A novel fluorescent probe for retaining galactosyltransferases

Article abstract of DOI:10.1002/cbic.201000013

Glycosyltransferases (GTs) are a large class of carbohydrate-active enzymes that are involved, in both pro- and eukaryotic organisms, in numerous important biological processes, from cellular adhesion to carcinogenesis. GTs have enormous potential as mole

Full text of DOI:10.1002/cbic.201000013

DOI: 10.1002/cbic.201000013
A Novel Fluorescent Probe for Retaining
Galactosyltransferases
Thomas Pesnot,[a] Monica M. Palcic,[b] and Gerd K. Wagner*[a, c]
Glycosyltransferases (GTs) are a large class of carbohydrate-
active enzymes that are involved, in both pro- and eukaryotic
organisms, in numerous important biological processes, from
cellular adhesion to carcinogenesis. GTs have enormous poten-
tial as molecular targets for chemical biology and drug discov-
ery. For the full realisation of this potential, operationally
simple and generally applicable GT bioassays, especially for in-
hibitor screening, are indispensable tools. In order to facilitate
the development of GT high-throughput screening assays for
the identification of GT inhibitors, we have developed novel,
fluorescent derivatives of UDP-galactose (UDP-Gal) that are rec-
ognised as donor analogues by several different retaining gal-
actosyltransferases (GalTs). We demonstrate for one of these
derivatives that fluorescence emission is quenched upon spe-
cific binding to individual GalTs, and that this effect can be
used as the read-out in ligand-displacement experiments. The
novel fluorophore acts as an excellent sensor for several differ-
ent enzymes and is suitable for the development of a new
type of GalT bioassay, whose modular nature and operational
simplicity will significantly facilitate inhibitor screening. Impor-
tantly, the structural differences between the natural donor
UDP-Gal and the new fluorescent derivatives are minimal, and
the general assay principle described herein may therefore
also be applicable to other GalTs and/or proteins that use nu-
cleotides or nucleotide conjugates as their cofactor.
Introduction
Glycosyltransferases (GTs) are a large class of enzymes that are
required, in both pro- and eukaryotic organisms, for the bio-
synthesis of complex carbohydrates and glycoconjugates.[1–4]
GTs catalyse the transfer of a mono- or oligosaccharide from a
glycosyl donor, generally a sugar nucleotide, to a diverse range
of acceptor substrates, including saccharides, proteins, lipids
and secondary metabolites.[1–5] GTs are involved in numerous
fundamental biological processes, from cellular adhesion to
carcinogenesis and neurobiology,[6–8] and it has been estimated
that this class of enzymes accounts for up to 1% of ORFs
across all sequenced genomes.[3] However, compared to other
enzyme classes of similar size and biological significance (e.g.,
the protein kinases), the considerable potential of GTs as mo-
lecular targets for chemical biology and drug discovery has yet
to be fully realised. Towards this goal, operationally simple GT
bioassays, for example, for studies on donor/acceptor selectivi-
ty, function analysis and inhibitor screening, are indispensable
tools, and considerable progress has recently been made in
the development of such assays.[9–13]
(GlcNAc) transferases.[15,16] However, the considerable steric
demand of the fluorescein moiety necessitated the design of a
separate fluorophore for each enzyme and limits the general
applicability of these probes. In view of these results, the de-
velopment of a generally applicable fluorophore that can be
used for the simultaneous screening of an entire group of GTs
(e.g., several galactosyltransferases) in a single experiment is
an attractive proposition. Herein, we describe, for the first
time, a broadly applicable GT fluorophore that might have the
potential to be used in such a general assay.
UDP-a-d-galactose (UDP-Gal, Table 1) is the general sugar-
nucleotide donor for all Leloir-type galactosyltransferases
(GalTs).[17] Individual GalTs have been identified as molecular
targets in a number of therapeutic areas, including cancer and
infection.[18,19] For example, the inhibition of bacterial GalTs in-
volved in the biosynthesis of lipopoly- and lipooligosaccharides
of the Gram-negative cell envelope is a promising new strat-
[a] Dr. T. Pesnot, Dr. G. K. Wagner
School of Pharmacy, University of East Anglia
Norwich, NR4 7TJ (UK)
In view of the size of the GT enzyme family, a considerable
need remains for assay formats that are generally applicable
with different GTs. A general GT ligand-displacement assay, for
example, will allow the rapid screening of multiple enzymes in
parallel and the efficient selectivity profiling of substrate ana-
logues and inhibitors. Fluorescence-based formats are particu-
larly attractive for such applications,[14] due to their operational
simplicity and their adaptability for high-throughput screening
(HTS). In pioneering work in this area, the Walker group have
developed UDP-GalNAc/fluorescein conjugates as fluorophores
for ligand-displacement assays with two N-acetyl glucosamine
Fax: (+44)1603-592003
[b] Prof. M. M. Palcic
Carlsberg Research Centre
Gamle Carlsberg Vej 10, 2500 Valby Copenhagen (Denmark)
[c] Dr. G. K. Wagner
Current address: School of Biomedical and Health Sciences
King’s College London, Waterloo Campus
Franklin-Wilkins Building, Stamford Street, London, SE1 9NH (UK)
E-mail: gerd.wagner@kcl.ac.uk
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201000013.
1392
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1392 – 1398

Probe for Retaining Galactosyltransferases
Results and Discussion
Table 1. Fluorescence and enzymological properties of UDP-Gal and
base-modified UDP-Gal derivatives 1a–d.
Using our previously developed synthetic methodology for the
Suzuki–Miyaura cross-coupling of unprotected sugar nucleo-
tides,[22,23] we prepared a series of novel UDP-Gal derivatives
with an additional aromatic or heteroaromatic substituent in
position 5 of the uracil base (Table 1, 1a–d). In contrast to the
practically nonfluorescent parent UDP-Gal, the 5-(hetero)aryl-
substituted derivatives 1a–d are fluorescence emitters, and
their fluorescence properties can be modulated by the nature
of the 5-substituent. Crucially, while the phenyl- and furyl-sub-
stituted derivatives 1a–c showed only moderate to weak fluo-
rescence, thienyl derivative 1d was much more strongly fluo-
rescent. The quantum yield of 1d is 25 times greater than that
of 1a, and almost 6000 times greater than the quantum yield
of the parent UDP-Gal (Table 1). We anticipated that, with
these fluorescence characteristics, 1d might be a suitable fluo-
rophore for a fluorescence-based GalT ligand-displacement
assay, provided that 1d was recognised as a ligand by the
target GalTs.
Cmpd
X
Quantum yield
F
B. taurus a-(1!3)-GalT
Km [mm]
118 ꢁ14
kcat [sꢀ1
]
UDP-Gal
H
4.5ꢁ10ꢀ5
0.98
1a
<0.01
96ꢁ8
2.1ꢁ10ꢀ3
1b
0.024
82ꢁ11
4.0ꢁ10ꢀ3
1c
0.04
0.26
69ꢁ13
13ꢁ1
4.4ꢁ10ꢀ3
1.9ꢁ10ꢀ3
1d
In order to assess the influence of the additional substituent
in position 5 on GalT recognition and binding, we carried out
enzymological studies with donor analogues 1a–d and a rep-
resentative bovine a-(1!3)-GalT (Table 1). Pleasingly, we found
that although the turnover of 1a–d was considerably lower
than for UDP-Gal, the Michaelis–Menten constant of the base-
modified analogues was of a similar order of magnitude as for
UDP-Gal. These results suggested that the additional substitu-
ent in position 5 is not detrimental for binding of these donor
analogues at a-(1!3)-GalT, a key criterion for their potential
application in a GalT ligand binding assay. This interpretation
has subsequently been confirmed by structural studies with
1d and a blood-group galactosyltransferase.[24] As the most
promising analogue, with regard to both its strong binding
affinity and pronounced fluorescence, the thienyl-substituted
derivative 1d was selected for proof-of-principle investigations
into the suitability of this novel type of fluorophore for assay
development.
egy for anti-bacterial drug discovery.[19] Known GalT inhibitors
are generally donor or acceptor substrate analogues with only
limited “drug-likeness”, which could compromise their applica-
bility for, for example, cellular studies.[20] The identification of
novel, drug-like GalT-inhibitor chemotypes would be greatly
facilitated by a general GalT-screening assay. Recently, fluores-
cent derivatives of uracil nucleotides have been generated, as
probes for nucleic acid chemistry, by installation of a compact,
heteroaromatic substituent in position 5 of the uracil base.[21]
We reasoned that this phenomenon could also be harnessed
for the development of autofluorescent UDP-Gal derivatives, as
a novel type of fluorophore for GalT bioassays. We anticipated
that such base-modified UDP-Gal derivatives might be broadly
recognised by a range of different GalTs, due to the minimal
steric demand of the fluorogenic substituent and the strongly
conserved architecture of the nucleotide binding domain in
different GalTs.[2–4]
The fluorescence emission of a given fluorophore is modu-
lated by its microenvironment.[25] We therefore speculated that,
upon binding to a target GalT, the fluorescence emission of 1d
would either be enhanced or attenuated, and that the differ-
ence in fluorescence between protein-bound and free fluoro-
phore could be used as the read-out for a GalT ligand-displace-
ment assay. In order to test this hypothesis, we first carried out
titration experiments with a fixed concentration of 1d and
with bovine a-(1!3)-GalT as a representative target enzyme.
We found that the fluorescence of 1d is indeed quenched in
the presence of enzyme in a concentration-dependent manner
(Figure 1A, circles).
The implementation of this strategy has now resulted in the
identification of the novel UDP-Gal derivative 1d (Table 1).
Compound 1d is a strong fluorescence emitter and is recog-
nised as a high-affinity ligand by various GalTs. Importantly, the
fluorescence of 1d is quenched upon binding to protein, and
we have exploited this effect for the development of an opera-
tionally simple GalT ligand-displacement assay. This new GalT
assay format allows for screening against bacterial and mam-
malian enzymes in parallel, and thus the evaluation of inhibitor
candidates for their potency and selectivity in the same experi-
ment. To the best of our knowledge, this is the first example of
the use of this type of fluorophore in a ligand-displacement
assay, and the general assay principle we describe herein
might also be applicable to other (sugar)-nucleotide-depen-
dent proteins.
To assess the specificity of this quenching effect, we next
performed a range of control experiments. Like many other
GTs, B. taurus a-(1!3)-GalT requires a divalent metal such as
Mn2+ to bind the sugar-nucleotide donor.[26] Importantly, in the
absence of Mn2+, no fluorescence quenching was observed
upon titration of fluorophore 1d with a-(1!3)-GalT (Fig-
ure 1A, squares). We concluded from this result that binding of
ChemBioChem 2010, 11, 1392 – 1398
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1393

G. K. Wagner et al.
quenching, of 1d, they also allowed the determination of IC50
values for these known GalT ligands (Table 2). The order of
potency observed for UDP-Gal, UDP, UMP and uridine was in
agreement, qualitatively and quantitatively, with literature data
for inhibition of GalTs by these ligands.[20] We saw these results
as an important validation of our assay design and evidence
for its reliability. As our experimental protocol allowed us to
discriminate between a-(1!3)-GalT binders with strong (UDP-
Gal, UDP), moderate (UMP) and poor (uridine) affinity, we con-
cluded that, in principle, this assay set-up might also be suita-
ble for inhibitor screening.
Before carrying out library screening experiments, we decid-
ed to assess the generality of this new GT assay principle. We
therefore performed the requisite fluorescence quenching and
control experiments with three other GalTs. This panel of en-
zymes covered a variety of GalT activities, including the mam-
malian blood-group enzymes GTB and AA(Gly)B, and the bac-
terial a-(1!4)-GalT LgtC. AA(Gly)B is a dual-specificity enzyme
that can utilise either UDP-Gal or UDP-GalNAc as a donor sub-
strate, producing either blood group A or B structures.[24] Sig-
nificantly, we consistently observed a strong fluorescence-
quenching effect for 1d with all of these enzymes (Figure S2).
As in the case of bovine a-(1!3)-GalT, titration with UDP-Gal,
UDP and, to a lesser extent, UMP also restored the fluores-
cence of 1d in the presence of the human enzymes GTB and
AA(Gly)B, as well as the bacterial enzyme LgtC (Figure 2B–D).
On the other hand, uridine was, as expected, not an effective
competitive binder for 1d at these enzymes. Notably, the IC50
values obtained for UDP-Gal in these competition experiments
were in good agreement, for all four enzymes, with Km values
determined in other assays (Table 2). Thus, these findings dem-
onstrated not only the reliability, but also the broad applicabili-
ty of our ligand-displacement assay protocol.
Figure 1. A) Fluorescence emission of fluorophore 1d upon titration with
a-(1!3)-GalT in the presence or absence of 10 mm MnCl2. B) Fluorescence
emission of fluorophore 1d upon titration with MnCl2, in the presence or ab-
sence of 0.58 mm a-(1!3)-GalT. Conditions: 200 nm 1d, 50 mm Tris buffer
pH 7, 308C, incubation 15 min.
1d occurs specifically at the donor binding site of a-(1!3)-
GalT. This implies that the observed fluorescence-quenching
effect also is specific and not due to, for example, nonspecific
binding of 1d on the protein surface. In keeping with this
interpretation, no significant fluorescence quenching was ob-
served upon titration of 1d with bovine serum albumin (BSA),
a protein lacking a binding site for UDP sugars (Figure S1 in
the Supporting Information). To investigate the possibility that
the reduced fluorescence in the presence of Mn2+ might be
due to direct quenching by the divalent metal,[27] we also re-
peated the initial titration experiment with variable concentra-
tions of Mn2+, at fixed concentrations of 1d and a-(1!3)-GalT.
Significantly, fluorescence quenching was only observed in the
presence of all three binding partners, 1d, Mn2+, and a-(1!3)-
GalT (Figure 1B, circles). In contrast, fluorescence emission re-
mained strong upon titration of 1d with increasing concentra-
tions of Mn2+ but in the absence of enzyme (Figure 1B,
squares). Taken together, these results provided strong support
for the notion that the observed fluorescence quenching was
due to the specific binding of 1d at the donor binding site of
a-(1!3)-GalT.
With a novel fluorophore for different GalTs in hand, we
investigated its suitability for the identification and selectivity
profiling of new GalT inhibitors. In a proof-of-concept experi-
ment, we screened a small library of drug-like inhibitor candi-
dates in parallel against three different enzymes. For this
screen, we selected a structurally diverse set of thiazolidinones
as inhibitor candidates (Table 3), as thiazolidinones had previ-
ously been reported as inhibitors for other GTs.[15,28] Candidate
compounds, at a concentration of 50 mm, were co-incubated
on a single microplate with fluorophore 1d and three different
GalTs (i.e., the model a-(1!3)-GalT from B. taurus, the H. sa-
piens a-(1!3)-GalT GTB, and the N. meningitidis a-(1!4)-GalT
LgtC; Figure 3A). As expected, the competitive displacement
of 1d from the GalT donor binding site by high-affinity binders
resulted in an increase in fluorescence. To quantify the dis-
placement of fluorophore by individual binders, the fluores-
cence increase observed for the natural donor UDP-Gal was
used as a reference. Using this procedure, we were able to es-
tablish an order for the inhibitor candidates according to their
binding affinity and, at the same time, to assess their GalT-se-
lectivity profile. To validate this approach, we determined com-
plete binding curves with all three GalTs for the representative
thiazolidinone inhibitor 2b (Figure 3B). The IC50 values extract-
ed from these binding curves suggest a slightly greater affinity
This interpretation was further substantiated by the finding
that the fluorescence of 1d could be restored by titration with
nonfluorescent, competitive a-(1!3)-GalT ligands, including
UDP-Gal (Figure 2A). While these experiments confirmed the
specificity of the binding, and concomitant fluorescence
1394
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1392 – 1398

Probe for Retaining Galactosyltransferases
Figure 2. Titration of 1d (200 nm) and A) a-(1!3)-GalT, B) GTB, C) LgtC and D) AA(Gly)B with UDP-Gal, UDP, UMP and uridine, plus control experiments with-
out enzyme.
Table 2. IC50 values for different GalTs/GalT ligands, determined with fluo-
rophore 1d. Km values for UDP-Gal are given for direct comparison.
Table 3. Molecular structures of the thiazolidinone inhibitor library.
GalT
Km [mm]
IC50 [mm]
UDP-Gal
UDP-Gal UDP
UMP
uridine
B. taurus a-(1!3)-GalT 118ꢁ14[a]
109ꢁ20 74ꢁ30 240ꢁ110 >1000
Code R1
R2
Code R1
2g
R2
GTB
27[b]
28ꢁ5 18ꢁ7
7ꢁ6 12ꢁ5
48ꢁ5
22ꢁ3
>1000
>1000
AA(Gly)B
LgtC
0.7ꢁ0.1[a]
2a
2b
2c
2d
2e
2 f
CH2CO2H
CH2CO2H
3-pyr
4-BnO-C6H4 2h
CH2ꢀCH=CH2 3-pyr
18[c]
26ꢁ8 83ꢁ49 293ꢁ91 >1000
H
4-BnO-C6H4
CH2ꢀCH=CH2 4-pyr
2i
2j
2k
2l
CH2ꢀCH=CH2 2-pyr
[a] This study, HPLC-based assay. [b] Ref. [33]. [c] Ref. [19].
CH2CO2H
H
4-pyr
Ph
2-pyr
CH2ꢀCH=CH2 4-BnO-C6H4
Ph
NH2
3,4-(BnO)-C6H3
4-BnO-C6H4
CH2CO2H
of 2b for a-(1!3)-GalT and LgtC than for GTB. Significantly,
these results are in keeping with the selectivity profile ob-
served for 2b in the single-concentration screen, thus confirm-
ing the suitability of the experimental set-up chosen for the
library-screening experiment. Taken together, results from this
screen provide a proof-of-principle that this experimental set-
up allows the reliable discrimination between low- and high-af-
finity GalT binders. Our protocol therefore offers an extremely
rapid and simple method for the identification and target
profiling of novel, drug-like GalT inhibitors.
In summary, we have developed the new and broadly appli-
cable fluorophore 1d for the HTS of retaining GalTs. We have
demonstrated that 1d is suitable for the screening of several
different enzymes in parallel and allows the simultaneous
profiling of inhibitor candidates for both potency and selectivi-
ChemBioChem 2010, 11, 1392 – 1398
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1395

G. K. Wagner et al.
Fluorophore 1d was prepared by Suzuki–Miyaura cross-coupling of
5-iodo UDP-a-d-galactose and (5-formylthien-2-yl)boronic acid, as
previously described.[24] UDP-Gal derivatives 1a–c were prepared in
analogous fashion, and full synthetic details will be reported else-
where. All target compounds were purified by ion-pair and/or ion-
exchange chromatography, on Lichroprep RP-18 or Macro Prep
1
25Q resin, respectively, and characterised analytically by H, 13C and
31P NMR and HR ESI-MS (for the analytical characterisation of 1a–c
see the Supporting Information). Chemical shifts (d) are referenced
to methanol (dH 3.34, dC 49.50 for solutions in D2O).
1
5-(5-Formylthien-2-yl)-UDP-a-d-galactose (1d): H NMR (400 MHz,
D2O): d=1.27 (2.1 equiv of TEA, t, J=6.8 Hz), 3.19 (2.1 equiv of
TEA, q, J=6.8 Hz), 3.66–3.72 (m, 2H; H-6’’), 3.72–3.76 (m, 1H; H-2’’),
3.84 (dd, J=3.2, 10.2 Hz, 1H; H-3’’), 3.95 (d, J=3.0 Hz, 1H; H-4’’),
4.10–4.13 (m, 1H; H-5’’), 4.28–4.31 (m, 2H; H-5’), 4.32–4.34 (m, 1H;
H-4’), 4.40–4.48 (2t, J=5.1, 5.1 Hz, 2H; H-2’, H-3’), 5.62 (dd, J=3.4,
7.1 Hz, 1H; H-1’’), 6.04 (d, J=4.9 Hz, 1H; H-1’), 7.74 (d, J=4.2 Hz,
1H; Th), 8.01 (d, J=4.1 Hz, 1H; Th), 8.46 (s, 1H; H-6), 9.79 (s, 1H;
CHO); 13C NMR (125 MHz, D2O): d=9.0 (TEA), 47.5 (TEA), 61.7, 65.7
(d, JC,P =4.6 Hz), 69.0 (d, JC,P =6.7 Hz), 69.7, 70.0, 70.3, 72.6, 74.9,
84.3 (d, JC,P =7.3 Hz), 89.7, 96.4 (d, JC,P =5.4 Hz), 109.6, 126.0, 139.2,
140.3, 142.0, 144.8, 151.2, 163.5, 187.8; 31P NMR (121.5 MHz, D2O):
d=ꢀ11.2 (d, JP,P =22.5 Hz), ꢀ12.7 (d, JP,P =21.2 Hz); MS (ESI): m/z
C20H25N2O18P2S1: calcd 675.0304 [MꢀH]ꢀ, found: 675.0305.
Biochemistry: Proteins were expressed and purified as previously
described.[24,31] For donor kinetics, B. taurus a-(1!3)-GalT, UDP-Gal
or 5-substituted UDP-Gal derivatives 1a–d (0.6–400 mm), lactose
(2 mm) and MnCl2 (10 mm) in Tris/HCl buffer (50 mm, pH 7.5) were
incubated at 378C (total volume 100 mL, all concentrations are final
concentrations). Enzyme concentrations and reaction times were
chosen so as to avoid depletion of donor in excess of 10% (see
Table 4). After the appropriate time, the reactions were stopped by
cooling in dry ice, and samples were analysed immediately by
HPLC.
Figure 3. A) Screening of small molecular inhibitor candidates against three
different GalTs, using fluorophore 1d. The potency of the inhibitors is given
relative to that of UDP-Gal (indicated with a line). See Table 3 for inhibitor
structures 2a–l. B) Displacement of fluorophore 1d from a-(1!3)-GalT, GTB
and LgtC by thiazolidinone inhibitor 2b.
ty. This assay design significantly facilitates the identification of
novel GalT inhibitor chemotypes, for example, for antibacterial
drug discovery, and obviates the need for the time-consuming
evaluation of candidate molecules in separate GT bioassays.[29]
Moreover, the modular nature of this screening format allows,
in principle, the continuous addition of new enzymes to this
assay. These enzymes may include other retaining and, poten-
tially, inverting GalTs as well as other UDP-Gal-dependent en-
zymes, such as the epimerase GalE.[30] While not all of these en-
zymes might tolerate the additional fluorogenic substituent at
the uracil base of 1d as well as the GalTs used in this study,
this potential limitation can very likely be addressed by gener-
ating mutants of the proteins in question. Beyond carbohy-
drate-active and glycoprocessing enzymes, the general assay
principle described herein could also be applicable to other
proteins that use nucleotides or nucleotide conjugates as their
cofactor. Studies exploring the scope of the new fluorophore
and some of its derivatives for such applications are ongoing.
Table 4. GalT activities and incubation times for enzyme kinetics.
Cmpd
5-Substituent R
a-(1!3)-GalT [mU]
tinc [min]
UDP-Gal
1a
1b
1c
1d
H
0.16
3.2
3.2
3.2
8
5
30
60
60
10
phenyl
4-MeO-C6H4
2-furanyl
5-(2-formyl)thienyl
HPLC analyses were performed on a PerkinElmer Series 200 ma-
chine equipped with a column oven, a diode array detector and a
Supelcosil LC-18-T column (5 mm, 25 cmꢁ4.6 mm). Each sample
(injection volume 40 mL) was eluted at 308C, at a flow rate of
1.5 mLminꢀ1, with a gradient of methanol (2–15%) against phos-
phate buffer (0.5m, adjusted to pH 8 with triethylamine). The de-
pletion of donor (UDP-Gal, 1a–d) and the formation of nucleoside
diphosphate, the secondary product of the glycosylation reaction,
were monitored at 430 nm. Km and vmax values were determined by
fitting data points to a Michaelis–Menten curve (v=vmax ꢁSꢁ(Km +
S)ꢀ1) by using GraFit 5.0.10. To assess the hydrolytic stability of 1d,
two separate control experiments were carried out in the absence
of either 1) enzyme (to account for potential chemical hydrolysis)
or 2) acceptor (to account for potential enzymatic glycohydrolase
activity). No significant degree of hydrolysis was observed in these
experiments over a period of 24 h.
Experimental Section
Synthetic chemistry: All chemicals and reagents were obtained
commercially and used as received unless stated otherwise. Thiazo-
lidinone inhibitors 2a–l were prepared as previously described.[28]
1396
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1392 – 1398

Probe for Retaining Galactosyltransferases
Photophysical experiments
Thus, for all wells, fluorescence intensity after GalT addition
(second reading) was subtracted from fluorescence intensity
before GalT addition (first reading) to give DFI. For each GalT, the
maximum (no displacement of 1d, e.g., wells A1 and A2) and mini-
mum (displacement of 1d by UDP-Gal, e.g., wells A3 and A4)
changes in fluorescence (DFImax and DFImin) were calculated. The
change in fluorescence in the presence of individual thiazolidi-
nones (DFIT, wells C1–H12) was used to quantify the displacement
of 1d from each GalT, by each thiazolidinone, relative to the dis-
placement of 1d by UDP-Gal from the same enzyme, according to
the following Equation (1):
General: UV absorbance spectra were recorded on a PerkinElmer
Lambda 25 UV/Vis spectrometer at ambient temperature in FarUV
quartz cells (path length=1.0 cm). Fluorescence spectra were re-
corded on a PerkinElmer LS-45 spectrometer at ambient tempera-
ture in a quartz micro fluorescence cell (path length=1.0 cm).
Quantum yields: UDP-Gal derivatives 1a–d were serially diluted in
H2O (10, 20, 30, 40, and 50 mm for absorbance measurements, and
0.2, 0.4, 0.6, 0.8, 1 mm for fluorescence measurements), and UV ab-
sorbance and fluorescence emission (with lmax absorbance=lex
fluorescence) were recorded for all samples. To determine quantum
yields,[32] for each absorbance and fluorescence spectrum the area
under the curve (AUC) was calculated by numerical integration, ap-
plying the midpoint rule. For each compound, AUCabs and AUCfluo
were then plotted over compound concentration according to
AUCabs =Aꢁ[conc]+B and AUCfluo =A’ꢁ[conc]+B’. From these
linear plots, the gradients A and A’ were extracted, and for each
compound the specific quantum yield Fs, under these experimen-
tal conditions, was calculated as the ratio A’/A. Quantum yields de-
termined with this protocol for two reference compounds, 2-ami-
nopyridine (0.60) and l-tryptophan (0.14) were in exact agreement
with literature values.[32] The quantum yields for reference com-
pounds were used to calculate the general quantum yield Fg for
each compound 1a–d, according to Fg =Fref ꢁ(A’/A)/(A’/A)ref.
1ꢀDFIT
ð1Þ
% inhibition ¼
ꢂ 100 %
DFImaxꢀDFImin
Acknowledgements
We thank Durita Djurhuus, Dr. Dietlind Adlercreutz and Mette H.
Bien (Copenhagen) for the isolation of enzymes, and Sebastian
Gehrke, Adam Wiscombe and Ben Young (Norwich) for the syn-
thesis of the thiazolidinone compound library. The plasmid for
the N. meningitidis a-(1!4)-GalT was a generous gift from Dr.
Warren W. Wakarchuk (Ottawa). This work was supported by the
EPSRC (First Grant EP/D059186/1, to G.K.W.), the MRC (Discipline
Hopping Award G0701861, to G.K.W.), the Royal Society (research
grant to G.K.W.) and the Danish Natural Science Research Council
(to M.M.P.). G.K.W. is a Leverhulme Trust Research Fellow (RF/4/
RFG/2008/0544). We thank the EPSRC National Mass Spectrome-
try Service Centre, Swansea, for the recording of mass spectra.
Microplate assays: Fluorescence-intensity measurements with 1d
were carried out in black NUNC F96 MicroWell polystyrene plates
on a BMG LABTECH PolarStar microplate reader equipped with a
350ꢁ5 nm excitation filter and a 430ꢁ5 nm emission filter. The
number of flashes per well was set to 50, the gain to 2240, and the
position delay to 0.2 s. Prior to readings, the plates were incubated
for 10 min at 308C. Prior to fluorescence readings, shaking of the
microplate was performed in a double orbital for 10 s (shaking
width 4 mm). The gain adjustment for fluorescence readings was
performed on the entire microplate. Results were visualised with
BMG LABTECH data-analysis software Mars 1.10 and analysed with
GraphPad Prism 5.
Keywords: bioassays
· enzymes · fluorescent probes ·
glycobiology · glycosyltransferases
[1] J. T. Weadge, M. M. Palcic in Wiley Encyclopedia of Chemical Biology,
Vol. 2 (Ed.: T. P. Begley), Wiley, New York, 2009, pp. 198–211.
[2] L. L. Lairson, B. Henrissat, G. J. Davies, S. G. Withers, Annu. Rev. Biochem.
2008, 77, 521–555.
[3] B. Schuman, J. A. Alfaro, S. V. Evans, Top. Curr. Chem. 2007, 272, 217–
257.
[4] C. Breton, L. Snajdrova, C. Jeanneau, J. Koca, A. Imberty, Glycobiology
2006, 16, 29R–37R.
[5] D. Bowles, J. Isayenkova, E. K. Lim, B. Poppenberger, Curr. Opin. Plant.
Biol. 2005, 8, 254–263.
[6] J. D. Marth, P. K. Grewal, Nat. Rev. Immunol. 2008, 8, 874–887.
[7] D. H. Dube, C. R. Bertozzi, Nat. Rev. Drug. Discovery 2005, 4, 477–488.
[8] J. E. Rexach, P. M. Clark, L. C. Hsieh-Wilson, Nat. Chem. Biol. 2008, 4, 97–
106.
[9] M. M. Palcic, K. Sujino, Trends Glycosci. Glycotechnol. 2001, 13, 361–370.
[10] O. Von Ahsen, U. Voigtmann, M. Klotz, N. Nifantiev, A. Schottelius, A.
Ernst, B. Mꢂller-Tiemann, K. A. Parczyk, Anal. Biochem. 2008, 372, 96–
105.
Assay protocol: Samples were pipetted into the requisite wells of a
black NUNC 96-well plate as shown in Tables S1–S3. Key: B: Tris/
HCl buffer (50 mm, pH 7; 40 mL); F: fluorophore 1d (200 nm in Tris/
HCl buffer; 40 mL); M: MnCl2 (10 mm in Tris/HCl buffer; 80 mL); E:
galactosyltransferase (in Tris/HCl buffer; 40 mL); I: inhibitor (in Tris/
HCl buffer; 40 mL); total volume per well: 200 mL; all concentrations
are final concentrations per well. For the calculation of IC50 values,
requisite data points were fitted to a four-parameter curve with
GraphPad Prism 5. All experiments were carried out in triplicate,
unless indicated otherwise.
Protocol for library screening: Samples were pipetted into individual
wells of a black NUNC 96-well plate as shown in Table S4. Key: M:
MnCl2 (10 mm in Tris/HCl buffer; 80 mL); F: fluorophore 1d (200 nm
in Tris/HCl buffer; 40 mL); T1–T12: thiazolidinone inhibitors 2a–l
(50 mm in Tris/HCl buffer containing 5% DMSO; 40 mL); U: UDP-Gal
(5 mm in Tris/HCl buffer; 40 mL); total volume per well: 160 mL; all
concentrations are final concentrations per well. The microplate
was incubated for 10 min at 308C, and the fluorescence emission
was recorded (first reading). Tris/HCl buffer (B), a-(1!3)-GalT (E1),
LgtC (E2) or GTB (E3; 40 mL) was then added to the requisite wells,
as shown in Table S5. The microplate was incubated for another
10 min at 308C, and a second fluorescence emission reading was
taken (second reading). The relative increase in fluorescence from
first to second reading was attributed to the degree of binding of
1d at the individual GalT in the presence or absence of inhibitor.
[11] M. Yang, M. Brazier, R. Edwards, B. G. Davis, ChemBioChem 2005, 6, 346–
357.
[12] J. Wongkongkatep, Y. Miyahara, A. Ojida, I. Hamachi, Angew. Chem.
2006, 118, 681–684; Angew. Chem. Int. Ed. 2006, 45, 665–668.
[13] M. Persson, M. M. Palcic, Anal. Biochem. 2008, 378, 1–7.
[14] P. Gribbon, A. Sewing, Drug Discovery Today 2003, 8, 1035–1043.
[15] J. S. Helm, Y. Hu, L. Chen, B. Gross, S. Walker, J. Am. Chem. Soc. 2003,
125, 11168–11169.
[16] B. J. Gross, B. C. Kraybill, S. Walker, J. Am. Chem. Soc. 2005, 127, 14588–
14589.
[17] T. Hennet, Cell. Mol. Life Sci. 2002, 59, 1081–1095.
ChemBioChem 2010, 11, 1392 – 1398
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1397

G. K. Wagner et al.
[18] J. R. Brown, F. Yang, A. Sinha, B. Ramakrishnan, Y. Tor, P. K. Qasba, J. D.
Esko, J. Biol. Chem. 2009, 284, 4952–4959.
[19] K. Persson, H. D. Ly, M. Dieckelmann, W. W. Wakarchuk, S. G. Withers,
N. C. J. Strynadka, Nat. Struct. Biol. 2001, 8, 166–175.
[20] X. Qian, M. M. Palcic in Carbohydrates in Chemistry and Biology: Part 2:
Biology of Saccharides, Vol. 3 (Eds.: B. Ernst, G. Hart, P. Sinay¨), Wiley-VCH,
Weinheim, 2000, pp. 293–328.
[29] Y. Hu, J. S. Helm, L. Chen, C. Ginsberg, B. Gross, B. Kraybill, K. Tiyanont,
X. Fang, T. Wu, S. Walker, Chem. Biol. 2004, 11, 703–711.
[30] S. Bernatchez, C. M. Szymanski, N. Ishiyama, J. Li, H. C. Jarrell, P. C. Lau,
A. M. Berghuis, N. M. Young, W. W. Wakarchuk, J. Biol. Chem. 2005, 280,
4792–4802.
[31] K. Sujino, T. Uchiyama, O. Hindsgaul, N. O. L. Seto, W. W. Wakarchuk,
M. M. Palcic, J. Am. Chem. Soc. 2000, 122, 1261–1269.
[32] J. R. Lakowicz, Principles of Fluorescence Spectroscopy, 3rd ed., Springer,
Heidelberg, 2006.
[21] N. J. Greco, Y. Tor, Nat. Protocols 2007, 2, 305–316.
[22] T. Pesnot, G. K. Wagner, Org. Biomol. Chem. 2008, 6, 2884–2891.
[23] A. Collier, G. K. Wagner, Chem. Commun. 2008, 178–180.
[24] T. Pesnot, R. Jørgensen, M. M. Palcic, G. K. Wagner, Nat. Chem. Biol.
2010, 6, 321–323.
[33] J. A. Alfaro, R. B. Zheng, M. Persson, J. A. Letts, R. Polakowski, Y. Bai, S. N.
Borisova, N. O. L. Seto, T. L. Lowary, M. M. Palcic, S. V. Evans, J. Biol.
Chem. 2008, 283, 10097–10108.
[25] R. W. Sinkeldam, Y. Tor, Org. Biomol. Chem. 2007, 5, 2523–2528.
[26] Y. N. Zhang, G. J. Swaminathan, A. Deshpande, E. Boix, R. Natesh, Z. H.
Xie, K. R. Acharya, K. Brew, Biochemistry 2003, 42, 13512–13521.
[27] R. F. Steiner, E. P. Kirby, J. Phys. Chem. 1969, 73, 4130–4135.
[28] T. K. Smith, B. L. Young, H. Denton, D. L. Hughes, G. K. Wagner, Bioorg.
Med. Chem. Lett. 2009, 19, 1749–1752.
Received: January 7, 2010
Published online on June 8, 2010
1398
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1392 – 1398