Cytotoxic and anticonvulsant aryloxyaryl Mannich bases and related compounds

Article abstract of DOI:10.1016/j.ejmech.2003.09.011

A series of 1-(4-aryloxyphenyl)-3-diethylamino-1-propanone hydrochlorides 3a-3e and related compounds 3f, 3g and 4a-4d were synthesised. In addition, a group of 4-(4-aryloxyphenyl)-3-(4-aryloxyphenylcarbonyl)-1-ethyl-4-piperidinol hydrochlorides 6a-6e were prepared which incorporated most of the structural features of 3a-3e. All of these compounds displayed cytotoxic properties towards murine L1210 cells as well as human Molt 4/C8 and CEM T-lymphocytes. A number of these compounds possessed noteworthy potencies towards seven human colon cancer cell lines. Some correlations were noted between the IC ₅₀ values generated in the different screens and the σ, π and molar refractivity constants of the aryl substituents as well as with the volumes and solvent accessible surface areas of various basic groups. Molecular modelling of representative compounds revealed structural features, which may have contributed to the varying potencies noted. In general, the compounds in series 6 were well tolerated when administered to mice. Anticonvulsant properties were demonstrated by a number of compounds in the maximal electroshock (MES) screen when administered intraperitoneally to mice while 4c and 6e afforded protection in the MES test when given orally to rats.

Full text of DOI:10.1016/j.ejmech.2003.09.011

European Journal of Medicinal Chemistry 39 (2004) 27–35
www.elsevier.com/locate/ejmech
Original article
Cytotoxic and anticonvulsant aryloxyaryl Mannich bases
and related compounds
Sarvesh C. Vashishtha a, Gordon A. Zello a, Kurt H. Nienaber b, Jan Balzarini c,
Erik De Clercq c, James P. Stables d, Jonathan R. Dimmock a,*
a College of Pharmacy and Nutrition, University of Saskatchewan, 110 Science Place, Saskatoon, S7N 5C9, Sask., Canada
b Department of Biochemistry, University of Saskatchewan, 107 Wiggins Road, Saskatoon, S7N 5E5, Sask., Canada
c Rega Institute for Medical Research, Katholieke Universiteit Leuven, 3000 Leuven, Belgium
d National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD 20892-9020, USA
Received 10 July 2003; received in revised form 6 September 2003; accepted 10 September 2003
Abstract
A series of 1-(4-aryloxyphenyl)-3-diethylamino-1-propanone hydrochlorides 3a–3e and related compounds 3f, 3g and 4a–4d were
synthesised. In addition, a group of 4-(4-aryloxyphenyl)-3-(4-aryloxyphenylcarbonyl)-1-ethyl-4-piperidinol hydrochlorides 6a–6e were
prepared which incorporated most of the structural features of 3a–3e. All of these compounds displayed cytotoxic properties towards murine
L1210 cells as well as human Molt 4/C8 and CEM T-lymphocytes. A number of these compounds possessed noteworthy potencies towards
seven human colon cancer cell lines. Some correlations were noted between the IC50 values generated in the different screens and the r, p and
molar refractivity constants of the aryl substituents as well as with the volumes and solvent accessible surface areas of various basic groups.
Molecular modelling of representative compounds revealed structural features, which may have contributed to the varying potencies noted. In
general, the compounds in series 6 were well tolerated when administered to mice. Anticonvulsant properties were demonstrated by a number
of compounds in the maximal electroshock (MES) screen when administered intraperitoneally to mice while 4c and 6e afforded protection in
the MES test when given orally to rats.
© 2003 Elsevier SAS. All rights reserved.
Keywords: Mannich bases; Cytotoxicity; Molecular modelling; Structure–activity relationships; Murine toxicity; Anticonvulsant activity
1. Introduction
[6], extrusion of dimethylamine hydrobromide followed by
thiolation likely accounts for the formation of 2 [7]. In
addition, various members of series 1 inhibited respiration in
isolated rat mitochondria [5]. These modes of action, namely
thiol alkylation and inhibition of mitochondrial respiration,
are different from the principal biochemical mechanisms
displayed by currently available anticancer drugs whether the
biological responses are beneficial effects or unwanted tox-
icity. Thus, Mannich bases such as 1 and related compounds
have the potential to possess two advantages over contempo-
rary anticancer medications. First, if the bioactivities of Man-
nich bases of aryl alkyl ketones are due to their being pro-
drugs of conjugated enones, interactions with nucleic acids
should be absent. This prediction is based on the affinity of
a,b-unsaturated ketones for thiols in contrast to hydroxy and
amino groups, which are present in DNA and RNA [8,9].
Hence, the genotoxic side effects of traditional antineoplastic
agents, such as the alkylating agents, should be absent. Sec-
A major emphasis in our laboratories is on the design and
syntheses of various Mannich bases as candidate cytotoxic
and anticancer agents [1]. In particular, several 1-aryl-3-
dialkylamino-1-propanone hydrohalides 1 displayed cyto-
toxic properties towards various murine and human tumour
cell lines [2–4]. Investigations into the modes of action of
representative compounds in series 1 disclosed the following
results. Incubation of 1-(4-methoxyphenyl)-3-dimethyla-
mino-1-propanone hydrobromide (1, R1 = OCH3, R2 = H,
R3 = CH3, X = Br) with 2-mercaptoethanol (simulating a
cellular constituent containing mercapto and hydroxy
groups) led to the formation of the corresponding thioether 2
[5]. Since Mannich bases often readily undergo deamination
* Corresponding author.
E-mail address: dimmock@skyway.usask.ca (J.R. Dimmock).
© 2003 Elsevier SAS. All rights reserved.
doi:10.1016/j.ejmech.2003.09.011

28
S.C. Vashishtha et al. / European Journal of Medicinal Chemistry 39 (2004) 27–35
R2
ring. Should the biodata of these compounds prove to be
R3
R3
CH CH SCH CH OH
2
H3CO
C
O
R1
C
O
noteworthy, the relative positions of the proximal and distal
aryl rings could be altered by changing the nature of the
spacer group. These considerations led to the decision to
launch an investigation involving the preparation of a number
of prototypic molecules. Thus, the preparation and bioevalu-
ation of series 3 was planned.
2
2
2
CH2CH2N
HX
1
2
R2
R2
(B)
HO
R1
R1
C
O
An additional line of investigation involved the syntheses
of the compounds in series 4 for the following reasons. If
antineoplastic activity was displayed by the compounds in
series 3, the result could be due principally to the intact
molecules or to the a,b-unsaturated ketones released on
deamination. In the first case, i.e. if cytotoxicity is due to the
Mannich bases per se, then interaction of the basic group (or
the related protonated species) with a binding site is likely.
This interaction will be controlled to a significant extent by
the sizes of the basic or protonated basic groups, which in
turn would be predicted to affect potencies. The sizes of the
diethylamino, dimethylamino, pyrrolidino, piperidino and
morpholino groups (the corresponding protonated species
are given in parentheses) which are found in 3b and 4a–4d,
respectively, were determined by molecular modelling and
found to be 87.5 (89.6), 55.5 (57.8), 78.0 (80.1), 86.2 (88.4)
and 94.9 (96.5) Å3, respectively. In addition, the solvent
accessible surface area figures for these five substituents
(protonated forms are in parentheses) also determined by
modelling are 234.4 (247.9), 163.6 (184.9), 196.7 (218.5),
208.8 (229.9) and 220.0 (241.1), respectively. Thus investi-
gations pertaining to possible correlations between the sizes
of the terminal groups in 3b and 4a–4d with potencies was
planned. On the other hand, if the antineoplastic activity
would be principally due to the a,b-unsaturated ketone de-
H
N
HCl
C2H5
(A)
5
Fig. 1. Structures of the compounds in series 1, 2 and 5.
ond, the differences in the mode of action of 1 and analogues
from the anticancer drugs used today should enable cross-
resistance to be absent. The observation that some
melphalan-resistant tumour cells were not cross-resistant to
various Mannich bases of conjugated styryl ketones [10]
reinforces this conclusion.
The objective of the present investigation was to expand
series 1 in the following ways. The aryl ring in series 1 may
be designated the proximal ring. It is likely that this group
either in 1, or in the a,b-unsaturated ketones released from
the Mannich bases, interacts at a binding site by van der
Waals bonding. Such a putative area on the receptor may be
referred to as the proximal aryl binding site. It is possible that
an additional aryl binding site close to the proximal one
exists. In order to evaluate this hypothesis, the decision was
made to attach a second aryl moiety (subsequently referred to
as the distal aryl ring) via a spacer group to the proximal aryl
+
F
CCH3
O
R
XH
(i)
R
X
CCH3
O
(iv)
(iii)
(ii)
HO
O
R
C CH2CH2X HCl
F
O
R
O
C
O
C CH2CH2N(C2H5)2 HCl
O
R
X
O
H
N
HCl
C2H5
3a-g
4a-d
6a-e
a : R = H
a : X = N(CH3)2
c : X =N
d : X =N
a : R = H, X = O e : R = CH3, X = O
b : R = F
c : R = Cl
d : R = Br
b : R = F, X = O
f : R = H, X = S
b : X =
O
N
c : R = Cl, X = O
g : R = F, X = S
d : R = Br, X = O
e : R = CH3
Fig. 2. The synthetic routes used in preparing the compounds in series 3, 4 and 6. The reactants (i)–(iv) were as follows: (i) K2CO3; (ii) HCHO/HN(C2H5)2HCl;
(iii) HCHO/various amine hydrochlorides (4a–4c), 4-methylenemorpholinium chloride (4d) and (iv) HCHO/C2H5NH2HCl.

S.C. Vashishtha et al. / European Journal of Medicinal Chemistry 39 (2004) 27–35
29
rived from 3b and 4a–4d, then the rates of release of this
enone may govern the potencies of these compounds. The
pKa values of diethylamine, dimethylamine, pyrrolidine, pip-
eridine and morpholine are 10.84, 10.73, 11.31, 11.12 and
8.50 [11], respectively. Since the rates of deamination are
inversely proportional to the pKa figures of the basic group,
the possibility of a correlation between the potencies of these
compounds and the rates of release of an identical enone was
planned to be investigated.
Several years ago, some 3-aroyl-4-aryl-1-ethyl-4-
piperidinol hydrochlorides 5 were prepared as rigid ana-
logues of 1 (R3 = C2H5) [4]. These substituted piperidines 5
may be considered as possessing two portions of the mol-
ecules designated as (A) and (B) both of which incorporate
most of the structural features of the acyclic analogues. In
general, when the same substituents were present in the aryl
ring, the compounds in both series exhibited similar IC50
values when evaluated towards murine L1210 cells and a
panel of human tumour cell lines [4]. Thus in the present
investigation, the preparation of 6a–6e was proposed with a
view to comparing the potencies of these compounds with
3a–3e, respectively. Furthermore the structures of the pip-
eridines 5 and 6 do not permit deamination to occur and
hence any bioactivity observed is likely due to the Mannich
bases per se.
ous interatomic distances as well as a bond angle and a
torsion angle were obtained. These data are presented in
Table 3.
3. Bioevaluations
All of the compounds in series 3, 4 and 6 were evaluated
against murine L1210 leukaemic cells as well as human Molt
4/C8 and CEM T-lymphocytes. These data along with the
figures for 7 and 8 (vide infra) and a reference compound
melphalan are presented in Table 1. The majority of the
compounds, as well as 7 and 8 and the reference drugs
5-fluorouracil and melphalan, were screened towards seven
human colon cancer cell lines. The results obtained are sum-
marised in Table 2. Evaluation of 3a–3f, 4a–4d and 6a–6e
for NT and lethality was undertaken. These compounds were
also examined for anticonvulsant activity in the maximal
electroshock (MES) and subcutaneous pentylenetetrazole
(scPTZ) screens.
4. Results and discussion
The initial evaluation of the Mannich bases 3, 4 and 6
utilised murine L1210 cells, which have been employed
extensively in evaluating candidate antineoplastic agents.
The choice to use Molt 4/C8 and CEM T-lymphocytes was
made in order to investigate whether the compounds dis-
played cytotoxic properties towards two human cell lines.
The data are presented in Table 1.
Previous studies with certain acyclic Mannich bases re-
vealed side effects of neurotoxicity (NT) and lethality
[12,13]. In addition, anticonvulsant properties were demon-
strated implying that penetration of the CNS occurred. Thus,
an evaluation of the in vivo tolerability of the compounds
proposed in this study was planned.
In summary, the principal objectives of the current inves-
tigation were to explore the potential of the series of com-
pounds 3, 4 and 6 as candidate cytotoxic agents and also to
determine the extent of murine toxicity displayed by these
compounds. The data generated may provide guidelines as to
which clusters of compounds should be developed further as
candidate anticancer agents.
The first question addressed was whether the attachment
of aryloxy or arylthio groups to the aryl ring of 1
(R1 = R2 = H, R3 = C2H5, X = Cl referred to hereafter as 7) led
to an increase in cytotoxicity. The results in Table 1 reveal
that 7 was more potent than 3a–3g (L1210 test), 3a–3e, 3g
(Molt 4/C8 screen) and 3g (CEM assay) while in the remain-
ing cases, equal potencies were observed. Thus in the case of
these three cell lines, the attachment of aryloxy or arylthio
groups to 7 led to reduction in potencies in two-thirds of the
comparisons made.
2. Chemistry
The IC50 values of 3a, 3b and 3f, 3g were examined with
a view to ascertaining whether an oxygen or sulphur spacer
atom between the proximal and distal aryl rings in series 3
was preferred. In four of the six comparisons, the analogue
containing the oxygen spacer had greater cytotoxicity and in
one case, equal potencies were observed suggesting that an
oxygen atom between the aryl rings was preferred. This
result may have been due to the variation in the relative
positions of the two aryl rings, which were dependent on the
nature of the spacer atom, since molecular modelling re-
vealed that the aryl ring–oxygen (sulphur)–aryl ring bond
angles in diphenylether and diphenylthioether were 48.2°
and –55.9°, respectively.
The aryloxyaryl or arylthioaryl methyl ketones required in
the preparation of 3b–3g, 4a–4d and 6b–6e were synthesised
from the appropriate phenol or thiophenol and
4-fluoroacetophenone. Various aryloxyaryl or arylthioaryl
ketones, formaldehyde and different secondary amine hydro-
chlorides underwent the Mannich reaction leading to the
formation of 3a–3g and 4a–4c. The synthesis of 4d was
achieved from the reaction between 1-(4-fluorophenylo-
xyphenyl)-1-ethanone and 4-methylenemorpholinium chlo-
ride. Condensation of ethylamine hydrochloride with excess
of both the appropriate aryloxyaryl ketone and formaldehyde
led to the cyclized products 6a–6e whose structures were
1
confirmed by high resolution H NMR spectroscopy. Mo-
lecular modelling was undertaken with 3a and 6a, and vari-
A number of statistical evaluations were undertaken by
constructing linear and semilogarithmic plots between vari-

30
S.C. Vashishtha et al. / European Journal of Medicinal Chemistry 39 (2004) 27–35
Table 1
Evaluation of the compounds in series 3, 4, 6–8 against murine L1210 cells and human Molt 4/C8 and CEM T-lymphocytes
Compound
IC50 (µM) a
L1210
Molt 4/C8
43.8 2.8
43.6 0.66
39.0 11.7
104 32
CEM
3a
3b
3c
3d
3e
3f
3g
50.5 4.4
46.0 2.1
51.5 2.3
165 108
46.9 3.8
38.2 2.5
212 37
43.3 3.0
37.4 2.2
43.5 0.8
38.7 1.7
40.8 1.3
72.8 52.6
129 35
43.8 1.5
133 121
182
9
4a
4b
4c
4d
6a
6b
6c
6d
88.2 81.1
52.7 12.5
41.3 1.5
48.0 3.9
15.1 1.8
8.69 1.2
9.11 0.61
8.15 0.27
8.75 0.18
20.3 3.1
6.94 3.08
2.13 0.03
97.5 33.1
36.0 9.3
43.8 4.7
62.3 19.9
16.3 2.7
13.2 2.7
8.58 0.47
8.94 1.18
22.3 16.4
17.4 4.1
12.5 6.6
3.24 0.79
40.2 2.5
108 50
38.2 4.2
37.1 1.1
20.7 1.3
12.3 6.6
7.33 0.35
7.66 0.59
22.0 7.8
38.5 34.3
15.8 1.8
2.47 0.79
6e
7
8
Melphalan
a The IC50 figures indicate the concentrations required to inhibit cell proliferation by 50%.
ous physicochemical parameters and cytotoxic potencies.
Any relationships observed were classified as highly signifi-
cant (P < 0.01 and P < 0.001), significant (P < 0.05) or trends
towards significance (P < 0.1 and P < 0.15). Details, includ-
ing Pearson’s correlation coefficients (i.e. “r” values), are
given in Section 6.
The potencies of 3a–3e are likely influenced by the mag-
nitude of the physicochemical properties of the aryl substitu-
ents. Plots were made between the IC50 values in each of the
three screens and the r, p and molar refractivity (MR)
constants of the aryl substituents, which reflect the elec-
tronic, hydrophobic and steric properties, respectively, of the
aryl substituents. However, no correlations were noted
(P > 0.15).
The volumes and SASA figures of the protonated form of
the basic groups of 3b, 4a–4d were negatively correlated
with the IC50 figures in the L1210 screen (P < 0.05). In
addition, the volumes (P < 0.05) and SASA values (P < 0.1)
of the protonated amines found in 3b, 4a–4d were negatively
correlated with the IC50 values in the L1210 assay. No corre-
lations were noted when the two T-lymphocytes were evalu-
ated. Thus, future molecular modifications should consider
increasing the size of the basic group, such as using hexam-
ethyleneimine in the Mannich reaction.
Table 2
The cytotoxicity of representative compounds in series 3, 4, 6–8 towards various human colon cancer cell lines
Compound
IC50 (µM) a
COLO 205
HCC-2998
HCT-116
HCT-15
HT29
KM12
SW-620
Average
potency
17.2
3a
3b
3c
3e
3g
4b
4d
6a
6b
6c
6e
7 b
17.8
19.1
24.5
23.4
20.9
22.9
22.9
13.5
16.6
13.2
12.0
–
17.8
20.0
19.1
19.1
32.4
44.7
37.2
6.61
11.5
12.6
7.94
56.2
>100
52.5
0.91
17.8
16.6
17.8
17.4
18.2
60.3
22.4
9.33
13.5
3.80
6.46
17.4
17.0
39.8
5.37
17.4
18.2
18.2
18.6
17.4
41.7
19.1
12.3
11.5
7.41
3.98
20.0
19.1
36.3
12.0
19.1
20.0
21.4
19.1
21.9
32.4
18.6
6.92
10.7
5.13
3.31
28.2
20.4
70.8
7.94
13.2
12.3
14.5
11.0
15.8
30.2
17.4
4.90
4.37
2.57
6.31
43.7
41.7
57.5
12.0
17.4
18.2
21.9
18.6
20.0
79.4
30.2
12.0
15.5
12.9
12.9
20.0
15.8
26.9
25.7
17.8
19.6
18.2
20.9
44.5
24.0
9.34
12.0
8.23
7.56
30.9
8
21.4
32.4
7.41
>33.6
45.2
Melphalan
5-Fluorouracil
10.2
a The IC50 figures indicate the concentrations required to inhibit cell proliferation by 50%.
b Evaluated as the hydrobromide salt.

S.C. Vashishtha et al. / European Journal of Medicinal Chemistry 39 (2004) 27–35
31
In order to determine whether the relative rates of deami-
nation of 3b, 4a–4d correlated with cytotoxic potencies,
plots were constructed between the pKa values of the liber-
ated amines and the IC50 figures in the L1210, Molt 4/C8 and
CEM assays. No correlations were observed (P > 0.15).
The data for 6a–6e in Table 1 revealed that these com-
pounds were, in general, the most potent Mannich bases
towards L1210, Molt 4/C8 and CEM cells. The average IC50
values for 6a–6e in these three screens were 9.96, 13.9 and
14.0 µM, respectively. These figures reflect potencies which
are approximately one-fifth that of melphalan in these three
assays. Comparisons were made between the IC50 values of
each of the compounds 6a–6e with the analogues in series 3
which had the same aryl substituent, e.g. the figures for 6a
were compared with the data for 3a, 6b with 3b and so forth.
In each case, greater potencies were displayed by the com-
pounds in series 6. Furthermore, the average IC50 values of
3a–3e in each of the L1210, Molt 4/C8 and CEM assays were
72.0, 54.8 and 40.7 µM, respectively. Hence the potency of
6a–6e were seven, four and three times that of 3a–3e in each
of these three cytotoxicity screens. Thus from these initial
data 6a–6e, which contain some of the structural features of
series 3a–3e, represent an important structural modification
of series 3. In addition, plots were made between the IC50
values of 6a–6e in each of these three screens and the r, p
and MR values of the aryl substituents. The results indicated
that the r constants were negatively correlated with the IC50
values in the Molt 4/C8 (P < 0.001) and CEM (P < 0.05)
tests, while no other correlations (P > 0.15) were noted.
Thus, the placement of strongly electron-withdrawing sub-
stituents into the distal aryl rings, such as the
4-trifluoromethyl and 4-nitro groups, which possess rp val-
ues of 0.54 and 0.78, respectively [14], should be considered
when amplifying the project.
pound towards the remaining six neoplasms was 30.9 µM.
The average IC50 value of 3a–3c, 3e, 3g toward the same cell
lines was 18.4 µM. This observation revealed that the attach-
ments of aryloxy or arylthio groups to the aromatic ring of 7
led to compounds possessing nearly double the potency of 7
towards colon cancer cell lines. The 4-fluorophenoxy Man-
nich base 3b possessed marginally greater potencies than the
corresponding sulphur analogue 3g, which is in tandem with
the results obtained in the L1210, Molt 4/C8 and CEM
screens. The relative potencies of the acyclic Mannich bases
containing diethylamino (3b), pyrrolidino (4b) and mor-
pholino (4d) groups were 3b >4d >4b. Thus the two most
potent compounds, i.e. 3b and 4d, had larger volumes and
solvent accessible surface areas as well as lower pKa values
than the pyrrolidino analogue 4b. Plots were constructed
between the IC50 values of 3a–3c, 3e towards each of the
colon cell lines as well as the average potencies and the r, p
and MR values of the aryl substituents. Positive correlations
were noted between the IC50 figures generated towards
COLO 205 cell lines and the p (P < 0.01) and MR (P < 0.05)
values. In addition, positive trends towards significance were
observed between the r constants and the IC50 figures gener-
ated towards HT29 (P < 0.1) and KM12 (P < 0.1) cells as
well as between the p values and SW20 cells (P < 0.15) and
average potencies (P < 0.1). Thus, future modifications of
series 3 should consider the insertion of small, hydrophilic
groups into the aryl ring with a view to increasing cytotoxic
potencies.
The average IC50 values of 3a–3c, 3e and 6a–6c, 6e were
18.2 and 9.28 µM, respectively, indicating that the pip-
eridines 6 possessed double the potencies of the acyclic
derivatives 3. In addition, the average potency of 6a–6c, 6e
was more than three times the figure of 8, revealing that the
insertion of aryloxyaryl groups into the phenyl rings of 8 led
to substantial increases in cytotoxic potencies. A negative
correlation was noted between the r values of the aryl sub-
stituents of 6a–6c, 6e and the IC50 values towards KM12
cells when linear (P < 0.01) and semilogarithmic (P < 0.05)
plots were made. Negative trends towards significance were
observed with the p and MR values and HCT-116 cells
(P < 0.15) and between the MR figures and HCT-15 cell lines
(P < 0.10). Thus, development of series 6 should include the
use of strongly electron-withdrawing aryl substituents,
which are preferably hydrophobic and large in size.
An examination was made of the effect on potencies by
adding aryloxy substituents onto the phenyl groups of 5,
R1 = R2 = H (referred to hereafter as 8), which led to series 6.
The average IC50 values of 6a–6e in the L1210, Molt 4/C8
and CEM screens were 9.96, 13.9 and 14.0 µM, respectively,
which were virtually identical to the figures obtained for 8.
Thus, in the case of series 6, the attachment of aryloxy groups
to the proximal rings of 8 had negligible effects on potencies
when these three cell lines were considered.
One of the interests of this laboratory is the discovery of
compounds possessing antineoplastic activity towards colon
cancers [3,4]. In the present investigation, over two-thirds of
the compounds described in this report were evaluated
against seven human colon cancer cell lines. These data are
presented in Table 2. Comparisons were made between the
IC50 values of different groups of compounds in order to
glean some understanding of the structural features govern-
ing potencies. The results obtained using the colon cancer
cells revealed some similarities and also various differences
from the data summarised in Table 1.
The data in Table 2 revealed that the potencies of the
compounds in series 3, 4 and 6 were greater than that of the
established anticancer alkylating agent melphalan. The
maximum activity was displayed by 6a–6c, 6e which, on
average, were nearly five times more potent than melphalan.
Furthermore, comparisons were made with 5-fluorouracil
which finds extensive use in treating colon cancers [15]. Each
of the piperidines 6a, 6c, 6e had lower average IC50 figures
than 5-fluorouracil indicating that development of these
compounds with specificity for colonic neoplasms is clearly
warranted.
No datum was available for the evaluation of 7 towards
COLO 205 cells, while the average IC50 value of this com-
An investigation was undertaken with a view to determin-
ing the reasons for the greater potencies of the piperidines 6

32
S.C. Vashishtha et al. / European Journal of Medicinal Chemistry 39 (2004) 27–35
than of the acyclic analogues 3. The approach taken was to
compare the shapes of two representative compounds 3a and
6a. The assumptions were made that the observed bioactivi-
ties of the compounds in series 3 and 6 were due to interac-
tions at the same binding site and also that the relative
locations of two aryl rings and the nitrogen atom principally
accounted for the disparity in potencies between the two
series of compounds. In other words, interactions could oc-
cur at three areas of a putative binding site designated B1, B2
and B3 as indicated in Fig. 3(a).
The compounds in series 3 and 6 likely exist as a mixture
of the free bases and protonated species. The Mannich bases
3a and 6a were modelled with quaternary nitrogen atoms
although conceivably interaction at the binding site could be
achieved by the free bases. In the case of 6a, both aryloxyaryl
groups may be orientated in such a way as to interact at the
three areas of the proposed binding site; these arrangements
are illustrated in Fig. 3(b,c) and are referred to as 6a-A and
6a-B, respectively.
The possibility exists that 6a in one or both of its orienta-
tions has a better fit than 3a at the different areas on the
binding site. This hypothesis implies that certain crucial
interatomic distances differed between 3a and 6a. In order to
evaluate this possibility, carbon atoms were placed in the
centres of the aryl rings A and B (3a, 6a-A, 6a-B) and C and
D (6a-A, 6a-B) and designated CA, CB, CC and CD, respec-
tively. In addition, axis 1 was constructed by extending the
plane of the carbon atoms marked with asterisks in ring A
(3a, 6a-A) and ring C (6a-B) as indicated in Fig. 3(b,c). The
following distances were measured, namely d1 [CA–CB (3a,
6a-A) or CC–CD (6a-B)], d2 [CA–N (3a, 6a-A) or CC–N
(6a-B), d3 [CB–N (3a, 6a-A) or CD–N (6a-B)] and d4 [the
perpendicular distances from N to axis 1 (3a, 6a-A, 6a-B)].
Furthermore, as indicated in Fig. 3(c), the bond angle w1,
comprising the two aryl rings separated by an oxygen atom,
as well as the torsion angle h1, created by the proximal rings
and the adjacent oxygen atom, were measured. The results
are presented in Table 3.
The interatomic distances d1 were virtually identical in 3a,
6a-A and 6a-B. The increased flexibility of 3a probably
accounts for its d2 distance being larger than is found in either
orientations of 6a. The d3 distances of 3a and 6a-B are very
similar, while the figure for 6a-A is 6% larger than the span
for 3a. The major difference in the alignment of 3a and 6a at
a binding site is the distance of the nitrogen atom from axis 1.
Thus the d4 figures for the orientations 6a-A and 6a-B are
20% and 12%, respectively, greater than the case of 3a and
may contribute significantly to the differences in potencies
between the compounds in series 3 and 6. The molecular
modelling studies suggest that use of ketones containing one
or more methylene groups between the carbonyl function and
the proximal aryl ring in the Mannich reaction will lead to
even greater d4 distances. Such compounds may have in-
creased potencies towards a variety of tumour cell lines.
The bond angles w1 were virtually identical. However, the
data in Table 3 indicate that the torsion angles h1 for 6a were,
on average, 20% greater than the corresponding measure-
ment for 3a. Thus in the future, the placement of substituents
Interactions of rings A or C
B1
Interactions of rings B or D
B2
Interaction with the nitrogen atom
.
.
.
.
. .
.
B3
(a)
C
O
C
O
A
O
D
θ1
D
*
*
AXIS 1
*
AXIS 1
ψ
1
H
O
B
d1
d2
d4
O
B
A
=
C
OH
C
O
d3
(+)
N
O
H
H
(+)
N
H
C2H5
C2H5
H
(b)
(c)
Fig. 3. (a) The putative areas of the binding site of 6a. (b) The orientation of 6a is so that rings A and B and the nitrogen atom align at B1, B2 and B3. (c) The
orientation of 6a is so that rings C and D align at B1, B2 and B3.
Table 3
Some interatomic distances d1–d4 (Å), bond angles w1 (°) and torsion angles h1 (°) of 3a and 6a determined by molecular modelling
Structure
3a
6a-A
6a-B
d1
d2
d3
d4
w1
h1
4.81
4.80
4.78
10.31
9.89
10.03
5.87
6.20
5.75
5.61
6.72
6.26
117.2
117.1
116.3
44.7
–49.5
57.4

S.C. Vashishtha et al. / European Journal of Medicinal Chemistry 39 (2004) 27–35
33
of varying sizes in the ortho position of the aryl ring attached
to the oxygen atom would be predicted to increase the h1
figures. It is conceivable that such molecular modifications
of 6a would lead to compounds displaying greater potencies
than 6a; in addition, correlations between the magnitude of
the h1 values and potencies may emerge.
Some of the conclusions that may be drawn from the
toxicity and anticonvulsant screens were as follows. First, the
data afford noteworthy evidence that mice tolerated single
doses up to and including 300 mg/kg of the compounds of
series 6, in marked contrast to members of series 3 and 4
which demonstrated NT and lethality. Second, anticonvul-
sant activity in the MES screen of the latter compounds was
observed, and bearing in mind the oral activity of 4c, some
future development may be worthwhile with 3 and 4 as
candidate anticonvulsants.
A final portion of this study involved the evaluation of
whether NT and/or lethal properties were present in the
compounds of series 3, 4 and 6 in order to guide future
aspects of the project. In addition, their inclusion in the MES
and scPTZ screens was planned since, if protection was
observed, an indication of the ability of the compounds to
penetrate the CNS would be demonstrated. Should this be the
case, the potential of the compounds to treat brain tumours in
vivo exists. On the other hand, CNS penetration may lead to
unwanted toxicity. Accordingly, doses of 30, 100 and
300 mg/kg of 3a–3f, 4a–4d and 6a–6e were administered
intraperitoneally to mice and the animals were observed at
0.5 and 4 h. The doses indicated in the following discussion
reflect the quantity of compound required to elicit anticon-
vulsant activity in 50% or more of the animals.
After 0.5 h, NT was detected for 3e and 4c (30 mg/kg) and
for the remaining compounds of series 3 and 4 when a dose of
100 mg/kg was used. After 0.5 h, deaths in some or all of the
animals were caused by all of the compounds in series 3 and
4 at 100 mg/kg (3a, 3b, 4b, 4c) or 300 mg/kg (3c–3f, 4a, 4d).
Neither NT nor mortalities were observed when doses up to
and including 300 mg/kg of 6a–6e were administered to
mice. Anticonvulsant activity in the MES screen was ob-
served by various compounds. After 0.5 h, protection was
afforded by 3a–3f and 4b, 4c (30 mg/kg) and 4a, 4d
(100 mg/kg). When the mice were examined 4 h after admin-
istration of the compounds, 4b, 4c (30 mg/kg), 3b–3d, 4d
(100 mg/kg) and 6a–6c, 6e (300 mg/kg) demonstrated anti-
convulsant activity. None of the compounds afforded protec-
tion in the scPTZ screen.
5. Conclusions
This study reports the synthesis and some bioevaluations
of three series of novel compounds 3, 4 and 6. Cytotoxic
evaluation using murine L1210 lymphocytic leukaemia cells,
human Molt 4/C8 and CEM T-lymphocytes and human colon
cancer cell lines revealed that all three series of compounds
possessed cytotoxic properties. However, the piperidines 6
were more potent than 3 and 4 and are clearly useful lead
molecules. This conclusion was enhanced further by the
excellent tolerance of the compounds in series 6 when ad-
ministered to mice. Thus, while 3 and 4 displayed moderate
cytotoxic properties and possess interesting anticonvulsant
activity, it is series 6, which should be considered as a
template for future development of candidate antineoplastic
agents. Correlations were established between some of the
tumour cell lines and one or more physicochemical constants
of the aryl ring as well as the volumes and SASA figures of
the basic groups. These observations, along with the data
provided by molecular modelling, provided guidelines for
future molecular modifications of these novel compounds.
6. Experimental protocols
Three representative compounds, namely 3f, 4b and 4c,
which gave complete protection in the MES screen at the
minimum dose of 30 mg/kg 0.5 h after administration, were
examined using doses of 3 and 10 mg/kg with a view to
determine whether bioactivity would be demonstrated at
lower doses. After 0.5 h, 3f displayed neither anticonvulsant
activity nor toxicity while 4b and 4c afforded protection in
the MES screen unaccompanied by murine toxicity when
10 mg/kg of each compound was administered. A quantita-
tive evaluation of 4c revealed that the ED50 and TD50 figures
(95% confidence intervals in parentheses) were 7.37 (6.64–
8.13) and 26.5 (25.2–28.1) mg/kg, respectively, indicating a
protection index (TD50/ED50) of 3.60. The ED50 and TD50
values of a reference drug phenytoin were 6.32 and
41.2 mg/kg, respectively [16], which generated a protection
index of 6.52 or approximately twice that of 4c. Compounds
4b, 4c and 6e were administered orally to rats using doses of
12.5 mg/kg (4b) and 30 mg/kg (4c, 6e). The animals were
observed after 0.25, 0.5, 1, 2 and 4 h in the MES screen and
for NT. Protection was displayed by 4c (0.25, 1 and 2 h) and
6e (4 h). No toxicity was noted by visual observation.
6.1. Chemistry
Melting points (m.p.) are uncorrected. Temperatures were
recorded in Celsius degrees. Elemental analyses (carbon,
hydrogen and nitrogen) were undertaken by Mr. K. Thoms,
Department of Chemistry, University of Saskatchewan, on
3a–3g, 4a–4d and 6a–6e and were within 0.4% of the calcu-
lated values. Compounds 3a, 3b and 6b were obtained as the
monohydrates and 3e as the hemihydrate. The yields were
based on the lowest molar fractions of the reactants. 1-(4-
Phenyloxyphenyl)-1-ethanone was obtained from the Ald-
rich Chemical Company, Milwaukee, USA. TLC using silica
gel plastic-backed sheets with a fluorescent indicator re-
1
vealed that the compounds were homogenous. H NMR
spectra were determined using Varian T-60 and Bruker AM
300 FT instruments. Spectra obtained on some of the inter-
mediates and on all of the compounds submitted for bio-
evaluation confirmed the identity of the products. The pro-
tons of the proximal and distal aryl rings are designated as

34
S.C. Vashishtha et al. / European Journal of Medicinal Chemistry 39 (2004) 27–35
unprimed and primed, respectively. The recrystallisation sol-
vents were designated by the letters A–I and were as follows,
namely A, acetone–ethanol; B, diethyl ether–ethanol; C,
n-hexane–acetone; D, acetonitrile; E, diethyl ether–metha-
nol; F, methanol; G, ethanol; H, acetone–methanol and I,
diethyl ether–acetone.
ortho H), 6.8–7.2 (m, 2H, meta H and 4′H ortho and meta H),
2.7–3.9 (m, 8H, COCH2CH2; 2-CH2, 6-CH2 of piperidine
ring), 1.5–2.5 (br m, 4H, 3-CH2, 5-CH2 of piperidine ring),
1.2–1.4 (m, 2H, 4-CH2 of piperidine ring).
6.1.2. Synthesis of series 6
A mixture of the appropriate 1-(4-aryloxyphenyl)-1-
ethanone (0.04 mol), paraformaldehyde (0.08 mol), ethy-
lamine hydrochloride (0.01 mol), hydrochloric acid (37%
w/v, 0.04 ml) and isopropanol (100 ml) was heated under
reflux for varying periods of time (vide infra). Evaporation in
vacuo gave a residue which was triturated with dry diethyl
ether and upon recrystallisation led to the following com-
pounds (time of heating under reflux, recrystallisation sol-
vent, m.p. and % yield in parentheses), namely 6a (48 h, F,
180–182 °C, 38), 6b (48 h, G, 172–174 °C, 25), 6c (72 h, H,
179–181 °C, 18), 6d (72 h, F, 191–193 °C, 22) and 6e (29 h,
I, 175–177 °C, 20). The 1H NMR spectrum (300 MHz) of a
representative compound 6a is given below. In this case, the
hydrogen atoms of the aroyl proximal and distal aryl rings
are primed and unprimed, respectively, while the proximal
and distal aryl hydrogen atoms of the group at position 4 of
the piperidine ring are double and triple primed, respectively.
The subscripts a and e refer to hydrogen atoms in the axial
and equatorial conformations, respectively. d (CDCl3): 8.02–
8.10 (d, 2H, ortho H), 7.30–7.45 (m, 2H, meta H; 2′H, meta
H), 7.15–7.30 (m, 3′′′H, ortho and para H), 6.95–7.05 (m,
3′H, ortho and para H), 6.75–6.85 (d, 2″H, ortho H), 6.90–
6.95 (m, 2′″H, meta H; 2″H, meta H), 5.55 (dd, 1H, C3Ha, J
3a/2a = 11.8 Hz, J 3a/2a = 3.7 Hz), 5.10 (d, 1H, OH,
J = 2.54 Hz), 3.40–3.48 (m, 2H, C2He, C6He), 3.25–3.35 (m,
2H, C2Ha, C6Ha), 3.00–3.20 (m, 2H, N CH2CH3), 2.74–2.90
(m, 1H, C5Ha), 1.92 (br d, 1H, C5He, J = 14.7 Hz), 1.50 (t,
3H, NCH2CH3, J = 7.3 Hz).
6.1.1. Synthesis of series 3 and 4
The 1-(4-aryloxyaryl)-1-ethanones and 1-(4-arylthioa-
ryl)-1-ethanones required for the syntheses of 3b–3g, 4a–4d
and 6b–6e were prepared as follows. A mixture of the appro-
priate phenol or thiophenol (0.11 mol), 4-fluoroacetophe-
none (0.10 mmol), anhydrous potassium carbonate
(0.15 mol) and dimethylacetamide (75 ml) was heated under
reflux for approximately 12 h at 160 °C under nitrogen. On
cooling, water (100 ml) was added and the mixture was
extracted with chloroform (2 × 100 ml) and the combined
organic extracts were washed with aqueous sodium hydrox-
ide solution (4% w/v) and water. The solution was dried
using anhydrous magnesium sulphate and evaporation of the
solvent led to the isolation of an oil which was used without
1
further purification. The H NMR spectrum (60 MHz) of a
representative intermediate, namely 1-(4-fluorophenyloxy-
phenyl)-1-ethanone was as follows: d (CDCl3) 7.8–8.0 (m,
2H, ortho H), 6.8–7.2 (m, 2H, meta H and 4′H, ortho and
meta H), 2.5 (s, 3H, CH3).
The preparation of a representative Mannich base 3a was
as follows. A mixture of 1-(4-phenyloxyphenyl)-1-ethanone
(0.015 mol), paraformaldehyde (0.035 mol), diethylamine
hydrochloride (0.01 mol), hydrochloric acid (36.5% w/v,
0.04 ml) and acetonitrile (50 ml) was heated under reflux for
24 h. Removal of the solvent gave an oil which was triturated
several times with diethyl ether. The residue obtained was
recrystallised from acetone to give 3a, m.p. 98–99 °C in 75%
yield. A similar route was utilised to give the following
compounds (molar ratio of ketone, paraformaldehyde and
amine, time of heating under reflux, recrystallisation solvent,
m.p. and % yield in parentheses), namely 3b (0.03, 0.02,
0.01, 46 h,A, 131–133 °C, 64), 3c (0.015, 0.03, 0.01, 14 h, B,
125–127 °C, 37), 3d (0.015, 0.03, 0.01, 14 h, B, 126–128 °C,
50), 3e (0.03, 0.03, 0.01, 24 h, C, 134–136 °C, 78), 3f (0.03,
0.05, 0.01, 48 h, C, 94–96 °C (Ref. [17] m.p. 90–92 °C), 72),
3g (0.03, 0.05, 0.01, 48 h, C, 136–138 °C, 76), 4a (0.02, 0.02,
0.01, 24 h, D, 162–164 °C, 34), 4b (0.025, 0.025, 0.01, 24 h,
E, 165–167 °C, 34) and 4c (0.01, 0.02, 0.01, 6 h, D, 174–
175 °C, 49).
6.1.3. Syntheses of 7 and 8
The synthesis of 7 and the corresponding hydrobromide
salt as well as 8 has been described previously [4].
6.1.4. Statistical analyses
The SASA values and volumes of the diethylamino, dim-
ethylamino, pyrrolidino, piperidino and morpholino groups
(along with the related protonated species) were obtained
from MacroModel 4.5 [19]. The r, p and MR values were
taken from Ref. [14]. Linear and semilogarithmic plots were
made using a commercial software package [20].
The preparation of 4d was undertaken as follows. A mix-
ture of 4-methylenemorpholinium chloride (0.01 mol),
which was prepared by a literature procedure [18], 1-(4-
flurorophenyloxyphenyl)-1-ethanone (0.01 mol) and aceto-
nitrile (~50 ml) was heated under reflux for 2 h. On cooling,
the precipitate was collected, washed with cold acetonitrile
and dry diethyl ether. Recrystallisation from acetone gave
4d, m.p. 184–185 °C in 70% yield.
For each group of compounds where correlations and
trends towards significance were noted, the physicochemical
constant, assay, linear (l) or semilogarithmic (sl) plots, Pear-
son’s correlation coefficient and P-value are indicated as
follows: 3b, 4a–4d: volume of basic group (protonated),
L1210, l, –0.935, 0.020; volume of basic group (protonated),
L1210, sl, –0.925, 0.024; volume of basic group (free base),
L1210, l, –0.931, 0.022; volume of basic group (free base),
L1210, sl, –0.921, 0.026; SASA (protonated), L1210, l,
–0.910, 0.032; SASA (protonated), L1210, sl, –0.900, 0.037;
1
The H NMR spectrum (60 MHz) of a representative
compound 4c was as follows: d (CDCl3): 7.9–8.1 (m, 2H,

S.C. Vashishtha et al. / European Journal of Medicinal Chemistry 39 (2004) 27–35
35
SASA (free base), L1210, l, –0.875, 0.052; SASA (free
base), L1210, sl, –0.867, 0.057; 6a–6e: r, Molt 4/C8, l,
–0.995, 0.0001; r, Molt 4/C8, sl, –0.996, 0.001; r, CEM, l,
–0.942, 0.017; r, CEM, sl, –0.946, 0.015; 3a–3c, 3e: r,
HT29, l, 0.905, 0.095; r, HT29, sl, 0.906, 0.094; r, KM12, l,
0.921, 0.079; r, KM12, sl, 0.920, 0.080; p, COLO 205, l,
0.999, 0.001; p, COLO 205, sl, 0.998, 0.002; p, SW-620, sl,
0.842, 0.146; p, average potency of human colon cancer cell
lines (AP), l, 0.910, 0.090; p, AP, sl, 0.915, 0.085; MR,
COLO 205, l, 0.980, 0.020; MR, COLO 205, sl, 0.977, 0.023;
6a–6c, 6e: r, KM12, l, –0.997, 0.003; r, KM12, sl, –0.977,
0.023; p, HCT-116, sl, –0.871, 0.129; MR, HCT-116, l,
–0.893, 0.107; MR, HCT-116, sl, –0.898, 0.102; MR, HCT-
15, l, –0.905, 0.095.
tutes of Health Research to J.R. Dimmock. The National
Cancer Institute (USA) kindly provided the data presented in
Table 2. The excellent technical assistance of Mrs. Lisette
van Berckelaer is gratefully recognised (L1210, Molt 4/C8
and CEM assays).
References
[1] J.R. Dimmock, P. Kumar, Curr. Med. Chem. 4 (1997) 1–22.
[2] J.R. Dimmock, K.M. Advikolanu, H.E. Scott, M.J. Duffy, R.S. Reid,
J.W. Quail, Z. Jia, R.A. Hickie, T.M. Allen, J.M. Rutledge, M.L. Tem-
pest, A.B. Oreski, J. Pharm. Sci. 81 (1992) 1147–1152.
[3] S.C. Vashishtha, T.M. Allen, S. Halleran, J. Szydlowski, C.L. Santos,
E. De Clercq, J. Balzarini, J.R. Dimmock, Pharmazie 56 (2001)
390–393.
[4] J.R. Dimmock, S.C. Vashishtha, J.W. Quail, U. Pugazhenthi,
Z. Zimpel, A.M. Sudom, T.M. Allen, G.Y. Kao, J. Balzarini, E. De
Clercq, J. Med. Chem. 41 (1998) 4012–4020.
6.1.5. Molecular modelling
Models of compounds were built using MacroModel 8.0
[21] followed by a Monte Carlo search for the lowest energy
conformations using an Amber force field of 1000 initial
conformations. Measurements were made on the lowest en-
ergy conformations.
[5] J.R. Dimmock, K. Shyam, N.W. Hamon, B.M. Logan, S.K. Raghavan,
D.J. Harwood, P.J. Smith, J. Pharm. Sci. 72 (1983) 887–894.
[6] F.A. Carey, R.J. Sundberg, Advanced Organic Chemistry, Part B,
Plenum Press, New York, 1977, pp. 44 et 46.
[7] J. March, Advanced Organic Chemistry, third ed, Wiley Eastern Lim-
ited, New Delhi, 1985, p. 298.
[8] B. Mutus, J.D. Wagner, C.J. Talpas, J.R. Dimmock, O.A. Phillips,
R.S. Reid, Anal. Biochem. 177 (1989) 237–243.
6.2. Bioevaluations
[9] A. Baluja, A.M. Municio, S. Vega, Chem. Ind. (1964) 2053–2054.
[10] J.R. Dimmock, P. Kumar, J.W. Quail, U. Pugazhenthi, J. Yang,
M. Chen, R.S. Reid, T.M. Allen, G.Y. Kao, S.P.C. Cole, G. Batist,
J. Balzarini, E. De Clercq, Eur. J. Med. Chem 30 (1995) 209–217.
[11] A. Albert, E.P. Serjeant, The Determination of Ionization Constants,
third ed, Chapman and Hall, London, 1984, pp. 151–152.
[12] J.R. Dimmock, S.S. Jonnalagadda, O.A. Phillips, E. Erciyas,
K. Shyam, H.A. Semple, J. Pharm. Sci. 81 (1992) 436–440.
[13] J.R. Dimmock, S.A. Patil, K. Shyam, Pharmazie 46 (1991) 538–539.
[14] C. Hansch,A.J. Leo, Substituent Constants for CorrelationAnalysis in
Chemistry and Biology, Wiley, New York, 1979, p. 49.
[15] W.A. Remers, 10th ed, in: J.N. Delgado, W.A. Remers (Eds.), Wilson
and Gisvold’s Textbook of Organic and Medicinal Chemistry,
Lippincott-Raven, Philadelphia, 1998, p. 359.
[16] R.N. Puthucode, U. Pugazhenthi, J.W. Quail, J.P. Stables, J.R. Dim-
mock, Eur. J. Med. Chem. 33 (1998) 595–607.
6.2.1. Cytotoxicity determinations
Literature methodologies were followed when evaluating
various compounds against L1210, Molt 4/C8 and CEM cells
[22] as well as human colon cancer cell lines [23].
6.2.2. Evaluation of various compounds for toxicity
and anticonvulsant properties
The screening of 3a–3f, 4a–4d and 6a–6e for murine
toxicity and anticonvulsant properties was undertaken by a
literature procedure [24]. Initially, doses of 30, 100 and
300 mg/kg of each compound were injected into mice using
the intraperitoneal route. The animals were observed after
0.5 and 4 h. NT was determined by the rotorod procedure
[25] and visual observations. Anticonvulsant properties were
assessed using the MES and scPTZ screens.
[17] N. Wassermann, M. Calarasanu, R. Waitman, Rev. Chim. (Bucherest)
11 (1960) 349–350; Chem. Abstr. 57 (1962) 14991b.
[18] H. Bohme, D. Hartke, Chem. Ber. 93 (1960) 1305–1309.
[19] MacroModel, Version 4.5, Department of Chemistry, Columbia Uni-
versity, New York, August 1994.
[20] Statistical Package for Social Sciences, SPSS for Windows, Standard
Version, Release 10.0.5, SPSS Inc., Chicago, IL, USA, 1999.
[21] MacroModel, Version 8.0, Department of Chemistry, Columbia Uni-
versity, New York, April 2002.
[22] J. Balzarini, E. De Clercq, M.P. Mertes, D. Shugar, P.F. Torrence,
Biochem. Pharmacol. 31 (1982) 3673–3682.
[23] M.R. Boyd, K.D. Paull, Drug Dev. Res. 34 (1995) 91–109.
[24] J.P. Stables, H.J. Kupferberg, in: G. Vanzini, P. Tanganelli, M. Avoli
(Eds.), Molecular and Cellular Targets for Antiepileptic Drugs, John
Libbey and Company Ltd., London, 1997.
[25] M.S. Dunham, T.A. Miya, J. Am. Pharm. Assoc. Sci. Ed. 46 (1957)
208–209.
Acknowledgements
The award of a Graduate Scholarship to S.C. Vashishtha
from the University of Saskatchewan is gratefully acknowl-
edged. Funding for this project was provided by the Bio-
medical Research Programme of the European Community
and the Belgian Fonds voor Geneeskundig Wetenschappelijk
Onderzeok to J. Balzarini and E. De Clercq and by the
Medical Research Council of Canada and the Canadian Insti-