Synthesis and glycosidase inhibitory activity of 5- thioglucopyranosylamines. Molecular modeling of complexes with glucoamylase

Article abstract of DOI:10.1016/S0008-6215(99)00177-9

The synthesis of a series of 5-thio-D-glucopyranosylarylamines by reaction of 5-thio-D-glucopyranose pentaacetate with the corresponding arylamine and mercuric chloride catalyst is reported. The products were obtained as anomeric mixtures of the tetraacet

Full text of DOI:10.1016/S0008-6215(99)00177-9

www.elsevier.nl/locate/carres
Carbohydrate Research 321 (1999) 143–156
Synthesis and glycosidase inhibitory activity of
5-thioglucopyranosylamines. Molecular modeling of
complexes with glucoamylase
Karla D. Randell a, Torben P. Frandsen b,1, Bjarne Stoffer c, Margaret A. Johnson a,
Birte Svensson b, B. Mario Pinto a,*
a Department of Chemistry and Institute of Molecular Biology and Biochemistry, Simon Fraser Uni6ersity, Burnaby,
BC, Canada V5A 1S6
b Department of Chemistry, Carlsberg Laboratory, Gamle Carlsberg Vej 10, DK-2500 Valby, Denmark
c Department of Medicinal Chemistry, Royal Danish School of Pharmacy, Uni6ersitetsparken 2,
DK-2100 Copenhagen, Denmark
Received 1 September 1998; revised 10 May 1999; accepted 3 June 1999
Abstract
The synthesis of a series of 5-thio-D-glucopyranosylarylamines by reaction of 5-thio-D-glucopyranose pentaacetate
with the corresponding arylamine and mercuric chloride catalyst is reported. The products were obtained as anomeric
mixtures of the tetraacetates which can be separated and crystallized. The tetraacetates were deprotected to give a/b
mixtures of the parent compounds which were evaluated as inhibitors of the hydrolysis of maltose by glucoamylase
G2 (GA). A transferred NOE NMR experiment with an a/b mixture of 7 in the presence of GA showed that only
the a isomer is bound by the enzyme. The Ki values, calculated on the basis of specific binding of the a isomers, are
0.47 mM for p-methoxy-N-phenyl-5-thio-
nosylamine (8), 0.27 mM for p-nitro-N-phenyl-5-thio-
N-phenyl-5-thio- -glucopyranosylamine (10), and the Km values for the substrates maltose and p-nitrophenyl
a- -glucopyranoside are 1.2 and 3.7 mM, respectively. Methyl 4-amino-4-deoxy-4-N-(5%-thio-a- -glucopyranosyl)-a-
-glucopyranoside (11) is a competitive inhibitor of GA wild-type (Ki 4 mM) and the active site mutant Trp120“Phe
D
-glucopyranosylamine (7), 0.78 mM for N-phenyl-5-thio-
D-glucopyra-
D
-glucopyranosylamine (9) and 0.87 mM for p-trifluoromethyl-
D
D
D
D
GA (Ki 0.12 mM). Compounds 7, 8, and 11 are also competitive inhibitors of a-glucosidase from brewer’s yeast, with
Ki values of 1.05 mM, \10 mM, and 0.5 mM, respectively. Molecular modeling of the inhibitors in the catalytic site
of GA was used to probe the ligand–enzyme complementary interactions and to offer insight into the differences in
inhibitory potencies of the ligands. © 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Glucoamylase; a-Glucosidase; 5-Thio-
NMR
D
-glucopyranosylamines; Inhibitors; Molecular modeling; Transferred NOE
1. Introduction
Glucoamylase (GA) (1,4-a- -glucan gluco-
D
hydrolase, EC 3.2.1.3) from Aspergillus niger
* Corresponding author. Tel.: +1-604-291-4327; fax: +1-
604-291-3765.
E-mail address: bpinto@sfu.ca (B.M. Pinto)
1 Present address: Novo-Nordisk, DK-2880 Bagsvaerd,
Denmark.
is an exo-acting and inverting carbohydrase,
catalyzing the release of b- -glucopyranose
D
from non-reducing ends of starch and related
poly- and oligosaccharides. GA tolerates a
0008-6215/99/$ - see front matter © 1999 Elsevier Science Ltd. All rights reserved.
PII: S0008-6215(99)00177-9

144
K.D. Randell et al. / Carbohydrate Research 321 (1999) 143–156
rently unknown; also, details concerning
residues involved in catalysis and substrate
binding are lacking.
broad range of aglycon moieties, cleaving a-
1,6-glucosidic bonds and aryl glucosides with
500- and 50-fold lower activity, respectively,
than a-(1“4)-linked oligosaccharides [1–3].
GA from A. niger is secreted as two glycosy-
lated forms [4]: G1 containing a catalytic do-
main, a highly O-glycosylated region, and a
starch-binding domain (82.7 kDa, residues 1–
616) [5]; and G2 a smaller form lacking the
starch-binding domain (72 kDa, residues 1–
512) [6]. Crystallographic structure determina-
tion of the 94% identical GA from Aspergillus
awamori var. X100 shows that the catalytic
domain (residues 1–471) folds into an (a/a)6-
barrel, where six highly conserved a“a loop
regions connect segments of an inner and an
outer a6 barrel [7]. These conserved segments,
whose architecture creates the funnel-shaped
active site and details of residues involved in
catalysis, substrate binding, and transition-
state stabilization, are described by crystallog-
raphy in several GAꢀinhibitor complexes
[8–10], using site-directed mutagenesis [2,11–
15], inhibitor binding thermodynamics [16,17],
and by molecular recognition of substrate
analogs [12,18–20].
Glucosidases, inverting and retaining, cata-
lyze reactions proceeding through transition
states having substantial oxo-carbenium ion
character [23–25]. Thus, the various classes of
enzymes are efficiently inhibited by substrate
analogs mimicking either the charge and/or
the shape of the transition state; thus glucosi-
dases are often inhibited by analogs having a
half-chair conformation or by sugars bearing
a basic nitrogen adjacent to the C-1% atom of
disaccharides or the C-1 atom of monosac-
charides [26,27]. Electrostatic interactions be-
tween protonated inhibitors and negatively
charged side-chains in the active site cleft thus
might contribute significantly to the mecha-
nism and strengths of inhibition. We have
recently reported that heteroanalogs of
methyl maltosides containing sulfur and sele-
nium were inhibitors of A. niger with Ki val-
ues in the millimolar range [28], whereas an
analog containing sulfur in the non-reducing
ring and nitrogen in the glycosidic linkage
showed much stronger inhibition with a Ki of
4 mM [29]. It was of interest, therefore, to
examine the general class of compounds con-
taining S/N acetal functions as potential in-
hibitors of GA from A. niger and brewer’s
yeast a-glucosidase. In the present study we
report the synthesis and enzyme inhibitory
a-Glucosidases (a-
lase, EC 3.2.1.20) are also exo-acting carbo-
hydrases, catalyzing the release of -gluco-
D
-glucoside glucohydro-
D
pyranose from non-reducing ends of various
substrates having an a-glucosidic linkage,
such as disaccharides, oligosaccharides, aryl
glucosides and even starch [21]. a-Glucosi-
dases are widespread in nature and have been
isolated from fungi, animals, and higher
plants [21]. They generally display broad sub-
strate specificity, and are conventionally
classified into three types dependent on dis-
tinct properties; i.e., class I, typical a-glucosi-
dases hydrolyzing heterogeneous substrates
such as aryl glucosides; class II, maltases hy-
drolyzing homogeneous substrates like mal-
tose; and class III, like the maltases, but
capable of also attacking a-glucans [21]. a-
Glucosidase from brewer’s yeast is a class I
enzyme, having a 120-fold preference, based
activity of a series of a- -glucopyranosy-
D
lamines containing simple arylamines as the
aglycons.
2. Results and discussion
Synthesis.—1,2,3,4,6-Penta-O-acetyl-5-thio-
-glucopyranose (1) [30,31] was treated with
D
hydrazine acetate [32] to afford 2 in 90% yield.
The tetraacetate 2 was then reacted with the
respective arylamine. The reactions were cata-
lyzed by 10% HgCl2 (Scheme 1) and pre-
sumably proceed via stabilization of an acyclic
intermediate by complexation of Hg2+ to the
sulfur atom. It is significant that reactions
catalyzed by protic acids such as acetic acid
did not proceed smoothly. Reaction condi-
tions varied slightly for 3–6, but all reactions
on kcat/Km, for p-nitrophenyl a- -glycopyra-
D
noside compared with that of maltose [22]. In
contrast to GA, which acts by inversion of
configuration, the exact mechanism of hydrol-
ysis by brewer’s yeast a-glucosidase is cur-

K.D. Randell et al. / Carbohydrate Research 321 (1999) 143–156
145
Scheme 1. (i) H2NNH2·AcOH, DMF; (ii) p-XꢀC6H4NH2, 10% HgCl2; (iii) 5:1:1 MeOH–H2O–Et3N.
were completed at 50 °C or less and in less
than 20 h. The resulting anomeric mixtures
were separated and purified by chromatogra-
phy and the pure isomers were obtained by
recrystallization.
protein environment was simulated with
bovine serum albumin (BSA), showed nega-
tive NOE effects from H-1 of both a and b
isomers of the free ligand 7. A further negative
enhancement due to the transferred NOE was
observed at both mixing times by comparing
peak intensities with those of the control sam-
ple. The normalized peak intensity for the a
anomer increased up to 100% in the sample
containing enzyme relative to the sample con-
taining BSA, indicating the presence of inten-
sity due to transferred NOE. In contrast, b
peak intensities decreased slightly or did not
change. For example, at a mixing time of 205
ms, the a H-1–H-2 contact peak intensity
increased from 0.070 in the control to 0.098
with glucoamylase (relative to the H-1 diago-
nal peak), an increase of 40%. In contrast, the
b H-1–H-3 contact peak intensities were 0.035
and 0.034 in the control and with glucoamy-
lase, respectively. For the H-1–aromatic con-
tact, the a peak intensity increased 95% from
0.073 to 0.142, while the b intensity decreased
from 0.068 to 0.050 (see Fig. 1). This evidence
suggests [33] that GA binds only the a isomer.
Enzyme inhibition.—The Ki values for inhi-
bition of GA by compounds 7–11 are shown
in Table 1. In addition, compounds 7 and 8
were chosen as representative of the first set of
compounds and their inhibition of a-glucosi-
dase from brewer’s yeast was compared with
the inhibitory effect of 11 (Table 1). Because
11 was a much better inhibitor of GA than 7
or 8, it was of interest to examine whether this
trend would hold for a related enzyme. These
values were determined by assuming that only
the a isomers in the anomeric mixtures would
be accepted as inhibitors of glucoamylase. GA
is inhibited by compounds 7–11, all with Ki
values under 1 mM (Table 1). The Km values
for the substrates maltose and p-nitrophenyl
The tetraacetates 3–6 were deprotected with
a 5:1:1 mixture of methanol–water–triethy-
lamine [29] to give 7–10, respectively. The
deprotected sugars were obtained as anomeric
mixtures that could not be separated. The
percentages of the a isomers in equilibrated
1
samples in D2O were assessed by H NMR
spectroscopy and were used in the calculation
of the Ki values, as discussed below. The a:b
ratios for 7–10 were 1:2.2, 1:2, 1:0.9, and
1:2.4, respectively (Scheme 1).
Preferential binding of the h isomers.—As
was the case in our earlier study of glucoamy-
lase binding of the maltoside heteroanalog 11
containing sulfur in the non-reducing ring and
nitrogen in the interglycosidic linkage [29],
NMR studies with an a/b mixture of com-
pound 7 in the presence of GA indicated
significant line-broadening effects for the sig-
nals of only the a isomer. Thus, when glu-
coamylase was added to the sample containing
7, selective broadening (2.4–3.9 Hz) of the
H-1 resonance of the a isomer was observed.
Other resonances of the a isomer were also
broadened such that Dw1/2 was not measur-
able. Linewidths for the resonances b isomer
did not change.
In order to confirm that the a isomer was
preferentially bound, transferred NOE experi-
ments [33] were performed. A control NOE
experiment, in which the viscosity of the
a- -glucopyranoside are 1.2 and 3.7 mM, re-
D

146
K.D. Randell et al. / Carbohydrate Research 321 (1999) 143–156
Fig. 1. Traces from TRNOESY spectra (mixing time 205 ms) of compound 7. (1 and 2): at a H-1 resonance; (1) with
glucoamylase, and (2) with BSA. (3 and 4): at b H-1 resonance; (3) with glucoamylase, and (4) with BSA.
modynamic penalty associated with burying
uncompensated negative charge [9]. One might
have expected that the Ki values would in-
crease with decreasing pKa values.
Thermodynamic analysis of the complexa-
tion between GA and various inhibitors has
shown several interesting features of the mech-
anism and strength of inhibition [16]. The
binding process is both enthalpically and en-
tropically driven; thus, both hydrogen bond-
ing and dehydration of protein–oligo-
saccharide surfaces drives the complex forma-
spectively. The differences in Ki may reflect
the influence of the substituents on the pKa,
and thus the ionization state of the nitrogen
atom. The pKa values for the isolated aglycon
differ significantly: 4.60 for aniline, 5.31 for
p-anisidine, 2.45 for p-trifluoromethylaniline
and 1.00 for p-nitroaniline [34]. Protonation
of the inhibitor may take place and this is
favorable for binding in two ways: the forma-
tion of an ion pair with an active site carboxy-
late and also introduction of positive charge in
the active site is not accompanied by a ther-
Table 1
Inhibition of glucoamylase and a-glucosidase by 7–11
Inhibitor
Ki (mM)
Glucoamylase
Ki (mM)
a-Glucosidase
p-Methoxy-N-phenyl 5-thio-
N-Phenyl 5-thio- -glucopyranosylamine (8)
p-Nitro-N-phenyl 5-thio- -glucopyranosylamine (9)
p-Trifluoromethyl-N-phenyl 5-thio- -glucopyranosylamine (10)
D
-glucopyranosylamine (7)
0.4790.12 a
0.7890.27
1.0590.18
\10
D
D
0.2790.02
nd b
D
0.8790.12
0.004490.003
nd
0.590.15
Methyl 4-amino-4-deoxy-4-N-(5%-thio-a-
D
-glucopyranosyl)-a-
D
-glucopyranoside (11) [29]
a Standard deviation.
b nd, not determined

K.D. Randell et al. / Carbohydrate Research 321 (1999) 143–156
147
11. The significant difference in Ki values be-
tween compounds 7 and 8 for brewer’s yeast
a-glucosidase (Table 1) may, in this case,
reflect the influence of the substituent on the
pKa value of the nitrogen atom. Alternatively,
a hydrogen bond to the methoxy substituent
may be formed at the periphery of binding
subsite +1.
tion [16]. pH-Dependence studies of the com-
plex formation between GA and 1-deoxynojir-
imycin and acarbose suggested that the
complex formation was most efficient at pH
values close to the pKa of the inhibitor nitro-
gen [16]. Binding of acarbose and 1-deoxyno-
jirimycin at pH 4.5 is accompanied by the
release of 0.590.1 proton, probably due to
partial protonation of both inhibitor and gen-
eral acid catalyst in the active site, whereas
0.790.1 proton is taken up at pH 7.5 [16].
Formation of an ion pair in the active site,
consisting of a protonated inhibitor and an
anionic residue can, in principle, take place in
two ways: a neutral form of the inhibitor
could be bound by the enzyme, and subse-
quently protonated by the general acid cata-
lyst, or alternatively, a protonated form of the
inhibitor could form an ion pair with an active
site carboxylate. Thus, although the glycosidic
nitrogen atom in the p-nitro analogs might be
unprotonated, it may still be involved in a
strong ion pair with the general acid catalyst,
accounting for the low Ki value.
The relatively higher Ki values for 7–10
compared with that of 11 can be rationalized
based on the analysis of complementary inter-
actions between ligands and enzyme at sub-
sites −1 and +1. It is of note that analysis
of recognition of deoxygenated maltoside
analogs by GA revealed key polar group inter-
actions of OH-4%, -6%, and -3 at substrate bind-
ing subsites −1 and +1, respectively [19].
Elimination of the OH-3 interacting with NH-
1 of Arg305 [9] resulted in loss of transition-
state stabilization of 11.3 kJ/mol [19]. We
speculate, therefore, that the higher Ki values
for 7–10 compared with that of 11 are due in
part to the loss of hydrogen bonding from
Arg305 to the aglycon at the binding subsite
+1. Alternatively, the greater inhibition by 11
could result from its higher basicity relative to
7–10. In contrast to glucoamylase, brewer’s
yeast a-glucosidase has no key polar group
interactions to the substrate aglycon at bind-
ing subsite +1, and thus binding strongly
depends on charged hydrogen bonds to all
OH groups of the non-reducing sugar ring
[35]. Compound 7, which lacks possible OH-
group partners, is therefore essentially as good
an inhibitor of brewer’s yeast a-glucosidase as
Mutational and structural analysis of GA
suggested Trp120 to be involved in transition-
state stabilization through hydrogen bonding
to the general acid catalyst Glu179, and by
direct stacking towards the sugar in subsite
+2 [2,9,10,36]. Mutation of Trp120 to Phe
thus decreased activity 80-fold compared with
wild-type GA using maltose as substrate, ac-
companied, however, by no changes in Km [2].
In the present study, compound 11 showed
competitive inhibition towards Trp120“Phe
GA with a Ki value of 0.12 mM. Although Km
on maltose for wild-type and Trp120“Phe
GA are similar, Ki for 11 is 30-fold higher for
Trp120“Phe than for the wild-type enzyme.
The effect of Trp120“Phe was shown previ-
ously, using stopped-flow fluorescence spec-
troscopy, to be substrate-length dependent;
thus, the mutation is less severe on small
substrates like maltose compared with longer
oligosaccharides [36]. Assuming that the mu-
tation using a short inhibitor, like 11, only
affects the hydrogen bond to the general acid
catalyst Glu179, the present study reinforces
the importance for efficient inhibition of a salt
linkage from Glu179 to the basic nitrogen,
stabilized in the wild-type enzyme by Trp120,
but lost in the mutant. The area surrounding
Glu179 is probably extremely sensitive to mu-
tation and the actual ionization state of
Glu179 has in fact been questioned by crystal-
lographic studies of the GA–acarbose com-
plex [9]. In addition, acarbose binding was
affected in Trp120“Phe GA. The association
constant for acarbose is thus decreased
markedly (10−5-fold) in Trp120“Phe GA
compared with wild-type GA [17], a combina-
tion probably due to the loss of hydrophobic
stacking against the sugar, and of the hydro-
gen bond to Glu179. Significantly, a pH–ac-
tivity dependence study for Trp120“Phe has
shown that the pKa of Glu179 is not changed
in the substrate bound form [37]. It is there-

148
K.D. Randell et al. / Carbohydrate Research 321 (1999) 143–156
,
Fig. 2. Stereoview of the active site of glucoamylase with 11 bound. Hydrogen bonds less than 3 A are represented by dashed lines.
The catalytic water molecule is represented by an X.
fore likely that the elimination of the Trp120
to Glu179 H bond in the mutant enzyme leads
to an unfavourable orientation of the catalytic
acid Glu179 and a higher Ki value for 11.
Molecular modeling.—Compounds 7–9 and
11 were used in molecular modeling studies to
investigate the ligand–enzyme interactions
and to offer insight into the differences in
inhibitory potencies of the ligands. The com-
pounds were introduced into the active site of
As compared with 11, the aniline analog 8
has lost three hydrogen-bond interactions due
to substitution of the reducing-end residue with
an aromatic moiety (Fig. 3). The hydrogen-
bond distance of OH-4% to Arg54 NH-1 is
the
D
-gluco-dihydroacarbose (DGA, 12)–glu-
coamylase complex [38]. Overall the models
have the same hydrogen-bond pattern for the
non-reducing ends as seen for the DGA glu-
coamylase complex both before and after min-
imization. Furthermore, before minimization
the reducing end of 11 shows the same hydro-
gen bonds as seen for ring B in 12. The only
exception is a hydrogen-bond contact between
OH-6 and Lys108 through a water molecule.
After minimization (Fig. 2), the hydrogen
bond from OH-2 to Glu180 is changed from
,
lengthened from 2.88 to 3.02 A. The aromatic
ring of 8 stacks with the side-chain of Tyr311
with an angle of 36°. The increase in Ki of this
compound as compared with 11 is probably
due to the loss of the three hydrogen bonds in
the active site.
In compound 9, the aromatic ring again
stacks with Tyr311, with an angle of 33°. In
addition, a hydrogen bond between Glu180
OE-2 and the nitro functionality (Fig. 4) was
,
2.84 to 3.38 A and the hydrogen bond from
OH-2% to Arg305 NH-1 is changed from 3.00
,
detected. This hydrogen bond is 2.97 A and it
,
to 3.30 A. All other hydrogen-bond interac-
does not change upon minimization, although
we note that the hydrogen bond is not in the
s-plane of the aromatic ring. It is not clear,
therefore, whether the decrease in Ki relative to
that of 8 could be attributed to this additional
hydrogen bond.
tions are approximately unchanged upon min-
imization.
D-Gluco-dihydroacarbose (12) is
the second strongest inhibitor known for glu-
coamylase, with a Ki value of 10−8 M. The
similarity in hydrogen bonding and conforma-
tional patterns between 11 and 12 explains
why it is also a good inhibitor.
Compound 7 was also found to have the
same hydrogen bond to the Arg54 NH-1 as in

K.D. Randell et al. / Carbohydrate Research 321 (1999) 143–156
149
,
Fig. 3. Stereoview of the active site of glucoamylase with 8 bound. Hydrogen bonds less than 3 A are represented by dashed lines.
The catalytic water molecule is represented by an X.
8, but the O-methyl function has an additional
apparatus and are uncorrected. Optical rota-
tions were measured with a Rudolph Research
weak hydrogen bond with Glu180 OE-2 of 3.47
1
13
,
A (Fig. 5). The aromatic ring stacks with
Autopol II automatic polarimeter. H and C
NMR spectra were recorded on a Bruker
AMX-400 NMR spectrometer at 400.13 and
100.6 MHz, for proton and carbon, respec-
tively, unless otherwise stated. The spectra
were recorded in CDCl3 or D2O. Chemical
shifts are given in ppm downfield from Me4Si
for those measured in CDCl3 and from 2,2-
dimethyl-2-silapentane-5-sulfonate (DSS) for
those spectra measured in D2O. Chemical
Tyr311 with an angle of 33°. The decrease and
increase in Ki values as compared with those of
8 and 9, respectively, can be attributed to this
weak H bond of the O-methyl function.
3. Experimental
General methods.—Melting points were de-
termined on a Fisher–Johns melting point
,
Fig. 4. Stereoview of the active site of glucoamylase with 9 bound. Hydrogen bonds less than 3 A are represented by dashed lines.
The catalytic water molecule is represented by an X.

150
K.D. Randell et al. / Carbohydrate Research 321 (1999) 143–156
shifts and coupling constants were obtained
from a first-order analysis of the spectra. An-
alytical thin-layer chromatography (TLC) was
performed on aluminum plates precoated with
Silica Gel 60F-254 (E. Merck) as the adsor-
bent. The developed plates were air-dried, ex-
posed to UV light and/or sprayed with a
solution containing 1% Ce(SO4)2 and 1.5%
molybdic acid in 10% aq H2SO4 and heated at
150 °C. All compounds were purified by flash
column chromatography on Kieselgel 60
(230–400 mesh). Solvents were distilled before
use and were dried as necessary, by literature
procedures. Solvents were evaporated under
reduced pressure and below 40 °C.
Enzymes.—Recombinant A. niger wild-type
(AnMT 833) and Trp120“Phe GA were ob-
tained as culture filtrates from Novo-Nordisk
(Bagsvaerd, Denmark) and purified on indi-
vidual acarbose–Sepharose columns [39], fol-
lowed by ion-exchange chromatography using
Hiload Q-Sepharose to separate G1 and G2
[40]. G2 was used throughout the present
study. Brewer’s yeast a-glucosidase (Type VI,
EC. 3.2.1.20) was obtained as a lyophilized
powder from Sigma (G-6136; Lot No.
21F8105). After resuspension in 0.05 M phos-
phate, pH 6.8 and dialysis overnight at 4 °C
against 0.05 M phosphate, pH 6.8, the enzyme
(360 U/mL) was used for inhibition studies
without further purification.
Enzyme inhibition assays.—The initial rates
of GA G2-catalyzed hydrolysis of maltose
(10 different substrate concentrations in the
range 0.19–22.7 mM) were followed in the
absence and in the presence of the different
inhibitors (5 different concentrations in the
range 0.3–8 mM) in 0.1 M sodium acetate,
pH 4.5 at 45 °C and a final enzyme concentra-
tion in the range 15–90 nM. The glucose
released was analyzed in aliquots removed at
appropriate time intervals using a glucose oxi-
dase assay adapted to microtiter plate reading
and using a total reaction volume for the
enzyme reaction mixtures of 150 or 300 mL
[4,14,41,42]. Absorbances were read at 450 or
490 nm after 1 h incubation at room tempera-
ture (rt) and quantitated using
standard.
D
-glucose as a
The initial rates of a-glucosidase-catalyzed
hydrolysis of p-nitrophenyl a- -glucopyra-
D
noside were followed in 0.05 M phosphate,
pH 6.8 and at 30 °C at ten different substrate
concentrations in the range 0.049–3.92 mM,
in the absence and presence of inhibitor. The
p-nitrophenol released was analyzed in
aliquots removed at appropriate time intervals
and quenched in 0.1 M sodium borate, pH
9.4. Absorbances were read at 405 nm and
quantitated using p-nitrophenol as a stan-
dard.
,
Fig. 5. Stereoview of the active site of glucoamylase with 7 bound. Hydrogen bonds less than 3 A are represented by dashed lines.
The catalytic water molecule is represented by an X.

K.D. Randell et al. / Carbohydrate Research 321 (1999) 143–156
151
then linked together, resulting in a model for 8.
This model complex was then adjusted to
minimize close van der Waals contacts.
The inhibitors were all competitive, and the
constants of inhibition were calculated from
Km%=Km(1+([I]/Ki)), where Km% and Km are
the Michaelis–Menten constants determined
in the presence and the absence of inhibitor,
using the software ENZFITTER [43], and [I] is
the concentration of inhibitor. Vmax and Km, in
the absence and presence of inhibitor, were
obtained by fitting the velocities as a function
of substrate concentration to the Michaelis–
Menten equation. kcat was derived from kcat=
The model of 9 was constructed by superim-
posing the p-nitroaniline group from 1pip.pdb
[45] on the aniline moiety of 8. The non-reduc-
ing ring of 11 and the superimposed p-ni-
troaniline group were then linked together,
resulting in a model of 9; minimization was
then performed. Furthermore, after the adjust-
ment, an oxygen atom from the nitro function-
ality was positioned within hydrogen-bonding
distance to OE2 of Glu180; the aromatic por-
tion of the molecule was stacked with the
aromatic moiety of Tyr311 in a 33–36° angle.
The model of 7 was constructed from 9, by
replacing the nitro functionality with the O-
methyl group. The complex was minimized,
bringing the methyl group within hydrophobic
contact distance to Tyr311.
V
max/[Eo] where Vmax is the maximum velocity
and [Eo] the molecular enzyme concentration.
With 5-thio- -glucose, the glucose oxidase had
51% of the activity toward -glucose and
neither 5-thio- -glucose nor the glucoamylase
D
D
D
inhibitors tested were inhibitors of the glucose
oxidase.
Molecular modeling.—Docking and mini-
mization procedures are outlined below. The
5-thio-
D
-glucopyranosyl amines were intro-
-gluco-dihy-
After initial construction of the 5-thio- -
D
duced into the active site of
D
glucopyranosylarylamine–glucoamylase mod-
els, hydrogens were added using X-PLOR
version 3.851 [46] and the CHARMM 22
parameter and topology parameter set, assum-
ing the g-carboxylate function of Glu180 was
protonated. Finally, to fill out cavities created
upon deletion of residues from the original
DGA–GA complex and to give a uniform
surface, the models were soaked in a water
droacarbose (DGA 12)–glucoamylase com-
,
plex, determined at 1.7 A resolution [38], using
the following procedure. First, the DGA com-
plex was prepared for docking of ligands and
minimization by removing all the hydrogen
atoms and all residues of the low-occupancy
conformation [38]. The ligand 11 was con-
structed using standard procedures in INSIGHT
II version 97.0 (Molecular Simulations Inc.,
,
shell of 2.5 A. Parameters for standard groups
San Diego, CA) from two molecules of a- -
D
were generated from existing parameters of
equivalent groups.
glucose linked by an a-(1“4) bond followed
by replacing the glucosidic oxygen with a ni-
trogen and the endocyclic oxygen of the non-
reducing ring with a sulfur. Finally, an
a-O-methyl group was added to the reducing
end of the glucose unit. The C-2, C-3, and C-4
atoms of the non-reducing ring of 11 were
superimposed upon C-2, C-3, and C-4 of ring
A of DGA 12. The model of 11 was then
adjusted to minimize close van der Waals
contacts. The adjustment placed the reducing-
end residue of 11 in an equivalent position as
seen for ring B of DGA 12.
The models were subjected to 100 cycles of
conventional Powell minimization using X-
PLOR version 3.851. The positions of the
amino acids in the active site, the atoms of the
ligands, and the water-shell molecules were all
optimized. The gradients changed from about
50 to about 10 after 100 cycles of optimization.
Further refinement of the models may be ob-
tained by molecular dynamics simulations, but
the present models were considered to be suffi-
cient for the purposes of the present work.
Inspection and adjusting of the models were
The model of 8 was constructed by superim-
posing the nitrogen and six carbons of the
aniline group from 1eld.pdb [44] on the N-4,
C-4, C-3, C-2, C-1, O-5, and C-5 atoms of the
reducing-end residue of 11. The non-reducing
ring and the superimposed aniline group were
done using TURBO FRODO version OpenGL.1
–
(CNRS-AFMB, Marseille, France).
Transferred NOE experiments.—Spectra
were acquired on a Bruker AMX600 spec-
trometer at 600 MHz. One-dimensional spec-

152
K.D. Randell et al. / Carbohydrate Research 321 (1999) 143–156
mediately. The solvent was then evaporated,
the residue was dissolved in CH2Cl2 (50 mL)
and washed with NaHCO3 (10 mL) and H2O
(10 mL), and dried over Na2SO4. The solvent
was evaporated to yield an orange foam. The
a isomer was crystallized from 95% EtOH, as
fine needles. The remaining mixture was sepa-
rated by column chromatography, using a 4:1
mixture of toluene–EtOAc as the eluant. Both
3a and 3b were then recrystallized from 95%
EtOH (56%, 3a:3b 8.8:1).
tra were recorded spinning at 300 K; 32K data
points were acquired over a spectral width of
6 ppm. For 1D spectra, 1 mg of an a/b
mixture of compound 7 was dissolved in 0.6
mL of phosphate buffered saline–D2O (pH
7.5) to 5.4 mM (1.8 mM a, 3.6 mM b). For
each spectrum, 64 scans preceded by 16
dummy scans were acquired, with a relaxation
delay of 3.5 s. For 2D TRNOESY spectra, 1
mg of an a/b mixture of compound 7 was
dissolved in 0.6 mL of phosphate buffered
saline–D2O (pH 7.5) along with 5.5 mg glu-
coamylase G2 (0.12 mM), giving a ratio of
approximately 15:1 a–enzyme. A control sam-
ple was prepared by dissolving 1 mg of the a/b
mixture of compound 7 in 0.6 mL of phos-
phate buffered saline–D2O (pH 7.5) along
with 5.4 mg BSA (0.14 mM). 2D TRNOESY
spectra were acquired nonspinning at 298 K,
in phase-sensitive mode using TPPI [47]. The
spectral width was 8 ppm. Spectra were ac-
quired under identical conditions for both
samples. One set of spectra used a mixing time
of 125 ms and 48 scans per increment; the
other used a mixing time of 205 ms and 96
scans per increment. All experiments were pre-
ceded by 16 dummy scans. Water suppression
was achieved by presaturation during the re-
laxation delay (2 s) and a 180° pulse in the
middle of the mixing time followed by a 5 ms
homospoil pulse. An 18 ms T1r filter [48] was
applied after the excitation pulse to relax
protein resonances.
3a: Mp 198–200 °C; [h]2D2 326° (c 0.53,
1
CHCl3); H NMR (CDCl3): 2.01 (s, 3 H,
OCOCH3), 2.02 (s, 3 H, OCOCH3), 2.03 (s, 3
H, OCOCH3), 2.04 (s, 3 H, OCOCH3), 3.47
(ddd, 1 H, J4,5 10.8, J5,6B 3.0, J5,6A 4.1 Hz,
H-5), 3.76 (s, 3 H, OCH3), 3.98 (dd, 1 H,
J6A,6B 12.1 Hz, H-6B), 4.10 (d, 1 H, JNH,1 4.9
Hz, NH), 4.44 (dd, 1 H, H-6A), 4.79 (dd, 1 H,
J1,2 3.7 Hz, H-1), 5.34 (dd, 1 H, J3,4 9.0 Hz,
H-4), 5.40 (dd, 1 H, J2,3 10.1 Hz, H-2), 5.47
(dd, 1 H, H-3), 6.75–6.87 (dd, 4 H, Ar); 13C
NMR (CDCl3): 20.76–20.88 (4-OCOCH3),
38.79 (C-5), 56.01 (OCH3), 58.38 (C-1, JC-1–H-1
153.85 Hz), 61.83 (C-6), 71.49, 72.65, 74.13
(C-2, C-3, or C-4), 115.19 (C-m), 116.81 (C-o),
138.90 (C-i ), 154.23 (C-p), 169.51–170.71
(CꢁO). Anal. Calcd for C21H27NO9S: C, 53.71;
H, 5.81; N, 2.98. Found: C, 53.75; H, 5.68; N,
2.80.
3b: Mp 203–204 °C; [h]2D2 −66° (c 0.52,
1
CHCl3); H NMR (CDCl3): 1.98 (s, 3 H,
OCOCH3), 2.01 (s, 3 H, OCOCH3), 2.02 (s, 3
H, OCOCH3), 2.03 (s, 3 H, OCOCH3), 3.37
(ddd, 1 H, J4,5 10.7, J5,6B 3.4, J5,6A 5.0 Hz,
H-5), 3.69 (d, 1 H, JNH,1 10.9 Hz, NH), 3.74
(s, 3 H, OCH3), 4.08 (dd, 1 H, J6A,6B 11.9 Hz,
H-6B), 4.32 (dd, 1 H, H-6A), 4.69 (dd, 1 H,
J1,2 10.0 Hz, H-1), 5.18 (dd, 1 H, J2,3 9.6 Hz,
H-3), 5.31 (dd, 1 H, J3,4 9.5 Hz, H-4), 5.32
(dd, 1 H, H-2), 6.67–6.8 (dd, 4 H, Ar); 13C
NMR (CDCl3): 20.48–20.66 (4-OCOCH3),
41.77 (C-5), 55.73 (OCH3), 60.62 (C-1, JC-1–H-1
150.3 Hz.), 61.52 (C-6), 72.26, 73.50, 74.63
(C-2, C-3, or C-4), 114.92 (C-m), 115.81 (C-o),
138.35 (C-i ), 153.69 (C-p), 169.45–170.99
(CꢁO). Anal. Calcd for C21H27NO9S: C, 53.71;
H, 5.81; N, 2.98. Found: C, 53.69; H, 5.72; N,
2.76.
Processing of data was performed with
standard UXNMR and XWINNMR (Bruker)
software. Zero-filling of the acquired data
(512 t1 values and 2K data points in t2) led to
a final data matrix of 1K×2K (F1×F2) data
points. Chemical shifts were referenced to
sodium 2,2-dimethyl-2-silapentane-5-sulfonate
(DSS). The baseline was corrected in F1 and
F2 with a fifth-order polynomial prior to inte-
gration of peak volumes.
p-Methoxy-N-phenyl 2,3,4,6-tetra-O-acetyl-
h/i-5-thio- -glucopyranosylamine (3a and
D
3b).—To a solution of 2,3,4,6-tetra-O-acetyl-
5-thio- -glucopyranose (2) (470 mg, 1.16
D
mmol) in MeOH (10 mL) was added anisidine
(1.16 mmol, 143 mg) and HgCl2 (34 mg, 0.12
mmol). The reaction mixture was stirred at rt
under nitrogen and a precipitate formed im-
N-Phenyl 2,3,4,6-tetra-O-acetyl-h/i-5-thio-
D
-glucopyranosylamine (4a and 4b).—To a so-
lution of 2,3,4,6-tetra-O-acetyl-5-thio- -gluco-
D

K.D. Randell et al. / Carbohydrate Research 321 (1999) 143–156
153
pyranose (2) (700 mg, 1.72 mmol) in MeOH
(10 mL) was added aniline (1.72 mmol, 156
mL) and HgCl2 (0.17 mmol, 47 mg). The
reaction mixture was stirred at rt under nitro-
gen for 30 min. The solvent was evaporated,
the product was dissolved in CH2Cl2 (50 mL)
and washed with NaHCO3 (10 mL) and H2O
(10 mL), and dried over Na2SO4. The solvent
was evaporated to yield an orange foam. The
a isomer was crystallized from 95% EtOH, as
a cottony fiber. The remaining mixture was
separated by column chromatography, using a
4:1 mixture of toluene–EtOAc as the eluant.
Both 4a and 4b were then recrystallized from
95% EtOH (58%, 4a:4b 4.8:1).
p-Nitro-N-phenyl 2,3,4,6-tetra-O-acetyl-h/
i-5-thio- -glucopyranosylamine (5a and 5b).
—To a solution of 2,3,4,6-tetra-O-acetyl-5-
thio- -glucopyranose (2) (1.0 g, 2.46 mmol) in
D
D
MeOH (20 mL) was added p-nitroaniline (340
mg, 2.46 mmol) and HgCl2 (134 mg, 0.49
mmol). The reaction mixture was stirred at
50 °C, under nitrogen for 18 h. The solvent
was then evaporated, the product was dis-
solved in CH2Cl2 (50 mL) and washed with
NaHCO3 (10 mL) and water (10 mL), and
dried over Na2SO4. The solvent was evapo-
rated to yield a yellow/brown foam. The mix-
ture was separated by column chroma-
tography, using a 4:1 mixture of toluene–
EtOAc as the eluant. Both 5a and 5b were
then recrystallized from 95% EtOH (8.4%,
5a:5b 1:1).
4a: Mp 207–209 °C; [h]2D2 400° (c 0.52,
1
CHCl3); H NMR (CDCl3): 2.01 (s, 3 H,
OCOCH3), 2.03 (s, 3 H, OCOCH3), 2.04 (s, 3
H, OCOCH3), 2.05 (s, 3 H, OCOCH3), 3.44
(ddd, 1 H, J4,5 10.9, J5,6B 3.2, J5,6A 4.2 Hz,
H-5), 3.98 (dd, 1 H, J6A,6B 12.1 Hz, H-6B),
4.28 (d, 1 H, JNH,1 3.8 Hz, NH), 4.43 (dd, 1 H,
H-6A), 4.88 (dd, 1 H, J1,2 4.0 Hz, H-1), 5.34
(dd, 1 H, J3,4 8.9 Hz, H-4), 5.41 (dd, 1 H, J2,3
10.1 Hz, H-2), 5.47 (dd, 1 H, H-3), 6.7–7.3
5a: Mp 215–217 °C; [h]2D2 353° (c 0.38,
1
CHCl3); H NMR (CDCl3): 2.03 (s, 3 H,
OCOCH3), 2.04 (s, 3 H, OCOCH3), 2.05 (s, 3
H, OCOCH3), 2.07 (s, 3 H, OCOCH3), 3.38
(ddd, 1 H, J4,5 10.8, J5,6B 3.2, J5,6A 4.5 Hz,
H-5), 4.0 (dd, 1 H, J6A,6B 12.2 Hz, H-6B), 4.41
(dd, 1 H, H-6A), 4.96 (dd, 1 H, J1,2 4.4 Hz,
H-1), 5.21 (d, 1 H, JNH,1 4.4 Hz, NH), 5.35
(dd, 1 H, J3,4 9.2 Hz, H-4), 5.40 (dd, 1 H, J2,3
10.3 Hz, H-2), 5.45 (dd, 1 H, H-3), 6.7–7.3
(dd, 4 H, Ar); 13C NMR (CDCl3): 20.71–
20.80 (4-OCOCH3), 39.36 (C-5), 56.27 (C-1,
JC-1–H-1 154.9 Hz), 61.24 (C-6), 70.98, 72.44,
73.48 (C-2, C-3, or C-4), 113.01 (C-o), 126.10
(C-m), 140.80 (C-p), 150.28 (C-i ), 169.43–
170.59 (CꢁO). Anal. Calcd for C20H24N2O10S:
C, 49.57; H, 5.00; N, 5.78. Found: C, 49.68;
H, 5.14; N, 5.58.
13
(m, 5 H, Ar); C NMR (CDCl3): 20.51–20.61
(4-OCOCH3), 38.69 (C-5), 57.18 (C-1, JC-1–H-1
154.52 Hz), 61.29 (C-6), 71.22, 72.49, 73.71
(C-2, C-3, or C-4), 114.88 (C-o), 119.89 (C-p),
129.30 (C-m), 144.82 (C-i ), 169.25–170.43
(CꢁO). Anal. Calcd for C20H25NO8S: C, 54.65;
H, 5.74; N, 3.19. Found: C, 54.75; H, 5.60; N,
3.06.
4b: Mp 146–148 °C; [h]2D2 −71° (c 0.53,
1
CHCl3); H NMR (CDCl3): 1.98 (s, 3 H,
OCOCH3), 2.03 (s, 3 H, OCOCH3), 2.04 (s, 3
H, OCOCH3), 2.06 (s, 3 H, OCOCH3), 3.28
(ddd, 1 H, J4,5 10.6, J5,6B 3.4, J5,6A 5.0 Hz,
H-5), 4.03 (d, 1 H, JNH,1 10.0 Hz, NH), 4.10
(dd, 1 H, J6A,6B 11.9 Hz, H-6B), 4.33 (dd, 1 H,
H-6A), 4.78 (dd, 1 H, J1,2 10.0 Hz, H-1), 5.19
(dd, 1 H, J2,3 9.6 Hz, H-3), 5.31 (dd, 1 H, J3,4
9.4 Hz, H-4), 5.33 (dd 1 H, H-2), 6.0–7.7 (m,
5 H, Ar); 13C NMR (CDCl3): 20.48–20.63
(4-OCOCH3), 41.77 (C-5), 59.31 (C-1, JC-1–H-1
151.70 Hz), 61.48 (C-6), 72.23, 73.45, 74.62
(C-2, C-3, or C-4), 114.06 (C-o), 119.65 (C-p),
129.38 (C-m), 144.47 (C-i ), 169.46–171.15
(CꢁO). Anal. Calcd for C20H25NO8S: C, 54.65;
H, 5.74; N, 3.19. Found: C, 54.67; H, 5.61; N,
3.04.
5b: Mp 183–185 °C; [h]2D2 −22° (c 0.51,
1
CHCl3); H NMR (CDCl3): 2.01 (s, 3 H,
OCOCH3), 2.03 (s, 3 H, OCOCH3), 2.05 (s, 3
H, OCOCH3), 2.06 (s, 3 H, OCOCH3), 3.34
(ddd, 1 H, J4,5 10.6, J5,6B 3.4, J5,6A 5.3 Hz,
H-5), 4.13 (dd, 1 H, J6A,6B 12.0 Hz, H-6B),
4.34 (dd, 1 H, H-6A), 4.78 (dd, 1 H, J1,2 9.6
Hz, H-1), 4.94 (d, 1 H, JNH,1 9.3 Hz, NH),
5.21 (dd, 1 H, J3,4 9.6 Hz, H-3), 5.31 (dd, 1 H,
H-4), 5.32 (dd, 1 H, J2,3 9.6 Hz, H-2), 6.7–8.0
(dd, 4 H, Ar); 13C NMR (CDCl3): 20.60–
20.76 (4-OCOCH3), 42.34 (C-5), 58.50 (C-1,
JC-1–H-1 154.5 Hz), 61.49 (C-6), 72.30, 73.35,
74.94 (C-2, C-3, or C-4), 113.01 (C-o), 126.33
(C-m), 140.57 (C-p), 150.15 (C-i ), 169.59–

154
K.D. Randell et al. / Carbohydrate Research 321 (1999) 143–156
171.74 (CꢁO). Anal. Calcd for C20H24N2O10S:
C, 49.57; H, 5.00; N, 5.78. Found: C, 49.68;
H, 5.07; N, 5.62.
Ar), 7.45 (d, 2 H, Ar); 13C NMR (CDCl3):
20.36–20.50 (4-OCOCH3), 41.10 (C-5), 58.65
(C-1, JC-1–H-1 155.9 Hz), 61.37 (C-6), 72.20,
73.29, 74.69 (C-2, C-3, C-4), 113.31 (C-o),
114.22 (C-p), 126.70 (C-m), 147.20 (C-i ),
169.3–171.3 (CꢁO). Anal. Calcd for
C21H24F3NO8S: C, 49.70; H, 4.77; N, 2.76.
Found: C, 49.88; H, 4.80; N, 2.58.
p-Trifluoromethyl-N-phenyl 2,3,4,6-tetra-O-
acetyl-h/i-5-thio- -glucopyranosylamine (6a
D
and 6b).—To a solution of 2,3,4,6-tetra-O-
acetyl-5-thio- -glucopyranose (2) (500 mg,
D
1.23 mmol) in dry MeOH (10 mL) was added
p-trifluoromethylaniline (170 mL, 1.35 mmol)
and HgCl2 (34 mg, 0.12 mmol). The reaction
mixture was stirred at 50 °C, under nitrogen
for 1 h, during which a white precipitate
formed. The solvent was then evaporated, the
product was dissolved in CH2Cl2 (50 mL) and
washed with NaHCO3 (10 mL) and H2O (10
mL), and dried over Na2SO4. The solvent was
evaporated to yield a light brown foam. The
a isomer was crystallized from 95% EtOH, to
yield a cottony fiber. The remaining mixture
was separated by column chromatography,
using a 4:1 mixture of toluene–EtOAc as the
eluant. Both 6a and 6b were then recrystal-
lized from 95% EtOH (44%; 6a:6b 8:1).
p-Methoxy-N-phenyl h/i-5-thio- -glucopy-
D
ranosylamine (7).—The tetraacetate 3 (300
mg, 0.64 mmol) was dissolved in a 5:1:1 mix-
ture of MeOH–water–Et3N (21 mL) and the
mixture was stirred overnight at rt. The sol-
vent was removed and the residue was
purified by flash column chromatography us-
ing a 10:1 mixture of EtOAc–MeOH as the
eluant to give a white solid (155 mg, 80% a/b
1:1.7).
1H NMR (D2O): 2.98 (m, 2 H, H-5 a/b),
3.38 (dd, 1 H, J2,3=J3,4 9.1 Hz, H-3b), 3.62
(m, 2 H, H-4 a/b), 3.70 (m, 2 H, H-2b/H-3a),
3.76 (s, 6 H, OCH3), 3.85 (m, 3 H, H-6Ab,
H-6Bb), 3.91 (dd, 1 H, J5,6A 5.2, J6A,6B 11.9
Hz, H-3), 4.13 (dd, 1 H, J2,3 9.8 Hz, H-2a),
4.63 (d, 1 H, J1,2 9.7 Hz, H-1b), 4.81 (d, 1 H,
J1,2 4.5 Hz, H-1a), 6.90 (m, 8 H, Ar); 13C
NMR (D2O): 45.91 (C-5a), 48.79 (C-5b),
58.93 (OCH3), 62.93 (C-1a, JC-1–H-1 152.81
Hz), 63.12 (C-6a), 63.22 (C-1b, JC-1–H-1 156.38
Hz), 63.30 (C-6b), 76.20 (C-4b), 76.64 (C-4a),
76.92 (C-2a/C-3a), 79.04 (C-2b), 80.54 (C-3b),
118.18 (C-m a/b), 119.40 (C-o b), 119.77 (C-o
a), 142.90 (C-i a/b), 155.49 (C-p a/b). Anal.
Calcd for C13H19NO5S: C, 51.81; H, 6.35; N,
4.65. Found: C, 51.66; H, 6.18; N, 4.81.
6a: Mp 231–234 °C; [h]2D2 261° (c 0.53,
1
CHCl3); H NMR (CDCl3): 2.02 (s, 3 H,
OCOCH3), 2.03 (s, 3 H, OCOCH3), 2.04 (s, 3
H, OCOCH3), 2.05 (s, 3 H, OCOCH3), 3.39
(ddd, 1 H, J4,5 10.7 Hz, H-5), 3.99 (dd, 1 H,
J5,6B 3.1, J6A,6B 12.1 Hz, H-6B), 4.43 (dd, 1 H,
J5,6A 4.4 Hz, H-6A), 4.68 (d, 1 H, JNH,1 4.3
Hz, NH), 4.90 (dd, 1 H, J1,2 4.1 Hz, H-1),
5.33 (dd, 1 H, J3,4 8.9 Hz, H-4), 5.41 (dd, 1 H,
J2,3 10.2 Hz, H-2), 5.45 (dd, 1 H, H-3), 6.84
(d, 2 H, Ar), 7.5 (d, 2 H, Ar); 13C NMR
(CDCl3): 20.46–20.54 (4-OCOCH3), 38.89 (C-
5), 56.46 (C-1, JC-1–H-1 154.43 Hz), 61.10 (C-
6), 70.93, 72.32, 73.45 (C-2, C-3, or C-4),
114.21 (C-o), 126.5 (C-m), 147.40 (C-i ),
169.19–170.37 (CꢁO). Anal. Calcd for
C21H24F3NO8S: C, 49.70; H, 4.77; N, 2.76.
Found: C, 49.71; H, 4.81; N, 2.74.
N-Phenyl h/i-5-thio- -glucopyranosylamine
D
(8).—The tetraacetate 4 (200 mg, 0.46 mmol)
was dissolved in a 5:1:1 mixture of MeOH–
water–Et3N (14 mL) and the mixture was
stirred overnight at rt. The solvent was re-
moved and the residue was purified by flash
column chromatography using a 10:1 mixture
of EtOAc–MeOH as the eluant to give a
white solid (105 mg, 85%, a:b 1:1.6).
6b: Mp 189–192 °C; [h]2D2 −24° (c 0.21,
CHCl3); 1H NMR (CDCl3): 2.0 (s, 3 H,
OCOCH3), 2.03 (s, 3 H, OCOCH3), 2.05 (s, 3
H, OCOCH3), 2.07 (s, 3 H, OCOCH3), 3.30
(ddd, 1 H, H-5), 4.12 (dd, 1 H, J5,6B 3.4, J6A,6B
11.9 Hz, H-6B), 4.33 (dd, 1 H, J5,6A 5.2 Hz,
H-6A), 4.45 (d, 1 H, JNH,1 9.5 Hz, NH), 4.78
(dd, 1 H, J1,2 9.7 Hz, H-1), 5.20 (dd, 1 H, J3,4
9.6 Hz, H-3), 5.31 (dd, 1 H, J4,5 9.5 Hz, H-4),
5.32 (dd, 1 H, J2,3 9.5 Hz, H-2), 6.73 (d, 2 H,
1H NMR (D2O): 3.02 (m, 2 H, H-5 a/b),
3.39 (dd, 1 H, J2,3=J3,4 9.1 Hz, H-3b), 3.62
(dd, 1 H, J4,5 10.5 Hz, H-4b), 3.66 (dd, 1 H,
J3,4 9.1, J4,5 10.5 Hz, H-4a), 3.72 (m, 2 H,
H-2b, H-3a), 3.77 (dd, 1 H, J5,6B 5.9 Hz,
H-6Bb), 3.81 (dd, 1 H, J5,6B 5.3 Hz, H-6Ba),
3.85 (m, 1 H, H-6Ab), 3.91 (dd, 1 H, J5,6A 5.2,

K.D. Randell et al. / Carbohydrate Research 321 (1999) 143–156
155
eluant to give a white solid (128 mg, 77%, a/b
1:2.24).
J6A,6B 12.0 Hz, H-6Aa), 4.05 (dd, 1 H, J2,3 9.9
Hz, H-2a), 4.73 (d, 1 H, J1,2 9.7 Hz, H-1b),
4.91 (d, 1 H, J1,2 4.4 Hz, H-1a), 6.88 (m, 6 H,
Ar), 7.27 (m, 4 H, Ar). 13C NMR (D2O):
46.01 (C-5a), 48.79 (C-5b), 61.62 (C-1a,
JC-1–H-1 152.9 Hz), 62.02 (C-1b, JC-1–H-1 149.5
Hz), 63.01 (C-6a), 63.19 (C-6b), 76.19 (C-4b),
76.63 (C-4a), 76.83 (C-2a), 76.94 (C-3a),
78.98 (C-2b), 80.52 (C-3b), 117.47 (C-o b),
117.48 (C-o a), 122.18 (C-p a), 122.24 (C-p b),
132.39 (C-m a), 132.44 (C-m b), 148.66 (C-i
a/b). Anal. Calcd for C12H17NO4S: C, 53.12;
H, 6.32; N, 5.16. Found: C, 53.15; H, 6.18; N,
5.02.
1H NMR (D2O): 3.00 (ddd, 1 H, J5,6A 3.1,
J5,6B 5.2 Hz, H-5 a), 3.05 (ddd, 1 H, J5,6 3.4,
J5,6B 5.4 Hz, H-5 b), 3.39 (dd, 1 H, J2,3 =J3,4
9.1 Hz, H-3b), 3.62 (dd, 1 H, J4,5 10.3 Hz,
H-4b), 3.66 (dd, 1 H, J4,5 10.5 Hz, H-4a), 3.75
(m, 3 H, H-2b, H-3a, H-6Ba), 3.83 (dd, 1 H,
J6A,6B 12.0 Hz, H-6Bb), 3.89 (dd, 1 H, J6A,6B
12.3 Hz, H-6Ab), 3.91 (dd, 1 H, H-6Aa), 4.08
(dd, 1 H, J2,3 9.9 Hz, H-2a), 4.80 (d, 1 H, J1,2
9.7 Hz, H-1b), 4.95 (d, 1 H, J1,2 4.8 Hz,
H-1a), 6.98 (m, 4 H, Ar), 7.5 (m, 4 H, Ar).
13C NMR (CD3OD): 45.36 (C-5a), 47.89 (C-
5b), 59.43 (C-1a, JC-1–H-1 151.0 Hz), 60.25
(C-1b, JC-1–H-1 151.2 Hz), 62.60 (C-6a/b),
75.70 (C-4b), 75.77, 75.83, 76.01 (C-4a/C-2a/
C-3a), 77.80 (C-2b), 79.86 (C-3b), 114.16 (C-o
b), 114.78 (C-o a), 126.97, 127.14, 127.17,
127.74 (C-p a/b, C-m a/b), 151.13 (C-i a/b).
Anal. Calcd for C13H16F3NO4S: C, 46.02; H,
4.75; N, 4.13. Found: C, 45.78; H, 4.90; N,
3.95.
p-Nitro-N-phenyl h/i-5-thio- -glucopyra-
D
nosylamine (9).—The tetraacetate 5 (100 mg,
0.21 mmol) was dissolved in a 5:1:1 mixture
of MeOH–water–Et3N (7 mL) and the mix-
ture was stirred overnight at rt. The solvent
was removed and the residue was purified by
flash column chromotagraphy using a 10:1
mixture of EtOAc–MeOH as the eluant to
give a yellow solid (45 mg, 69%, a:b 1:1.75).
1H NMR (D2O): 3.03 (ddd, 1 H, J5,6A 3.2,
J5,6B 5.4 Hz, H-5 a), 3.10 (ddd, 1 H, J5,6 3.5,
J5,6B 5.7 Hz, H-5 b), 3.40 (dd, 1 H, J2,3 =J3,4
9.1 Hz, H-3b), 3.63 (dd, 1 H, J4,5 10.5 Hz,
H-4b), 3.68 (dd, 1 H, J3,4 9.1, J4,5 10.5 Hz,
H-4a), 3.77 (m, 3 H, H-2b, H-3a, H-6Ba),
3.84 (dd, 1 H, J6A,6B 11.9 Hz, H-6Bb), 3.90
(m, 2 H, H-6Aa, H-6Ab), 4.09 (dd, 1 H, J2,3
9.9 Hz, H-2a), 4.84 (d, 1 H, J1,2 9.7 Hz,
H-1b), 5.01 (d, 1 H, J1,2 5.0 Hz, H-1a), 6.90
(m, 4 H, Ar), 8.1 (m, 4 H, Ar). 13C NMR
(CD3OD): 45.57 (C-5a), 48.00 (C-5b), 58.85
(C-1a, JC-1–H-1 152.0), 59.60 (C-1b, JC-1–H-1
152.1 Hz), 62.39 (C-6a/b), 75.48 (C-4b),
75.62, 75.76 (C-4a/C-2a/C-3a), 77.62 (C-2b),
79.74 (C-3b), 113.45 (C-o b), 114.06 (C-o a),
126.60 (C-p a), 126.87 (C-p b), 139.49 (C-m
a/b), 154.30 (C-i a/b). Anal. Calcd for
C12H16N2O6S: C, 45.56; H, 5.10; N, 8.86.
Found: C, 45.31; H, 5.15; N, 8.90.
Acknowledgements
We are grateful to the Natural Sciences and
Engineering Research Council of Canada for
financial support and to NATO for an Inter-
national Collaborative Research Grant No.
931165.
References
[1] K. Hiromi, K. Takahashi, Z.-i. Hamauzu, S. Ono, J.
Biochem., 59 (1966) 469–475.
[2] M.R. Sierks, C. Ford, P.J. Reilly, B. Svensson, Protein
Eng., 2 (1989) 621–625.
[3] J. Ermer, K. Rose, G. Hu¨bner, A. Schellenberger, Biol.
Chem. Hoppe-Seyler, 374 (1993) 123–128.
[4] B. Svensson, T.G. Pedersen, I. Svendsen, T. Sakai, M.
Ottesen, Carlsberg Res. Commun., 47 (1982) 55–69.
[5] B. Svensson, K. Larsen, I. Svendsen, E. Boel, Carlsberg
Res. Commun., 48 (1983) 529–544.
[6] B. Svensson, K. Larsen, A. Gunnarsson, Eur. J.
Biochem., 154 (1986) 497–502.
[7] A. Aleshin, A. Golubev, L.M. Firsov, R.B. Honzatko, J.
Biol. Chem., 267 (1992) 19291–19298.
[8] E.M.S. Harris, A.E. Aleshin, L.M. Firsov, R.B. Honza-
tko, Biochemistry, 32 (1993) 1618–1626.
[9] A.E. Aleshin, L.M. Firsov, R.B. Honzatko, J. Biol.
Chem., 269 (1994) 15631–15639.
p-Trifluoromethyl-N-phenyl h/i-5-thio- -
D
glucopyranosylamine (10).—The tetraacetate 6
(250 mg, 0.49 mmol) was dissolved in a 5:1:1
mixture of MeOH–water–Et3N (21 mL) and
the mixture was stirred at rt overnight. The
solvent was removed and the residue was
purified by flash column chromotagraphy us-
ing a 10:1 mixture of EtOAc–MeOH as the
[10] B. Stoffer, A.E. Aleshin, L.M. Firsov, B. Svensson, R.B.
Honzatko, FEBS Lett., 358 (1995) 57–61.

156
K.D. Randell et al. / Carbohydrate Research 321 (1999) 143–156
[11] M.R. Sierks, C. Ford, P.J. Reilly, B. Svensson, Protein
Eng., 3 (1990) 193–198.
[30] H. Driguez, B. Henrissat, Tetrahedron Lett., 22 (1981)
5061–5062.
[12] M.R. Sierks, C. Ford, P.J. Reilly, B. Svensson, Protein
Eng., 6 (1993) 75–79.
[13] M.R. Sierks, B. Svensson, Biochemistry, 32 (1993) 1113–
1117.
[14] T.P. Frandsen, C. Dupont, J. Lehmbeck, B. Stoffer,
M.R. Sierks, R.B. Honzatko, B. Svensson, Biochemistry,
33 (1994) 13808–13816.
[15] T.P. Frandsen, T. Christensen, B. Stoffer, J. Lehmbeck,
C. Dupont, R.B. Honzatko, B. Svensson, Biochemistry,
34 (1995) 10162–10169.
[16] B.W. Sigurskjold, C.R. Berland, B. Svensson, Biochem-
istry, 33 (1994) 10191–10199.
[17] C.R. Berland, B.W. Sigurskjold, B. Stoffer, T.P. Frand-
sen, B. Svensson, Biochemistry, 34 (1995) 10153–10161.
[18] M.R. Sierks, B. Svensson, Protein Eng., 5 (1992) 185–
188.
[19] R.U. Lemieux, U. Spohr, M. Bach, D.R. Cameron, T.P.
Frandsen, B.B. Stoffer, B. Svensson, M.M. Palcic, Can.
J. Chem., 74 (1996) 319–335.
[20] T.P. Frandsen, B.B. Stoffer, M.M. Palcic, S. Hof, B.
Svensson, J. Mol. Biol., 263 (1996) 79–89.
[21] (a) S. Chiba, in Amylase Research Society of Japan
(Eds.), Handbook of Amylases and Related Enzymes,
Pergamon Press, Oxford, 1988, pp. 104–116. (b) T.P.
Frandsen, B. Svensson, Plant Mol. Biol., 37 (1998) 1–13.
[22] K. Yoshikawa, K. Yamamoto, S. Okada, Biosci. Biotech.
Biochem., 58 (1994) 1392–1398.
[23] M.L. Sinnot, Chem. Re6., 90 (1990) 1171–1202.
[24] J.D. McCarter, S.G. Withers, Curr. Opin. Struct. Biol., 4
(1994) 885–892.
[31] C.-W. Chiu, R.L. Whistler, J. Org. Chem., 38 (1973)
832–834.
[32] G. Excoffier, D. Gagnaire, J.-P. Utille, Carbohydr. Res.,
39 (1975) 368–373.
[33] (a) F. Ni, Prog. NMR Spectr., 26 (1994) 517–606. (b) T.
Peters, B.M. Pinto, Curr. Opin. Struct. Biol., 6 (1996)
710–720.
[34] D.D. Perrin, Dissociation Constants of Organic Bases in
Aqueous Solution, Butterworths, London, 1965.
[35] T.P. Frandsen, M.M. Palcic, B. Svensson, unpublished
data.
[36] K. Olsen, U. Christensen, M.R. Sierks, B. Svensson,
Biochemistry, 32 (1993) 9686–9693.
[37] T.P. Frandsen, M.M. Palcic, C. Dupont, B. Svensson,
Carbohydr. Res., 314 (1998) 127–133.
[38] A.E. Aleshin, B. Stoffer, L.M. Firsov, B. Svensson, R.B.
Honzatko, Biochemistry, 35 (1996) 8319–8328.
[39] A.J. Clarke, B. Svensson, Carlsberg Res. Commun., 49
(1984) 559–566.
[40] B. Stoffer, T.P. Frandsen, P.K. Busk, P. Schneider, I.
Svendsen, B. Svensson, Biochem. J., 292 (1993) 197–202.
[41] J.D. Fox, J.F. Robyt, Anal. Biochem., 195 (1991) 93–96.
[42] M.M. Palcic, T. Skrydstrup, K. Bock, N. Le, R.U.
Lemieux, Carbohydr. Res., 250 (1993) 87–92.
[43] R.J. Leatherbarrow, ENZFITTER: A Non-linear Regression
Data Analysis Program for IBM PC, Elsevier, Amster-
dam, The Netherlands, 1987.
[44] C. Mattos, D.A. Giammona, G.A. Petsko, D. Ringe,
Biochemistry, 34 (1995) 3193–3203.
[45] A. Yamamoto, K. Tomoo, M. Doi, H. Ohishi, M. Inoue,
T. Ishida, D. Yamamoto, S. Tsuboi, H. Okamoto, Y.
Okada, Biochemistry, 31 (1992) 11305–11309.
[46] A.T. Brunger, X-PLOR, Version 3.1: A System for X-ray
Crystallography and NMR, Yale University Press, New
Haven, CT, 1992.
[47] D. Marion, K. Wu¨thrich, Biochem. Biophys. Res. Com-
mun., 113 (1983) 967–974.
[48] T. Scherf, J. Anglister, Biophys. J., 64 (1993) 754–761.
[25] B. Svensson, M. Sogaard, J. Biotechnol., 29 (1993) 1–37.
[26] K. Bock, B.W. Sigurskjold, Stud. Nat. Prod. Chem., 7
(1990) 29–86.
[27] G. Legler, Ad6. Carbohydr. Chem. Biochem., 48 (1990)
319–384.
[28] S. Mehta, J.S. Andrews, B.D. Johnston, B. Svensson,
B.M. Pinto, J. Am. Chem. Soc., 117 (1995) 9783–9790.
[29] J.S. Andrews, T. Weimar, T.P. Frandsen, B. Svensson,
B.M. Pinto, J. Am. Chem. Soc., 117 (1995) 10799–10804.
.