Synthesis and biological evaluation of analogues of butyrolactone I and molecular model of its interaction with CDK2.

Article abstract of DOI:10.1039/b403052d

A series of analogues of butyrolactone I, a natural product isolated from Aspergillus terreus that selectively inhibits the CDK2 and CDK1 kinases and that has been found to exhibit an interesting antiproliferative activity, have been synthesized. Its anti

Full text of DOI:10.1039/b403052d

View Article Online / Journal Homepage / Table of Contents for this issue
Synthesis and biological evaluation of analogues of butyrolactone I
and molecular model of its interaction with CDK2†
Miguel F. Braña,*a M. Luisa García,a Berta López,a Beatriz de Pascual-Teresa,a Ana Ramos,a
Jose M. Pozuelob and M. Teresa Domínguezb
a Departamento de CC Químicas, Facultad de CC, Experimentales y de la Salud,
Universidad San Pablo CEU, Urb. Montepríncipe, Boadilla del Monte, 28668 Madrid, Spain
b Departamento de Biología, Facultad de CC, Experimentales y de la Salud,
Universidad San Pablo CEU, Urb. Montepríncipe, Boadilla del Monte, 28668 Madrid, Spain
Received 1st March 2004, Accepted 22nd April 2004
First published as an Advance Article on the web 16th June 2004
A series of analogues of butyrolactone I, a natural product isolated from Aspergillus terreus that selectively inhibits
the CDK2 and CDK1 kinases and that has been found to exhibit an interesting antiproliferative activity, have been
synthesized. Its antitumor activity has been tested. Molecular models of the complex between butyrolactone I
and the CDK2 active site have been built using a combination of conformational search and automated docking
techniques. The stability of the resulting complexes has been assessed by molecular dynamics simulations and the
experimental results obtained for the synthesized analogues are rationalized based on the molecular models.
the hydroxyl groups in the para position of the aromatic rings,
Introduction
showed an activity (IC50 = 55.0 µg mlϪ1) that is about 190 times
lower than butyrolactone I.
Cyclin dependent kinases (CDKs) play a central role in
the regulation of the cell division cycle, which makes them a
promising target for the development of cancer therapeutic
agents. A big effort has been made in the last few years in the
search of low molecular weight specific inhibitors of CDKs.
These inhibitors block cell cycle progression and display
interesting antitumor activities.1
Butyrolactone I, a natural product isolated from Aspergillus
terreus var. africans IFO 8835 in 1977,2 has been found to
exhibit antiproliferative activity against colon and pancreatic
carcinoma, human lung cancer3 and prostatic cancer4 cell lines.
It selectively inhibits CDK2 and CDK1 kinases, both of which
play important roles in cell progression from G1 to S phase and
from G2 phase to M phase, respectively, in mammalian cells.
However, it has little effect on mitogen-activated protein kinase,
protein kinase C, cyclic-AMP dependent kinase, casein kinase
II, casein kinase I or epidermal growth factor-receptor tyrosine
kinase.5 It behaves as an ATP competitive inhibitor.5 Due to the
structural complexity of this compound, little is known about
its binding mode to its target.
In this work, we have synthesized a series of analogues of
butyrolactone I, where the hydroxyl groups in the aromatic
rings have been substituted by a variety of electron-donor and
electron-withdrawing groups, with the aim of obtaining active
compounds by an accessible synthetic path, and establish
a SAR for the substitution at that position. Synthesized com-
pounds have been evaluated for their antitumor and CDK1
inhibitory activity, and computer-based techniques have been
used to propose a binding mode of butyrolactone I to CDK2
and to rationalize the experimental results obtained for the
synthesized analogues.
Results and discussion
Chemistry
Butyrolactones
2 were synthesized by treatment of the
corresponding aryl pyruvic methyl esters 1 with K2CO3 in
acetone, or DBU in DMF (Scheme 1).
Butyrolactones 2a and 2b were previously described by an
analogous method, using saturated K2CO3 aqueous solution.7
The required α-ketoesters 1 were readily accessible from
commercially available nitriles or aldehydes, or by previously
described methods, as depicted in Scheme 2. Thus, treatment
of the appropriately substituted aryl acetonitriles 3c and 3f–h
with diethyl oxalate furnished ethyl 3-aryl-3-cyanopyruvates
4c, 4f–h,8 which were hydrolyzed and decarboxylated to give
α-hydroxycinnamic acids 5c, 5f–h. α-Hydroxycinnamic acids
5d,e were synthesized from the corresponding aryl carboxalde-
hydes, following the method described by H. N. C. Wong.9
The first step involves the Erlenmeyer synthesis of 4-benzyl-
ideneoxazol-5(4H )-ones 7d,e from benzaldehydes 6d,e and
N-acetylglycine. Compounds 7d,e were refluxed with 3 M
hydrochloric acid until complete conversion to α-hydroxy-
cinnamic acids 5d,e, which was monitored by thin layer
chromatography. All the α-hydroxycinnamic acids 5 were
converted into the corresponding aryl pyruvic methyl esters
by reaction with methyl iodide in the presence of DBU
(compounds 1a–c, 1f–h) or methanol in the presence of HCl gas
(compounds 1d,e).
Looking at the work developed by Morishima et al.6 it seems
that suppression of the alkenyl chain does not markedly affect
the anti-CDK1 activity of butyrolactone I. These authors
found that compound 2b showed an inhibitory activity against
CDK1 (IC50 = 4.75 µg mlϪ1) only 16 fold lower than the lead
compound (IC50 = 0.29 µg mlϪ1). However, analogue 2a, lacking
† Electronic supplementary information (ESI) available: general
experimental procedure for preparation, physical and spectral
characterizations (1H/13C NMR, mass spectrometry, and IR data) of
compounds 1a–h, 4c, 4f, 4h, 5c–f, 5h, and 7d–e. AMBER parameters
and partial atomic charges for butyrolactone I (Tables 2 and 3).
Docking parameters (Table 4). See http://www.rsc.org/suppdata/ob/b4/
b403052d/
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 8 6 4 – 1 8 7 1
T h i s j o u r n a l i s © T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 4
1864

View Article Online
Scheme 1 Synthesis of butyrolactones 2a–h.
Scheme 2 Synthesis of α-ketoesters 1a–h.
Table 1 Antiproliferative activities of butyrolactones 2
Biological evaluation
Butyrolactone I has been shown to present antitumor effects on
several lung cancer cell lines with IC50 values in the order of
0.12 µM (Nishio3) and prostatic cancer cell lines with IC50
values in the order of 70 µM,4 suggesting that CDK1 kinase
inhibitors may be active against human solid tumors.
The aim of this work was to obtain structurally simplified
analogues, where the 3-methylbut-2-enyl chain present in
butyrolactone I, has been eliminated. Compound 2a, bearing
two unsubstituted aromatic rings, showed only moderate
activity against the HT-29 cell line, and was completely inactive
against the rest of the cell lines tested (Table 1). The intro-
duction of a polar amino group (compound 2d) in the para
position of both aromatic rings did not bring about any
improvement in activity. However, compounds 2c and 2f, with
an electron-withdrawing group in the aromatic rings, showed
moderate activity against some of the cell lines tested. This
activity is maintained by compounds 2e, 2g and 2h, where an
electron-donor substituent has been introduced in the same
position. These results suggest that the electron nature of the
substituents does not affect the antitumor activity.
HT-29a
IC50, µM
HeLab
IC50, µM
PC-3c
IC50, µM
Compounds
R
2a
2b
2c
2d
2e
2f
H
OH
Cl
46
> 1000
25
> 1000
42.15
> 1000
57
100
> 1000
14.5
> 1000
> 1000
57.3
> 1000
10.96
36.39
263
NH2
97.1
But
39.63
32.61
55
CF3
OCH3
CH3
2g
2h
55
50
157
a Human colon carcinoma cell line. b Human cervical carcinoma cell
line. c Human prostate carcinoma cell line.
Finally, compound 2b, where the only structural deviation
from butyrolactone I is the lack of the 3-methylbut-2-enyl
chain, was completely inactive against all the cell lines tested.
In summary, no structure–activity relationship has been
observed in relation to the aromatic substitution, and our
results suggest that the presence of the alkenyl chain is essential
for antitumor activity.
All the synthesized compounds were tested against CDK1 by
Prof. Laurent Meijer (C.N.R.S., Station Biologique, Roscoff )
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 8 6 4 – 1 8 7 1
1865

View Article Online
and they did not show any significant inhibitory activity. This
result was surprising, especially in the case of the hydroxylated
compound 2b, previously reported to be active against this
enzyme. However, it is in accordance with the results of anti-
tumor activity, and prompted us to propose that the alkenyl
chain plays an important role in the stability of the CDK1–
butyrolactone I complex. A computational study has been
carried out in order to get information about this complex, and
rationalize the experimental results.
Molecular modeling
We have carried out a molecular modeling study on the mode
of interaction of both enantiomers of butyrolactone I with
CDK2 making use of automated docking methods and molecu-
lar dynamics simulations. Although the inhibitory activity has
been tested against CDK1, these studies have been carried
out with CDK2 because only X-ray structures of CDK2 in
complexes with small molecule ligands have been solved to date,
which limits the structure-based studies to this cyclin-depend-
ent kinase. However, the two protein kinases have nearly 66%
sequence identity overall, and their binding pockets are quite
similar.10 It is therefore possible to assume that these com-
pounds will not show selectivity between these two kinases, as is
the case for butyrolactone I.
First the ligands were submitted to a Monte Carlo conform-
ational search and the lowest energy conformations were taken
as the starting conformations for the automated docking into
CDK2. The lowest energy complex for each enantiomer
predicted by AUTODOCK is shown in Fig. 1. As shown in
this figure both enantiomers bind to the active site of CDK2
locating the aromatic and lactone rings in approximately the
same areas. Both are able to establish hydrogen bonds with
different residues in the receptor as shown schematically in
Fig. 2. The atoms from the ligand (O1, O23, O14 and O37) and
the residues in the receptor (Lys33, Asp86 and Lys129) involved
in this type of interaction are common in both enantiomers, but
(R)-butyrolactone I forms extra hydrogen bonds with Thr14.
This difference in the binding affinity is overcome in the (S)-
enantiomer by a larger van der Waals energy term leading to a
1.13 Kcal molϪ1 difference in the binding energy in favour of
the (S)-butyrolactone I–CDK2 complex. However, this energy
difference is sufficiently small as to consider both as candidates
to occupy the active site of CDK2.
Fig.
2
Schematic representation of the binding mode of (R)-
butyrolactone I (top) and (S)-butyrolactone I (bottom) to CDK2.
Color code: residues with which the ligand interacts through hydrogen
bonds (blue) or through van der Waals interactions (magenta).
square (rms) deviations of the coordinates of the solutes with
respect to the initial structures was measured (Fig. 3). The
progression of this parameter indicates that the complexes do
not experience large conformational changes during the
sampling time and thus can be considered to be in a state near
equilibrium. In order to evaluate the relative contributions of
the different residues to complex stabilization the 125 structures
collected from the last 500 ps of the simulations were averaged
and energy-minimized, and the interaction energy between
butyrolactone I and the binding site was decomposed on a
residue basis using the ANAL module of AMBER (Figs. 4
and 5). A superimposition of the energy-minimized average
structures and the initial structures is shown in Fig. 6. As
expected from the observed rms deviations shown in Fig. 3,
both complexes remain stable along the simulation time, thus
the conformational changes are not very relevant, except in the
area of the 3-methyl-2-butenyl group, where there seems to be a
wide pocket in the binding site that allows the rotation of the
benzyl group, while maintaining the overall conformation of
the ligands and the interactions with the receptor. The models
we propose for butyrolactone I give rise to rather strong electro-
static interactions with Lys33 and Asp86 in each of the two
enantiomers. It can be seen that the major contributors to the
Fig. 1 (R)-Butyrolactone I (blue) and (S)-butyrolactone I (red) bound
in the CDK2 active site. Residues relevant to the discussion are
displayed as thick sticks in yellow.
The evidence that proteins, which constitute therapeutic
targets, are mobile is substantial.11 In order to take into account
protein flexibility the behaviour of both complexes was studied
in a dynamic context and the progression of the root-mean-
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 8 6 4 – 1 8 7 1
1866

View Article Online
Fig. 3 Time evolution (ps) of the root-mean-square deviation (Å) between the simulated structures and the corresponding initial structures (all
solute non-hydrogen atoms were included in the comparisons) for complexes (R)-butyrolactone I–CDK2 (grey) and (S)-butyrolactone I–CDK2
(black).
Fig. 4 Residue-based van der Waals energy term (Kcal molϪ1) of the interaction between (R)-butyrolactone I (blue) and (S)-butyrolactone I (red)
and the CDK2 active site.
Fig. 5 Residue-based electrostatic energy term (Kcal molϪ1) of the interaction between (R)-butyrolactone I (blue) and (S)-butyrolactone I (red) and
the CDK2 active site.
van der Waals energy term are Gly13, Tyr15, Val18, Lys129,
Gln131, Asn132 and Lys133, and that the relative importance
of Gly11, Glu12, Phe82, Leu134, and Asp145 depends on the
enantiomer considered. From this energy decomposition it can
be clearly deduced that the van der Waals energy contribution
is greater for the (R)-butyrolactone I–CDK2 complex. The
electrostatic energy term shows the relevance of Asp86 and
Lys33 in the molecular recognition of the ligand in the active
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 8 6 4 – 1 8 7 1
1867

View Article Online
Fig. 6 View of superimposed C(α) traces of the energy-minimized average structure of the last 500 ps of the MD simulation (green) and the initial
structure (magenta) for CDK2–(R)-butyrolactone I (left) and CDK2–(S)-butyrolactone I (right) complexes. Butyrolactones are shown as sticks.
site of CDK2. The resulting total interaction energy for the two
energy minimized average structures is 134 Kcal molϪ1 more
stable for the (R)-enantiomer.
To analyse in more detail this effect, we measured, for all the
synthesized compounds in the (R)-configuration, and for the
natural product in the docked orientation predicted by
AUTODOCK, the torsion angle depicted in Fig. 8 as D1. In
addition, the value of this angle was monitored along the
simulation time for the (R)-butyrolactone I–CDK2 complex.
The results are summarized in Fig. 8. It is interesting to point
out that this value lies between Ϫ120.3Њ and Ϫ170.3Њ for the
butyrolactone I and between 50.8Њ and 103.8Њ for compounds 2.
In Fig. 8, we have represented the corresponding Newman
projections, where it can be appreciated that in the proposed
binding mode of the natural product, the ester group lies below
A recent study on several metabolites of butyrolactone I has
allowed the assignment of the configuration of the natural
product to be the (R)-enantiomer.12 According to this study,
both binding modes are chemically reasonable, although the
(R)-enantiomer leads to a more stable CDK2 complex.
A similar docking procedure was used to explore the possible
binding modes of the synthesized analogues 2 in the (R)-con-
figuration. First the ligands were submitted to a Monte Carlo
conformational search and the lowest energy conformations
were taken as the starting conformations for the automated
docking into CDK2. In all cases, due to the lack of the 3-methyl-
2-butenyl group, the analogues adopt a different lowest energy
conformation and, in consequence, a different orientation in the
binding modes predicted by AUTODOCK. As an example,
Fig. 7 shows the binding mode of compound 2b, superimposed
with butyrolactone I.
Fig. 8 Newman projections along dihedral angle D1 for the natural
product (left) and for compounds 2 (right), all of them as the (R)-
enantiomers. Values for each compound are summarized below.
Fig. 7 Docked (R)-butyrolactone I (pale blue) and compound 2b
(dark blue) in the CDK2 active site.
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 8 6 4 – 1 8 7 1
1868

View Article Online
the Ar2 ring, whereas in all other derivatives, the butyrolactone
moiety is the one which lies below this system. This effect seems
to be related to the steric hindrance of the 3-methyl-2-butenyl
group and could explain the absence of activity of the syn-
thesized compounds. In consequence, the design and synthesis
of new ligands where this group is substituted by other bulky
groups will be undertaken in future work.
Methyl 4-hydroxy-3-(4-hydroxyphenyl)-2-(4-hydroxyphenyl-
methyl)-5-oxo-2H-furan-2-carboxlate (2b). By method A, 1b
(1.131 g, 5.82 mmol) and K2CO3 (1.6 g, 11.64 mmol) yielded a
white solid. Flash chromatography of the crude product using
hexane–AcOEt (1 : 1) as eluent gave 2b (0.63 g, 61%), mp 213–
215 ЊC (from AcOH) (Found: C, 63.85; H, 4.54. C19H16O7
requires C, 64.04; H, 4.52%); νmax(KBr)/cmϪ1 3440, 3300, 1750
and 1710; δH (300 MHz; CD3OD; Me4Si) 7.59 (2 H, d, J 8.55,
ArH), 6.85 (2 H, d, J 8.55, ArH), 6.63 (2 H, d, J 8.55, ArH),
6.50 (2 H, d, J 8.55, ArH), 3.77 (3 H, s, CH3) and 3.46 (2 H, s,
CH2); δC (300 MHz; DMSO-d6; Me4Si) 169.82, 167.98, 157.94,
156.32, 138.13, 131.20, 128.84, 127.45, 123.19, 121.04, 115.90,
114.64, 84.73, 53.56 and 38.00.
Conclusions
Butyrolactone I has been found to exhibit antiproliferative
activity against several cell lines by selectively inhibiting CDK2
and CDK1 kinases. Interest in the synthesis of analogues of
this drug is warranted, not only by its antitumor profile, but
also by its structural complexity. Nevertheless, no structural
details of the interaction of butyrolactone I with CDK2
are available. Our simulations show the importance of the
3-methyl-2-butenyl chain in the overall conformation of the
natural product and it could explain the fact that the
synthesized analogues do not maintain the antitumor and
CDK1 inhibitory activity.
Methyl
4-hydroxy-3-(4-chlorophenyl)-2-(4-chlorophenyl-
methyl)-5-oxo-2H-furan-2-carboxylate (2c). By method A, 1c
(0.45 g, 2.11 mmol) and K2CO3 (0.59 g, 4.22 mmol) yielded a
white solid. Flash chromatography of the crude product using
hexane–AcOEt (7 : 3) as eluent gave 2c (0.3 g, 72%), mp 210–
212 ЊC (from EtOH) (Found: C, 57.98; H, 3.63. C19H14Cl2O5
requires C, 58.03; H, 3.58%); νmax(KBr)/cmϪ1 3280 and 1750;
δH (300 MHz; DMSO-d6; Me4Si) 11.48 (1 H, br s, OH), 7.69
(2 H, d, J 8.55, ArH), 7.58 (2 H, d, J 8.55, ArH), 7.25 (2 H, d,
J 8.55, ArH), 6.85 (2 H, d, J 8.55, ArH), 3.76 (3 H, s, CH3), 3.62
(1 H, d, J 14.64, CH2) and 3.55 (1 H, d, J 14.64, CH2); δC (300
MHz; DMSO-d6; Me4Si) 169.06, 167.17, 140.96, 133.27, 132.21,
131.97, 131.94, 129.15, 128.70, 128.61, 127.89, 124.75, 84.49,
53.75 and 37.79.
Experimental
General methods
Melting points (uncorrected) were determined on a Stuart
Scientific SMP3 apparatus. Infrared (IR) spectra were recorded
1
with a Perkin-Elmer 1330 infrared spectrophotometer. H and
13C NMR: δ were recorded on a Bruker 300-AC instrument.
Chemical shifts (δ) are expressed in parts per million relative to
internal tetramethylsilane; coupling constants (J) are in hertz.
Elemental analyses (C, H, N) were performed on a Perkin
Elmer 2400 CHN apparatus at the Microanalyses Service of the
University Complutense of Madrid. Thin-layer chromato-
graphy (TLC) was run on Merck silica gel 60 F-254 plates.
Unless stated otherwise, starting materials used were high-
grade commercial products.
Methyl
4-hydroxy-3-(4-aminophenyl)-2-(4-aminophenyl-
solution of
methyl)-5-oxo-2H-furan-2-carboxylate (2d).
A
K2CO3 (4.5 g, 32.66 mmol) in water (20 cm3) was added to 1d
(3.0 g, 13.06 mmol) and the mixture was stirred at room tem-
perature for 30 min. The solution was neutralized (1 M HCl)
and the resulting precipitate was filtered, washed with water
and dried yielding 2d as a yellow solid (1.3 g, 56%), mp 104 ЊC
dec; νmax(KBr)/cmϪ1 3450, 3380, 1755 and 1740; δH (300 MHz;
CDCl3; Me4Si) 7.54 (2 H, d, J 8.80, ArH), 6.73 (2 H, d, J 8.80,
ArH), 6.65 (2 H, d, J 8.25, ArH), 6.45 (2 H, d, J 8.25, ArH),
3.77 (3 H, s, CH3), 3.53 (1 H, d, J 14.50, CH2), 3.46 (1 H, d,
J 14.50, CH2); δC (300 MHz; CDCl3; Me4Si) 169.96, 169.21,
147.26, 145.06, 136.21, 131.26, 129.23, 129.04, 122.76, 119.79,
115.12, 114.93, 85.82, 53.43 and 38.62.
Synthesis of butyrolactones 2; General procedure A
To a solution of ester 1 (5 mmol) in acetone (30 cm3) was added
K2CO3 (10 mmol) and the mixture was stirred at room temper-
ature until the reaction was completed (TLC). After the acetone
was evaporated, 1 M HCl was added, and the solution was
extracted with Et2O (3 × 20 cm3). The combined organic layers
were washed with brine, dried (MgSO4), and evaporated to
dryness.
Methyl
4-hydroxy-3-(4-tert-butylphenyl)-2-(4-tert-butyl-
phenylmethyl)-5-oxo-2H-furan-2-carboxylate (2e). By method
A, 1e (3.19 g, 13.61 mmol) and K2CO3 (3.76 g, 27.22 mmol)
yielded a white solid. Flash chromatography of the crude
product using hexane–AcOEt (7 : 3) as eluent gave 2e (1 g,
50%), mp 185–186 ЊC (from hexane) (Found: C, 74.14; H, 7.43.
C27H32O5 requires C, 74.29; H, 7.39%); νmax(KBr)/cmϪ1 3300
and 1740; δH (300 MHz; CDCl3; Me4Si) 7.67 (2 H, d, J 8.82,
ArH), 7.49 (2 H, d, J 8.82, ArH), 7.15 (2 H, d, J 8.25, ArH),
6.82 (2 H, d, J 8.25, ArH), 3.77 (3 H, s, CH3), 3.66 (1 H, d,
J 14.85, CH2), 3.53 (1 H, d, J 14.85, CH2), 1.36 (9 H, s, CH3)
and 1.24 (9 H, s, CH3); δC (300 MHz; DMSO-d6; Me4Si) 169.52,
169.12, 152.57, 149.96, 138.06, 130.17, 129.53, 127.82, 127.46,
126.54, 125.96, 124.89, 86.04, 53.53, 38.87, 34.84, 34.35, 31.24
and 31.11.
General procedure B
To a solution of ester 1 (5 mmol) in DMF (20 cm3) was added
DBU (2.5 mmol). The reaction mixture was stirred at room
temperature until the reaction was completed (TLC), and then
was acidified (1 M HCl). The resulting precipitate was filtered
and purified by column chromatography or recrystallization.
Methyl
2-benzyl-4-hydroxy-5-oxo-3-phenyl-2H-furan-2-
carboxylate (2a). By method A, 1a (2.0 g, 11.2 mmol) and
K2CO3 (3.01 g, 22.4 mmol) yielded a white solid. Flash chroma-
tography of the crude product using hexane–AcOEt (7 : 3) as
eluent gave 2a (1.027 g, 57%), mp 156–157 ЊC (from EtOH)
(Found: C, 70.75; H, 4.99. C19H16O5 requires C, 70.36; H,
4.97%); νmax(KBr)/cmϪ1 3290 and 1735; δH (300 MHz; CDCl3;
Me4Si) 7.72–7.69 (2 H, m, ArH), 7.49–7.42 (3 H, m, ArH),
7.16–7.10 (3 H, m, ArH), 6.84–6.81 (2 H, m, ArH), 3.79 (3 H, s,
CH3), 3.68 (1 H, d, J 14.30, CH2) and 3.58 (1 H, d, J 14.30,
CH2); δC (300 MHz; CDCl3; Me4Si) 169.38, 169.12, 138.92,
132.49, 130.31, 129.47, 129.23, 128.95, 127.93, 127.70, 127.57,
127.25, 86.01, 53.57 and 39.08.
Methyl 4-hydroxy-3-(4-trifluoromethylphenyl)-2-(4-trifluoro-
methylphenylmethyl)-5-oxo-2H-furan-2-carboxylate (2f ). By
method A, 1f (1.0 g, 4.06 mmol) and K2CO3 (1.12 g, 8.11 mmol)
yielded a white solid. Flash chromatography of the crude
product using CHCl3–EtOH (30 : 1) as eluent gave 2f (0.706 g,
76%), mp 199–200 ЊC (from benzene) (Found: C, 54.97; H, 3.21.
C21H14F6O5 requires C, 54.79; H, 3.06%); νmax(KBr)/cmϪ1 3330,
1770 and 1745; δH (300 MHz; CD3OD; Me4Si) 7.93 (2 H, d,
J 8.55, ArH), 7.79 (2 H, d, J 8.55, ArH), 7.44 (2 H, d, J 8.55,
ArH), 7.05 (2 H, d, J 8.55, ArH), 3.82 (3 H, s, CH3), 3.76 (1 H,
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 8 6 4 – 1 8 7 1
1869

View Article Online
d, J 14.64, CH2) and 3.70 (1 H, d, J 14.64, CH2); δC (300 MHz;
DMSO-d6; Me4Si) 168.84, 166.94, 142–118, 84.57, 53.87 and
38.28.
For the molecular dynamics simulations the same starting
structure was used, however, some preliminary model building
and energy refinement had to be undertaken. The ten missing
amino acids from 1hck crystal structure (Ala 31 through Thr
41) were directly taken from the CDK2–cyclinA–ATP complex
(ref. 1fin) after superimposition of common parts in both
structures. Hydrogens were added using standard geometries
and their positions were optimised using the molecular mech-
anics program AMBER.22 A short optimisation was then
undertaken where only the residues connecting the added
residues were allowed to move (Thr39 through Val 44 and
Glu28 through Lys33) in a continuum medium of relative
permittivity ε = 4rij for imitating the solvent environment.
Finally an optimisation run restraining all non-H atoms to their
initial coordinates allowed readjustment of covalent bonds and
van der Waals contact without changing the overall conform-
ation of the protein, which was considered the starting struc-
ture for the molecular dynamics simulations described below.
Methyl
4-hydroxy-3-(4-methoxyphenyl)-2-(4-methoxy-
phenylmethyl)-5-oxo-2H-furan-2-carboxylate (2g). By method
B, 1g (1.70 g, 8.16 mmol) and DBU (0.62 g, 4.08 mmol) yielded
a white solid. Flash chromatography of the crude product using
hexane–AcOEt (1 : 1) as eluent gave 2g (1.50 g, 96%), mp 183–
185 ЊC (from EtOH) (Found: C, 65.44; H, 5.27. C21H20O7
requires C, 65.61; H, 5.24%); νmax(KBr)/cmϪ1 3280 and 1740;
δH (300 MHz; CDCl3; Me4Si) 7.68 (2 H, d, J 8.80, ArH), 6.99 (2
H, d, J 8.80, ArH), 6.76 (2 H, d, J 8.80, ArH), 6.65 (2 H, d,
J 8.80, ArH), 6.14 (1 H, s, OH), 3.87 (3 H, s, CH3), 3.78 (3 H, s,
CH3), 3.72 (3 H, s, CH3), 3.60 (1 H, d, J 14.85, CH2) and
3.52 (1 H, d, J 14.85, CH2); δC (300 MHz; CDCl3; Me4Si)
169.65, 169.13, 160.10, 158.67, 137.15, 131.38, 129.33, 127.76,
124.53, 122.11, 114.44, 113.37, 85.95, 55.31, 55.04, 53.53 and
38.98.
Docking. All docking studies were performed with the
program AUTODOCK (version 3.0).23 The target in each
docking run was CDK2 obtained as described above. Affinity
grid files were generated using the auxiliary program AUTO-
GRID (version 3.0). The centre of the removed ATP molecule
bound to the active site was chosen as the centre of the grids,
and the dimensions of the grid were 60 × 60 × 60 Å3 with grid
points separated by 0.375 Å. The original Lennard-Jones and
hydrogen bonding potentials provided by the program were
used. The parameters for the docking using the Lamarckian
Genetic Algorithm (LGA) were identical for all docking jobs
and are summarized in Table 5 (supplementary information).
After docking, the 100 solutions were clustered in groups
with RMS deviations lower than 1.0 Å. The clusters were
ranked by the lowest energy representative of each cluster. The
docking experiments were performed on an Octane R12000 and
the average CPU time for each compound was 3 h.
Methyl
4-hydroxy-3-(4-methylphenyl)-2-(4-methylphenyl-
methyl)-5-oxo-2H-furan-2-carboxylate (2h). By method B, 1h
(1.70 g, 8.84 mmol) and DBU (0.67 g, 4.42 mmol) yielded a
white solid. Flash chromatography of the crude product using
hexane–AcOEt (7 : 3) as eluent gave 2h (1.23 g, 79%), mp 200–
202 ЊC (from EtOH–H2O) (Found: C, 71.59; H, 5.83. C21H20O5
requires C, 71.57; H, 5.72%); νmax(KBr)/cmϪ1 3330 and 1740; δH
(300 MHz; CDCl3; Me4Si) 7.61 (2 H, d, J 8.25, ArH), 7.27 (2 H,
d, J 8.25, ArH), 6.92 (2 H, d, J 8.25, ArH), 6.73 (2 H, d, J 8.25,
ArH), 6.51 (1 H, br s, OH), 3.77 (3 H, s, CH3), 3.63 (1 H, d,
J 14.31, CH2), 3.54 (1 H, d, J 14.31, CH2), 2.41 (3 H, s, CH3)
and 2.24 (3 H, s, CH3); δC (300 MHz; CDCl3; Me4Si) 169.55,
169.09, 139.59, 138.03, 136.82, 130.26, 129.74, 129.44, 128.70,
127.78, 127.59, 126.69, 86.01, 53.54, 38.84, 21.45 and 21.03.
Computational details
Molecular modeling and conformational search of butyro-
lactone I and compounds 2. Butyrolactone I and compounds 2
were model built in SYBYL 6.7 using standard geometries.
Conformational searches and energy minimizations were
performed using Macromodel version 5.5.13 The Macromodel
implementation of the AMBER14 all atom force field was used
(denoted AMBER*). All calculations were performed using the
implicit water GB/SA solvation model of Still et al.15 Conform-
ational searches were performed using the Monte Carlo method
of Goodman and Still.16 All esters were required to be cis; those
deviating more than 90Њ being rejected as energetically improb-
able. For each search 1000 starting structures were generated
and minimized to an energy convergence of 0.05 (kJ molϪ1) ÅϪ1
using the Polak–Ribiere conjugate gradient minimization
method implemented in Macromodel. Duplicated structures
and those greater than 50 kJ molϪ1 above the global minimum
were discarded. The lowest energy conformer for each
compound was then fully optimized by means of the ab initio
quantum mechanical program Gaussian 9817 and the 3-21G
basis set. RHF/6-31G*//3-21G RESP charges18 were derived
together with appropriate bonded and nonbonded parameters
consistent with the AMBER force field19 (Tables 3 and 4, in
supporting information). These low energy minimized con-
formers were selected to be used as input for the ligand docking
process.
Energy refinement of the CDK2–ligand complexes. The
selected complexes from the docking experiments were grad-
ually refined in AMBER using a cutoff of 10.0 Å. The receptor
structure was modified including missing amino acids as
described above. The initial complexes were refined by progres-
sively minimizing their potential energy. The optimisations were
carried out in a continuum medium of relative permittivity ε
= 4rij for imitating the solvent environment. The resulting
complexes were considered the initial structures for the molec-
ular dynamics simulations in water.
Molecular dynamics in water. Each molecular system was
placed into a 20 Å in radius spherical cap of ∼340 TIP3P water
molecules centered on the centre of mass of the bound ligand.
The initial complexes were refined by progressively minimizing
their potential energy. 1000 ps unrestrained MD simulations at
298 K and 1 atm were then run for all complexes using the
SANDER module in AMBER. SHAKE24 was applied to all
bonds involving hydrogens and an integration step of 2 fs was
used throughout. The simulation protocol involved a series of
progressive energy minimizations followed by a 10 ps heating
phase and 10 ps equilibration period before data collection.
System coordinates were saved every 2 ps for further analysis.
Analysis of the MD trajectories. Three-dimensional struc-
tures and trajectories were visually inspected using the com-
puter graphics program SYBYL.21 Root-mean-square (rms)
deviations from the initial structures and interatomic distances
were monitored using CARNAL.22 An average structure from
the last 500 ps of the MD simulations in water for all complexes
was refined by means of steepest-descent energy minimizations
and the resulting complexes were used for the energy analysis
using ANAL.
CDK2 model building and energy refinement. For the flexible
ligand docking the CDK2–ATP complex structure was directly
retrieved from the Protein Data Bank (PDB)20 (ref. 1hck). The
ATP molecule, the magnesium ion and water molecules were
manually removed. Polar hydrogens and Kollman united-atom
partial atomic charges were added by means of the SYBYL
program.21
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 8 6 4 – 1 8 7 1
1870

View Article Online
Mohan, S. D. Dhanvantri, M. Sitaramkumar, J. M. Babu, K. Vyas
Acknowledgements
and G. O. Reddy, Chem. Pharm. Bull., 2000, 48, 559–562.
13 F. Mohamadi, N. G. J. Richards, W. C. Guida, R. Liskamp,
M. Lipton, C. Caufield, G. Chang, T. Hendrickson and W. C. Still,
J. Comput. Chem., 1990, 11, 440–467.
14 S. J. Weiner, P. A. Kollman, D. A. Case, U. C. Singh and C. Ghio et
al., J. Am. Chem. Soc., 1984, 106, 765–784.
15 W. C. Still, A. Temczyk, R. C. Hawely and T. Hendrickson, J. Am.
Chem. Soc., 1990, 112, 6127–6129.
16 J. M. Goodman and W. C. Still, J. Comput. Chem., 1991, 12,
1110–1117.
17 Gaussian 98 (Revision A.7), M. J. Frisch, G. W. Trucks,
H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman,
V. G. Zakrzewski, J. A. Montgomery, R. E. Stratmann, J. C. Burant,
S. Dapprich, J. M. Millam, A. D. Daniels, K. N. Kudin,
M. C. Strain, O. Farkas, J. Tomasi, V. Barone, M. Cossi, R. Cammi,
B. Mennucci, C. Pomelli, C. Adamo, S. Clifford, J. Ochterski,
G. A. Petersson, P. Y. Ayala, Q. Cui, K. Morokuma, D. K. Malick,
A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. Cioslowski,
J. V. Ortiz, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz,
I. Komaromi, R. Gomperts, R. L. Martin, D. J. Fox, T. Keith,
M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, C. Gonzalez,
M. Challacombe, P. M. W. Gill, B. G. Johnson, W. Chen,
M. W. Wong, J. L. Andres, M. Head-Gordon, E. S. Replogle and
J. A. Pople, Gaussian, Inc., Pittsburgh, PA, 1998.
We thank Prof. Laurent Meijer (C.N.R.S., Station Biologique,
Roscoff, France) for biological activity determinations and
Dr Kensuke Nakamura (Nara Institute of Science and Tech-
nology, Nara, Japan) for his help in dealing with Japanese
patents. We are grateful to A. Olson for providing
AUTODOCK. Part of the molecular modeling included in this
work was awarded a Normon Prize, given by the Real
Academia de Farmacia (Spain). This research received financial
support from Universidad San Pablo-CEU (Grant 4/01 U.S.P.).
We thank the CIEMAT (Spain) for computer time and
facilities.
References
1 A. Huwe, R. Mazitschek and A. Giannis, Angew. Chem., Int. Ed.,
2003, 42, 2122–2138.
2 N. Kiriyama, K. Nitta, Y. Sakaguchi, Y. Taguchi and Y. Yamamoto,
Chem. Pharm. Bull., 1977, 25, 2593–2601.
3 K. Nishio, T. Ishida, H. Arioka, H. Kurokawa, K. Fukuoka,
T. Nomoto, H. Fukumoto, H. Yokote and N. Saijo, Anticancer Res.,
1996, 16, 3387–3396.
4 M. Suzuki, Y. Hosaka, H. Matsushima, T. Goto, T. Kitamura and
K. Kawabe, Cancer Lett., 1999, 138, 121–130.
5 M. Kitagawa, T. Okabe, H. Ogino, H. Matsumoto, I. Suzuki-
Takahashi, T. Kokubo, H. Higashi, S. Saitoh and Y. Taya, Oncogene,
1993, 8, 2425–2432.
6 H. Morishima, K. Fujita, M. Nakano, S. Atsumi, M. Ookubo,
M. Kitagawa, H. Matsumoto, A. Okuyama and T. Okabe, (Banyu
Pharma Co. Ltd.), Jpn. Kokai Tokkyo Koho 06 100, 445, 1994.
7 K. Nitta, N. Fujita, T. Yoshimura, K. Arai and Y. Yamamoto,
Chem. Pharm. Bull., 1983, 31, 1528–1533.
8 D. Cagniant, Ann. Chim., 1952, 442–457.
9 H. N. C. Wong, Z. L. Xu, H. M. Chang and C. M. A Lee, Synthesis,
1992, 793–797.
10 P. A. Sims, C. F. Wong and J. A. McCammon, J. Med. Chem., 2003,
46, 3314–3325.
11 S. J. Teague, Nat. Rev. Drug Discovery, 2003, 2, 527–541.
12 K. V. Rao, A. K. Sadhukhan, M. Veerender, V. Ravikumar, E. V. S.
18 C. I. Bayly, P. Cieplack and P. A. Kollman, J. Phys. Chem., 1993, 97,
10269–10280.
19 W. D. Cornell, P. Cieplack, C. I. Bayly, I. R. Gould, K. M. Merz,
D. M. Ferguson, D. C. Spellmeyer, T. Fox, J. W. Caldwell and
P. A. Kollman, J. Am. Chem. Soc., 1995, 117, 5179–5197.
20 H. M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T. N. Bhat,
H. Weissig, I. N. Shindyalov and P. E. Bourne, Nucleic Acids Res.,
2000, 28, 235–242.
21 SYBYL®; 6.7. ed., Tripos Inc., 1699 South Hanley Road, St. Louis,
Missouri 63144; http://www.tripos.com.
22 D. A. Case, D. A. Pearlman, J. W. Caldwell, T. E. Cheatham III,
W. S. Ross, C. L. Simmerling, T. A. Darden, K. M. Merz,
R. V. Stanton, A. L. Cheng, J. J. Vicent, M. Crowly, D. M. Ferguson,
R. J. Radmer, G. L. Seibel, U. C. Singh, P. K. Weiner and
P. A. Kollman, AMBER 5 : University of California, San Francisco.
23 G. M. Morris, J. Comput. Chem., 1998, 19, 1639–1662.
24 J.-P. Ryckaert, G. Ciccotti and H. J. C. Berendsen, J. Comput. Phys.,
1977, 23, 327–341.
O r g . B i o m o l . C h e m . , 2 0 0 4 , 2, 1 8 6 4 – 1 8 7 1
1871