Synthesis, enzymatic activity, and X-ray crystallography of an unusual class of amino acids

Article abstract of DOI:10.1016/j.bmc.2006.09.004

The synthesis of two novel amino acids, nitrogen analogues of the naturally occurring glycosidase inhibitor, salacinol, containing a carboxylate inner salt are described, along with the crystal structure of one of these analogues in the active site of Dro

Full text of DOI:10.1016/j.bmc.2006.09.004

Bioorganic & Medicinal Chemistry 14 (2006) 8332–8340
Synthesis, enzymatic activity, and X-ray crystallography
of an unusual class of amino acids
Wang Chen,a Douglas A. Kuntz,b Tamika Hamlet,b Lyann Sim,b David R. Roseb
and B. Mario Pintoa,*
aDepartment of Chemistry, Simon Fraser University, Burnaby, BC, Canada V5A 1S6
bDepartment of Medical Biophysics, University of Toronto and Division of Cancer Genomics and Proteomics,
Ontario Cancer Institute, Toronto, Ont., Canada M5G 1L7
Received 18 July 2006; revised 28 August 2006; accepted 7 September 2006
Available online 28 September 2006
Abstract—The synthesis of two novel amino acids, nitrogen analogues of the naturally occurring glycosidase inhibitor, salacinol,
containing a carboxylate inner salt are described, along with the crystal structure of one of these analogues in the active site of
Drosophila melanogaster Golgi mannosidase II (dGMII). Salacinol, a naturally occurring sulfonium ion, is one of the active prin-
cipals in the aqueous extracts of Salacia reticulata that are traditionally used in Sri Lanka and India for the treatment of diabetes.
The synthetic strategy relies on the nucleophilic attack of 2,3,5-tri-O-benzyl-1,4-dideoxy-1,4-imino L- or D-arabinitol at the least
hindered carbon of 5,6-anhydro-2,3-di-O-benzyl-L-ascorbic acid to yield coupled adducts. Deprotection, stereoselective catalytic
reduction, and hydrolysis of the coupled products give the target compounds. The compound derived from D-arabinitol inhibits
dGMII, one of the critical enzymes in the glycoprotein processing pathway, with an IC50 of 0.3 mM. Inhibition of GMII has been
identified as a target for control of metastatic cancer. An X-ray crystal structure of the complex of this compound with dGMII pro-
vides insight into the requirements for an effective inhibitor. The same compound inhibits recombinant human maltase glucoamy-
lase, one of the key intestinal enzymes involved in the breakdown of glucose oligosaccharides in the small intestine, with a Ki value of
21 lM.
ꢀ 2006 Published by Elsevier Ltd.
1. Introduction
Many alkaloid sugar mimics with a nitrogen in the ring
have been isolated from plants and microorganisms,
and inhibit various glycosidases.2–4 1-Deoxynojirimycin
(1), which is a D-glucose analogue with an NH group
in place of the ring oxygen atom, has been shown to
inhibit intestinal a-glucosidases and pancreatic a-amy-
lase both in vitro and in vivo, as well as a-glucosidases
I and II involved in N-linked oligosaccharide process-
ing.5 Two N-alkylated analogues of deoxynojirimycin,
namely miglitol (2) and N-butyldeoxynojirimycin (3),
are currently in use as drugs for the treatment of Type
II diabetes and Gaucher’s disease, respectively. Both
drugs act by inhibition of glucosidase enzymes. 1,4-
Dideoxy-1,4-imino-D-arabinitol (D-AB1) (4), which
was first isolated from the fruits of the legume Angylo-
calyx boutiquenus, was found to be a potent inhibitor
of hepatic glycogen phosphorylase.6 Its synthetic
L-enantiomer (L-AB1) (5) is a powerful inhibitor of
mammalian a-D-glucosidases.7,8 The naturally occur-
ring glycosidase inhibitor acarbose (6),9 which contains
a nitrogen atom in one of the linkages between the
Glycosidases are involved in several important biologi-
cal processes, such as digestion, the biosynthesis of gly-
coproteins, and the catabolism of glycoconjugates.1
Since glycosidase inhibitors have shown antiviral, insect
antifeedant, antidiabetic, and anticancer effects, as well
as immune modulatory properties, they have attracted
considerable attention. The transition-state structure in
the enzyme-mediated hydrolysis of glycosides is believed
to be the oxacarbenium ion intermediate with a distort-
ed conformation. Thus, mimicking this distorted, posi-
tively charged species is one factor that should lead to
an effective inhibitor of glycosidase enzymes.
Keywords: Glycosidase inhibitors; Salacinol; Nitrogen analogues;
Carboxylate counterions; Amino acids; Iminoarabinitols; Epoxide;
Amino acids; L-Ascorbic acid; Stereoselective catalytic reduction;
Golgi mannosidase II; Human maltase glucoamylase.
*
Corresponding author. Tel:. +1 604 291 4152; fax: +1 604 291
4860; e-mail: bpinto@sfu.ca
0968-0896/$ - see front matter ꢀ 2006 Published by Elsevier Ltd.
doi:10.1016/j.bmc.2006.09.004

W. Chen et al. / Bioorg. Med. Chem. 14 (2006) 8332–8340
8333
H
N
H
N
OH
HO
HO
N
1 R = H
R
HO
HO
HO
OH
HO
OH
2 R = CH2CH2OH
3 R = CH2CH2CH2CH3
OH
4
5
sugar and pseudosugar units, is the highest-affinity car-
bohydrate analogue for a binding protein and has also
been used for the treatment of type-2 diabetes.10,11 It is
generally believed that this strong binding originates
from electrostatic interactions of the positively charged,
protonated nitrogen atom with carboxylate residues in
the enzyme active site.5
than the singly charged, anionic species.17 Thus, the
zwitterionic salacinol should be quite different from con-
ventional glycosidase inhibitors which mimic just the
positive charge of the transition state. Structural modifi-
cation of salacinol represents a promising approach in
the search for new glycosidase inhibitors. One strategy
is to replace the sulfur atom in salacinol with a nitrogen
atom, and we have reported the synthesis of nitrogen
analogues (11 and 12) of salacinol and their evaluation
as glycosidase inhibitors.18 Compound 11 also selective-
ly inhibits the lysosomal a-glucosidases.19
OH
HO
HO
H3C
HO
OH
O
OH
OH
OH
HN
HO
OH
OH
OH
OH
O
O
OH
HO
S
OSO3
S
OSO3 OH
HO
OH
O
HO
O
HO
OH
OH
HO
OH
HO
OH
6
9
10
OH
OH
OH
HO
N
N
OH
HO
HO
OH
OH
NH
OSO3
OH
NH
OSO3
OH
HO
HO
7
8
HO
OH
HO
OH
11
12
A similar mode of action has been suggested for the nat-
urally occurring indolizidine alkaloids castanospermine
(7) and swainsonine (8).
The fact that salacinol has greater inhibitory activity
and specificity against a-glucosidases than the methyl
sulfonium ion (13) indicates that the sulfate group is
important.20 Yuasa et al.21 reported that docking of
salacinol into the binding site of glucoamylase indicat-
ed close contacts between the sulfate ion with Arg305.
Crystallographic analysis of the interactions of Dro-
sophila melanogaster Golgi a-mannosidase II (dGMII)
with salacinol and its analogues shows that the sulfate
group does interact with residues in the enzyme active
site.22 Compound 14, isolated from a marine sponge in
Japan, was also reported to be a strong inhibitor of
a-glucosidase.23 The sulfate groups in 14 may play a
role similar to that proposed for the sulfate group of
salacinol.21
Sulfonium-ion mimics of the oxacarbenium ion contain
a permanent positive charge, which could make strong
interactions with the active-site carboxylate residues.
The most interesting glycosidase inhibitors in the form
of cyclic sulfonium ions are perhaps the naturally occur-
ring compounds salacinol (9) and kotalanol (10).12,13
The a-glucosidase inhibitory activity of salacinol was
confirmed to be as strong as that of acarbose, which is
used clinically.14 The zwitterionic structure of salacinol
is unique in that a negative charge is positioned at the
sulfate group and the positive charge is centered at the
sulfur atom.12 Molecular dynamics simulations have
shown that enzyme charge distribution plays an impor-
tant role in guiding charged ligands to the active site of
Torpedo californica acetylcholinesterase.15 Zhou et al.
have shown that the electrostatic potential within the ac-
tive site can be used to predict the electrostatic rate
enhancement for acetylcholinesterase-substrate bind-
ing.16 Of note, zwitterionic inhibitors have been predict-
ed to bind to a neuraminidase enzyme more effectively
An intriguing question is whether the corresponding car-
boxylate analogues of salacinol will act as inhibitors of
glucosidases. We now report the synthesis of novel ami-
no acids that are nitrogen analogues (15 and 16) of sal-
acinol containing a carboxylate inner salt.

8334
W. Chen et al. / Bioorg. Med. Chem. 14 (2006) 8332–8340
I
S
HO
HO
OH
13
O
O
OSO3Na
OSO3Na
N
NaO3SO
O
14
OH OH
OH OH
CO
2
CO H
2
Cl
OH OH
NH
OH OH
NH
HO
HO
HO
OH
HO
OH
15
16
2. Results and discussion
the resulting inorganic salts were removed using Sepha-
dex G-10 chromatography to yield 15. The overall yield
for the two steps was 78%. The structure of the zwitter-
ion 15 was confirmed by MALDI-TOF mass spectro-
metry, microanalysis data, and 1H and 13C NMR
spectroscopy.
2.1. Synthesis
Retrosynthetic analysis indicated that amino acids A
could be obtained by alkylation of the iminoarabinitols
B at the nitrogen atom (Scheme 1). The alkylating agent
could be an epoxide C, whereby regioselective attack of
the amine at the least hindered primary center should af-
ford the desired amino acids.24 The epoxide C could be
synthesized from inexpensive vitamin C (D).
Compound 16, the diastereomer of 15, was similarly -
obtained by reaction of the amine 26 with the epoxide
20 to produce the protected compound 29 in 74% yield
(Scheme 5). Deprotection, stereoselective catalytic
reduction, and hydrolysis, and exchange of Na+ ion with
excess cation-exchange resin gave compound 16 in 62%
yield. In this case, the compound was obtained as a chlo-
ride salt, as confirmed by MALDI-TOF mass spectro-
metry, microanalysis data, and 1H and 13C NMR
spectroscopy.
The epoxide 20 was synthesized using a simplified proce-
dure of Raic-Malic et al. (Scheme 2).25 The iminoarabin-
itols 23 and 26 were synthesized from D-xylose and L-
xylose, respectively, following a similar strategy that has
been described previously in the literature (Scheme 3).26–28
Coupling of 2,3,5-tri-O-benzyl-1,4-dideoxy-1,4-imino-L-
arabinitol 23 with the benzyl-protected L-ascorbic acid
epoxide 20 in dry acetonitrile at 70 ꢁC gave the protected
compound 27 in 82% yield (Scheme 4). No side products
were obtained. Debenzylation of the coupled product 27
by hydrogenolysis and subsequent stereoselective cata-
lytic reduction of the C40–C50 double bond of the
L-ascorbic acid moiety, using a widely employed proce-
dure,29–31 afforded 28. Catalytic hydrogenation of the
L-ascorbic acid was reported to proceed with complete
diastereoselectivity.29,30 The reduction of the double
bond in 27 was monitored by MALDI-TOF mass spec-
trometry as a hydrogen chloride salt. Even though a
high pressure of H2 was used, the reduction was com-
plete only after 4 days. Without further purification,
the crude compound 28 was treated with aqueous potas-
sium carbonate. After hydrolysis of the lactone ring in
28 and neutralization of potassium carbonate with acid,
NMR spectra were performed on samples of com-
pounds 15 and 16 in deuterated water, made basic with
small amounts of sodium deuteroxide to give the corre-
sponding amines, to ensure the peaks were more defined.
We believe that the broadening of the peaks in the ab-
sence of base is due to chemical exchange between the
ammonium salts and the corresponding tertiary amines,
a process that is in the intermediate-exchange regime on
the NMR time scale. In the presence of base, only the
rapidly inverting tertiary amines are present, and con-
formationally averaged NMR spectra in the fast-
exchange regime are observed.
2.2. Enzyme inhibitory activity
We measured the enzyme inhibitory activity of the
amino acids 15 and 16. Compound 16 inhibits recombi-
nant human maltase glucoamylase (MGA), a critical

W. Chen et al. / Bioorg. Med. Chem. 14 (2006) 8332–8340
8335
OH OH
CO2
H
N
O
OH
HO
O
OH OH
O
NH
O
O
BnO
HO
BnO
OBn
BnO
HO
OBn
OH
HO
OH
A
B
C
D
Scheme 1.
OH
OH
OH
O
Ts
O
O
O
O
HO
HO
O
O
O
O
O
K CO
TsCl
2
3
BnBr, K CO
2
3
Pyridine
CH CN
3
DMF
HO
BnO
18 (90%)
BnO
19 (70%)
OH
OBn
OBn
BnO
OBn
17
20 (81%)
Scheme 2.
H
OMs
N
OMs
N
BnO
BnO
5 steps
BnO
D-xylose
ref (26)
allyl amine
OBn
(PPh3)3RhCl
OBn
22 (82%)
BnO
BnO
OBn
BnO
21
23 (74%)
H
OMs
N
OMs
N
BnO
BnO
(PPh3)3RhCl
OBn
5 steps
BnO
L-xylose
ref (26)
allyl amine
BnO
OBn
BnO
OBn
BnO
25
26 (61% for two steps)
24
Scheme 3.
OH
O
O
H
O
O
N
N
O
BnO
H2 , Pd/C
BnO
70 ˚C
BnO
OBn
OBn
+
CH3CN
AcOH/H2O
BnO
BnO
OBn
OBn
BnO
23
20
(82%)
27
OH OH
OH
CO
O
2
O
1. K2CO3
2. Sephadex G-10
OH OH
NH
N
HO
HO
HO
OH
HO
HO
OH
15 (78% for two steps)
OH
28
Scheme 4.

8336
W. Chen et al. / Bioorg. Med. Chem. 14 (2006) 8332–8340
OH
O
O
H
O
O
N
N
O
BnO
BnO
H
, Pd/C
OBn
2
70 ˚C
BnO
OBn
+
CH CN
AcOH/H O
3
2
BnO
BnO
OBn
OBn
BnO
(
29
)
26
20
OH OH
OH
CO H
O
2
Cl
O
1. NaOH
OH OH
N
NH
HO
HO
2. Ion exchange
resin
HO
OH
OH
HO
HO
OH
(62 for two steps)
16
30
Scheme 5.
fold) over the inhibition measured for other salacinol
analogues which all inhibited dGMII with an IC50 of
approximately 7.5 mM (9, 11, 31–33, Scheme 6).22 Com-
pound 15 is not active on either enzyme; similar results
were obtained for other salacinol analogues derived
from anhydro-L-heteroarabinitol moieties and presum-
ably reflect minimal contacts of the enantiomeric five-
membered rings in the enzyme active sites.18,22
OH
OH
OH
OH
X
OSO3
HO
Se
OSO3
HO
HO
OH
HO
OH
32 X = S
33 X = Se
31
2.3. X-ray crystallography
Scheme 6.
We have solved the crystal structure of 16 bound in the
active site of dGMII. Statistics for data collection and
refinement are presented in Table 1. The electron density
of the bound compound 16 is shown in Figure 1. Close
intestinal glucosidase involved in the processing of oli-
gosaccharides of glucose into glucose itself, with a Ki
value of 21 lM. Salacinol itself has a Ki value of
0.2 lM. Compound 16 is also active against Drosophila
melanogaster Golgi a-mannosidase II (dGMII) with an
IC50 of 0.3 mM. This is a significant improvement (25-
˚
interactions (with a distance of less than 3.2 A) are high-
lighted in Figure 2 and the corresponding distances in
comparison with the nanomolar inhibitor swainsonine
are detailed in Table S1 (Supporting information).
Table 1. Statistics for data collection and refinement
PDB code/HET symbol
2FYV/W72
Space Group
Cell dimensions (A)
Data collection (values in parentheses represent high resolution shell)
P212121
68.81 · 108.64 · 137.38
˚
˚
Resolution (A)
30–1.90 (1.95–1.90)
81152/5.7 (5388/2.7)
12.4 (2.5)
Unique reflections/redundancy
I/sigma I
% completeness
R merge
99.6 (95.4)
0.092 (0.41)
13.3
2
Wilson B (A )
˚
Structure refinement
Rworkt/Rfree (reflections for Rfree
)
Amino acids/alternate conformers
0.162/0.212 (1848)
1044/10
1048/49
Water molecules/heteroatoms
˚
rmsd bonds (A)/rmsd angles (ꢁ)
Average B factors (A )
0.2/1.9
2
˚
Overall
Protein main chain/side chain
Water
16.8
14.5/16.4
26.7
Inhibitor (range)
Zn/MPD/PO4
25.6(10–44)
10.8/21.4/42.9

W. Chen et al. / Bioorg. Med. Chem. 14 (2006) 8332–8340
8337
other salacinol analogues, and it is these novel interac-
tions which may account for the increased potency of
16 in comparison to its parent compound 11. The
hydroxyl groups of this chain form extensive contacts
with both side chains and water molecules in the active
site. The electron density in this region is more ill defined
than that of the ring region and this indicates that there
is mobility of the chain. Flexibility of this region is also
reflected in the temperature (B) factors which are a mea-
sure of atomic mobility. B-factors in the ring region are
2
˚
in the range of 10–15 A (the zinc bound OH1 is below
2
10 A ) while those in the tail region approach 44 A .
2
˚
˚
Hydrogen bonds in the acyclic region occur between O8
and the catalytic nucleophile D204, as well as R228 and
Y269. O9 interacts with the acid–base catalyst residue
D341 and two waters. O11 hydrogen bonds to one water
while O12 makes close contacts with two waters. The
carbonyl O13 interacts with D340 and 2 water mole-
cules, one of which is shared with O9 and the other
shared with O12, D340 and D270.
Figure 1. Stereoview of 16 in the active site of Drosophila melanogaster
Golgi mannosidase II and its surrounding electron density. The
electron density was determined as a simulated annealing omit map
(Fo ꢀ Fc) and is contoured at 2 sigma (red) or 5 sigma (blue). The
active-site zinc ion is shown in gray.
Similar to the other salacinol analogues (and in contrast
to most other inhibitors bound to dGMII) only a single
hydroxyl group (OH2) interacts with the active-site zinc
atom. Also, as seen in other salacinol analogues, Tryp-
tophan95 stacks on top of the ring portion of 16, and
the ring hydroxyl groups form hydrogen bonds with
aspartate residues (D92, D204, and D472 with OH1
and D472 with OH2) and tyrosine (Y727 OH with
OH2). The C6 OH forms hydrogen bonds with the car-
bonyl oxygen of R876 as well as a bound water mole-
cule. D204 makes a hydrogen bond with the nitrogen
atom of compound 16.
Figure 3A shows an overlay of 16 (this work) and swains-
onine (from PDB 1HWW) bound in the crystal structure
of dGMII. Although the reason for the potency of
swainsonine has not been clearly determined, it is one
of the best inhibitors of dGMII, with an IC50 value in
the range of 20–40 nM, and is believed to closely mimic
the oxacarbenium ion which occurs in the reaction path-
way. The position of the nitrogen moiety, which is de-
signed to serve as the mimic of the positive charge on
the oxacarbenium ion, is almost identical in the two
bound structures. However, in contrast with swainso-
nine, only a single hydroxyl group of 16 is in contact with
the active site zinc ion. As well, the orientation of the sec-
ond hydroxyl group, which forms hydrogen bonds with
It is in the acyclic chain of 16 that the interactions of the
inhibitor exhibit the most significant differences from the
˚
Figure 2. Interactions of compound 16 with Drosophila GMII. Only interactions less than 3.2 A are indicated. The zinc ion in the active site is
depicted as a black ball and water molecules are shown as gray spheres. Distances are given in Angstrom units. Numbering of the inhibitor is as it
occurs in the PDB file.

8338
W. Chen et al. / Bioorg. Med. Chem. 14 (2006) 8332–8340
Figure 3. Overlay of compounds bound in the active site of dGMII. Compound 16 (cyan) is overlaid with A. Swainsonine (pink, PDB 1HWW) B.
Ghavamiol 11 (magenta, PDB 1TQU) C. Salacinol diastereomer 32 (gray, PDB 1TQT) or D. N-Benzyl mannostatin (green, PDB 2F7P).
D472 and Y727, differs in the two structures, and it is
possible that this geometry is not ideal for forming
strong interactions. While the position of the head group
of 16 is comparable to those of the other salacinol ana-
logues22 the region of space occupied by the acyclic tail
region is quite different. Figure 3 shows overlays of 16
bound in the crystal structure of dGMII with bound gha-
vamiol 11 (Fig. 3B) or the diastereomer of salacinol 32
(Fig. 3C) solved previously (PDBs 1TQU and 1TQT22).
In both cases the position of the sulfate group is quite dif-
ferent from that of the carboxyl group, and the space
through which the aliphatic chain passes is also quite dif-
ferent. Interestingly, the region of the active site through
which the aliphatic chain of 16 passes is very much com-
parable to that of a recently solved benzyl-mannostatin
A–dGMII complex.32 The overlay of the two complexes
(Fig. 3D) shows them to intertwine, although the nature
of the interactions formed by the two tail moieties is dif-
ferent. The benzyl tail reduced the potency of the man-
nostatin to which it was attached33, while in the present
case the carboxylate tail greatly increased the inhibitory
activity of the salacinol head group. Given the much bet-
ter inhibitory properties of the mannostatin derivatives
(nM) in comparison to the salacinol derivatives (mM)
it is possible that attachment of a similar carboxylate tail
to a mannostatin head group would lead to a very effec-
tive mannosidase inhibitor.
analysis of MGA inhibition was performed using malt-
ose as the substrate and measuring the release of glu-
cose. Reactions were carried out in 100 mM Mes
buffer, pH 6.5, at 37 ꢁC. The reaction was stopped by
boiling for 3 min. Twenty microliter aliquots were taken
and added to 100 lL of glucose oxidase assay reagent
(Sigma) in a 96-well plate. Reactions were developed
for 1 h and absorbance was measured at 450 nm to
determine the amount of glucose produced by MGA
activity in the reaction. All reactions were performed
in triplicate and absorbance measurements were aver-
aged to give a final result.
3.2. Enzyme kinetics
Kinetic parameters of recombinant MGA were deter-
mined using the glucose oxidase assay to follow the pro-
duction of glucose upon addition of enzyme (15 nM) at
increasing maltose concentrations (from 2.5 to 30 mM)
with a reaction time of 15 min. The Ki value was deter-
mined by measuring the rate of maltose hydrolysis by
MGA at varying inhibitor concentrations. Data were
plotted in Lineweaver–Burk plots (1/rate vs 1/[substrate])
and the Ki value was determined by the equation Ki =
Km[I]/ (Vmax)s ꢀ Km, where ‘s’ is the slope of the line.
The Ki reported was an average of the Ki values obtained
from each of the different inhibitor concentrations.
3.3. Structure determination of dGMII–16 complex
3. Experimental
Preparation of dGMII crystals soaked with 16 was car-
ried out essentially as described previously.22 In this case
however, the crystals were first washed with reservoir
buffer containing phosphate instead of Tris, to reduce
any effects of Tris binding in the active site. The crystals
were soaked for 24 h with a 2 mM solution of 16 in
phosphate containing reservoir buffer. The crystals were
3.1. Enzyme activity assays
Measurement of dGMII inhibition was carried out as
outlined previously.22 Analysis of recombinant MGA
inhibition and determination of the kinetic constants
for competitive inhibition have been described.34 Briefly,

W. Chen et al. / Bioorg. Med. Chem. 14 (2006) 8332–8340
8339
passed through phosphate-buffered cryo-solutions con-
taining 1 mM inhibitor prior to rapid freezing in a liquid
nitrogen stream. X-ray diffraction data were collected at
100 K with a Bruker X8 Proteum system consisting of a
CCD detector and a Bruker Microstar rotating anode
generator. Data were integrated and scaled using the
Proteum suite of programs (Bruker AXS, Madison
WI). Structure solution and refinement were carried
out using the programs CNS35 and O36 as previously de-
scribed.22 Diagrams were rendered in Pymol.37
ly washed with H2O. The combined filtrates were
concentrated under vacuum. Concentrated hydrochloric
acid (1 mL) was added and the mixture was concentrated
by high vacuum. The obtained solid was dissolved in
aqueous K2CO3 solution (1 mL, pH 9.0) and the mixture
was stirred for 3 h. The solution was neutralized with di-
lute hydrochloric acid and then concentrated. The residue
was purified by Sephadex G-10 chromatography column
to give 15 as an amorphous solid (104 mg, 78%). [a]D +10ꢁ
(c 0.1, H2O). 1H NMR (D2O, pH = 12.0): d 4.03 (1H, d,
J4 ;5 ¼ 4:5 Hz, H-50), 3.98 (1H, br d, H-2), 3.83 (1H, d,
H-40), 3.79 (2H, br d, H-3, H-20), 3.64 (1H, m, H-30),
3.61 (2H, m, 2H-5), 2.99 (1H, d, J1a,1b = 11.1 Hz, H-1a),
0
0
3.4. N-Allyl-2,3,5-tri-O-benzyl-1,4-dideoxy-1,4-imino-L-
arabinitol (22), N-allyl-2,3,5-tri-O-benzyl-1,4-dideoxy-
1,4-imino-D-arabinitol (25), 2,3,5-tri-O-benzyl-1,4-dide-
oxy-1,4-imino-L-arabinitol (23), 2,3,5-tri-O-benzyl-1,4-
dideoxy-1,4-imino-D-arabinitol (26)
2.91 (1H, dd, J1 a;1 b ¼ 12:8 Hz, J1 a;2 ¼ 5:0 Hz, H-10a),
0
0
0
0
2.73 (1H, dd, J1b,2 = 5.0 Hz, H-1b), 2.47 (1H, m, H-4),
2.43 (1H, dd, J1 b;2 ¼ 7:1 Hz, H-10b). 13C NMR (D2O,
0
pH 12.0): d 179.13 (C-60), 76.50 (C-3), 73.35 (C-2),
71.45 (C-50), 70.27 (C-40), 69.86 (C-4), 69.35 (C-30),
68.39 (C-20), 58.46 (C-5), 57.44 (C-1), 54.98 (C-10). MAL-
DI-TOF MS: m/e 334.47 (M++Na), 312.43 (M++H).
Anal. Calcd for C11H21O9N: C, 42.44; H, 6.80; N, 4.50.
Found: C, 42.19; H, 6.66; N, 4.36.
Compounds 22, 23, 25, and 26 were synthesized accord-
ing to the original literature procedures.26–28
3.5. 60-((2,3,5-Tri-O-benzyl-1,4-dideoxy-1,4-imino-L-ara-
binitol)-4-N-yl)-20,30-di-O-benzyl-60-deoxy-L-ascorbic
acid (27)
3.7. 60-((2,3,5-Tri-O-benzyl-1,4-dideoxy-1,4-imino-D-ara-
binitol)-4-N-yl)-20,30-di-O-benzyl-60-deoxy-L-ascorbic
acid (29)
A mixture of 2,3,5-tri-O-benzyl-1,4-dideoxy-1,4-imino-
L-arabinitol 23 (404 mg, 1.0 mmol) and 5,6-anhydro-
2,3-di-O-benzyl-L-ascorbic acid 20 (340 mg, 1.0 equiv)
was dissolved in dry CH3CN (5 mL). The mixture was
stirred in a round-bottomed flask in an oil-bath (70 ꢁC)
overnight. The solvent was removed under reduced pres-
sure, and the product was purified by column chromatog-
raphy (hexanes–EtOAc, 3:1) to afford 27 (612 mg, 82%) as
a yellow oil. [a]D +23ꢁ (c 0.6, CH2Cl2). 1H NMR (CDCl3):
d 7.11–7.30 (25H, m, Ar), 5.12 and 5.04 (2H, 2d,
JA,B = 11.8 Hz, C@C–OCH2Ph), 5.01 and 4.98 (2H, 2d,
A mixture of 2,3,5-tri-O-benzyl-1,4-dideoxy-1,4-imino-
D-arabinitol 26 (444 mg, 1.1 mmol) and 5,6-anhydro-
2,3-di-O-benzyl-L-ascorbic acid 20 (374 mg, 1.0 equiv)
was dissolved in dry CH3CN (5 mL). The mixture was
stirred in a round-bottomed flask in an oil-bath (70 ꢁC)
overnight. The solvent was removed under reduced pres-
sure, and column chromatography (hexanes–EtOAc, 3:1)
of the crude product gave 29 (605 mg, 74%) as a yellow
oil. [a]D +6ꢁ (c 0.7, CH2Cl2). 1H NMR (CDCl3): d 7.13–
7.30 (25H, m, Ar), 5.22 and 5.04 (2H, 2d, JA,B = 11.7 Hz,
C@C–OCH2Ph), 5.11 and 5.05 (2H, 2d, JA,B = 11.3 Hz,
C@C–OCH2Ph), 4.51 (1H, br d, H-30), 4.50–4.40 (6H,
m, 3CH2Ph), 3.96 (1H, dt, J1a,2 = 1.7 Hz, J1b,2 = 5.5 Hz,
H-2), 3.88 (1H, dd, J2,3 = 1.8 Hz, J3,4 = 4.4 Hz, H-3),
3.87 (1H, m, H-20), 3.53 (1H, dd, J4,5a = 5.3 Hz,
J5a,5b = 9.8 Hz, H-5a), 3.51 (1H, dd, J4,5b = 5.6 Hz, H-
5b), 3.23 (1H, d, J1a,1b = 10.4 Hz, H-1a), 3.17 (1H, dd,
0
0
JA,B = 11.3 Hz, C@C–OCH2Ph), 4.57 (1H, d, J2 ;3
¼
1:5 Hz, H-30), 4.44 and 4.42 (2H, 2d, JA,B = 6.3 Hz,
CH2Ph), 4.39 and 4.36 (2H, 2d, JA,B = 9.4 Hz, CH2Ph),
4.35 and 4.33 (2H, 2d, JA,B = 12.0 Hz, CH2Ph), 3.89
(1H, ddd, J1a,2 = 1.6 Hz, J1b,2 = 5.5 Hz, J2,3 = 1.8 Hz,
0
0
0
0
H-2), 3.84 (1H, ddd, J1 a;2 ¼ 8:0 Hz, J1 b;2 ¼ 6:0 Hz,
H-20), 3.76 (1H, dd, J3,4 = 3.9 Hz, H-3), 3.46 (1H, dd,
J4,5a = 5.9 Hz, J5a,5b = 9.8 Hz, H-5a), 3.43 (1H, dd,
J4,5b = 5.9 Hz, H-5b), 3.09 (1H, dd, J1a,1b = 10.5 Hz, H-
1a), 2.88 (1H, dd, J1 a;1 b ¼ 13:0 Hz, H-10a), 2.84 (1H, dt,
H-4), 2.77 (1H, dd, H-1b), 2.73 (1H, dd, H-10b). 13C
NMR (CDCl3): d 169.57 (C-60), 157.44 (C-40), 137.86,
137.79, 137.73, 135.85, 135.32 (5Cipso), 128.76–127.24
(25CAr), 120.68 (C-50), 84.59 (C-3), 81.63 (C-2), 75.54
(C-30), 73.60 and 72.98 (2C@C–OCH2Ph), 72.87, 71.09,
70.78 (3CH2Ph), 70.34 (C-5), 68.95 (C-4), 66.93 (C-20),
58.00 (C-1), 56.69 (C-10). MALDI-TOF MS: m/e 763.85
(M++Na), 742.075 (M++H). Anal. Calcd for
C46H47O8N: C, 74.43; H, 6.38; N, 1.89. Found: C,
74.20; H, 6.35; N, 2.14.
J1 a;1 b ¼ 12:5 Hz, J1 a;2 ¼ 10:9 Hz, H-10a), 2.88 (1H, q,
0
0
0
0
0
0
0
0
H-4), 2.64 (1H, dd, H-1b), 2.52 (1H, dd, J1 b;2 ¼ 3:5 Hz,
13
H-10b). C NMR (CDCl3): d 169.90 (C-60), 157.43 (C-
40), 138.24, 138.21, 138.19, 136.36, 135.80 (5Cipso),
128.96-127.64 (25CAr), 121.22 (C-50), 84.83 (C-3), 82.03
(C-2), 75.94 (C-30), 74.20 and 73.62 (2C@C–OCH2Ph),
73.49, 71.79, 71.38 (3CH2Ph), 70.28 (C-5), 69.12 (C-4),
66.28 (C-20), 57.40 (C-1), 57.10 (C-10). MALDI-TOF
MS: m/e 764.03 (M++Na), 741.95 (M++H). Anal. Calcd
for C46H47O8N: C, 74.43; H, 6.38; N, 1.89. Found: C,
74.27; H, 6.39; N, 2.03.
3.6. 60-((1,4-Dideoxy-1,4-imino-L-arabinitol)-4-N-ammo-
3.8. 60-((1,4-Dideoxy-1,4-imino-D-arabinitol)-4-N-ammo-
nium)-60-deoxy-L-gulonate (15)
nium)-60-deoxy-L-gulonic acid hydrochloride (16)
The protected compound 27 (300 mg, 0.4 mmol) was dis-
solved in AcOH–H2O (4:1, 6 mL) and stirred with Pd/C
(30 mg) under H2 (70 psi). After 4 days, the reaction mix-
ture was filtered through a cotton, which was subsequent-
The protected compound 29 (600 mg, 0.8 mmol) was
dissolved in AcOH–H2O (4:1, 10 mL) and stirred with
Pd/C (50 mg) under H2 (70 psi). After 4 days, the reac-
tion mixture was filtered through cotton, which was sub-

8340
W. Chen et al. / Bioorg. Med. Chem. 14 (2006) 8332–8340
7. Fleet, G. W. J.; Nicholas, S. J.; Smith, P. W. Tetrahedron
Lett. 1985, 26, 3127–3130.
8. Scofield, A. M.; Fellows, L. E.; Nash, R. J.; Fleet, G. W. J.
Life Sci. 1986, 39, 645–650.
9. Bock, K.; Sigurskjold, B. Stud. Nat. Prod. Chem 1990, 7,
29–86.
10. Holman, R. R.; Cull, C. A.; Turner, R. C. Diabetes Care
1999, 22, 960–964.
11. Jacob, G. S. Curr. Opin. Struct. Biol. 1995, 5, 605–611.
12. Yoshikawa, M.; Murakami, T.; Shimada, H.; Matsuda,
H.; Yamahara, J.; Tanabe, G.; Muraoka, O. Tetrahedron
Lett. 1997, 38, 8367–8370.
13. Yoshikawa, M.; Murakami, T.; Yashiro, K.; Matsuda, H.
Chem. Pharm. Bull. 1998, 46, 1339–1340.
14. Asano, N. Glycobiology 2003, 13(10), 93R–104R.
15. Antosiewicz, J.; Wlodek, S. T.; McCammon, J. A.
Biopolymers 1996, 39, 85–94.
16. Zhou, H.-X.; Briggs, J. M.; McCammon, J. A. J. Am.
Chem. Soc. 1996, 118, 13069–13070.
sequently washed with H2O. The combined filtrates were
concentrated under vacuum. Concentrated hydrochloric
acid (2 mL) was added and the mixture was concentrat-
ed by high vacuum. The obtained solid was dissolved in
aqueous NaOH solution (2 mL, pH 9.0) and the mixture
was stirred for 3 h. The solution was neutralized with di-
lute hydrochloric acid. Na+ ion was removed with excess
Amberlite IR-120-P (H+ form) and the resin was re-
moved by filtration. The aqueous solution was concen-
trated to give 16 as an amorphous solid (165 mg,
62%). [a]D +2ꢁ (c 0.5, H2O). 1H NMR (D2O, pH
12.0): d 3.94 (1H, d, J4 ;5 ¼ 4:8 Hz, H-50), 3.89 (1H, m,
H-2), 3.75–3.71 (3H, m, H-40, H-3, H-20), 3.57–3.52
(2H, m, 2H-5), 3.50 (1H, m, H-30), 2.90 (1H, dd,
J1a,1b = 10.4 Hz, J1a,2 = 1.1 Hz, H-1a), 2.76 (1H, dd,
0
0
J1 a;1 b ¼ 13:0 Hz, J1 a;2 ¼ 9:4 Hz, H-10a), 2.57 (1H, dd,
0
0
0
0
J1b,2 = 5.6 Hz, H-1b), 2.36 (1H, m, H-10b), 2.33 (1H,
13
m, H-4); C NMR (D2O, pH 12.0): d 179.00 (C-60),
17. Bonnet, P.; Bryce, R. A. Protein Sci. 2004, 13, 946–957.
18. Ghavami, A.; Johnston, B. D.; Jensen, M. T.; Svensson,
B.; Pinto, B. M. J. Am. Chem. Soc. 2001, 123, 6268–6271.
19. Li, Y.; Scott, C. R.; Chamoles, N. A.; Ghavami, A.; Pinto,
B. M.; Turecek, F.; Gelb, M. H. Clin. Chem. 2004, 50,
1785–1796.
79.24 (C-3), 75.95 (C-2), 74.18 (C-50), 72.71 (C-40),
72.30 (C-4, C-30), 70.06 (C-20), 60.70 (C-5), 58.99 (C-
1), 57.54 (C-10). MALDI-TOF MS: m/e 334.52
(M++Na), 312.32 (M++H). Anal. Calcd for
C11H22O9NCl: C, 38.00; H, 6.37; N, 4.03. Found: C,
38.15; H, 6.45; N, 3.86.
20. Yuasa, H.; Takada, J.; Hashimoto, H. Bioorg. Med.
Chem. Lett. 2001, 11, 1137–1139.
21. Yuasa, H.; Saotome, C.; Kanie, O. Trends Glycosci.
Glycotechnol. 2002, 14, 231–251.
22. Kuntz, D. A.; Ghavami, A.; Johnston, B. D.; Pinto, B.
M.; Rose, D. R. Tetrahedron: Asymmetry 2005, 16, 25–
32.
23. Nakano, Y.; Maki, T.; Matsunaga, S.; van Soest, R. W.
M.; Fusetani, N. Tetrahedron 2000, 56, 8977–8987.
24. Efange, S. M. N.; Michelson, R. H.; Dutta, A. K.;
Parsons, S. M. J. Med. Chem. 1991, 34, 2638–2643.
25. Raic-Malic, S.; Svedruzic, D.; Gazivoda, T.; Marunovic,
A.; Hergold-Brundic, A.; Nagl, A.; Balzarini, J.; De
Clercq, E.; Mintas, M. J. Med. Chem. 2000, 43, 4806–
4811.
26. Satoh, H.; Yoshimura, Y.; Sakata, S.; Miura, S.; Machida,
H. Bioorg. Med. Chem. Lett. 1998, 8, 989–992.
27. Veerapen, N.; Yuan, Y.; Sanders, D. R. A.; Pinto, B. M.
Carbohydr. Res. 2004, 339, 2205–2217.
Acknowledgments
We are grateful to Blair D. Johnston for helpful discus-
sion and to the Natural Sciences and Engineering
Research Council of Canada and the Canadian Insti-
tutes for Health Research for financial support.
Supplementary data
Details of X-ray crystal structure determination, Table
comparing interatomic distances in Drosophila Golgi
mannosidase II (1 page), and CIF file. Supplementary
data associated with this article can be found, in the
online version, at doi:10.1016/j.bmc.2006.09.004.
28. Overkleeft, H. S.; Wiltenburg, J.; Pandit, U. K. Tetrahe-
dron 1994, 50, 4215–4224.
29. Czarnocki, Z.; Mieczkowski, J. B.; Ziolkowski, M. Tetra-
hedron: Asymmetry 1996, 7, 2711–2720.
References and notes
30. Andrews, G. C.; Crawford, T. C.; Bacon, B. E. J. Org.
Chem. 1981, 46, 2976–2977.
31. Joseph, C. C.; Regeling, H.; Zwanenburg, B.; Chittenden,
G. J. F. Tetrahedron 2002, 58, 6907–6912.
32. Kawatkar, S. P.; Kuntz, D. A.; Woods, R. J.; Rose, D.
R.; Boons, G.-J. J. Am. Chem. Soc. 2006, 128,
8310–8319.
33. Li, B.; Kawatkar, S. P.; GeorgeS, S.; Strachan, H.;
Woods, R. J.; Siriwardena, A.; Moremen, K. W.; Boons,
G.-J. Chembiochem 2004, 5, 1127–1220.
1. For leading references Dwek, R. A. Chem. Rev. 1996, 96,
683–720; Essentials of Glycobiology; Varki, A., Cumming,
R., Esko, J., Freeze, H., Hart, G., Marth, J., Eds.; Cold
Spring Harbor Laboratory Press: Cold Spring Harbor,
NY, 1999.
2. Elbein, A. D. Annu. Rev. Biochem. 1987, 56, 497–534.
3. Fellows, L. E.; Kite, G. C.; Nash, R. J.; Simmonds, M. S.
J.; Scofield, A. M. Castanospermine, Swainsonine, and
Related Polyhydroxy Alkaloids: Structure, Distribution
and Biological Activity. In Plant Nitrogen Metabolism;
Poulton, J. E., Romero, J. T., Conn, E. E., Eds.; Plenum:
New York, 1989; pp 395–427.
4. Legler, G. Adv. Carbohydr. Chem. Biochem. 1990, 48, 319–
384.
5. Iminosugars as Glycosidase Inhibitors: Nojirimycin and
Beyond; Stutz, A. E., Ed.; Wiley-VCH: Weinheim: New
York, 1999.
34. Liu, H.; Sim, L.; Rose, D. R.; Pinto, B. M. J. Org. Chem.
2006, 71, 3007–3013.
35. Brunger, A. T.; Adams, P. D.; Clore, G. M.; DeLano, W.
L.; Gros, P.; Grosse-Kunstleve, R. W.; Jiang, J. S.;
Kuszewski, J.; Nilges, M.; Pannu, N. S.; Read, R. J.; Rice,
L. M.; Simonson, T.; Warren, G. L. Acta Crystallogr. D
Biol. Crystallogr. 1998, 54, 905–921.
36. Jones, T. A.; Zou, J. Y.; Cowan, S. W.; Kjeldgaard, M.
Acta Crystallogr. A 1991, 47, 110–119.
37. Delano, W. L. The PyMOL Molecular Graphics System
DeLano Scientific, San Carlos, CA, 2002.
6. Fosgerau, K.; Westergaard, N.; Quistorff, B.; Grunner,
N.; Kristiansen, M.; Lundgren, K. Arch. Biochem. Bio-
phys. 2000, 380, 274–284.