Mechanistic insight into enzymatic glycosyl transfer with retention of configuration through analysis of glycomimetic inhibitors

Article abstract of DOI:10.1002/anie.200905096

(Figure Presented) Structural "valid"-ation: The mechanism of enzyme-catalyzed glycosyl transfer with retention of anomeric configuration continues to baffle, a situation compounded by the lack of insightful 3-D structures of ternary enzyme complexes. Syn

Full text of DOI:10.1002/anie.200905096

Communications
DOI: 10.1002/anie.200905096
Glycosyltransferase Inhibitors
Mechanistic Insight into Enzymatic Glycosyl Transfer with Retention
of Configuration through Analysis of Glycomimetic Inhibitors**
James C. Errey, Seung Seo Lee, Robert P. Gibson, Carlos Martinez Fleites, Conor S. Barry,
Pierre M. J. Jung, Anthony C. OꢀSullivan, Benjamin G. Davis,* and Gideon J. Davies*
One of the fundamental challenges in the field of glycobiol-
ogy remains the dissection of the catalytic mechanism of
nucleotide-sugar dependent glycosyltransferases (GTs), espe-
cially those that act with retention of anomeric configuration.
For a class of enzymes that contains > 20000 putative
members in over 90 families it is remarkable that our
understanding of the mechanistic origin of the notably high
stereoselectivity of retaining GTs is so poor; to date only
glimpses of trapped pseudo- (off pathway) intermediates,[1]
mutational analyses,[2] or unexpected inhibitor configura-
tions[3] have given hints. None unambiguously support the
origin of net retention being either through double inversion
(SN2 ꢀ 2) or internal return (SNi) mechanisms. Theoretical
analyses[4] have shown that both pathways are accessible and
that the SNi pathway provides a lower energy manifold as
result of active-site geometrical constraints.
synthesis and screening of pseudo-disaccharide inhibitors of
the trehalose-6-phosphate synthase, a classical retaining
glycosyltransferase of the “GT-B” fold. Enzyme kinetics in
the absence and presence of different inhibitors revealed one
compound with an affinity similar to that of the substrates
that was subsequently used to obtain a ternary complex with
this synergistic inhibitor.
The disaccharide trehalose (1, a-d-glucopyranosyl-a-d-
glucopyranoside; Scheme 1) and its 6-phosphate (2) are non-
reducing a,a-1,1 disaccharides with considerable importance
A major barrier to the study of these enzymes is the lack
of non-hydrolysable substrates or inhibitors, with affinities
similar to that of the substrates, which would thus allow
structural access to the ternary complex or something that
resembles it. Glycosyltransferase inhibitors are rare; rarer
still are those which do not harness portions of the nucleotide
donor.[5,6] Only one bisubstrate analogue of a retaining GT
has previously been described;[3] surprising inhibition profiles
were observed, inconsistent with a double-displacement
mechanism. Partially as a consequence of this lack of suitable
compounds, there are no 3-D structures of intact ternary
complexes with which to describe the catalytic centre and
reveal geometry of the transfer process. Here we describe the
Scheme 1. Trehalose (1), trehalose-6-phosphate (T6P, 2, the product of
OtsA), and comparative pseudodisaccharide mimetics based on the a-
glucopyranosyl moiety. Numbering shown is that of trehalose to aid
comparison.
in nature.[7,8] Given their absence in mammalian biology,
trehalose synthesising and processing enzymes offer attrac-
tive inhibition targets. For example, trehalase is the target of
the commercial agricultural fungicide validamycin,[9] and
trehalose-6-phosphate synthase was chosen by Bayer scien-
tists as a target for developing a new fungicide.[10] Trehalose is
also central to many bacterial and fungal strategies to
overcome environmental stress, and it is a major constituent
of the unusual glycolipids of mycobacteria, including Myco-
bacterium tuberculosis.[11] Most success has been achieved
with inhibitors of the hydrolytic trehalase enzyme. Com-
pounds such as validoxylamine A (3) have proven to be tight-
binding and structurally informative[12] trehalase inhibitors by
virtue, in part, of the resemblance of the pseudodisaccharide
scaffold of 3, for example, to 1.
[*] Dr. R. P. Gibson, Dr. C. Martinez Fleites, Prof. G. J. Davies
Department of Chemistry, The University of York
Heslington, York, YO10 5YW (UK)
Fax: (+44)1904-328-266
E-mail: davies@ysbl.york.ac.uk
Homepage: http://www.york.ac.uk/res/daviesgroup/index.html
Dr. J. C. Errey, Dr. S. S. Lee, Dr. C. S. Barry, Prof. B. G. Davis
Department of Chemistry, University of Oxford, Chemistry Research
Laboratory, Mansfield Road, Oxford, OX1 3TA (UK)
E-mail: ben.davis@chem.ox.ac.uk
Homepage: http://www.chem.ox.ac.uk/researchguide/
bgdavis.html
The predominant[13] pathway for the biosynthesis of
trehalose involves the initial formation of trehalose-6-phos-
phate (2) through the action of a glycosyltransferase, fre-
quently termed OtsA, found in family GT20[14] of the CAZY
(www.cazy.org) classification. This enzyme, which acts with
net retention of anomeric configuration catalyses the forma-
tion of the a,a-1,1 linkage with UDP-glucose (UDP = uridyl-
diphosphate) as the donor and Glc-6-P as the acceptor.[15] In
contrast to the success with trehalase inhibition, there has
been conspicuously less success in finding chemical tools to
similarly probe the active centre of OtsA.
Dr. P. M. J. Jung, Dr. A. C. O’Sullivan
Syngenta AG, WST-820.1.09, 4332 Stein (Switzerland)
[**] We thank Dr. M Yang for technical assistance. This work was funded
by the BBSRC (BB/E004350/1, BB/D006112/1) and EPSRC (EP/
E000614/1). G.J.D. and B.G.D. are Royal Society–Wolfson Research
Merit Award recipients.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200905096.
1234
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 1234 –1237

Angewandte
Chemie
The trehalose-6-phosphate synthase, OtsA, was the first
retaining nucleotide-sugar dependent glycosyltransferase
structure solved of the “GT-B” fold (one of two glycosyl-
transferase topologies). The structure of E. coli OtsA features
a fold similar to that seen for the PLKP-dependent enzyme
glycogen phosphorylase,[16] with N-terminal acceptor and C-
terminal UDP-Glc donor domains that act, with conforma-
tional flexibility, to achieve the synthesis of trehalose-6-
phosphate (2).[15,17] Structures have subsequently been solved
in complex with Glc-6-P and UDP and as the binary complex
with a UDP-2FGlc donor but, as with most glycosyltransfer-
ase structural analysis, no ternary complex exists.
In order to generate potential disaccharide mimetics that
would allow access to these first such ternary complexes, we
prepared compounds 4–7 (Scheme 2, see Supporting Infor-
mation for full details). These were based on the previously
successful scaffold mimetic validoxylamine 3 and utilized
features of transition state mimicry suitable for inhibition:[4]
1) bisubstrate mimicry; 2) flattening of transferred sugar
mimic (modelling of the SNi TS[4] has suggested near-cyclo-
hexenyl C5-O5-C1 and C2-C1-O5 bond angles of 121.28 and
123.78, respectively); 3) modulated pKa; 4) functionality to
prevent processing (heteroatom replacement) and to engage
putative phosphate binding pockets (charged phosphate or
uncharged tetrahedral sulfamate at O-6’ or 4’). Starting from
3, the pseudo-symmetrical cyclohexyl and cyclohexenyl
moieties were initially differentiated using regioselective
benzylidene acetal formation[18] at OH-4’,6’ and peracetyla-
tion to yield 8. Regioselective access to OH-4’ and OH-6’ was
then achieved in two ways. Firstly, OH-4’-free heptaacetate 10
was synthesized from 8 in four steps and 46% overall yield
through appropriate protecting-group manipulation that took
advantage of concomitant acid-catalyzed regioselective acetyl
migration.[19] Secondly, OH-6’-free heptabenzyl 9 was synthe-
sized from 8 in six steps and 52% overall yield through acetal
hydrolysis, regioselective tritylation, global deacetylation–
benzylation and final detritylation. Final access from 9 and 10
to target compounds 4–7 was then achieved through appro-
priate phosphorylation or sulfamoylation and global ammi-
nolytic or Birch deprotection, respectively. The low reactivity
of the central, pseudo-glycosidic secondary amine rendered
protection of NH-1 unnecessary.[18,20]
OtsA “pseudo-single substrate” kinetics were performed
using an assay in which UDP release was coupled through
pyruvate kinase and lactate dehydrogenase to NADH for-
mation. Under these conditions, OtsA yielded kcat = 34 Æ
1 sÀ1, and KM values of 1.7 Æ 0.3 mm (UDP-Glc) and of 7.3 Æ
0.6 mm (Glc-6-P). An initial evaluation revealed that com-
pounds 4, 5, and 7 (Table 1) all inhibited E. coli OtsAwith IC50
Table 1: IC50 values for compounds 4–7.
R1
R2
IC50 [mm]
2À
4
5
6
7
PO3
H
H
5.3Æ1.4
17Æ3.5
n/d[a]
28Æ8
SO2NH2
2À
H
H
PO3
SO2NH2
[a] Very slow onset of inhibition precluded IC50 determination; estimated
as >100 mm.
values between 5 and 28 mm. Consistent with intended
structural analogy, better inhibition was observed for the 6’-
phosphomimetics 4 and 5 than for 7. Compound 6 showed
only weak inhibition with slow onset.
Although less potent in numerical terms than other GT
inhibitors, the 5 mm IC50 values obtained for 4 were respect-
able when compared to the KM values of OtsA for its natural
substrates. Initial screens (IC50) also suggested inhibition by
UDP and putative synergistic inhibition and prompted more
detailed investigation of the inhibitory modes of OtsA using
both 4 and UDP. Double reciprocal analyses (see Supporting
Information, Figures S3 to S5) revealed that 4 alone com-
petitively inhibited UDP-Glc (Ki = 1.3 Æ 0.2 mm) and non-
competitively inhibited Glc-6-P (Ki = 4.2 Æ 0.2 mm). UDP
alone competitively inhibited only UDP-Glc (Ki = 140 Æ
10 mm) but not Glc-6-P. Together these suggested an ordered
“bi-bi mechanism”[22] with UDP-Glc binding first followed by
Glc-6-P and the effectiveness of 4 as a bisubstrate inhibitor
likely operating through a direct inhibitory equilibrium
between E and E·4.[21] Furthermore, addition of NDP
(UDP) showed synergistic enhancement (Figure 1) of the
inhibition of 4—approximately 100-fold improvement in the
IC50 value, to yield an IC50 of 41 mm at 0.15 mm UDP (a
concentration around UDPꢁs own Ki). It should be noted that
in other than the leading examples of a-2,6-sialyltransferase
inhibition[23,24] (which use NMP sugars), good inhibition of
GTs has only been shown with a bisubstrate analogue towards
Scheme 2. a) PhCH(OMe)2, DMF, TsOH, 39%; b) py, Ac2O, 76%;
c) AcOH (70% aq.), 608C, 79%; d) TrtCl, py, 408C, 97%; e) Ac2O, py,
96%; f) AcOH (80% aq.), 63%; g) iPr2NP(OBn)2, tetrazole, then
MCPBA, CH2Cl2, 82%; h) TMSBr, iPr2NEt, CH2Cl2, 95%; i) NH3,
MeOH, 47%; j) H2NSO2Cl, 98%; k) NH3, MeOH, 30%; l) NH3,
MeOH, 94%; m) BnBr, Bu4NI, NaH, 83%; n) AcOH (70% aq.), 90%;
o) iPr2NP(OBn)2, tetrazole, then MCPBA, CH2Cl2, 82%; p) Li, NH3,
58%; q) H2NSO2Cl, 99%; r) Li, NH3, 77%. DMF=dimethylformamide,
Ts =p-toluenesulfonyl, py=pyridine, Trt=trityl, Bn=benzyl, MCPBA=
meta-chloroperbenzoic acid, TMS=trimethylsilyl.
Angew. Chem. Int. Ed. 2010, 49, 1234 –1237
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1235

Communications
Figure 1. Relative OtsA activity at varying concentrations of 4 and
~
*
*
UDP. [UDP]=0 ( ), 0.05 (&), 0.1 ( ), 0.15 mm ( ).
Scheme 3. Diagram (partial) of the interactions of UDP and pseudo-
disaccharide 4 in the active centre of OtsA.
the inhibition of inverting a1,3-fucosyltransferase.[25] Such
inhibition similarly required the synergistic presence of
appropriate NDP (GDP). Such synergistic inhibition in the
presence of NDP is, of course, of reduced potential clinical
relevance since NDP concentration is not only uncontrollable
in vivo but often unknown. However, here, in screening for a
putative TS-like ligand this binding was highly encouraging.
Indeed, the inhibition, in the absence of UDP, shown here
by 4 of OtsA is highly reminiscent of the weak binding of the
inhibitor acarbose to glycogen and maltodextrin phosphor-
ylases which also allowed access to the key ternary complex of
that enzyme.[16] The structure of OtsA was therefore deter-
mined (statistics in Table S1) in a ternary complex with UDP
and bisubstrate mimic 4. The structure, solved at 2.2 ꢂ
resolution, provides unambiguous electron density for both
UDP and the pseudo-disaccharide 4.
The complex between OtsA and 4 illuminates the
geometry present around the anomeric centre (Figure 2 and
Scheme 3). This is especially important when formulating
reaction mechanisms consistent with glycosyltransfer with net
retention of anomeric configuration. These mechanisms
remain very poorly understood, not least because a mecha-
nism involving a double displacement with a covalent
intermediate has little experimental support forcing enzymol-
ogists to consider “front face” mechanisms such as those akin
to internal return “SNi” mechanisms[26] (discussed extensively
in Ref. [27]). Recently, Goedl and Nidetzky have provided
convincing evidence in a change in kinetic and chemical
mechanism, from a double displacement to a “front side”
mechanism in a sucrose phosphorylase (GH13) variant.[28]
A “front face” mechanism poses many problems. In
particular, what is the catalytic geometry and what acts as the
catalytic base to deprotonate the acceptor for nucleophilic
attack at the donor anomeric carbon (in the direction of
synthesis)? Intact nucleotide sugar glycosyltransferase ter-
nary complexes are exceedingly rare and where they have
been achieved, notably with the UDP-2FGal/3-deoxylactose
complex of the GT-A fold enzyme LgtC,[2] the geometry of
the crucial phosphorous-C1-acceptor is undefined.
The 3-D structure of OtsAwith 4 shows that the glucose-6-
P mimicking portion binds essentially identically as observed
in the Glc-6-P/UDP complex previously[15] (Figure 2). The
flattened cyclohexenyl moiety of 4 binds in an approximate E3
conformation. Of particular interest is the 2.8 ꢂ H-bond (13,
bond c; Scheme 4) between the donor phosphate oxygen O3B
(leaving group oxygen in the direction of synthesis and
nucleophile in the reverse direction) to the “glycosidic” NH
of the pseudo-disaccharide (Figure 2). This interaction shows,
we believe for the first time on a UDP-sugar transferase of
this kind, that the leaving-group phosphate oxygen indeed
contacts the acceptor nucleophile (here the NH of the
disaccharide inhibitor) in a manner consistent with the
phosphate acting as a base to deprotonate the acceptor in
the “internal return like” mechanism (Scheme 4), as semi-
nally invoked by Sinnott and Jencks for the solvolysis of
glycosyl fluoride by trifluoroethanol[26] and as predicted by
modelling of retaining GTs (12, bond c, Scheme 4).[4] Indeed,
the LgtC ternary structure is also completely consistent with
such a mechanism if the absent 3-OH is simply modelled in its
expected position. The more selective, concerted “internal-
Figure 2. Observed electron density (2FobsÀFcalcd, contoured at 1s, in
divergent “wall eyed” stereo) for the ternary complex of OtsA with
UDP (only a portion of which is shown) and validoxylamine (4). 4 is
shown in yellow with Glc-6-P (cyan bonds) and UDP-2FGlc (orange
bonds) from previous structure determinations overlaid for compar-
ison.
1236
www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 1234 –1237

Angewandte
Chemie
provided in the protein data base (pdb no. 2WTX). Additional
synthetic details are given in the Supporting Information.
Received: September 11, 2009
Published online: January 14, 2010
Keywords: bisubstrate mimetic · enzymes ·
.
glycosyltransferases · internal return · trehaloses
[1] L. L. Lairson, C. P. C. Chiu, H. D. Ly, S. He, W. W. Wakarchuk,
N. C. J. Strynadka, S. G. Withers, J. Biol. Chem. 2004, 279, 28339.
[2] K. Persson, H. D. Ly, M. Dieckelmann, W. W. Wakarchuk, S. G.
Withers, N. C. J. Strynadka, Nat. Struct. Mol. Biol. 2001, 8, 166.
[3] B. Waldscheck, M. Streiff, W. Notz, W. Kinzy, R. R. Schmidt,
Angew. Chem. 2001, 113, 4120; Angew. Chem. Int. Ed. 2001, 40,
4007.
[4] I. Tvaroska, ACS Symp Ser. 2006, 930, 285.
[5] P. Compain, O. R. Martin, Bioorg. Med. Chem. 2001, 9, 3077.
[6] K.-H. Jung, R. R. Schmidt in Carbohydrate-based Drug Discov-
ery (Ed.: C.-H. Wong), Wiley-VCH, Weinheim, 2005, p. 609.
[7] J. C. Arguelles, Arch. Microbiol. 2000, 174, 217.
[8] A. D. Elbein, Y. T. Pan, I. Pastuszak, D. Carroll, Glycobiology
2003, 13, 17R.
[9] N. Asano, T. Yamaguchi, Y. Kameda, K. Matsui, J. Antibiot.
1987, 40, 526.
[10] F. Spaltmann, M. Blunck, K. Ziegelbauer, Drug Discovery Today
1999, 4, 17.
[11] D. Chatterjee, Curr. Opin. Chem. Biol. 1997, 1, 579.
[12] R. P. Gibson, T. M. Gloster, S. Roberts, R. A. J. Warren, I. Storch
de Gracia, A. Garcꢃa, J. L. Chiara, G. J. Davies, Angew. Chem.
2007, 119, 4193; Angew. Chem. Int. Ed. 2007, 46, 4115.
[13] K. A. L. De Smet, A. Weston, I. N. Brown, D. B. Young, B. D.
Robertson, Microbiology 2000, 146, 199.
[14] P. Coutinho, E. Deleury, G. J. Davies, B. Henrissat, J. Mol. Biol.
2003, 328, 307.
[15] R. P. Gibson, J. P. Turkenburg, R. M. Lloyd, S. J. Charnock, G. J.
Davies, Chem. Biol. 2002, 9, 1337.
[16] M. OꢁReilly, K. A. Watson, L. N. Johnson, Biochemistry 1999, 38,
5337.
[17] R. Gibson, C. A. Tarling, S. Roberts, S. G. Withers, G. J. Davies,
J. Biol. Chem. 2004, 279, 1950.
Scheme 4. An “SNi-like” mechanism for retaining glucosyltransfer. The
boxed ternary species 12 and 13 equate to that proposed for SNi and
modelled by Tvaroska[4] and that seen experimentally in the OtsA
complex with UDP and 4, respectively.
return-like”[26] mechanism[29] is likely additionally favoured in
a desolvated, less polar active centre, and involves the
deprotonation of the incoming nucleophile by the departing
phosphate oxygen.[30] The extent and timing of the making
and breaking of bonds a–d in 12 (Scheme 4) is unclear and
extremes as represented by 14 may be possible. Nonetheless,
structure 13 determined here for the first time gives direct
evidence of a bond geometry consistent with a TS such as 11.
[18] S. Ogawa, T. Nose, T. Ogawa, T. Toyokuni, Y. Iwasawa, T. Suami,
J. Chem. Soc. Perkin Trans. 1 1985, 2369.
[19] T. Horrobin, C. H. Tran, D. Crout, J. Chem. Soc. Perkin Trans. 1
1998, 1069.
[20] Y. Miyamoto, S. Ogawa, J. Chem. Soc. Perkin Trans. 1 1991, 2121.
[21] M. Yu, M. L. B. Magalhaes, P. F. Cook, J. S. Blanchard, Bio-
chemistry 2006, 45, 14788.
[22] W. W. Clelend, The Enzymes, Vol. 2, 3rd ed. (Ed.: P. D. Boyer),
Academic Press, New York, 1970.
[23] B. Mꢄller, C. Schaub, R. R. Schmidt, Angew. Chem. 1998, 110,
3021; Angew. Chem. Int. Ed. 1998, 37, 2893.
[24] R. Schwꢅrer, R. R. Schmidt, J. Am. Chem. Soc. 2002, 124, 1632.
[25] L. Qiao, B. W. Murray, M. Shimazaki, J. Schultz, C.-H. Wong, J.
Am. Chem. Soc. 1996, 118, 7653.
[26] M. L. Sinnott, W. Jencks, J. Am. Chem. Soc. 1980, 102, 2026.
[27] L. Lairson, B. Henrissat, G. J. Davies, S. G. Withers, Annu. Rev.
Biochem. 2008, 77, 521.
[28] C. Goedl, B. Nidetzky, ChemBioChem 2009, 10, 2333.
[29] W. A. Cowdrey, E. D. Hughes, C. K. Ingold, S. Masterman, A. D.
Scott, J. Chem. Soc. 1937, 1252.
[30] K. Okamoto, K. Takeuchi, T. Inoue, J. Chem. Soc. Perkin Trans. 2
1980, 842.
[31] Collaborative Computational Project Number 4, Acta Crystal-
logr. Sect. D 1994, 50, 760.
Experimental Section
Trehalose-6-phosphate synthase activity was assayed spectrophoto-
metrically by coupling the formation of UDP to the reactions of
pyruvate kinase, and lactate dehydrogenase inhibition experiments
were carried out at varying concentrations of both substrates and
whilst varying the concentration of inhibitors. Initial velocity kinetic
data were then fitted using GRAFIT5 to allow full double reciprocal
analyses, including Ki determinations or OtsA activity and inhibition
by 4 and UDP. Rate measurement when UDP was used as a sole
inhibitor or a synergistic inhibitor determined OtsA activity by
coupling the formation of UDP to the reactions of pyruvate kinase
and lactate dehydrogenase (PK/LDH) in a background/product-
calibrated stopped assay format (see Supporting Information for
further details). Structures of OtsA were crystallised from protein at
16 mgmLÀ1 in a buffer of 20 mm TrisHCl pH 8 and 200 mm NaCl with
10 mm UDP and 5 mm 5. Data, to 2.2 ꢂ resolution were collected at
100 K at the European Synchrotron Radiation Facility. Structures
were refined using the CCP4 suite.[31] Data and structure quality
statistics are given in Table S1, other experimental information is
Angew. Chem. Int. Ed. 2010, 49, 1234 –1237
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1237