Five-membered iminocyclitol α-glucosidase inhibitors: Synthetic, biological screening and in silico studies

Article abstract of DOI:10.1016/j.bmc.2013.01.030

The design and synthesis of a small library of pyrrolidine iminocyclitol inhibitors with a structural similarity to 1,4-dideoxy-1,4-imino-d-arabitol (DAB-1) is reported. This library was specifically designed to gain a better insight into the mechanism of inhibition of glycosidases by polyhydroxylated pyrrolidines or iminocyclitols. Pyrrolidine-3,4-diol 15a and pyrrolidine-3,4-diol diacetate 15b had emerged as the most potent α-glucosidase inhibitors in the series. Docking studies performed with an homology model of α-glucosidase disclosed binding poses for compounds 15a, 15b, 16a, and 16a′ occupying the same region as the NH group of the terminal ring of acarbose and suggest a closer and stronger binding of compound 15a and 15b with the enzyme active site residues. Our studies indicate that 2 or 5-hydroxyl substituents appear to be vital for high inhibitory activity.

Full text of DOI:10.1016/j.bmc.2013.01.030

Bioorganic & Medicinal Chemistry 21 (2013) 1911–1917
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Five-membered iminocyclitol
a-glucosidase inhibitors: Synthetic,
biological screening and in silico studies
Luis R. Guerreiro a, Elisabete P. Carreiro a, Luis Fernandes a, Teresa A. F. Cardote b, Rui Moreira b,
Ana T. Caldeira a, Rita C. Guedes b, A. J. Burke a,
⇑
a Departamento de Química and Centro de Química de Évora, Universidade de Évora, Rua Romão Ramalho, 59, 7000 Évora, Portugal
b Research Institute for Medicines and Pharmaceutical Sciences-iMed.UL, Faculty of Pharmacy, University of Lisbon, Av. Prof. Gama Pinto, 1649-003 Lisbon, Portugal
a r t i c l e i n f o
a b s t r a c t
Article history:
The design and synthesis of a small library of pyrrolidine iminocyclitol inhibitors with a structural sim-
Received 16 November 2012
Revised 10 January 2013
Accepted 14 January 2013
Available online 26 January 2013
ilarity to 1,4-dideoxy-1,4-imino-D-arabitol (DAB-1) is reported. This library was specifically designed to
gain a better insight into the mechanism of inhibition of glycosidases by polyhydroxylated pyrrolidines
or iminocyclitols. Pyrrolidine-3,4-diol 15a and pyrrolidine-3,4-diol diacetate 15b had emerged as the
most potent
a-glucosidase inhibitors in the series. Docking studies performed with an homology model
of
a
-glucosidase disclosed binding poses for compounds 15a, 15b, 16a, and 16a0 occupying the same
Keywords:
Iminocyclitol
Small molecule inhibitor
region as the NH group of the terminal ring of acarbose and suggest a closer and stronger binding of com-
pound 15a and 15b with the enzyme active site residues. Our studies indicate that 2 or 5-hydroxyl sub-
stituents appear to be vital for high inhibitory activity.
a-Glucosidase
Ó 2013 Elsevier Ltd. All rights reserved.
Enantiopure compound
(3,4)-Dihydroxypyrrolidine
1. Introduction
cult.6 For this reason many types of five-membered iminocyclitols
have been synthesized and screened. An examination of the litera-
The iminocyclitols—polyhydroxylated pyrrolidines and piperi-
dines—are a family of important pharmacologically active com-
pounds that are both potent glycosidase and glycosyltransferase
inhibitors due to their mimicry of the transition state of the enzy-
matic reaction, including serendipitous electrostatic binding inter-
actions.1,2 For this reason, they have been selected as therapeutic
agents in several areas such as cancer, viral infections (particularly
influenza) and diabetes, etc. For example, deoxinojirimicin (DNJ) 1
ture showed that a number of diverse structural types have been
screened for -glucosidase inhibition. This includes those devel-
a
oped by Davis, for example, 46 (Fig. 1) obtained from 3-pyrroline
and tartaric acid, and showed no significant inhibition (3–13% at
the 1 mM level). The five-membered iminocyclitol amide deriva-
tives 5 developed by Wong’s group showed inhibition with a Ki va-
lue of 53 nM7 (which was better than that exhibited by the parent
structure 6) and the 2-alkylated analogues of type 7 (Fig. 1) were
is an inhibitor of endoplasmic reticulum
1,4-dideoxy-1,4-imino- -arabinitol (DAB-1) 2 and 2,5-dideoxy-
2,5-imino-
-mannitol (DMDP) 35 are powerful inhibitors of
-glucosidases4 (Fig. 1). The synthesis of more potent novel
a
-glucosidases I,3 and
also shown by Wong1b to be weak
(3–54% inhibition at 200 M). Davis has also shown that a library
of N-acyl(aroyl)-2-carboxyamide substituted pyrrolidine imino-
cyclitols of type 8, showed little or no inhibition at 100
M.8 Wong
a-glucosidase inhibitors
D
l
D
a
l
analogues of these compounds is an important goal in medicinal
chemistry, not only for targeting human disease, but also as tools
to probe the mechanism of glucosidase function. In most cases,
these molecules function as pure enantiomers.
It has been previously pointed out that structure–activity rela-
tionships for iminocyclitol glycosidase inhibitors are difficult to
elucidate, making rational inhibitor design a difficult task.1 It is
also known that five-membered iminocyclitols can give rise to
higher inhibition than their six-membered counterparts and subtle
selectivities may be observed for five- over six-membered systems,
thus making logical design based upon structural analogy diffi-
has prepared and tested pyrrolidine iminocyclitols of type 9 with
side-chains in both the 2 and the 5 positions,9 in which some were
very potent inhibitors. Calveras et al.10 showed that a library of
iminocyclitols of type 10 inhibited this enzyme at 1.6–4.2 nM level.
In most cases the presence of a hydroxymethyl appendage seems
to give significant inhibition, whereas when substituted by an alkyl
group, as in 4 and 7 (Fig. 1), the inhibition is weaker or even absent,
indicating that this substituent must be relevant to approximate
the structures of the natural
a
-glucosidase sugar substrate.6 This
point has been echoed by Wong9 and indeed, this hypothesis has
been supported somewhat by the work of both Bols11 and Lundt,12
who prepared the 2,5-non-substituted pyrrolidine iminocyclitols
11, 12 and 13, with the hydroxymethyl appendage transposed to
⇑
Corresponding author. Tel.: +351 266745310
0968-0896/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmc.2013.01.030

1912
L. R. Guerreiro et al. / Bioorg. Med. Chem. 21 (2013) 1911–1917
OH
H
N
HO
OH
HO
OH
HO
OH
OH
OH
N
H
R
N
H
HO
OH
N
H
HO
OH
2
3
1
R = Et, Ph
Deoxynojirimicin (DNJ)
DAB-1
4
DMDP
HO
O
OH
HO
OH
CHX
HO
OH
HO
OH
OH
OH
OH
OH
R
N
N
H
N
H
N
H
COR1
HN
R
H2N
HN
X = H, CH2OH, CH2NHAc
ROC
6
7
5
8
OH
OH
OH
OH
HO
OH
HO
OH
R2
HO
HO
HO
HO
HO
OH
R1
N
N
N
H
N
N
H
H
H
H
OH
9
10
12
11
13
Figure 1. Some natural and iminocyclitol inhibitors.
the 3-position and showing weak and unspecific inhibition (at
3.8 mM for 11 and 2 mM in the case of 12)11 and a
-glucosidase
AcO
OAc
HO
OH
HO
N
OH AcO
OAc
a
inhibition in the case of 13 (at 0.8 mM).12 It must be noted that
the 3,4-hydroxy groups in these iminocyclitols had a cis relative
configuration. However, on the other hand, in the case of the study
conducted by Calveras et al.10 the results seemed to imply that the
hydroxymethyl group could have a deleterious effect on the activ-
ity of the enzyme.
N
N
H
N
H
N
H2 OCOCF3
2 OCOCF3
Bn
Bn
20
16b-(3S,4S)
16b'-(3R,4R)
16a-(3S,4S)
15b-(S,S)
15b'-(R,R)
15a-(3S,4S)
15a'-(3R,4R)
16a'-(3 ,4
)
R
R
Figure 3. Library of five-membered iminocyclitols derived from L- or D-tartaric
acid.
We were curious to know how important these structural/func-
tional appendages were for
a-glucosidase inhibition and thus set
about preparing a very small library of simple pyrrolidine imino-
cyclitols lacking functionality in the 2-position and with or without
a hydrophobic group on the pyrrolidine nitrogen. The diol unit had
the trans-configuration. Asano’s group also studied this question
several years ago, and speculated that glycoside inhibitors have
structures which resemble those of the respective cations.13 In fact,
the binding mode of the five-membered inhibitors has yet eluded
characterization. Work with the endoglycoceramidase model sys-
tems by Withers’ group2 using a variety of five- and six-membered
iminocyclitol inhibitors has shown that imitation of the substrate’s
2-hydroxy group by the inhibitor’s 6-hydroxy group and by elec-
trostatic interactions between the ring nitrogen and the glutamic
In order to probe the inhibition mechanism of such inhibitors
-glucosidases, we decided to use a small library of enantiome-
rically pure five-membered iminocyclitol inhibitors of the enantio-
meric series derived from - and -tartaric acids (Fig. 3, all with the
trans-relative stereochemistry), lacking substituents in either the 2
or 5-positions and some containing acetate groups to increase the
hydrophobicity of the inhibitor and to protect the hydroxyl groups.
Molecular modeling studies involving a docking protocol were
undertaken on a completely new and validated homology model
on
a
L
D
of
a-glucosidase with the purpose to clarify these compounds
inhibitory potency with yeast maltase. An additional goal was to
test some simple, easily prepared pyrrolidine diols aswell.
acid residues (which are also present in
important.
a
-glucosidases14) are
2. Results and discussion
2.1. Chemistry
For testing purposes, Davis and co-workers synthesized a small
library of anisomycin analogues 14 (Fig. 2). These were designed
on the basis that a hydrophobic group at/or near the putative agly-
cone binding site might augment enzyme inhibition.6 In fact, the
application of anisomycin and analogues as inhibitors against gly-
cosidases has been very poorly studied. These compounds indeed
The (3R,4R)-pyrrolidine-3,4-diol 15a (Scheme 1) was synthe-
tized using the method of Nagel.16 This molecule has previously
been used as an intermediate in the synthesis of certain asparagine
modified pyrrolidine iminocyclitols used as Multiple Sclerosis
Antigenic probes.15 Incidentally 16a has been synthesized previ-
ously in our group, by an innovative, but less direct route which
uses the commercial dioxane diester 17 (Scheme 1) derived from
the tartaric acid as precursor.17 16b was obtained in a yield of
92% from 16a by stirring in the presence of pyridine and acetic
anhydride (Scheme 1). The diacetate derivative 15b (as the trifluo-
roacetate salt) was obtained despite many efforts according to the
procedure shown in Scheme 2. Many unsuccessful debenzylations
were carried out, via hydrogenation of the N-benzyl diacetate
derivative 16b with palladium on carbon, or with iron chloride.18
displayed competitive inhibition of a
-glucosidases.6
MeO
HO
OH
MeO
AcO
OH
N
H
N
H
14
Anisomycin
Figure 2. Anisomycin and anisomycin analogues tested by Davis.6

L. R. Guerreiro et al. / Bioorg. Med. Chem. 21 (2013) 1911–1917
1913
An alternative, more indirect synthetic route was used that in-
volved the transformation of both enantiomers 15a and 15a0 to
the N-Boc protected intermediates (structures not shown) followed
by successful acetylation to the N-Boc protected diacetates 19a,
and 19a0 (Scheme 2). The deprotection of 19a was investigated
using several methods, but it was successfully transformed to the
corresponding (3S,4S)-3,4-diacetoxypyrrolidinium trifluoroacetate
15b in 98% yield using TFA in dichloromethane. The other methods
which failed to function, included, (a) HCl 4 M in THF,19 (b) TBAF in
THF,20 (c) Na2CO3/DME/H2O21 and (d) Amberlyst 15/NH4OH.22 In
the case of (a), both deprotection of the N-Boc and acetyl groups
occurred to give 15a. In the case of (b) and (c) the acetyl groups
were removed. In the case of (d), compound 15a was obtained
due to hydrolysis of the acetyl groups by the aqueous basic condi-
tions employed. Compound 20 was synthetized using 1 equiv of
TFA and pyrrolidine in dichloromethane. This compound was used
as a reference in the enzymatic bioassays.
between the o-acetylated and non-acetylated derivatives (compare
15a with 15b in Table 1) was virtually nil, this close similarity we
believe might have been a consequence of the hydrolysis of the
acetate groups during the assays. Based on these observations,
and from the analogous work of Withers and Davis24,25 on the
inhibition of the glucosidase hydrolase endocellulase Cel5A from
Bacillus agaradhaerens using isofagomine as the probe, and in
the knowledge that the catalytic residues of the active site of the
a-glucosidase from S. cerevisiae consists of three amino acids,
two aspartic acids (Asp214 and Asp349) and one glutamic acid
(Glu276) which acts as a proton donor,23 including key Tyr71,
His111 and His348 units in the substrate binding domain of the
enzyme,23 we present the following putative preliminary model
(Fig. 5), which on the basis of the weakly acidic conditions of the
assay (pH 6.7) and literature precedent should be protonated on
the pyrrolidine nitrogen.
Continuing with the same argument as given above, observing
that the difference in the inhibitory activities for both 16a and
16a0 was less than twofold, it is hard to draw any strong conclu-
sions as to the stereochemical requirements for inhibition.
In order to gain a better insight into the process of inhibition at
the molecular level we conducted some molecular modeling stud-
ies or more precisely, docking studies.
The same procedures were used to obtain compounds 15b0, 16a0
and 16b0 (the enantiomeric series) derived from
D-tartaric acid. The
NMR and other physical data for these compounds were in agree-
ment with that obtained from the literature.15
2.2. Biological screening studies
These compounds were then assayed for the inhibition of the
2.3. Molecular modeling studies
a
-glucosidase from Saccharomyces cerevisiae using inhibitor
concentrations of between 0.19 ꢀ 10ꢁ3 and 46 mM. This
a
-
In order to rationalize the described biological results, docking
studies of compounds 15a, 15b, 16a-(3S,4S), and 16a0-(3R,4R) and
16b were undertaken. Molecular docking simulations were per-
formed, using GOLD software version 5.1.0,26 to predict the inter-
actions and binding modes of the synthesized compounds on the
glucosidase contains 584 amino acid residues (vide infra). In
our bioassays, eight compounds (15a, 15b, 15b0, 16a, 16a0, 16b,
16b0 and 20) were tested, but only four (15a, 15b, 16a, 16a0)
showed an inhibition profile in the concentration range used,
exhibiting over 50% inhibition down to a concentration of
P10.9 mM (Fig. 4). The debenzylated compounds 15a and 15b
were the best inhibitors, with IC50’s of 10.9 and 12.5 mM, respec-
tively (Table 1). In the case of compounds 15b and 15b0, which
are enantiomers, only 15b showed an inhibition profile in the
concentration range used, exhibiting >50% inhibition at a concen-
tration of 25 mM (80% of inhibition) and at this concentration
15b0 showed only 40% inhibition. By using 20 as a control, it
was established that the ꢁOCOCF3 counter-ion had no effect on
a-glucosidase active site and to evaluate their relative binding
affinities. To better understand the important enzyme–ligand
interactions, poses inside the binding pocket were performed. As
a means of testing the adopted scoring function, some known
inhibitors like acarbose (the first a-glucosidase inhibitor approved
for type 2 diabetes treatment), DAB-1, and DMDP were also docked
within the MAL12 model active site.
Although the X-ray crystal structures of some bacterial
sidases have been reported, structural information is still unavail-
able for the eukaryotic -glucosidase enzyme from Baker’s yeast
a-gluco-
the
a-glucosidase activity, no inhibitory activity was observed
a
at over 50% of the inhibition concentration. Compounds 16a
and 16a0, inhibited the enzyme over the 50% of the inhibition
concentration, and had IC50’s of 38 and 25 mM, respectively
(Table 1). Compound 16a0 had the best inhibition profile. In the
case of 16b and 16b0 no inhibitory activity was observed at the
concentrations studied.
(the enzyme used in our biological assays). However, only a few
homology models have been previously developed for this enzyme.
Based on the fact that the catalytic residues (Asp214, Glu276, and
Asp349) and other important residues for substrate binding are
highly conserved in the GH13 family, Park et al.28 developed a
MAL12 model based on oligo-1,6-glucosidase enzyme from Bacillus
cereus (PDB code 1UOK), however this enzyme only shares a very
small sequence identity (ꢂ38%). Also Ferreira and co-workers,23
In summary, compound 15a showed the greatest inhibitory
activity, whilst 16a was the least active. We also screened the
pseudotetrasaccharide
a-glucosidase inhibitor, acarbose, which is
very recently developed another Mal12 homology model from
29
a known competitive inhibitor, and used as a cheap benchmark
for the level of inhibition exhibited by our iminocyclitols. An IC50
value of 0.641 mM was observed for acarbose, much higher than
that of our best inhibitor 15a. Acarbose was shown to be active
against this enzyme by Ferreira and co-workers,23 showing an
IC50 value of 0.109 mM, but this was obtained under different
experimental conditions. Comparing the compounds 16a and
16a0 with 16b and 16b0 it would appear that the presence of free
hydroxyl groups in the 3 and 4-positions is important in the inhi-
bition process, possibly due to favorable secondary hydrogen-
bonding interactions.
Given that the inhibition values for the unsubstituted pyrrol-
idines were only 2–3 times stronger than for the benzylated ana-
logues (Table 1), it is hard at this juncture to clearly conclude
that the presence of a free amine group is necessary for good
inhibition of this enzyme. The difference in inhibitory activities
Thermotoga maritima 4-a-glucanotransferase (PDB code 1LWJ).
None of these models have their 3D coordinates publicly released.
Therefore we built by homology modeling a completely new -glu-
a
cosidase MAL12 model based on the crystal structure of isomaltase
from S. cerevisiae (3AJ7.pdb)27 recently released, isolated and in
complex with its competitive inhibitor maltose. This enzyme
shares 72.4% sequence identity with
a-glucosidase enzyme from
Baker’s yeast (Fig. 2 in Supplementary data). The
a
-glucosidase
MAL12 model developed in our study consists of a single polypep-
tide chain constituted of 584 residues (Fig. 6). A more complete
description of the developed homology model can be found in
the Experimental section.
After completing the MAL12 model, the
a-glucosidase–ligand
complexes were generated by docking our compounds into the en-
zyme active site. All the docked compounds were previously en-
ergy-minimized and subjected to 1000 docking runs (see more

1914
L. R. Guerreiro et al. / Bioorg. Med. Chem. 21 (2013) 1911–1917
OMe
details in the Experimental section). The 10 top solutions (the ones
with the highest Goldscore) were visually and critically analyzed.
Interestingly, we found that all the tested compounds occupy
almost the same region of the binding pocket as the amino group
and its corresponding cyclohexenyl ring of acarbose. This proves
that our five-membered iminocyclitol derivatives are molecules
that can be favorably positioned in the binding cavity, having the
capacity for forming important contacts with the enzyme active
site residues.
Compound 15a, the most potent compound in the series, forms
an H-bond between the pyrrolidine nitrogen and Asp214 (1.6 Å),
and also strongly interacts with Glu276 (catalytic site). The free hy-
droxyl groups interact with Arg439 and Asp68 and an additional
interaction between the iminocyclitol and Tyr71 is also estab-
lished. A very similar pose is observed for compound 15b, however,
this compound is closer to the catalytic binding site than the for-
mer due to strong interactions of the acetate groups with Arg439
and Asp68 that pushes this compound deep inside the binding site.
This compound makes H-bonds with Asp214 (1.5 Å), Asp349
(1.9 Å), and Glu276 (1.6 Å).
MeO2C
MeO2C
O
O
OMe
17
ref. 17
AcO
(S)
OAc
(S)
HO
OH
HO2C(R) OH
ref. 16a
(i)
(S)
(S)
HO (R)CO2H
N
N
Bn
18-(L)
Bn
16b-(3S,4S)
16a-(3S,4S)
ref. 16a
HO
OH
(S)
(S)
N
H
15a-(3S,4S)
Compound 16a interacts with the same residues of the catalytic
site, however with slightly longer interactions (ꢂ2 Å), including
those in the binding region—Arg439 and Asp68. Compound 16a0’s
position is almost the same as 16a the only difference being, the
iminocyclitol ring is slightly tilted, but it establishes more interac-
tions with the residues in the binding pocket, and thus would be
expected to have a better biological activity profile than 16a, which
was observed from the biological assays. Furthermore, compound
AcO
OAc
HO
OH
HO2C(S) OH
As above
(R)
(R)
As above
(R)
(R)
HO (S)CO2H
N
N
Bn
Bn
18'-(D)
16a'-(3R,4R)
16b'-(3R,4R)
As above
OH
16b, inactive against a-glucosidases, has a completely different po-
HO
sition inside the pocket, in this case a single interaction is estab-
lished with Glu276, with one of the acetate groups pointing to
the inside of the binding pocket and rotating the conformation,
contrarily to the other docked compounds where the iminicyclitol
ring occupies this position. This completely different pose of com-
pound 16b might explain the absence of activity of this compound.
The predicted poses for compounds 15a, 15b, 16a, 16a0 (with
inhibitory activity), and 16b (inactive) and their interaction with
the MAL12 model active site are shown in Figure 7.
(R)
(R)
N
H
15a'-(3R,4R)
Scheme 1. Reagents and conditions: (i) Ac2O, pyridine, rt, 8 h.
AcO
OAc
AcO
OAc
HO
OH
(iii)
(i), (ii)
As described above it can be seen that for all the bioactive com-
pounds the iminocyclitol moiety is positioned almost at the same
location, with the pyrrolidine nitrogen directed to the catalytic site
(Asp214, Glu276, Asp349) and leaving the hydroxyl and acetate
groups free to interact with the Asp68 and Arg439 residues. In fact,
we see that the most active compounds of the series (15a and 15b)
are at H-bond or short distances from the catalytic residues, obvi-
ously due to the fact that these two compounds have their pyrrol-
idine nitrogen unsubstituted thus permitting a closer interaction
with the catalytic residues. In the case of compounds 16a and
16a0 these interactions exist, however with longer distances. What
might explain the reduced inhibitory activity of these compounds?
One possible explanation for the activity of these four compounds
N
N
N
H
H2 O2CCF3
O
O
19a-(S,S)
19a'-(R,R)
15a-(S,S)
15a'-(R,R)
15b-(S,S)
15b'-(R,R)
Scheme 2. Reagents and conditions: (i) BOC2O, EtOH, 0 °C, (ii) pyridine, Ac2O rt, 8 h,
(iii) TFA, DCM, rt, 1.5 h.
15b, despite the fact that our docking studies (without water) seem
to predict that the latter would be more active.
3. Experimental Section
3.1. General chemical
against
a-glucosidases is certainly the establishment of strong
(considering the short distances) interactions like, presumably,
hydrogen bonds; also, the molecules seem to fit perfectly in the ac-
tive site pocket. Also in agreement with bioassays our calculations
All reagents were obtained from Aldrich, Fluka, Alfa Aesar or Ac-
ros. Solvents were dried using common laboratory methods.
(3S,4S)-Pyrrolidine-3,4-diol 15a and (3R,4R)-Pyrrolidine-3,4-diol
15a0 were synthesized according to the method of Nagel.16 The
compounds 16a and 16b (and their respective enantiomers) were
synthesized using the precursors (3R,4R)-N-benzyl-3,4-dihy-
droxy-2,5-dioxopyrrolidine and (3S,4S)-N-benzyl-3,4-dihydroxy-
2,5-dioxopyrrolidine, respectively, both enantiomers had an
enantiomeric purity of 99%ee as determined by HPLC. TLC was
carried out on aluminium backed Kiselgel 60 F254 plates (Merck).
Plates were visualised either by UV light or with phosphomolybdic
suggest that compound 16b (inactive against
adopts a completely different orientation within the binding pock-
et when compared to the active compounds.
a-glucosidase)
We consider that one limiting factor in this docking procedure
was the absence of water molecules in the docking calculations.
As observed in the crystallographic structure (3AJ7.PDB) this pock-
et accommodates several water molecules that usually help in the
establishment of H-bond interactions with the free hydroxyl
groups what could push the inhibitors closer to the Glu276, thus
improving the interactions within the active site. This might be
one reason to explain why 15a has a slightly better activity than

L. R. Guerreiro et al. / Bioorg. Med. Chem. 21 (2013) 1911–1917
1915
O
Glu
Possible secondary
H
N
interactions
with key amino acid
units
O
HO
HO
H
Figure 5. Schematic model to explain the possible positive secondary interactions
between the -glucosidase and inhibitor 15a.
a
ment methodology originally introduced by Needleman and
Wunsch.30 All the default settings in the MOE-Align panel were
used for the sequence alignment. Our calculations identified the
Saccharomyces cerevisiae isomaltase crystallographic structure
(PDB code 3AJ7, 1.30 Å resolution)27 with 72.4% of sequence iden-
tity with the target as the most suitable template. Yamamoto
et al.27 resolved the structures of isomaltase of Saccharomyces cere-
visiae (PDB code 3AJ7, 1.30 Å) and in a complex with maltose (PDB
code 3A4A, 1.60 Å). The catalytic site is highly conserved in both
structures (as can been seen in Fig. 1 of the Supplementary data).
The 3D homology models were built with MODELLER31 software
using only a single template. A set of 10 intermediate models were
generated and refined with AMBER99 (R-field) forcefield32 and the
best of them was selected and evaluated to investigate the stereo-
chemical fitness of our model with MOE software. The stereochem-
ical quality of the enzyme backbone and side chains was validated
by Ramachandran plots. To validate our model we first docked
maltose, into the structure’s active site and compared the final
docked complex with the crystallographic structure obtained by
Yamamoto et al.27 We confirmed that in the MAL12 model, maltose
was placed in the same position as that adopted in the crystallo-
graphic structure (PDB ID: 3A4A.pdb, 1.60 Å) interacting with the
active site residues in an analogous manner.
Figure 4. Inhibitory effects of compounds 15a, 15b, 16a, 16a0and acarbose against
-glucosidase activity.
a
acid in ethanol. The 1H NMR and 13C NMR spectra were recorded
on a Bruker Avance instrument (1H: 400 MHz and 13C: 100 MHz)
using CDCl3 as solvent and the signal from residual CHCl3 as an
internal standard. Mass spectra were recorded using the electro-
spray ionization (ESI) technique on a Bruker Daltonics Apex-Qe
instrument (CACTI). Specific rotations were measured on a
Perkin–Elmer 241 polarimeter.
3.2. General enzymatic assays
The assay mixture had the following composition: 50
0.1 M phosphate buffer (pH 6.7), 50 L of 2 mM p-nitrophenyl-
-glucopyranoside (the method was optimized giving Km and Vmax
values of 0.258 mM and 0.0418 mM minꢁ1 mgꢁ1, respectively) and
50 L of Baker’s yeast -glucosidase solution (0.04 U/mL). All the
samples were run in 5 replicas. The test samples were dissolved
lL of
l
a-
D
l
a
in 2% DMSO in 0.1 M phosphate buffer (pH 6.7) and 50 lL solutions
having the following range of concentration: 0.19 ꢀ 10ꢁ3 and
46 mM. The a-glucosidase activity was determined by monitoring
3.3.2. MAL12 binding site definition
the p-nitrophenol released from p-nitrophenyl-a-D-glycopyranos-
MAL12 binding site is composed of Asp214, Glu276, and Asp349
catalytic residues. In addition to the catalytic residues, molecular
docking studies confirm Asp68, Tyr71, and Arg439 as important
ide spectrophotometrically at 405 nm over 20 min. The IC50 values
were obtained from the inhibition curves. These determinations
were made using an Elisa Microplate spectrophotometer UV/Vis
(Ref. ELX800G) from Bio-Tek Instruments Inc.
residues in the a-glucosidase inhibition.
3.3.3. Docking calculations
3.3. Molecular modeling
The molecular structures of the 5-membered iminocyclitols
were built and optimized with the MMFF94x forcefield as imple-
mented in the MOE software (version 2011.10). These compounds
were docked into the MAL12 model enzyme binding site. Molecu-
lar docking simulations were performed, using GOLD (Genetic
Optimization Ligand Docking) software (version 5.1.0). GOLD uses
an evolutionary genetic algorithm to optimize the docked confor-
mation of the flexible inhibitor within the enzyme. For each com-
pound, 1000 docking runs were performed. The following genetic
algorithm parameters were used: population size = 100; selected
pressure = 1.1; number of operations = 1000; number of is-
lands = 5; niche size = 2; migrate = 10; mutate = 95; crossover = 95.
Each conformation was ranked according to its goldscore scoring
3.3.1. MAL12 model building and validation
Homology modeling of
yeast was carried out to predict its 3D-structure. The amino acid
sequence of -glucosidase (MAL12) from baker’s yeast comprises
a-glucosidase (MAL12) from baker’s
a
584 amino acid residues and was retrieved from UniProt protein
resource data bank (http://www.uniprot.org/), under the access
code
P53341
(http://www.uniprot.org/uniprot/P53341.html).
Using the Molecular Operating Environment program (MOE) ver-
sion 2011.10 (http://www.chemcomp.com/software-moe2011.
htm) we searched for proper structural templates on the PDB data-
base of protein structures and sequences and aligned the obtained
results. MOE-Align implements a modified version of the align-
function. The top solutions (the ones with the highest goldscore33
)
were visually inspected and critically evaluated and for each inhib-
itor the highest scoring conformation was chosen as the actual
binding conformation (Fig. 7).
Table 1
IC50 values of 15a, 15b, 16a and 16a0 for
(Saccharomyces cerevisiae)
a
-glucosidase inhibition
IC50 (mM)
3.4. Synthesis of pyrrolidine iminocyclitol inhibitors
Inhibitor
Acarbose
15a
15b
16a
16a0
0.641
10.9
12.5
38.0
25.0
3.4.1. (3S,4S)-N-Benzyl-3,4-dihydroxypyrrolidine 16a
(3S,4S)-N-Benzyl-3,4-dihydroxypyrrolidine 16a was synthe-
sized according to the method of Nagel.16
½
a 2D0
+40 (c 0.7 in Meth-
ꢃ
anol) [½a 2D0
ꢃ
+32.4 (c 4.2 in Methanol)].16a

1916
L. R. Guerreiro et al. / Bioorg. Med. Chem. 21 (2013) 1911–1917
Figure 6. Graphic representation of the homology modeling structure of MAL12 with catalytic residues highlighted.
Figure 7. Binding mode of compounds 15a, 15b, 16a, 16a0, 16b and acarbose in the MAL12 active site of the homology model.
3.4.2. (3R,4R)-N-Benzyl-3,4-dihydroxypyrrolidine 16a0
170.4 ppm, 137.5, 128.9, 128.4, 127.3, 77.7, 59.7, 58.1, 21.0. MS
(ESI-TOF), 278.14 (M+1).
(3R,4R)-N-Benzyl-3,4-dihydroxypyrrolidine 16a0 was synthe-
sized according to the method of Nagel.16
Methanol).
½
a 2D0
ꢃ
ꢁ28.2 (c 1.05 in
3.4.4. (3R,4R)-N-Benzyl-3,4-pyrrolidinediol diacetate 16b0
Using the same procedure as described previously with 16a0 as
the substrate, and in the same quantities, the title compound was
obtained as a light brown oil (0.96 g, 81.5%). The NMR data were
similar to 16b.
3.4.3. (3S,4S)-N-Benzyl-3,4-pyrrolidinediol diacetate 16b
(3S,4S)-N-Benzyl-3,4-dihydroxypyrrolidine 16a (2 g, 10 mmol)
was dissolved in pyridine (10.5 mL) and acetic anhydride
(7.7 mL). The solution was stirred for 23 h at rt. The solvents were
removed in vacuo, and the mixture was dissolved in water (25 mL)
and extracted with EtOAc (2 ꢀ 15 mL). The organic phases were
collected, dried with MgSO4, filtered and the solvent was removed,
the crude product was purified by column chromatography (SiO2,
(2:1) to (1:1) Hex:EtOAc) to give the title compound as a light
brown oil (1.12 g, 93%). 1H NMR (CDCl3, 400 MHz) d: 2.05 ppm(s,
6H); 2.52 (dd, J 4, 12 Hz, 2H); 3.04–3.08 (m, 2H); 3.59 (d, J 12 Hz,
1H, AB system); 3.66 (d, J 12 Hz, 1H, AB system); 5.12 (s broad,
2H); 7.25–7.31 (m, 5H). 13C NMR (CDCl3, 100.61 MHz) d:
3.4.5. (3R,4R)-1-(tert-Butoxycarbonyl)-3,4-pyrrolidinediol
diacetate 19a0
(3R,4R)-Pirrolidine-3,4-diol 15a0 (0.848 g, 8.23 mmol) was dis-
solved in ethanol (35 mL) this was followed by the slow addition
of Boc-anhydride at 0 °C. The mixture was stirred for 23 h at 0 °C.
The reaction was warmed to 50 °C and stirred for over 1 h, and
the solvents were removed in vacuum to give a brown solid. To
the crude product was added pyridine (4.5 mL) and Ac2O
(5.6 mL) and stirred for 19 h at rt. The solvent was removed in

L. R. Guerreiro et al. / Bioorg. Med. Chem. 21 (2013) 1911–1917
1917
vacuum and the crude product was purified by column chromatog-
raphy [SiO2, (2:1) to (1:1) Hex:EtOAc] to give the title compound
19a0 as a colorless oil (1.8 g, 76%) 1H NMR (400 MHz, CDCl3):
d = 1.43 ppm (s, 9H), 2.04 (s, 6H), 3.38 (d, J 12 Hz, 1H), 3.48 (d, J
12 Hz, 2H), 3.66–3.63 (m, 2H), 5.09 (s broad, 2H). 13C NMR
(100 MHz, CDCl3): d = 20.8 ppm, 28.3, 49.6, 50.1, 74.2, 75.1, 79.9,
154.2, 169.6, 169.7. MS (ESI-TOF), 288.15 (M+1).
personnel of the mass spectrometry unit at C.A.C.T.I (Univ. of Vigo,
Spain) are acknowledged for mass spectrometric analyses. Dr. Oli-
via Furtado of the Laboratorio Nacional de Energia e Geologia
(LNEG), is acknowledged for the optical rotation measurements.
We also thank Miss Mariana Moreira, for her assistance in the
exploratory deprotection studies leading to 15b and its
enantiomer.
3.4.6. (3S,4S)-1-(tert-Butoxycarbonyl)-3,4-pyrrolidinediol
diacetate 19a
Supplementary data
Using the same procedure as described previously: 15a (0.750 g,
7.27 mmol), Boc-anhydride (1.22 g, 10.2 mmol) in ethanol (20 mL)
as solvent. Acetylation was conducted with pyridine (5.6 mL) and
acetic anhydride (4.5 mL) to give the title compound 19a as a color-
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmc.2013.01.030.
References and notes
less oil (0.868 g, 42%). The NMR data was similar to 19a0. ½a D20
ꢃ
+30
1. For key references. (a) Lillelund, V. H.; Jensen, H. H.; Liang, X.; Bols, M. Chem.
Rev. 2002, 102, 515; (b) Sugiyama, M.; Hong, Z.; Liang, P.-H.; Dean, S. M.;
Whalen, L. J.; Greenberg, W. A.; Wong, C.-H. J. Am. Chem. 2007, 129, 14813. , and
references cited therein; For more specific references see: Moorthy, N. S. H. N.;
Bras, N. F.; Ramos, M. J. Bioorg. Med. Chem. 2012, 20, 6945; Bello, C.; Cea, M.; Dal
Bello, G.; Garuti, A.; Rocco, I.; Cirmena, G.; Moran, E.; Nahimana, A.; Duchosal,
M. A.; Fruscione, F.; Pronzato, P.; Grossi, F.; Patrone, F.; Ballestrero, A.; Dupuis,
M.; Sordat, B.; Nencioni, A.; Vogel, P. Bioorg. Med. Chem. 2010, 18, 3320;
Moreno-Clavijo, E.; Carmona, A. T.; Vera-Ayoso, Y.; Moreno-Vargas, A. J.; Bello,
C.; Vogel, P.; Robina, I. Org. Biomol. Chem. 2009, 7, 1192.
2. Caines, M. E. C.; Hancock, S. M.; Tarling, C. A.; Wrodnigg, T. M.; Stick, R. V.;
Stütz, A. E.; Vasella, A.; Withers, S. G.; Strynadka, N. C. J. Angew. Chem. 2007, 46,
4474.
3. Inouye, S.; Tsurouka, T.; Ito, T.; Niida, T. Tetrahedron 1968, 24, 2125.
4. Kim, Y. J.; Kido, M.; Bando, M.; Kitahara, T. Tetrahedron 1997, 53, 7501.
5. Schofield, A. M.; Fellows, L. E.; Nash, R. J.; Fleet, G. W. J. Life Sci. 1986, 39, 645.
6. Chapman, T. M.; Courtney, S.; Hay, P.; Davis, B. G. Chem. Eur. J. 2003, 9, 3397.
and references cited therein.
(c 1.08 in CHCl3).
3.4.7. (3R,4R)-3,4-Diacetoxypyrrolidinium 2,2,2-trifluoroacetate
15b0
(3R,4R)-1-(tert-Butoxycarbonyl)-3,4-pyrrolidinediol diacetate
19a0 (0.4 g, 1.39 mmol) was dissolved in dichloromethane (4 mL)
this was followed by the addition of TFA (4 mL) and the mixture
was stirred at rt for 1.5 h (TLC analysis revealed the presence of
no substrate). The mixture was evaporated in vacuo to give a
brown oil (0.255 g, 98%). 1H NMR (400 MHz, CDCl3) d: 2.09 ppm
(s, 6H), 3.51 (d, J 12 Hz, 2H), 3.65 (dd, J 4, 12 Hz, 2H), 5.25 (d, J
4 Hz, 2H), 10.23 (s broad, 2H). 13C NMR (100 MHz, CDCl3):
d = 20.4 ppm, 49.0, 74.2, 169.5. MS (ESI-TOF), 188.1 (M).
7. Liang, P.-H.; Cheng, W.-C.; Lee, Y.-L.; Yu, H.-P.; Wu, Y.-T.; Lin, Y.-L.; Wong, C.-H.
ChemBioChem 2006, 7, 165.
8. Chapman, T. M.; Davies, I. G.; Gu, B.; Block, T. M.; Scopes, D. I. C.; Hay, P. A.;
Courtney, S. M.; McNeill, L. A.; Schofield, C. J.; Davis, B. G. J. Am. Chem. Soc. 2005,
127, 506.
9. Takebayashi, M.; Hiranuma, S.; Kanie, Y.; Kajimoto, T.; Kanie, O.; Wong, C.-H. J.
Org. Chem. 1999, 64, 5280.
10. Calveras, J.; Egido-Gabás, M.; Gómez, L.; Casas, J.; Parella, T.; Joglar, J.; Bujons,
J.; Clapés, P. Chem. Eur. J. 2009, 15, 7310.
3.4.8. (3S,4S)-3,4-Diacetoxypyrrolidinium 2,2,2-trifluoroacetate
15b
Using the same procedure as described previously: 19a (0.300 g,
1.04 mmol), CH2Cl2 (2 mL) and TFA (2 mL) the title compound 15b
was obtained as a brown oil (0.190 g, 98%). The NMR data was sim-
ilar to 15b0.
11. Bols, M. Tetrahedon Lett. 1996, 37, 2097.
4. Conclusions
12. Godskesen, M.; Lundt, I. Tetrahedon Lett. 1998, 39, 5841.
13. Asano, N.; Oseki, K.; Kizu, H.; Matsui, K. J. Med. Chem. 1984, 37, 3701.
14. Protein Data Bank. http://www.rcsb.org.
15. Nuti, F.; Paolini, I.; Cardona, F.; Chelli, M.; Lolli, F.; Brandi, A.; Goti, A.; Rovero,
P.; Papini, A. M. Bioorg. Med. Chem. 2007, 15, 3965.
16. (a) Nagel, U.; Kinzel, E.; Andrade, J.; Prescher, G. Chem. Ber. 1986, 119, 3326; (b)
Beck, W.; Nagel, U. U.S. patent 4634,775, 1985.
17. Martins, J. G.; Barrulas, P. C.; Marques, C. S.; Burke, A. J. Synth. Commun. 2008,
38, 1365.
18. Guerreiro, L. R.; Simões, M.; Carreiro, E. P.; Burke, A. J. unpublished results.
19. Greene, T. W.; Wuts, P. G. M. Protective Groups in Organic Synthesis, 3rd ed.;
John Wiley & Sons, Inc: New York, 1999.
20. Kazzouli, S. E.; Koubachi, J.; Berteina-Raboin, S.; Mouaddib, A.; Guillaumet, G.
Tetrahedron Lett. 2006, 47, 8575.
21. Routier, S.; Saugé, L.; Ayerbe, N.; Coudert, G.; Mérour, J.-Y. Tetrahedron Lett.
2002, 43, 589.
22. Liu, Y.-S.; Zhao, C.; Bergbreiter, D. E.; Romo, D. J. Org. Chem. 1998, 63, 3471.
23. Ferreira, S. B.; Sodero, A. C. R.; Cardoso, M. F. C.; Lima, E. S.; Kaiser, C. R.; Silva, F.
P.; Ferreira, V. F. J. Med. Chem. 2010, 53, 2364.
We have successfully synthesized a small library of trans-pyr-
rolidine
the 2 or 5 positions. These compounds were assayed for baker’s
yeast -glucosidadse inhibition using acarbose as a reference. It
a-glucosidase inhibitors, lacking functionality in either
a
was the non-benzylated diol 15a which showed the highest inhibi-
tion (IC50 = 10.9 mM). It seems that the presence of a free NH group
and free hydroxyl groups in the 3 and 4 positions are important for
favorable interaction with the enzyme active site. However, this
study reinforces the fact that in order to have micromolar levels
of inhibition or better (like in the examples shown above) it is nec-
essary to have substituents in the 2 and/or 5 positions.
A molecular modeling study using a completely new MAL12 en-
zyme homology model was carried out and explains the inhibitory
activity of the tested compounds. Our docking studies explain the
improved inhibitory activity of 15a and 15b based on the distances
and interactions that these compounds establish with the active
site catalytic residues. We are currently investigating these mole-
24. Varrot, A.; Tarling, C. A.; Macdonald, J. M.; Stick, R. V.; Zechel, D. L.; Withers, S.
G.; Davies, G. J. J. Am. Chem. 2003, 125, 7496.
25. Zechel, D. L.; Withers, S. G. Acc. Chem. Res. 2000, 33, 11.
26. Jones, G.; Willett, P.; Glen, R. C. J. Mol. Biol. 1995, 245, 43; Jones, G.; Willett, P.;
Glen, R. C.; Leach, A. R.; Taylor, R. J. Mol. Biol. 1997, 267, 727.
27. Yamamoto, K.; Miyake, H.; Kusunoki, M.; Osaki, S. FEBS J. 2010, 277, 4205.
28. Park, J.-H.; Ko, S.; Park, H. Bull. Korean Chem. Soc. 2008, 29, 921.
29. Roujeinikova, A.; Raasch, C.; Sedelnikova, S.; Liebl, W.; Rice, W. J. Mol. Biol.
2002, 321(1), 149.
30. Needleman, S. B.; Wunsch, C. D. J. Mol. Biol. 1970, 48, 443.
31. Eswar, N.; Eramian, D.; Webb, B.; Shen, M. -Y.; Sali, A. Protein Structure
Modeling with MODELLER. In Methods in Molecular Biology, Kobe, B. G. M. H. T.,
Ed.; Springer, 2008; Vol. 426, pp 145–159.
cules and other analogues for the inhibition of mammalian a-glu-
cosidases, and other carbohydrate-processing enzymes. We are
also looking at saturation transference difference-NMR (STD-
NMR) as another method to probe these very interesting
interactions.
32. Wang, J.; Cieplak, P.; Kollman, P. A. J. Comput. Chem. 2000, 21, 1049.
33. GoldScore performs a force field based scoring function and is made up of four
components: (1) protein–ligand hydrogen bond energy (external H-bond); (2)
protein–ligand van der Waals energy (external vdw); (3) ligand internal van
der Waals energy (internal vdw); (4) ligand intramolecular hydrogen bond
energy (internal-H-bond).
Acknowledgements
EPC thanks the Fundação para a Ciência e a Tecnologia (FCT) for
a post-doctoral research fellowship (SFRH/BPD/72182/2010). The