Inhibition of O-GlcNAcase by PUGNAc is dependent upon the oxime stereochemistry

Article abstract of DOI:10.1016/j.bmc.2005.09.013

The potent O-GlcNAcase inhibitor PUGNAc was synthesized and two isomers based on the E and Z stereochemistry of the oxime moiety were separated, defined, and tested for activity. Several lines of evidence were examined in an effort to define the correct s

Full text of DOI:10.1016/j.bmc.2005.09.013

Bioorganic & Medicinal Chemistry 14 (2006) 837–846
Inhibition of O-GlcNAcase by PUGNAc is dependent upon
the oxime stereochemistry
Melissa Perreira,a Eun Ju Kim,b Craig J. Thomasa,* and John A. Hanoverb
aChemical Biology Core Facility, National Institute of Diabetes and Digestive and Kidney Disorders,
National Institutes of Health, Bethesda, MD 20892, USA
bLaboratory of Cell Biochemistry and Biology, National Institute of Diabetes and Digestive and Kidney Disorders,
National Institutes of Health, Bethesda, MD 20892, USA
Received 21 July 2005; revised 1 September 2005; accepted 2 September 2005
Available online 7 October 2005
Abstract—The potent O-GlcNAcase inhibitor PUGNAc was synthesized and two isomers based on the E and Z stereochemistry of
the oxime moiety were separated, defined, and tested for activity. Several lines of evidence were examined in an effort to define the
correct stereochemical assignments of each form of PUGNAc. The ability of the Z stereoisomer to undergo the Beckmann rear-
rangement was ultimately the most definitive proof. It was determined via both in vitro and intact cell experiments that the Z form
of PUGNAc was vastly more potent an inhibitor of O-GlcNAcase than the E form.
ꢀ 2005 Elsevier Ltd. All rights reserved.
1. Introduction
well-characterized inhibitor of O-GlcNAc transferase is
Alloxan (1) (Fig. 1).3 Additionally, there are few known
inhibitors of O-GlcNAcase including streptozotocin
(2) and O-(2-acetamido-2-deoxy-D-glucopyranosylid-
ene)amino-N-phenylcarbamate (PUGNAc) (3).4
Among the assorted post-translational modifications of
nuclear and cytoplasmic proteins are O-linked N-acetyl-
glucosamine (O-GlcNAc) addition and removal on ser-
ine and threonine residues by O-GlcNAc transferase
andO-GlcNAcase, respectively.1 The adornment of O-
GlcNAc upon protein surfaces has been suggested to
take part in a functional interplay involving both ki-
nases and phosphatases with implications on cellular
signaling and regulation.2 Disruptions of these multifac-
eted interactions likely have consequences in maladies
including diabetes mellitus and cellular events such as
apoptosis. Conversely, manipulation of these pathways
could provide a significant advantage within the man-
agement of numerous diseases.
PUGNAc was initially noted for its potent inhibition of
b-N-acetylglucosaminidases and it was suggested that
inhibition was the result of substrate mimicry.4a Com-
posed of a hydroximolactone N-acetyl triol moiety,
PUGNAc does indeed mimic the structure of native glu-
cosamine substrates such as UDP-GlcNAc (4). The abil-
ity of PUGNAc to mimic several glucosamine substrates
undoubtedly plays a role in its inhibition of O-GlcNAc-
ase and could be a useful reagent in delineating the phe-
notypical response to exaggerated levels of O-linked N-
acetylglucosamine modifications of proteins. Recently, it
has been reported that prolonged cellular exposure to
PUGNAc has resulted in the amplified incidence of O-
GlcNAc incorporation to proteins and an ensuing resis-
tance to insulin.5 While the elucidation of the biochem-
ical potential of PUGNAc continues to receive
attention, complete understanding of the role of the
oxime stereochemistry of PUGNAc has been largely ig-
nored. Herein we synthesize, isolate, and define each of
the PUGNAc isomers and show that the ability of
PUGNAc to inhibit the action of O-GlcNAcase is
dependent upon the stereochemistry of the oxime.
There are many potent and selective small molecule
inhibitors and activators for the assorted protein phos-
phatases and, in particular, kinases. The development
of selective activators or inhibitors of O-GlcNAcase
(EC 3.2.1.52) and O-GlcNAc transferase (EC 2.4.1.94)
remains minuscule by comparison. Currently, the only
Keywords: PUGNAc; O-GlcNAcase; Oxime stereochemistry.
*
Corresponding author. Tel.: +1 301 496 40761; fax: +1 301 402
0008; e-mail: craigt@niddk.nih.gov
0968-0896/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2005.09.013

838
M. Perreira et al. / Bioorg. Med. Chem. 14 (2006) 837–846
Figure 1. Structures of alloxane (1), streptozocin (2), PUGNAc (3), and UDP-GlcNAc (4).
2. Results
PUGNAc seemed important to determine. Following
HPLC separation of the two isomers of PUGNAc, we
set out to determine their absolute stereochemistry. We
initially designated the stereoisomers based upon their
HPLC elution times; for example, as the ‘‘F’’ or fast-
running isomer and the ‘‘S’’ or slow-running isomer.
Following several failed attempts to produce acceptable
crystals for X-ray analysis of either the ‘‘F’’ or ‘‘S’’ iso-
mers of PUGNAc, we concluded that the designation of
the absolute stereochemistry by other means would be
necessary. We explored several established lines of evi-
dence including melting point comparisons, compound
stability, and NMR spectral shifts with limited success
due to inconsistencies between oxime and hydroximo-
lactone structures and lack of precedence in the case
of hydroximolactones. Ultimately, the most straightfor-
ward line of evidence was the capability of the Z isomer
of PUGNAc to undergo the Beckmann rearrangement.
Finally, the ‘‘F’’ PUGNAc was designated as the Z
stereoisomer (3a) and the ‘‘S’’ PUGNAc was the E
stereoisomer (3b) (Fig. 2).
2.1. Chemistry
Synthetic elucidation of PUGNAc was accomplished
employing the original pathway developed by Vasella
and co-workers.6 Following the coupling of the hydrox-
imolactone (5) with phenyl isocyanate to form the corre-
sponding phenylcarbamoyl (6) (Scheme 1), we noted the
presence of two products by TLC and HPLC. The sep-
aration of each isomer was difficult by flash chromatog-
raphy, however, their isolation by HPLC proved
uncomplicated. The two isomers of 6a and 6b were
formed in approximately a 5:2 ratio based upon HPLC
integration and comparison of the NMR integration of
corresponding peaks from an aliquot of the principal
product mixture. The original product mixture of 6
was utilized in the subsequent deacetylation reaction
to produce 3a and 3b, and we found that the product ra-
tio was maintained based upon HPLC integration. Fol-
lowing purification of the two isomers 6a and 6b,
aliquots of each were treated with ammonia to effect
deacetylation and it was verified based upon HPLC
retention time alignment that the oxime stereochemistry
was unchanged during deacetylation. Previously, the
isolation of PUGNAc relied upon flash chromatography
and a subsequent crystallization to yield a single isomer.
PUGNAc has been generally referred to as the Z isomer,
although we are unaware of any crystallographic evi-
dence or NMR correlations that would allow exact
stereochemical definition.
Several inferences can be made based upon the experi-
mental evidence such as melting points and molecular
stability.7 However, these reports are often in disagree-
ment with one another based upon the specific structural
elements of the oxime and/or hydroximolactone. For
example, several studies of oximes and hydroximolac-
tones suggest that the E stereoisomer is more stable than
its Z counterparts,7b,d while others present evidence that
the Z isomers of oximes or hydroximolactones are the
more robust.7a,c,e,f In our hands, the Z isomer (3a) of
PUGNAc decomposed completely in a 0.1% trifluoro-
acetic acid aqueous solution after only 8 h, while the E
isomer (3b) remained intact up to and exceeding 7 days.
Given the plethora of data concerning the importance of
lock and key models for ligand–protein interactions, the
biochemical relationship between the two structures of
Scheme 1.

M. Perreira et al. / Bioorg. Med. Chem. 14 (2006) 837–846
839
de-shield the proton and cause a downfield shift in the
signals of the E-isomer.8 However, this effect is largely
dependent of the dihedral angle between the oxime oxy-
gen and the C2 methine proton (H–C–C@N–O–R), and
ambiguity based thereupon is plausible due to the as-
sumed psuedochair conformations of the hydroximolac-
tone. For example, studies have presented evidence of E
hydroximolactones with C2-H shifts upfield of their Z
counterparts.9 We were able to correctly identify the
1
C2 methine proton through H–1H correlation NMR
based upon coupling to the nitrogen proton of C2-
NHAc group. The recorded H signal for the C2-H car-
Figure 2. Structural dissimilarity of the E and Z isomers of PUGNAc.
1
bon of the Z isomer (3a) was seen at 4.36–4.38 ppm and
the corresponding signal of the E isomer (3b) was
recorded at 4.62–4.64 ppm (Fig. 3), which are in agree-
ment with the established trend. The chemical shifts
seen, however, in the peracetylated compounds 6a
(4.78–80 ppm) and 6b (4.71–4.73 ppm) are in conflict
with the conventional precedent. It should be noted that
the results for the deacetylated versions (3a, 3b) are in
alignment with several hydroximolactones presented
by Vasella and co-workers.7e Ultimately, ambiguity sur-
rounding the accurate prediction of the deshielding ef-
fect for these molecules and the discrepancies between
the recorded signals in this study and in other published
work pressed us to rely on evidence other than the C2-H
shifts.
Further, the Z isomer was seen to decompose in neutral
conditions (1:1 solution of H2O and CH3CN) over a
period of 5 weeks, while the E isomer, again, remained
whole. The same trend was noticed in the peracetylated
analogues (6a and 6b).
Melting point variations have been used as determinants
of stereochemistry as well. Several studies suggest that Z
oximes will generally have lower melting points than E
oximes.7a,c,e,f Again, there are studies suggesting just
the opposite.7b,e Researchers point to several explana-
tions and have often invoked the use of melting point
difference to assign the otherwise ambiguous stereo-
chemistry of oximes. Our determination of the melting
points of the E isomer (3b) was lower (115–119 ꢁC) than
that of the Z isomer (3a) (181–185 ꢁC).
The shifting of the C1 signal within the 13C spectrum of
oximes can also be descriptive. The pattern is based
upon steric compression and the observed trend suggests
that within the more sterically compressed oxime the C1
signal will be shifted upfield.10 However, Sivertsen and
co-workers show just the opposite for a small group of
hydroximolactones.7b The E and Z assignments of
numerous hydroximolactones have since been made
Inference can also be made based upon the crystalliza-
tion timescale for each isomer. Following HPLC separa-
tion of the two stereoisomers of PUGNAc, we effected
crystallization in the same manner as previously pub-
lished studies involving only the active form of PUG-
NAc. In our hands, the Z isomer (3a) was noted to
fully crystallize in 90 min. The E isomer (3b) took
approximately 18 h to form crystals. This is in alignment
with our supposition that all previous reports of PUG-
NAc involved exclusively Z-PUGNAc.
1
based solely on this study and the use of the H C2-H
spectral shift.7e,11 We observed that the C1 signals for
the E form of PUGNAc (3b) (162.6 ppm) and the per-
acetylated PUGNAc (6b) (158.3 ppm) were shifted
downfield relative to their Z counterparts (3a, 6a)
(158.8 ppm and 154.1 ppm, respectively) (Fig. 3). These
observations are in alignment with the work of Siversten
and co-workers but in direct conflict with the observa-
tion that the sterically compressed hydroximolactone
will result in the upfield shift of C1.
While these data are descriptive, ambiguity and dis-
agreement are prevalent and physical trends such as
melting points that are based upon the stereochemistry
of the oxime will often be superseded due to the nature
of the substrate as a whole. Thus, reliance on physical
properties to ascertain the structure of these compounds
is inadequate and often misleading. The far more com-
pelling evidences lie within the numerous studies sug-
gesting strong trends in the NMR spectral shifts of
selected carbon and proton atoms within the structures
of hydroximolactones and oximes. It is unfortunate that
among the limited crystallographic data only one form
(often the Z stereoisomer) of any given hydroximolac-
tone structure exists so no direct correlation can be
made between the absolute structure and significant
NMR signals.7e
Each of the aforementioned lines of evidence is sus-
pect due to inconsistencies within the literature. How-
ever, our survey of the literature suggested that there
is general agreement for the trends maintained for
the C2 13C NMR chemical shifts of hydroximolactone
and oxime structures. Repeated studies show that the
13C signals of the a-syn carbon (C2) of oximes akin to
the E stereoisomer of PUGNAc have a significant up-
field shift relative to the a-anti-counterparts of the Z
stereoisomer (Fig. 2). This is based upon steric com-
pression (also called the c effect) causing an upfield
shift of the 13C nuclei in close proximity to a bulky
moiety; in this case, the N-phenylcarbamate. This re-
sult was first described by Roberts and co-workers
and Nelson and co-workers,12 and numerous studies
since have provided good precedence for this effect.13
The 1H chemical shifts of the C2 methine protons for the
E and Z isomers of PUGNAc have notable deviations in
their chemical shifts. Precedence in similar oxime and
hydroximolactone systems suggests that proximity to
the oxygen of the oxime in the a-syn configuration will

840
M. Perreira et al. / Bioorg. Med. Chem. 14 (2006) 837–846
Figure 3. Selected NMR data; comparison of 13C signals of C2 and comparison of 1H signals of C2-H for E and Z PUGNAc, and E and Z (OAc)3
PUGNAc.
In our study, the correct assignment of the C2 carbon
line of evidence, however, is not precedence based upon
NMR data (mostly based upon oxime precedence), but
is the capability of only the Z isomer to undergo the Beck-
mann rearrangement (Fig. 4).14 It was initially noted that
HPLC purification of the undefined mixture of the ‘‘F’’
and ‘‘S’’ isomers of PUGNAc was complicated when, in
a 0.1% trifluoroacetic acid aqueous buffer, the ‘‘F’’ isomer
degraded over a period of 6–8 h. Following purification at
neutral conditions, the ‘‘F’’ isomer was reevaluated within
the acidic buffer conditions and the degradation products
observed via LCMS. Two products were noted; one with a
strong UV adsorption that maintained a mass of 136.2
and a second product with no associated chromaphore
that showed a mass of 220.1 (Fig. 5). These low resolution
masses and their associated UV/vis signatures corre-
1
signals was accomplished by H–13C correlation NMR
based upon coupling to the methine C2 proton. The
recorded 13C signal for the C2 carbon of the E isomer
(3b) was seen at 57.4 ppm and the corresponding sig-
nal of the Z isomer (3a) was recorded at 51.8 ppm
(Fig. 3). This trend was also seen in the peracetylated
compounds 6b (53.1 ppm) and 6a (47.1 ppm). Again,
the 13C shifts for C2 of the E and Z PUGNAc and
the peracetylated E and Z PUGNAc are not in agree-
ment with accepted precedence.
Based solely on the alignment of the NMR data with
accepted trends it would be difficult to unambiguously as-
sign the E and Z forms of PUGNAc. The most compelling
Figure 4. Description of the Beckmann rearrangement for Z PUGNAc into 7, 8, and 9.

M. Perreira et al. / Bioorg. Med. Chem. 14 (2006) 837–846
841
Figure 5. LCMS analysis of the rearrangement products of Z PUGNAc in 0.1% TFA-buffered water within a linear gradient of increasing amounts
of acetonitrile.
sponded well with the assumed products of the Beckmann
rearrangement 7 (m/z = 219.1) and 8 (m/z = 137.1).
ed by the lack of fluorophore to allow diode array
detection. A general collection based upon assumed
retention times allowed for the collection of a compound
with the appropriate low resolution mass and proton
NMR signals for the non-hydrated product 7. Observa-
tion of the hydrated product (9) via mass analysis was
not seen. It should be noted that the proton NMR we
have observed could well represent products other than
7 including products formed via abnormal Beckmann
rearrangements. We do not believe, however, that the
existence of other Beckmann rearrangement type prod-
ucts alters our assignment of the stereochemistry of
PUGNAc.
Following this observation, we tested the generality of
rearrangement by exposing both isomers to several con-
centrations of trifluoroacetic acid. It was observed that
the ‘‘S’’ isomer was stable up to and including exposure
to a 50% trifluoroacetic acid solution. Conversely, the
‘‘F’’ isomer broke down at all trifluoroacetic acid con-
centrations tested. We next analyzed the products of
the presumed Beckmann rearrangement via high resolu-
tion mass analysis and proton NMR. Following treat-
ment of 8 mg of the ‘‘F’’ isomer with 25%
trifluoroacetic acid in methanol, the resulting products
were purified by HPLC. Analysis of the two products
collected confirmed the existence of an aromatic com-
pound with appropriate mass and proton NMR signals.
The collection and analysis of the presumed sugar
byproduct (7 or 9), observed as the non-hydrated prod-
uct 7 via low resolution LCMS analysis, was complicat-
With this evidence in hand, we returned to the original
characterization of PUGNAc by Vasella and co-workers
to compare the NMR signatures of both the Z and E
PUGNAc (3a and 3b), and the Z and E acetylated ver-
1
sions (6a and 6b). It was confirmed that the H and 13C
signals for both 3a and 6a corresponded well with the

842
M. Perreira et al. / Bioorg. Med. Chem. 14 (2006) 837–846
published signals for PUGNAc. The E PUGNAc ana-
logues 3b and 6b, however, had several significant devi-
ations. Most notable were the observed signals for the
C5-H multiplet of 6b, which was shifted from 4.65–
4.67 to 5.25–5.28, and the C6-H multiplet of 3b that
shifted from 3.76–3.78 to 4.53–4.55.
PUGNAc (6a and 6b) (from a stock ethanol solution
diluted to a final concentration of 50 lM) followed by
incubation at 37 ꢁC for 48 h and lysis of each culture
and a control were followed by an analysis of O-Glc-
NAc integration via Western blot analysis. The peracet-
ylated forms were utilized to achieve a more efficient
cellular incorporation of each reagent and the ethanol
treatment was mimicked within the control culture.
The level of O-GlcNAc incorporation was quantified
via infrared imaging of an IR-labeled O-GlcNAc specific
antibody.
The characterization of the Beckmann rearrangement
products of 3a and good alignment of the NMR signals
from our study to that of Vesella and co-workers have
allowed us to unambiguously assign the ‘‘F’’ isomers
with the Z stereochemical configuration.
The Z form of the peracetlyated PUGNAc (6a) was
noted to induce an approximately 1.4-fold increased
level of O-GlcNAc incorporation upon proteins within
the HeLa cell systems and similar results were found
for HEK cells (Fig. 7 [gel B: lane 4 and gel D: lane
4]). The E isomer (6b), however, was seen to have
no effect on O-GlcNAc levels (Fig. 7 [gel B: lane 3
and gel D: lane 3]). The experiments were repeated
and the 1.4-fold increase was observed for a second
time with the Z form, while the E form, again, pro-
duced no increased incidence of O-GlcNAc
incorporation.
2.2. Biological evaluation
With a satisfactory assignment of the E and Z stereoiso-
mers of PUGNAc complete, we concerned ourselves
with the biochemical activities of each isomer. Both 3a
and 3b were examined in the presence of recombinant
O-GlcNAcase and ÔcagedÕ fluorogenic substrate, fluores-
cein di-N-acetyl-b-D-glucosaminide (FDGlcNAc) 10
(Fig. 6). We recently reported the action of O-GlcNA-
case on 10 and the measurement of O-GlcNAcase activ-
ity based upon direct fluorescence measurement.15 The
level of inhibition was determined based upon the quan-
tification of fluorescence measured in the absence and
presence of the PUGNAc isomers (intensity of fluores-
3. Discussion
cence was measured at
k
kex = 485 nm and at
em = 535 nm) (Fig. 6). The Z isomer of PUGNAc (3a)
The ability of the Z form of PUGNAc to more potently
inhibit the action of O-GlcNAcase relative to the E form
is a significant finding. Consideration of the similarities
of the PUGNAc structure and the native substrate
would strongly advocate a binding mechanism whereby
the hydroximolactone N-acetyl triol moiety provides a
significant portion of ligand recognition based upon
substrate mimicry. Given that this moiety remains un-
changed in both the E and Z forms of PUGNAc, we
may well envision a mechanism of enzyme inhibition
by Z PUGNAc that is moderately tenuous.
was seen to strongly inhibit the action of O-GlcNAcase
(approx 20% of O-GlcNAcase native activity in the pres-
ence of 1 lM of 3a), while the E isomer of PUGNAc
(3b) had a marked decrease in inhibition (approx 90%
of O-GlcNAcase native activity in the presence of
1 lM of 3b).
We further examined the abilities of both the E and Z
forms of PUGNAc to evoke an increased incidence of
O-GlcNAc incorporation upon proteins via methods
previously reported by Philipsberg and co-workers.16
Treatment of separate cultures of both HEK and HeLa
cells with aliquots of the peracetylated forms of E and Z
It is worth noting that PUGNAc has been viewed within
the crystallographic analysis of glycogen phosphorylase b
Figure 6. Structure of fluorescein-di-b-GlcNAc (10) and activity of O-GlcNAcase based upon fluorescence: (1) control—no inhibitor, (2) with 1.0 lM
Z-PUGNAc (3a), and (3) with 1.0 lM E-PUGNAc (3b).

M. Perreira et al. / Bioorg. Med. Chem. 14 (2006) 837–846
843
Figure 7. Effect of the oxime stereochemistry of PUGNAC on protein incorporation of O-GlcNAc. Ponceau S staining of the nitrocellulose
membranes (gels A and C) reveals equivalent proteins were loaded. Western blot proteins from HeLa (gel B) and HEK (gel D) indicate an increased
incidence of O-GlcNAc protein incorporation. Lane 1, protein marker; lanes 2–4, proteins from cells cultured for 48 h in the absence of inhibitor and
presence of 50 mM E-(OAc)3-PUGNAc (6b) and Z-(OAc)4-PUGNAc (6a), respectively. IR quantification revealed an approximately 1.4-fold
increase in O-GlcNAc incorporation upon proteins when both HeLa and HEK cells were treated with Z-(OAc)3-PUGNAc (6a) (gel B—lane 4 and
gel D—lane 4, respectively).
and was seen to exist as the Z stereoisomer.17 However, a
compound similar to PUGNAc (N,N0-diactyl-chit-
obionoxime-N-phenylcarbamate (HM508)) was recently
viewed within the crystallographic analysis of ChiB as
the E stereoisomer.18 The exact ligand–enzyme docking
arrangement for Z PUGNAc and O-GlcNAcase is worth
further exploration.
isomer binds in a manner that better positions the N-
phenylcarbamate moiety in terms of ligand–enzyme con-
tacts and steric accommodation.
Further, a recent report21 and congruent observations in
our own laboratory22 have noted the success of a Glc-
NAc-thiazoline as a potent inhibitor of O-GlcNAcase.
The success of this analogue as an inhibiting agent
would suggest that a substrate assisted mechanism
(Fig. 8) is being utilized by O-GlcNAcase and that
obstruction of such an enzymatic mechanism is a key
necessity for small molecule inhibitors like PUGNAc.23
The positioning of the N-acetyl group and the reactivity
of anomeric ether linkage are both imperative to this
type of mechanism. The oxime component of both E
and Z PUGNAc would likely block any such intermedi-
ate from forming and any inhibition enhancement of the
Z form seems unlikely to be based upon this possible
substrate–enzyme interaction. The enhanced potency
of Z-PUGNAc might indicate that the positioning of
the N-acetyl group is a key binding motif based upon
the assumed steric clash of this moiety with the phenyl
carbamate in the E-PUGNAc form (Fig. 8). Clearly,
visualization of the active form of PUGNAc within
the active site of O-GlcNAcase would resolve many of
these questions and allow for refinement of Z-PUGNAc
into a more specific and potent enzyme inhibitor.
Structural activity relationships based solely on stereo-
chemistry are widespread throughout medicinal chemis-
try and in modern drug discovery.19 Recent reports,
including observations of the stereospecific antitumor
activity of a radicicol oxime analogue and the stereospe-
cific agonistic activity of oxime-piperidine derivatives at
the CCR5 receptor, demonstrate that oxime stereochem-
istry can be an important mediator of structure and
function.20
The assignment of stereochemistry within E and Z oxi-
mes and/or hydroximolactones can be deceptive. The
methods utilized to delineate between the two sterioi-
somers has evolved from simple qualitative analyses
to, in the best case scenario, crystallographic methods.
Herein, we describe how both qualitative methods and
NMR signal comparison can be misleading. Further,
the E and Z isomers of PUGNAc represent important
counter-examples to several of the relied upon NMR
methods to predict oxime stereochemistry.
The understanding that PUGNAc relies on the correct
oxime stereochemistry to potently inhibit O-GlcNAcase
compels us to reconsider of the means of inhibition by Z
PUGNAc and the mechanism of action of O-GlcNA-
case. Certainly, the fundamentally different shapes of
the two stereoisomers will have reverberations in the
binding ability of E and Z PUGNAc to O-GlcNAcase
and ultimately their inhibitory potential. Clearly, the Z
4. Conclusion
Herein, we report that the inhibition of O-GlcNAcase
by PUGNAc is highly dependent on the stereochemical
configuration of the oxime portion of the PUGNAc
chemical structure. Further, through experimental evi-
dence, primarily the observation that the Z stereoiso-
mers will undergo the Beckmann rearrangement, we

844
M. Perreira et al. / Bioorg. Med. Chem. 14 (2006) 837–846
Figure 8. Example of substrate assisted mechanism and structural comparison of E and Z PUGNAc.
have shown that the Z form of PUGNAc is the active
isomer. Importantly, only the Z form of PUGNAc
was shown to amplify the incorporation of O-GlcNAc
on proteins within both HeLa and HEK cell systems.
4.78–4.80 (m, 1H), 5.22–5.26 (m, 1H), 5.33–5.36 (m,
1H), 7.02–7.05 (m, 1H), 7.29–7.32 (m, 2H), 7.48–7.49
(m, 2H), 8.61 (d, JHH = 7.68 Hz, 1H), 9.74 (s, 1H); 13C
NMR (d6-DMSO): d 18.8, 18.1, 18.9, 20.9, 47.1, 60.0,
65.5, 69.3, 74.2, 117.3, 121.4, 127.2, 136.8, 149.8,
154.1, 167.5, 167.7, 168.0, 168.4; mass spectrometry
(TOF); m/z = 480.1603 (M+H) (theoretical 480.1613).
5. Experimental
5.1. General
5.3. O-(2-Acetamido-3,4,5-tri-O-acetyl-D-glucopyranosy-
lidene)amino-E-N-phenylcarbamoyl (6b)
The synthetic elaboration was done in an identical fash-
ion as reported in Ref. 6b. 1H NMR data were recorded
on a Varian Gemini300, and 13C NMR and COSY data
were recorded on a Brooker 600. Spectra were recorded
in d6-DMSO and/or D2O and were referenced to the
residual solvent peak at 2.50 and 4.79, respectively. Re-
verse-phase (C18) HPLC was carried out using an Agi-
lent HPLC with a ZorbaxTM SP-C18 semi-prep column.
All melting points were determined with a Thomas-
Hoover apparatus and are uncorrected. High-resolution
mass spectroscopy measurements were performed on a
Micromass/Waters LCT Premier Electrospray TOF
mass spectrometer.
Synthesis was accomplished in an identical fashion and
with equivalent yields as in Ref. 6b. Following purifica-
tion of the isomer mixture via flash chromatography, an
aliquot was removed for HPLC separation of the iso-
mers. HPLC purification was achieved using a linear
gradient of water containing increasing amounts of
CH3CN (0 ! 15 min, linear gradient from 30% to 60%
CH3CN at a flow rate of 4 mL/min: Rt 10.2 min). The
triacetyl-E-PUGNAc (6b) was obtained as a white solid
1
following lyophilization. H NMR (d6-DMSO): d 1.90
(s, 3H), 1.99 (s, 3H), 2.04 (s, 3H), 2.05 (s, 3H), 4.12–
4.15 (m, 1H), 4.54–4.57 (m, 1H), 4.71–4.73 (m, 1H),
5.14–5.17 (m, 1H), 5.25–5.28 (m, 1H), 5.34–5.35 (m,
1H), 7.02–7.04 (m, 1H), 7.28–7.31 (m, 2H), 7.47–7.48
(m, 2H), 8.90 (d, JHH = 7.56 Hz, 1H), 9.84 (s, 1H); 13C
NMR (d6-DMSO): d 18.9, 19.0, 20.7, 53.1, 60.4, 65.3,
72.2, 80.0, 117.6, 121.4, 127.2, 136.9, 149.8, 158.3,
167.3, 167.5, 167.7, 168.6; mass spectrometry (TOF);
m/z = 480.1617 (M+H) (theoretical 480.1613).
5.2. O-(2-Acetamido-3,4,5-tri-O-acetyl-D-glucopyranosy-
lidene)amino-Z-N-phenylcarbamoyl (6a)
Synthesis was accomplished in an identical fashion and
with equivalent yields as in Ref. 6b. Following purifica-
tion of the isomer mixture via flash chromatography, an
aliquot was removed for HPLC separation of the iso-
mers. HPLC purification was achieved using a linear
gradient of water containing increasing amounts of
CH3CN (0 ! 15 min, linear gradient from 30% to 60%
CH3CN at a flow rate of 4 mL/min: Rt 9.6 min). The tri-
acetyl-Z-PUGNAc (6a) was obtained as a white solid
5.4. O-(2-Acetamido-2-deoxy-D-glucopyranosylid-
ene)amino-Z-N-phenylcarbamoyl (3a)
Synthesis was accomplished in an identical fashion as in
Ref. 6b. Following removal of the solvent by rotary
evaporation, the crude product mixture was purified
by HPLC without further workup. HPLC purification
was achieved using a linear gradient of water containing
1
following lyophilization. H NMR (d6-DMSO): d 1.87
(s, 3H), 2.00 (s, 3H), 2.03 (s, 3H), 2.05 (s, 3H), 4.19–
4.21 (m, 1H), 4.37–4.40 (m, 1H), 4.65–4.67 (m, 1H),

M. Perreira et al. / Bioorg. Med. Chem. 14 (2006) 837–846
845
increasing amounts of CH3CN (0 ! 15 min, linear
gradient from 10% to 25% CH3CN at a flow rate of
4 mL/min: Rt 12.6 min). Z-PUGNAc (3a) was obtained
as a white solid following lyophilization. Crystallization
of a 20 mg aliquot of the sample was achieved in a
3.57 (m, 1H), 3.89–3.95 (m, 1H), 4.13–4.16 (m, 1H)
7.57–7.59 (s, 1H).
5.7. Expression and purification of recombinant
O-GlcNAcase
solution
of
methanol/ethyl
acetate/hexanes
(10 mL:30 mL:30 mL). Crystals formed in approximate-
ly 90 min. Mp 181–185 ꢁC. 1H NMR (d6-DMSO): d 1.89
(s, 3H), 3.56–3.69 (m, 3H), 3.76–3.78 (m, 1H), 3.93–3.95
(m, 1H), 4.36–4.38 (m, 1H), 4.94 (br s, 1H), 5.59 (br s,
2H), 7.01–7.03 (m, 1H), 7.27–7.30 (m, 2H), 7.48–7.49
(m, 2H), 8.33 (d, J = 7.98 Hz, 1H), 9.59 (s, 1H); 13C
NMR (D2O at 15 ꢁC): d 23.8, 51.8, 60.7, 69.2, 72.9,
83.1, 119.3, 123.5, 129.4, 139.2, 152.4, 158.8, 169.9; mass
spectrometry (TOF); m/z = 354.1289 (M+H) (theoretical
354.1296).
Cultures of TOP 10 cells containing pBAD/HisA human
O-GlcNAcase clone expression vector were grown over-
night in Luria–Bertani (LB) broth (Digene) supplement-
ed with ampicillin (50 lg/ml). Induction of O-
GlcNAcases was initiated by adding arabinose solution
(0.02%) when the OD600 of cultures reached around 0.5.
The cultures were further grown for 4 h at 30 ꢁC at
200 rpm. The cells were harvested by centrifugation at
4000g for 15 min in a Sorvallꢂ RC 5C Plus (DuPont)
centrifuge. The pellet was subjected to freeze–thaw and
suspended in 1/50 of the original volume in 0.1 mg/ml
lysozyme, 20 mM Tris–Hcl, pH 8.0, 1 mM DTT, 0.1%
Triton X-100, and an EDTA-free protease inhibitor
cocktail tablet (1 tablet/10 ml). The suspension was incu-
bated at room temperature for 5 min and the lysate was
sonicated on ice (4· 10 s, setting 3, Misonix Ultrasonic
Processor). The supernatant was obtained after centrifu-
gation at 20,000g for 15 min. Expressed His6-tagged O-
GlcNAcase in the supernatant was purified on His-Trap
HP column (Amersham Biosciences) using the slightly
modified manufacturerÕs protocol. Fractions (1 ml) col-
lected throughout the separation were assayed for pro-
tein (absorbance at 280 nm) and were further assayed
for O-GlcNAcase activity according to the protocol de-
scribed below. Fractions showing enzyme activity were
pooled and concentrated using a Centricon 100 micro-
concentrator (Amicon). Pierce BCA protein assay
reagent was used to estimate protein concentration.
Purity of O-GlcNAcase was confirmed by Ponceau S
staining of the membrane (Sigma) and Western blot
analysis using anti His-Tag antibody (Abcam).
5.5. O-(2-Acetamido-2-deoxy-D-glucopyranosylid-
ene)amino-E-N-phenylcarbamoyl (3b)
Synthesis was accomplished in an identical fashion as
in Ref. 6b. Following removal of the solvent by rotary
evaporation, the crude product mixture was purified
by HPLC without further workup. HPLC purification
was achieved using a linear gradient of water contain-
ing increasing amounts of CH3CN (0 ! 15 min, linear
gradient from 10% to 25% CH3CN at a flow rate of
4 mL/min: Rt 14.5 min). E-PUGNAc (3b) was ob-
tained as a white solid following lyophilization. Crys-
tallization of a 30 mg aliquot of the sample was
achieved in a solution of methanol/ethyl acetate/hex-
anes (10 mL:30 mL:30 mL). Crystals formed in
approximately 18 h. Mp 115–119 ꢁC. 1H NMR
(d6-DMSO): d 1.87 (s, 3H), 3.50–3.53 (m, 1H), 3.67–
3.69 (m, 1H), 3.83–3.86 (m, 1H), 4.11–4.12 (m, 1H),
4.53–4.55 (m, 1H), 4.62–4.64 (m, 1H), 4.69 (br s,
1H), 5.03 (br s, 1H), 5.92 (br s, 1H), 7.00–7.03 (m,
1H), 7.27–7.30 (m, 2H), 7.48–7.49 (m, 2H), 8.66 (d,
J = 7.74 Hz, 1H), 9.73 (br s, 1H); 13C NMR (d6-
DMSO): d 23.1, 57.4, 63.6, 68.8, 73.3, 86.0, 119.4,
123.5, 129.4, 139.2, 152.3, 162.6, 169.5; mass spec-
trometry (TOF); m/z = 354.1316 (M+H) (theoretical
354.1296).
5.8. O-GlcNAcaseÕs enzymatic assay
Four microliters of purified recombinant O-GlcNAcase
was incubated in 0.1 M citrate/phosphate buffer, pH
6.5, containing 30 lM of the fluorogenic substrate
(10), in a final volume of 100 ll in the presence of
1 lM of inhibitors (3a and 3b) and in the absence of
the inhibitor at 37 ꢁC for 30 min. The assays were termi-
nated by adding 900 ll of 0.5 M Na2CO3 solution. Two
hundred microliters of assay solution was transferred
into a 96-well plate and fluorescence was measured at
the excitation wavelength of 485 nm and at the emission
wavelength of 535 nm on the Wallac 1420 fluorometer.
5.6. Beckmann rearrangement product characterization
To a solution of methanol and trifluoroacetic acid (3 mL
MeOH:1 mL TFA) was added 3a (8 mg, 0.022 mmol)
and the resulting solution was allowed to stir for
10 min at room temperature. Solvent was removed un-
der reduced pressure and the reaction mixture was dis-
solved in 2 mL of a 1:1 mixture of acetonitrile and
water. HPLC purification was achieved using a linear
gradient of water containing increasing amounts of
CH3CN (0 ! 10 min, linear gradient from 10% to 80%
CH3CN at a flow rate of 4 mL/min). Time of retention
5.9. Cell culture and inhibition
HEK and HeLa cells were cultured in a DMEM (Gibco)
supplemented with 10% FBS (Invitrogen). Aliquots of
inhibitors (6a and 6b in a 50 lL of a stock in 100% eth-
anol) were delivered into 10 mL tissue culture flasks to
yield a final concentration of 50 lM of inhibitor and
the ethanol was evaporated. The cells were incubated
in humidified 95% air and 5% CO2 at 37 ꢁC for 48 h.
Cells were harvested and were pooled by centrifugation
(200g, 5 min). Cells were washed twice with PBS, pH 7.0
1
for 8 was 5.1 min. Compound 8 H NMR (d6-DMSO):
d 6.97–7.04 (m, 1H), 7.23–7.38 (m, 2H), 7.42–7.53 (m,
2H); mass spectrometry (TOF); m/z = 137.0444 (M+)
(theoretical 137.0476) and m/z = 179.0759 (M+CH3CN)
(theoretical 179.0820). Time of retention for 7 was
1
1.16 min. Compound 7 H NMR (d6-DMSO); d 1.86
(s, 3H), 3.22–3.24 (m, 1H), 3.41–3.44 (m, 1H), 3.55–

846
M. Perreira et al. / Bioorg. Med. Chem. 14 (2006) 837–846
7. (a) Johnson, J. E.; Springfield, J. R.; Hwang, J. S.; Hayes,
L. J.; Cunningham, W. C.; McClaugherty, D. L. J. Org.
Chem. 1971, 36, 284; (b) Kjaer, A.; Larson, I. K.;
Sivertsen, P. Acta Chem. Scand. Ser. B 1977, 31, 415; (c)
Bradshaw, J.; Webb, G. B. U.S. Patent # 3903113,
1975.(d) Aebischer, B. M.; Hanseen, H. W.; Vasella, A.
T.; Schweizer, W. B. J. Chem. Soc. Perkin Trans. 1 1982,
2139; (e) Beer, D.; Vasella, A. Helv. Chim. Acta 1985, 68,
2254; (f) Kim, H. Y.; Lantrip, D. A.; Fuchs, P. L. Org.
Lett. 2001, 3, 2137.
8. (a) Baji, H.; Flammang, M.; Kimny, T.; Gasquez, F.;
Compagnon, P. L.; Delcourt, A. Eur. J. Med. Chem. 1995,
30, 617; (b) Emami, S.; Shafiee, A. Heterocycles 2001, 55,
2059.
9. Uhlmann, P.; Harth, E.; Naughton, A. B.; Vasella, A.
Helv. Chim. Acta 1994, 77, 2335.
10. Gordon, M. S.; Sojka, S. A.; Krause, J. G. J. Org. Chem.
1984, 49, 97.
11. (a) Takahashi, Y.; Vasella, A. Helv. Chim. Acta 1992, 75,
1563; (b) Vasella, A.; Witzig, C.; Waldraff, C.; Uhlmann,
P.; Briner, K.; Bernet, B.; Panza, L.; Husi, R. Helv. Chim.
Acta 1993, 76, 2847.
12. (a) Levy, G. C.; Nelson, G. L. J. Am. Chem. Soc. 1972, 94,
4897; (b) Hawkes, G. E.; Herwig, K.; Roberts, J. D.
J. Org. Chem. 1974, 39, 1017.
(10 mL) and pelleted (200g, 5 min). A culture of control
cells was treated in the same manner as above, except
that aliquots of ethanol were delivered into the tissue
culture flask and the ethanol was evaporated and the
cultures contained no inhibitors.
5.10. Western blot analysis
The pellet was suspended in 0.8 mL lysis buffer containing
20 mM Tris–HCl, pH 8.0, 1 mM DTT, 0.1% Triton X-
100, and an EDTA-free protease inhibitor cocktail (1 tab-
let/10 mL). The suspension was incubated at room tem-
perature for 5 min and sonicated on ice (4· 10 s, setting
3, Misonix Ultrasonic Processor). The supernatant was
obtained after centrifugation at 20,000g for 15 min. Pierce
BCA protein assay reagent was used to estimate protein
concentration. Samples were loaded onto a 4–12% Nu-
PAGE pre-cast gel and proteins were separated by elec-
trophoresis and then transferred to a nitrocellulose
membrane for immunoblotting. The blot was probed with
O-GlcNAc specific antibody RL2 (mouse IgG1, Affinity
Bioreagents, Inc.) at 1:1000 dilution and infrared (IR)-la-
beled anti-mouse IgG Goat antibody (Rockland, Inc.).
The blot was scanned on the Odyssey Infrared Imager
(LI-COR Biosciences). Relative intensity of O-GlcNAc
level was quantitated using the Odyssay Infrared Imaging
System application software.
13. (a) Unterhalt, B.; Koehler, H. Archiv. Der Pharmazie
1978, 311, 366; (b) Khouri, N.; Usubillaga, A.; Delgado,
J. N. Spectrosc. Lett. 1993, 26, 859; (c) Sohar, P.; Bodor,
A.; Schwartz, R. Steroids 1999, 64, 246; (d) Afonin, A.
V.; Ushakov, I. A.; Tarasova, O. A.; Shmidt, E. Y.;
Mikhaleva, A. I.; Voronov, V. K. Russ. J. Org. Chem.
2000, 36, 1777; (e) Holzer, W.; Hahn, K.; Brehmer, T.;
Claramunt, R. M.; Perez-Torralba, M. Eur. J. Org.
Chem. 2003, 1209.
Acknowledgments
14. Nguyen, M. T.; Raspoet, G.; Vanquickenborne, L. G.
J. Am. Chem. Soc. 1997, 119, 2552.
15. Kim, E.-J.; Kang, D. O.; Love, D. C.; Hanover, J. A.
Biochemistry, submitted.
16. Haltiwanger, R. S.; Grove, K.; Philipsberg, G. A. J. Biol.
Chem. 1998, 273, 3611.
17. Barford, D.; Schwabe, J. W. R.; Oikonomakos, N. G.;
Acharya, K. R.; Hajdu, J.; Papageorgiou, A. C.; Martin,
J. L.; Knott, J. C. A.; Vasella, A.; Johnson, L. N.
Biochemistry 1988, 27, 6733.
The authors thank Kenneth A. Jacobson for helpful dis-
cussions during the preparation of this manuscript.
Appreciation is also expressed to Victor Livengood
and Sonya Hess for assistance in obtaining the HRMS
data. We gratefully acknowledge the support of the
NIH Intramural Research Program.
References and notes
18. Vaaje-Kolstad, G.; Vasella, A.; Peter, M. G.; Netter, C.;
Houston, D. R.; Westereng, B.; Synstad, B.; Eijsink, V. G.
H.; van Aalten, D. M. F. J. Biol. Chem. 2004, 279, 3612.
19. Agranat, I.; Craner, H.; Caldwell, J. Nat. Rev. Drug Disc.
2002, 1, 753.
1. (a) Comer, F. I.; Hart, G. W. J. Biol. Chem. 2000, 275,
29179; (b) Wells, L.; Vosseller, K.; Hart, G. W. Science
2001, 291, 2376; (c) Wells, L.; Hart, G. W. FEBS Lett.
2003, 546, 154.
20. (a) Soga, S.; Sharma, S. V.; Shiotsu, Y.; Shimizu, M.;
Tahara, H.; Yamaguchi, K.; Ikuina, Y.; Murakata, C.;
Tamaoki, T.; Kurebayashi, J.; Schulte, T. W.; Neckers, L.
M.; Akinaga, S. Cancer Chemother. Pharmacol. 2001, 48,
435; (b) Palani, A.; Shapiro, S.; Josien, H.; Bara, T.;
Clader, J. W.; Greenlee, W. J.; Cox, K.; Strizke, J. M.;
Baroudy, B. M. J. Med. Chem. 2002, 45, 3143.
21. Macauley, M. S.; Whitworth, G. E.; Debowski, A. W.;
Chin, D.; Vocadlo, E. J. J. Biol. Chem. 2005, 280,
25313.
22. Kim, E.-J.; Darout, E.; Thomas, C. J.; Knapp, S.;
Hanover, J. A. manuscript in preparation.
23. Tews, I.; Terwisscha van Scheltinga, A. C.; Perrakis, A.;
Wilson, K. S.; Dijkstra, B. W. J. Am. Chem. Soc. 1997,
119, 7954.
2. (a) Hanover, J. A. FASEB J. 2001, 15, 1865; (b) Zachara,
N. E.; Hart, G. W. Chem. Rev. 2002, 102, 431; (c) Slawson,
C.; Hart, G. W. Curr. Opin. Struct. Biol. 2003, 13, 631.
3. Konrad, R. J.; Zhang, F.; Hale, J. E.; Knierman, M. D.;
Becker, G. W.; Kudlow, J. E. Biochem. Biophys. Res.
Commun. 2002, 293, 207.
4. (a) Horsch, M.; Hoesch, L.; Vasella, A.; Rast, D. M. Eur.
J. Biochem. 1991, 197, 815; (b) Dong, D. L.-Y.; Hart, G.
W. J. Biol. Chem. 1994, 269, 19321; (c) Konrad, R. J.;
Mikolaenko, I.; Tolar, J. F.; Liu, K.; Kudlow, J. E.
Biochem. J. 2001, 356, 31.
5. Arias, E. B.; Kim, J.; Cartee, G. D. Diabetes 2004, 53, 921.
6. (a) Beer, D.; Maloisel, J.-L.; Rast, D. M.; Vasella, A. Helv.
Chim. Acta 1990, 73, 1918; (b) Mohan, H.; Vasella, A.
Helv. Chim. Acta 2000, 83, 114.