP displacement assay as well
as a coupled protease protection assay[16,32]. These efforts resulted in a class of quinolone-6-
sulfonate-based (Q6S) compounds,[33] which were recently advanced to yield promising
compounds with nanomolar in vitro potency[34]. However, many of the inhibitors arising from these
screens adversely affect cell viability and likely suffer from significant off-target effects[35]. Efforts
to improve the potency of the Q6S scaffold over the past decade have yielded good success,
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
however, the resulting lead has a high molecular weight that may impair brain permeability and
other in vivo applications. Ideally, as has often proven useful, the identification and advancement
of new and diverse scaffolds that can be optimized are needed in order to ultimately identify
complementary tool compounds for in vivo use.
Despite its importance in contributing to cellular physiology, and the complexity of the
mechanisms by which OGT function must accordingly be regulated, studying this enzyme remains
challenging in part because of a lack of convenient assays and chemical biology tools[14,36]. Here
we describe the development of a simple and sensitive fluorescence-based assay that directly
and precisely measures glycosyl transfer catalyzed by OGT. We validate this assay and illustrate
its amenability to high-throughput screening in microplate format. After implementing the assay in
a trial screen, we identify a known bioactive inhibitor that also antagonizes OGT and confirm its
ability to directly inhibit OGT.
Results and Discussion
OGT is able to use uridine diphosphate N-acetylglucosamine (UDP-GlcNAc, 1) analogues
having various 2-N-acyl groups in place of the 2-acetamido substituent[16,37]. UDP-GlcNAc
analogues bearing large fluorophores at this 2-position were previously found only to bind to OGT
rather than be turned over[16], however, judicious selection of linkers revealed appending a
nitrobezoxadiazole (NBD) fluorophore to this position was tolerated.[37] Based on these
observations, we reasoned that using suitable linkers might enable incorporation of other
fluorophores having different physicochemical and photophysical properties. We therefore
elected to prepare a BODIPY-FL dye linked using an N-hexanoyl linker (2, UDP-GlcN-BODIPY,
Figure 1). We were attracted to this option because of the relatively small size of this fluorophore
in combination with its desirable photophysical properties, including high photostability and
quantum yield. We therefore prepared UDP-GlcN-BODIPY (Scheme 1) by modification of
established methods starting with glucosamine hydrochloride (3). After protection of the hydroxyl
groups and masking of the amine as an azide functionality, we installed the anomeric phosphate
using MacDonald phosphorylation conditions[38]. The acetate protecting groups were then
removed, the azide reduced, and the resulting amine coupled to BODIPY-FL caproic acid using
HBTU peptide coupling conditions. Finally, the desired nucleoside, UDP-GlcN-BODIPY, was
prepared by coupling of intermediate 8 with UMP-morpholidate. We found after exploration of
several routes that installation of the fluorophore prior to the morpholidate coupling permitted
facile purification by silica chromatography and enabled suitable scale-up of this reaction. Details
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
regarding the synthesis, including preparation of BODIPY-FL caproic acid, are included in the
Supporting Information (SI).
O
OH
O
OH
O
NH
HO
HO
O
NH
HO
HO
O
O
N
O
NH
O
O
O
N
O
O
NH
O
O
P
O
O
P
O
P
O
O
P
O
5
OH
OH
HN
OH
OH
O
HO
OH
HO
OH
1
2
F
F
B
N
N
Figure 1: Structure of UDP-GlcNAc (1), the natural donor substrate used by OGT, and the
fluorescent analogue UDP-GlcN-BODIPY (2).
OH
OAc
OAc
i
iii
O
O
O
HO
HO
AcO
AcO
AcO
AcO
ii
O
NH2 OH
.HCl
N3 OAc
N3
O
P
OH
3
4
5
OH
iv
OH
OH
OH
v
O
vi
O
O
HO
HO
HO
HO
HO
HO
O
O
P
NH2
N3
O
O
NH
O
O
P
OH
OH
OH
O
P
OH
6
OH
7
OH
5
OH
HN
O
8
O
F
F
vii
B
O
NH
N
HO
HO
N
O
P
O
N
O
NH
O
O
P
O
O
O
5
OH
OH
HN
O
HO
OH
2
F
F
B
N
N
Scheme 1: Synthetic route towards UDP-GlcN-BODIPY (2). Reagents and conditions: i)
imidazole-1-sulfonyl azide•HCl, CuSO4•5H2O, K2CO3, MeOH, 2 h, 92%; ii) Ac2O, pyridine, 10 h,
98%; iii) H3PO4 (s), vacuum, 80 mTorr, 60°C, 44%; iv) NaOMe, MeOH, 92%; v) H2, Pd/C, MeOH,
98%; vi) BODIPY-FL-X, HBTU, DIPEA, DMF, 37%; vii) UMP-morpholidate, N-
methylimidazole•HCl, DMF, 45%.
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
With UDP-GlcN-BODIPY (2) in hand, we assessed its competency as a donor substrate
for OGT using a well studied OGT peptide substrate derived from Casein Kinase II (CKII). Using
this TAMRA-conjugated CKII peptide (TAMRA-NH-YPGGSTPVSSANMM-OH) we observed,
using HPLC-based analysis, a modest level of transfer of the GlcN-BODIPY moiety to the peptide
that was confirmed by mass spectrometry (MS) analysis of the reaction products (Figure S1). To
understand the basis for molecular recognition of UDP-GlcN-BODIPY by OGT and how this
enzyme tolerated this larger BODIPY moiety, we crystallized a truncated construct of OGT in
complex with substrate UDP-GlcN-BODIPY[39]. This data (Table S1) enabled building a clear
model revealing how the caproic acid linker is sufficiently long for the BODIPY fluorophore to
completely extrude from the active site, causing minimal disruption of the active site architecture
(Figure 2, Figure S2). Strong difference density was observable at the end of the linker and this
is likely attributable to the BODIPY fluorophore, however, the fluorophore cannot be modelled
with confidence (Figure S3). This ambiguity is likely a result of the BODIPY fluorophore occupying
multiple conformations. A second BODIPY fluorophore can be modelled near the substrate
(Figure 2), but this may stem from OGT-catalyzed hydrolysis of UDP-GlcN-BODIPY or,
alternatively, a second molecule of substrate being present in proximity to the active site. The
uridine and pyranose ring are positioned within the active site in a manner similar to those
structures previously reported with bound UDP, except that there is a slight rotation in the α-
phosphate. This may stem from the altered position of Q839 leading to rotation of the α-phosphate
to within hydrogen bonding distance of T921.
Figure 2: Crystallographic structure of OGT with substrate 2 bound. The OGT active site (grey)
is depicted as a surface representation demonstrating how it is able to accommodate the caproic
acid linker of the substrate (light blue). The fluorophore portion of a second substrate molecule
can be modeled nearby.
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
Encouraged by these results, we set out to implement a convenient microplate-based
assay (Figure 3). We envisioned we could incubate a biotinylated peptide as an OGT acceptor
substrate along with glycosyl donor 2 and initiate the reaction through the addition of recombinant
OGT to generate a BODIPY-modified glycopeptide. The modified and unmodified peptides can
then be captured on streptavidin-coated surfaces after specified times. In this way, we reasoned
we could directly monitor the glycosyl transfer catalyzed by OGT. We set out to use the same
CKII peptide as noted above, which is a well-studied substrate of OGT and is readily prepared by
solid-phase peptide synthesis, in its biotinylated form (Table S3) as a convenient acceptor
substrate for the assay. The N-terminal biotin handle permitted immobilization on streptavidin-
coated surfaces, which in turn facilitated subsequent washing to remove unreacted UDP-GlcN-
BODIPY. The fluorescent signal remaining after washing could thus correspond to the BODIPY-
modified glycopeptide, which can be obtained using a standard microplate reader (Figure S4). In
this way, a measure of direct glycosyl transfer to a peptide acceptor can be obtained and a wide
range of parameters can be varied to optimize the sensitivity of the assay. We found that both
commercial streptavidin-coated microplates and streptavidin-coated magnetic beads could be
used as a capture medium for the assay. We decided to use magnetic beads for subsequent
experiments, since this format provided greater signal over background.
OH
O
OH
O
+
HO
HO
OGT
HO
HO
O
HN
O
NH
OUDP
HO
O
OH
O
HO
HO
Biotinylated
peptide acceptor
Fluorophore
Fluorophore
HN
O
OUDP
(excess)
Fluorophore
1) Streptavidin immobilization
2) Wash cycle
Fluorescence
scan
Kinetic
assays
Screening
Figure 3: Methodology for activity-based OGT assay and inhibitor screen. Incubation of a peptide
substrate with OGT results in transfer of a fluorescent sugar analogue. Immobilization on
streptavidin-coated beads followed by washing permits direct measurement of transfer by
fluorescence in 96- or 384-well plate format.
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
We recognized that one key factor in assay reproducibility would be to stop the assay at
a defined time point by preventing further O-GlcNAc incorporation during the biotin-streptavidin
capture. We reasoned that UDP, which is known to competitively inhibit OGT (Ki of 1.3 µM)[40],
could be used to stop the enzymatic reaction prior to washing. Using this assay, we found that 2
mM UDP completely inhibits the reaction when concentrations of UDP-GlcN-BODIPY in the assay
were equal or less than 25 µM (Figure S5). After optimizing the assay conditions and
concentrations of both substrates and OGT, we found that the assay produced fluorescence
signals 50-fold greater than control reactions performed either in the presence of 2 mM UDP or
in the absence of CKII peptide acceptor. By monitoring the reaction as a function of time at various
concentrations of 2, we were able to perform Michaelis-Menten analyses to obtain kinetic
parameters for both the UDP-GlcN-BODIPY donor and CKII peptide acceptor (Figures 4A and
4B). As expected, the presence of the BODIPY fluorophore and linker impaired the rate of the
reaction. This effect was quite modest, however, since the kcat/KM value for UDP-GlcN-BODIPY
when using CKII as the acceptor was found to be ~7-fold lower than the value previously reported
for {3H}-UDP-GlcNAc when using a similar CKII peptide acceptor in a radioactivity-based assay
(Table S2)[39]. Interestingly, the KMapp for UDP-GlcN-BODIPY (KM = 2.9 ± 0.3 µM) was found to be
similar to than reported for UDP-GlcNAc (KM =2.3 ± 0.4 µM) and the reduction in kcat value was
approximately 6-fold (0.05 ± 0.002 min-1) as compared to the value measured for UDP-GlcNAc
(0.29 ± 0.01 min-1) (Table S2). The similar KMapp values observed for UDP-GlcNAc and UDP-GlcN-
BODIPY, coupled with the structural data, suggest that OGT tolerates binding of the fluorophore
and linker in its active site, and that the reduction in second order rate constant is largely driven
by other factors affecting catalytic transfer to the peptide. We note that in the binary complex we
observe the linker of compound 2, and by extension the unmodeled fluorophore, reaches across
the active site cleft to where the incoming peptide acceptor has been observed in ternary
complexes[39]. We therefore reason that the linker or fluorophore may partially hinder binding of
peptide acceptors, which may in turn explain the reduction in kcat values observed using UDP-
GlcN-BODIPY.
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
Figure 4: UDP-GlcN-BODIPY (2) is a substrate for OGT, permitting convenient determination of
kinetic parameters. A, Michaelis-Menten curve of glycosyl donor 2 at saturating concentrations of
peptide acceptor (>10-fold above KM). B, Michaelis-Menten curve of Biotin-CKII peptide acceptor
in the presence of saturating concentrations of glycosyl donor. C, Lineweaver-Burk plot showing
inhibition of OGT by UDP at varying concentrations of glycosyl donor 2, with a calculated Ki value
of 6.1 µM. Error bars represent S.E.M. of triplicate measurements.
Next, having established initial conditions, we evaluated a small panel of biotinylated
peptide acceptors that had been previously identified as OGT substrates (Figure S6). A TAK1-
binding protein 1 (TAB-1) peptide[41] was found to have comparable kinetic parameters to those
measured for CKII, whereas a peptide derived from host cell factor 1 (HCF-1)[42] was a 22-fold
better substrate as determined by comparing its kcat/KM value to that obtained for CKII (Table S3).
We therefore chose this peptide for subsequent studies since it produced greater signal to noise,
even at lower concentrations of enzyme. We next investigated the utility of this assay for analysis
of inhibitors, both in performing IC50 studies as well as for determining mode-of-inhibition and Ki
values (Figure 4C). The inhibition of OGT using UDP and various concentrations of UDP-GlcN-
BODIPY in the presence of saturating concentrations of the CKII peptide acceptor gave a Ki value
of 6.9 ± 0.7 µM, which is in close agreement with previous reports using the radioactivity-based
glycosyl transfer assay[43]. Lineweaver-Burk graphical analysis of this data also clearly showed
the expected pattern of competitive inhibition. In contrast, when inhibition was assessed over
varying concentrations of peptide acceptor, a non-competitive pattern of inhibition was obtained
with respect to UDP (Figure S7). Importantly, these comparisons demonstrated the utility of our
assay for the convenient determination of Michaelis-Menten parameters across various
substrates and conditions as well as in detailed inhibition studies to establish the site at which an
inhibitor binds within the active site of OGT.
Having demonstrated the applicability and high sensitivity of our assay in kinetic
experiments, we next adapted the assay for use in 384-well microplates to enable economic use
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
of materials as well as with a view to apply it to high-throughput screening for small-molecule
inhibitors of OGT. The concentration of donor and acceptor substrates used were at
approximately KM for each substrate in order to make the assay sensitive to compounds that might
compete for binding at either the donor or acceptor binding site. The reaction buffer was also
changed to include 0.01% (v/v) Triton X-100 so as to reduce potential non-specific inhibition by
compounds prone to colloidal aggregation, which is a common problem encountered when
performing screening assays[19], and we reduced the total reaction volume to 25 µL. These
changes had no significant effect on assay behaviour. We also optimized the automated plate
washing protocol to minimize disruption of the streptavidin-coated magnetic beads in the wells by
by introducing a short delay between wash cycles, which allows the beads to settle. We also
adjusted the height at which aspiration was performed above the well base and found that a 6
mm height struck a balance between effective washing of the beads without aspirating them out
of the wells, resulting in satisfactory signal-to-noise (Figure S8). Although we recognized that
interference in screening could arise from antagonists that disrupt the biotin-streptavidin
interaction, felt this extremely high affinity interaction would be one of the most robust binding
interactions since the resulting complex has a half-life in the order of hours[44]. Accordingly, we
reasoned that a 30 minute incubation time for capture would be sufficient to avoid interference.
To test this assumption, we showed that desthiobiotin, a biotin analogue with a KD of
approximately 10-11 M towards streptavidin[45], at concentrations of up to 25 µM did not influence
the performance of the assay (Figure S9). This result suggested that the capture of the
biotinylated substrate would not be impaired by even potent streptavidin antagonists, including
less potent candidate antagonists that might be encountered during HTS.
Using this optimized assay, we next set out to validate this assay in high-throughput format
by performing a trial screen. As a proof of concept, we used a targeted library containing 4480
known bioactive compounds obtained commercially from TargetMol Corp (Wellesley Hills, MA).
We were able to conduct this assay over the course of one day in a fully automated format, which
validated the suitability of this assay for high-throughput screening. The overall Z’ score obtained
for the assay was 0.67 (Figure 5) and we observed a clear Gaussian distribution of compound
activities. Because the discovery of glycosyltransferase inhibitors from screening libraries has
proved difficult, and identifying tractable inhibitors from past efforts to screen OGT has been
challenging, we chose to screen using a relatively high compound concentration of 100 µM. This
translated to an overall hit rate of 6.1%, with hits being defined as those compounds resulting in
OGT activity that was less than three standard deviations from the mean of our 100% activity
control, which corresponds to 58.5% enzyme activity relative to the positive control vehicle only
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
standards. Of the initial 275 hits we identified from primary screening, a short-list of 88 compounds
for re-testing in triplicate was prepared by first excluding compounds in the library with a high
likelihood of pan-assay interference (PAINS)[46] such as those that might function as detergents,
or have fluorescent moieties or reactive functional groups. From these confirmed hits, we further
prioritized 11 compounds that displayed desirable properties and obtained them commercially to
determine inhibition constants. Compounds were also subjected to a secondary thermal shift
assay using differential scanning fluorimetry (DSF)[47] to confirm direct binding of the materials to
the enzyme.
Figure 5: Summary of screening results against OGT from a targeted library of 4480 known
chemical probes. Compounds were screened at a final concentration of 100 µM. µpos corresponds
to the mean signal of uninhibited positive control wells. µneg corresponds to the mean signal of
negative control wells in the absence of enzyme. Hits were defined as compounds resulting in
less than 58.5% OGT activity, corresponding to at least three standard deviations from the
positive control signal for activity. Prioritized hits which were selected for follow-up studies are
marked as blue circles. Structures of these compounds and bioassay results are listed in the
Supporting Information (Table S4).
Using our secondary binding assay in combination with our activity assay, we rejected the
majority of our selected 11 hit compounds due to non-specific behaviour such as detergent
sensitivity, solubility issues, or failure to replicate activity in both assay formats. However, we
identified one compound (Dyngo®-4a, 9, Table 1) which displayed low-micromolar affinity towards
OGT in both our primary enzyme assay and secondary DSF assay. This compound was initially
developed as part of a chemical family of inhibitors targeting the dynamin GTPase, an enzyme
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
responsible for the rate-limiting scission of vesicles at the plasma membrane that occurs in many
forms of endocytosis just prior to vesicle internalization[48,49]. Notably, Dyngo-4a represents a great
improvement over its predecessor, known as Dynasore, both in terms of its potency against
dynamin and consequent effects on endocytosis. Furthermore, Dyngo-4a exhibits desirable
chemical properties including a lack of detergent sensitivity[49,50]. Given its utility in the inhibition
of endocytosis, Dyngo-4a and other dynamin inhibitors have demonstrated efficacy in multiple
proof-of-concept disease studies[51,52]. We were surprised to find that Dyngo-4a, which is a
comparatively small molecule, also bound and inhibited OGT with an IC50 value of 6.3 ± 0.7 µM
(Figure 6A). Using detailed kinetic assays, we showed that Dyngo-4a inhibits OGT non-
competitively with respect to both the glycosyl donor and the peptide acceptor (Figure S10). Given
the phenolic nature of Dyngo-4a, we speculated that it could lead to time dependent inactivation
of OGT, but our kinetic studies show no time-dependent loss of activity, suggesting inhibition of
OGT by Dyngo-4a is reversible (Figure S11). Additionally, the presence of 1 mM DTT in the assay
buffer reduces the likelihood of oxidation of the phenolic hydroxyls to a quinone moiety.
Furthermore, the positive shift in melting temperature observed when OGT is incubated with
Dyngo-4a at a concentration of 10 µM, coupled with the lack of detergent sensitivity and the Hill
slope of -1, indicates that the compound directly binds and stabilizes the protein, rather than
inactivating the enzyme non-specifically (Figure 6A and 6B).
Figure 6: Dyngo-4a binds and inhibits OGT with low-micromolar affinity. A, Concentration-
response showing Dyngo-4a inhibition of recombinant OGT. Error bars represent S.E.M. (n=4).
B, DSF-based thermal shift assay showing thermal stabilization of the OGT catalytic domain in
the presence of 10 µM Dyngo-4a. Dashed lines indicate S.E.M (n=3). A temperature ramp of 25°C
to 95°C was used at a gradient of 0.05°C/sec.
Although we were encouraged by the ability of Dyngo-4a to inhibit OGT, we recognized
that this off-target inhibition would complicate future studies of this tool compound, both for
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
potential use as an OGT inhibitor but also for use as a dynamin inhibitor. We therefore aimed to
identify structurally related compounds that would inhibit each target in a selective manner by
testing previously synthesized Dyngo-4a analogues against OGT. Through screening of a small
library of 167 Dyngo-4a analogues, we identified an additional 9 compounds that showed at least
50% inhibition against OGT at a concentration of 40 µM (Figure S12). Two of these compounds
display minimal activity towards dynamin, but largely retain their potency toward OGT (Figure 7).
Compound 10, which differs from the parent compound by the removal of one hydroxyl group,
retains an IC50 value for OGT of 19 ± 3 µM (Table 1). Compound 11 displays less potency towards
OGT but also displays selectivity, as no inhibition towards dynamin was observed at
concentrations up to 100 µM (Table 1, Figure S13). These compounds should serve as a small
novel scaffold from which more potent analogues can be derived. Additionally, the results
obtained here highlight the utility of our assay for the quick identification of Dyngo-4a analogues
that do not show inhibition towards OGT, suggesting that the development of more potent and
selective effectors of dynamin activity is also feasible.
Figure 7: Screening of a series of Dyngo-4a analogues reveals two additional compounds with
selectivity for OGT. Concentration response curves show inhibition of recombinant OGT by
Dyngo-4a (9, black), and structurally related analogues 10 (blue), and 11 (red). Error bars
represent S.E.M. of triplicate measurements.
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
IC50
(OGT)
6.3 ±
IC50
(DynI)
2.7 ±
Structure
Compound
OH
OH
OH
0.7 µM
0.7 µM
O
Dyngo-4a (9)
N
N
H
OH
HO
19 ± 2
µM
33± 0.7
µM
O
Dyngo-4-070
(10)
N
N
OH
H
OH
58± 9
µM
HO
OH
OH
> 100
µM
O
Dyngo-4-183
(11)
N
N
H
Table 1: Comparison of biological activity observed from Dyngo-4a (9) and related analogues
(10,11).
Conclusions
Owing to the important roles played by glycosyltransferases, the development of inhibitors
of these enzymes is currently a topic of high interest. A prerequisite for successful identification
of such inhibitors is the development of robust assays for this class of enzymes. Remarkably few
GT assays are known in which the assay reports directly on the enzyme catalyzed transfer of the
carbohydrate unit. Here we exploit the substrate tolerance of OGT to develop what is, to our
knowledge, the first fluorescence-based assay that reports directly on the transfer of the sugar
unit. Optimization of this assay yields a highly sensitive and accurate assay that can be
conveniently used to obtain Michalis-Menten kinetic parameters for both substrates and inhibitors.
This assay is furthermore amenable for use in high-throughput screening. After HTS-based
discovery of a known endocytosis inhibitor in a screening campaign, we used this assay to rapidly
assess a family of analogues in order to identify a selective OGT inhibitor. While these leads show
relatively modest potency, the low molecular weight of these compounds may make them a useful
starting point to generate tool compounds for OGT and provide a novel scaffold for development
of more potent analogues. We expect this assay will have great utility, both for routine kinetic
analysis of OGT activity as well as for additional HTS campaigns for inhibitors of OGT.
While the substrates used in this assay are tailored for OGT, we note that many
glycosyltransferases that have been studied in this regard demonstrate tolerance for pendant
groups[4,15,17,53]. Although the degree of tolerance is highly dependent on the individual enzyme, it
is likely that similar fluorescent substrates can be developed for other GTs but may require
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
optimization of both the type of linker and its site of attachment. Therefore, we believe this assay
format could be readily adapted to other families of GTs, including those that have already been
shown to tolerate pendent groups including fucosyltransferases[4,54,55], sialyltransferases[4,55], and
polypetide GalNAc transferases[56]. Combining the use of such fluorescent donor sugars with
biotinylated acceptors will permit the study of these other classes of GTs. An advantage of our
assay is that it is compatible for use with GTs that use non-Leloir glycosyl donors, which cannot
be monitored by coupled assays that detect nucleotide diphosphate reaction products.
Additionally, as a screening assay, our approach is capable of robust and efficient HTS and can
detect weak (>50 µM) binders, which is useful for GTs as these are notoriously difficult to target.
The ability to screen with low concentrations of enzyme (20 nM), as well as low donor (3 µM) and
acceptor substrates (10 µM), makes this approach reagent conservative and therefore readily
feasible for high throughput screening. Notably, this flexibility also permits the design of screens
that are biased toward a particular type of inhibition – for example, by increasing the concentration
of one substrate or the other above KM in order to select for non-competitive or allosteric inhibitors.
We also note that the strategy of directly incorporating the fluorescent reporter to a substrate,
adding magnetic beads to the reaction wells, and then measuring on-bead fluorescence after
washing offers a simplified assay design that minimizes reagent transfer and liquid handling steps.
We therefore expect this general assay format will aid in the mission of discovering improved
chemical tools for the important superfamily of glycosyltransferases. Finally, we believe that this
technique will have applications for other group transfer enzymes that are difficult to monitor
directly including, for example, acyltransferases and nucleic acid polymerases.
Acknowledgements
D.J.V. is supported by a Tier I Canada Research Chair in Chemical Glycobiology and an
E.W.R. Steacie Memorial Fellowship. Additional financial support was provided by the Canadian
Glycomics Network (R675184) and a Discovery Grant from the Natural Sciences and Engineering
Research Council (RGPIN-2015-05426). The miniaturization of the assay and the high throughput
screening campaign were performed in the Center for High Throughput Chemical Biology (HTCB)
at Simon Fraser University. The HTCB is supported by grants from the Canadian Foundation for
Innovation (Innovation Fund Grant Number 33097, JELF Grant Number 36766, and JELF Grant
Number 37241), the British Columbia Knowledge Development Foundation (Grant Numbers 662-
805223 and 862-805636), and the Y.P. Heung Foundation. We would like to thank Diamond Light
Source for access to beamline I04-1 (proposal number mx18598). GJD thanks the Royal Society
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
for the Ken Murray Research Professorship and RWM for the associated PDRA funding
(RP\EA\180016). P.J.R. and A.M are supported by the National Health & Medical Research
Council Australia (GNT1162515 and GNT1137064), the Australian Research Council
(DP180101781) and for equipment from the Australian Cancer Research Foundation, the
Ramaciotti Foundation and the Cancer Institute NSW.
Competing Interests
Dyngo® is a registered trademark of Newcastle Innovation Ltd and Children’s Medical Research
Institute and some of the Dyngo® compounds are commercially available to researchers via
Abcam Biochemicals® (Cambridge, UK).
References:
[1]
[2]
A. Varki, Glycobiology 2017, 27, 3–49.
L. L. Lairson, B. Henrissat, G. J. Davies, S. G. Withers, Annu. Rev. Biochem. 2008, 77,
521–555.
[3]
[4]
S. S. Pinho, C. A. Reis, Nat. Rev. Cancer 2015, 15, 540–555.
C. D. Rillahan, S. J. Brown, A. C. Register, H. Rosen, J. C. Paulson, Angew. Chemie -
Int. Ed. 2011, 50, 12534–12537.
[5]
[6]
C. W. Hall, T. F. Mah, FEMS Microbiol. Rev. 2017, 41, 276–301.
J. Marshall, Y. Sun, D. S. Bangari, E. Budman, H. Park, J. B. Nietupski, A. Allaire, M. A.
Cromwell, B. Wang, G. A. Grabowski, et al., Mol. Ther. 2016, 24, 1019–1029.
[7]
[8]
G. K. Wagner, T. Pesnot, ChemBioChem 2010, 11, 1939–1949.
Z. L. Wu, C. M. Ethen, B. Prather, M. Machacek, W. Jiang, Glycobiology 2011, 21, 727–
733.
[9]
O. von Ahsen, U. Voigtmann, M. Klotz, N. Nifantiev, A. Schottelius, A. Ernst, B. Müller-
Tiemann, K. Parczyk, Anal. Biochem. 2008, 372, 96–105.
[10] W. A. Lubas, J. A. Hanover, J. Biol. Chem. 2000, 275, 10983–10988.
[11] J. J. M. Bergeron, R. A. Rachubinski, R. A. Sikstrom, B. I. Posner, J. Paiement, J. Cell
Biol. 1982, 92, 139–146.
[12] S. Gosselin, M. Alhussaini, M. H. Streiff, K. Takabayashi, M. M. Palcic, Anal. Biochem.
1994, 220, 92–97.
[13] S. Zegzouti, H., Engel, L., Hennek, J., Alves, J., Vidugiris, G., Goueli, S., Glycobiology
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
2014, 24, 1051.
[14] J. Qi, R. Wang, Y. Zeng, W. Yu, Y. Gu, Prep. Biochem. Biotechnol. 2017, 47, 699–702.
[15] X. Zhang, F. Chen, A. Petrella, F. Chacón-Huete, J. Covone, T.-W. Tsai, C.-C. Yu, P.
Forgione, D. H. Kwan, ACS Chem. Biol. 2019, 14, 715–724.
[16] B. J. Gross, B. C. Kraybill, S. Walker, J. Am. Chem. Soc. 2005, 127, 14588–14589.
[17] Z. Gao, O. G. Ovchinnikova, B.-S. Huang, F. Liu, D. E. Williams, R. J. Andersen, T. L.
Lowary, C. Whitfield, S. G. Withers, 2019, DOI 10.1021/jacs.8b10940.
[18] M. G. Acker, D. S. Auld, Perspect. Sci. 2014, 1, 56–73.
[19] B. Y. Feng, A. Simeonov, A. Jadhav, K. Babaoglu, J. Inglese, B. K. Shoichet, C. P.
Austin, J. Med. Chem. 2007, 50, 2385–2390.
[20] W. A. Lea, A. Simeonov, Expert Opin. Drug Discov. 2011, 6, 17–32.
[21] A. Wochner, J. Glökler, Biotechniques 2007, 42, 578–582.
[22] T. Carmen-Rosa, G. W. Hart, J. Biol. Chem. 1984, 259, 3308–3317.
[23] X. Yang, K. Qian, Nat. Rev. Mol. Cell Biol. 2017, 18, 452–465.
[24] J. C. Trinidad, D. T. Barkan, B. F. Gulledge, A. Thalhammer, A. Sali, R. Schoepfer, A. L.
Burlingame, Mol. Cell. Proteomics 2012, 11, 215–229.
[25] S. Brimble, E. Wollaston-Hayden, C. Teo, A. Morris, L. Wells, Curr. Signal Transduct.
Ther. 2010, 5, 12–24.
[26] J. A. Groves, A. Lee, G. Yildirir, N. E. Zachara, Cell Stress Chaperones 2013, 18, 535–
558.
[27] F. Capotosti, S. Guernier, F. Lammers, P. Waridel, Y. Cai, J. Jin, J. W. Conaway, R. C.
Conaway, W. Herr, Cell 2011, 144, 376–388.
[28] J. A. Hanover, W. Chen, M. R. Bond, J. Bioenerg. Biomembr. 2018, 50, 155–173.
[29] C. M. Ferrer, T. Y. Lu, Z. A. Bacigalupa, C. D. Katsetos, D. A. Sinclair, M. J. Reginato,
Oncogene 2017, 36, 559–569.
[30] T. P. Lynch, C. M. Ferrer, S. R. E. Jackson, K. S. Shahriari, K. Vosseller, M. J. Reginato,
J. Biol. Chem. 2012, 287, 11070–11081.
[31] R. M. De Queiroz, R. Madan, J. Chien, W. B. Dias, C. Slawson, J. Biol. Chem. 2016, 291,
18897–18914.
[32] B. J. Gross, J. G. Swoboda, S. Walker, J. Am. Chem. Soc. 2008, 130, 440–441.
[33] R. F. Ortiz-Meoz, J. Jiang, M. B. Lazarus, M. Orman, J. Janetzko, C. Fan, D. Y. Duveau,
Z. W. Tan, C. J. Thomas, S. Walker, ACS Chem. Biol. 2015, 10, 1392–1397.
[34] S. E. S. Martin, Z. W. Tan, H. M. Itkonen, D. Y. Duveau, J. A. Paulo, J. Janetzko, P. L.
Boutz, L. Törk, F. A. Moss, C. J. Thomas, et al., J. Am. Chem. Soc. 2018, 140, 13542–
13545.
[35] R. Trapannone, K. Rafie, D. M. F. Van Aalten, Biochem. Soc. Trans. 2016, 44, DOI
10.1042/BST20150189.
This article is protected by copyright. All rights reserved.

10.1002/anie.202000621
Angewandte Chemie International Edition
[36] E. J. Kim, L. K. Abramowitz, M. R. Bond, D. C. Love, D. W. Kang, H. F. Leucke, D. W.
Kang, J. S. Ahn, J. A. Hanover, Bioconjug. Chem. 2014, 25, 1025–1030.
[37] H. Y. Tan, R. Eskandari, D. Shen, Y. Zhu, T. W. Liu, L. I. Willems, M. G. Alteen, Z.
Madden, D. J. Vocadlo, J. Am. Chem. Soc. 2018, 140, 15300–15308.
[38] D. L. MacDonald, J. Org. Chem. 1962, 27, 1107–1109.
[39] M. B. Lazarus, Y. Nam, J. Jiang, P. Sliz, S. Walker, Nature 2011, 469, 564–569.
[40] K. Rafie, A. Gorelik, R. Trapannone, V. S. Borodkin, D. M. F. van Aalten, Bioconjug.
Chem. 2018, 29, 1834–1840.
[41] S. Pathak, V. S. Borodkin, O. Albarbarawi, D. G. Campbell, A. Ibrahim, D. M. van Aalten,
EMBO J. 2012, 31, 1394–1404.
[42] M. B. Lazarus, J. Jiang, T. M. Gloster, W. F. Zandberg, G. E. Whitworth, D. J. Vocadlo, S.
Walker, Nat. Chem. Biol. 2012, 8, 966–968.
[43] D. L. Shen, T. M. Gloster, S. A. Yuzwa, D. J. Vocadlo, J. Biol. Chem. 2012, 287, 15395–
15408.
[44] A. Chilkoti, P. S. Stayton, Molecular Origins of the Slow Streptavidin-Biotin Dissociation
Kinetics, 1995.
[45] J. D. Hirsch, L. Eslamizar, B. J. Filanoski, N. Malekzadeh, R. P. Haugland, J. M.
Beechem, R. P. Haugland, Anal. Biochem. 2002, 308, 343–357.
[46] J. B. Baell, J. W. M. Nissink, ACS Chem. Biol. 2018, 13, 36–44.
[47] F. H. Niesen, H. Berglund, M. Vedadi, Nat. Protoc. 2007, 2, 2212–2221.
[48] G. J. Doherty, H. T. McMahon, Annu. Rev. Biochem. 2009, 78, 857–902.
[49] A. Mccluskey, J. A. Daniel, G. Hadzic, N. Chau, E. L. Clayton, A. Mariana, A. Whiting, N.
N. Gorgani, J. Lloyd, A. Quan, et al., Traffic 2013, 14, 1272–1289.
[50] E. Macia, M. Ehrlich, R. Massol, E. Boucrot, C. Brunner, T. Kirchhausen, Dev. Cell 2006,
10, 839–850.
[51] C. B. Harper, S. Martin, T. H. Nguyen, S. J. Daniels, N. A. Lavidis, M. R. Popoff, G.
Hadzic, A. Mariana, N. Chau, A. McCluskey, et al., J. Biol. Chem. 2011, 286, 35966–
35976.
[52] Y. Y. Li, X. N. Chen, X. X. Fan, Y. J. Zhang, J. Gu, X. W. Fu, Z. H. Wang, X. F. Wang, Z.
Xiao, Synapse 2015, 69, 67–77.
[53] M. Noel, P. A. Gilormini, V. Cogez, N. Yamakawa, D. Vicogne, C. Lion, C. Biot, Y.
Guérardel, A. Harduin-Lepers, ChemBioChem 2017, 18, 1251–1259.
[54] T. Maeda, S.-I. Nishimura, Chem. - A Eur. J. 2008, 14, 478–487.
[55] S. Hong, P. Sahai-Hernandez, D. G. Chapla, K. W. Moremen, D. Traver, P. Wu, Angew.
Chemie Int. Ed. 2019, 58, 14327–14333.
[56] H. C. Hang, C. Yu, M. R. Pratt, C. R. Bertozzi, J. Am. Chem. Soc. 2004, 126, 6–7.
This article is protected by copyright. All rights reserved.