Structure-activity relationships in aminocyclopentitol glycosidase inhibitors

Article abstract of DOI:10.1039/b315704k

Aminocyclopentitol analogs of β-D-glucose, β-D-galactose and α-D-galactose bearing alkyl substituents as aglycon mimics on the amine function were prepared and tested for inhibition of various glycosidases. N-benzyl-β-D-gluco derivatives 1-4 and N-benzyl-β-D-galacto derivative 5 inhibited β-galactosidase and β-glucosidase. N-benzyl-α-D- galacto aminocyclopentitol 6 strongly inhibited α-galactosidase. The inhibitory activities observed were generally stronger compared to those of their primary amine analogs. A structure-activity relationship analysis was carried out including data from thirty-five different aminocyclopentitol glycosidase inhibitors. The strongest inhibitions reported for any enzyme were associated with a perfect stereochemical match between aminocyclopentitol and glycosidase, including the α- or β-configuration of the amino-group corresponding to the enzyme's anomeric selectivity.

Full text of DOI:10.1039/b315704k

View Article Online / Journal Homepage / Table of Contents for this issue
Structure–activity relationships in aminocyclopentitol glycosidase
inhibitors
Lucas Gartenmann Dickson, Emmanuel Leroy and Jean-Louis Reymond*
Department of Chemistry & Biochemistry, University of Bern, Freiestrasse 3, 3012 Bern,
Switzerland. E-mail: jean-louis.reymond@ioc.unibe.ch; Fax: ϩ41 31 631 80 57
Received 4th December 2003, Accepted 27th February 2004
First published as an Advance Article on the web 19th March 2004
Aminocyclopentitol analogs of β--glucose, β--galactose and α--galactose bearing alkyl substituents as aglycon
mimics on the amine function were prepared and tested for inhibition of various glycosidases. N-benzyl-β--gluco
derivatives 1–4 and N-benzyl-β--galacto derivative 5 inhibited β-galactosidase and β-glucosidase. N-benzyl-α--
galacto aminocyclopentitol 6 strongly inhibited α-galactosidase. The inhibitory activities observed were generally
stronger compared to those of their primary amine analogs. A structure–activity relationship analysis was carried out
including data from thirty-five different aminocyclopentitol glycosidase inhibitors. The strongest inhibitions reported
for any enzyme were associated with a perfect stereochemical match between aminocyclopentitol and glycosidase,
including the α- or β-configuration of the amino-group corresponding to the enzyme’s anomeric selectivity.
Introduction
Glycosidase inhibitors1 can be used for treating diabetes,
cancer, viral (HIV, influenza) and bacterial infections, and as
insecticides.2 In conjunction with our interest in raising
glycosidase catalytic antibodies,3,4 we have been interested in
aminocyclopentitol glycosidase inhibitors because these com-
pounds also resemble the transition state of glycosidic bond
hydrolysis. Aminocyclopentitols incorporate the entire stereo-
chemical pattern of glycopyranosides, this in an elegantly
compact manner by which each of the five cyclopentane sub-
stituents represents one of the five pyranoside substituents in a
stereodefined manner. In particular the amino group mimics
the protonated form of the leaving group oxygen atom in an
α- or β-orientation. Early studies with N-alkyl substituted
α--fuco and β--fuco configured aminocyclopentitols showed
that N-alkyl substitution enhanced inhibitory potency and
selectivity. Herein we report the synthesis and evaluation of
N-alkyl substituted aminocyclopentitols 1–6 with β-gluco,
β-galacto and α-galacto configuration (Fig. 1). A comparative
analysis of known aminocyclopentitols shows that the strongest
aminocyclopentitol inhibitor of any α- or β-glycosidase is
always an N-alkyl derivative of the stereochemically corre-
sponding aminocyclopentitol, which clearly establishes the
stereochemical design principle underlying inhibition in these
compounds.
Results and discussion
Design
Aminocyclopentitol substructures are found in a number of
natural products such as mannostatin and trehazolin.5 The
neuraminidase inhibitor BCX-1812 in clinical development to
treat influenza is also an aminocyclopentitol.6 Synthetic routes
to aminocyclopentitols were developed for the total synthesis of
Fig. 1 Structure of aminocyclopentitols.
such compounds.5 Inspired by an early report of glycosidase
inhibition,7 we have investigated aminocyclopentitols as trans-
ition state analog inhibitors of glycosidic bond cleavage.
The protonated amino group acts as a mimic of the leaving
group exocyclic C(1) oxygen atom and interacts favorably with
the enzyme’s catalytic acidic groups by electrostatic and hydro-
gen bonding interactions (Fig. 2). The complete stereochemical
pattern of the glycoside is encoded in the relative configuration
of the cyclopentitol substituents. In particular, the anomeric
β- or α-orientation of the leaving group is represented by the
relative configuration of the amino group. This feature is absent
from most other glycosidase inhibitor structures, in particular
analogs of the intermediate oxocarbonium cation where the
anomeric configuration is absent. Anomeric stereochemistry
is a key feature of glycosidases if one considers that these
enzymes are usually completely selective for either β- or
T h i s j o u r n a l i s © T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 4
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 2 1 7 – 1 2 2 6
1217

View Article Online
Fig. 2 Structural analogy between β- and α-glycosides undergoing
enzymatic cleavage and N-alkyl aminocyclopentitols in the example of
phenyl β--glucosides and α--galactoside and compounds 2 and 6.
The enzyme’s catalytic groups AH (acidic residue) and Nu (nucleophilic
residue) are usually a pair of carboxylate side chains.
α-glycosides. The cyclopentitol’s amino group can be alkylated
to provide a mimic of the aglycon leaving group.
Inhibition studies with aminocyclopentitols 7–12 bearing
a primary amine with β- and α--fuco,8 β--galacto,9 and
β--gluco10 configuration (Fig. 1) indicated that inhibitory
selectivity among the different glycosidases was governed by
the stereochemistry of the cyclopentane substituents, with
the hydroxyl group configuration corresponding to the
carbohydrate type and the amino group configuration to the
anomeric selectivity. In the case of the α--fuco aminocyclo-
pentitol 9, addition of a benzyl substituent at nitrogen to give
10 led to a dramatic enhancement of inhibitory potency against
various α--fucosidases, while the stereoisomeric β--fuco
aminocyclopentitol 7 lost its potency against α--fucosidases
upon N-alkylation to 8. To probe the generality of this anomer-
selective inhibitor design, we set out to prepare N-alkylated
β- and α-configured aminocyclopentitols either in the galacto-
or gluco- configuration since β- and α-selective glycosidases are
available in both series. During the course of this study, a series
of closely related aminocyclopentitols was reported by Jäger
et al.11
Scheme 1 Synthesis of N-alkyl β--gluco aminocyclopentitols 1–4.
Reagents and conditions: a) BnONH3Cl, MeOH, 60 ЊC, 15 h (95%); b)
Im2CS, benzene, reflux, 4 h (96%); c) AIBN, Bu3SnH, benzene, reflux, 4
h (55% 15, 35% 16); d) 12 bar H2, Pd/C, AcOH, 25 ЊC, 4 d; DMAP,
Ac2O, 20 h (80%); e) 1.2 M HCl, reflux, 12 h (100%); f ) CsOHؒH2O,
DMF, MS 4 Å, BrCH2C6H4CO2H, 80 ЊC, 12 h (16%); g) Al2O3,
PhCHO, 50 ЊC, 6 h, then MS 4 Å, NaBH4, 25 ЊC, 24 h (43%); h) as g)
with PhCH2CHO (49%); i) as g) with PhCH2CH2CHO (51%).
phenethyl and N-phenylpropyl derivatives 2–4. All compounds
were purified by preparative reverse-phase HPLC.
Our synthetic approach to N-benzyl-β--galacto amino-
cyclopentitols followed our previous synthesis of β--galacto-
aminocyclopentitol 11 (Scheme 2).9 Oxidative cleavage of the
Synthesis
The β--gluco configured aminocyclopentitols were prepared
starting from tetra-O-benzyl glucose using a radical cyclization
procedure modified from the original report of Bartlett et al.
(Scheme 1).12 Reaction with O-benzylhydroxylamine gave oxime
13. Reaction with thiocarbonyl diimidazole gave the thiocarb-
amate 14. Reaction in refluxing benzene and slow addition of
AIBN and Bu3SnH effected radical cyclization to produce a
stereoisomeric 1.6 : 1 mixture of the β--gluco and β--ido
aminocyclopentitols 15 and 16. The stereoisomers were
separated by column chromatography. The major β--gluco
isomer 15 was deprotected by catalytic hydrogenation in acetic
acid with 12 bars of H2 for 4 days, which resulted in cleavage
of the NO bond and removal of all benzyl protecting groups.
The product was isolated as the acetylated derivative 17 after
acetylation of the crude product. This simple deprotection
procedure avantageously replaced an earlier two-step process
beginning with the rather sluggish reductive cleavage of the
N–O bond with zinc–HOAc.10 The primary aminocyclopentitol
12 was obtained quantitatively from the acetylated product
17 by acidic hydrolysis. Direct alkylation with methyl (bromo-
methyl)benzoate followed by hydrolysis of the ester gave the
carboxybenzyl derivative 1. On the other hand, imine formation
with aldehydes and reduction following the procedure described
by Jäger et al. for similar compounds,11 gave the N-benzyl, N-
Scheme 2 Synthesis of N-benzyl-β-galacto aminocyclopentitol 5 from
-lyxose. Reagents and conditions: a) 2,2-dimethoxypropane, acetone,
MeOH, 0 ЊC, cat. HClO4, 5 h (84%); b) PCC (4 equiv.), benzene, reflux,
15 h (39%); c) LiCH2PO(OMe)2, THF, Ϫ80 ЊC, 2 h, then Ϫ20 ЊC, 20
min (39%); d) H2O2, MeOH, NaOH, Ϫ50 ЊC,
1 h (79%); e)
BrCH3PPh3–NaNH2, THF, 25 ЊC, 1 h (quant.); f ) benzylamine, 25 ЊC,
1.5 h, then Boc2O, NaHCO3, EtOAc–H2O (24%); g) BH3ؒTHF, THF,
0 ЊC, 2 h, then aq. NaBO3, 25 ЊC, 1.5 h (75%); h) aq. 0.3 N HCl, 60 ЊC,
12 h (quant.).
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 2 1 7 – 1 2 2 6
1218

View Article Online
Table 1 Inhibition data on glycosidases. Inhibition data in µM for compounds 1–6. IC50 values are marked (*) and were determined at [S] = 2.5 mM
and [I] = 100, 10, 1, 0.1 and 0.01 µM. Percents (%) indicate inhibition at [I] = 100 µM and [S] = 2.5 mM. No data are given when less than 5%
inhibition was observed at [I] = 100 µM and [S] = 2.5 mM. “—”: not determined. Competitive inhibition constants Ki were determined from Dixon
replots (see Fig. 1) at [S] = 2.5, 1.67, 1.11, 0.74 and 0.49 mM. The inhibitor concentration [I] was in the range of the previously determined
IC50. Hydrolysis of 4-nitrophenyl-glycosides at 30 ЊC was followed at 405 nm in individual wells of flat-bottomed, 96-well, half-area polystyrene
microtiter plates (Costar) using a UV Spectramax 250 instrument from Molecular Devices. 100 µL assays contained both β-glucosidases and the
β-galactosidase at 0.1 U mLϪ1, the β-mannosidase at 0.75 U mLϪ1, the α-glucosidase at 1 U mLϪ1, the α-galactosidase at 0.1 U mLϪ1, the
α-mannosidase at 0.05 U mLϪ1 and α--fucosidase at 2 mU mLϪ1. The α-glucosidase was buffered in PBS pH 7.4, all other enzymes in 0.1 M HEPES
buffer at pH 6.8 and 30 ЊC
Enzyme
Origin
Ki (1) × 10Ϫ6 M Ki (2) × 10Ϫ6 M Ki (3) × 10Ϫ6 M Ki (4) × 10Ϫ6 M Ki (5) × 10Ϫ6 M Ki (6) × 10Ϫ6 M
α-glucosidase
β-glucosidase
β-glucosidase
α-galactosidase green coffee beans
β-galactosidase bovine liver
yeast
14*
0.66
0.27
14*
0.018
0.022
15*
0.007
0.024
6*
120
0.51
almonds
0.004
0.017
7%
0.035
—
80
0.012
C. s. a
—
0.43
1.5
2
0.024
0.025
α-mannosidase jack beans
β-mannosidase snail acetone powder
12%
38%
50*
10*
9%
α--fucosidase
human placenta
—
—
—
—
a C. s. = Caldocellum saccharolyticum.
protected furanoside 18 with excess PCC gave lactone 19, which
reacted with lithium dimethyl methylene phosphonate to
provide enone 20 following a protocol adapted from the
literature.13 Stereoselective epoxidation with basic H2O2 at low
temperature gave epoxide 21. The ketone was then converted
quantitatively to the volatile allylic epoxide 22 using instant
ylide.14 Aminolysis of this sensitive intermediate with benzyl-
amine followed by Boc protection gave carbamate 23. Hydro-
boration with borane–THF gave the corresponding alcohol 24
stereoselectively. Finally, acidic deprotection with aqueous HCl
gave the N-benzyl derivative 5.
The N-benzyl-α-galacto derivative 6 was prepared from
epoxide intermediate 22 following a sequence similar to that
used for the corresponding -fuco compound (Scheme 3).8
Treatment with LiBr in acetic acid resulted in regioselective
opening of the epoxide at the allylic carbon to give bromo-
hydrin 25. The secondary alcohol function was then converted
to the corresponding N-benzyl carbamate 26 by reaction with
benzyl isocyanate. Hydroboration was then carried out with
BH3ؒTHF, followed by a mild oxidative workup with perboric
acid,15 to give the primary alcohol 27. The neutral oxidation
conditions were chosen since the desired product was not
obtained upon oxidative workup of the borane intermediate
with NaOH–H2O2, presumably due to the sensitivity of the
bromide functional group. The primary alcohol was then
protected as the TBS ether 28. At that stage the α-configured
nitrogen substituent was introduced by intramolecular dis-
placement of the bromide by the N-benzyl carbamate function,
which took place smoothly upon treatment with NaH in THF
at 0 ЊC to give 29. Deprotection was achieved by sequential
treatment with aqueous NaOH to open the cyclic carbamate to
30, followed by aqueous acidic treatment with HCl to give
N-benzyl aminocyclopentitol 6. Attempts to remove the
N-benzyl group by reductive procedures (H2 and various
catalysts, or Li–NH3–THF) were unsuccessful.
Glycosidase inhibition measurements
All compounds were tested for inhibition against a series of
glycosidases using chromogenic nitrophenyl glycosides as
substrates in aqueous buffer at pH 6.8 or pH 7.4, under which
conditions all enzymes tested showed satisfactory activities.
Initial inhibitor screening was done at 100 µM concentration.
Compounds showing significant inhibition under these con-
ditions were characterized more closely by determining the
competitive inhibition constant (Table 1 and Fig. 3). Remark-
ably, most compounds inhibited at least one enzyme in
the nanomolar range. The tightest inhibitions were found
for the N-benzyl-β--galacto configured inhibitor 5 with
β-galactosidase from bovine liver (Ki = 12 nM), and for the
N-phenylpropyl-β--gluco configured inhibitor
β-glucosidase from almonds (Ki = 4 nM).
4
with
Scheme 3 Synthesis of N-benzyl-α-galacto aminocyclopentitol 6.
Reagents and conditions: a) LiBr, AcOH, 25 ЊC, 2 h (53%); b) BnNCO,
Et3N, CH2Cl2, 25 ЊC, 15 h (69%); c) BH3ؒTHF, THF, 25 ЊC, 2 h, then aq.
NaBO3, 25 ЊC, 1.5 h (75%); d) TBSOTf, 2,6-lutidine, CH2Cl2, 25 ЊC, 2 h
(84%); e) NaH, THF, 0 ЊC, 2 h (91%); f ) NaOH, EtOH, 60 ЊC, 7 h
(quant.); g) aq. 0.3 M HCl, 60 ЊC, 12 h, then prep. RP-HPLC (quant.).
Fig. 3 Lineweaver–Burk plot and Dixon replot (insert) of almond
β-glucosidase inhibition by 4. The Ki was determined at [S] = 2.5, 1.67,
1.11, 0.74 and 0.49 mM and [I] = 10, 3.3, 1.1, 0.37 and 0 nM.
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 2 1 7 – 1 2 2 6
1219

View Article Online
Structure–activity relationships
The structure–activity relationship of glycosidase inhibition
was investigated by comparative analysis including a broad
family of aminocyclopentitols reported in the literature, includ-
ing data from our group together with data by Ogawa et al.16
and by Jäger et al.11 (Fig. 4). Inhibitory data for α-glucosidase,
β-glucosidase, α-galactosidase, β-galactosidase and α-manno-
sidase were considered since these were available for almost all
compounds. For each compound the strongest inhibitory data
reported in each enzyme class was used as entry irrespective of
the exact enzyme type used, and reported as pIC50 (= Ϫlog
(IC50)) or pKi (= Ϫlog (Ki)). An IC50 value of 10 mM was used in
the case of missing data or no detected inhibition. The inhibitor
series was then ordered by clustering using the multivariate
analysis software Vista.17 The inhibition pattern was visualized
by a grayscale rendering of inhibition constants. In addition,
the relationship among the different inhibitors was rendered by
the hierarchical tree (Table 2). Inhibitors were grouped in
clusters of two to eleven inhibitors displaying similar inhibition
profiles, with only a few inhibitors showing unique patterns of
inhibition.
Fig. 5 XY-plot of inhibitors according to the gluco/galacto-inhibition
selectivities. The coordinates for each compound are calculated as
x = log Ki(β-glucosidase) Ϫ log Ki(β-galactosidase); y = log Ki(α-
glucosidase) Ϫ log Ki(α-galactosidase) using the data in Tables 1 and 2.
An IC50 value of 10 mM was assumed in the case of missing data or no
detected inhibition. Compound structures in Figs. 1 and 4.
pronounced β-gluco/β-galacto selectivity and was in fact the
only inhibitor showing no detectable cross-reactivity with
β-galactosidase, as measured with the enzyme from bovine liver.
This group of α-glucosidase reactive inhibitors also contained
compound 44 reported by Ogawa, which showed the most
pronounced selectivity for a β-galactosidase (enzyme from
bovine liver) against a β-glucosidase (enzyme from almonds)
due to an absence of cross-reactivity with the latter enzyme.
A second group of inhibitors appeared at zero or positive
y-coordinates, corresponding to either no α-type inhibition
or significant α-galactosidase inhibition. This group contained
mostly galacto-configured inhibitors. However no aminocyclo-
pentitol showed a complete selectivity for α-galactosidase
against α-glucosidase, including the α-galacto-configured
aminocyclopentitol 6 reported here. This compound was never-
theless noteworthy since it displayed the strongest activity
against α-galactosidase in the series, in agreement with its
stereochemistry.
Anomer selectivity
The issue of anomer selectivity of the different inhibitors was
analyzed by constructing a second plot reporting the α-/β-
glucosidase selectivity on the x-axis and the α-/β- galactosidase
selectivity on the y-axis (Fig. 6). The best β-selectivity in both
the gluco and galacto series occurred in the N-bromobenzyl-β-
galacto aminocyclopentitol 42 reported by Jäger,11 which was
also the most potent inhibitor. The strongest α-selectivities were
observed with the N-benzyl-α-galacto aminocyclopentitol 6
reported here for α-galactosidase and the α-gluco-configured
derivative 44 of Ogawa for α-glucosidase (enzyme from yeast)
against β-glucosidase (enzyme from almonds).
The weak cross-reactivity with α-mannosidases reported for
seven compounds in the series could not be rationalized on the
basis of structural features or other inhibitory activities since it
was associated with aminocyclopentitols of various stereo-
chemistries and substitution patterns also showing other inhib-
ition activities. Also, none of the aminocyclopentitols in the
series possessed an α-manno stereochemistry. These weak activ-
ities (Ki = 9.1 µM in the best case of 35) could however be
compared with a competitive inhibition constant Ki = 0.062 µM
reported for the N-methyl-α-manno aminocyclopentitol 58
(Fig. 7) against jack bean α-mannosidase.7 This comparison
clearly showed that the correct stereochemical pattern was also
correlated with strong inhibitory activity against jack bean
α-mannosidase. No data were reported for a β-mannosidase
Fig. 4 Structures of glycosidase inhibitors.
Gluco-galacto selectivity
The dataset was first analyzed in terms of gluco/galacto selectiv-
ity since most inhibitors showed pronounced cross-reactivities
with both enzyme types. A two-dimensional plot was con-
structed in which the x-coordinates represented the β-gluco-
sidase/β-galactosidase selectivity and the y-coordinate the
corresponding α-glucosidase/α-galactosidase selectivity (Fig.
5). Two groups of inhibitors were visible in this display. A first
group appeared at negative y-coordinates representing inhibi-
tors with α-glucosidase inhibition and comprised compounds
with either α-gluco or β-gluco stereochemistry. The N-carb-
oxybenzyl substituted aminocyclopentitol 1 showed the most
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 2 1 7 – 1 2 2 6
1220

View Article Online
Table 2 Glycosidase inhibition constants (µM) for aminocyclopentitol inhibitors (structures in Figs. 1 and 4). No data are given if the inhibitor was
reported to be inactive for the given enzyme. “—” indicates missing data. Grayscale representation of table data at right. The inhibitors were ordered
for similarity by hierarchical clustering of inhibition data (hierarchical tree at right). a Relative configuration of inhibitors in analogy to glycosides.
f
i
Origin of enzymes: b yeast, c Caldocellum saccharolyticum, d bovine liver, e almonds, E. coli, g jack beans, h green coffee beans, Aspergillus niger,
k Aspergillus oryzae
inhibition by this compound. Similarly, the possible activity of
α--fuco aminocyclopentitols 9 and 10 against β--fucosidases
has not been tested. Thus, although our N-benzyl-α-galacto
aminocyclopentitol 6 was clearly the strongest and most select-
ive aminocyclopentitol in the series against α-galactosidase, its
significant cross-inhibition against bovine liver β-galactosidase
and two different β-glucosidases might indicate a general limi-
tation of N-alkyl aminocyclopentitols at providing α-selective
inhibitors.
The relatively weaker anomer selectivity displayed by the
α-galacto aminocyclopentitol 6 in comparison to the β-con-
figured analogs might be caused by conformational flexibility
of the N-alkyl group. This N-alkyl substituent may be con-
sidered as a mimic of the aglycon leaving group, and structural
flexibility probably allows it to adopt a pseudo-equatorial
conformation suitable to bind to the active site of β-selective
glycosidases. Such structural flexibility might also explain the
fact that monoclonal antibodies produced against the α--fuco
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 2 1 7 – 1 2 2 6
1221

View Article Online
Experimental section
General
Enzymes and reagents were purchased from Fluka, Sigma
or Aldrich. All solvents used in reactions were bought in p.a.
quality or distilled and dried prior to use. Solvents for extrac-
tions were distilled from technical quality. Sensitive reactions
were carried out under nitrogen or argon, the glassware was
heated under high vacuum. Chromatographic purifications
(flash) were performed with Silicagel 60 from Merck or Fluka
(0.04–0.063 nm; 230–400 mesh ASTM). HPLC was carried out
with mixtures of MilliQ deionized water containing 0.1% TFA
and HPLC-grade acetonitrile. Analytical HPLC was performed
in a Waters chromatography system (996 Photodiode Array
Detector) using a Vydac column (218TP54 RP-C18, 300 Å pore
size, 0.4 × 22 cm) with a flow rate of 1.0 mL minϪ1 (isocratic
mode). Preparative HPLC was done with a Waters Delta Prep
LC4000 system (Waters 486 Tunable Absorbance Detector)
using a Waters prepak Cartridge 500 g column (RP-C18 20
mm, 300 Å pore size), flow rate 100 mL minϪ1 (gradients 1%
minϪ1 CH3CN). TLC controls were performed with pre-coated
TLC plates Sil G-25 UV254 from Macherey-Nagel followed by
coloration with cerium solution (10.5 g Ce()-sulfate, 21 g
phosphomolybdic acid, 60 mL conc. H2SO4 in 900 mL water)
or ninhydrin (5 g in 100 mL of ethanol) followed by heating.
Optical rotations were determined in a Perkin Elmer 241 polar-
imeter using a 10 cm cell. Infrared spectra were recorded in a
Perkin Elmer 1600 series FTIR. MS and HRMS analyses were
provided by the “Service of Mass Spectrometry” of the
Department of Chemistry and Biochemistry. 1H- and 13C-
NMR spectra were registered on Bruker AC-300 (300 MHz)
and DRX 500 instruments. Chemical shifts δ are given in ppm,
coupling constants J in hertz (Hz).
Fig.
6
XY-plot of inhibitors according to the α-/β-inhibition
selectivities. The coordinates for each compound are calculated as
x = log Ki(β-glucosidase) Ϫ log Ki(α-glucosidase); y = log Ki(β-
galactosidase) Ϫ log Ki(α-galactosidase) using the data in Tables 1 and
2. An IC50 value of 10 mM was assumed in the case of missing data
(“—” in Table 2) or no detected inhibition (no data in Table 2).
Compound structures in Figs. 1 and 4.
Fig.
7
Structures of N-methyl-α-manno aminocyclopentitol 58,
α--fuco hapten 59 and 1-deoxy-galacto-nojirimycin 60.
(1R,2S,3S,4R,5R)-4-[(2,3,4-Trihydroxy-5-hydroxymethyl-
cyclopentylamino)methyl]benzoic acid (1). A solution of 12
(10.9 mg, 0.06 mmol) in DMF (0.3 mL) with CsOH mono-
hydrate (18.9 mg, 0.11 mmol) and 4 Å molecular sieve (30 mg)
was heated to 80 ЊC for 20 min. At 25 ЊC 4-(bromomethyl)-
benzoic acid (13.1 mg, 0.6 mmol) was added. The mixture was
heated to 80 ЊC for 12 hours, concentrated under vacuum, taken
up in water, acidified with TFA, filtrated and purified by
RP-HPLC. The compound 1 was obtained (4.5 mg, 16%) after
lyophilization as a white solid; [α]2D0 = ϩ11.2 (c = 0.42, H2O); 1H
NMR (300 MHz, MeOD): δ = 8.06 (d, 2H, J = 8.5), 7.59 (d, 2H,
J = 8.5), 4.43 (dd, 2H, J = 37.7, 13.4), 4.00 (t, 1H, J = 8.1), 3.88
aminocyclopentitol 59 are not catalytic.18 The bicyclic deriv-
ative 4416 is clearly an α-selective and very potent aminocyclo-
pentitol inhibitor. This α-selectivity might be caused by its
bicyclic nature providing structural rigidity in the sense of a
locked α-orientation of the leaving group mimic. However
the corresponding bicyclic α-galacto aminocyclopentitol 4616
shows only weak β-glucosidase and β-galactosidase activity
and no activity against α-galactosidase or α-glucosidase. The
problem of selective and potent α-galactosidase inhibition is
elegantly solved by the galacto-deoxynojirimycin 60 (Ki
=
(dd, 1H, J = 11.6, 4.2), 3.8–3.55 (m, 4H), 2.32–2.21 (m, 1H); 13
C
0.0016 µM for α-galactosidase from green coffee bean, Ki = 12.5
µM for β-galactosidase from E. coli, Ki = 540 µM for β-gluco-
sidase from almonds).19
NMR (75 MHz, MeOD): δ = 169.0, 137.6, 131.4, 130.8, 82.3,
77.9, 75.4, 62.8, 60.0, 51.3, 45.6; HR-ESI-MS calcd for
C14H20NO6 [M ϩ 1]: 298.1291; found 298.1303.
Conclusion
General procedure for the synthesis of 2–4 by reductive alkyl-
ation of 12. A solution of 12 (7.8 mg, 0.04 mmol) in MeOH (0.5
mL) was added to Al2O3 (32.9 mg) and concentrated under
vacuum. Then an excess of the corresponding aldehyde (0.4
mmol) was added. The mixture was stirred at 50 ЊC for 6 hours.
MeOH (1 mL), dried 4 Å molecular sieve (45 mg) followed by
small portions of NaBH4 (38 mg, 1 mmol) were added. After
stirring at 25 ЊC for 12 hours the mixture was concentrated
under vacuum, taken up in water, acidified with TFA, filtrated
and purified by RP-HPLC. Lyophilisation gave the desired
compounds as white TFA salts.
A series of N-alkyl aminocyclopentitols with β-gluco, β-galacto
and α-galacto- stereochemistry was prepared and evaluated
for glycosidase inhibition. The most potent inhibitions were
observed whenever the stereochemistry was matched between
aminocyclopentitol and glycosidase, including the α- or β-con-
figuration of the N-alkyl-amino group mimicking the proton-
ated leaving group at the anomeric position. This principle also
emerged from a comparative study including thirty-five differ-
ent aminocyclopentitols with varying stereochemistry and
substitution pattern. Since the source of each enzyme also
influences inhibition, the details of the classification of inhibi-
tors produced cannot be used to predict inhibitor selectivities
against any glycosidase enzyme. Most interestingly, the struc-
ture–activity relationships show that the selectivity and potency
of glycosidase inhibition also depend on the nature of the
aglycon mimic. Systematic variations of this group might be
of interest to optimize aminocyclopentitols against enzymes of
therapeutic relevance.
(1R,2S,3S,4R,5R)-4-Benzylamino-5-hydroxymethyl-cyclo-
pentane-1,2,3-triol (2). Compound 2 was obtained as a white
TFA salt in 43% yield; [α]2D0 = ϩ13.6 (c = 0.58, H2O); 1H NMR
(300 MHz, MeOD): δ = 7.55–7.42 (m, 5H), 4.38 (dd, 2H, J =
41.2, 13.2), 4.02 (t, 1H, J = 8.1), 3.92 (dd, 1H, J = 11.4, 4.1),
3.82–3.76 (m, 4H), 2.34–2.23 (m, 1H); 13C NMR (75 MHz,
MeOD): δ = 137.6, 131.4, 130.8, 82.3, 77.9, 75.4, 62.8, 60.0,
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 2 1 7 – 1 2 2 6
1222

View Article Online
51.3, 45.6; MS (FABϩ): m/z (%) = 254 (Mϩ, 94), 176 (8), 154
(100), 149 (14), 107 (30); HR-ESI-MS calcd for C13H20NO4
[Mϩ1]: 254.1392; found 254.1387.
isomer, integration = 0.5), 7.4–7.2 (m, 25H), 6.95 (H-1 Z isomer,
integration = 0.1), 5.11 (s, 1H), 4.82–4.35 (m, 10H), 3.95 (m,
1H), 3.85–3.72 (m, 2H), 3.65–3.55 (m, 2H); 13C NMR (75 MHz,
MeOD): δ = 129.39, 129.33, 129.31, 129.19, 128.98, 128.94,
128.82, 128.78, 128.64, 128.57, 91.79, 82.93, 81.71, 79.23, 76.54,
75.92, 74.35, 73.71, 71.15, 70.04. MS (FABϩ): m/z (%) = 646
(Mϩ, 100), 538 (20), 181 (70), 154 (37), 136 (33); HR-LSIMS
calcd for C41H44NO5 [Mϩ1]: 646.3138; found 646.3167.
(1S,2S,3R,4R,5R)-4-Hydroxymethyl-5-(2-phenylethylamino)-
cyclopentane-1,2,3-triol (3). Compound 3 was obtained as a
white TFA salt in 49% yield; [α]2D0 = ϩ10.2 (c = 0.52, H2O); H
1
NMR (300 MHz, MeOD): δ = 7.37–7.23 (m, 5H), 3.97–3.44 (m,
8H), 3.12–2.96 (m, 2H), 2.31–2.22 (m, 1H); 13C NMR (75 MHz,
MeOD): δ = 137.77, 130.03, 129.79, 128.33, 82.10, 77.81, 74.98,
62.73, 59.56, 49.75, 45.47, 33.25; HR-ESI-MS calcd for
C14H22NO4 [Mϩ1]: 268.1549; found 268.1543.
Imidazole-1-carbothioic acid O-[(1R,2R,3R,4S )-2,3,4-tris-
benzyloxy-5-benzyloxyimino-1-benzyloxymethyl]pentyl
ester
(14). Oxime ether 13 (6.6 g, 10.22 mmol) and 1,1Ј-thiocarb-
onyldiimidazole (2.8 g, 15.73 mmol) were dissolved in benzene
(102 mL). The mixture was heated to reflux under an atmos-
phere of nitrogen for 4 hours. After evaporation of the solvent
FC (hexane–AcOEt, 2 : 1) gave a 5 : 1 mixture of E/Z isomers
of imidazolide 14 (7.4 g, 96%) as a colorless oil. Rf = 0.42
(hexane–AcOEt, 2 : 1); [α]2D7 = ϩ30.5 (c = 0.96 in CHCl3); IR
(neat): ν = 3090w, 3065m, 3032m, 2926m, 2870m, 1737s, 1497s,
1455s, 1391s, 1328s, 1287s, 1244s, 1232s; 1H NMR (300 MHz,
CDCl3): δ = 8.17 (s, 1H, E isomer integration = 0.84 and Z
isomer integration = 0.17), 7.53 (d, 1H, J = 8.5), 7.47 (s, 1H),
7.4–7.1 (m, 25H), 5.75 (m, 1H), 5.10 (s, 2H), 4.73–4.32 (m,
10H), 4.18 (t, 1H, J = 4.2), 4.01 (dd, 1H, J = 2.6, 11.4), 3.82 (m,
2H); 13C NMR (75 MHz, CDCl3): δ = 181.2, 151.6, 149.19,
137.52, 131.40, 129.20, 129.12, 129.08, 129.04, 129.03, 129.01,
128.99, 128.96, 128.87, 128.78, 128.68, 128.58, 128.56, 128.51,
128.48, 128.47, 128.38, 128.36, 128.31, 118.71, 83.77, 79.97,
78.10, 77.10, 76.70, 75.15, 75.06, 73.83, 72.17, 71.84, 67.96;
HR-LSIMS calcd for C45H45N3O6S [Mϩ1]: 756.3107; found
756.3114.
(1S,2S,3R,4R,5R)-4-Hydroxymethyl-5-(3-phenylpropyl-
amino)cyclopentane-1,2,3-triol (4). Compound 4 was obtained
as a white TFA salt in 51% yield; [α]2D0 = ϩ12.4 (c = 0.51, H2O);
1H NMR (300 MHz, MeOD): δ = 7.35–7.15 (m, 5H), 3.93–3.68
(m, 4H), 3.63–3.57 (t, 1H, J = 7.7), 3.56–3.46 (t, 1H, J = 9.0),
3.12–3.0 (m, 1H), 2.76–2.66 (t, 2H, J = 7.5), 2.28–2.18 (m, 1H),
2.09–1.96 (m, 2H); 13C NMR (75 MHz, MeOD): δ = 141.6,
129.7, 129.4, 129.3, 127.4, 82.2, 77.9, 75.1, 62.9, 59.6, 48.3, 45.6,
33.6, 28.9. HR-ESI-MS calcd for C15H24NO4 [Mϩ1]: 282.1705;
found 282.1694.
(1S,2S,3S,4R,5R)-4-Benzylamino-5-(hydroxymethyl)cyclo-
pentane-1,2,3-triol (5). A solution of 24 (5 mg, 13 µmol) in
water (2 mL) and 0.3 M aq. HCl (0.25 mL) was heated at 60 ЊC
for 12 hours and lyophilized to give 5 (3.2 mg, 100%) as a
colorless solid. [α]2D0 = ϩ9.54 (c = 1.23, CHCl3); H NMR (300
1
MHz, D2O): 7.45 (m, 5H), 4.24 (m, 4H), 3.97 (dd, 1H, J = 5.88,
4.41), 3.76 (dd, 1H, J = 11.40, 6.99), 3.64 (dd, 1H, J = 11.37,
6.96), 3.54 (dd, 1H, J = 7.74, 7.71), 2.37 (m, 1H); 13C NMR (75
MHz, D2O ϩ 5% DMSO-d6): 132.2, 131.8, 80.2, 74.1, 72.5,
62.0, 61.1, 52.8, 47.2; MS: 253, 91, 75.
O-Benzyl-N-[(1R,2S,3S,4R,5R)-2,3,4-tris-benzyloxy-5-
benzyloxymethylcyclopentyl]-N-hydroxylamine (15) and O-
benzyl-N-[(1R,2S,3S,4R,5S )-2,3,4-tris-benzyloxy-5-benzyloxy-
methylcyclopentyl]-N-hydroxylamine (16). A solution of imid-
azolide 14 (7.13 g, 9.43 mmol) in benzene (500 mL) was heated
under an atmosphere of nitrogen to reflux. A solution of AIBN
(778 mg, 4.74 mmol) and Bu3SnH (5.4 mL, 20.38 mmol) in
benzene (23 mL) was added dropwise over 30 min. After 4 h the
solvent was evaporated. In a first FC (hexane–EtOAc 4 : 1) the
two isomers (5.88 g, 99%) were separated from impurities and
then separated by FC to give 15 (3.2 g, 55%) and 16 (2.06 g,
35%) as colorless oils. Data for 15: Rf = 0.55 (hexane–EtOAc 4 :
1); [α]2D7 = Ϫ15.6 (c = 1.04 in CHCl3); IR (neat): ν = 3089w,
(1S,2S,3S,4S,5R)-4-Benzylamino-5-(hydroxymethyl)cyclo-
pentane-1,2,3-triol (6). A solution of amine 30 (1 mg; 3.4 µmol)
in water (2 mL) and aq. HCl (0.3 M, 250 µl) was heated at 60 ЊC
for 12 hours and evaporated. The residue was purified by RP-
HPLC. Lyophilization gave 6 (0.9 mg, 100%) as a colorless
1
solid. H NMR (300 MHz, D2O): 7.42 (m, 5H), 4.24 (m, 4H),
3.97 (dd, 1H, J = 5.9, 4.4), 3.76 (dd, 1H, J = 11.0, 7.0), 3.64 (dd,
1H, J = 11.0, 7.0), 3.55 (dd, 1H, J = 8.1, 7.7), 2.37 (m, 1H); 13
C
NMR (75 MHz, D2O ϩ 5% DMSO-d6): 132.2, 131.2, 131.1,
130.8, 79.2, 73.1, 71.5, 61.0, 60.0, 52.0, 46.2.
1
3065m, 3032m, 2866m, 1737m, 1497m, 1455s, 1363s; H NMR
(1R,2S,3S,4R,5R)-4-Amino-5-hydroxymethyl-cyclopentane-
1,2,3-triol hydrochloride (12). Pentaacetate 17 (132.5 mg, 0.36
mmol) was dissolved in 1.2 M HCl (4 mL) and heated to reflux
for 15 hours. Evaporation gave 12 (66.2 mg, 100%) as a color-
less solid. 1H NMR (300 MHz, MeOD): δ = 3.88–3.72 (m, 4H),
3.62 (dd, 1H, J = 8.5, 7.7), 3.50 (t, 1H, J = 9.2); 13C NMR (75
MHz, MeOD): δ = 81.8, 78.2, 75.7, 59.7, 56.2, 45.3; HR-ESI-
MS calcd for C6H14NO4 [M ϩ 1]: 164.0923; found 164.0927.
(300 MHz, CDCl3): δ = 7.4–7.2 (m, 25H), 6.0 (s, 1H), 4.60–4.40
(m, 10H), 3.98 (m, 2H), 3.80 (m, 1H), 3.61 (m, 3H), 2.53 (m,
1H); 13C NMR (75 MHz, CDCl3): δ = 138.8, 138.4, 138.0, 128.6,
128.54, 128.5, 128.1, 128.0, 127.9, 127.7, 89.9, 84.2, 83.2, 76.5,
73.4, 72.2, 72.1, 71.1, 68.0, 63.9, 44.1; MS (FABϩ): m/z (%) =
630 (Mϩ, 87), 522 (16), 503 (18), 463 (24), 445 (35), 387 (30), 329
(22); HR-ESI-TOF-MS calcd for C41H44NO6 [Mϩ1]: 630.3246;
found 630.3254.
Data for 16: Rf = 0.50 (hexane–EtOAc 4 : 1); [α]D27 = ϩ8.7 (c =
1.01 in CHCl3); 1H NMR (300 MHz, CDCl3): δ = 7.4–7.2 (m, 25
H), 4.70–4.59 (m, 10H), 4.04–3.95 (m, 3H), 3.81 (dd, 1H, J =
9.2, 7.4), 3.68 (dd, 1H, J = 9.2, 6.6), 3.45 (dd, 1H, J = 8.8, 5.9),
2.61–2.50 (m, 1H); 13C NMR (75 MHz, CDCl3): δ = 139.33,
139.20, 139.02, 138.58, 129.05, 129.03, 129.01, 128.99, 128.98,
128.96, 128.45, 128.40, 128.36, 128.30, 128.27, 128.19, 128.15,
87.19, 85.63, 81.68, 77.13, 73.89, 72.59, 72.44, 72.39, 69.24,
68.13, 43.28.
(2S,3R,4R,5R)-2,3,4,6-Tetrakis-benzyloxy-5-hydroxyhexanal
O-benzyloxime (13). A suspension of tetra-O-benzyl-gluco-
pyranose (6.03 g, 9.34 mmol) and O-benzylhydroxylamine
hydrochloride (2.13 g, 13.34 mmol) in MeOH (85 mL) and
pyridine (6 mL) was stirred at 60 ЊC for 15 hours. After concen-
tration under vacuum the resulting residue was dissolved in
AcOEt (240 mL) and washed with 1 M HCl, saturated
NaHCO3 and brine (240 mL each). The aqueous layers were
washed with AcOEt (2 × 240 mL). The organic layers were
dried over MgSO4 and concentrated under vacuum. FC (hexa-
ne–AcOEt, 4 : 1) gave a 5 : 1 mixture of E/Z isomers of oxime
ether 13 (6.84 g, 95%) as a colorless oil. Rf = 0.45 (hexane–
AcOEt, 3 : 1); [α]2D7 = ϩ24.3 (c = 0.70 in CHCl3); IR (neat): ν =
3499w br, 3090w, 3064m, 3032m, 2867m, 1497s, 1455s; 1H
NMR (300 MHz, CDCl3): δ = 7.50 (dd, 1H, J = 8.1, 1.8; H-1 E
Acetic acid O-(1R,2R,3S,4S,5R)-2,3,4-triacetoxy-5-acetyl-
amino-cyclopentylmethyl ester (17). Hydrogenation of -amino-
cyclopentitol hydroxylamine 15 (1.08 g, 1.72 mmol) with 10%
Pd/C (249.7 mg) in acetic acid (22 mL) in a 500 mL autoclave
under 12 bar of hydrogen at 25 ЊC during 4 days resulted in
complete debenzylation. The mixture was filtrated over Celite
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 2 1 7 – 1 2 2 6
1223

View Article Online
and concentrated under vacuum. After adding DMAP (9.1 mg,
0.07 mmol) and pyridine (5 mL) the solution was cooled to 0 ЊC
and acetic anhydride (5 mL) was slowly added. After 20 hours
stirring at 25 ЊC the mixture was concentrated under vacuum.
FC (hexane–EtOAc 1 : 4) gave the pentaacetate 17 (514 mg,
was stirred at Ϫ50 ЊC during 1 hour and quenched by addition
of dichloromethane (10 mL) at Ϫ50 ЊC. The mixture was
poured onto sat. aq. NH4Cl. Extraction (aq. NH4Cl–CH2Cl2)
followed by FC (hexane–AcOEt 2 : 1, Rf = 0.63) gave 21 (84 mg,
79%) as a colorless oil, [α]2D0 = Ϫ84.92 (c = 0.455, CHCl3); IR
(KBr): 2990m, 2939m, 2361m, 2343m, 1763s, 1376s, 1213s,
1156m, 1086s, 969w, 865m; MS: 170, 155, 143, 129, 113, 85, 71,
59, 43; 1H NMR (300 MHz, CDCl3): 4.90 (br d, 1H, J = 5.52),
4.42 (br d, 1H, J = 5.52), 4.02 (br d, 1H, J = 2.19), 3.58 (br d,
1H, J = 2.19 Hz), 1.41 (s, 3H), 1.37 (s, 3H); 13C NMR (75 MHz,
CDCl3): 204.2, 115.0, 78.1, 76.1, 58.7, 55.1, 27.7, 25.7.
80%) as a yellow oil. Rf = 0.22 (hexane–EtOAc 1 : 4); [α]2D7
=
Ϫ31.3 (c = 0.88 in CHCl3); IR (neat): ν = 3500–3400w, 3300m
br, 3074w, 2966w, 1745s, 1661m; 1H NMR (300 MHz, CDCl3):
δ = 6.67 (d, 1H, J = 6.7), 5.35–5.25 (m, 2H), 5.10 (t, 1H, J = 4.8),
4.55 (dd, 1H, J = 16.6, 6.6), 4.25 (dd, 1H, J = 11.6, 4.4), 4.08 (dd,
1H, J = 11.8, 4.4), 2.65 (m, 1H); 13C NMR (75 MHz, CDCl3):
δ =172.12, 171.16, 171.13, 170.87, 170.57, 78.72, 78.27, 76.79,
62.21, 53.34, 43.72, 23.56, 21.60, 21.47, 21.43, 21.39; MS: m/z
(%) = 372 (0.1), 538 (0.4), 330 (0.5), 314 (0.8), 193 (45), 151
(100); HR-ESI-TOF-MS calcd for C16H23NO9 [Mϩ1]: 374.1451;
found 374.1438.
(2S,3S,4S,5R)-1-Methylene-2,3-O-isopropylidene-2,3-di-
hydroxy-4,5-epoxycyclopentane (22). A solution of epoxyketone
21 (62.1 mg, 0.365 mmol) in THF (1 mL) was added to a solu-
tion of “instant ylide” (PPh3CH2Br–NaNH2, 304.0 mg, 0.365
mmol, 1 equiv.) in THF (1.5 mL). The reaction mixture was
stirred at 25 ЊC for 1 hour, diluted in diethyl ether (10 mL), and
filtered over Celite. Evaporation and purification by FC (Et2O–
Methyl 2,3-O-isopropylidene-ꢀ-D-lyxofuranoside (18). To a
solution of -(Ϫ)-lyxose (10.0 g, 66.6 mmol) in acetone (80 mL)
and 2,2-dimethoxypropane (20.0 mL, 266 mmol) 70% aq.
HClO4 (4 mL) was added at 0 ЊC. After stirring for 2 hours at
25 ЊC methanol (14 mL) was added. The mixture was stirred for
another 3 hours at 25 ЊC. The reaction was cooled to 0 ЊC and
quenched with aq. Na2CO3 (9.6 g in 30 mL H2O). The precipi-
tate was filtered and the filtrate was concentrated. Extraction
(Et2O–brine) followed by vacuum distillation gave 18 (10.4 g,
84%) as a pale yellow oil; 1H NMR (300 MHz, CDCl3): 4.94 (s,
1H), 4.78 (dd, 1H, J = 5.88, 3.68), 4.58 (d, 1H, J = 5.88), 4.00
(m, 3H), 3.34 (s, 3H), 1.54 (s, 3H), 1.31 (s, 3H); 13C NMR (75
MHz, CDCl3): 113.3, 107.7, 85.8, 80.9, 80.0, 61.7, 55.3, 26.6,
25.2; MS: 204, 187, 173, 171, 123, 85, 59, 43.
pentane) gave 22 (61.4 mg, 100%) as a colorless oil, [α]2D0
=
Ϫ132.2 (c = 1.495, CHCl3); IR (KBr): 3049w, 2952m, 2924m,
1
1735w, 1585w, 1486w, 1376w, 1270w; H NMR (300 MHz,
CDCl3): 5.62 (d, 1H, J = 1.86 Hz), 5.53 (d, 1H, J = 1.86), 4.73 (d,
1H, J = 5.88), 4.67 (d, 1H, J = 5.88), 3.84 (d, 1H, J = 1.83), 3.80
(d, 1H, J = 2.2), 1.43 (s, 3H), 1.37 (s, 3H); 13C NMR (75 MHz,
CDCl3): 147.4, 118.0, 113.5, 81.1, 79.8, 60.0, 59.8, 28.2, 26.2.
(1R,2S,3R,4S )-1-(N-Benzyl)-(N-tert-butoxycarbonyl)amino-
3,4-O-isopropylidene-5-methylenecyclopentane-2,3,4-triol (23).
Epoxyalkene 22 (67 mg, 0.4 mmol) was stirred in neat benzyl-
amine (2.5 mL) for 90 min at 25 ЊC. The benzylamine was evap-
orated under vacuum. After FC (hexane–AcOEt 1 : 1, Rf = 0.67)
the product still was contaminated with benzylamine. 1H NMR
(300 MHz, CDCl3): 7.32 (m, 5H), 5.52, 5.35 (2s, 1H), 4.91 (d,
1H, J = 6.60), 4.38 (ddd, 1H, J = 6.24, 1.83, 1.08), 4.04 (dd, 1H,
J = 3.66, 1.83), 3.94, 3.74 (2d, 1H, J = 12.87), 3.39 (d, 1H, J =
2.58), 1.41 (s, 3H), 1.33 (s, 3H); HR-MS calcd for C16H21NO3:
275.1521; found: 275.1524. This intermediate was dissolved in
water (2 mL) and AcOEt (2 mL) and treated with di-tert-butyl
dicarbonate (58 mg, 0.27 mmol) and NaHCO3 (40 mg). The
mixture was stirred vigorously overnight at 25 ЊC. Extraction
(brine–AcOEt) followed by FC (hexane–AcOEt, Rf = 0.25) gave
23 (36 mg, 24%) as a colorless oil, [α]2D0 = ϩ6.94 (c = 1.56,
CHCl3); IR (KBr): 3684w, 3618w, 3444w br, 3019s, 2978m,
2935w, 2400w, 1686m, 1522w, 1418m, 1368m, 1214s, 1163m,
1071m, 928m; 1H NMR (300 MHz, CDCl3): 7.31 (m, 5H), 5.43
(1s, 1H), 5.10 (br s, 2H), 4.85 (d, 1H, J = 6.99), 4.34 (s, 1H), 4.06
(m, 2H), 3.92 (m, 1H), 1.43 (s, 12H), 1.33 (s, 3H); HR-MS calcd
for C21H29NO5: 375.2046; found 375.2048.
(3S,4S,8R)-6,6-Dimethyl-3-methoxy-2,5,7-trioxabicyclo-
[3.3.0]octanone (19). Pyridinium chlorochromate (52.7 g, 308
mmol) was added to a stirred solution of 18 (12.6 g; 615 mmol)
in benzene (500 mL). The reaction mixture was heated to reflux
in a Dean–Stark apparatus overnight. The organic phase was
decanted and recovered. The residues were stirred in EtOAc
(100 mL) several times. The combined organic phases were
filtrated over Celite and evaporated. The crude product was
purified by FC (hexane–EtOAc 7 : 1, Rf = 0.2) to give 19 (4.52 g,
39%) as a colorless solid; [α]2D0 = ϩ47.52 (c = 0.52, CHCl3); IR
(KBr): 2996w, 2942w, 1809m, 1455w, 1376w, 1359w, 1210w,
1156w, 1122m, 1105m, 1048w, 973w, 931w, 797s, 774m, 755m;
1H NMR (300 MHz, CDCl3): 5.18 (s, 1H), 4.65 (d, 1H, J = 5.52
Hz), 4.41 (d, 1H, J = 5.52), 3.35 (s, 3H), 1.25 (s, 3H), 1.19 (s,
3H); 13C NMR (75 MHz, CDCl3): 171.2, 111.8, 102.8, 76.9,
72.0, 54.5, 24.1, 23.1; HR-MS calcd for C8H13O5 [Mϩ1]:
189.0762; found 189.0769.
(4R,8R)-6,6-Dimethyl-5,7-dioxabicyclo[3.3.0]oct-2-en-1-one
(20). A solution of dimethyl methylphosphonate (0.28 mL, 1
equiv.) in THF (20 mL) at Ϫ78 ЊC was treated with n-BuLi
(1.73 mL, 1.52 M in hexane). Then a solution of lactone 19 (493
mg, 2.62 mmol) in THF was added. The reaction mixture was
stirred 150 min at Ϫ78 ЊC, 20 min at Ϫ20 ЊC and then quenched
with EtOAc (50 mL). Extraction (AcOEt–water) and purifi-
cation by FC (hexane–AcOEt 2 : 1, Rf = 0.46) gave 20 (158 mg,
39%) as a colorless crystalline solid, mp: 69–70 ЊC; IR: 2993w,
2937w, 1735s, 1550m, 1382m, 1374m, 1210m, 1156w, 1096m,
1068w, 1007w; 1H NMR (300 MHz, CDCl3): 7.61 (dd, 1H, J =
5.88, 2.22), 6.22 (d, 1H, J = 5.88), 5.27 (ddd, 1H, J = 5.52, 2.22,
0.75), 4.46 (d, 1H, J = 5.52), 1.42 (s, 6H); 13C NMR (75 MHz,
CDCl3): 203.0, 159.6, 134.3, 115.5, 78.6, 76.6, 27.4, 26.1; MS:
154, 129, 96, 85, 68, 43.
(1S,2R,3S,4R,5R)-1,2-O-Isopropylidene-4-(N-benzyl)-(N-
tert-butoxycarbonyl)-amino-5-(hydroxymethyl)-cyclopentane-
1,2,3-triol (24). BH3ؒTHF (0.4 ml, 0.4 mmol) was added to a
solution of 23 (15.4 mg, 40.3 µmol) in THF (0.5 mL) at 0 ЊC
and the solution stirred for 2 hours at 25 ЊC. Water (5 mL) and
NaBO3 (186 mg, 30 equiv.) were added and the reaction was
stirred for 90 minutes at 25 ЊC. Extraction (AcOEt–water) and
purification by FC (hexane–AcOEt 2 : 1, Rf = 0.28) gave 24
(12.1 mg, 75%) as a colorless oil. [α]2D0 = ϩ13.6 (c = 1.84, CHCl3);
1H NMR (300 MHz in CDCl3): 7.33 (m, 5H), 4.72 (dd, 1H, J =
7.74, 7.35), 4.46 (br s, 2H), 4.35 (dd, 1H, J = 7.31, 4.8), 4.20 (m,
1H), 3.74 (m, 3H), 1.56 (s, 3H), 1.46 (s, 9H), 1.32 (s, 3H), 1.45 (s,
9H); 13C NMR (75 MHz, CDCl3): 139.9, 129.6, 128.1, 127.3,
113.3, 84.6, 81.4, 77.9, 64.8, 60.4, 29.0, 27.1, 24.7; HR-MS calcd
for C21H31NO6: 393.2151; found 393.2170.
(2R,3R,4S,5R)-2,3-O-Isopropylidene-2,3-dihydroxy-4,5-
epoxycyclopentanone (21). A solution of 20 (97 mg, 6.3 mmol)
in methanol (5 mL) cooled to Ϫ50 ЊC was treated with 30% aq.
H2O2 (143 µl, 2 equiv.) and 2 M NaOH (133 µl). The reaction
(1R,2R,3R,4S )-1-Bromo-3,4-O-isopropylidene-5-methylene-
cyclopentane-2,3,4-triol (25). To a solution of epoxyolefin 22 (83
mg, 0.50 mmol) in acetic acid (1.67 mL) was added LiBr (215
mg, 5 equiv.) and the mixture was stirred at room temperature
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 2 1 7 – 1 2 2 6
1224

View Article Online
during 2 hours. Co evaporation with toluene and purification
by FC (hexane–EtOAc 2 : 1, Rf = 0.53) gave 25 (65.4 mg, 53%)
as a colorless oil. [α]2D0 = ϩ43.20 (c = 1.32, CHCl3); IR (KBr):
2958s, 2928s, 2873m, 2854m, 1451w, 1381w, 1209w, 1158w,
1077w, 868w; 1H NMR (300 MHz, CDCl3): 5.63 (ddd, 2H, J =
4.0, 1.8, 1.5 Hz), 5.00 (dd, 1H, J = 1.5, 1.1), 4.50 (m, 2H), 4.39
(dd, 1H, J = 4.4, 2.9), 1.56 (s, 3H), 1.36 (s, 3H); 13C NMR (75
MHz, CDCl3): 147.6, 119.6, 113.5, 85.5, 83.3, 80.4, 53.0, 27.2,
25.6; MS: 251, 249 ([Mϩ1]), 155.
cyclopentane-1,2,3-triol (29). A solution of intermediate 28
(14.8 mg, 28.8 µmol) in dry THF (2 mL) at 0 ЊC was treated
with NaH (4.3 mg, 0.12 mmol). The reaction mixture was
stirred during 2 hours at 0 ЊC. Evaporation of the solvent and
purification by FC (hexane–AcOEt 8 : 1, Rf = 0.14) gave 29
(11.4 mg, 91%) as a colorless oil; [α]2D0 = ϩ17.38 (c = 0.1, CHCl3);
1H NMR (300 MHz, CDCl3): 7.33 (m, 5H), 4.79 (d, 1H, J =
15.1), 4.31 (d, 1H, J = 15.1), 2.36 (m, 1H), 1.57 (s, 3H), 1.36 (s,
1H), 1.28 (s, 1H), 1.26 (s, 3H), 0.91 (s, 9H), 0.09 (s, 6H); 13C
NMR (75 MHz, CDCl3): 157.7, 137.0, 129.5, 128.7, 128.6,
112.3, 85.4, 84.6, 81.8, 62.5, 62.4, 52.9, 46.9, 27.6, 25.2, 26.7,
14.8, Ϫ4.6, Ϫ4.7; HR-MS calcd for C23H36BrNO5Si [Mϩ1]:
434.2362; found 434.2365.
(1R,2R,3S,4S )-1-Bromo-2-O-(N-benzylcarbamoyl)-3,4-O-
isopropylidene-5-methylenecyclopentane-2,3,4-triol (26). To a
solution of bromoolefin 25 (15.7 mg, 63.0 µmol) in dry di-
chloromethane (0.5 mL) were added at 0 ЊC, benzyl isocyanate
(8.9 µl, 100 µmol) and dry triethylamine (3.5 µl, 25.2 µmol).
After stirring overnight at 25 ЊC, the mixture was diluted in
diethyl ether. The organic phase was washed with water (5 mL)
and the aqueous phase was extracted with diethyl ether (2 × 10
mL). The recombined organic phases were dried (Na2SO4) and
evaporated. Purification by FC (hexane–AcOEt, Rf = 0.62) gave
26 (16.6 mg, 69%) as a colorless oil. [α]2D0 = ϩ2.3 (c = 0.82,
CHCl3); IR (KBr): 3453w, 3307w br, 2959m, 2932m, 2870m,
1734s, 1725s, 1540m, 1522m, 1508m, 1455m, 1381m, 1256s,
(1S,2R,3S,4S,5R)-4-Benzylamino-5-(hydroxymethyl)-1,2-
isopropylidenecyclopentane-1,2,3-triol (30). A solution of 29
(2 mg, 4.6 µmol) in EtOH (0.3 mL), and 2 M NaOH (0.35 mL)
was heated at 60 ЊC for 7 hours. The reaction was evaporated
under vacuum and the residue was purified by FC (AcOEt–
MeOH 5 : 1, Rf = 0.61) to give 30 (1.4 mg, 100%) as a colorless
1
oil; H NMR (300 MHz, CDCl3): 7.33 (m, 5H), 4.65 (dd, 1H,
J = 5.52, 5.52), 4.49 (d, 1H, J = 5.88), 4.03 (d, 1H, J = 4.02), 3.90
(m, 4H), 3.37 (dd, 1H, J = 11.4, 4.05), 2.19 (m, 1H), 1.40 (s, 3H),
1.27 (s, 3H); 13C NMR (75 MHz, CDCl3): 129.4, 129.2, 128.5,
111.0, 84.1, 80.6, 73.0, 62.2, 62.1, 52.9, 46.3, 26.6, 24.2; HR-MS
calcd for C16H23BrNO4: 293.1627; found: 293.1634.
1
1034s, 795s; H NMR (300 MHz, CDCl3): 7.32 (m, 5H), 5.61
(br d, 2H, J = 4.41), 5.30 (dd, 1H, J = 2.19, 1.11), 5.07 (br s, 1H),
4.99 (br d, 1H, J = 6.63), 4.60 (br d, 2H, J = 4.41), 4.38 (d, 2H,
J = 5.88), 1.61 (s, 3H), 1.34 (s, 3H); 13C NMR (75 MHz, CDCl3):
155.3, 148.6, 129.4, 128.4, 128.3, 120.2, 119.7, 113.5, 84.7, 84.1,
81.4, 48.9, 46.0, 26.8, 25.2; MS: 384, 382 ([Mϩ1]).
Enzyme measurements
Enzymes were purchased from Fluka and Sigma. The α-
glucosidase (yeast, EC 3.2.1.20) was buffered in PBS at pH 7.4;
all other glycosidases were buffered in 0.1 M HEPES at pH 6.8:
β-glucosidase (almonds and Caldocellum saccharolyticum, EC
3.2.1.21), α-galactosidase (green coffee beans, EC 3.2.1.22),
β-galactosidase (E. coli, EC 3.2.1.23), α-mannosidase (jack
beans, EC 3.2.1.24), β-mannosidase (snail acetone powder, EC
3.2.1.25) and α--fucosidase (human placenta, EC 3.2.1.51). All
buffers and solutions were prepared using MilliQ deionized
water. Substrates were used as 25 mM stock solutions in buffer
and inhibitors as 10 mM stock solutions in water.
Enzymes and inhibitors were mixed. After 30 minutes these
solutions were added to the substrates. The 100 µl assays were
followed in individual wells of flat-bottomed 96-well half-area
polystyrene cell culture plates (Costar) using a UV Spectramax
250 instrument from Molecular Devices. The release of 4-nitro-
phenol was followed at 405 nm over 30 min at pH 7.4 or pH 6.8
and 30 ЊC. The concentration of each enzyme in the assays was
adjusted so as to give an approximately 0.2 to 1 OD increase as
given by the instrument. The initial rate of reaction of the first
10 minutes was linear and was used to calculate the rate. Rates
were expressed in relative units as milliOD minϪ1. The competi-
tive inhibition constants Ki were determined by Dixon replot of
inhibition data.
(1R,2R,3S,4S,5R)-1-Bromo-2-O-(N-benzylcarbamoyl)-5-O-
(hydroxymethyl)-3,4-O-isopropylidenecyclopentane-2,3,4-triol
(27). BH3ؒTHF (403 µL, 10 equiv.) was added to a solution of
bromoolefin 26 (15.4 mg, 40.3 µmoL) in dry THF (0.5 mL) at
0 ЊC and the reaction was stirred for 2 hours at 25 ЊC. Water
(5 mL) and NaBO3 (186 mg, 30 equiv.) were then added and the
reaction was stirred for another 90 minutes at 25 ЊC. Aqueous
workup (H2O–AcOEt) and column chromatography gave 27
(12.1 mg, 75%) as a colorless oil, [α]2D0 = Ϫ16.48 (c = 0.3, CHCl3);
IR (KBr): 3362w, 2934w, 2340w, 1730m, 1546m, 1510m, 1454m,
1
1382m, 1245m, 1208s, 1004m; H NMR (300 MHz, CDCl3):
7.34 (m, 5H), 5.37 (s, 1H), 5.03 (br s, 1H), 4.80 (dd, 1H, J = 5.88,
5.52), 4.59 (d, 1H, J = 5.88), 4.38 (s, 1H), 4.28 (d, 1H, J = 5.88),
4.01 (m, 2H), 2.55 (m, 1H), 1.58 (s, 3H), 1.29 (s, 3H); 13C NMR
(75 MHz, CDCl3): 129.5, 128.4, 128.3, 112.9, 85.3, 84.2, 80.9,
62.0, 51.6, 48.1, 45.9, 26.0, 24.3; MS: 402, 400 ([Mϩ1]), 386, 384,
320, 91.
(1R,2R,3S,4S,5R)-2-O-(N-Benzylcarbamoyl)-1-bromo-5-(O-
tert-butyldimethylsilyl)-hydroxymethyl-3,4-isopropylidenecyclo-
pentane-1,2,3-triol (28). To a solution of bromoalcohol 27 (25.4
mg, 63.5 µmol) in dry dichloromethane (0.5 mL), 2,6-lutidine
(15 µl, 127 µmol) and tert-butyldimethylsilyl triflate (22 µl, 95
µmol) were added at 0 ЊC. After 2 hours at 25 ЊC, the reaction
mixture was evaporated. Aqueous workup (H2O–AcOEt) and
purification by FC (hexane–AcOEt = 8 : 1, Rf = 0.14) gave 28
(27 mg, 84%) as a colorless solid; mp 106 ЊC; [α]2D0 = Ϫ17.7 (c =
0.26, CHCl3); IR (KBr): 3446m, 3019m, 2929w, 2400w, 1729s,
1511s, 1472w, 1383w, 1214s, 1124w, 1080s; 1H NMR (300
MHz, CDCl3): 7.33 (m, 5H), 5.37 (s, 1H), 4.98 (br d, 1H, J =
5.16), 4.75 (dd, 1H, J = 5.52, 5.49), 4.56 (d, 1H, J = 5.88), 4.39
(m, 1H), 4.38 (d, 2H, J = 5.88), 4.27 (d, 1H, J = 5.52 Hz), 3.97
(m, 2H), 2.45 (m, 1H), 1.57 (s, 3H), 1.27 (s, 3H), 0.91 (s, 9H),
0.10 (s, 3H), 0.09 (s, 3H); 13C NMR (75 MHz, CDCl3): 155.3,
138.7, 129.5, 128.4, 112.6, 85.6, 84.6, 81.2, 62.1, 52.0, 49.2, 45.9,
26.7, 26.0, 24.2, 12.1, Ϫ4.6, Ϫ4.7; HR-MS calcd for C23H37Br-
NO5Si [Mϩ1]: 514.1624; found 514.1628.
Acknowledgements
This work was financially supported by the Swiss National
Science Foundation, the Swiss Office Fédéral de l’Education et
de la Science, the COST program D13 and the Roche Research
Foundation.
References
1 (a) T. D. Heightman and A. T. Vasella, Angew. Chem., 1999, 111,
795; T. D. Heightman and A. T. Vasella, Angew. Chem., Int. Ed.,
1999, 38, 750; (b) Y. Bourne and B. Henrissat, Curr. Opin. Struct.
Biol., 2001, 11, 593; (c) G. J. Davies and B. Henrissat, Biochem. Soc.
Trans., 2002, 30, 291.
2 (a) P. Compain and O. R. Martin, Curr. Top. Med. Chem., 2003, 3,
541; (b) N. Asano, Glycobiology, 2003, 13, 93R; (c) N. Asano,
R. J. Nash, R. J. Molyneux and G. W. J. Fleet, Tetrahedron:
(1S,2S,3S,4S,5R)-4-Benzylamino-3-O-4-N-(oxomethylidene)-
1,2-O-isopropylidene-5-((tert-butyl-dimethylsilyloxy)methyl)-
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 2 1 7 – 1 2 2 6
1225

View Article Online
Asymmetry, 2000, 11, 1645; (d ) P. Sears and C.-H. Wong, Angew.
Chem., Int. Ed., 1999, 38, 2300; (e) V. H. Lillelund, H. H. Jensen,
X. Liang and M. Bols, Chem. Rev., 2002, 102, 515; ( f ) M. Bols, Acc.
Chem. Res., 1998, 31, 1; (g) G. Davies, M. L. Sinnott, S. G. Withers,
pp. 119–208 in Comprehensive Biological Catalysis, Ed. M. Sinnott,
Academic Press, London, 1998.
3 (a) J.-L. Reymond, K. D. Janda and R. A. Lerner, Angew. Chem.,
Int. Ed. Engl., 1991, 30, 1711; (b) D. Shabat, S. C. Sinha, J.-L.
Reymond and E. Keinan, Angew. Chem., 1996, 35, 2628.
4 (a) P. G. Schultz and R. A. Lerner, Science, 1995, 269, 1835; (b)
R. A. Lerner, S. J. Benkovic and P. G. Schultz, Science, 1991, 252,
659; (c) N. R. Thomas, Nat. Prod. Rep., 1996, 479; (d ) J.-L.
Reymond, Top. Curr. Chem., 1999, 200, 59.
5 A. Berecibar, C. Grandjean and A. Siriwardena, Chem. Rev., 1999,
99, 779.
6 B. J. Smith, J. L. McKimm-Breshkin, M. McDonald, R. T. Fernley,
J. N. Varghese and P. M. Colman, J. Med. Chem., 2002, 45,
2207.
7 R. A. Farr, N. P. Peet and M. S. Kand, Tetrahedron Lett., 1990, 31,
7109.
9 E. Leroy and J.-L. Reymond, Org. Lett., 1999, 1, 775.
10 O. Boss, E. Leroy, A. Blaser and J.-L. Reymond, Org. Lett., 2000, 2,
151.
11 (a) M. Kleban, P. Hilgers, J. N. Greul, R. D. Kugler, S. Picasso,
P. Vogel and V. Jäger, Chem. Bio. Chem., 2001, 2, 365; (b) J. N. Greul,
M. Kleban, B. Schneider, S. Picasso and V. Jäger, Chem. Bio. Chem.,
2001, 2, 368.
12 P. A. Bartlett, K. L. McLaren and P. C. Ting, J. Am. Chem. Soc.,
1988, 110, 1633.
13 S. M. Ali, K. Ramesh and R. T. Borchardt, Tetrahedron Lett., 1990,
31, 1509.
14 M. Schlosser and B. Schaub, Chimia, 1982, 10, 396.
15 G. W. Kabalka, T. M. Shoup and N. M. Goudgaon, Tetrahedron
Lett., 1989, 30, 1483.
16 C. Uchida, H. Kimura and S. Ogawa, Bioorg. Med. Chem., 1997, 5,
921.
17 http://www.visualstats.org/.
18 None of more than five hundred hybridoma cell lines generated
from an immunization against a protein conjugate of 59 and
producing strong hapten-binding antibodies showed any α--
fucosidase activity. L. Gartenmann Dickson, Ph.D. thesis, Bern,
2004.
8 (a) A. Blaser and J.-L. Reymond, Helv. Chim. Acta, 1999, 82, 760;
(b) A. Blaser and J.-L. Reymond, Org. Lett., 2000, 2, 1733; (c)
A. Blaser and J.-L. Reymond, Helv. Chim. Acta, 2001, 84, 2119.
19 G. Legeler, Adv. Carbohydr. Chem. Biochem., 1990, 48, 319.
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 2 1 7 – 1 2 2 6
1226