Integration of high speed microwave chemistry and a statistical 'design of experiment' approach for the synthesis of the mitotic kinesin Eg5 inhibitor monastrol

Article abstract of DOI:10.1016/j.tet.2007.12.056

The rapid preparation of the mitotic kinesin Eg5 inhibitor monastrol has been performed using multicomponent chemistry and combining microwave-assisted synthesis and a statistical design of experiments (DoEs) approach. By variation of the solvent, catalys

Full text of DOI:10.1016/j.tet.2007.12.056

Available online at www.sciencedirect.com
Tetrahedron 64 (2008) 2035e2041
www.elsevier.com/locate/tet
Integration of high speed microwave chemistry and a statistical
design of experiment’ approach for the synthesis of the mitotic
kinesin Eg5 inhibitor monastrol
‘
a
Toma N. Glasnov , Heather Tye , C. Oliver Kappe
b
a,*
a
Christian Doppler Laboratory for Microwave Chemistry (CDLMC) and Institute of Chemistry, Karl-Franzens University of Graz,
Heinrichstrasse 28, A-8010 Graz, Austria
b
Evotec (UK) Ltd, 114 Milton Park, Abingdon, Oxfordshire, OX14 4RZ, UK
Received 2 November 2007; received in revised form 17 December 2007; accepted 20 December 2007
Available online 4 January 2008
Abstract
The rapid preparation of the mitotic kinesin Eg5 inhibitor monastrol has been performed using multicomponent chemistry and combining
microwave-assisted synthesis and a statistical design of experiments (DoEs) approach. By variation of the solvent, catalyst type and concentra-
tion, reaction time, and temperature a matrix of experiments for the DoE method was derived. As a result of 29 experiments with reaction times
ranging from 10 to 30 min, an optimized procedure was derived that allowed the preparation of monastrol in 82%.
Ó 2008 Elsevier Ltd. All rights reserved.
Keywords: Dihydropyrimidines; Multicomponent reaction; Design of experiments; Microwave irradiation; Heterocycles
1
. Introduction
separate factor at time’) could be misleading, being one-
dimensional, thus often missing the optimal set of conditions
and/or leading to different implications. On the contrary, ap-
plying DoE provides the scientist with an organized approach
concerning rational reaction variables’ exploration, in multi-
dimensional fashion additionally taking into account possible
factor interactions. Another concern when using DoE can be
the number of experiments needed in order to obtain a plausi-
ble model. However, this issue is easily resolved by the use of
commercially available robotic synthesizers.
As statistical methods play an ever increasing important
role in today’s research programs, the so-called ‘design of
experiments’ (DoEs) approach has become a valuable instru-
ment in the hands of process anddmore recentlydof discov-
ery oriented synthetic chemists. After being implemented in
2
praxis in the late 1950s, DoE has further evolved so that today
no in-depth knowledge of statistics is necessary for an organic
chemist to succeed in applying these methods. By integrating
statistical DoE protocols into sophisticated software pack-
1
Microwave-assisted organic synthesis (MAOS) has
emerged during the past two decades as another valuable tech-
3
ages, this method has become easily accessible and a comfort-
4
5
able tool for reaction optimizations. Still, by tradition,
chemists prefer to rely on their intuition whenever facing
the issue of reaction optimization. Optimizing a reaction by
randomly varying factors in a sequence (the so-called
OVATd‘one variable at a time’, or COSTd‘changing one
nology for synthetic organic and medicinal chemists. Re-
placing the oil bath with a dedicated microwave reactor
provides the opportunity to perform reactions in dramatically
shortened time (more experiments and the corresponding
data point generation from the statistical point of view) as
well as increasing yields by using conditions not attainable un-
der conventional heating. Combining high speed microwave
synthesis with a DoE approach, therefore, appears to be an
ideal tool for obtaining truly optimized reaction conditions
4
,5
*
Corresponding author. Tel.: þ43 316 3805352; fax: þ43 316 3809840.
E-mail address: oliver.kappe@uni-graz.at (C.O. Kappe).
URL: http://www.maos.net
0040-4020/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tet.2007.12.056

2036
T.N. Glasnov et al. / Tetrahedron 64 (2008) 2035e2041
in a short period of time. Despite this fact, only a very limited
number of publications so far have described applications
using this attractive combination of enabling technologies.
1999 report, numerous procedures have been reported in
which different mediators and solvents have been introduced
11,12
for the preparation of monstrol.
1
a,6
In this context, we herewith wish to report the use of a DoE
approach in combination with automated sequential micro-
wave synthesis for optimizing the Biginelli dihydropyrimidine
To fully optimize the dihydropyrimidine-2-thione forma-
tion, we have now examined a number of experimental vari-
ables using a statistical DoE software package (MODDE8
7
3
synthesis. Specifically, we have applied this technology for
the synthesis of the mitotic kinesin inhibitor monastrol
from Umetrics) in combination with an automated single
mode microwave reactor, which allowed unattended process-
8
Scheme 1, 1), a dihydropyrimidine derivative on which we
4,5
ing of reaction vessels. For the initial screening, we focused
(
have previously performed microwave-assisted optimization
studies by the above mentioned conventional OVAT/COST
method.
our attention on a few Lewis acid catalystsdCu(OTf) , CuBr ,
2
2
LaCl , and Yb(OTf) , and few standard organic solventsdace-
3
3
9
,10
tonitrile, ethanol, dichloromethane, and tetrahydrofuran. For
this initial screening, we decided to apply the previously opti-
9
mized microwave conditionsdcontinuous irradiation for
OH
OH
ꢀ
3
0 min at 120 C of 1.5 equiv of ethyl acetoacetate, 1 equiv
of 3-hydroxybenzaldehyde, 1 equiv of thiourea, 0.1 equiv of
the catalyst, and 1 mL of solvent, as a starting point. Using
the DoE software package, a Full Factorial Design (mixed)
was chosen, where catalysts and solvents were selected as
qualitative factors and conversion and purity were the desired
responses, providing 16 main experiments plus 3 center points
O
O
H
catalyst, solvent
O
EtO
Me
NH
time, temperature
EtO
NH2
S
N
H
S
Me
O
H2N
1
Scheme 1. Biginelli three-component synthesis of monastrol.
(center points serve to give measure of the reproducibility of
the results and allow for more reliable statistical modeling),
where the main effects and all interactions are not confounded.
One advantage of the DoE software packages is the possibility
for graphical representation of the results by automatically
generated plots. Hence, an interaction plot with the results
2
. Results and discussion
As we have previously demonstrated, monastrol (1) can be
easily synthesized using a sealed vessel microwave protocol
employing the Biginelli cyclocondensation of ethyl acetoace-
tate, 3-hydroxybenzaldehyde, and thiourea, promoted by the
(Fig. 1) was produced, showing that both LaCl
in acetonitrile indeed perform very well as catalysts in terms
of conversion and purity (in contrast to CuBr ).
From the data shown in Figure 1, it was determined that in
this case the solvent of choice clearly is ethanol, in particular
3 3
and Yb(OTf)
2
9
use of Yb(OTf) in MeCN as solvent. This approach shortens
3
the reaction time required by conventional heating methods
from hours to a few minutes. These reaction conditions were
specifically optimized for the preparation of monastrol al-
though similar conditions are applicable for preparing other
in combination with LaCl or Yb(OTf) as a Lewis acid. It
3 3
should be noted that previous empirical optimizations have
9
found acetonitrile to be a suitable solvent. Based on a price
1
0
13
analysis of LaCl and Yb(OTf) , we have decided to use
3 3
dihydropyrimidine-2-thione derivatives. Since the original
Figure 1. Interaction effects plot of the catalyst/solvent screening for the optimized synthesis of monastrol for (a) conversion and (b) purity. This type of plot
enables the effect of both solvent and catalyst to be visualized simultaneously. Clearly the result in terms of conversion and purity is highly dependent on
both the choice of solvent and catalyst.

T.N. Glasnov et al. / Tetrahedron 64 (2008) 2035e2041
2037
2
2
LaCl for all further optimization studies as this is less expen-
3
of the model displayed acceptable values for R and Q for
both conversion (0.91 and 0.71) and purity (0.96 and 0.70)
representing the model as good for our studies, which was
also confirmed by ‘Lack-of-Fit’ and ANOVA plots (data not
sive. With this information in hand, we decided to next exam-
ine the effects of temperature, time, and catalyst concentration
on the monastrol synthesis using LaCl /ethanol as the reaction
3
1
4
medium. A second experimental design was devised. In this
model, the effects and interactions between the reaction tem-
perature and the reaction time were investigated. These two
parameters were selected as quantitative factors for the design.
shown). Figure 2 shows a plot for the coefficients of the
model. As it can be seen, both temperature and time are signif-
icant (error bars small relative to the size of the bar) and have
a positive impact on both conversion and purity. The addi-
tional square and interaction terms are not highly significant
(error bars are large and span the axis), however, they are
required to give a good model. It should be noted that the in-
fluence of these terms, particularly the timeꢁtime term differs
for each of the responses. The contour plot (Fig. 3) shows
a ‘map’ of the reaction space based on the results obtained
by experimental design assessing temperature and time effects
on the monastrol synthesis. Due to the presence of a slight cur-
vature in the response for purity it can be predicted that at
A three level factorial design was selected using 100e120e
ꢀ
1
40 C for the temperature variation and 10e20e30 min for
the time variation. The design consisted of eight runs plus
two center point experiments (Table 1).
Table 1
Three level factorial design and results for a temperatureetime dependence
a
study of the Biginelli reaction (Scheme 1)
ꢀ
b
Conversion (%)
c
Purity (%)
Exp
Temp ( C)
Time (min)
ꢀ
elevated temperatures (>140 C) and prolonged reaction
time (30 min) the product purity will drop.
1
2
3
4
5
6
7
8
9
100
140
100
140
100
140
120
120
120
120
10
10
30
30
20
20
10
30
20
20
67
86
84
94
77
95
78
96
89
85
51
68
64
74
60
79
60
71
69
66
ꢀ
The fact that higher reaction temperatures (>120 C) in mi-
crowave-assisted Biginelli reactions using protic solvents such
as ethanol will lead to reduced dihydropyrimidine product
purities has been experimentally demonstrated previously
and can be rationalized by partial decomposition of the urea/
1
0,15
thiourea components.
With this information in hand, we
1
0
decided to perform one additional experimental design to op-
timize the catalyst concentration. For this purpose another ex-
perimental matrix was assembled. In this model the effects and
interactions between the reaction temperature and the quantity
of catalyst employed in the reaction were studied.
a
Software-generated set of experiments including 8 main experimental
points and 2 center points (experiments 9 and 10). For details on experimental
conditions, see main text.
b
HPLC conversion (peak area percent, 215 nm) based on 3-hydroxybenzal-
dehyde and product peak integration.
c
These two parameters were selected as quantitative factors
for the design, and conversion and purity were depicted as
responses. For the variation of the factors, we selected
HPLC purity (peak area percent, 215 nm), unreacted starting materials are
considered as impurities.
ꢀ
As in the first design the molar ratio of ethyl acetoacetate/3-
hydroxybenzaldehyde/thiourea/LaCl was kept at 1.5/1.0/1.0/
100e120e140 C steps for the temperature variation and 5e
10e15 mol % for the catalyst concentration. Again, we selected
a three level factorial design to efficiently explore the reaction
space. The set of suggested experiments contained eight design
runs plus three center points (Table 2). The reagent ratio and
amount of solvent were kept identical as in the previous design.
3
0
.1, the mixture being dissolved in 1 mL pure ethanol. After
microwave heating of the reaction mixtures 1e10 under the
defined conditions (Table 1), the model was fitted with the
values obtained for conversion and purity. The summary plot
Figure 2. Coefficient plot for temperatureetime effect screening for (a) conversion and (b) purity. Each plot contains the terms used in the statistical model. Posi-
tive bars indicate a positive impact of the term on the measured response. Negative bars indicate a negative effect. The significance of each term in the model is
determined by the size of the bar relative to the error bar. Non-significant terms are small and have an error bar, which spans the axis.

2038
T.N. Glasnov et al. / Tetrahedron 64 (2008) 2035e2041
Figure 3. Response surfaces for (a) conversion and (b) purity in dependence of temperature and time. Each contour line indicates a level of the response, e.g.,
conversion for a given combination of the variables, in this case temperature and time. Blue areas indicate a low response and red areas indicate a high response,
thus the optimal regions are red in these plots. Curvature of the lines indicates that the response is non-linear with respect to one or both of the factors. The purity
response is strongly curved on the time axis, which is consistent with the significant timeꢁtime term in the model.
After microwave heating the reaction mixtures using the de-
fined conditions, the model was fitted with the values obtained
for conversion and purity. However, fitting the results gave
optimized model, we were able to get somewhat better results.
Still, the complemented model was not able to adequately
describe the studied response. In the MODDE8 software the
so-called ‘Normal Probability Plot of Residuals’ is available
as part of the model analysis tools, which display the distribu-
tion of the responses on a double log scale. On this plot it is pos-
sible to detect obvious ‘outliers’, which do not fit well within
the distribution of responses. By examining the Normal Proba-
bility Plot of Residuals for the results from experiments six and
seven were depicted as strong outliers. Exclusion of these
results and re-fitting of the model substantially improved the
2
a rather poor model with a low Q value and poor reproducibil-
ity. To solve this problem the choice was either to attempt
a more sophisticated type of design or to improve the prediction
capability of the existing model by complementing the model
with additional runs. Since the main purpose of DoE is to opti-
mize a reaction with minimal effort and maximal efficiency, we
opted for the second possibility. Using the MODDE8 software
capabilities design was complemented with five additional runs
2
2
(experiments 12e15, Table 2). As anticipated, inclusion of the
results from the additional five experiments and re-fitting the
values for R and Q for both conversion (0.92 and 0.74) and
purity (0.92 and 0.76), the validity and reproducibility of the
model were correspondingly above 87%.
Table 2
Complemented three level factorial design for a temperatureecatalyst concen-
tration dependence study in the Biginelli reaction (Scheme 1)
From the coefficient plot (Fig. 4) the pronounced tempera-
ture effect was evident, in particular on purity. The squared
terms have a much greater impact on the conversion (large
bars) than they do on the purity (small bars and large error
bars). This suggests that the conversion is non-linear with re-
spect to both temperature and catalyst concentration. The 2D
surface plot (Fig. 5) represented the dependence of the conver-
sion and purity of monastrol formation on the temperature and
catalyst concentration. As it is evident from the plot, the purity
has an almost linear dependence of the catalyst concentration
and temperature variation (consistent with the coefficient plot
shown in Fig. 4b). However, in the case of conversion we ob-
serve a strong curvature in the response particularly with re-
spect to the catalysts concentration (this is again consistent
with the coefficient plot shown in Fig. 4a). These plots taken
together suggest that optimal performance in the synthesis of
monastrol (in terms of purity) is predicted when the catalyst
a
ꢀ
b
Conversion (%)
c
Purity (%)
Exp
Temp ( C)
Catalyst (mol %)
1
2
3
4
5
100
140
100
140
100
140
120
120
120
120
120
100
140
100
140
120
5
5
88
91
93
92
91
88
92
93
93
93
93
87
93
93
95
92
77
86
77
90
77
86
79
86
82
81
85
73
91
78
91
83
15
15
10
10
5
d
6
d
7
8
15
10
10
10
5
e
9
e
1
1
1
1
1
1
1
0
1
2
3
4
5
6
a
e
f
f
f
f
f
5
15
15
10
ꢀ
concentration and temperature are at a maximumd140 C
and 15 mol % of LaCl (Fig. 5b), but the best results in terms
Software-generated set of experiments including eight main experimental
3
points, two center points, and five complementing experimental points. For
details on experimental conditions, see main text.
ꢀ
of conversion could be achieved at around 125 C and
10 mol % LaCl3.
b
HPLC conversion (peak area percent, 215 nm) based on 3-hydroxybenz-
aldehyde and product peak integration.
The final optimized conditions for the synthesis of monas-
trol (Scheme 1) were determined using the MODDE8 sweet-
spot function (Fig. 6). We targeted maximum conversion and
purity (conversion being more important). However, the
sweet-spot region for the temperatureecatalyst design was
c
HPLC purity (peak area percent, 215 nm), unreacted starting materials are
considered as impurities.
d
Strong outliers excluded from the model.
Center points.
Complementing runs.
e
f

T.N. Glasnov et al. / Tetrahedron 64 (2008) 2035e2041
2039
Figure 4. Coefficient plot for temperatureecatalyst amount effect screening: for (a) conversion and (b) purity. Each plot contains the terms used in the statistical
model. Positive bars indicate a positive impact of the term on the measured response. Negative bars indicate a negative effect. The significance of each term in the
model is determined by the size of the bar relative to the error bar. Non-significant terms are small and have an error bar, which spans the axis.
Figure 5. Response surfaces for (a) conversion and (b) purity in dependence of temperature and catalyst concentration.
narrower than the one for the temperatureetime model (still
including the desired area from the first onedFig. 6b), as a re-
sult the temperatureecatalyst sweet spot was selected as the
the optimal set of conditions for the reaction. For the duration
of the microwave irradiation 30 min were considered as opti-
mum. The response for conversion was predicted to be 93%
and the obtained response was 94%, leading to 82% isolated
yield after chromatographic work-up.
one of higher importance. A sweet spot corresponding to
ꢀ
1
40 C and 12 mol % LaCl was selected from Figure 6a as
3
Figure 6. Sweet-spot prediction plots for (a) temperature versus catalyst amount (the set desired values for conversion are in the range 93e95% and 90e100% for
purity), and (b) temperature versus time (the set desired values for conversion are in the range 90e96% and 72e80% for purity). Areas in red meet both given
criteria, areas in blue meet one of the criteria and in the white area no criteria is met.

2040
T.N. Glasnov et al. / Tetrahedron 64 (2008) 2035e2041
3
. Conclusion
2 min, the vial was closed, placed in the microwave cavity,
ꢀ
and irradiated at 140 C for 30 min (fixed hold time). After
In summary, combining the advantages of automated high
the reaction was completed, an aliquot sample of the reaction
mixture was taken and subjected to HPLC analysis (215/
254 nm) in order to determine the conversion/purity for the
corresponding run.
speed microwave synthesis and a statistical design of experi-
ments (DoEs) approach has allowed the rapid screening for
an optimum catalyst/solvent system and for temperaturee
timeecatalyst concentration conditions for the synthesis of
the mitotic kinesin Eg5 inhibitor monastrol. The now proposed
4.3. Experimental procedure for design 2 (temperatureetime
LaCl /ethanol catalyst/solvent system was demonstrated to be
3
optimization/three level factorial design)
as effective as the more traditionally employed Yb(OTf) /ace-
3
tonitrile combination. As a final remark it should be mentioned
that these statistically driven studies required a set of 29 exper-
iments in total. By using HPLC as rapid analytical tool, the
full set of optimizations could be performed in a very short
time span.
The design matrix consists of 10 runs (8 main runs plus 2
center points, see Table 1), automatically generated by
the MODDE8 software. These were performed in a random
fashion in accordance to the software-generated worksheet ta-
ble. The screening runs were performed as follows: in a 10 mL
Pyrex microwave vial, 3-hydroxybenzaldehyde (100 mg,
0.82 mmol, 1 equiv), ethyl acetoacetate (157 mL, 1.23 mmol,
4
. Experimental
1.5 equiv), thiourea (60.7 mg, 0.82 mmol, 1 equiv), and
4
.1. General
10 mol % LaCl3 (24.5 mg, 0.082 mmol, 0.10 equiv) were
added to 1 mL ethanol. After magnetic stirring for 2 min,
the vial was closed, placed in the microwave cavity, and irra-
Microwave-assisted reactions were performed on a Discover
CEM Corporation) single mode microwave instrument at
450 MHz controlled irradiation using standard sealed micro-
ꢀ
(
2
diated at 100, 120, or 140 C for 10, 20, or 30 min (fixed hold
time). After the reaction was finished, an aliquot sample of the
reaction mixture was taken and subjected to HPLC analysis
(215/254 nm) in order to determine the conversion/purity for
the corresponding run.
wave glass vials. For the statistical model generation and eval-
uation of the responses, MODDE8 (Umetrics) software was
used. Reaction temperatures were monitored by an IR sensor
on the outside wall of the reaction vials. Reaction times refer
to hold times at the selected set temperature, not to total irra-
4.4. Experimental procedure for design 3 (temperaturee
catalyst amount optimization/three level factorial design)
1
diation times. H NMR spectra were recorded on a Bruker
AMX 360. Low resolution mass spectra were obtained in
the atmospheric pressure chemical ionization (positive or neg-
ative APCI mode). Analytical HPLC analysis was carried out
The design matrix consists of 11 runs (8 main runs plus 3
center points, see Table 2), automatically generated by
the MODDE8 software. These were performed in a random
fashion in accordance to the software-generated worksheet
table. The screening runs were performed as follows: in
a 10 mL Pyrex microwave vial, 3-hydroxybenzaldehyde
(100 mg, 0.82 mmol, 1 equiv), ethyl acetoacetate (157 mL,
1.23 mmol, 1.5 equiv), thiourea (60.7 mg, 0.82 mmol, 1 equiv),
on a LiChrospher 100 C18 reversed-phase analytical column
ꢀ
(
119ꢁ3 mm, 5 mm particle size) at 25 C, using mobile phase
A (water/MeCN 9/1 (v/v)þ0.1% TFA) and phase B
MeCNþ0.1% TFA), with linear gradient from 30% B to
00% B in 8 and 2 min with 100% phase B. Chromatographic
(
1
9
product purification was performed as previously described.
Melting points were recorded on a GALLENKAMP melting
point apparatus.
and LaCl d5 mol % (10.1 mg, 0.041 mmol, 0.05 equiv),
3
10 mol % (24.5 mg, 0.082 mmol, 0.10 equiv), or 15 mol %
(30.1 mg, 0.123 mmol, 0.15 equiv)dwere added to 1 mL etha-
nol. After magnetic stirring for 2 min, the vial was closed,
4
.2. Experimental procedure for design 1 (catalyst/solvent
optimization/full factorial designdmixed)
placed in the microwave cavity, and irradiated at 100, 120, or
ꢀ
1
40 C for 30 min (fixed hold time). After the reaction is fin-
The design matrix consists of 19 runs (16 main runs plus 3
center points), automatically generated by the MODDE8
software. These were performed in a random fashion in accor-
dance to the software-generated worksheet table. The screening
runs were performed as follows: in a 10 mL Pyrex microwave
vial, 3-hydroxybenzaldehyde (100 mg, 0.82 mmol, 1 equiv),
ethyl acetoacetate (157 mL, 1.23 mmol, 1.5 equiv), thiourea
ished, an aliquot sample of the reaction mixture was taken
and subjected to HPLC analysis (215/254 nm) in order to deter-
mine the conversion/purity for the corresponding run. Five more
experimental runs were performed for complementing the de-
sign after the first fitting (see Table 2). These were automatically
generated and added to the previous set by the MODDE8 soft-
ware. They were performed in a random fashion in accordance
(
60.7 mg, 0.82 mmol, 1 equiv), and the corresponding catalyst-
dLaCl3 (24.5 mg, 0.082 mmol, 0.10 equiv), Cu(OTf)2
29.7 mg, 0.082 mmol, 0.12 equiv), CuBr2 (18.3 mg,
.082 mmol, 0.12 equiv), or Yb(OTf) (50.9 mg, 0.082 mmol,
3
.12 equiv) were added to 1 mL ethanol, tetrahydrofuran,
to the software-generated worksheet table and include the fol-
ꢀ
lowing experimental points: 100 C with 5 mol % LaCl ,
3
ꢀ ꢀ
140 C with 10 mol % LaCl , 100 C with 15 mol % LaCl
3 3,
(
0
0
ꢀ
ꢀ
140 C with 15 mol % LaCl , and 120 C with 10 mol %
3
LaCl . The five complementary runs were analyzed in the man-
3
dichloroethane, or acetonitrile. After magnetic stirring for
ner described above.

T.N. Glasnov et al. / Tetrahedron 64 (2008) 2035e2041
2041
4
.5. Ethyl 4-(3-hydroxyphenyl)-6-methyl-2-thioxo-1,2,3,4-
D. R. S. Mini-Rev. Med. Chem. 2003, 3, 449; (d) Shipe, W. D.; Wolken-
berg, S. E.; Lindsley, C. W. Drug Discov. Today: Technol. 2005, 2,
tetrahydropyrimidine-5-carboxylate (monastrol, 1):
optimized protocol
1
55; (e) Kappe, C. O.; Dallinger, D. Nature Rev. Drug Discov. 2006, 5, 51.
6
. (a) Evans, M.; Ring, J.; Schoen, A.; Bell, A.; Edwards, P.; Berthelot, D.;
Nicewonger, R.; Baldino, C. M. Tetrahedron Lett. 2003, 44, 9337; (b) Tye,
H.; Whittaker, M. Org. Biomol. Chem. 2004, 2, 813; (c) McLean, N. J.;
Tye, H.; Whittaker, M. Tetrahedron Lett. 2004, 45, 993; (d) Gopalsamy,
A.; Shi, M.; Nilakantan, R. Org. Process Res. Dev. 2007, 11, 450; (e)
Sarotti, A. M.; Spanevello, R. A.; Su a´ rez, A. G. Green Chem. 2007, 9,
In a 10 mL Pyrex microwave vial, 3-hydroxybenzaldehyde
(
100 mg, 0.82 mmol, 1 equiv), ethyl acetoacetate (157 mL,
.23 mmol, 1.5 equiv), thiourea (60.7 mg, 0.82 mmol, 1 equiv),
and LaCl (24.5 mg, 0.082 mmol, 0.10 equiv) were added to
1
3
1
137.
1 mL anhydrous ethanol. After magnetic stirring for 2 min,
the vial was closed, placed in the microwave cavity, and irradi-
7
. (a) Biginelli, P. Gazz. Chim. Ital. 1893, 23, 360; (b) Kappe, C. O. Eur. J.
Med. Chem. 2000, 35, 1043; (c) Kappe, C. O. Acc. Chem. Res. 2000, 33,
879; (d) Kappe, C. O. Multicomponent Reactions; Zhu, J., Bienayme, H.,
Eds.; Wiley-VCH: Weinheim, 2005; p 95.
ꢀ
ated at 140 C for 30 min (fixed hold time). After the reaction
was finished, the solvent was removed under reduced pressure
and the residue subjected to silica gel chromatography (chloro-
8
. (a) Mayer, T. U.; Kapoor, T. M.; Haggarty, S. J.; King, R. W.; Schreiber,
S. L.; Mitchison, T. J. Science 1999, 286, 971; (b) Haggarty, S. J.; Mayer,
T. U.; Miyamoto, D. T.; Fathi, R.; King, R. W.; Mitchison, T. J.; Schreiber,
S. L. Chem. Biol. 2000, 7, 275; (c) For a recent review on mitotic kinesin
inhibitors, see: Sarli, V.; Giannis, A. ChemMedChem 2006, 1, 293.
. Dallinger, D.; Kappe, C. O. Nat. Protoc. 2007, 2, 317.
form/acetone¼5/1), resulting in the isolation of the pure product
1
in 82% yield. H NMR (360 MHz, CDCl ): d 1.12 (t, J¼6.9 Hz,
3
3
1
9
1
H), 2.27 (s, 3H), 4.02 (q, J¼6.6 Hz, 2H), 5.51 (d, J¼1.44 Hz,
9
H), 6.61e6.70 (m, 3H), 7.06e7.17 (m, 1H), 9.44 (s, 1H),
9
.60 (s, 1H), 10.29 (s, 1H). Mp 183e185 C (MeCN), lit. mp
1
0. For general microwave-assisted reaction optimization and library synthe-
sis of dihydropyrimidines via the Biginelli reaction, see: (a) Stadler, A.;
Kappe, C. O. J. Comb. Chem. 2001, 3, 624; (b) Dallinger, D.; Kappe,
C. O. Nat. Protoc. 2007, 2, 1713.
ꢀ
ꢀ
84e186 C.
Acknowledgements
11. (a) Dondoni, A.; Masso, A.; Sabbatini, S. Tetrahedron Lett. 2002, 43,
5
2
2
013; (b) Kappe, C. O.; Shishkin, O. V.; Uray, G.; Verdino, P. Tetrahedron
000, 56, 1859; (c) Bose, D. S.; Kumar, R. K.; Fatima, L. A. Synlett 2004,
79; (d) De, S. K.; Gibbs, R. A. Synthesis 2005, 1748; (e) Pasha, M. A.;
We thank the Austrian Science Fund (Project I18-N07) for
supporting this work.
Swamy, N. R.; Jayashankara, V. P. Indian J. Chem., Sect. B 2005, 44, 823;
f) Srinivas, K. V. N. S.; Das, B. Synthesis 2004, 2091; (g) Bose, D. S.;
(
References and notes
Sudharshan, M.; Chavhan, S. W. ARKIVOC 2005, iii, 228; (h) Salehi,
P.; Dabiri, M.; Zolfigol, M. A.; Fard, M. A. B. Tetrahedron Lett. 2003,
44, 2889; (i) Bose, D. S.; Chary, M. V.; Mereyala, H. B. Heterocycles
2006, 68, 1217; (j) Rafiee, E.; Jafari, H. Bioorg. Med. Chem. Lett. 2006,
16, 2463; (k) Russowsky, D.; Canto, R. F. S.; Sanches, S. A. A.; D’Oca,
M. G. M.; de Fatima, A.; Pilli, R. A.; Kohn, L. K.; Antonio, M. A.;
Ernesto de Carvalho, J. Bioorg. Chem. 2006, 34, 173; (l) Godoi, M. N.;
Costenaro, H. S.; Kramer, E.; Machado, P. S.; Montes, D.; Marcelo, G.;
Russowsky, D. Quim. Nova 2005, 28, 1010.
1
2
. For reviews on DoE, see: (a) Tye, H. Drug Discov. Today 2004, 9, 485; (b)
Gooding, O. W. Curr. Opin. Chem. Biol. 2004, 8297.
. (a) Montgomery, D. C. Design and Analysis of Experiments, 6th ed.; John
Wiley and Sons: New York, NY, 2005; (b) Haaland, P. D. Experimental
Design in Biotechnology; Marcel Dekker: New York, NY, 1989.
. Several DoE software packages are commercially available. Four of the
most commonly used being MODDE from Umetrics (www.umetrics.
com), Design Expert from Statese (www.statease.com), Fusion Pro from
S-Matix (www.smatrix.com), and JMP from the SAS Institute
3
12. For an enantioselective synthesis of monastrol, see: Huang, Y.; Yang, F.;
Zhu, C. J. Am. Chem. Soc. 2005, 127, 16386.
(
www.jmp.com).
13. Yb(OTf)
3
ꢁH
2
O (ꢂ97% purity), 5 g for 79.10V; LaCl
3
ꢁH O (ꢂ98.5%
2
4
. (a) Kappe, C. O. Angew. Chem., Int. Ed. 2004, 43, 6250; (b) Kappe, C. O.;
Stadler, A. Microwaves in Organic and Medicinal Chemistry; Wiley-VCH:
Weinheim, 2005; (c) Microwaves in Organic Synthesis, 2nd ed.; Loupy,
A., Ed.; Wiley-VCH: Weinheim, 2006.
purity), 25 g for 14.30V (Aldrich Chemical Company).
2
14. R is a statistical parameter, which defines the goodness of fit of the
2
model. The closer to 1 this value is the better the fit. Q defines the good-
ness of prediction of the model. Again this should be as close to 1 as pos-
2
2
5
. (a) Larhed, M.; Hallberg, A. Drug Discov. Today 2001, 6, 406; (b)
Wathey, B.; Tierney, J.; Lidstr o¨ m, P.; Westman, J. Drug Discov. Today
sible. Only a model with good R and Q can be reliably used in
a predictive sense.
2002, 7, 373; (c) Al-Obeidi, F.; Austin, R. E.; Okonya, J. F.; Bond,
15. Stadler, A.; Kappe, C. O. J. Chem. Soc., Perkin Trans. 2 2000, 1363.