Discovery of O-GlcNAc transferase inhibitors

Full text of DOI:10.1021/ja0555217

Published on Web 09/29/2005
Discovery of O-GlcNAc Transferase Inhibitors
Benjamin J. Gross,‡ Brian C. Kraybill,† and Suzanne Walker*,†
Department of Microbiology and Molecular Genetics, HarVard Medical School, 200 Longwood AVenue, Boston,
Massachusetts 02115, and Department of Chemistry and Chemical Biology, HarVard UniVersity, 12 Oxford Street,
Cambridge, Massachusetts 02138
Received August 12, 2005; E-mail: suzanne_walker@hms.harvard.edu
Many nuclear and cytosolic proteins are transiently glycosylated
by an enzyme known as O-GlcNAc transferase (OGT), which
transfers N-acetylglucosamine from UDP-GlcNAc to selected serine
and threonine residues. O-GlcNAcylation affects such diverse
cellular processes as transcription, translation, organelle targeting,
and protein-protein interactions,1 and is believed to play a role in
a variety of signaling cascades that mediate glucose homeostasis
and stress responses.2 Specific inhibitors of OGT could be valuable
tools to probe the biological functions of O-GlcNAcylation, but
the inability to obtain significant quantities of enzyme, combined
with the lack of a high-throughput assay, has impeded efforts to
identify such compounds.3 We have developed conditions to express
large quantities of the catalytic domain of active OGT for the first
Figure 1. (a) OGT constructs expressed in E. coli. Blue columns represent
R-helices, two of which comprise a TPR. ncOGT and sOGT are identical
to known splice variants; ∆mOGT is 50 residues shorter than mOGT. (b)
time, and we report a high-throughput donor displacement assay
for the enzyme along with the discovery of a set of small-molecule
inhibitors. This work lays the foundation for both structural and
PAGE of these three constructs after IMAC purification.
functional analysis of the catalytic domain of OGT.
The human, rat, and mouse ogt genes have previously been
expressed in baculovirus,4 several mammalian cell lines,5,6 and
Escherichia coli,7 but good expression levels were not achieved.
Therefore, our first goal was to develop conditions to produce large
amounts of pure, active OGT. We chose to focus our efforts on
expression in E. coli because the potential to obtain large amounts
of protein is greater than in eukaryotic expression systems. OGT
is a bipartite protein consisting of a C-terminal glycosyltransferase
(Gtf) domain and an N-terminal protein-protein interaction domain
comprised of 12 tetratricopeptide (TPR) repeats. To optimize ex-
pression, we synthesized the gene for human OGT using preferred
E. coli codons, and then made constructs based on three known
splice variants of OGT (ncOGT, mOGT, and sOGT; Figure 1).8
These constructs were cloned into a modified pET-24b (Novagen)
vector for expression as C-terminal His8 fusions. Good expression
for all constructs could be achieved by late log induction in BL21-
(DE3) at low temperature (16 °C; OD ) 1.2; 0.4 mM IPTG), but
soluble sOGT was expressed at much higher levels than the other
proteins. We obtained 10-12 mg of sOGT/L of culture in >95%
purity after a single step Ni2+-IDA IMAC purification. Activity
was evaluated using a known peptide substrate. The Km of UDP-
GlcNAc was found to be 6.7 ( 0.5 µM, nearly identical to the
value reported for a construct of the rat enzyme containing six
TPRs.4 The specific activity, however, is over 150-fold higher than
that reported for the rat enzyme (161 nmol min-1 mg-1 vs 1.06
nmol min-1 mg-1), which may reflect differences in phosphorylation
or glycosylation states between OGT produced in E. coli and insect
cells.4,9 Nup62, a known ncOGT substrate, is also glycosylated by
sOGT. We note that this is the first report of sOGT activity in
vitro.
assays in which fluorescence polarization (FP) is monitored are
being used increasingly for high-throughput screening (HTS)
because they are technically simple to implement if an appropriate
ligand can be identified. Furthermore, they result in hits that are
biased toward compounds that bind in the same location as the
fluorescent probe, which can simplify the analysis of structure-
activity relationships.10-12 We have previously developed a fluo-
rescent UDP-GlcNAc displacement assay for a Gtf involved in
peptidoglycan biosynthesis (MurG),10 and we wondered whether a
similar assay could be used to screen OGT. If so, we wanted to
know whether the hits obtained would be selective for OGT relative
to MurG.
There is no structure of the Gtf domain of OGT to guide the
design of a fluorescent UDP-GlcNAc analogue, but it has been
proposed that OGT is structurally related to MurG.13 Therefore,
we evaluated whether the fluorescent probe used to screen MurG
(1, Figure 2) could also be used in an OGT screen.14,15 The FP of
a 50 nM solution of 1 in the presence of increasing amounts of
sOGT did not change significantly until high concentrations of
protein were added. Hypothesizing that the short linker between
the sugar and the fluorophore interfered with binding to the enzyme,
we prepared two additional fluorescent UDP-GlcNAc analogues
containing longer linkers (2 and 3, Figure 2). We observed
significant FP changes for both 2 and 3 in the presence of sOGT,
but the change for 3 was larger over a wider range of protein
concentrations. We selected this compound as our probe. From the
change in polarization as a function of sOGT concentration, we
calculated a dissociation constant of 1.3 ( 0.1 µM for 3. Addition
of unlabeled UDP-GlcNAc or UDP to a pre-equilibrated mixture
of sOGT and 3 resulted in a decrease in polarization, and both
compounds completely displaced 3 from sOGT at high concentra-
tions. Dissociation constants were calculated from the displacement
curves and found to be 1.5 ( 0.4 µM for UDP-GlcNAc, which
implies that the fluorophore on 3 does not affect binding, and 0.8
The ability to obtain large amounts of protein allowed us to
investigate possible high-throughput screens. Ligand displacement
† Harvard Medical School.
‡ Harvard University.
9
14588
J. AM. CHEM. SOC. 2005, 127, 14588-14589
10.1021/ja0555217 CCC: $30.25 © 2005 American Chemical Society

C O M M U N I C A T I O N S
Figure 2. Donor analogue displacement probes used in this study.
( 0.3 µM for UDP, which agrees well with previously reported
values.16
Figure 3. Validated OGT inhibitors found in this HTS.
The above experiments established the feasibility of a donor
displacement assay for HTS of OGT, and we adapted the assay to
a 384-well microplate format and screened 64 416 commercial
library compounds at the Institute of Chemistry and Cell Biology
(ICCB) at Harvard Medical School. The libraries were screened in
duplicate at a final concentration of 25 µg/mL using a Perkin-Elmer
Envision microplate reader. Included in the compounds screened
were 12 390 molecules that had previously been screened against
MurG.10 This subset contained 58% of the hits that were identified
in the MurG screen. Each plate contained a positive control well
containing sOGT, 3, and 1 mM UDP-GlcNAc and a negative
control well containing sOGT and 3. Compounds that reproducibly
caused a significant decrease in FP without a corresponding change
in fluorescence intensity were scored as hits. Using this criterion,
102 compounds were scored as positives, for a hit rate of 0.2%.
The positive compounds were then evaluated for OGT inhibition
using a radiometric assay that involves monitoring transfer of 14C-
GlcNAc to an OGT acceptor peptide containing an N-terminal
(Lys)3 tag that enables capture on phosphocellulose filter disks.4
Nineteen of these 102 compounds inhibited sOGT >40% at 25
µM. These molecules do not share obvious common structural
features. However, this may be a consequence of library diversity
since fewer than five compounds with the same core are present in
the screened libraries for almost all of the 19 inhibitors. IC50 values
were determined for several compounds, and the mode of inhibition
was determined for two of the best; both were found to be
competitive with respect to UDP-GlcNAc (see Supporting Informa-
tion). All of the compounds examined also inhibited the full-length
construct, ncOGT.
Remarkably, none of compounds that were identified as hits in
the MurG screen, which was based on displacement of UDP-
GlcNAc analogue 1, were found to displace UDP-GlcNAc analogue
3 from OGT. Furthermore, none of the OGT inhibitors identified
in this screen were found to inhibit MurG. Thus, there is no overlap
in the compounds selected in the two high-throughput screens, even
though both screens were based on displacement of the same
glycosyl donor, UDP-GlcNAc, and led to the discovery of
compounds that compete with this donor. We conclude that there
are substantial differences in the binding pockets for UDP-GlcNAc
in these enzymes that can be exploited to develop specific inhibitors.
The ability to use the same screening strategy against different Gtfs
could have clear advantages for the rapid discovery of orthogonal
inhibitors for enzymes that use similar substrates.
be useful tools for probing the biological functions of OGT. In the
meantime, the ability to obtain large quantities of the catalytic
domain of OGT enables structural analysis of this biologically
important enzyme.
Acknowledgment. This work was supported by NIH grant
AI44854 and funds from the Camille and Henry Dreyfus Founda-
tion. We thank Dr. Caroline Shamu and the staff at the ICCB,
Harvard Medical School, for their assistance. We are also grateful
to Kittichoat Tiyanont for synthetic expertise.
Supporting Information Available: OGT expression, purification,
and kinetic characterization of ncOGT and sOGT with peptide and
protein substrates; synthetic scheme for 2 and 3; experimental details
for primary and secondary screening; assay conditions; IC50 values and
inhibition pattern for 6. This material is available free of charge via
the Internet at http://pubs.acs.org.
References
(1) Zachara, N. E.; Hart, G. W. Chem. ReV. 2002, 102, 431-8.
(2) Zachara, N. E.; Hart, G. W. Biochim. Biophys. Acta 2004, 1673 (1-2),
13-28.
(3) Alloxan inhibits OGT with IC50 ) 100 µM: Konrad, R. J.; Zhang, F.;
Hale, J. E.; Knierman, M. D.; Becker, G. W.; Kudlow, J. E. Biochem.
Biophys. Res. Commun. 2002, 293, 207-12.
(4) The peptide acceptor substrate, KKKYPGGSTPVSSANMM, is derived
from a sequence described by Hart and co-workers: Kreppel, L. K.; Hart,
G. W. J. Biol. Chem. 1999, 274, 32015-22.
(5) Marshall, S.; Duong, T.; Wu, T.; Hering, M. A.; Yada, J.; Higgins, S.;
Orbus, R. J.; Yan, Z. H.; Rumberger, J. M. Anal. Biochem. 2003, 319,
304-13.
(6) Yang, X.; Su, K.; Roos, M. D.; Chang, Q.; Paterson, A. J.; Kudlow, J. E.
Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 6611-6.
(7) Lubas, W. A.; Hanover, J. A. J. Biol. Chem. 2000, 275, 10983-8.
(8) Hanover, J. A.; Yu, S.; Lubas, W. B.; Shin, S. H.; Ragano-Caracciola,
M.; Kochran, J.; Love, D. C. Arch. Biochem. Biophys. 2003, 409, 287-
97.
(9) Tai, H. C.; Khidekel, N.; Ficarro, S. B.; Peters, E. C.; Hsieh-Wilson, L.
C. J. Am. Chem. Soc. 2004, 126, 10500-1.
(10) Helm, J. S.; Hu, Y.; Chen, L.; Gross, B.; Walker, S. J. Am. Chem. Soc.
2003, 125, 11168-9.
(11) Hu, Y.; Helm, J. S.; Chen, L.; Ginsberg, C.; Gross, B.; Kraybill, B.;
Tiyanont, K.; Fang, X.; Wu, T.; Walker, S. Chem. Biol. 2004, 11, 703-
11.
(12) Soltero-Higgin, M.; Carlson, E. E.; Phillips, J. H.; Kiessling, L. L. J. Am.
Chem. Soc. 2004, 126, 10532-3.
(13) Wrabl, J. O.; Grishin, N. V. J. Mol. Biol. 2001, 314, 365-74.
(14) Pohl and co-workers have reported a route to rapidly access such probes:
Mizanur, R. M.; Jaipuri, F. A.; Pohl, N. L. J. Am. Chem. Soc. 2005, 127,
836-7.
(15) Bertozzi and co-workers have shown that OGT will accept substitutions
at the C2 amine of UDP-GlcNAc: Vocadlo, D. J.; Hang, H. C.; Kim, E.
J.; Hanover, J. A.; Bertozzi, C. R. Proc. Natl. Acad. Sci. U.S.A. 2003,
100, 9116-21.
(16) Haltiwanger, R. S.; Blomberg, M. A.; Hart, G. W. J. Biol. Chem. 1992,
267, 9005-13.
We are currently investigating the effects of these compounds
in cell culture. If they reduce O-GlcNAcylation in cells, they could
JA0555217
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14589