Synthesis, biological evaluation, and molecular modeling of natural and unnatural flavonoidal alkaloids, inhibitors of kinases

Article abstract of DOI:10.1021/jm201727w

The screening of the ICSN chemical library on various disease-relevant protein kinases led to the identification of natural flavonoidal alkaloids of unknown configuration as potent inhibitors of the CDK1 and CDK5 kinases. We thus developed an efficient and modular synthetic strategy for their preparation and that of analogues in order to determine the absolute configuration of the active natural flavonoidal alkaloids and to provide further insights on the structure-activity relationships in this series. The structural determinants of the interaction between some flavonoidal alkaloids with specific kinases were also evaluated using molecular modeling.

Full text of DOI:10.1021/jm201727w

Article
pubs.acs.org/jmc
Synthesis, Biological Evaluation, and Molecular Modeling of Natural
and Unnatural Flavonoidal Alkaloids, Inhibitors of Kinases
Thanh Binh Nguyen,† Olivier Lozach,‡ Georgiana Surpateanu,† Qian Wang,§ Pascal Retailleau,†
Bogdan I. Iorga,† Laurent Meijer,*,∥,‡ and Francoise Gueritte*,†
̧
́
†Centre de Recherche de Gif, Institut de Chimie des Substances Naturelles (ICSN), CNRS, 91198 Gif-sur-Yvette Cedex, France
‡“Protein Phosphorylation & Human Disease”, CNRS, Station Biologique, Place G. Teissier, 29682 Roscoff Cedex, France
§Institute of Chemical Sciences and Engineering, Ecole Polytechnique Fed
́ ́
erale de Lausanne, EPFL-SB-ISIC-LSPN, 1015 Lausanne,
Switzerland
∥ManRos Therapeutics, Centre de Perharidy, 29680 Roscoff, France
S
* Supporting Information
ABSTRACT: The screening of the ICSN chemical library on
various disease-relevant protein kinases led to the identification
of natural flavonoidal alkaloids of unknown configuration as
potent inhibitors of the CDK1 and CDK5 kinases. We thus
developed an efficient and modular synthetic strategy for their
preparation and that of analogues in order to determine the
absolute configuration of the active natural flavonoidal
alkaloids and to provide further insights on the structure−
activity relationships in this series. The structural determinants of the interaction between some flavonoidal alkaloids with specific
kinases were also evaluated using molecular modeling.
1. INTRODUCTION
Cyclin-dependent kinases (CDKs),1 glycogen synthase kinase-3
(GSK3),2 dual specificity tyrosine-phosphorylation-regulated
kinase 1A (DYRK1A),3 and cdc2-like kinase CLK14 are serine−
threonine kinases that attract considerable interest because of
their involvement in many essential physiological processes and
multiple human diseases such as cancer and neurodegenerative
diseases.1 Today, around 250 small molecule kinase inhibitors
are undergoing clinical trials for the treatment of cancer and
other diseases, showing that kinases constitute important
therapeutic targets. Our interest in the synthesis of kinase
inhibitors comes from the screening of the ICSN−CNRS
chemical library, comprising about 4000 natural and synthetic
products. The screening on five kinases, CDK1, CDK5, GSK3,
CLK1, and DYRK1A, led to the selection of two natural
flavonoidal alkaloids which possess a good inhibitory activity on
the cyclin-dependent kinases CDK1/cyclin B and CDK5/p25
by ATP competition: O-demethylbuchenavianine (1)5,6 and
N,O-didemethylbuchenavianine (2)5,6 (Figure 1) found in the
fruits of Buchenavia macrophylla and in the leaves of Buchenavia
capitata.
Figure 1. Natural flavonoids (1−6) and synthetic flavonoidal alkaloids
(7−10).
interaction with these kinases. It should be noted that the
flavonoidal alkaloids 1−5 were optically active, but their
absolute configuration was unknown at this time. The structure
of these molecules is reminiscent of flavopiridol (7), the first
synthetic cCDK inhibitor to enter clinical trials as an anticancer
drug.8 Similar to flavopiridol, a new synthetic flavonoidal
alkaloid (P276−00) 8 enters clinical studies as a small-molecule
CDK inhibitor9 (Figure 1). Since the discovery of the
Capitavine (3),5 isolated from the seeds of B. capitata, N-
demethylbuchenavianine (4),5 and buchenavianine (5)5,6 as
well as chrysin (6)7 (Figure 1), also present in the ICSN
chemical library, were found less active on the five kinases (IC50
> 10 μM) (Table 1). These preliminary results showed that (1)
the 8-substituted regioisomer 1 was more active than its 6-
substituted analogue 3 and (2) the amino moiety and the free
hydroxyl group at C-5 of flavonoidal alkaloids were essential for
Received: December 22, 2011
Published: February 21, 2012
© 2012 American Chemical Society
2811
dx.doi.org/10.1021/jm201727w | J. Med. Chem. 2012, 55, 2811−2819

Journal of Medicinal Chemistry
Article
substituted adduct increased when the reaction was performed
at elevated temperature.
Table 1. Synthesis of 6- and 8-Substituted Flavonoidal
Alkaloids from 6
Having an important quantity of 3 in hand, we were looking
for a convenient alternative approach to its 8-substituted
regioisomer 1. The Wessely−Moser rearrangement15 has been
known as an important method in structure elucidation and
synthesis of flavonoids, involving the interconversion between
8- and 6-substituted flavone under acidic conditions.16 Thus,
heating neat triflate salt of capitavin at 180 °C afforded cleanly a
mixture of both regioisomers (6-sub/8-sub = 3:1) without
noticeable degradation. In that way, compound 1 was obtained
from capitavin with a 22% yield. We then prepared adducts
derived from chrysin, hexanal, and methylamine (Scheme 2).
The 6-substituted aminoalkylflavone 17 was obtained predom-
inantly (78%) when 6 was reacted with hexanal (2 equiv) and
methylamine (6 equiv). When methylamine was used in excess
(20 equiv), the ratio of 17 to the 8-substituted regioisomer 18
decreases to 2:1 (Scheme 2). In all cases, both regioisomers
could be separated efficiently by flash column chromatography
on silica gel.
entry
conditions
6-sub/8-sub
yield (%)
1a
1b
2a
2b
3a
3b
4a
4b
40 °C, H2O/THF
110 °C, H2O/THF
40 °C, H2O
80 °C, H2O/THF
40 °C, H2O/THF
120 °C, H2O/THF
40 °C, H2O/THF
110 °C, H2O/THF
100/0
5/1
95 (13)
nd (13)/nd (16)
80 (12)
100/0
5/1
13 (12)/7 (15)
87 (14)
100/0
5/1
11 (2)/nd (14)
78 (3)
100/0
24/1
nd (3)/ nd (1)
antitumor activity of flavopiridol, studies concerning the
structure−activity relationships have been realized on nitro-
gen-containing flavonoids active on kinases. For example, the
piperidinylmethyl analogues of flavopiridol 910 as well as that of
baicalein 1011 (Figure 1) are good inhibitors of CDK1/cyclin
B. These studies have shown that the position of the piperidine
ring as well as the position of the atom nitrogen play critical
role in the kinase inhibitory activity.
In the present study, we report the synthesis of natural and
non-natural flavonoidal alkaloids as well as the absolute
configuration of the active natural flavonoidal alkaloids. We
also provide further insights on the structure−activity relation-
ships in this series as well as the structural determinants of the
interaction between these flavonoids with specific kinases
identified using molecular modeling.
We focused next our efforts on the enantiomeric separation
of the three most active racemic 8-substituted adducts: 1, 15,
and 18. These separations, which were realized by SFC HPLC
using IC column, afforded the desired enantiomers in
satisfactory enantiomerical purities. Absolute configuration of
a single enantiomer was determined unambiguously by X-ray
crystallography. Thus, (+)-1 and (+)-15 have the (R)-
configuration, whereas (−)-18 was (S) (see Supporting
Information). Natural 1 has a negative optical rotation
2. SYNTHESIS AND KINASE INHIBITORY ACTIVITY
23
([α]D −21 (c, 0.13, CHCl3), and its HPLC analysis shows
2.1. Synthesis of 6- and 8-Substituted Flavonoidal
Alkaloids. Modifications were carried out by varying the
substitution position of the alkaloid moiety, the ring size, ring
vs open chain structure, and alkylation degree of the nitrogen
atom. We recently reported the phenolic Mannich reaction
between 6 and the bench-stable iminic compound 11a or
iminium precursor 11b as a method of choice for a one-step,
straightforward access to the 6-substituted 3,5 14,5 12,12,13 and
13 or 8-substituted 1612,13 and 25,6 flavonoidal alkaloids
(Scheme 1).14 We also describe here the synthesis of ficine
an ee of 20% enriched in (S) isomer (more retained). We thus
assumed that during the isolation process, which was carried
out in highly basic conditions (aqueous ammonia solution),
partial racemization occurred.
2.2. Biological Evaluation of Flavonoidal Alkaloids.
The compounds were evaluated for their potential inhibitory
effects on purified kinases (CDK1, CDK5, GSK3, CLK1, and
DYRK1A) as described in the Experimental Section. The 6-
substituted flavones 3, 12, 13, 14, and 17 were found to much
more less active (IC50 ≫ 10 μM) than their 8-substituted
regioisomers 1, 15, 16, 2, and 18, respectively. Thus, Table 2
summarizes the biological evaluation on the five kinases of the
8-substituted isomers together with that of each enantiomers of
1, ficine 15, and 18. The IC50 values were determined from
dose−response curves and compared to those of the reference
molecule, flavopiridol.
Scheme 1. Synthetic Route to compounds 1−3 and 12−1614
Like for flavopiridol, the cyclin-dependent kinases CDK1 and
CDK5 are more sensitive to the flavonoidal alkaloids than
GSK3 and CLK1. One important point is that the N-
methylation of the amino moiety is essential for a better
interaction with kinases. Indeed, 1 and 15 are 50 and 30 times
more active on CDK1 than their N-demethylated analogues 2
and 16, and more active also on the other kinases CDK5,
GSK3, and CLK1. Additionally, a free hydroxyl group at the C-
5 position is essential for a good interaction with kinases
because the IC50 of 5 was superior to 10 μM. These
observations are in full agreement with the interactions revealed
by molecular docking calculations (see below). It should be
noted that 1 and 15 have a CDK1/cyclin B inhibitory activity
similar to that of bacalein analogues containing a nitrogen ring
at C-8 and a free hydroxyl group at C-7, such as 10.11 Thanks
to the separation of each enantiomer of the racemic mixtures of
1 and 15 and X-ray analysis of one of the enantiomers, we
15 and isoficine 12 under the same conditions. Thus, the
reaction performed in aqueous medium at 40 °C, in the mixture
of H2O−THF, was highly regioselective and afforded the 6-
substituted adduct as major or unique product in excellent
yields for all iminic starting materials (Scheme 1 and Table
1).14 As shown in Table 1 (entries 1b−4b), the yield of 8-
2812
dx.doi.org/10.1021/jm201727w | J. Med. Chem. 2012, 55, 2811−2819

Journal of Medicinal Chemistry
Article
a
Scheme 2. Synthesis of 6- and 8-Substituted Aminoalkylflavones
a
Reaction conditions: H2O−THF, 40 °C, 5 h
Table 2. Kinases Activity of 8-Substituted Flavonoidal
Alkaloids
Information). These structures present well conserved geo-
metries, with backbone root-mean-square deviations (rmsd) of
2.05, 2.33, 1.98, and 2.63 Å for GSK3, CLK1, DYRK1A, and
CDK1, respectively, compared with CDK5.
kinases inhibitory activities (IC50 in μM)
compd
CDK1
CDK5
GSK3
CLK1
DYRK1A
The overall sequence identity between these kinases
sequences ranges from 32% to 57%, with values of 30−76%
for the binding site region (defined as a sphere with a 6.0 Å
radius around the aspartate residue from the DFG motif). A
notable exception is DYRK1A, which shows sensibly lower
values, 12−15% and 31−34%, respectively. Selected residues
from the ATP binding site presenting amino acid variability are
shown in Table S1 of the Supporting Information). Some of
these residues play an important role in the ligand selectivity
determined experimentally.
From the ligand side, the crystal structure of a synthetic
flavonoid analogue, flavopiridol (7, Figure 1), in complex with
CDK9/CycT1 (PDB entry 3BLR), was recently reported.21
This ligand binds to the ATP-binding pocket of CDK9 and has
a remarkable structural similarity with the compounds studied
in this work. The overall sequence identity of CDK9 with
CDK1, CDK5 and GSK3 is about 40−46% and only 22% with
CLK1, whereas for the binding site region it increases to 53−
68% and 30%, respectively. Therefore, all docking results
obtained in this study were critically compared with the
flavopiridol/CDK9 complex, which can be considered as
representative for the interaction of kinases with flavonoid
compounds.
The docking procedure was first evaluated by the direct
docking of flavopiridol into the native CDK9 structure and by
cross-docking with the other kinase structures included in this
study. The protein−ligand complex was correctly reproduced in
direct docking (rmsd 0.6 Å) and the cross-docking provided
similar conformations, thus validating the procedure that will be
further used for the docking of new compounds (Table 2 and
Figure S3 in the Supporting Information). In the following
paragraphs, only the interactions between the active com-
pounds (1, 2, 15, 16, 18) and the corresponding proteins are
discussed. The inactive compounds generally show unspecific
binding modes inside or outside the ATP binding pocket. A
particular example is DYRK1A, for which docking calculations
could evidence no specific interactions with any of the ligands
included in this work, in agreement with the experimental data.
However, 7 is a notable exception to this rule, being the only
member of this class of compounds showing a biological
activity against DYRK1A. A possible explanation for this
behavior is the presence of a Val172 residue in DYRK1A just
before the DFG motif, whereas in the equivalent positions of
CDK1 and CDK5, the smaller residues Ala145 and Ala143 are
present, respectively. As a consequence, in the latter case this
region of the binding site can accommodate the substituents
(e.g., NMe in 1 or 15) surrounding the protonated amino
group, while the Val172 would induce steric clashes with these
1
0.09
1.1
1.7
0.95
0.05
5.5
5.1
≥10
1.6
8.5
>10
2.1
>10
>10
>10
>10
>10
>10
>10
>10
>10
>10
>10
>10
>10
0.3
(R)-(+)-1
(S)-(−)-1
( )-2
0.03
5
>10
>10
>10
1.3
>10
>10
>10
1.6
4
>10
>10
0.05
0.39
0.04
1.6
>10
>10
0.09
0.57
0.04
1.1
5
( )-15
(R)-(+)-15
(S)-(−)-15
( )-16
( )-18
(R)-(+)-18
(S)-(−)-18
7
2.6
7.4
0.72
>10
5.2
0.67
>10
>10
>10
>10
1.3
1.1
0.52
0.27
2.1
0.52
5.4
4.1
>10
0.36
0.09
0.06
could assign the S configuration to the most active enantiomers
of 1 and 15. If racemic 1 and ficine 15 turned out to inhibit
CDK1 and CDK5 as potently as flavopiridol, their (S)-
enantiomer was slightly more active than flavopiridol, inhibiting
CDK1 and CDK5 with IC50 value around 0.04 μM. These
flavonoidal alkaloids were generally less active on the GSK3 and
CLK1 kinases. Only (S)-1 retains a good activity, with an IC50
similar to that of flavopiridol on these two kinases. Moreover,
the flavonoidal alkaloids are inactive toward DYRK1A at 10
μM, while flavopiridol, the reference compound, showed a
micromolar inhibition of this enzyme. Compound 18 with an
opened amino side chain at C-8 is also less active than
compounds 1 and 15 possessing a nitrogen-containing
heterocycle at that position. After separation of the racemic
mixture 18 and X-ray analysis of one of the enantiomers, we
could assign the (R) configuration to the most active
enantiomer, which is contradictory to the results obtained for
ficine and (O)-demethylbuchenavianine.
To better understand the interaction pattern at the molecular
level between the flavonoidal alkaloids and these kinases,
molecular modeling calculations were carried out and the
results compared with the experimental data.
3. MOLECULAR MODELING
Three-dimensional crystal structures are available in the Protein
Data Bank (PDB)17 for four proteins included in this study:
CDK5 (entry 1UNL),18 GSK3 (entry 1J1B),19 CLK1 (entry
1Z57), and DYRK1A (entry 2VX3). As no structure was
available for CDK1, a homology model was built using the
DFG-in structure of CDK2 (entry 1YKR,20 sharing 65%
sequence identity with CDK1) as template and the sequence
alignment presented in Figure S2 of the Supporting
2813
dx.doi.org/10.1021/jm201727w | J. Med. Chem. 2012, 55, 2811−2819

Journal of Medicinal Chemistry
Article
Figure 2. Docking poses for complexes of CDK1 with compounds 1 (a), 2 (b), 15 (c), 16 (d), and 18 (e).
substituents, which can explain the absence of biological activity
observed experimentally for the compounds included in this
study against DYRK1A (Table 2). On the contrary, 7 has the
NMe group in a different position on the ring and thus the
clashes with Val172 are absent.
3.1. Docking with CDK1 and CDK5. Compounds 1, 2, 15,
16, and 18 were identified to interact favorably with the CDK1
and CDK5 binding sites (Figures 2 and 3). These two binding
sites are very similar, the only differences being Leu83/Cys83
and Ser84/Asp84 in CDK1/CDK5, respectively. The mutation
D144N in the CDK5 crystallographic structure should also be
noted. However, all these differences do not influence the main
binding pattern and these five compounds (1, 2, 15, 16, and
18) interact in a similar manner with CDK1 and CDK5. The
phenyl substituent points out toward the outside of the binding
pocket, and the flavone ring system establishes hydrogen bonds
with the NH backbone group of Leu83/Cys83 via the
chromone oxygen and with Asp146/Asn144 via the C-7
hydroxyl group. In CDK1, this hydroxyl group also interacts
with the side chain of Lys33. Another conserved feature in this
system is the interaction between the protonated amino group
of the ligand with the backbone carbonyl group of Gln132/
2814
dx.doi.org/10.1021/jm201727w | J. Med. Chem. 2012, 55, 2811−2819

Journal of Medicinal Chemistry
Article
Figure 3. Docking poses for complexes of CDK5 with compounds 1 (a), 2 (b), 15 (c), 16 (d), and 18 (e).
Gln130. Both configurations were considered in the docking
process. For compound 18, the isomer with R configuration
was found to establish better interactions with the active site
residues, whereas for compounds 1, 2, 15, and 16, the S isomer
was predicted to be more potent. These results are in
agreement with the biological activities and X-ray diffraction
data (Table 2).
3.2. Docking with GSK3. Molecular docking calculations
showed that GSK3 is able to interact favorably only with
compounds 1, 15, and 18. The binding modes are generally
2815
dx.doi.org/10.1021/jm201727w | J. Med. Chem. 2012, 55, 2811−2819

Journal of Medicinal Chemistry
Article
Figure 4. Docking poses for complexes of GSK3 with compounds 1 (a), 15 (b), and 18 (c).
Figure 5. Docking poses for complexes of CLK1 with compounds 1 (a) and 15 (b).
similar with those evidenced in the case of CDK1 and CDK5,
with the phenyl substituent pointing out toward the outside of
the active site, but with the notable difference that the ligand is
buried more deeply into the ATP binding pocket. This is
mainly due to the presence of Arg 141 (instead of Lys in CDK1
and CDK5, smaller and more flexible) and to a different
conformation of the loop Val 61−Tyr 71 in GSK3. The
consequence of these changes in the binding site conformation
is that the ligands (compounds 1, 15, and 18) do not have
anymore optimal interactions with neighboring residues and
2816
dx.doi.org/10.1021/jm201727w | J. Med. Chem. 2012, 55, 2811−2819

Journal of Medicinal Chemistry
Article
(Whatman) using a FilterMate harvester (Packard), and papers were
washed in 1% phosphoric acid. Scintillation fluid was added and the
radioactivity measured in a Packard counter.
GSK-3α/β (porcine brain, native) was purified (Primot et al.,
2000)23 and assayed, as described for CDK1 but using a GSK-3
specific substrate (GS-1: YRRAAVPPSPSLSRHSSPHQSpEDEEE)
(pS stands for phosphorylated serine), synthesized by Proteogenix
(Oberhausbergen, France).
DYRK1A (rat, recombinant, expressed in Escherichia coli as a GST
fusion protein) was purified by affinity chromatography on
glutathione−agarose and assayed as described for CDK1/cyclin B
using Woodtide (KKISGRLSPIMTEQ) (1 μg/assay) as a substrate.
CLK1 (mouse, recombinant, expressed in E. coli as GST fusion
proteins) was assayed in buffer A with RS peptide (GRSRSRSRSRSR)
(1 μg/assay).
only a limited number of hydrogen bonds is observed either
with the DFG or the hinge residues (Figure 4). Again, the R
isomer is predicted to be more active for compound 18 and the
S isomer in the case of compounds 1 and 15. These results are
also in agreement with the kinase activity (Table 2).
3.3. Docking with CLK1. Compared with the previous
three proteins, CLK1 presents similar features within the
binding pocket. However, the solvent-exposed surface is slightly
different, which is mainly due to a conformation of the loop
Thr 166−Val 176 similar to those observed for GSK3, and to
the presence of Asp 250 (instead of Arg or Lys) and Leu 243
(instead of Phe or Tyr). These changes in the active site
conformations allow the interaction with only a few ligands
(compounds 1 and 15) and only through the DFG loop of the
binding pocket (Figure 5). The general orientation of the
ligand is similar to the one observed previously, with the phenyl
substituent toward the outside of the binding pocket.
Additionally, the S isomer of compounds 1 and 15 was
predicted to establish more favorable interactions with the
active site residues than the R isomer. Again, this is in line with
the activity of flavonoids 1−18 found for kinase CLK1 (Table
2).
5.2. Chemistry. Reagents were obtained from commercial
suppliers and used without further purification unless otherwise
noted. Analytical thin layer chromatography (TLC) was purchased
from Merck KGaA (silica gel 60 F254). Visualization of the
chromatogram was performed by UV light (254 nm) or
phosphomolybdic acid and vanilline stains. Flash column chromatog-
raphy was carried out using Kieselgel 35−70 μm particle sized silica gel
(230−400 mesh).
1H chemical shifts are reported in ppm (δ) relative to
tetramethylsilane (TMS) with the solvent resonance employed as
the internal standard (CDCl3, δ 7.24 ppm). Data are reported as
follows: chemical shift, multiplicity (s = singlet, d = doublet, t = triplet,
q = quartet, m = multiplet), coupling constants (Hz), and integration.
13C chemical shifts are reported in ppm (δ) from TMS, with the
4. CONCLUSION
The screening of the ICSN−CNRS chemical library has
resulted in the selection of two natural flavonoidal alkaloids,
1 and 2, which possess a good inhibitory activity on the cyclin-
dependent kinases CDK1/cyclin B and CDK5/p25. The results
of this screening showed also that the substitution of chrysin by
piperidine or pyrrolidine was to be located at C-8 and not C-6
to get active flavonoidal alkaloids. This led us to develop
syntheses for an easy access to racemic natural and unnatural
flavonoidal alkaloids, in order to gain more insight into the
structure−activity relationships in this series. Thank to SFC
chromatography and X-ray crystallography, each enantiomer of
the three racemic mixtures could be separated for biological
evaluation. Thus, it was shown that 1 and 15 bearing a
piperidine and pyrrolidine ring, respectively, were most active
on kinases CDK1, CDK5, GSK3, and CLK1, in their S-
configuration. On the contrary, the most active enantiomer, 18,
with an opened amino side chain, has the R configuration. If 1
and 18 displayed potent inhibition of CDK1 and CDK5, they
are less active on GSK3 and CLK1 and inactive on DYRK1A at
10 μM. Only 15 is as active as flavopiridol on the four kinases
CDK1, CDK5, GSK3, and CLK1 but inactive on DYRK1A at
10 μM. The differences in activity of these flavonoidal alkaloids
could be explained by molecular docking studies of each
enantiomer at the active site of each kinase.
solvent resonance as the internal standard (CDCl3, δ 77.0 ppm).
Mass spectra were determined on an AEI MS-9 using electron spray
ionization (ESI). Piperideine24 and pyrroline25 were synthesized by
the known methods. Compounds 2, 3, and 12−14 were prepared
according to the procedure published by our group.14
Analytical and preparative high performance liquid chromatography
(HPLC) was performed on TharSFC superchrom equipped with
diode array UV detectors, using IC (4.6 mm × 250 mm) column.
Optical rotation measurement was performed on a MCP 300/500
Anton Paar digital polarimeter.
CCDC 846440 (1), CCDC 846439 (15), and CCDC 846441
(compound 18) contain the supplementary crystallographic data for
this paper. These data can be obtained free of charge from The
Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/
data_request/cif.
5,7-Dihydroxy-8-(1-methylpiperidin-2-yl)-2-phenyl-4H-chromen-
4-one (( )-1). Triflic acid (250 μL, 426 mg, 2.84 mmol) was added
dropwise to a solution of 3 (1.00 g, 2.84 mmol) in CH2Cl2 (4 mL) at 0
°C. The resulting solution was concentrated in vacuo. The yellow salt
was then heated under agon at 180 °C for 16 h. The reaction mixture
cooled to rt was treated cautiously with a saturated aqueous solution of
NaHCO3 (5 mL) and AcOEt (20 mL) with vigorous stirring. The
separated aqueous layer was extracted with AcOEt (2 × 10 mL). The
combined organic layers were dried (Na2SO4) and concentrated. The
residue (1/3 = 1:3 by 1H NMR) was purified by flash column
chromatography (AcOEt/MeOH 10:0 to 95:5) to afford F1 ( )-1
(217 mg, 22%) and F2 3 (652 mg, 65%) as yellow crystals. Data for
5. EXPERIMENTAL SECTION
1
( )-1: H NMR (300 MHz, CDCl3) δ 13.10 (s, 1H), 7.87−7.84 (m,
5.1. Kinase Preparations and Assays. Kinase activities were
assayed in buffer A (10 mM MgCl2, 1 mM EGTA, 1 mM DTT, 25
mM Tris-HCl pH 7.5, 50 μg heparin/mL, supplemented extempora-
neously with 0.15 mg BSA/mL), at 30 °C, at a final ATP
concentration of 15 μM. Blank values were subtracted and activities
expressed in % of the maximal activity, i.e., in the absence of inhibitors.
Controls experiments were performed with appropriate dilutions of
DMSO.
2H), 7.51−7.46 (m, 3H), 6.60 (s, 1H), 6.46 (s, 1H), 3.80 (dd, J = 11.4,
3.7 Hz, 1H), 3.24−3.16 (m, 1H), 2.30 (s, 3H), 2.30−2.11 (m, 1H),
1.95−1.61 (m, 5H), 1.49−1.33 (m, 1H). 13C NMR (75 MHz, CDCl3)
δ 182.6, 165.5, 163.7, 158.5, 157.3, 131.8, 129.2, 126.4, 110.3, 105.7,
104.4, 100.6, 94.9, 61.1, 55.0, 43.8, 30.9, 25.6, 23.9. HRMS: m/z [M +
H]+ calcd for C21H22NO4, 352.1549; found, 352.1550. F1 was next
subjected to SFC HPLC separation using, column IC, 30% cosolvents
(iPrOH/MeOH/diethylamine 50:50:1), flow rate 4 mL/min; tR1
=
CDK1/cyclin B (M phase starfish oocytes, native) and CDK5/p25
(human, recombinant) were prepared as previously described.22 Their
kinase activity was assayed in buffer A, with 25 μg of histone H1, in the
presence of 15 μM [γ-33P] ATP (3,000 Ci/mmol; 10 mCi/ml) in a
final volume of 30 μL. After 30 min incubation at 30 °C, the reaction
was stopped by harvesting onto P81 phosphocellulose papers
10.1 min, 96 mg, ee > 99%, [α]D23 +122 (c, 0.1, CHCl3), configuration
R (determined by X-ray diffraction); tR2 = 17.7 min (102 mg, ee = 96%,
23
[α]D −104 (c, 0.1, CHCl3) configuration S. Natural product 1 has a
negative optical rotation ([α]D23 −21 (c, 0.13, CHCl3), and its HPLC
analysis shows an ee of 20% enriched in (S) isomer (more retained).
2817
dx.doi.org/10.1021/jm201727w | J. Med. Chem. 2012, 55, 2811−2819

Journal of Medicinal Chemistry
Article
5,7-Dihydroxy-8-(1-methylpyrrolidin-2-yl)-2-phenyl-4H-chro-
men-4-one (( )-15). A mixture of chrysin (2.03 g, 8 mmol) and
freshly prepared 2-hydroxy-1-methylpyrrolidine (1.21 g, 12 mmol) in
water (4 mL) and THF (4 mL) was heated at 80 °C in a sealed tube
for 5 h and then cooled to rt and diluted with CH2Cl2 (30 mL) and
H2O (10 mL). The mixture was filtered to remove trace of insoluble
chrysin. The separated aqueous layer of the filtrate was extracted with
CH2Cl2 (2 × 10 mL). The combined organic layers were dried
molecular docking step. For both PDB structures and homology
models, the hydrogen atoms were added and energy minimized with
OPLS2005 force field (standard parameters from Schrodinger’s
̈
Protein Preparation Wizard Workflow), keeping all heavy atoms fixed.
Three-dimensional structures of ligands were generated using
CORINA28 v3.44, and then all combinations of stereoisomers and
protonation states at pH 7.0 2.0 were calculated with the LigPrep
module from the Schrodinger Suite.29 Whenever the absolute
configuration of a chiral center in the ligand structure was unknown,
both isomers were considered in the docking process. The amino
groups at C-8 in all flavonoidal compounds were used in the
protonated form throughout this work.
1
(Na2SO4) and concentrated. The residue (15/12 = 1:5 by H NMR)
was purified by flash column chromatography (AcOEt/MeOH 10:0 to
95:5) to afford F1 ( )-15 (347 mg, 13%) and F2 (12) (1923 mg,
71%) as yellow crystals. Data for ( )-15: 1H NMR (300 MHz,
CDCl3) δ 12.65 (s, 1H), 7.81−7.78 (m, 2H), 7.54−7.24 (m, 3H), 6.62
(s, 1H), 6.26 (s, 1H), 4.13 (t, J = 8.5 Hz, 1H), 3.41−3.34 (m, 1H),
2.56−2.37 (m, 2H), 2.43 (s, 3H), 2.05−1.86 (m, 3H). 13C NMR (75
MHz, CDCl3) δ 182.5, 166.0, 163.1, 161.5, 155.1, 132.0, 131.9, 129.4,
126.2, 106.1, 104.7, 102.6, 100.9, 64.7, 55.7, 40.7, 32.9, 23.4. HRMS:
m/z [M + H]+ calcd for C20H20NO4, 338.1392; found, 338.1381. Then
200 mg of F1 was next subjected to SFC HPLC separation using
column IC, 50% cosolvents (iPrOH/diethylamine 99:1), flow rate 4
mL/min; tR1 = 5.8 min, 78 mg, ee > 99%, [α]D23 −160 (c, 0.1, CHCl3),
configuration S; tR2 = 6.9 min, 85 mg, ee = 92%, [α]D23 +144.5 (c, 0.11,
CHCl3), configuration R (determined by X-ray diffraction).
Molecular docking calculations were performed using GOLD 5.0
software30 with both GoldScore and Chemscore_Kinase scoring
functions. After 200 genetic algorithm (GA) runs with 100000 GA
operations per docking, clusters were created at 2.0 Å rmsd. Ligand
flexibility was allowed to explore ring conformations and flip ring
corners. After docking, the complex was minimized using Schro-
̈
dinger’s Maestro framework31 taking into account protein residues
within 8.0 Å from the ligand. Visual inspection of the resulting clusters
showed that in most cases the lowest energy cluster presented the
correct positioning in the binding site, with a conformation similar to
the one adopted by flavopiridol in the CDK9 crystal structure (see
Figure S3, Supporting Information). In the other few instances, this
correct positioning related to flavopiridol was found in the second
cluster (and once in the third cluster). In these latter cases, the
differences between the score of the selected complex and the score of
the lowest energy complex were not significant (less than 10%).
Images of the protein−ligand complexes were rendered using
PyMol version 0.99.32
5,7-Dihydroxy-8-(1-(methylamino)hexyl)-2-phenyl-4H-chromen-
4-one (( )-18) and 5,7-Dihydroxy-6-(1-(methylamino)hexyl)-2-
phenyl-4H-chromen-4-one (( )-17). A mixture of chrysin (2.03 g, 8
mmol) and hexanal (1.00 g, 10 mmol) in an aqueous solution MeNH2
(40% in water, 6.2 g, 80 mmol) was heated in a sealed tube at 40 °C
for 5 h and then cooled to rt and diluted with CH2Cl2 (30 mL) and
H2O (10 mL). The mixture was filtered to remove trace of insoluble
chrysin. The separated aqueous layer of the filtrate was extracted with
CH2Cl2 (2 × 10 mL). The combined organic layers were dried
(Na2SO4) and concentrated. The residue (8-substituted: 6-substituted
ASSOCIATED CONTENT
* Supporting Information
■
S
1
= 1:2 by H NMR) was purified by flash column chromatography
Experimental details, NMR, mass, HPLC chromatograms of
racemic and enantiomers data, X-ray data, X-ray structures of
(−)-1, (+)-15, and (−)-18, table with residue variability in the
binding sites of kinases and sequence alignment used for
homology modeling. This material is available free of charge via
the Internet at http://pubs.acs.org.
(AcOEt/MeOH 10:0 to 95:5) to afford F1 5,7-dihydroxy-8-(1-
(methylamino)hexyl)-2-phenyl-4H-chromen-4-one (793 mg, 27%)
and F2 5,7-dihydroxy-6-(1-(methylamino)hexyl)-2-phenyl-4H-chro-
men-4-one (2.42 g, 67%) as yellow crystals. Under other conditions
in which a mixture of chrysin (254 mg, 1 mmol), hexanal (200 mg, 2
mmol), and MeNH2 (40% in water, 0.46 mL, 6 mmol) was heated in a
sealed tube at 40 °C for 16 h then cooled to rt and subjected to the
same treatment, 5,7-dihydroxy-6-(1-(methylamino)hexyl)-2-phenyl-
4H-chromen-4-one was obtained preponderantly (286 mg, 78%).
( )-18: 1H NMR (300 MHz, CDCl3) δ 12.72 (s, 1H), 7.82−7.80 (m,
2H), 7.54−7.24 (m, 3H), 6.63 (s, 1H), 6.25 (s, 1H), 4.38 (t, J = 6.8
Hz, 1H), 2.47 (s, 3H), 1.87−1.79 (m, 2H), 1.49−1.16 (m, 6H), 0.77
(t, J = 7 Hz, 3H). 13C NMR (75 MHz, CDCl3) δ 182.6, 166.2, 163.1,
161.5, 155.2, 131.9, 129.4, 126.2, 106.0, 104.8, 102.8, 100.9, 59.1, 35.8,
34.6, 31.8, 25.9, 22.7, 14.1. HRMS: m/z [M + H]+ calcd for
C22H26NO4, 368.1862; found, 368.1850. ( )-17: 1H NMR (300 MHz,
CDCl3) δ 13.09 (s, 1H), 7.85−7.81 (m, 2H), 7.48−7.36 (m, 3H), 6.58
(s, 1H), 6.36 (s, 1H), 4.25 (t, J = 6.8 Hz, 1H), 2.44 (s, 3H), 1.83−1.66
(m, 2H), 1.49−1.16 (m, 6H), 0.77 (t, J = 7 Hz, 3H). 13C NMR (75
MHz, CDCl3) δ 182.6, 166.6, 163.7, 159.6, 157.3, 131.7, 129.2, 126.4,
107.9, 105.6, 95.1, 77.7, 77.2, 76.8, 58.3, 35.1, 34.3, 31.9, 25.6, 22.7,
14.2. HRMS: m/z [M + H]+ calcd for C22H26NO4, 368.1862; found,
368.1869. Enantiomeric separation of ( )-18 by chiral HPLC: 400 mg
of F1 was next subjected to SFC HPLC separation using column IC,
20% cosolvents (MeOH/diethylamine 98:2), flow rate 4 mL/min; tR1
AUTHOR INFORMATION
Corresponding Author
■
*For chemistry (F.G.): phone, +33 (0) 1 69 82 45 80; E-mail,
gueritte@icsn.cnrs-gif.fr. For biology (L. M.): phone, +33 (0) 2
98 29 23 39; E-mail, meijer@manros-therapeutics.com.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
Kinase assays were supported by grants from “Association
■
France-Alzheimer” (Comite
(Institut National du Cancer) “Recherches Biomed
́
du Finister
̀
e) (L.M.), INCa
icales”
́
(GLIOMER project) (L.M.), “Fonds Unique Interministeriel”
(PHARMASEA project) (L.M.), and “Ligue Nationale contre le
Cancer” (Comite
́ ̀
du Finistere) (L.M.).
23
= 9.7 min, 165 mg, ee = 96%, [α]D +62.3 (c, 0.12, CHCl3),
23
ABBREVIATIONS USED
configuration R; tR2 = 13.3 min, 159 mg, ee = 99%, [α]D −65.4 (c,
■
0.26, CHCl3), configuration S (determined by X-ray diffraction).
5.3. Molecular Modeling. Crystal structures were downloaded
from the Protein Data Bank (PDB), and residues not belonging to the
protein were removed, including water molecules. The homology
model of CDK1 was built with MODELER26 v9.7 using the X-ray
structure of CDK2 (PDB code 1YKR) and the sequence alignment
shown in Figure S2 of the Supporting Information), generated using
the ClustalW27 software (v1.8). Several homology models (100) were
built, and the one with the best DOPE energy score retained for the
CDK, cyclin-dependent kinase; GSK3, glycogen synthase
kinase-3; DYRK1A, dual specificity tyrosine-phosphorylation-
regulated kinase 1A; CLK1, CDC-like kinase 1; ATP, adenosine
5′-triphosphate; SFC, supercritical fluid chromatography;
HPLC, high-performance liquid chromatography; THF,
tetrahydrofuran; [α], specific rotation; μM, micromolar; IC50,
half-maximum inhibitory concentration; PDB, Protein Data
Bank; rmsd, root-mean-square deviation; Val, valine; Ala,
2818
dx.doi.org/10.1021/jm201727w | J. Med. Chem. 2012, 55, 2811−2819

Journal of Medicinal Chemistry
Article
(17) Berman, H. M.; Battistuz, T.; Bhat, T. N.; Bluhm, W. F.;
Bourne, P. E.; Burkhardt, K.; Feng, Z.; Gilliland, G. L.; Iype, L.; Jain,
S.; Fagan, P.; Marvin, J.; Padilla, D.; Ravichandran, V.; Schneider, B.;
Thanki, N.; Weissig, H.; Westbrook, J. D.; Zardecki, C. The Protein
Data Bank. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2002, 58, 899−
907.
(18) Mapelli, M.; Massimiliano, L.; Crovace, C.; Seeliger, M. A.; Tsai,
L. H.; Meijer, L.; Musacchio, A. Mechanism of CDK5/p25 binding by
CDK inhibitors. J. Med. Chem. 2005, 48, 671−679.
(19) Aoki, M.; Yokota, T.; Sugiura, I.; Sasaki, C.; Hasegawa, T.;
Okumura, C.; Ishiguro, K.; Kohno, T.; Sugio, S.; Matsuzaki, T.
Structural insight into nucleotide recognition in tau-protein kinase I/
glycogen synthase kinase 3 beta. Acta Crystallogr., Sect. D: Biol.
Crystallogr. 2004, 60, 439−446.
(20) Hamdouchi, C.; Zhong, B.; Mendoza, J.; Collins, E.; Jaramillo,
C.; De Diego, J. E.; Robertson, D.; Spencer, C. D.; Anderson, B. D.;
Watkins, S. A.; Zhang, F.; Brooks, H. B. Structure-based design of a
new class of highly selective aminoimidazo[1,2-a]pyridine-based
inhibitors of cyclin dependent kinases. Bioorg. Med. Chem. Lett.
2005, 15, 1943−1947.
(21) Baumli, S.; Lolli, G.; Lowe, E. D.; Troiani, S.; Rusconi, L.;
Bullock, A. N.; Debreczeni, J. E.; Knapp, S.; Johnson, L. N. The
structure of P-TEFb (CDK9/cyclin T1), its complex with flavopiridol
and regulation by phosphorylation. EMBO J. 2008, 27, 1907−1918.
(22) Bettayeb, K.; Oumata, N.; Echalier, A.; Ferandin, Y.; Endicott, J.
A.; Galons, H.; Meijer, L. CR8, a potent and selective, roscovitine-
derived inhibitor of cyclin-dependent kinases. Oncogene 2008, 27,
5797−5807.
(23) Primot, A.; Baratte, B.; Gompel, M.; Borgne, A.; Liabeuf, S.;
Romette, J. L.; Costantini, F.; Meijer, L. Purification of GSK-3 by
affinity chromatography on immobilised axin. Protein Expression Purif.
2000, 20, 394−404.
alanine; Leu, leucine; Cys, cysteine; Asp, aspartic acid; Asn,
asparagine; Lys, lysine; Gln, glutamine; Arg, arginine; Tyr,
tyrosine; Thr, threonine; Phe, phenylalanine; mM, millimolar;
EGTA, ethylene glycol tetraacetic acid; DTT, dithiothreitol;
BSA, bovine serum albumin; DMSO, dimethyl sulfoxide; GST,
glutathione S-transferase; TLC, thin layer chromatography; UV,
ultraviolet; TMS, tetramethylsilane; s, singlet; d, doublet; t,
triplet; q, quartet; m, multiplet; ESI, electron spray ionization;
AcOEt, ethyl acetate; MeOH, methanol; NMR, nuclear
magnetic resonance; HRMS, high resolution mass spectrome-
try; iPrOH, isopropanol; ee, enantiomeric excess; rt, room
temperature; tr, retention time; MeNH2, methylamine; DOPE,
discrete optimized protein energy; GA, genetic algorithm
REFERENCES
■
(1) Diaz-Padilla, I.; Siu, L. L.; Duran, I. Cyclin-dependent kinase
inhibitors as potential targeted anticancer agents. Invest. New Drugs
2009, 27, 586−594.
(2) Meijer, L.; Flajolet, M.; Greengard, P. Pharmacological inhibitors
of glycogen synthase kinase 3. Trends Pharmacol. Sci. 2004, 25, 471−
480.
(3) Park, J.; Song, W.-J.; Chung, K. C. Function and regulation of
Dyrk1A: towards understanding Down syndrome. Cell. Mol. Life Sci.
2009, 66, 3235−3240.
(4) Moeslein, F. M.; Myers, M. P.; Landreth, G. E. The CLK family
kinases, CLK1 and CLK2, phosphorylate and activate the tyrosine
phosphatase, PTP-1B. J. Biol. Chem. 1999, 274, 26697−26704.
(5) Ahond, A.; Fournet, A.; Moretti, C.; Philogene, E.; Poupat, C.;
Thoison, O.; Potier, P. Premiers Alcaloides vrais isoles
Combretacees, Buchenavia macrophylla Eichl. et Buchenavia capitata
Eichl. Bull. Soc. Chim. Fr. 1984, 1−2 (Pt2), 41−45.
́
de
́
́
(24) Gravel, E.; Poupon, E.; Hocquemiller, R. Biomimetic
investigations from reactive lysine-derived C5 units: one step synthesis
of complex polycyclic alkaloids from the Nitraria genus. Tetrahedron
2006, 62, 5248.
(6) Beutler, J. A.; Cardellina, J. H. II; McMahon, J. B.; Boyd, M. R.
Anti-HIV and cytotoxic alkaloids from Buchenavia capitata. J. Nat.
Prod. 1992, 55, 207−213.
(25) Nomura, Y.; Ogawa, K.; Takeuchi, Y.; Tomoda, S. One-Step
Synthesis and Structural Confirmation of 1-Pyrroline Trimer. Chem.
Lett. 1977, 6, 693.
(7) Khoo, B. Y.; Chua, S. L.; Balaram, P. Apoptotic Effects of Chrysin
in Human Cancer Cell Lines. Int. J. Mol. Sci. 2010, 11, 2188−2199.
(8) Senderowicz, A. M. Flavopiridol: the first cyclin-dependent kinase
inhibitor in human clinical trials. Invest. New Drugs 1999, 17, 313−320.
(9) Raje, N.; Hideshima, T.; Mukherjee, S.; Raab, M.; Vallet, S.;
Chhetri, S.; Cirstea, D.; Pozzi, S.; Mitsiades, C.; Rooney, M.; Kiziltepe,
T.; Podar, K.; Okawa, Y.; Ikeda, H.; Carrasco, R.; Richardson, P. G.;
Chauhan, D.; Munshi, N. C.; Sharma, S.; Parikh, H.; Chabner, B.;
Scadden, D.; Anderson, K. C. Preclinical activity of P276−00, a novel
smallmolecule cyclin-dependent kinase inhibitor in the therapy of
multiple myeloma. Leukemia 2009, 23, 961−970.
̌
(26) Sali, A.; Potterton, L.; Yuan, F.; van Vlijmen, H.; Karplus, M.
Evaluation of comparative protein modeling by MODELLER. Proteins:
Struct., Funct., Bioinf. 1995, 23, 318−326.
(27) Thompson, J. D.; Higgins, D. G.; Gibson, T. J. CLUSTAL W:
improving the sensitivity of progressive multiple sequence alignment
through sequence weighting, position-specific gap penalties and weight
matrix choice. Nucleic Acids Res. 1994, 22, 4673−4680.
(28) CORINA, version 3.44; Molecular Networks GmbH: Erlangen,
Germany, 2008; http://www.molecular-networks.com (Accessed
December 15, 2011).
(10) Liu, T.; Xu, Z.; He, Q.; Chen, Y.; Yang, B.; Hu, Y. Nitrogen-
containing flavonoids as CDK1/cyclin B inhibitors: design, synthesis,
and biological evaluation. Bioorg. Med. Chem. Lett. 2007, 17, 278−281.
(11) Zhang, S.; Ma, J.; Bao, Y.; Yang, P.; Zou, L.; Li, K.; Sun, X.
Nitrogen-containing flavonoid analogues as CDK1/cyclin B inhibitors:
synthesis, SAR analysis, and biological activity. Bioorg. Med. Chem.
2008, 16, 7127−7132.
(29) LigPrep, version 2.5; Schrodinger, LLC: New York, 2011; http://
̈
www.schrodinger.com (Accessed December 15, 2011).
(30) Verdonk, M. L.; Cole, J. C.; Hartshorn, M. J.; Murray, C. W.;
Taylor, R. D. Improved protein−ligand docking using GOLD. Proteins:
Struct., Funct., Bioinf. 2003, 52, 609−623.
(31) Maestro, version 9.2; Schrodinger, LLC: New York, 2011; http://
̈
www.schrodinger.com (Accessed December 15, 2011).
(32) The PyMOL Molecular Graphics System, version 0.99; DeLano
Scientific: Palo Alto, CA, 2006; http://www.pymol.org (Accessed
December 15, 2011).
(12) Johns, S. R.; Russel, J. H.; Hefferman, M. L. Ficine, A novel
flavonoidal alkaloid from Ficus pantoniana. Tetrahedron Lett. 1965, 24,
1987−1991.
(13) Leete, E. One-Step Synthesis of Ficine and Isoficine. J. Nat.
Prod. 1982, 45, 605−607.
́
(14) Nguyen, T.-B.; Wang, Q.; Gueritte, F. An efficient one-step
synthesis of piperidin-2-yl and pyrrolidin-2-yl flavonoidal alkaloids
through phenolic Mannich reaction of flavonoid and cyclic imines. Eur.
J. Org. Chem. 2011, 7076−7079.
(15) Wessely, F.; Moser, G. H. Synthese und Konstitution des
Skutellareins. Monatsh. Chem. 1930, 56, 97−105.
(16) Shinomiya, K.; Hano, Y.; Nomura, T. Mechanism on one-sided
Wessely−Moser Rearrangement Reaction. Heterocycles 2000, 53, 877−
886.
2819
dx.doi.org/10.1021/jm201727w | J. Med. Chem. 2012, 55, 2811−2819