Cytotoxic 2,6-bis(arylidene)cyclohexanones and related compounds

Article abstract of DOI:10.1016/S0223-5234(00)01173-9

A number of 2-arylidenecyclohexanones 1, 2,6-bis(arylidene)cyclohexanones 2 and related Mannich bases 3-5 were prepared. Various torsion angles as well as atomic charges on olefinic carbon atoms were determined by molecular modelling on all compounds. These molecules showed cytotoxicity towards murine P388 and L1210 cells as well as to human Molt 4/C8 and CEM T-lymphocytes. The average cytotoxicity of the dienones 2 was more than three times greater than was found with the monoarylidene analogues 1, and, in general, were slightly more cytotoxic than the Mannich bases 3-5. A number of the compounds displayed potency towards a panel of human tumour cell lines and most of the representative compounds in series 2-5 were selectively toxic to colon cancers and leukaemic cells. (C) 2000 Editions scientifiques et medicales Elsevier SAS.

Full text of DOI:10.1016/S0223-5234(00)01173-9

Eur. J. Med. Chem. 35 (2000) 967–977
´
© 2000 Editions scientifiques et me´dicales Elsevier SAS. All rights reserved
PII: S0223-5234(00)01173-9/FLA
Original article
Cytotoxic 2,6-bis(arylidene)cyclohexanones and related compounds
Jonathan R. Dimmocka*, Praveen Kumara, Adil J. Nazaralia, Narasimhan L. Motaganahallia,
Travis P. Kowalchuka, Michael A. Beazelya, J. Wilson Quailb, Eliud O. Oloob,
Theresa M. Allenc, Jennifer Szydlowskic, Erik DeClercqd, Jan Balzarinid
aCollege of Pharmacy and Nutrition, University of Saskatchewan, 110 Science Place, Saskatoon, Saskatchewan, Canada S7N 5C9
bDepartment of Chemistry, University of Saskatchewan, Saskatoon, Saskatchewan, Canada S7N 5C9
cDepartment of Pharmacology, University of Alberta, Edmonton, Alberta, Canada T6G 2H7
dRega Institute for Medical Research, Katholieke Universiteit Leuven, B-3000 Leuven, Belgium
Received 28 February 2000; accepted 5 May 2000
Abstract – A number of 2-arylidenecyclohexanones 1, 2,6-bis(arylidene)cyclohexanones 2 and related Mannich bases 3–5 were
prepared. Various torsion angles as well as atomic charges on olefinic carbon atoms were determined by molecular modelling on all
compounds. These molecules showed cytotoxicity towards murine P388 and L1210 cells as well as to human Molt 4/C8 and CEM
T-lymphocytes. The average cytotoxicity of the dienones 2 was more than three times greater than was found with the
monoarylidene analogues 1, and, in general, were slightly more cytotoxic than the Mannich bases 3–5. A number of the compounds
displayed potency towards a panel of human tumour cell lines and most of the representative compounds in series 2–5 were
´
selectively toxic to colon cancers and leukaemic cells. © 2000 Editions scientifiques et me´dicales Elsevier SAS
unsaturated ketones / Mannich bases / molecular modelling / cytotoxicity
1. Introduction
cytotoxicity to be displayed towards cancers which
are resistant to established alkylating agents. Support
for this viewpoint was obtained when two cell lines
which were resistant to the antineoplastic agent mel-
phalan were shown to be virtually free from cross
resistance to a number of Mannich bases of conju-
gated styryl ketones [5]. Thus, the preparation of a
number of prototypic enones and related compounds
as candidate cytotoxic and anticancer agents was con-
sidered a profitable avenue to pursue.
A number of years ago, the cytotoxicity of E-2-
benzylidenecyclohexanone 1a towards an epidermoid
carcinoma of the nasopharynx (KB screen) was de-
scribed in which this enone had an ED50 of 1.34 mM
[6]. Thus, elaboration to give series 1 was planned in
which the aryl substituents were chosen with diver-
gent Hammett s, Hansch p and molecular refractivity
(MR) values with a view to finding out whether
cytotoxicity was related to the electronic, hydropho-
bic or steric properties of these groups. Furthermore,
the theory of sequential cytotoxicity, which originated
The principal aim of research in this laboratory is
the discovery of novel cytotoxic and anticancer alkyl-
ating agents. The currently available antineoplastics
that act by alkylation of cellular nucleophiles suffer
from a number of significant disadvantages, many of
which are related to their interactions with nucleic
acids [1]. In contrast, various a,b-unsaturated ketones
react preferentially or exclusively with thiols but not
with amino or hydroxy groups [2, 3]. Hence, since
thiols are not part of the nucleic acid structures,
conjugated enones may be significantly less mutagenic
and carcinogenic than conventional drug strategies
using alkylating agents [4]. In addition, the mode of
action of enones by thiol alkylation should enable
* Correspondence and reprints:
E-mail address: dimmock@skyway.usask.ca
(J.R. Dimmock).

968
J.R. Dimmock et al. / European Journal of Medicinal Chemistry 35 (2000) 967–977
in this laboratory [7], states that successive chemical
attacks of cellular constituents may be highly delete-
rious to malignant cells and, thus, the preparation
and bioevaluation of the potential bis alkylators 2
was considered. In fact, several studies have shown
that various neoplasms are more vulnerable to multi-
ple chemical insults than the corresponding normal
cells [8, 9], and hence compounds with selective toxic-
ity for neoplastic tissues may evolve using this ap-
proach.
Another design of compounds with potential pref-
erential lethality to cancers was as follows. The pH of
a number of tumours is lower than the corresponding
normal cells [10] and hence an increase in the elec-
trophilicity of the olefinic carbon atoms in enones
under acidic conditions should enable greater thiola-
tion in cancerous cells to occur. An examination of
available Hammett sm values revealed that the figure
for the dimethylaminomethyl group is 0.00 while the
corresponding protonoted species — CH2NH(CH3)2
has a sm figure of 0.40 [11]. Since piperidine is more
basic than dimethylamine (pKa of 11.12 versus 10.73
[12]), the placement of a 1-piperidinylmethyl sub-
stituent in an aryl ring should give rise to an even
higher percentage of charged species under acidic
conditions than utilization of a dimethylaminomethyl
group. These considerations led to the decision to
prepare series 3.
Other prototypic molecules were considered,
namely 4 and 5a,b. Certain Mannich bases of conju-
gated styryl ketones react much more rapidly with
thiols than the precursor enones [13]. Hence, the
preparation of 4 was suggested, which should be
appreciably more cytotoxic than 1f if thiol alkylation
contributes significantly to potency. The dienone 5a
would be predicted to have greater cytotoxicity than
2f, if the presence of a 1-piperidinylmethyl substituent
markedly increases the capacity for alkylation of cel-
lular nucleophiles. Should this be the case, further
evidence of this phenomenon would be obtained if the
activity of 5a would be greater than that of the
dimethylamino analogue 5b (6ide supra).
In summary, the aim of the present investigation
was the preparation of a number of prototypic
Figure 1. Structures and synthesis of series 1–5. i=4-hydroxybenzaldehyde/hydrochloric acid; ii=dipiperidinomethane; iii=1-
methylenepiperidinium chloride; iv=N,N,N%,N%-tetramethylethylaminediamine.

J.R. Dimmock et al. / European Journal of Medicinal Chemistry 35 (2000) 967–977
969
Figure 2. ORTEP diagram of 1f.
molecules 1–5 as candidate cytotoxic agents with a
view to determining the important structural features
of the molecules that contribute to potency. The data
obtained may provide guidance for the ways in which
this area of investigation can be expanded on a ratio-
nal basis.
as well as human Molt 4/C8 and CEM T-
lymphocytes. These data are summarized in table II.
In addition, a number of compounds were examined
using a panel of approximately 55 human cancer cell
lines from a variety of neoplastic diseases. The results
are presented in table III, together with an indication
of whether selective toxicity to colon and/or
leukaemic cells was observed. The Mannich bases
3a,d were also evaluated using human Jurkat
leukaemia cells. The IC50 for 3a,d against this cell line
were 4.47 and 5.36 mmoles respectively, while the IC80
values of 7.15 and 8.57 mmoles, respectively, were
used in subsequent biochemical determinations. The
apoptotic indices [(number of cells with apoptotic
nuclei/number of cells counted)×100] for 3a,d were
50.3 and 64.7%, respectively, while 3a inhibited the
syntheses of RNA and protein by 73.2 and 28.9%,
respectively.
2. Chemistry
The compounds in series 1 were prepared by a
Claisen–Schmidt condensation between cyclohex-
anone and the appropriate aryl aldehyde. A similar
reaction using 1f in place of cyclohexanone led to
series 2. Condensation of 2a–e with dipiperidino-
methane gave rise to the corresponding Mannich
bases 3a–e, while 4 was synthesized by reaction of 1f
with the Mannich reagent 1-methylenepiperidinium
chloride. The reaction of 2f with either dipiperidino-
methane or N,N,N%,N%-tetramethylethylenediamine
led to 5a and 5b, respectively. These reaction se-
quences are illustrated in figure 1. The X-ray crystal-
lographic data obtained for 1f are displayed in figure
2. Molecular modelling was undertaken on all com-
pounds and the specific torsion angles and charge
densities determined are illustrated in figure 3, while
the details are summarized in table I.
4. Results and discussion
The synthesis of the compounds in series 1–5 was
achieved successfully. The evidence from thin-layer
3. Cytotoxicity
Figure 3. Torsion angles (u1, u2) and atomic charges (CA, CB)
of compounds 1–5 determined by molecular modelling.
Compounds 1a–g, 2a–g, 3a–c,e, 4, 5a,b were eval-
uated against murine P388 and L1210 leukaemic cells

970
J.R. Dimmock et al. / European Journal of Medicinal Chemistry 35 (2000) 967–977
Table I. Torsion angles (u1, u2) and atomic charges (CA, CB)
of compounds 1–5 determined by molecular modelling.a
1
chromatography (TLC) and H-NMR spectroscopy
indicated that the compounds were isomerically pure.
The absorptions of the olefinic protons in the H-
1
Compound
u1
9.07
9.02
8.78
8.90
9.18
u2
CA
CB
NMR spectra were located at 7.3–7.8 ppm, which is
indicative of compounds possessing the E configura-
tion [14]; for the corresponding Z isomers, these
protons would be predicted to be located at higher
field. For example, the olefinic hydrogen atoms of
Z-2-phenylmethylenecylohexanone and Z-2-phenyl-
methylene-6,6-diphenylcyclohexanone absorb at 6.27
ppm [15] and 6.22 ppm [16], respectively. In addition,
X-ray crystallography of a representative compound
1f revealed that it had the E configuration (figure 2),
which is consistent with similar studies involving a
2-arylidenecyclohexanone [17] and a 2,6-bis-(aryli-
dene)cyclohexanone [18]. The location of the 1-pipe-
ridinylmethyl and dimethylaminomethyl substituents
in series 3 and 5 ortho to the aryl hydroxy group was
1a
1b
1c
1d
1e
1f
2a
2b
2c
2d
2e
2f
3a
3b
3c
3d
3e
4
–
–
–
–
–
–
0.055
0.053
0.047
0.058
0.062
0.063
0.044
0.042
0.036
0.047
0.051
0.052
0.044
0.042
0.036
0.047
0.051
0.064
0.052
0.052
–
–
–
–
–
–
8.87
13.27
13.31
13.38
13.23
13.39
13.08
13.17
13.14
13.22
13.07
12.94
9.44
−13.06
−13.08
−13.10
−13.07
−13.06
−13.08
−14.61
−15.37
−15.40
−15.32
−14.47
–
0.053
0.055
0.056
0.053
0.052
0.052
0.054
0.055
0.056
0.052
0.053
–
1
assigned on the basis of H-NMR spectroscopy and
1
5a
5b
14.52
13.91
−15.38
−13.90
0.052
0.052
literature precedents [19, 20]. H-NMR spectroscopy
of 4 confirmed that condensation of 1f with 1-
methylenepiperidinium chloride led to substitution on
the alicyclic rather than aryl ring.
If the compounds exert their cytotoxicity by elec-
trophilic attack of cellular constituents, two structural
a The torsion angles (o) and atomic charges (esu) in series 1–5
are indicated in figure 3. Torsion angles are indicated as
positive or negative if the rotation is anticlockwise or clock-
wise, respectively.
Table II. Cytotoxicity of 1, 2, 3a–c,e, 4 and 5 towards murine P388 and L1210 cells and human Molt 4/C8 and CEM
T-lymphocytes.
Compound
IC50 (mM)
P388 cells
L1210 cells
Molt 4/C8 cells
CEM cells
Av. of four
cell lines
1a
19.2590.72
6.4890.28
3.8790.54
30.7690.27
20.5290.96
16.6590.23
3.5890.21
3.1390.43
2.9390.21
5.6990.37
3.8490.07
5.8090.68
2.0490.11
2.7990.24
2.1690.07
2.3990.07
0.16190.01
0.58190.10
1.2890.04
0.2290.01
0.4990.01
77.7914.6
10.490.3
2.9190.01
13.490.7
16.890.1
45.293.8
7.3290.87
6.4491.14
7.1590.01
7.4190.86
6.8591.18
13.592.3
6.9690.61
15.290.29
10.290.2
7.2890.57
2.6090.47
318926
38.192.3
4.2191.08
1.0690.41
13.490.2
14.890.2
39.294.6
5.1791.83
4.9090.67
7.7190.61
5.2291.72
4.1891.92
5.5292.12
7.1490.75
15.596.9
7.8290.73
7.2190.75
5.8092.25
49892
36.893.0
3.5090.17
3.3790.07
14.490.3
15.890.1
26.096.8
5.8290.83
2.6990.39
4.2392.05
4.3191.16
5.6891.65
6.8490.14
4.3590.96
8.2190.09
8.1290.59
5.6290.88
3.8592.50
31691
42.96
6.15
2.80
17.99
16.98
31.76
5.47
4.29
5.51
5.66
5.14
7.92
5.12
10.43
7.08
5.63
3.10
283
1b
1c
1d
1e
1f
2a
2b
2c
2d
2e
2f
3a
3b
3c
3e
4
5a
5b
6.2690.76
2.1390.03
0.2890.14
6.9891.60
3.2490.79
2393.0
5.8592.49
2.4790.30
8.990.43
5.09
2.02
8.17
Melphalan
5-Fluorouracil

J.R. Dimmock et al. / European Journal of Medicinal Chemistry 35 (2000) 967–977
971
features of the compounds in series 1–5 may deter-
mine the relative potencies. First, the orientation of
the aryl rings will be expected to influence the align-
ment of the molecules at critical sites of the biomacro-
molecules. Thus, the measurement of the torsion
angles u1 and u2 between the aryl rings and the
adjacent olefinic groups, as illustrated in figure 3, was
considered of interest since previous studies have
demonstrated that potency is influenced by coplanar-
ity, or lack thereof, between different groups [21].
Second, the atomic charges on carbon atoms CA and
CB (figure 3) will permit a prediction of relative po-
tencies, if thiol alkylation is the principal mode of
action of these compounds. For example, a reduction
in the electron density on the olefinic carbon atoms
would be expected to lead to increased cytotoxicity.
The data generated by molecular modelling are
summarized in table I. The average u1 value in series
1 was 8.97°, while the introduction of a second aryli-
dene group into the alicyclic ring leading to 2 led to
an approximately 50% increase in the u1 figures. The
u2 values in series 2 were virtually constant, being
slightly lower than the u1 figures. The increase in
torsion angles in 2 compared to 1 was difficult to
rationalize since a lowering of the u1 and u2 values in
2 (as well as series 3 and 5) would lead to increased
conjugation in these dienones. The presence of the
3-(1-piperidinyl)methyl group in series 3 led to a 13%
increase in the u2 figures, on average, compared to
series 2. This observation may be attributed to a
buttressing effect, i.e. the substituents in the meta
position forced the adjacent ortho protons towards
the olefinic bond, thereby increasing steric hindrance
to coplanarity of the aryl ring with the adjacent
unsaturated linkage. The percentage variation of the
u1, u2 and CB values within each in the series 1–3 was
invariably less than 8%. There was no linear correla-
tion between the average potencies of the compounds
in series 2 and the u2 values (P\0.1). However, a
linear relationship between the u2 figures of 3a–c,e
and average cytotoxicities was observed (PB0.1)
whereby potencies increased as the size of the inter-
planar angles diminished. On the other hand, the
atomic charges at CA within each of these three series
varied by 34, 44 and 42, respectively. Thus, within
each series of compounds, variation in cytotoxicity is
likely to be influenced principally by the charges on
the CA atoms. Linear plots between the atomic
charges on the CA atoms and the Hammett s values
of aryl ring A revealed that correlations occurred
within each of the series 1–3 (PB0.05) as indicated
in figure 4.
All of the compounds with the exception of 3d were
evaluated against murine P388 and L1210 cells as well
as human Molt 4/C8 and CEM T-lymphocytes. The
murine cell lines were chosen on the basis of the claim
that they are good indicators of a number of clinically
useful antineoplastic agents [22], while evaluation
Table III. Evaluation of various compounds against a panel of human tumour cell lines.
Compound
All cell lines
Colon cancer cells
Leukaemic cells
MG MIDa (mM)
MG MIDa (mM)
SIb
MG MIDa (mM)
SIb
1c
21.9
67.6
97.7
3.16
4.26
16.2
6.64
6.63
3.98
6.31
6.61
23.5
32.6
1.45
22.2
72.0
100
1.0
0.9
1.0
1.5
2.1
1.6
2.0
1.8
1.9
1.8
1.7
0.5
4.1
1.0
19.4
21.6
100
1.1
3.1
1.0
2.1
1.7
2.5
1.3
2.0
2.7
1.7
3.3
5.1
1.2
2.3
1d
1e
2c
2.15
2.07
10.3
3.27
3.76
2.13
3.49
3.97
44.6
7.90
1.42
1.53
2.56
6.43
5.03
3.38
1.46
3.63
2.02
4.65
27.6
0.620
3a
3b
3d
3e
4
5a
5b
Melphalan
5-Fluorouracil
Helenalin
a The letters MG MID refer to the mean graph midpoint values. This term is explained in Section 4.
b The letters SI indicate the selectivity index, i.e. the ratio between the MG MID figure for all cell lines and the MG MID value for
either the colon cancers or leukaemic cells.

972
J.R. Dimmock et al. / European Journal of Medicinal Chemistry 35 (2000) 967–977
lines for 1a–f and 2a–f were 19.77 and 5.67 mM,
respectively, i.e. the compounds in series 2 are 3.5
times more potent than the analogues in series 1
suggesting the viability of the theory. Furthermore,
the IC50 figures for the cell lines for 1a–c,e and 3a–c,e
were 17.22 and 7.07 mM, respectively, revealing a
2.4-fold increase in potency for the dienones, which
also supports the theory of sequential cytotoxicity.
The insertion of a 3-(1-piperidyl)methyl group into an
aryl ring in series 3 did not lead to an increase in
potency over the precursor a,b-unsaturated ketones in
series 2. Thus, the average IC50 figures of the four cell
lines for 2a–c,e and 3a–c,e were 5.10 and 7.07 mM,
respectively, indicating a nearly 40% reduction in
cytotoxicity. Similarly, the average cytotoxicities of 2f
and 5a were 7.92 and 283 mM, respectively, revealing
an even greater decrease in potency. It is conceivable
that a bulky substituent in the meta position of the
aryl rings impedes alignment at a binding site in the
cells, or that the increase in the u2 figures of 3a–c,e
and 5a, compared to the analogues in series 2, is
detrimental to cytotoxicity. As mentioned previously,
Mannich bases of conjugated styryl ketones react
against the T-lymphocytes may reveal whether cyto-
toxicity to human cancer cells is likely. The data
generated are summarized in table II along with the
relevant figures for the reference drugs melphalan and
5-fluorouracil.
A review of the data in table II indicated that most
compounds were markedly cytotoxic to the four cell
lines except 5a, which had only very weak activity in
the L1210, Molt 4/C8 and CEM screens. The most
potent compounds (potencies compared to melphalan
and 5-fluorouracil, respectively) in the P388, L1210,
Molt 4/C8 and CEM screens were 4 (1.4, 3.0), 4 (0.82,
0.11), 1c (3.1, 21.7) and 2b (0.92, 3.3), respectively.
Hence, development of analogues of 1c, 2b and 4 is
warranted. If the potencies of compounds 1–5 are
mediated by interaction with thiols on the olefinic
carbon atoms, then series 2 would be expected to be
twice as potent as series 1, having double the number
of alkylation sites. However, if the theory of sequen-
tial cytotoxicity is validated in this cluster of com-
pounds, then the average IC50 value of 2a–f should be
substantially less than half of the average IC50 figure
for 1a–f. The average IC50 figures for the four cell
Figure 4. Correlations between the average potencies of 1, 2 and 3a–c,e and the Hammett s values. The triangles represent the data
for series 1 while the results for series 2 and 3a–c,e were superimposable and are presented as squares.

J.R. Dimmock et al. / European Journal of Medicinal Chemistry 35 (2000) 967–977
973
much more rapidly with thiols than the analogous
unsaturated ketones [13]. The fact that compound 4
was 10 times more potent than the precursor ketone
1f supports the view that an important mode of
action of these compounds is based upon interference
with cellular thiols. In addition, interaction of the
piperidyl group of 4 with a binding site may account
for its greater potency than 1f.
In order to seek correlations between the potencies
of various groups of compounds and the electronic,
hydrophobic and steric properties of the aryl sub-
stituents, linear plots were made between the average
IC50 figures of 1a–f, 2a–f, 3a–c,e and the Hammett
s, Hansch p and MR constants in each series of
compounds. The test for zero correlation [23] was
employed. The potencies of 1a–f were correlated with
the s (PB0.1), p (PB0.05) and MR (PB0.05) con-
stants and, in addition, the cytotoxicity of 2a–f was
related to the p constants (PB0.1). These correla-
tions were all positive, i.e. potencies increased as the
physicochemical constants rose. Thus, development of
series 1 and 2 with the aim of increasing cytotoxicity
should bear in mind these interrelationships. No other
correlations (P\0.1) were noted.
Over half of the compounds were evaluated against
a panel of approximately 55 human tumour cell lines
from different groups of neoplasms, including
leukaemia, melanoma, non-small cell lung, colon,
central nervous system, ovarian, renal, prostate and
breast cancers. The concentrations of compounds
used in this assay were normally log 10−8 to log 10−4
M, and the amounts of compounds required to inhibit
the growth of the cells by 50% were noted. Mean
graphs were constructed indicating the sensitivities of
individual tumours to the compounds relative to the
average IC50 figures for all cell lines. However, if the
IC50 values were greater than log 10−4 M for one or
more tumours, this figure of log 10−4 M was used in
calculating the average cytotoxicity towards all cell
lines. Hence, in some cases, the average IC50 figures
against all cell lines were not generated and are conse-
quently referred to as mean graph midpoint (MG
MID) values [24].
agent helenalin, whose mode of action is reported to
be by thiol alkylation [25]. The observation that the
potencies of the dienones 2c and 3d,e are appreciably
more than double those of the analogues 1c–e having
capacity for monoalkylation supports the hypothesis
of sequential cytotoxicity. An examination of the
mean graphs [24] revealed that many of the com-
pounds demonstrated greater potencies to colon can-
cers and leukaemia than towards other cell lines.
Thus, a comparison of the average cytotoxicity figures
for either colon or leukaemia cell lines with the MG
MID values for all tumours gave the selectivity index
(SI) for each compound. With the exception of 1c–e,
all of the compounds had a SI figure of 1.50 or
greater when colon tumours were considered. This
criterion for preferential cytotoxicity, i.e. SI=1.50 or
more, was displayed against leukaemia by all the
compounds except for 1c,e and 3d. Thus, 94% of the
compounds in series 2–5 that were evaluated in this
screen displayed a 50% or more increased toxicity to
both colon and leukaemic neoplasms. Hence, these
compounds are templates for molecular modification
with a view to obtaining analogues with even greater
selective toxicity to these important groups of cancers.
A number of anticancer agents act by inducing
apoptosis [26] as well as by inhibiting the biosynthesis
of RNA and proteins [27]. Work from this laboratory
using human Jurkat T cells revealed that cytotoxicity
could be explained, at least in part, by these mecha-
nisms of action for various conjugated arylidene ke-
tones [14] and a Mannich base of an unsaturated
ketone [28]. In the present investigation, the IC80
concentrations of compounds were used in order to
cause the death of the majority of the Jurkat T cells.
Under these conditions, the apoptotic indices of two
representative compounds 3a,d were 50 and 65%,
respectively. A further investigation with 3a revealed
that RNA and protein syntheses were inhibited by 73
and 29%, respectively. The reference compounds mel-
phalan (an antineoplastic alkylating agent), actino-
mycin D (an inhibitor of RNA synthesis) and
cycloheximide (a protein synthesis inhibitor) also
caused apoptosis and were inhibitors of RNA and
protein syntheses, details of which are recorded in
Section 6 of this paper. The conclusion to be drawn is
the likelihood that the cytotoxicity of the compounds
prepared in this study is based upon induction of
apoptosis and interference with the biosyntheses of
RNA and proteins.
The cytotoxicity of various compounds in series
1–5, as well as three reference compounds, is pre-
sented in table III. The data reveal that, with the
exception of 1c–e, the compounds are, on average,
approximately 3.5 and 5 times more potent than
melphalan and 5-fluorouracil, respectively. The com-
pounds were less cytotoxic than the antineoplastic

974
J.R. Dimmock et al. / European Journal of Medicinal Chemistry 35 (2000) 967–977
5. Conclusions
water–methanol to give 1f, m.p. 172–173 °C (lit. [31],
m.p. 172 °C) in 27% yield. TLC using a solvent system
of cyclohexane:ether:methanol (5:3:2) revealed it to be a
single compound.
The investigation has revealed that a number of
conjugated arylidenecycloalkanones and related Man-
nich bases have promising cytotoxic properties. In
particular, 1c, 2b and 4 displayed significant activity
towards P388, L1210, Molt 4/C8 and CEM cells. All
of the compounds in series 2–5 that were evaluated
against a panel of human tumour cell lines were more
potent than the clinically useful drugs melphalan and
5-fluorouracil. Furthermore, in this screen, selective
toxicity to colon cancers and leukaemia was demon-
strated; this observation alone mandates that further
molecular modification of these compounds is war-
ranted. The modes of action of these compounds
likely include the induction of apoptosis and inhibi-
tion of RNA and protein syntheses.
6.1.2. Synthesis of series 2
Compounds 2a–e were prepared as follows. Hydro-
gen chloride was passed into a solution of the 2-aryli-
denecyclohexanone (0.02 mol) and 4-hydroxybenzal-
dehyde (0.02 mol) in a mixture of ether (100 mL) and
methanol (5 mL) for 35–40 min at room temperature.
After stirring at room temperature for 1 h, the solvent
was removed and the dark red precipitates which
formed were collected and washed with aqueous sodium
bicarbonate solution (0.1 M, 50 mL). Recrystallization
from methanol afforded the following 2,6-bis(aryli-
dene)cyclohexanones [m.p., yield (%)] namely 2a: 203–
208, 78; 2b: 227–229, 66; 2c: 164, 38; 2d: 188, 43 and 2e:
207–211, 59.
6. Experimental protocols
The preparation of 2f was accomplished by the fol-
lowing procedure. Hydrogen chloride was passed into a
solution of cyclohexanone (0.05 mol) and 4-hydroxyben-
zaldehyde (0.1 mol) in methanol (50 mL) for 1 h at
room temperature. After stirring at room temperature
for 1 h, evaporation of the solvent led to yellow precip-
itates which were filtered and washed with sodium bicar-
bonate solution (0.1 M, 50 mL). Recrystallization from
methanol gave 2f, m.p. 279–282 °C in 82% yield.
TLC using solvent systems of benzene:cyclohexane:
methanol (8:1:1) for 2a–e and cyclohexane:ether:
methanol (5:3:2) for 2f revealed the compounds to be
6.1. Chemistry
Melting points are uncorrected and are quoted in
degrees Celcius. Elemental analyses were undertaken on
1f, 2a–f (C,H) and 3a–e, 4 and 5a,b (C,H,N) by Mr K.
Thoms, Department of Chemistry, University of
Saskatchewan, and are within 0.4% of the calculated
values. TLC was employed using silica gel 60 F254-pre-
1
coated plastic sheets. H-NMR spectroscopy utilized a
Bruker AM 300FT instrument (300 MHz). Compounds
1f and 3c were obtained with one quarter and one half
the molecules of water of crystallization, respectively.
1
homogeneous. The H-NMR spectrum of a representa-
tive compound 2b in which the aryl rings containing the
chloro and hydroxy groups are referred to as rings A
and B, respectively, was as follows: l (CDCl3): 7.73 (d,
2H, 2,6 aryl A H, J=9.94 Hz), 7.60–7.35 (m, 6H, 3,5
aryl A H, 2,6 aryl B H, olefinic H), 6.87 (d, 2H, 3,5 aryl
B H, J=8.58 Hz), 2.90 (m, 4H, 3,5 cyclohexyl H), 1.82
(m, 2H, 4 cyclohexyl H).
6.1.1. Synthesis of series 1
The synthesis of 1a–e has been described previously
[29]. The preparation of 1f was undertaken by a modifi-
cation of a literature method [30] as follows. A solution
of 4-hydroxybenzaldehyde (0.1 mol) in aqueous sodium
hydroxide solution (18%, 25 mL) was added slowly to a
mixture of cyclohexanone (0.5 mol) and aqueous sodium
hydroxide solution (14%, 25 mL). After dilution with
water (300 mL), the reaction mixture was stirred at
room temperature for 100 h, then acidified with hydro-
chloric acid (1 M, 300 mL) and extracted with chloro-
form (3×30 mL). The combined organic extracts were
washed with water and dried (anhydrous magnesium
sulfate). After evaporation of some of the solvent (ꢀ20
mL), cyclohexane (100 mL) was added and the resultant
precipitate was collected, dried and recrystallized from
6.1.3. Synthesis of series 3
A solution of dipiperidinomethane (0.01 mol) in ace-
tonitrile (10 mL) was added to a solution of the 2,6-
bis(arylidene)cyclohexanone
2
(0.0075 mol) in
acetonitrile (150 mL) at 70–80 °C. The mixture was
heated under reflux and the progress of the reaction was
monitored by TLC using a solvent system of ethyl
acetate:chloroform:methanol (6:3:1). After 24 h, the re-
action mixture was concentrated to 25 mL and the
resultant yellow precipitates were collected and recrys-

J.R. Dimmock et al. / European Journal of Medicinal Chemistry 35 (2000) 967–977
975
tallized from methanol to give the following Mannich
bases [m.p., yield (%)]: 3a: 97, 42; 3b: 157, 70; 3c:
156–157, 40; 3d: 141, 65 and 3e: 173, 32. The com-
pounds were homogeneous by TLC using the same
solvent system as employed in monitoring the reaction.
After 18 h, the mixture was concentrated to 20 mL and
the resultant yellow precipitates were crystallized from
acetone–chloroform (5a) or methanol (5g) to give the
following compounds [m.p., yield (%)]: 5a: 228–232, 82;
5b: 170–173, 49. TLC revealed that the compounds
were homogeneous using the same solvent system as
employed in monitoring the reaction. The 1H-NMR
spectrum of a representative compound 5a was deter-
mined in which the aryl protons ortho to the 1-pipe-
ridinylmethyl, hydroxy and olefinic groups are designat-
ed as the aryl 2, 5 and 6 protons, respectively. l
(CDCl3): 7.70 (s, 2H, olefinic H), 7.35 (dd, 2H, aryl 5H,
J=8.75 Hz), 7.18 (s, 2H, aryl 2H), 6.82 (d, 2H, aryl 6H,
J=8.37 Hz), 3.69 (s, 4H, aryl-CH2-N), 2.90 (t, 4H,
3,5-cyclohexyl H), 2.08–2.80 (br s, 8H, 2,6-piperidinyl
H), 1.79 (quintiplet, 2H, 4-cyclohexyl H), 1.22–1.73 (br
m, 12H, 3,4,5-piperidyl H).
1
The H-NMR spectrum of a representative compound
3d was determined in which the aryl rings containing the
methyl and hydroxy groups are referred to as rings A
and B, respectively. l (CDCl3): 7.76 (s, 1H, CH=at-
tached to ring A), 7.72 (s, 1H, CH=attached to ring B),
7.36 (def d, 3H, 2,6 aryl A H, 6 aryl B H), 7.20 (d, 2H,
3,5 aryl A H, J=8.03 Hz), 7.13 (s, 1H, 2 aryl B H), 6.83
(d, 1H, 5 aryl B H), 3.70 (s, 2H, aryl B–CH2–N), 2.91
(t, 4H, 3,5 cyclohexyl H), 2.12–2.78 (broad s, 4H, 2,6
piperidinyl H), 2.37 (s, 3H, CH3 of aryl A), 1.79 (quin-
tiplet, 2H, 4 cyclohexyl H), 1.64 (def t, 4H, 3,5 pipe-
ridinyl H), 1.18–1.57 (broad s, 2H, 4 piperidyl H).
6.1.6. Charge density calculations
6.1.4. Synthesis of 4
The structures of the compounds 1–5 were built and
minimized using a HyperChem molecular modelling
programme [33]. The molecules were minimized using
the MM+ molecular mechanics module with the Po-
lak–Ribiere algorithm. A minimum energy difference
between cycles of less than 0.001 kcal/mol was achieved
for all compounds except for 1a,c,d,f and 2b,d, in which
cases the energy differences between conformations were
in the range of 0.0015–0.0380 kcal/mol. In order to
obtain the atomic charges, the SNDO semi-empirical
force field was used to calculate the wave functions of
the minimum energy conformers from which the charges
listed in table I were generated.
1-Methylenepiperidinium chloride (0.006 mol), pre-
pared by a literature procedure [32], was added to a
solution of 1f (0.003 mol) in acetonitrile (50 mL) at
40–50 °C and the mixture was heated under reflux. The
progress of the reaction was monitored by TLC using a
solvent system of ethyl acetate:methanol (7:3). After
heating under reflux for 2–3 h, the mixture was concen-
trated to 15 mL and the colourless crystals were col-
lected and recrystallized from ether–methanol to give
2-(4-hydroxyphenylmethylene)-6-(1-piperidinylmethyl)-
cyclohexanone hydrochloride 4, m.p. 168–170 °C in
64% yield. The compound was homogeneous by TLC
using the same solvents as employed in monitoring the
reaction. 1H-NMR (DMSO-d6): l 10.11 (def s, 1H,
OH), 9.91 (br s, 1H, NH), 7.36 (d, 2H, 2,6 aryl H,
J=8.24 Hz), 7.32 (s, 1H, olefinic H), 6.86 (d, 2H, 3,5
aryl H, J=8.51 Hz), 3.42–3.50 (m, 3H, cyclohexyl 3H,
piperidinyl 2H, 6H), 2.97 (m, 5H, cyclohexyl 3H, pipe-
ridinyl 2H, 6H, NCH2), 2.68–2.70 (m, 1H, cyclohexyl
6H), 2.26 (m, 1H, piperidinyl 4H), 1.64–1.86 (m, 8H,
cyclohexyl 4-CH2, 5-CH2; piperidinyl 3-CH2, 5CH2),
1.41 (m, 1H, piperidinyl 4H).
6.1.7. Statistical analyses
The s, p and MR values used in the statistical analy-
ses were taken from the literature [34].
6.1.8. X-ray crystallographic determination of 1f
Data collection was undertaken using an Enraf-Non-
ius CAD-4 diffractometer with an v scan. The structure
was solved by direct methods using NRCVAX [35] and
refined using SHELXL 97 [36]. Figure 2 was generated
using ORTEP [37] as implemented in XTAL 3.6 [38].
Atomic scattering factors and anomalous dispersion cor-
rections were obtained from the literature [39]. Non-hy-
drogen atoms were found on the E map and refined
anisotropically. Hydrogen atoms were placed by geome-
try and are not refined.
6.1.5. Synthesis of series 5
A solution of dipiperidinomethane or N,N,N%,N%-tetra-
methylethylenediamine (0.01 mol) in acetonitrile (10
mL) was added to a solution of 2f (0.0025 mol) in
acetonitrile (90 mL) at 70–80 °C and the mixture was
heated under reflux. The reaction was monitored by
TLC using a solvent system of ether:methanol (4:6).
Crystal data were as follows: C13H14O2, Mr=202.24,
0.3×0.3×0.04 mm3, orthorhombic, space group=

976
J.R. Dimmock et al. / European Journal of Medicinal Chemistry 35 (2000) 967–977
,
,
P21 a b. a=7.4879(10) A, b=11.2216(10) A, c=
2.07, 1.82–3.29; 3d: 6.03–7.25, 3.01–3.52, 3.67–6.38; 4:
2.95–5.01, 1.88–2.38, 0.887–2.03; melphalan: 19.1–26.9,
41.7–49.9, 2.82–7.16; 5-fluorouracil: 12.0–56.2, 3.09–
14.8, 8.05–52.9; and 3e: 3.72–12.3, 2.44–5.50, 1.73–4.47.
3
,
,
12.7460(10) A, V=1 071.00(19) A , Z=4, Dx=1.254
3
−1
,
mg/m , l (Mo Ka)=0.70930 A, m=0.083 mm
,
F(000)=432, T=288(2) K, qmax=25.0°, −85h51,
05k513, −155l55. A total of 1 114 reflections were
measured, of which 1 023 [R(int)=0.0091] were indepen-
dent. Refinement on F2, R1 and wR2 [F2\2s (F2)]=
0.0329 and 0.0721, respectively, R1 and wR2 (all
data)=0.0954 and 0.0849, respectively. Parameters
refined=137, w=1/[|2(F2o )+(0.0380P)2+0.07P], where
6.2.2. Measurement of apoptotic indices and inhibition
of RNA and protein syntheses in human Jurkat T cells
The cytotoxicity and apoptosis studies of 3a,d, as well
as the reference drugs melphalan, actinomycin D and
cycloheximide, were determined by a literature method
[28], except that nigrosin dye was used in place of trypan
blue in the cytotoxicity experiments and a solution of
ethidium bromide (0.1% w/v) and acridine orange (0.1 %
w/v) was used in the measurements of apoptosis. Inhibi-
tion of RNA and protein syntheses was undertaken by a
method which has been reported previously [43]. Deter-
minations were made at the end of 48 h (IC50 and IC80
experiments), 24 h (apoptosis studies) or 8 h (inhibition of
RNA and protein syntheses). The IC50 (mmol), IC80
(mmol), apoptotic index (%), inhibition of RNA synthesis
(%) and inhibition of protein synthesis (%) figures for
these compounds were as follows: 3a: 4.47, 7.15, 50.3,
73.294.71, 28.998.57; 3d: 5.36, 8.57, 64.7, –, –; mel-
phalan: 2.20, 3.52, 63.2, 85.194.71, 17.394.94; actino-
mycin D: 0.0045, 0.0072, 35.5, 90.592.15, 49.291.53;
and cycloheximide: 0.93, 1.54, 38.3, 93.393.56, 71.99
3.52.
P=(F2o +2 Fc2)/3; final (D/s)max=0.000. Dr in the final
difference map within +0.114 and −0.138 eA . Inter-
molecular hydrogen bonds produce spirals of molecules.
3
,
6.2. Screening
6.2.1. Cytotoxicity e6aluations
The examination of the cytotoxicity of various com-
pounds towards murine P388D1 cells was undertaken by
a literature procedure [40]. The method employed for
determining the bioactivity towards murine L1210 cells
and human Molt 4/C8 and CEM T-lymphocytes has been
described previously [41].
The procedure for evaluating compounds against a
panel of human tumour cell lines utilized a reported
procedure [42]. In this latter assay, the following neoplas-
tic cell lines were employed: leukaemia, melanoma, non-
small cell lung, colon, central nervous system, ovarian
and renal neoplasms. In addition, 1c,d,e were evaluated
against small cell lung tumour cell lines and, in the case of
the remaining compounds (including melphalan, 5-
fluorouracil and helenalin), prostate and breast cancers
were also employed. The highest concentration of com-
pounds was log 10−4 M except for one of the determina-
tions of 3e, whereby the maximum concentration used for
most of the cell lines was log 10−5 M. In addition, the
highest concentrations for melphalan and 5-fluorouracil
were log 10−3.6 and log 10−2.6 M, respectively. The
numbers of cell lines whose growth was inhibited by 50%
or more at the maximum concentration of compound
used per total number of cell lines were as follows: 1c:
46/46; 1d: 27/47; 1e: 2/47; 2c: 55/55; 3a: 55/55, 59/59; 3b:
56/56; 3d: 55/55, 58/59; 3e: 56/56, 59/59, 45/57, 57/57; 4:
56/56, 59/59; 5a: 57/57; 5b: 54/54; melphalan: 55/55,
59/59, 57/57; 5-fluorouracil: 44/50, 54/57, 56/58 and
helenalin: 51/51. The ranges of the MG MID figures for
all cell lines, colon cancer cells and leukaemic cells,
respectively, when the assays were conducted in duplicate
(3a,d and 4), triplicate (melphalan and 5-fluorouracil) or
quadruplicate (3e) were as follows: 3a: 3.02–5.50, 2.06–
Acknowledgements
The authors thank the following organizations for
providing financial assistance for the work described in
this study: Hoechst Marion Roussel Canada Research
Inc. (JRD, PK), CoCensys Inc, CA, USA (JRD, PK,
NLM, TPK), Medical Research Council of Canada
(MAB), Natural Sciences and Engineering Research
Council (AJN, JWQ, EOO), National Cancer Institute
(TMA, JS) and the Belgian Fonds voor Geneeskundig
Wetenschappelijk Onderzoek (JB, EDC). In addition,
gratitude is due to the National Cancer Institute, USA,
who provided the biodata presented in table III. We
thank Mrs. Lizette van Berckelaer for excellent technical
assistance.
References
[1] Silverman R.B., The Organic Chemistry of Drug Design and
Drug Action, Academic Press, Inc, San Diego, 1992, pp. 220–
222.

J.R. Dimmock et al. / European Journal of Medicinal Chemistry 35 (2000) 967–977
977
[2] Baluja A., Municio A.M., Vega S., Chem. Ind. (1964) 2053–
2054
sign, New Age International (P), Limited, New Delhi, 1997, pp.
73–74.
[3] Dimmock J.R., Raghavan S.K., Logan B.M., Bigam G.E., Eur.
J. Med. Chem. 18 (1983) 248–254.
[22] Suffness M., Douros J., in: DeVita V.T., Jr., Busch H. (Eds.),
Methods in Cancer Research, Volume XVI, Part A, Academic
Press, New York, 1979, pp. 84–85.
[4] Benvenuto J.A., Connor T.H., Monteith D.K., Laidlaw J.L.,
Adams S.C., Matney T.S., Theiss J.C., J. Pharm. Sci. 82 (1993)
988–991.
[23] Bolton S., Pharmaceutical Statistics: Practical and Clinical Ap-
plications, Second Edition, Marcel Dekker, New York, 1990,
pp. 250–252.
[5] Dimmock J.R., Kumar P., Quail J.W., Pugazhenthi U., Yang
J., Chen M., Reid R.S., Allen T.M., Kao G.Y., Cole S.P.C.,
Batist G., Balzarini J., De Clercq E., Eur. J. Med. Chem. 30
(1995) 209–217.
[24] Grever M.R., Schepartz S.A., Chabner B.A., Semin. Oncol. 19
(1992) 622–638.
[25] Hall I.H., Lee K.-H., Mar E.C., Starnes C.O., Waddell T.G., J.
Med. Chem. 20 (1977) 333–337.
[6] Dimmock J.R., Hamon N.W., Hindmarsh K.W., Sellar A.P.,
Turner W.A., Rank G.H., Robertson A.J., J. Pharm. Sci. 65
(1976) 538–543.
[26] Tsurusawa M., Saeki K., Fujimoto T., Int. J. Hematol. 66
(1997) 79–88.
[7] Dimmock J.R., Sidhu K.K., Chen M., Reid R.S., Allen T.M.,
Kao G.Y., Truitt G.A., Eur. J. Med. Chem. 28 (1993) 313–322.
[27] Foye W.O., Sengupta S.K., in: Foye W.O., Lemke T.L.,
Williams D.A. (Eds.), Principles of Medicinal Chemistry, 4th
Edition, Williams and Wilkins, Media, Pennsylvania, 1995, p.
829.
[8] Tsutsui K., Komuro C., Ono K., Nishidai T., Shibamoto Y.,
Takahashi M., Abe M., Int. J. Radiat. Oncol. Biol. Phys. 12
(1986) 1183–1186.
[28] Vashishtha S.C., Nazarali A.J., Dimmock J.R., Cell. Mol.
Neurobiol. 18 (1998) 437–445.
[9] Mitchell J.B., Russo A., Br. J. Cancer (suppl. VIII) 55 (1987)
96–104.
[29] Dimmock J.R., Sidhu K.K., Chen M., Li J., Quail J.W., Allen
T.M., Kao G.Y., J. Pharm. Sci. 83 (1994) 852–858.
[10] Wike-Hooley J.L., Haveman J., Reinhold H.S., Radiother.
Oncol. 2 (1984) 343–366.
[30] Papa D., Schwenk E., Villani F., Klingsberg E., J. Am. Chem.
Soc. 70 (1948) 3356–3360.
[11] Perrin D.D., Dempsey B., Serjeant E.P., pKa Prediction for
Organic Acids and Bases, Chapman and Hall, London, 1981, p.
116.
[31] Iyer R.N., Gopalchari R., Ind. J. Chem. 14B (1976) 194–198.
[32] Bohme H., Hartke D., Chem. Ber. 93 (1960) 1305–1309.
[12] Albert A., Serjeant E.P., The Determination of Ionization
Constants, Third Edition, Chapman and Hall, London, 1984,
p. 151.
[33] HyperChem Release 5.0 for Windows (1996), Hypercube, Inc.,
Waterloo, Ontario, Canada.
[13] Dimmock J.R., Smith L.M., Smith P.J., Can. J. Chem. 58
(1980) 984–991.
[34] Hansch C., Leo A.J., Substituent Constants for Correlation
Analysis in Chemistry and Biology, John Wiley and Sons, New
York, 1979, p. 49.
[14] Dimmock J.R., Kandepu N.M., Nazarali A.J., Kowalchuk
T.P., Motaganahalli N., Quail J.W., Mykytiuk P.A., Audette
G.F., Prasad L., Perje´si P., Allen T.M., Santos C.L., Szyd-
lowski J., De Clercq E., Balzarini J., J. Med. Chem. 42 (1999)
1358–1366.
[35] Gabe E.J., LePage Y., Charland J.-P., Lee F.L., White P.S., J.
Appl. Cryst. 22 (1989) 384–387.
[36] Sheldrick G.M., SHELXL 97, Programme for the Refinement
of Crystal Structures, University of Go¨ttingen, Germany, 1997.
[15] Smith P.J., Dimmock J.R., Turner W.A., Can. J. Chem. 51
(1973) 1458–1470.
[37] Johnson C.K., Ortep II Report ORNL-5138, Oak Ridge Na-
tional Laboratory, TN, USA, 1976.
[16] Hassner A., Mead T.C., Tetrahedron 20 (1964) 2201–2210.
[38] Hall S.R., King G.S.D., Stewart J.M., Xtal 3.6 System, Univer-
sity of Western Australia, Perth, Australia.
[17] Jia Z., Quail J.W., Dimmock J.R., Acta Crystallogr., Sect. C 46
(1990) 2467–2468.
[39] International Tables for X-ray Crystallography, Vol IV,
Kynoch Press, Birmingham, UK, 1974.
[18] Jia Z., Quail J.W., Arora V.K., Dimmock J.R., Acta Crystal-
logr., Sect. C 45 (1989) 285–289.
[40] Phillips O.A., Nelson L.A., Knaus E.E., Allen T.M., Fathi-
Afshar R., Drug Des. Deliv. 4 (1989) 121–127.
[19] Tramontini M., Angiolini L., Mannich Bases: Chemistry and
Uses, CRC Press, Boca Raton, Florida, 1994, p. 33.
[41] Balzarini J., De Clercq E., Mertes M.P., Shugar D., Torrence
P.F., Biochem. Pharmacol. 31 (1982) 3673–3682.
[20] Dimmock J.R., Erciyas E., Kumar P., Hetherington M., Quail
J.W., Pugazhenthi U., Arpin S.A., Hayes S.J., Allen T.M.,
Halleran S., De Clercq E., Balzarini J., Stables J.P., Eur. J.
Med. Chem. 32 (1997) 583–594.
[42] Boyd M.R., Paull K.D., Drug Dev. Res. 34 (1995) 91–109.
[43] Martin S.J., Lennon S.V., Bonham A.M., Cotter T.G., J.
Immunol. 145 (1990) 1859–1867.
[21] Pandeya S.N., Dimmock J.R., An Introduction to Drug De-
.