Solid-phase extraction on C18 silica as a purification strategy in the solution synthesis of a 1-Thio-β-D-galactopyranoside library

Article abstract of DOI:10.1016/S0968-0896(98)00087-X

A novel strategy for the purification of carbohydrate-based chemical libraries synthesized in solution was developed. Purification of reaction products was accomplished by means of solid-phase extraction enabled by protecting the 2-, 3-, 4-, and 6-hydroxyl groups of a galactose derivative as their hydrophobic O-laurates. The presence of multiple O-laurates allowed adsorption of reaction products onto C18 silica while reagents and by-products were washed away with MeOH. Products were quantitatively eluted with pentane. Purification of products using solid-phase extraction offers the combined advantages of solution synthesis (normal solution reactivity and ease of reaction monitoring) with those of solid-phase synthesis (facile product isolation permitting the use of large excesses of reagents). To demonstrate the utility of the hydrophobic recovery-procedure, tetra-O-lauroyl-β-d-galactopyranose-1-thiol was subjected to high-yielding reactions with a panel of Michael-acceptors and an α-chloro ketone. The resulting ketone adducts were then either reduced to the alcohols or reductively aminated with a selection of amino acids to give 30 different 1-thio-β-d-galactosides as mixtures of four diastereomers after removal of protecting groups. At each step, the product was separated from the reagents and their by-products by simple adsorption onto C18 silica, washing with MeOH and elution of product with pentane. After completion of the combinatorial chemistry sequence, the O-laurates were cleaved by methanolysis and the product methyl laurate in turn removed from the desired water-soluble products by C18 adsorption. Individual library members were thus conveniently produced on 10-30mg scales at purity levels of >90%. One of the 1-thio-β-d-galactosides thus produced was found to be a competitive inhibitor of the β-galactosidase from E. coli with K(i) value of 1.7μM. Copyright (C) 1998 Elsevier Science Ltd.

Full text of DOI:10.1016/S0968-0896(98)00087-X

BIOORGANIC &
MEDICINAL
CHEMISTRY
Bioorganic & Medicinal Chemistry 6 (1998) 1563±1575
Solid-phase Extraction on C18 Silica as a Puri®cation Strategy
in the Solution Synthesis of a 1-Thio-ꢀ-D-galactopyranoside
Library
Ulf J. Nilsson,y Eric J.-L. Fournier and Ole Hindsgaul*
Department of Chemistry, University of Alberta, Edmonton, Canada T6G 2G2
Received 17 February 1998; accepted 19 March 1998
AbstractÐA novel strategy for the puri®cation of carbohydrate-based chemical libraries synthesized in solution was
developed. Puri®cation of reaction products was accomplished by means of solid-phase extraction enabled by pro-
tecting the 2-, 3-, 4-, and 6-hydroxyl groups of a galactose derivative as their hydrophobic O-laurates. The presence of
multiple O-laurates allowed adsorption of reaction products onto C18 silica while reagents and by-products were
washed away with MeOH. Products were quantitatively eluted with pentane. Puri®cation of products using solid-phase
extraction o€ers the combined advantages of solution synthesis (normal solution reactivity and ease of reaction mon-
itoring) with those of solid-phase synthesis (facile product isolation permitting the use of large excesses of reagents). To
demonstrate the utility of the hydrophobic recovery-procedure, tetra-O-lauroyl-b-d-galactopyranose-1-thiol was sub-
jected to high-yielding reactions with a panel of Michael-acceptors and an a-chloro ketone. The resulting ketone
adducts were then either reduced to the alcohols or reductively aminated with a selection of amino acids to give 30
di€erent 1-thio-b-d-galactosides as mixtures of four diastereomers after removal of protecting groups. At each step, the
product was separated from the reagents and their by-products by simple adsorption onto C18 silica, washing with
MeOH and elution of product with pentane. After completion of the combinatorial chemistry sequence, the O-laurates
were cleaved by methanolysis and the product methyl laurate in turn removed from the desired water-soluble products
by C18 adsorption. Individual library members were thus conveniently produced on 10±30 mg scales at purity levels of
>90%. One of the 1-thio-b-d-galactosides thus produced was found to be a competitive inhibitor of the b-galactosi-
dase from E. coli with Ki value of 1.7 mM. # 1998 Elsevier Science Ltd. All rights reserved.
Introduction
anity of the natural ligand. While several examples
The design and synthesis of potent inhibitors of oligo-
saccharide-binding proteins is a challenging task, com-
plicated by the fact that most of the natural
carbohydrate ligands bind with low anity (Kd nor-
mally in the 0.1±1 mM range). The multistep synthesis
of oligosaccharide analogues, essentially a medicinal
chemistry approach, is one strategy to enhancing the
exist where signi®cant enhancements have been pro-
duced1 by this approach, sometimes guided by the
availability of protein crystal data and molecular mod-
eling,2 the process is nevertheless extremely laborious
and often fails.
Application of combinatorial chemistry to the enhance-
ment of ligand anity is an increasingly attractive
strategy whose main advantage is the ability to produce
a large number of active-site probes rapidly. There are
several reports on the production of oligosaccharide
libraries,3±5 including one wherein the anity of the a
disaccharide for a lectin was enhanced as a result of
appending functionality on the parent (or related) di-
saccharide bound by a lectin.5 This particularly impressive
Key words: Solid-phase extraction; C18; 1-thio-b-d-galacto-
pyranoside library; solution combinatorial chemistry.
*Corresponding author. Tel: +1 403 492 4171; Fax: +1 403
492 7705; E-mail: ole.hindsgaul@ualberta.ca
yCurrent address: Organic Chemistry 2, Lund Institute of
Technology, Lund University, P.O.B. 124, SE-221 00 Lund,
Sweden.
0968-0896/98/$Ðsee front matter # 1998 Elsevier Science Ltd. All rights reserved
PII: S0968-0896(98)00087-X

1564
U. J. Nilsson et al./Bioorg. Med. Chem. 6 (1998) 1563±1575
work produced over 3000 disaccharide derivatives which
were screened to yield the most active compound.
binatorial chemistry approaches, is schematically shown
in Figure 1(B) where the new molecular features are
introduced on a sca€old at the reducing end of the key
sugar, though it could in principle be introduced at any
position not making critical interactions with the protein.
The binding of an oligosaccharide by the combining site
of a protein, whether a lectin, selectin, antibody, toxin,
or enzyme, is schematically represented in Figure 1(A).
A key feature of the binding is that only a portion of the
oligosaccharide is usually in intimate contact with the
protein, the average combining site recognizing
approximately 2.5 sugar residues.6,7 Of these, often only
one is completely buried in the protein,8 the so called
`immunodominant sugar' for antibodies or `key sugar'
for other proteins. The key-monosaccharide usually
binds only weakly and 10±100 mM concentrations can
be required to inhibit the binding of the natural oligo-
saccharide. As shown in Figure 1(A), additional binding
energy is derived because the remaining sugar residues
on the oligosaccharide provide additional interaction
with the protein combining site.
The main strategic issues to be addressed in initiating
this project were:
(a) What types of structures should be considered as
sca€olds? Since proteins binding terminal sugars,
as shown schematically in Figure 1(A), normally
encounter a second sugar structure as the agly-
cone they should be easily able to sterically
accommodate cyclic sca€olds at this site. Scaf-
folds chosen were therefore 5-, 6-, and 7-mem-
bered rings.
(b) What types of molecular features are desirable for
A and B (Fig. 1(B))? This would also dictate the
types of chemical reactions to be used for their
preparation. In situations where little is known
about the target protein, a wide variety of struc-
tures should be examined and there should be no
bias based on the structure of the natural ligand
in the selection of the groups A and B (Figure
1(B)). Typical structures that might make bene-
®cial interactions with the protein combining site
or residues adjacent to it, include diverse polar-
neutral, hydrophobic, aromatic, cationic, or
anionic groups. The chemistry should therefore
be ¯exible in order to allow a large variety of
such structures to be covered by A and B.
(c) How can the intermediates and ®nal products be
eciently puri®ed? Combinatorial chemistry is
usually performed on a relatively small scale
with multiple reactions run in parallel. The chal-
lenge of rapid product puri®cation has been
addressed in several ways, especially using solid-
phase organic synthesis. Although solid-phase
synthesis9 allows for the very convenient removal
of excess reagents by simple ®ltration, it su€ers
from disadvantages relative to conventional
organic synthesis in solution. Reaction monitor-
ing is dicult, often anomalous reaction rates
are encountered, and insoluble reagents or cata-
lysts can usually not be used.
The most followed strategy for enhancing the anity of
an oligosaccharide ligand has therefore been to synthe-
size molecules containing all of the sugars that make
contact with the protein and then to alter these oligo-
saccharides, either through step-wise synthesis or com-
binatorial techniques, to yield additional interactions.
We present here an alternative approach, where only the
key sugar unit, despite being weakly inhibitory on its
own, is retained in the ligands in order to retain the
carbohydrate-binding speci®city of the natural struc-
ture. The approach is then to increase the anity of this
key sugar by the addition of additional molecular
structure that is unrelated to the sugars of the natural
oligosaccharide ligand. The resulting type of molecule,
which is particularly amenable to synthesis using com-
Recently, reports on solution-phase syntheses of chemi-
cal libraries have appeared. Some reports10,11 relied on
conventional silica gel column chromatography for
puri®cation, while others presented innovative proce-
dures for the puri®cation of products formed in solu-
tion. Immobilization of a substrate onto a precipitable
polymer enabled puri®cation of products by simple ®l-
tration.12 This method allowed the use of large excesses
of reagents as long as they were not precipitated.
Figure 1. Schematic of the binding of an oligosaccharide by a
protein. (A) The binding of the natural oligosaccharide show-
ing a buried key sugar residue and additional contacts with
adjacent sugar units. (B) A molecule containing the key sugar
residues and making additional non-carbohydrate interactions
with the proteins.

U. J. Nilsson et al./Bioorg. Med. Chem. 6 (1998) 1563±1575
1565
Solution synthesis of a carboxamide library has been
reported, where puri®cation relied on extraction of
reagents into acidic or alkaline aqueous phases leaving
the neutral reaction products in the organic phase.13
Such an approach does allow simple TLC monitoring of
reaction progress but requires that reagent excess and
by-products be either basic or acidic. A tetrasubstituted
ethylene library was synthesized in solution and the
products covalently captured on
a
resin.14 This
approach preserved the attractive features of performing
the reactions in solution while still taking advantage of
the easy puri®cation of a product attached to a resin.
However, if a second synthetic step is desired after resin
capture, it has to be performed on the resin. Similar
approaches, which capture excess reagents instead of
products, on a resin, have also been reported.15±17 Thus,
a second step could in this case be performed in solu-
tion. Finally, an elegant protocol, based on labeling
products with a per¯uorinated alkylsilyl group, was
recently introduced for solution synthesis of chemical
libraries.18,19 Three-phase liquid extraction with water,
dichloromethane, and per¯uorohexanes, allowed selec-
tive extraction of the labeled product into the per-
¯uorohexanes. The method was recently extended to
solid-phase, where ¯uorous reversed phase silica was
used to extract per¯uorinated alkylsilylated products
from reaction mixtures.20 Washing of the ¯uorous
reversed phase silica with acetonitrile removed reagents,
whereafter per¯uorinated alkylsilylated products were
eluted with hexane.
Figure 2. Solid-phase extraction of O-lauroylated sugars onto
C18 reverse-phase silica.
In order to validate this method for the puri®cation of
carbohydrate libraries, we selected the construction of a
small library terminating with the potential key sugar b-
d-Gal, thus yielding potential ligands for b-Gal-binding
lectins and b-galactosidase enzymes. A 1-thio-b-d-Gal
derivative (2), tagged with O-laurates, was synthesized
(Scheme 1) and used, after de-S-acetylation, in addition
reactions to four Michael acceptors (3±6) and one sub-
stitution of an a-halo-ketone (7). The anomeric oxygen
was replaced by the more nucleophilic sulfur in 2 to
permit these rapid and high-yielding reactions. The ®ve
carbonyl compounds thus formed (8±12, Scheme 2)
were subsequently reduced to the corresponding alcohol
or reductively aminated with glycine, b-alanine, l-leu-
cine, l-histidine, and l-tryptophan derivatives to pro-
duce a 1-thio-b-d-Gal based library of thirty (5Â6)
compounds produced as mixtures of four diastereomers
each (i.e. 120 di€erent 1-thio-b-d-galactosides, Fig. 4).
In the present work, we develop a novel approach to
purifying carbohydrate-based libraries synthesized in
solution by hydrophobic tagging of the carbohydrate
starting material to facilitate solid-phase extraction on
C18 silica. Hydrophobic tagging is a well-established
technique to enhance the chromatographic properties of
molecules, including carbohydrates,3,21 but has not yet
been used to enable solid-phase extraction for puri®ca-
tion of combinatorial libraries. Since the primary char-
acteristics of carbohydrates is their poly-hydroxylation,
we chose this feature to distinguish carbohydrate-con-
taining molecules from the organic reagents normally
used in combinatorial chemistry (to generate the scaf-
fold and groups A and B, Fig. 1(B)). This was achieved
by protecting the sugar OH-groups with long-chain
hydrophobic C-18 O-laurates which permitted the
quantitative adsorption of reaction products derived
from the starting sugar onto reverse-phase C18 silica gel
from which organic reagents and by products were
completely eluted by washing with methanol (Fig. 2).
Lauroylated carbohydrate-derived products were then
isolated by elution with pentane. The entire recovery
procedure including solvent exchanges and evaporation
takes 10±15 min per sample, and the reactions are easily
monitored by TLC.
This approach to a small b-d-Gal library was chosen for
several reasons. First, Michael additions and halide
Scheme 1. Reagents: (i) Lauroyl chloride, pyridine, DMAP,
pentane, 86%; (ii) AcSH, CF3SO3SiMe3, CH2Cl2, 77%.

1566
U. J. Nilsson et al./Bioorg. Med. Chem. 6 (1998) 1563±1575
Scheme 2. Reagents: (i) 3±7, CH2Cl2, Et2NH, 68±94%; (ii) NaBH4, (CH3)2CHOH, THF, 71±98%; (iii) Glycine t-butyl ester hydro-
chloride, b-alanine t-butyl ester hydrochloride, l-leucine t-buthyl ester hydrochloride, l-histidine methyl ester hydrochloride, or
l-tryptophan methyl ester hydrochloride, NaCNBH3, MeCN, THF, 73±99%.
substitution by thiols, as well as reductions and reduc-
tive aminations of ketones, are general and high-yield-
ing reactions. Second, a large number of Michael
acceptors, a-halo-carbonyl compounds and amines
(including amino acids) are commercially available,
which ensures rapid access to large and diverse libraries.
Solid-phase extractions. Reaction mixtures containing
tetra-O-lauroylated products (i.e. 8±42) were con-
centrated, the residue was re-dissolved in methanol and
applied onto the C18 silica. Reagents were eluted by
washing with methanol and the desired products were
eluted with pentane (Fig. 2). In all cases described in this
paper, no lauroylated products were present in the
MeOH washing according to TLC and no reagents
could be detected in the pentane eluate by 1H NMR,
MALDI-TOF MS, or TLC. This puri®cation protocol
has some similarities to that of solid-phase synthesis,
with the di€erence being that the product is attached to
a solid-phase via a noncovalent and reversible hydro-
phobic interaction in our protocol. Solid-phase extrac-
tion on C18 silica had the advantages in that
puri®cation is rapid and ecient and that large excesses
of reagents can be used to drive reactions to completion.
The advantages over solid-phase synthesis using this
approach is that it allows simple monitoring of each
reaction by TLC and characterization of each inter-
mediate and ®nal product by their MALDI-TOF MS
Results and Discussion
The 1-thio-b-d-Gal derivative 2, tagged with hydro-
phobic O-laurates to enable solid-phase extraction onto
C18 silica, was prepared in two steps from d-Gal
(Scheme 1). Acylation with lauroyl chloride, pyridine,
and DMAP, gave the per-O-lauroylated derivative 1 in
86% yield. Treatment of 1 with thioacetic acid and cat-
alytic trimethylsilyl tri¯uoromethanesulfonate in
dichloromethane furnished the key b-thioacetate 2
(77%). Formation of the corresponding a-thioacetate
could not be detected, which was in contrast to when
boron tri¯uoride etherate was used as a catalyst.

U. J. Nilsson et al./Bioorg. Med. Chem. 6 (1998) 1563±1575
1567
1
and H NMR spectra without the need for a cleavage
step. The solution-phase strategy allowed us to routinely
work on a scale providing 10±30 mg of ®nal compound
for each individual library member. The length of the
fatty acid ester protecting group was found to be critical
for successful solid-phase extraction, since the tetra-O-
octanoylated 1-thio-b-d-gal derivative, corresponding to
2, did not quantitatively adsorb on C18 silica when dis-
solved in methanol. Furthermore, C18 silica from sev-
eral suppliers were evaluated in this procedure. Waters
Prep C18 was found to produce the most consistent
results and at the same time was the easiest to handle
due to its large particle size.
Michael additions and ꢁ-chloro ketone substitutions
Our ®rst attempts to use the 1-thio-galactose derivative
2 in Michael additions involved selective removal of the
anomeric thioacetate22 to give the corresponding thiol,
followed by Michael addition catalysed by tetra-
butylammonium ¯uoride,23 sodium methoxide, or tri-
ethylamine. These approaches gave good yields of
Michael adducts in trial reactions with acrolein and 2-
cyclohexen-1-one (4), but both the de-S-acetylation and
Michael additions were always accompanied by an
interfering formation of the disul®de dimer Gal-S-S-
Gal. However, the disul®de formation could be sup-
pressed by performing the de-S-acetylation and Michael
addition or a-halo-carbonyl substitution in one pot.
This was accomplished using a modi®cation of the pro-
cedure of Bennett et al.,24 which involved treatment of a
mixture of 2 and the Michael acceptor or a-halo-carbo-
nyl compound with diethylamine or piperidine in
dichloromethane. Reproducibility and yields were con-
sistently better using dichloromethane as compared to
other solvents (DMF, THF, toluene, EtOAc, or MeCN).
Figure 3. Commercial Michael acceptors and a-halo ketones
used in the present study. (A) Compounds that gave high yields
and rapid reactions with a 1-thio-Gal nucleophile. (B) Com-
pounds of low reactivity.
Of the 21 alkylating agents of suitable reactivity
(Fig. 3(A)), we chose four commercial Michael accep-
tors (3±6) and one a-halo-carbonyl (7) compound to
demonstrate the ease of synthesis of a b-Gal-based
library. Cyclic electrophiles were chosen for the reasons
discussed above.
Reactions between 2 and the electrophiles 3±7 were
performed in a parallel fashion, providing ®ve, physi-
cally separated, ketones (8±12) (Scheme 2). As expected,
none of the electrophiles 3±7 showed any stereo-
selectivity in reaction with compound 2 and diastereo-
mers were formed in near equivalent amounts. Solid-
phase extraction on C18 silica of the ketones (8±12)
worked very well on a smaller scale (<0.5 g) providing
products in high yields and purities. However, column
chromatography on silica can be used as an alternative
more cost-ecient puri®cation method when reactions
are performed on larger scale (1±3 g). On the other
hand, further derivatization of the ketones 8±12, as
described below, involved 60 reactions (30 parallel
ketone derivatizations on 70±200 mg scales (13±42) fol-
lowed by 30 deblocking reactions (43±72)) for which
solid-phase extraction was essential for the synthesis to
be completed within a reasonable time. The ®nal galac-
tose-based library was then obtained by splitting each of
the ketones (8±12) in six batches, each of which was
further reacted in a parallel fashion.
With an ecient one-pot procedure for the activation of
2 and Michael addition/a-halo-carbonyl substitution in
hand, 35 di€erent commercially available Michael
acceptors and a-halo-carbonyl compounds were eval-
uated (Fig. 3). These experiments revealed that Michael
acceptors carrying bulky substituents on the g-carbon or
having tri- or tetra-substitution of the a,b-double bond
were less ecient in Michael additions with 2 (Fig. 3(B)).
Puri®cation reactions mixtures using solid-phase
extraction on C18 silica proved to be ecient. Not even
the more hydrophobic electrophiles (e.g. trans-4-phenyl-
3-buten-2-one, 3-nonen-2-one, 2-cyclohepten-1-one, or
trans-1-(4,5-dimethoxy-2-methylphenyl)-2-buten-2-one
in Figure 3(A)) adsorbed to the C18 silica during the
MeOH wash, while tetra-O-lauroylated products e-
ciently adsorbed to the C18 silica during the MeOH
wash and were recovered in high yields after elution
with pentane. This indicates that the presence of multi-
ple O-laurates is crucial for adsorbtion to the C18 silica.

1568
U. J. Nilsson et al./Bioorg. Med. Chem. 6 (1998) 1563±1575
Reduction of ketones to alcohols. The ketones 8±12 were
reduced to the corresponding alcohols (13±17) using
sodium borohydride in isopropanol/tetrahydrofuran
(Scheme 2). The purity of the alcohols 13±17 after solid-
phase extraction on C18 silica were >90% as assessed
derivatives (Scheme 2). Compounds 18±27 were >90%
pure according to 1H NMR, MALDI-TOF MS, and
TLC after solid-phase extraction on C18 silica. The use
of solvents other than acetonitrile (methanol, DMF,
neat THF, dichloromethane, ethyl acetate) resulted in
lower yields and sometimes uneven distribution of dia-
stereomers. Furthermore, reductive aminations of the
ketones 8±12 with a-substituted l-amino acids (l-leu-
cine as the t-butyl ester hydrochloride and l-histidine
and l-tryptophan as methyl ester hydrochlorides) under
these conditions also worked well, yielding the desired
N-alkylated l-leucine (28±32), l-histidine (33±37), and
l-tryptophan (38±42) derivatives (Scheme 2). However,
varying amounts (10±50%) of the corresponding alco-
hols (13±17) were formed as the only side-product in
these reactions. Presumably, formation of the inter-
mediate iminium ion is disfavoured by a-substituents on
the amino acids and competing reduction of the ketone
to the corresponding alcohol occurs to a signi®cant
extent. Addition of acids,26±29 in order to catalyse imi-
nium ion formation, or the use of sodium triacetoxy-
borohydride30,31 as reducing agent, did not decrease
the formation of the corresponding alcohols. However,
in cases where a mixture of a N-alkylated amino acid
and an alcohol was obtained, the alcohol was readily
1
by H NMR, MALDI-TOF MS, and TLC. The reduc-
tions proceeded with signi®cant stereoselectivity and
two out of the four possible diastereomers usually
dominated in the product mixture. In the ®nal step, the
O-laurates were removed with methanolic sodium
methoxide and the methyl laurate formed was separated
from the product by simply passing the mixture
through C18 silica with 80% methanol, conditions
under which the O-laurates quantitatively remained
bound to the C18 silica. The ®ve alcohols 43±47 (Fig. 4)
were obtained in purities of >90% and yields of 78±
99%.
Reductive aminations of ketones with amino acid esters.
Reductive aminations of the ketones 8±12 with the
unsubstituted amino acids (glycine and b-alanine t-butyl
ester hydrochlorides), using sodium cyanoborohydride25
in acetonitrile/THF, were high-yielding and produced
an even distribution of all four possible diastereomers of
the N-alkylated glycine (18±22) and b-alanine (23±27)
Figure 4. The ®nal 1-thio-b-d-galatoside library, consisting of 30 physically seperated compounds as mixtures of four diastereomers
each.

U. J. Nilsson et al./Bioorg. Med. Chem. 6 (1998) 1563±1575
1569
Figure 5. Evaluation of 43±73 as inhibitors of b-galactosidase from E. coli. The horizontal line shows the relative initial rate of
hydrolysis in the presence of the reference inhibitor cyclohexyl 1-thio-b-d-galactoside (73).
separated from the N-alkylated amino acid on a silica
Sep-Pak cartridge after removal protecting groups (O-
laurates and methyl or t-butyl esters) as described below.
The potential biological activity of this library was
evaluated, for demonstration purposes, against the b-
galactosidase from Escherichia coli, known to be inhib-
ited by 1-thio-b-d-galactosides.32 Cyclohexyl 1-thio-b-d-
galactoside 7333 was chosen as a reference inhibitor,
since it would re¯ect the e€ect of an anomeric sulfur
atom and a cycloalkane aglycon have on inhibitory
power. In order to rapidly obtain information of the
inhibitory power of the library members, an initial
screen was performed by brie¯y incubating the enzyme
with p-nitrophenyl b-d-galactoside in the presence of the
inhibitors 43±73. The optical density at 405 nm was used
as an approximation of the relative initial rates of
hydrolysis (Fig. 5). The initial rate of hydrolysis in the
presence of the reference inhibitor 73 was 28% of the
rate in the absence of an inhibitor, which is indicated
with a horizontal line in Figure 5. The important e€ect
of the aglycon on the inhibitory power individual library
members was clearly apparent. Eleven inhibitors were
better and 19 were worse than 73. Overall trends that
could be observed were that alcohols and N-alkylated
tryptophan derivatives, which are among the more
hydrophobic library members, were good inhibitors.
Furthermore, 1-thio-b-d-galactosides carrying a gem-
dimethylated cyclopentane aglycon (43, 53, 58, 63, and
68) were generally good inhibitors.
t-Butyl esters were removed from the protected N-alky-
lated glycine (18±22), b-alanine (23±27) and l-leucine
(28±32) derivatives using tri¯uoroacetic acid in dichloro-
methane, followed by removal of O-laurates using
methanolic sodium methoxide. The residual mixtures of
N-alkylated amino acid (48±62), methyl laurate and, in
some cases, the corresponding alcohol (43±47), were
applied onto silica normal-phase Sep-Pak cartridges.
The cartridges were washed with dichloromethane:
methanol 9:1 (to remove methyl laurate and the corre-
sponding alcohol) and the ®nal N-alkylated glycine (48±
52), b-alanine (53±57), and l-leucine (58±62) derivatives
(Figure 4) were eluted with dichloromethane:methanol:
water (65:35:5) in 31±99% yields and purities of >90%.
O-Laurates were removed from the protected N-alky-
lated l-histidine (33±37) and l-tryptophan (38±42) deri-
vatives using methanolic sodium methoxide. The methyl
laurate formed was thereafter removed by passing the
residue through C18 silica using 80% methanol as a
solvent. The eluate contained the desired product while
methyl laurate was adsorbed onto the C18 silica. Methyl
esters were then hydrolysed using aqueous LiOH and
the resulting N-alkylated amino acids (63±72) were
separated from traces of their corresponding alcohols
(43±47) on a silica Sep-Pak cartridges, as described
above for 48±62. The ®nal N-alkylated l-histidine (63±
67) and l-tryptophan (68±72) derivatives (Fig. 4) were
obtained in 45±93% yields and purities of >90%.
Each of the library reactions used produced mixtures of
diastereomers and in order to identify the active con-
stituent(s) the compounds would clearly have to be
resolved. The diastereomeric mixture of 13 could indeed
be partially resolved by chromatography on silica gel and,
on removal of the O-laurates, yielded two major isomers
43A and 43B (ꢀ90%) together with a mixture of the two
minor diastereomers (43CD). Their structures were
con®rmed with high-resolution mass spectroscopy and
Inhibition of ꢀ-galactosidase from E. coli. The 5Â6 1-
thio-b-d-Gal library thus prepared is shown in Figure 4.

1570
U. J. Nilsson et al./Bioorg. Med. Chem. 6 (1998) 1563±1575
preliminary 2-D ROESY NMR experiments established
that the aglycons of 43A and 43B had cis con®guration
(i.e. (1S,4S)-4-hydroxy-2,2-dimethyl-cyclopent-1-yl 1-thio-
b-d-galactopyranoside and (1R,4R)-4-hydroxy-2,2-
dimethyl-cyclopent-1-yl 1-thio-b-d-galactopyranoside).
Inhibitor evaluation of these compounds revealed that
43A and 43B were good inhibitors, while 43CD inhib-
ited the enzyme poorly (Fig. 6). This indicated that the
two major diastereomers, 43A and 43B, were respon-
sible for the strong inhibitory activity of the mixture of
four diastereomers (43).
and ecient isolation tool in the synthesis of carbohy-
drate libraries. The protocol combines several advan-
tages of solution synthesis (ease of reaction monitoring
and intermediate characterization) with those of solid-
phase synthesis (rapid puri®cation, use of excess
reagents to `drive' reactions). To demonstrate the utility
Table 1. Inhibition constants against b-galactosidase from E.
coli
Compd
Ki/mM
43A
43B
73
3.1
1.7
14
Compounds 43A and 43B were shown to be competitive
inhibitors and the inhibition constants determined to be
3.1 and 1.7 mM, respectively. Determination of the inhi-
bition constant of 73 (14 mM), showed that relatively
simple changes of the aglycon of a 1-thio-b-d-galacto-
side inhibitor (43B versus 73) had signi®cant e€ects on
Table 2. MALDI mass spectrometry data for lauroylated
library intermediates
the
inhibition
constant
(1.7
versus
14 mM;
ÁÁG&5.2 kJ/mol). The results parallel the well-known
observation that hydrophobic aglycons (alkyl or aryl
aglycons) signi®cantly enhance the binding of glycosides
to many lectins and enzymes, including b-galactosidase
from E. coli,32 and the relatively strong inhibition of
43B may be attributed to the hydrophobic nature of its
aglycon. However, the fact that the di€erent stereo-
isomers of 43 displayed di€erent inhibitor potencies and
that 43B was almost one order of magnitude better than
cyclohexyl 1-thio-b-d-galactopyranoside 73, suggests a
more speci®c hydrophobic interaction where the size
and shape of the aglycon is important. Furthermore, it
can not be excluded that the aglycon hydroxyl group of
43B is involved in a hydrogen bond to the enzyme.
Compd
Formula
M calcd
(M+X)+ founda
8
C61H110O10
S
S
S
S
S
S
S
S
S
S
1035.59
1021.57
1050.57
1035.59
1021.57
1037.61
1023.59
1052.63
1037.61
1023.59
1150.77
1136.75
1164.80
1150.77
1136.75
1164.80
1150.77
1178.83
1164.80
1150.77
1206.88
1192.86
1219.90
1206.88
1192.86
1188.78
1174.76
1201.75
1188.78
1174.76
1237.85
1223.83
1251.83
1237.85
1223.83
1060.6
1045.9
9
C
C
60H108O10
62H112O10
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1074.3
1058.9
C61H110O10
C
C
C
C
C
60H108O10
61H112O10
60H110O10
62H114O10
61H112O10
1045.3
1062.3
1044.7
1077.8
1060.6
C60H110O10
1047.0
C
C
67H123NO11
66H121NO11
S
S
1115.1,b 1170.2
1083.7b
1112.6b
1096.9,b 1153.2
1139.0
C468H125NO11
S
C67H123NO11
S
C66H121NO11
C
S
68H125NO11
S
S
S
S
S
S
S
S
S
S
1116.8b
1095.8b
1182.0
Conclusions
C67H123NO11
C69H127NO11
C68H125NO11
Solid-phase extraction on C18 silica, relying on the pre-
sence of multiple O-laurates, was developed as a novel
1109.1,b 1165.8
C
C
67H123NO11
71H131NO11
1192.8
1150.5b
1193.8
1167.6,b 1225.7
1151.5,b 1229.4
1137.2,b 1192.8
1188.4
C70H129NO11
C
C
C
72H133NO11
71H131NO11
70H129NO11
C68H121N3O11
S
S
S
S
S
S
S
S
C
C
C
67H119N3O11
69H123N3O11
68H121N3O11
1172.0
1201.0
1188.3
1173.9
C67H119N3O11
C
C
C
73H124N2O11
72H122N2O11
74H126N2O11
1239.7
1223.0
1252.7
1238.6
1223.9
C73H124N O11
C
S
2
72H122N2O11
S
Figure 6. The relative initial rate of hydrolysis of p-nitrophenyl
b-d-galactopyranoside by b-galactosidase from E. coli. in the
presence of 43A, 43B, and 43CD.
aX = H, Na, or K.
t
bLoss of Bu due to fragmentation.

U. J. Nilsson et al./Bioorg. Med. Chem. 6 (1998) 1563±1575
1571
Table 3. MALDI mass and 1H NMR spectroscopy data for the ®nal b-d-galactoside library
Compd
Formula
M calcd
(M+X)+ founda
d/ppmb
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
C13H24O6S
308.40
294.36
322.42
308.40
294.36
365.44
351.42
379.47
365.44
351.42
379.44
365.45
393.50
379.45
365.45
421.55
407.53
436.58
421.55
407.53
445.53
431.50
459.56
445.53
431.50
494.60
480.59
508.63
494.60
480.59
332.1
317.3
4.305, 4.310, 4.315, 4.34
4.38, 4.417, 4.422
4.32, 4.33, 4.34
C12H22O6S
C14H26O6S
C13H24O6S
346.6
332.1
4.381, 4.389, 4.394
4.34, 4.39, 4.43, 4.55
4.30, 4.320, 4.325, 4.330
4.36, 4.40, 4.44, 4.46
4.31, 4.34, 4.36, 4.38
4.38, 4.42, 4.45, 4.46
4.47, 4.48, 4.50, 4.58
4.310, 4.315, 4.33, 4.34
4.36, 4.43, 4.44, 4.46
4.32, 4.34, 4.35
C12H22O6S
318.8
368.0
375.5
C
15H27NO7S
14H25NO7S
16H29NO7S
C
C
380.6, 403.5
367.4, 389.9
353.5, 376.5
383.2
C15H27NO7S
C
C
14H25NO7S
16H29NO7S
C15H27NO7S
367.0, 389.9
399.3, 419.5
381.7, 403.5
367.4, 389.9
422.3
C
C
C
17H31NO7S
16H29NO7S
15H27NO7S
4.40, 4.46
4.34, 4.42, 4.44, 4.47
4.315, 4.320
C19H35NO7S
C
C
C
18H33NO7S
20H37NO7S
19H35NO7S
408.4
4.33, 4.40, 4.42, 4.46
4.34, 4.38
4.36, 4.43, 4.44
438.0, 461.4
421.7, 448.0
410.9, 435.5
447.6
C18H33NO7S
4.37, 4.41, 4.44, 4.49
4.30, 4.305, 4.32, 4.33
4.35, 4.39, 4.425, 4.43
4.34, 4.357, 4.364
4.32, 4.41, 4.44, 4.50
4.28, 4.40, 4.45, 4.52
4.13,4.20, 4.22, 4.26
4.24, 4.33, 4.35, 4.36
4.21, 4.26, 4.30
C
19H31N3O7S
18H29N3O7S
20H33N3O7S
19H31N3O7S
C
C
C
433.2
462.2
448.0
433.6
C18H29N3O7S
C24H34N2O7S
C23H32N2O7S
C25H36N2O7S
C24H34N2O7S
C23H32N2O7S
495.6
483.9
512.1
495.9
481.9, 505.3
4.19, 4.29, 4.37
4.09, 4.24, 4.40, 4.46
aX=H or Na.
bAnomeric protons. J=9.5±10 Hz. The observation of less than four anomeric signals is a consequence of overlapping signals.
of this solid-phase extraction protocol, high-yielding
procedures for Michael additions and a-halo-carbonyl
substitutions with a 1-thio-b-d-galactoside derivative, as
well as for reductive aminations with amino acids, were
employed in a practical solution-phase synthesis of a
small 1-thio-b-d-galactoside library containing 10±
30 mg of each individual library compound. The 1-thio-
b-d-galactoside library was screened for inhibitors
against b-galactosidase from E. coli, which showed that
the nature of the aglycon strongly in¯uenced the inhibi-
tory power of individual library members. Despite the
limited size of this demonstration library, an inhibitor
(43B) almost one order of magnitude better than the
reference compound cyclohexyl 1-thio-b-d-galactopyr-
anoside 73, was identi®ed. These results show that solid-
phase extraction on C18 silica is an ecient isolation
tool in the synthesis of carbohydrate libraries and com-
binatorial addition of diverse noncarbohydrate struc-
tures to a natural carbohydrate ligand is a feasible route
towards the discovery of potent inhibitors of carbo-
hydrate binding proteins.
Experimental
1
General. H NMR spectra were recorded with either a
Bruker AM-360 or a Varian UNITY 500 spectrometer.
MALDI-TOF mass spectra were recorded with a Kra-
tos Kompact MALDI instrument. Optical rotations
were measured with a Perkin-Elmer 241 polarimeter.
Solid-phase extraction were performed with Prep
C18, 55±105 mm, 125A (Waters Corp., Milford, USA).
Reactions were monitored by TLC on Silica Gel FG254
(E. Merck, Darmstadt, Germany). 1H NMR assign-
ments were derived from COSY experiments. Microtiter
plate absorbances were read on a Thermomax micro-
plate reader (Molecular Devices). b-Galactosidase from
E. coli was from SIGMA (G6512). The bu€er used for
enzyme assays was 0.15 M PBS and 0.02 mM MgCl2 at
pH 7.2.
Solid-phase extraction of lauroylated compounds. Fol-
lowing Michael addition, a-chloro substitution, reduc-
tion, or reductive amination, the reaction mixture was

1572
U. J. Nilsson et al./Bioorg. Med. Chem. 6 (1998) 1563±1575
concentrated, re-dissolved in methanol, and applied
onto C18 silica (1 g per 20 mg lauroylated carbohy-
drate), which had been pre-conditioned with pentane
(50 mL) and MeOH (50 mL). In some cases , when the
lauroylated product was poorly soluble in MeOH, the
residue was sonicated in MeOH to a milky suspension
before application onto the C18 silica. The C18 silica
was washed with methanol (10 mL/g C18 silica), where-
after the product was eluted with pentane (10 mL/g C18
silica). No remaining reagents could in any case be
detected by TLC, 1H NMR, or MALDI-TOF mass
spectroscopy.
was added Et2NH (0.45 mL). After 2 h, the mixture was
concentrated and puri®ed according to the solid-phase
extraction protocol to give 11 (203 mg, 90%).
Typical procedure for larger scale (>1 g) Michael
addition/ꢁ-halo-carbonyl substitution; 2,2-dimethylcyclo-
hexan-5-on-1-yl 2,3,4,6-tetra-O-lauroyl-1-thio-ꢀ-D-galacto-
pyranoside (10). To compounds 2 (1.02 g, 1.05 mmol)
and 5 (400 mg, 3.16 mmol) in dry CH2Cl2 (3 mL) under
argon, was added Et2NH (2 mL). After 3 h, the mixture
was concentrated and puri®ed by column chromato-
graphy (SiO2, pentane:EtOAc, 7:1) to give 10 (758 mg,
75%). In parallel reactions were obtained 2,2-dimethyl-
cyclopentan-4-on-1-yl 2,3,4,6-tetra-O-lauroyl-1-thio-b-
1,2,3,4,6-Pentꢁ-O-lauroyl-ꢁ-D-galactopyranose (1). To a
suspension of d-galactose (5.0 g, 27.8 mmol), pyridine
(60 mL), and 4-dimethylaminopyridine (cat.) in pentane
(195 mL) under argon atmosphere, was added lauroyl
d-galactopyranoside
2,3,4,6-tetra-O-lauroyl-1-thio-b-d-galactopyranoside
8 (71%), cyclohexan-3-on-1-yl
9
(88%), cycloheptan-3-on-1-yl 2,3,4,6-tetra-O-lauroyl-1-
thio-b-d-galactopyranoside 11 (79%), and cyclohexan-
2-on-1-yl 2,3,4,6-tetra-O-lauroyl-1-thio-b-d-galactopyr-
anoside 12 (94%).
ꢁ
chloride (60 mL, 252 mmol) at 78 C. The mixture was
allowed to reach ambient temperature. The white slurry
slowly dissolved and a ®ne precipitate of pyridinium
hydrochloride formed. After 40 h, the pyridinium
hydrochloride was ®ltered o€ and the pentane solution
was concentrated. Column chromatography (SiO2, pen-
tane:EtOAc 20:19:1 gradient) gave 1 (26.1 g, 86%),
Typical procedure for reduction; 2,2-dimethyl-4-hydroxy-
cyclopent-1-yl 2,3,4,6-tetra-O-lauroyl-1-thio-ꢀ-D-galacto-
pyranoside 13. To 8 (80 mg, 77 mmol) in dry tetra-
hydrofuran (2.0 mL) and isopropanol (0.5 mL) under
argon atmosphere, was added NaBH4 (30 mmol). After
25 min, AcOH (30 mL) was added and the mixture was
concentrated and puri®ed according to the solid-phase
extraction protocol to give 13 (65 mg, 81%). In parallel
reactions were obtained 3-hydroxy-cyclohex-1-yl
2,3,4,6-tetra-O-lauroyl-1-thio-b-d-galactopyranoside 14
(86%), 2,2-dimethyl-5-hydroxy-cyclohex-1-yl 2,3,4,6-
tetra-O-lauroyl-1-thio-b-d-galactopyranoside 15 (71%),
3-hydroxy-cyclohept-1-yl 2,3,4,6-tetra-O-lauroyl-1-thio-
b-d-galactopyranoside 16 (72%), and 2-hydroxy-cyclo-
hex-1-yl 2,3,4,6-tetra-O-lauroyl-1-thio-b-d-galactopyr-
anoside 17 (98%).
[a]d +39ꢁ (c 0.9, CHCl3). 1H NMR data (CDCl3) d
25
6.39 (d, 1H, J=2.4 Hz, H-1), 5.51 (br s, 1H, H-4), 5.35
(m, 2H, H-2 and H-3), 4.32 (br t, 1H, J=6.6 Hz, H-5),
4.08 (d, 2H, J=6.6 Hz, H-6a and H-6b), 2.39, 2.38, 2.30,
2.26 (4 t, 2H each, J=7.5 Hz, -CH2CO-), 2.21 (m, 2H,
-CH2CO-), 0.88 (t, 15 H, J=7.0 Hz, -CH3). Anal. calcd
for C66H122O11: C, 72.2; H, 11.3. Found: C, 72.6; H,
11.5.
1-S-Acetyl-2,3,4,6-tetra-O-lauroyl-1-thio-ꢀ-D-galactopyr-
anose (2). To 1 (20.0 g, 18.2 mmol) and thioacetic acid
(5.0 mL, 1.9 equiv) in dry dichloromethane (300 mL)
under argon, was added trimethylꢁsilyl tri¯uoromethane-
sulfonate (5.0 mL, 0.5 equiv) at 0 C. The cold-bath was
immediately removed and after 48 h the mixture was
diluted with dichloromethane, washed with satd
sodium hydrogencarbonate, dried (Na2SO4), and concen-
trated. Column chromatography (SiO2, pentane:EtOAc,
20:1) gave 2 (13.7 g, 77%), [a]d25 +21ꢁ (c 1, CHCl3). 1H
NMR data (CDCl3) d 5.47 (d, 1H, J=3.4 Hz, H-4), 5.32
(t, 1H, J=10.0 Hz, H-2), 5.25 (d, 1H, J=10.0 Hz, H-1),
5.12 (dd, 1H, J=3.4 and 10.0 Hz, H-3), 4.08 (m, 3H, H-
5, H-6a and H-6b), 2.14±2.43 (m, 8H, -CH2CO-), 2.37
(s, 3H, -SAc), 0.88 (t, 15 H, J=7.0 Hz, -CH3). Anal.
calcd for C56H102O10S: C, 69.5; H, 10.6; S, 3.3. Found:
C, 69.4; H, 10.8; S, 3.5.
Typical procedure for reductive amination with amino
acids; Nꢁ-[3-(2,3,4,6-tetra-O-lauroyl-1-thio-ꢀ-D-galacto-
pyranosyl)-cyclohept-1-yl]-L-tryptophan methyl ester (41).
To the ketone 11 (115 mg, 111 mmol) and l-tryptophan
hydrochloride methyl ester (283 mg, 1.11 mmol) in
dry MeCN (2.6 mL) and THF (0.9 mL), was added
NaCNBH3 (0.1 mmol). After 73 h, the mixture was
concentrated and puri®ed according to the solid-phase
extraction protocol to give 41 (129 mg, 95%). (Yields in
reductive aminations are calculated based on mass
recovery and products contain varying amounts (0
toꢀ50%) of contaminating alcohols 13±17 as described
under Results and Discussion.) In parallel reactions
were obtained Na-[4,4-dimethyl-3-(2,3,4,6-tetra-O-laur-
oyl-1-thio-b-d-galactopyranosyl)-cyclopent-1-yl]-glycine
t-butyl ester 18 (98%), Na-[3-(2,3,4,6-tetra-O-lauroyl-1-
thio-b-d-galactopyranosyl)-cyclohex-1-yl]-glycine t-butyl
ester 19 (90%), Na-[4,4-dimethyl-3-(2,3,4,6-tetra-O-
Typical procedure for small scale (<0.5 g) Michael
addition/ꢁ-halo-carbonyl substitution; cycloheptan-3-on-
1-yl 2,3,4,6-tetra-O-lauroyl-1-thio-ꢀ-D-galactopyranoside
(11). To compounds 2 (210 mg, 0.22 mmol) and 6
(48 mg, 0.43 mmol) in dry CH2Cl2 (1.8 mL) under argon,

U. J. Nilsson et al./Bioorg. Med. Chem. 6 (1998) 1563±1575
1573
lauroyl-1-thio-b-d-galactopyranosyl)-cyclohex-1-yl]-
glycine t-butyl ester 20 (94%), Na-[3-(2,3,4,6-tetra-O-
lauroyl-1-thio-b-d-galactopyranosyl)-cyclohept-1-yl]-
glycine t-butyl ester 21 (87%), Na-[2-(2,3,4,6-tetra-O-
lauroyl-1-thio-b-d-galactopyranosyl)-cyclohex-1-yl]-
glycine t-butyl ester 22 (93%), Nb-[4,4-dimethyl-3-
(2,3,4,6-tetra-O-lauroyl-1-thio-b-d-galactopyranosyl)-
cyclopent-1-yl]-b-alanine t-butyl ester 23 (97%), Nb-[3-
(2,3,4,6-tetra-O-lauroyl-1-thio-b-d-galactopyranosyl)-
cyclohex-1-yl]-b-alanine t-butyl ester 24 (79%), Nb-[4,4-
dimethyl-3-(2,3,4,6-tetra-O-lauroyl-1-thio-b-d-galacto-
pyranosyl)-cyclohex-1-yl]-b-alanine t-butyl ester 25
(82%), Nb-[3-(2,3,4,6-tetra-O-lauroyl-1-thio-b-d-galacto-
pyranosyl)-cyclohept-1-yl]-b-alanine t-butyl ester 26
(77%), Nb-[2-(2,3,4,6-tetra-O-lauroyl-1-thio-b-d-galacto-
pyranosyl)-cyclohex-1-yl]-b-alanine t-butyl ester 27 (73%),
Na-[4,4-dimethyl-3-(2,3,4,6-tetra-O-lauroyl-1-thio-b-d-
galactopyranosyl)-cyclopent-1-yl]-l-leucine t-butyl ester
28 (79%), Na-[3-(2,3,4,6-tetra-O-lauroyl-1-thio-b-d-
galactopyranosyl)-cyclohex-1-yl]-l-leucine t-butyl ester
29 (85%), Na-[4,4-dimethyl-3-(2,3,4,6-tetra-O-lauroyl-1-
thio-b-d -galactopyranosyl)-cyclohex-1-yl]-l -leucine
t-butyl ester 30 (95%), Na-[3-(2,3,4,6-tetra-O-lauroyl-1-
thio-b-d-galactopyranosyl)-cyclohept-1-yl]-l-leucine
t-butyl ester 31 (99%), Na-[2-(2,3,4,6-tetra-O-lauroyl-
1-thio-b-d-galactopyranosyl)-cyclohex-1-yl]-l-leucine
t-butyl ester 32 (93%), Na-[4,4-dimethyl-3-(2,3,4,6-tetra-
O-lauroyl-1-thio-b-d-galactopyranosyl)-cyclopent-1-yl]-
l-histidine methyl ester 33 (82%), Na-[3-(2,3,4,6-tetra-O-
lauroyl-1-thio-b-d-galactopyranosyl)-cyclohex-1-yl]-l-
histidine methyl ester 34 (92%), Na-[4,4-di-methyl-3-
(2,3,4,6-tetra-O-lauroyl-1-thio-b-d-galactopyranosyl)-c-
yclohex-1-yl]-l-histidine methyl ester 35 (87%), Na-[3-(2,
3, 4, 6-tetra-O-lauroyl-1-thio-b-d -galactopyranosyl)-
cyclohept-1-yl]-l-histidine methyl ester 36 (87%), Na-[2-
(2,3,4,6-tetra-O-lauroyl-1-thio-b-d-galactopyranosyl)-
cyclohex-1-yl]-l-histidine methyl ester 37 (quant),
Na-[4,4-dimethyl-3-(2,3,4,6-tetra-O-lauroyl-1-thio-b-d-
galactopyranosyl)-cyclopent-1-yl]-l-tryptophan methyl
ester 38 (95%), Na-[3-(2,3,4,6-tetra-O-lauroyl-1-thio-b-
d-galactopyranosyl)-cyclohex-1-yl]-l-tryptophan methyl
ester 39 (94%), Na-[4,4-dimethyl-3-(2,3,4,6-tetra-O-
lauroyl-1-thio-b-d-galactopyranosyl)-cyclohex-1-yl]-l-
tryptophan methyl ester 40 (95%), and Na-[2-(2,3,4,6-
tetra-O-lauroyl-1-thio-b-d-galactopyranosyl)-cyclohex-1-
yl]-l-tryptophan methyl ester 42 (92%).
125 A, 7 g). The C18 silica was washed with water
(50 mL), whereafter the ®nal de-protected alcohol was
eluted with 70% methanol (50 mL) and concentrated to
give a quantitative yield of 47. In parallel reactions were
obtained 2,2-dimethyl-4-hydroxy-cyclopent-1-yl 1-thio-
b-d-galactopyranoside 43 (78%), 3-hydroxy-cyclohex-1-yl
1-thio-b-d-galactopyranoside 44 (quant), 2,2-dimethyl-
5-hydroxy-cyclohex-1-yl 1-thio-b-d-galactopyranoside
45 (quant), 3-hydroxy-cyclohept-1-yl 1-thio-b-d-galacto-
pyranoside 46 (95%).
Typical procedure for deblocking of N-alkylated amino
acid t-butyl esters; Nꢁ-[3-(1-thio-ꢀ-D-galactopyranosyl)-
cyclohept-1-yl]-L-leucine (61). Compound 31 (119 mg,
98.4 mmol) was treated with tri¯uoroacetic acid (3.5 mL)
in dry dichloromethane (3.5 mL) for 330 min. n-Propyl
acetate (7 mL) and toluene (20 mL) were added and the
mixture was concentrated, then co-concentrated twice
with toluene. To the residue in dry methanol (7 mL) and
dichloromethane (1.2 mL) under argon atmosphere, was
added methanolic sodium methoxide (1 M, 197 mL).
After 20 h, the mixture was neutralized with Amberlite
IRC-50S (H+) resin, ®ltered and concentrated. The
residue was dissolved in dichloromethane:methanol
(9:1) and applied onto a Waters Sep-Pak Plus Longbody
SiO2 cartridge. The cartridge was washed with dichloro-
methane:methanol (9:1, 20 mL) to remove methyl lau-
rate and traces of the corresponding alcohol (46),
whereafter the 61 was eluted with dichloromethane:
methanol:water (65:35:5, 20 mL) and concentrated. The
residue was dissolved in water and applied onto a col-
umn of C18 silica (Waters Prep C18, 125 A, 7 g). The
C18 silica was washed with water (50 mL), whereafter
elution with methanol (50 mL) and concentration gave
61 (37.7 mg, 91%). In parallel reactions were obtained
Na-[4,4-dimethyl-3-(1-thio-b-d-galactopyranosyl)-cyclo-
pent-1-yl]-glycine 48 (56%), Na-[3-(1-thio-b-d-galacto-
pyranosyl)-cyclohex-1-yl]-glycine 49 (85%), Na-[4,4-
dimethyl-3-(1-thio-b-d-galactopyranosyl)-cyclohex-1-yl]-
glycine 50 (91%), Na-[3-(1-thio-b-d-galactopyranosyl)-
cyclohept-1-yl]-glycine 51 (79%), Na-[2-(1-thio-b-d-
galactopyranosyl)-cyclohex-1-yl]-glycine 52 (85%),
Nb-[4,4-dimethyl-3-(1-thio-b-d-galactopyranosyl)-cyclo-
pent-1-yl]-b-alanine 53 (32%), Nb-[3-(1-thio-b-d-galacto-
pyranosyl)-cyclohex-1-yl]-b-alanine 54 (65%), Nb-[4,4-
dimethyl-3-(1-thio-b-d-galactopyranosyl)-cyclohex-1-yl]-
b-alanine 55 (99%), Nb-[3-(1-thio-b-d-galactopyrano-
syl)-cyclohept-1-yl]-b-alanine 56 (99%), Nb[2-(1-thio-b-
d-galactopyranosyl)-cyclohex-1-yl]-b-alanine 57 (98%),
Na-[4,4-dimethyl-3-(1-thio-b-d-galactopyranosyl)-cyclo-
pent-1-yl]-l-leucine 58 (39%), Na-[3-(1-thio-b-d-
galactopyranosyl)-cyclohex-1-yl]-l-leucine 59 (55%),
Na - [4, 4 - dimethyl - 3 - (1 - thio - b - d - galactopyranosyl)-
cyclohex-1-yl]-l-leucine 60 (31%), and Na-[2-(1-
thio-b-d-galactopyranosyl)-cyclohex-1-yl]-l-leucine 62
(60%).
Typical procedure for deblocking of alcohols; 2-hydroxy-
cyclohex-1-yl 1-thio-ꢀ-D-galactopyranoside (47). To
compound 17 (77 mg, 75 mmol) in dry methanol (5.5 mL)
and dichloromethane (0.85 mL) under argon atmo-
sphere, was added methanolic sodium methoxide (1 M,
38 mL). After 24 h, the mixture was neutralized with
Amberlite IRC-50S (H+) resin, ®ltered and con-
centrated. The residue was dissolved in water and
applied onto a column of C18 silica (Waters Prep C18,

1574
U. J. Nilsson et al./Bioorg. Med. Chem. 6 (1998) 1563±1575
Typical procedure for deblocking of N-alkylated amino
acid methyl esters; Nꢁ-[3-(1-thio-ꢀ-D-galactopyranosyl)-
cyclohept-1-yl]-L-histidine (66). To compound 36
(100 mg, 84.5 mmol) in dry methanol (6.2 mL) and
dichloromethane (0.9 mL) under argon atmosphere, was
added methanolic sodium methoxide (1 M, 43 mL). After
2 h, the mixture was neutralized with Amberlite IRC-
50S (H+) resin, ®ltered and concentrated. The residue
was dissolved in 70% methanol and applied onto a col-
umn of C18 silica (Waters Prep C18, 125 A, 7 g),
whereafter the product was eluted with 70% methanol
(50 mL) and concentrated. To the residue in water
(3.1 mL) was then added aqueous lithium hydroxide
(1 M, 0.3 mL). After 80 min, the mixture was neutralized
with Amberlite IRC-50S (H+) resin, ®ltered and con-
centrated. The residue was dissolved in dichloro-
methane:methanol (9:1) and applied onto a Waters
Sep-Pak Plus Longbody SiO2 cartridge. The cartridge
was washed with dichloromethane:methanol (9:1,
10 mL) to remove traces of the corresponding alcohol
(46), whereafter 66 was eluted with dichloromethane:
methanol:water (65:35:5, 20 mL) and concentrated. The
residue was dissolved in water and applied onto a
column of C18 silica (Waters Prep C18, 125 A, 7 g). The
C18 silica was washed with water (50 mL), whereafter
elution with 70% methanol (50 mL) and concentration
gave 66 (30 mg, 80%). In parallel reactions were
obtained Na-[4,4-dimethyl-3-(1-thio-b-d-galactopyrano-
syl)-cyclopent-1-yl]-l-histidine 63 (68%), Na-[3-(1-
thio-b-d-galactopyranosyl)-cyclohex-1-yl]-l-histidine 64
(50%), Na-[4,4-dimethyl-3-(1-thio-b-d-galactopyranosyl)-
cyclohex-1-yl]-l-histidine 65 (56%), Na-[2-(1-thio-b-d-
galactopyranosyl)-cyclohex-1-yl]-l-histidine 67 (45%),
Na-[4,4-dimethyl-3-(1-thio-b-d-galactopyranosyl)-cyclo-
pent-1-yl]-l-tryptophan 68 (72%), Na-[3-(1-thio-b-d-
galactopyranosyl)-cyclohex-1-yl]-l-tryptophan 69 (56%),
Na-[4,4-dimethyl-3-(1-thio-b-d-galactopyranosyl)-cyclo-
hex-1-yl]-l-tryptophan 70 (48%), Na-[3-(1-thio-b-d-
galactopyranosyl)-cyclohept-1-yl]-l-tryptophan 71 (93%),
and Na-[2-(1-thio-b-d-galactopyranosyl)-cyclohex-1-yl]-
l-tryptophan 72 (76%).
fraction containing a mixture of the two major diaster-
eomers (13AB, 27.1 mg) and a slower moving fraction
(13CD, 7.6 mg) containing a mixture of the two minor
diastereomers. The two major diastereomers (13AB)
were further separated using column chromatography
(SiO2, CH2Cl2:Et2O, 19:1) to give pure diastereomers
13A (13.5 mg) and 13B (5.1 mg). To 13A (13.5 mg,
13.0 mmol), 13B (5.1 mg, 4.9 mmol), and 13CD (7.6 mg,
7.3 mmol) in dry methanol (1.0, 0.4, and 0.6 mL respec-
tively) and dichloromethane (130, 49, and 73 mL respec-
tively) under argon atmosphere, was added methanolic
sodium methoxide (7.1, 2.7, and 4.0 mL respectively,
1 M). After 195 min, the mixtures were neutralized with
AcOH and concentrated. The residues were dissolved in
70% methanol and applied onto Sep-Pak C18 car-
tridges, whereafter the products were eluted with 70%
methanol (10 mL) and concentrated. The residues were
dissolved in water and applied onto Sep-Pak C18 car-
tridges. The cartridges were washed with water (50 mL),
whereafter elution with methanol (50 mL) gave the two
major diastereomers 43A (2.4 mg, 60% from 13A) and
43B (0.7 mg, 31% from 13B), as well as a mixture of the
two minor diastereomers 43CD (1.2 mg, 53% from
13CD). 1H NMR data for 43A (500 MHz, CD3OD,
aglycon signals are indicated with *) d 4.30 (d, 1H, J
9.5 Hz, H-1), 4.20 (dddd, 1H, J=3.5, 6.3, 7.3, and
8.4 Hz, H-4*), 3.87 (dd, 1H, J=1.0 and 3.3 Hz, H-4),
3.72 (dd, 1H, J=6.7 and 11.3 Hz, H-6), 3.68 (dd, 1H,
J=5.5 and 11.3 Hz, H-6), 3.495 (t, 1H, J=9.3 Hz, H-2),
3.485 (ddd, 1H, J=1.0, 5.5, and 6.7 Hz, H-5), 3.42 (dd,
1H, J=3.3 and 9.3 Hz, H-3), 2.94 (dd, 1H, J=7.4 and
11.8 Hz, H-1*), 2.59 (ddd, 1H, J=7.3, 7.4, and 13.5 Hz,
H-5a*), 1.81 (dd, 1H, J=8.4 and 13.7 Hz, H-3a*), 1.79
(ddd, 1H, J=6.3, 11.8, and 13.5 Hz, H-5b*), 1.55 (dd,
1H, J=3.5 and 13.7 Hz, H-3b*), 1.09 and 1.07 (2 s, 3H
each, Me*). 1H NMR data for 43B (500 MHz, CD3OD)
d 4.29 (d, 1H, J=9.5 Hz, H-1), 4.20 (dddd, 1H, J=3.5,
6.3, 7.3, and 8.2 Hz, H-4*), 3.87 (dd, 1H, J=1.0 and
3.3 Hz, H-4), 3.71 (dd, 1H, J=6.7 and 11.3 Hz, H-6),
3.68 (dd, 1H, J=5.5 and 11.3 Hz, H-6), 3.50 (t, 1H,
J=9.3 Hz, H-2), 3.48 (ddd, 1H, J=1.0, 5.5, and 6.7 Hz,
H-5), 3.43 (dd, 1H, J=3.3 and 9.3 Hz, H-3), 3.02 (dd,
1H, J=7.4 and 11.8 Hz, H-1*), 2.61 (ddd, 1H, J=7.3,
7.4, and 13.3 Hz, H-5a*), 1.83 (dd, 1H, J=8.2 and
13.6 Hz, H-3a*), 1.78 (ddd, 1H, J=6.3, 11.8, and
13.3 Hz, H-5b*), 1.57 (dd, 1H, J=3.5 and 13.6 Hz, H-
3b*), 1.070 and 1.065 (2 s, 3H each, Me*). 1H NMR
data for 43CD (360 MHz, CD3OD) d 4.35 and 4.31 (2 d,
1H each, J=9.4 Hz, H-1), 4.29 and 4.26 (2 m, 1H each,
H-4*), 3.90 and 3.88 (dd, 1H, J=1.1 and 3.3 Hz, H-4),
3.74 and 3.73 (2 dd, 1H each, J=6.2 and 11.1 Hz, H-6),
3.70 (dd, 2H, J=5.9 and 11.1 Hz, H-6), 3.52 and 3.50 (2
t, 1H each, J=9.4 Hz, H-2), 3.51 and 3.49 (2 br t, 1H
each, H-5), 3.45 and 3.44 (2 dd, 1H each, J=3.3 and
9.3 Hz, H-3), 3.34 and 3.25 (2 dd, 1H each, J=8.3 and
11.5 Hz, H-1*), 2.20 to 2.06 (m, 4H, H-5a* and H-5b*),
Screening for inhibitors of ꢀ-galactosidase from E. coli.
To solutions of compounds 43±73 (3 mM) and p-nitro-
phenyl b-d-galactoside (2 mM, Km=0.04 mM) in
microtiter plate wells (50 mL/well) was added b-galacto-
sidase (5 u/mL, 50 mL/well). Aqueous sodium hydroxide
(1 M, 200 mL/well) was added after 8 min and the optical
density at 405 nm was used as an approximate value
related to the initial rates of hydrolysis. Each experi-
ment was done in triplicate.
Separation of diastereomers of 4-hydroxy-2,2-dimethyl-
cyclopent-1-yl 1-thio-ꢀ-D-galactopyranoside (43). Col-
umn chromatography (SiO2, toluene:EtOAc, 19:1) of
the diastereomeric mixture of 13 gave a faster moving

U. J. Nilsson et al./Bioorg. Med. Chem. 6 (1998) 1563±1575
1575
2.01 and 2.00 (2 dd, 1H each, J=7.0 and 13.7 Hz, H-
3a*), 1.48 and 1.46 (2 dd, 1H, J=3.5 and 13.6 Hz, H-
3b*), 1.17 and 1.15 (2 s, 3H each, Me*), 0.90 (s, 6H, 2
Me*). m/z calcd for C13H25O6S (M+H+), 309.1372,
found 309.1367.
7. Bundle, D. R.; Young, N. M. Curr. Opin. Struct. Biol. 1992,
2, 666.
8. Bundle, D. R. In Glycosciences: Status and Perspectives;
Gabius, H.-J.; Gabius S., Eds.; Chapman & Hall: Weinheim,
1997; pp 311±331.
9. Thompson, L. A.; Ellman, J. A. Chem. Rev. 1996, 96, 555.
Inhibition constants for 43A, 43B, and 73. Initial rates of
hydrolysis were measured at di€erent concentrations of
p-nitrophenyl b-d-galactoside (0.125±4 mM) and at the
same time varying the inhibitor concentration (0.0016±
0.1 mM) for each substrate concentration. The data
were ®tted to the Michaelis±Menten equation (according
to the Levenberg±Marquardt algorithm) using the program
Kaleidagraph (Abelbeck Software Inc.), which gave the
apparent Km values. Secondary plots of the apparent Km
values against [I] provided the Ki values (Table 1).
10. An, H.; Cummins, L. L.; Gri€ey, R. H.; Bharadwaj, R.;
Haly, B. D.; Fraser, A. S.; Wilson-Lingardo, L.; Risen, L. M.;
Wyatt, J. R.; Cook, P. D. J. Am. Chem. Soc. 1997, 119, 3696.
11. Sim, M. M.; Ganesan, A. J. Org. Chem 1997, 62, 3230.
12. Han, H.; Wolfe, M. M.; Brenner, S.; Janda, K. D. Proc.
Natl. Acad. Sci. U.S.A. 1995, 92, 6419.
13. Cheng, S.; Comer, D. D.; Williams, J. P.; Myers, P. L.;
Boger, D. L. J. Am. Chem. Soc. 1996, 118, 2567.
14. Brown, S. D.; Armstrong, R. W. J. Am. Chem. Soc. 1996,
118, 6331.
15. Flynn, D. L.; Crich, J. Z.; Devraj, R. V.; Hockerman, S. L.;
Parlow, J. J.; South, M. S.; Woodard, S. J. Am. Chem. Soc.
1997, 119, 4874.
Acknowledgements
16. Booth, R. J.; Hodges, J. C. J. Am. Chem. Soc. 1997, 119,
4882.
The authors Dr Albin Otter for recording the 500 MHz
1H NMR spectra. This work was supported by the
Natural Sciences and Engineering Research Council of
Canada (NSERC). U. J. N. was supported by a post-
doctoral fellowship from the Swedish Natural Research
Council (NFR) 1995±1997.
17. Kaldor, S. W.; Siegel, M. G.; Fritz, J. E.; Dressman, B. A.;
Hahn, P. J. Tetrahedron Lett. 1996, 37, 7193.
18. Studer, A.; Hadida, S.; Ferrito, R.; Kim, S.-Y.; Jeger, P.;
Wipf, P.; Curran, D. Science 1997, 275, 823.
19. Studer, A.; Curran, D. P. Tetrahedron 1997, 53, 6681.
20. Curran, D. P.; Hadida, S.; He, M. J. Org. Chem 1997, 62,
6714.
References and Notes
21. Palcic, M. M.; Heerze, L. D.; Pierce, M.; Hindsgaul, O.
Glycoconjugate J. 1988, 5, 49.
1. See for example; Lemieux, R. U. Chem. Soc. Rev. 1989, 18,
347. Kihlberg, J.; Hultgren, S. J.; Normark, S.; Magnusson, G.
J. Am. Chem. Soc. 1989, 111, 6364. Glaudemans, C. P. J.
Chem. Rev. 1991, 91, 25. Sierks, M. R.; Bock, K.; Refn, S.;
Svensson, B. Biochemistry 1992, 31, 8972. Kanie, O.; Crawley,
S. C.; Palcic, M. M.; Hindsgaul, O. Bioorg. Med. Chem. 1994,
2, 1231. Reck, F.; Meinjohanns, E.; Springer, M.; Wilkens, R.;
van Dorst, J. A. L. M.; Paulsen, H.; Moller, G.; Brockhausen,
I.; Schachter, H. Glycoconj. J. 1994, 11, 210.
22. Hasegawa, A.; Nakamura, J.; Kiso, M. J. Carbohydr.
Chem. 1986, 5, 11.
23. Kuwajima, I.; Murofushi, T.; Nakamura, E. Synthesis
1976, 602.
24. Bennett, S.; von Itzstein, M.; Kiefel, M. J. Carbohydr. Res.
1994, 259, 293.
25. Borch, R. F.; Bernstein, M. D.; Durst, H. D. J. Am. Chem.
Soc. 1971, 93, 2897.
2. von Itzstein, M.; Wu, W.-Y.; Kok, G. B.; Pegg, M. S.;
Dyason, J. C.; Jin, B.; Phan, T. V.; Smythe, M.; White, H. F.;
Oliver, S. W.; Colman, P. M.; Varghese, J. N.; Ryan, D. M.;
Woods, J. M.; Bethell, R. C.; Hotham, V. J.; Cameron, J. M.;
Penn, C. R. Nature 1993, 363, 418.
26. Coy, D. H.; Hocart, S. J.; Sasaki, Y. Tetrahedron 1988, 44,
835.
27. Tourwe, D.; Piron, J.; Defreyn, P.; Van Binst, G. Tetra-
hedron Lett. 1993, 34, 5499.
3. Kanie, O.; Barresi, F.; Ding, Y.; Labbe, J.; Otter, A.; Fors-
berg, L. S.; Ernst, B.; Hindsgaul, O. Angew. Chem. Int. Ed.
1995, 34, 2720.
28. Kim, S.; Oh, C. H.; Ko, J. S.; Ahn, K. H.; Kim, Y. J. J.
Org. Chem. 1985, 50, 1927.
29. Mattson, R. J.; Pham, K. M.; Leuck, D. J.; Cowen, K. A.
J. Org. Chem. 1990, 55, 1927.
4. Boons, G.-J.; Heskamp, B.; Hout, F. Angew. Chem. Int. Ed.
1996, 35, 2845.
30. Abdel-Magid, A. F.; Carson, K. G.; Harris, B. D.; Mar-
yano€, C. A.; Shah, R. D. J. Org. Chem. 1996, 61, 3849.
5. Liang, R.; Yan, L.; Loebach, J.; Ge, M.; Uozumi, Y.;
Sekanina, K.; Horan, N.; Gildersleeve, J.; Thompson, C.;
Smith, A.; Biswas, K.; Still, W. C.; Kahne, D. Science 1996,
274, 1520.
31. Ley, S. V.; Mynett, D. M.; Koot, W.-J. Synlett 1995, 1017.
32. De Bruyne, C. K.; Yde, M. Carbohydr. Res. 1977, 56, 153.
33. Helferich, B.; Turk, D. Chem. Ber. 1956, 89, 2215.
6. Quiocho, F. Pure Appl. Chem. 1989, 61, 1293.