Design, synthesis, and development of novel caprolactam anticonvulsants

Article abstract of DOI:10.1016/S0968-0896(03)00400-0

Epilepsy afflicts 1-2% of the world's population and often goes untreated; nearly 70% of those with a form of epilepsy fail to receive proper treatment. Therefore, there is great demand for the design of novel, effective anticonvulsants to combat epilepsy in its numerous forms. Previously, α-hydroxy-α-phenylcaprolactam was found to have rather potent antiepileptic activity [anti-maximal electroshock (MES) ED₅₀=63 mg/ kg and anti-subcutaneous Metrazol (scMet) ED₅₀=74 mg/kg] when administered intraperitoneally in mice. We focused our attention on the development of this compound through traditional medicinal chemistry techniques - including the Topliss approach, isosteric replacement, methylene insertion, and rigid analogue approach - in the hopes of determining the effect of caprolactam α-substitution and other structural modifications on anticonvulsant activity. A number of the desired targets were successfully synthesized and submitted to the Anticonvulsant Screening Program of the National Institute of Neurological Disorders and Stroke (NINDS). Phase I results were quite promising for at least three of the compounds: α-ethynyl-α-hydroxycaprolactam (10), α-benzyl-α-hydroxycaprolactam (11), and α-hydroxy-α-(phenylethynyl)caprolactam (13). Phase II results for 11 strongly suggested it as a new structural class for further development, as it exhibited an anti-MES T.I. in excess of 4.0. Further, the potent activity of 13 in all models also pointed to the substituted alkynylcaprolactams as a new anticonvulsant structural class.

Full text of DOI:10.1016/S0968-0896(03)00400-0

Bioorganic & Medicinal Chemistry 11 (2003) 4133–4141
Design, Synthesis, and Development of Novel Caprolactam
Anticonvulsants
Jonathan B. Grimm,a James P. Stablesb and Milton L. Browna,*
aDepartment of Chemistry, University of Virginia, McCormick Road, PO Box 400319, Charlottesville, VA 22904, USA
bAnticonvulsant Drug Development Program, National Institute of Neurological Disorders and Stroke, 6001 Rockville, MD 20892, USA
Received 12 March 2003;accepted 11 April 2003
Abstract—Epilepsy afflicts 1–2% of the world’s population and often goes untreated;nearly 70% of those with a form of epilepsy
fail to receive proper treatment. Therefore, there is great demand for the design of novel, effective anticonvulsants to combat epi-
lepsy in its numerous forms. Previously, a-hydroxy-a-phenylcaprolactam was found to have rather potent antiepileptic activity
[anti-maximal electroshock (MES) ED50=63 mg/kg and anti-subcutaneous Metrazol (scMet) ED50=74 mg/kg] when administered
intraperitoneally in mice. We focused our attention on the development of this compound through traditional medicinal chemistry
techniques—including the Topliss approach, isosteric replacement, methylene insertion, and rigid analogue approach—in the hopes
of determining the effect of caprolactam a-substitution and other structural modifications on anticonvulsant activity. A number of
the desired targets were successfully synthesized and submitted to the Anticonvulsant Screening Program of the National Institute
of Neurological Disorders and Stroke (NINDS). Phase I results were quite promising for at least three of the compounds:
a-ethynyl-a-hydroxycaprolactam (10), a-benzyl-a-hydroxycaprolactam (11), and a-hydroxy-a-(phenylethynyl)caprolactam (13).
Phase II results for 11 strongly suggested it as a new structural class for further development, as it exhibited an anti-MES T.I.
in excess of 4.0. Further, the potent activity of 13 in all models also pointed to the substituted alkynylcaprolactams as a new
anticonvulsant structural class.
# 2003 Elsevier Ltd. All rights reserved.
Introduction
Class I anticonvulsants bind to the neuronal voltage-
dependent sodium channel (NVSC), a heterotrimeric
transmembrane protein complex.3 The NVSC from a
typical mammalian brain consists of one a subunit and
two b subunits and is instrumental in the mediation of
action potentials in neurons and other electrically exci-
table cells. It is chiefly responsible for the change in ion
permeability of cell membranes and allows for the vol-
tage-stimulated depolarization of cells through the rapid
influx of sodium ions.3 Anticonvulsants belonging to
class II act on g-aminobutyric acid (GABA) receptors.
GABA itself is believed to be a central inhibitory neu-
rohormonal modulator,1 with the receptor being
responsible for local changes in ion conductance at
chemical synapses.3 The class III anticonvulsants func-
Of the countless neurological diseases and disorders that
have been identified, few are as pervasive and common
as epilepsy. Afflicting 1–2% of the world’s population, it
often goes untreated. The drugs that do exist to combat
epilepsy are far from perfect, as modern clinical anti-
convulsants are effective in reducing the severity and
number of seizures in less than 80% of treated epilep-
tics.1 The side effects that accompany the use of these
drugs are also quite significant in some cases. Therefore,
there is a great demand for the design of new, novel,
effective anticonvulsants that extend therapy to the mil-
lions suffering from epilepsy.
The anticonvulsants that serve to control the seizures
associated with epilepsy have previously been categor-
ized into a number of distinct classes on the basis of the
molecular mechanisms through which they function.2
tion through undetermined mechanisms. A more
recently disclosed class includes those anticonvulsants
that bind to excitatory amino acid receptors, such as
those for N-methyl-d-aspartate (NMDA).
Anti-Bredt bicyclic imides like 1 remain a class of com-
pounds that possess significant anticonvulsant activity.
*Corresponding author. Tel.: +1-434-982-3091;fax: +1-434-924-
0798;e-mail: mlb2v@virginia.edu
0968-0896/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0968-0896(03)00400-0

4134
J. B. Grimm et al. / Bioorg. Med. Chem. 11 (2003) 4133–4141
the binding of [3H]-batrachotoxinin A 20-a-benzoate
([3H]BTX-B) to the sodium channel.2 However, ip
administration of 4 in mice resulted in potent anti-
convulsant protection in both anticonvulsant models
tested (scMet and MES induced convulsions), exhibiting
an anti-MES ED50=63 mg/kg and an anti-scMet
ED50=74 mg/kg. These results encouraged a further
investigation6 into the possible anticonvulsant activity
of analogues of the lactam. After testing analogues with
ortho, meta, and para a-phenyl substitutions, one com-
pound was found to be particularly active: a-hydroxy-a-
4-chlorophenylcaprolactam (5). This lactam exhibited
an anti-MES effective dose of 21.8 mg/kg. Amazingly,
the toxic dose of this compound exceeded 250 mg/kg.
This corresponded to a therapeutic index (the ratio of
toxic dose to effective dose) greater than 11.5, implicat-
ing the lactam as a potential clinical agent.
Figure 1. Smissmanones and related structures.
These compounds were originally proposed by Edward
E. Smissman4—one of the fathers of modern medicinal
chemistry—and are referred to as ‘smissmanones.’ After
these unique compounds were first synthesized in the
late 1980s, they were tested for anticonvulsant activity
and found to be effective sodium channel binders (i.e.,
class I anticonvulsants).2 Interestingly enough, the
structurally related trimethadione (3,3,5-trimethyl-2,4-
oxazolidinedione, 2) has no effect on the NVSC and is
considered a class III anticonvulsant, whereas pheny-
toin (diphenylhydantoin, 3) is a class I anticonvulsant
with rather strong binding to the neuronal voltage-
dependent sodium channel (Fig. 1).5
With this in mind, we proposed the synthesis of com-
pounds 7–14 (see Schemes 1–3) to further explore the
effect of various substitutions (on the phenyl ring and
otherwise) on the anti-epileptic effects of 4. A number of
these targets were selected through the use of traditional
medicinal chemistry optimization techniques (Fig. 2).
The Topliss series (begun with 4, the unsubstituted
compound, and 5, the 4-chloro species) was completed
with the synthesis of 7 (3,4-dichloro) and 8 (4-methyl).
The 4-methoxy derivative, the final component of the
Topliss series, was synthesized previously.6 Target 11
was the product of a methylene insertion, whereas 9 was
the rigid analogue (di-ortho substitution). Isosteric
replacement of the phenyl ring with the character-
istically similar ethynyl group yielded 10. Initial anti-
convulsant results concerning the activities of 10 and 11
(see Results and Discussion) prompted the inclusion of
further targets, including the acyclic amide analogue 12.
In an attempt to determine what structural features
impart NVSC binding ability to the smissmanones, a
number of partial structures were synthesized and eval-
uated for anti-epileptic activity.2 One of the com-
pounds—a-hydroxy-a-phenylcaprolactam
particularly interesting. The lactam exhibited weak
(4)—was
NVSC binding with an IC50 >800 mM;IC is the con-
centration of anticonvulsant needed to inhibit 50% of
50
Scheme 1. Synthesis of a-substituted caprolactams.
Scheme 2. Synthetic route to acyclic amide analogue of 11.

J. B. Grimm et al. / Bioorg. Med. Chem. 11 (2003) 4133–4141
4135
Scheme 3. Sonogashira couplings involving 10.
Of particular interest was the alkyne moiety of 10 and
the potential for coupling any number of different
structures at this position. Compounds 13 and 14 were
therefore targeted via traditional Sonogashira cross-
coupling reactions.
reflex of the animal when placed on a rotating rod.
Other toxicities are also reported as observed. Further
evaluation of active compounds in Phase II provides
quantification of activity as an ED50 (the effective dose
needed to protect 50% of the mice or rats) in the MES
or scMet model and a TD50 (the dose required to reach
toxicity in 50% of the animals) in the Rotorod toxicity
model. The results of these evaluations revealed new,
active structural classes for the design of further targets.
The discovery of a number of active leads may also
help, in the future, to elucidate the mechanism of action
of this broad class of anticonvulsants.
Biology
The successfully synthesized compounds were screened
for anticonvulsant activity through the Anticonvulsant
Screening Program (ASP) of the National Institute of
Neurological Disorders and Stroke (NINDS). The ASP
consists of seven phases that thoroughly quantify the
anticonvulsant activity of the compound in question.
Phase I activity for each compound is reported as the
number of protected animals per group at concentration
mg/kg of drug measured over the duration of the antic-
onvulsant activity in mice (ip administration only) and
in rats (oral administration only). The maximal electro-
shock (MES) model involves the induction of seizures
through the application of an alternating current to
corneal electrodes in the eyes of a rodent, whereas the
subcutaneous pentylenetetrazol (PTZ, also known as
Metrazol) seizures are generated chemically by the
injection of PTZ. Neurotoxicity (at administered dose)
is reported in the Rotorod model by loss of the righting
Chemistry
Of the eight target compounds, seven are produced
through various reactions involving a-oxocaprolactam
(18). The synthesis of this compound has been well-
documented7 and is shown in Scheme 1. The first step
involves a bromination of the readily available e-capro-
lactam (15) to yield the dibrominated lactam 16.
Refluxing this compound in piperidine affords the
enamine, 17, that is then cleaved when flash chromato-
graphed on silica gel to yield a-oxocaprolactam (18).
This novel step allows for the simultaneous formation
and purification of the desired lactam.
Figure 2. Optimization techniques.

4136
J. B. Grimm et al. / Bioorg. Med. Chem. 11 (2003) 4133–4141
Scheme 1 also details the synthesis of the targets derived
directly from a-oxocaprolactam. The appropriate
Grignard reagents (e.g., 3,4-dichlorophenyl magnesium
bromide) are used to produce, through selective alkyl-
ation of the a-keto group, the a-phenyl, a-benzyl, and
a-ethynyl derivatives (7–11).
pound was not active against MES-induced seizures in
mice and showed no toxicity at or below 300 mg/kg.
However, the acetylene analogue was active in the mice
anti-scMet model at 300 mg/kg. This suggested a-ethy-
nyl-a-hydroxycaprolactam (10) as a selective inhibitor
of petite mal seizures. The initial identification of 10 as a
promising, active compound encouraged immediate
optimization and development. Alkynes 13 and 14 were
targeted for this reason;their syntheses exploited the
synthetically rich terminal alkyne moiety and are dis-
cussed below.
The synthesis of the acyclic amide 12 (Scheme 2) is
accomplished by first reacting valeronitrile with benzyl-
magnesium bromide in a Grignard reaction. The result-
ing ketone is then converted to the cyanohydrin through
the use of trimethylsilyl cyanide and an aqueous acid
workup. Subsequent hydrolysis with HCl gas yields the
desired hydroxyamide. The substituted alkynes 13 and
14 are afforded via traditional Sonogashira couplings of
the suitable aryl iodides (iodobenzene, 1-iodonaphtha-
lene) and a-ethynyl-a-hydroxycaprolactam (10) in the
presence of an appropriate palladium source [i.e.,
Pd(PPh3)4].8 A summary of important synthetic and
analytical results are listed in Table 1.
The two active a-substituted caprolactams 10 and 11
moved on to Phase II of the Anticonvulsant Screening
Program, and those results (oral activity in rats) are lis-
ted in Table 3. Compound 11 (a-benzyl-a-hydroxy-
caprolactam) displayed significant activity, with an anti-
MES ED50=86 mg/kg and an anti-scMet ED50
<83 mg/kg. Given that the Rotorod TD50 was greater
than 330 mg/kg, this particular lactam exhibited thera-
peutic indices exceeding 3.9 (MES) and 4.0 (scMet).
These
results
showed
that
a-benzyl-a-hydro-
xycaprolactam was effective in halting seizures,
although the effectiveness of this lactam appeared to be
lower than that of 5 (anti-MES ED50=22 mg/kg, T.I.
>11.5). However, because toxicity was not tested
beyond 300 mg/kg (at which dose there was still no
toxicity), the actual T.I. values may be higher than cal-
culated.
Results and Discussion
The first group of targets, lactams 7–11, were obtained
through Grignard reactions of a-oxocaprolactam.
Unfortunately, it was quickly found that these reactions
consistently gave low yields;significant amounts of the
starting material were required to produce samples of
the targets sufficient for biological testing (250–400 mg).
The especially low yield (<20%) of the 2,6-dimethyl
derivative was attributed to the steric hindrance of the
Grignard reagent. Nonetheless, 250 mg of each com-
pounds was synthesized and submitted to NINDS to
determine their anticonvulsant activity.
More surprising were the results seen for 10, where little
separation between activity and toxicity was seen. The
alkyne exhibited an anti-MES ED50 >120 mg/kg and a
Rotorod TD50 >120 mg/kg (Table 3). This compound
was ineffective at preventing MES-induced seizures at a
non-toxic dose, further emphasizing the importance of
the aryl moiety in the prevention of those seizures
mimicked by the MES model.
Phase I results for both mice (ip administration) and
rats (oral administration) have been obtained for these
compounds and best activities are listed in Table 2
along with the log P values of the targets (as predicted
by Crippen’s fragmentation).9 We note here that only
those compounds showing promising activity in Phase I
were evaluated in rats (Phase II) and listed in Table 3.
The synthesis of the acyclic amide (12), inspired by the
initially surprising activity of 11, was also completed.
The only major pitfall in the synthesis occurred early in
the synthetic scheme. The first intermediate is benzyl
butyl ketone;the initial attempt to synthesize this com-
pound involved a Grignard reaction between phenyla-
cetonitrile and butylmagnesium bromide. This
consistently failed to provide the desired ketone;the
starting material was isolated in every instance. It was
concluded that the reaction failed because the Grignard
reagent was deprotonating phenylacetonitrile at the
benzyl position, producing a highly resonance-stabilized
species that yielded the starting material upon acidifica-
tion of the reaction mixture. The logical solution to this
problem was a synthon switch to valeronitrile and ben-
zylmagnesium bromide. This new reaction was more
successful, affording the desired ketone in a 57% yield.
The ketone was carried on without difficulty to the
amide, which was promptly submitted to NINDS for
anticonvulsant screening. Unfortunately, this particular
compound showed little activity in both the MES and
scMet models when administered to mice (Table 2);
mild activity was seen at 100 mg/kg in less than 50% of
the rats in both models. Furthermore, the compound
Little separation between activity and toxicity was seen
in the para-methyl analogue 8, where anti-MES activity
was only seen in mice at a dose approaching 300 mg/kg.
Compounds 7, 9, and 11 provided more promising
results. For all three, anti-MES activity was seen at
100 mg/kg in mice;Rotorod toxicity was not seen until
the dose reached 300 mg/kg. All three were only mildly
active against the Metrazol-induced seizures (300 mg/
kg). These three targets were also active in the anti-MES
model when administered orally to rats. Analogues 7
and 9 were active over a longer duration than 11, and
the benzyl analogue appeared to be more potent and
effective (4/4 rats protected) than the 3,4-dichloro and
2,6-dimethyl lactams. These results suggested 11 for
further investigation, prompting the inclusion of the
acyclic amide 12 in this study. The potent activity of 11
when administered orally also implicated this com-
pound as a potential clinical anticonvulsant. The Phase
I results for 10 were also quite promising. The com-

Table 1. Summary of synthetic results and product characterization
Compd
Yield (%)
25
Mp (ꢀC)
1H NMR (CDCl3, d)
13C NMR (CDCl3, d)
MS (APCI, M+1)
273.9
7
134—136
7.50–7.41 (m, 2H, Ph), 7.23–7.16 (m, 1H, Ph), 6.92 (bs, 1H, NH), 5.02 (bs, 1H, OH),
3.21–3.05 (m, 1H, CH), 2.93–2.78 (m, 1H, CH), 2.60–2.48 (m, 1H, CH), 2.05–1.87 (m, 2H, CH),
1.77–1.64 (m, 1H, CH), 1.61–1.40 (m, 2H, CH)
178.2, 140.7, 133.8, 132.9, 131.7, 131.5,
129.1, 126.3, 42.9, 36.0, 29.0, 25.5
8
50
164—166
7.30–7.17 (m, 4H, Ph), 6.99 (bs, 1H, NH), 4.83 (bs, 1H, OH), 3.12–2.99 (m, 1H, CH),
2.94–2.80 (m, 1H, CH), 2.69–2.56 (m, 1H, CH), 2.35 (s, 3H, CH3), 2.02–1.84 (m, 2H, CH),
1.71–1.37 (m, 3H, CH)
179.5, 138.4, 137.2, 130.2, 126.7, 42.9,
36.0, 29.2, 25.7, 21.6
220.0
9
17
53
53
132–134
159–161
121–122
7.10–6.94 (m, 3H, Ph), 6.62 (bs, 1H, NH), 4.94 (bs, 1H, OH), 3.08–2.77 (m, 3H, CH),
2.48 (s, 6H, CH3), 2.02–1.31 (m, 5H, CH)
182.2, 138.0, 131.4, 127.7, 42.1, 38.8,
28.8, 25.4, 23.3
216.1 (MꢁOH)
154.1
10
11
6.91 (bs, 1H, NH), 5.02 (bs, 1H, OH), 3.78–3.60 (m, 1H, CH), 3.28–3.14 (m, 1H, CH),
175.0, 81.6, 76.6, 70.5, 43.0, 37.9,
29.4, 26.1
2.71 (s, 1H, H–C¼C), 2.20–1.79 (m, 4H, CH), 1.75–1.59 (m, 1H, CH), 1.50–1.32 (m, 1H, CH)
7.33–7.18 (m, 5H, Ph), 6.63 (bs, 1H, NH), 4.16 (bs, 1H, OH), 3.54–3.40 (td, 1H, CH),
3.31–3.18 (m, 2H, CH), 2.89–2.81 (d, 1H, CH), 2.10–1.35 (m, 6H, CH)
179.3, 136.3, 130.9, 128.5, 127.2, 43.3,
42.1, 36.6, 29.6, 24.9
220.1
12
13
68
31
99–101
7.38–7.20 (m, 5H, Ph), 6.46 (bs, 2H, NH2), 5.46 (bs, 1H, OH), 3.34–3.24 (d, 1H, CHPh),
2.86–2.76 (d, 1H, CHPh), 2.10–1.86 (m, 2H, CH), 1.63–1.20 (m, 4H, CH), 0.94–0.82 (t, 3H, CH3)
7.56–7.22 (m, 5H, Ph), 6.92 (bs, 1H, NH), 5.10 (bs, 1H, OH), 3.87–3.67 (m, 1H, CH),
3.34–3.15 (m, 1H, CH), 2.29–1.67 (m, 5H, CH), 1.53–1.35 (m, 1H, CH)
177.9, 136.0, 130.9, 129.1, 127.7, 79.2,
45.6, 39.1, 26.2, 23.4, 14.5
175.6, 132.3, 129.3, 128.9, 122.6, 88.4,
86.8, 71.0, 43.2, 38.2, 29.5, 26.3
222.1
230.1
116–118
14
30
195–197
8.34–8.25 (m, Ph, 1H), 7.90–7.37 (m, 7H, Ph), 6.62 (bs, 1H, NH), 4.98 (bs, 1H, OH),
3.94–3.76 (m, 1H, CH), 3.33–3.18 (m, 1H, CH), 2.45–1.38 (m, 6H, CH)
175.7, 133.8, 133.6, 131.4, 129.8, 128.9,
127.6, 127.0, 126.4, 125.6, 120.2, 91.7,
86.6, 71.4, 43.2, 38.4, 29.6, 26.5
280.2
16
17
79
93
162–164
139–144
6.84 (bs, 1H, NH), 3.43–3.33 (q, 2H, CH2), 2.79–2.70 (m, 2H, CH2), 2.04–1.63 (m, 4H, CH2)
169.0, 70.0, 46.5, 43.1, 28.9, 28.8
270.3, 272.2, 274.2
195.4
7.20 (bs, 1H, NH), 5.07–4.98 (t, 1H, CH), 3.23–3.14 (q, 2H, CH2), 2.78–2.70 (t, 4H, CH2),
2.17–2.06 (q, 2H, CH2), 1.79–1.41 (m, 8H, CH2)
172.1, 147.9, 105.8, 50.4, 47.0, 39.7,
30.6, 26.3, 25.8, 24.8, 21.8
18
20
56
57
60–64
—
7.71 (bs, 1H, NH), 3.30–3.22 (q, 2H, CH2), 2.60–2.53 (m, 2H, CH2), 1.97–1.68 (m, 4H, CH2)
204.2, 170.3, 40.0, 27.9, 23.4
128.1
7.40–7.14 (m, 5H, Ph), 3.67 (s, 2H, CH2), 2.51–2.39 (t, 2H, CH2), 1.62–1.48 (quint, 2H, CH2),
1.36–1.21 (m, 2H, CH2), 0.96–0.83 (t, 3H, CH3)
208.8, 135.0, 129.9, 129.2, 127.4, 20.6,
42.2, 26.3, 22.7, 14.3
177.2 (CI)
22
77
—
7.32–7.18 (m, 5H, Ph), 6.38 (bs, 1H, OH), 2.94 (s, 2H, CH2), 1.69–1.14 (m, 6H, CH),
0.88–0.78 (t, 3H, CH3) (DMSO)
134.1, 131.4, 131.1, 129.4, 128.7, 128.4,
121.4, 72.7, 46.8, 40.2, 26.7, 23.1, 14.4
177.1 (MꢁCN)

4138
J. B. Grimm et al. / Bioorg. Med. Chem. 11 (2003) 4133–4141
was noticeably toxic in the Rotorod model at 100 mg/kg
in 75% (6 out of 8) of the mice receiving the compound
by ip injection. When these results were compared to
those obtained for 11, it was clear that the ring opening
decreased the efficacy of the compound and increased its
toxicity. Removing the rigidity of the caprolactam ring
was not effective in improving the anticonvulsant activ-
ity of 11;the conformationally restricted seven-mem-
bered ring may play some role in site binding. Further
study is necessary to determine precisely how it is rele-
vant. Given the poor activity, it is unlikely we will fur-
ther develop the class of compounds typified by 12.
As another set of ‘second generation’ analogues, the
substituted alkynes shown in Scheme 3 were successfully
synthesized through Sonogashira couplings of 10 and
the appropriate aryl iodides. Employing a palladium
source of Pd(PPh3)4 as the catalyst, the reactions con-
sistently afforded the desired products in modest yields
(ꢂ30%). Both 13 and 14 were submitted to NINDS for
evaluation of anticonvulsant activity, and Phase I
results for both (Tables 2 and 3) are reported. Target 13
exhibited strong anti-MES and anti-scMet activity in
mice at 100 mg/kg. Rotorod toxicity was noted at
300 mg/kg in 4/4 mice. Similarly, rats receiving 13 were
Table 2. NINDS anti-epileptic drug development program results, Phase 1: best activity in mice
(ip administration)
Compd
Log P
Phase I: best activity in mice
(ip admin)
Phase I: best activity in rats
(oral admin)
MES
(duration)
scMet
(duration)
Rotorod
(duration)
MES
(duration)
Rotorod
(duration)
4a
5a
1.25
1.81
No data
No data
100 mg/kg
(0.25 h)
2/3 mice
protected at
100 mg/kg
(0.25 h)
300 mg/kg
(0.25 h)
3/5 mice
protected at
300 mg/kg
(0.25 h)
300 mg/kg
(0.25 h)
4/4 mice toxic at
300 mg/kg
(0.25 h)
4/4 rats
protected at
30 mg/kg
(2 h)
No toxicity at
30 mg/kg
6a
1.13
2.37
1.74
2.23
ꢁ0.27
1.53
2.22
1.52
2.52
1/3 mice
No activity at
100 mg/kg
No toxicity at
100 mg/kg
No data
No data
protected at
100 mg/kg
(0.25 h)
7
2/3 mice
1/1 mice
protected at
300 mg/kg
(0.5 h)
No activity at
300 mg/kg
4/4 mice toxic at
300 mg/kg
(0.5 h)
3/4 rats at
30 mg/kg
(1 h)
No toxicity at
30 mg/kg
protected at
100 mg/kg
(0.5 h)
8
1/1 mice
1/4 mice toxic at
300 mg/kg
(0.5 h)
No data
No data
protected at
300 mg/kg
(1 h)
9
3/3 mice
No activity at
300 mg/kg
3/4 mice toxic at
300 mg/kg
(0.5 h)
2/4 rats at
30 mg/kg
(1 h)
No toxicity at
30 mg/kg
protected at
100 mg/kg
(0.5 h)
No activity at
300 mg/kg
10
11
12
13
14
4/5 mice
protected at
300 mg/kg
(0.5 h)
No toxicity at
300 mg/kg
2/4 rats at
30 mg/kg
(4 h)
No toxicity at
30 mg/kg
2/3 mice
protected at
100 mg/kg
(0.25, 1 h)
1/3 mice
protected at
100 mg/kg
(0.25 h)
1/1 mice
4/4 mice toxic at
300 mg/kg
(0.5 h)
4/4 rats at
30 mg/kg
(0.25 h)
No toxicity at
30 mg/kg
protected at
300 mg/kg
(1 h)
1/5 mice
6/8 mice toxic at
100 mg/kg
(0.5 h)
No data
No data
protected at
100 mg/kg
(0.25 h)
3/3 mice
5/5 mice
4/4 mice toxic at
300 mg/kg
(0.5 h)
4/4 rats
protected at
30 mg/kg
(1 h)
No toxicity at
30 mg/kg
protected at
100 mg/kg
(0.5 h)
No activity at
300 mg/kg
protected at
100 mg/kg
(0.25 h)
No activity at
300 mg/kg
No toxicity at
300 mg/kg
No data
No data
Table 3. NINDS anti-epileptic drug development program results, Phase II: best activity in rats (oral administration)
Compd
Log P
Phase II: Quantification in rats (oral administration)
MES ED50
scMet ED50
Rotorod TD50
MES T.I.
scMet T.I.
10
11
1.81
1.25
>120 mg/kg
85.71 mg/kg
No data
<82.5 mg/kg
>120 mg/kg
>330 mg/kg
N/A
>3.85
N/A
>4.00

J. B. Grimm et al. / Bioorg. Med. Chem. 11 (2003) 4133–4141
4139
protected against MES-induced seizures at 30 mg/kg;no
toxicity was noted. The difference in activity between 10
and 13 was postulated to result, at least partly, from the
significant change in log P from the unsubstituted
terminal alkyne (ꢁ0.27) to the phenyl-substituted
alkyne (1.52). Unfortunately, the other alkyne, 14, dis-
played no activity (nor toxicity) in any model in rats or
mice. The large naphthyl group most likely interfered
sterically in the activity of this compound relative to
that of 13.
(25%) of 7 as a white powder. C, H, N analysis: calcd
C, 52.27;H, 4.78;N, 5.11;found C, 52.79;H, 4.87;N,
5.13.
ꢀ-Hydroxy-ꢀ-p-tolylcaprolactam (8). To a dry 100-mL
two-neck round-bottom flask was added ground mag-
nesium turnings (0.97 g, 0.040 mol) and iodine (a few
crystals). After melting 4-bromotoluene over a steam
bath, a small amount (0.5 mL, 0.004 mol) was added to
the flask. The remainder (3.2 mL, 0.026 mol) was dis-
solved in dry THF (30 mL) and added over 20 min. The
solution was then allowed to cool to room temperature
before cooling it in an ice bath and slowly adding 18
(1.27 g, 0.010 mol) in dry THF (20 mL). The ice bath
was removed and the reaction was allowed to stir at
room temperature under nitrogen for 96 h. The
Grignard was quenched by the addition of 5% HCl
(60 mL) before extracting with Et2O (3 ꢃ 125 mL). The
combined ether layers were dried over sodium sulfate
and concentrated in vacuo to yield a yellow powder.
The solid was recrystallized from CHCl3/hexanes to
collect 1.08 g (50%) of off-white crystals. C, H, N ana-
lysis: calcd C, 71.21;H, 7.81;N, 6.39;found C, 71.03;
H, 7.73;N, 6.29.
Conclusion
The synthesis of a series of caprolactams derived from
a-hydroxy-a-phenylcaprolactam (4) was completed, and
the compounds were submitted to the NINDS Antic-
onvulsant Screening Program. Phase I results for a
number of the targets were promising;the product of
isosteric replacement, a-ethynyl-a-phenylcaprolactam
(10), appeared to be selective for petit mal seizures. The
benzyl derivative 11 exhibited strong activity when
administered orally in rats, and when further evaluated
in Phase II, was sufficiently active to suggest further
development of its structural class. The activity of 10
prompted the inclusion of more targets, including the
Sonogashira coupling products 13 and 14. The phenyl-
substituted alkyne 13 demonstrated extremely promis-
ing anti-epileptic activity in rats and mice in Phase I and
will also be further developed in the future.
ꢀ-Hydroxy-ꢀ-2,6-dimethylphenylcaprolactam (9). After
placing ground magnesium turnings (1.46 g, 0.060 mol)
and iodine (a few crystals) in a dry 250-mL two-neck
round-bottom flask, 2-bromo-m-xylene (1.00 mL,
0.008 mol) was added. The remainder of the 2-bromo-
m-xylene (5.00 mL, 0.037 mol) was dissolved in dry THF
(40 mL) and added over a 30-min period. The solution
was allowed to cool to room temperature. After cool-
ing the reaction further in an ice bath, 18 (1.90 g,
0.015 mol) in dry THF (40 mL) was added slowly. The
solution was then stirred under nitrogen at room tem-
perature for 72 h. The reaction was quenched with 5%
HCl (150 mL) and extracted with Et2O (3 ꢃ 200 mL).
The combined organic layers were dried over sodium
sulfate and concentrated in vacuo to yield a yellow oil.
The yellow oil was placed on the HPLC (hexane/iso-
propanol, 0–100% isopropanol, gradient time=30 -
min, l=254 nm, flow rate=3 mL/min) and eluted for
35 min. The fraction containing the compound eluting
at 10 min was concentrated in vacuo to produce a clear
oil that crystallized upon standing, yielding 0.58 g
(17%) of 9 as colorless crystals. C, H, N analysis: calcd
C, 72.07;H, 8.21;N, 6.00;found C, 71.83;H, 8.09;N,
5.92.
Experimental
Melting points were recorded on an Electrothermal
capillary melting point apparatus and are uncorrected.
1H NMR and 13C NMR spectra were recorded on a QE
300 MHz FT NMR spectrometer at ambient tempera-
ture and with an internal standard of tetramethylsilane
(TMS). IR spectra were recorded on an FT-IR Impact
400D. Mass spectra were recorded on a Finnagan LCQ
Classic (APCI+ unless otherwise noted). Elemental
analyses were performed by Atlantic Microlabs in Nor-
cross, GA, USA. See Table 1 for characterizations of
products.
ꢀ-Hydroxy-ꢀ-3,4-dichlorophenylcaprolactam (7). Ground
magnesium turnings (0.49 g, 0.020 mol) were placed in
the bottom of a dry 100-mL two-neck round-bottom
flask with a few crystals of iodine. A small amount of 4-
bromo-1,2-dichlorobenzene (0.5 mL, 0.004 mol) was
added to the flask;the remainder (1.8 mL, 0.014 mol)
was dissolved in dry THF (20 mL) and added slowly.
After the reaction cooled to room temperature, it was
cooled in an ice bath while adding 18 (0.75 g, 0.006 mol)
in dry THF (15 mL). The solution was then allowed to
stir at room temperature under nitrogen for 48 h. The
Grignard was then quenched with 5% HCl (75 mL)
before extracting with Et2O (3 ꢃ 100 mL). The ether
layers were combined, dried over sodium sulfate, and
concentrated in vacuo to yield a yellow oil. Addition of
isopropanol and hexane (1:1) precipitated a solid that
was collected by vacuum filtration, producing 0.40 g
ꢀ-Ethynyl-ꢀ-hydroxycaprolactam (10). After placing
ethynylmagnesium bromide in THF (0.5 M, 120.0 mL)
in a dry 500-mL one-neck round-bottom flask, the
solution was cooled in an ice bath before adding 18
(2.50 g, 0.020 mol) in dry THF (30.0 mL) over 30 min.
The ice bath was removed, and the solution was stirred
at room temperature under nitrogen for 72 h. The reac-
tion was quenched by the gradual addition of 5% HCl
(100 mL). After extracting with Et2O (3 ꢃ 120 mL), the
ether layer was dried over magnesium sulfate. Removal
of the solvent on the rotary evaporatory yielded a
brown solid. Flash chromatography on silica gel (5 ꢃ
20cm) with 50% ethyl acetate/hexanes afford 1.61g (53%)

4140
J. B. Grimm et al. / Bioorg. Med. Chem. 11 (2003) 4133–4141
of 10 as a white power (Rf 0.40, 2:1 ethyl acetate/hex-
anes, permanganate stain). C, H, N analysis: calcd C,
62.73;H, 7.24;N, 9.14;found C, 62.83;H, 7.17;N, 9.04.
per(I)iodide (0.15 g, 0.0008 mol). The mixture was stir-
red for 20 h at 60 ꢀC under nitrogen. After removing the
solvent under vacuum, the resulting brown semisolid
was placed on a silica gel column (5 ꢃ 30 cm) and eluted
with a steady gradient of 10% ethyl acetate/hexanes to
60% ethyl acetate/hexanes. A single component (Rf
0.55, 2:1 ethyl acetate/hexanes) was collected. Evapora-
tion of the solvent yielded a white solid that was
recrystallized from hot ethyl acetate, giving 0.43 g (30%)
of white crystals. C, H, N analysis: calcd C, 77.40;H,
6.13;N, 5.01;found C, 77.20;H, 6.12;N, 5.04.
ꢀ-Benzyl-ꢀ-hydroxycaprolactam (11). Under dry condi-
tions, benzylmagnesium chloride in THF (2.0 M,
32.0 mL) was placed in a 250-mL three-neck round-
bottom flask. After cooling the solution in an ice bath,
18 (2.00 g, 0.016 mol) in dry THF (40.0 mL) was added
slowly over 1 h. The solution was stirred at room tem-
perature under nitrogen for 72 h before quenching the
Grignard with 5% HCl (75 mL). The mixture was
extracted with Et2O (3 ꢃ 150 mL);the combined ether
layers were dried over magnesium sulfate and con-
centrated in vacuo to produce a yellow oil that crystal-
lized on standing. The solid was recrystallized from hot
EtOAc to yield 1.87 g (53%) of a white powder. C, H, N
analysis: calcd C, 71.21;H, 7.81;N, 6.39;found C,
71.07;H, 7.79;N, 6.33.
3,3-Dibromo-azepan-2-one (16). To a stirring solution of
chloroform (400 mL) in a 1000-mL three-neck round-
bottom flask was added e-caprolactam (15, 15.00 g,
0.133 mol). The solution was cooled to 0–5 ꢀC through
the use of an ice bath and phosphorus pentachloride
(55.20 g, 0.265 mol) was added over a 30-min period.
Zinc iodide (1.53 g, 0.0048 mol) was then added. The
solution was removed from the ice bath and allowed to
reach room temperature as bromine (42.35 g, 0.265 mol)
was slowly added over the course of 30 min. After the
mixture was stirred at room temperature for 6 h, the
resulting orange solution was combined with ice water
(300 mL). The chloroform layer was washed with water
(1 ꢃ 200 mL) and 0.50 M sodium bisulfite (3 ꢃ 200 mL),
dried over sodium sulfate, and concentrated in vacuo to
yield a yellow solid. The solid was suspended in water,
filtered, and washed with water and Et2O to give 28.23 g
(79%) of a fluffy, white solid.
2-Benzyl-2-hydroxyhexanoic acid amide (12). The cya-
nohydrin 22 (1.00 g, 0.005 mol) was placed in a 25-mL
one-neck round-bottom flask. Concentrated HCl (37%,
1.00 mL) was added, and the reaction was stirred for 1 h
at 0 ꢀC while bubbling HCl gas through the solution.
The HCl gas line was removed, and the solution was
allowed to sit at room temperature overnight. The
resulting semisolid was diluted with water (50 mL) and
extracted with ethyl acetate (3 ꢃ 50 mL). After drying the
combined organics over magnesium sulfate, the solvent
was removed to yield a yellow oil. Flash chromatography
on silica gel (3 ꢃ 25 cm) with 2:1 hexanes/ethyl acetate
yielded a yellow oil (Rf 0.15, 2:1 hexanes/ethyl acetate)
that crystallized under vacuum to afford 0.75 g (68%) of
an off-white powder. C, H, N analysis: calcd C, 70.56;
H, 8.65;N, 6.33;found C, 70.26;H, 8.72;N, 6.23.
3-Piperidin-1-yl-1,5,6,7-tetrahydro-azepin-2-one (17). A
solution of 16 (20.00 g, 0.074 mol) in piperidine
(175.0 mL) was stirred and heated at reflux for 4.5 h in a
500-mL two-neck round-bottom flask. After removing
the heat, the solution was allowed to reach room tem-
perature before being combined with 0.50 M sodium
bisulfite (150.0 mL). The mixture was extracted with
chloroform (3 ꢃ 75 mL). The combined organic layer
was dried over sodium sulfate and concentrated in
vacuo to yield a brown, oily solid that crystallized upon
standing. The resulting solid was suspended in water,
filtered, and washed with water and Et2O to provide
13.31 g (93%) of 17 as a yellow-white, powdery solid.
ꢀ-Hydroxy-ꢀ-(phenylethynyl)caprolactam (13). Iodo-
benzene (1.45 g, 0.007 mol) and 10 (0.80 g, 0.005 mol)
were added to a solution of DMF (0.50 mL) and
CH3CN (1.00 mL) in a 25-mL one-neck round-bottom
flask. To the stirring solution were added triethylamine
(8.88 g, 0.088 mol), PPh3 (0.21 g, 0.0008 mol), Pd(PPh3)4
(0.92 g, 0.0008 mol), and copper(I)iodide (0.15 g,
0.0008 mol). The mixture was stirred under nitrogen at
60 ꢀC for 20 h. The solvent was removed under vacuum
to yield a viscous brown oil that was purified by flash
chromatography on silica gel (5 ꢃ 30 cm) with a gra-
dient of 10% ethyl acetate/hexanes to 50% ethyl ace-
tate/hexanes. Fractions with a component of Rf 0.50
(2:1 ethyl acetate/hexanes) were combined and evapo-
rated to afford a yellow solid. The solid, slightly impure
ꢀ-Oxocaprolactam (18). The enamine 17 (12.00 g,
0.006 mol) was dissolved in chloroform (40.0 mL) and
placed on an 8 ꢃ 24 cm silica gel column. The sample
was eluted with ethyl acetate (4 L) and 30-mL fractions
were collected. The fractions were monitored with TLC
(40% Et2O/chloroform);those fractions containing a
material with Rf 0.12 were combined and concentrated
in vacuo to provide a clear oil that crystallized on
standing. The solid was suspended in 1:1 Et2O/hexanes,
filtered, and washed with hexanes to yield the product
as a white solid (4.37 g, 56%).
1
by H NMR, was recrystallized from hot ethyl acetate
to give the product as 0.36 g (31%) of off-white crystals.
C, H, N analysis: calcd C, 73.34;H, 6.59;N, 6.11;found
C, 73.01;H, 6.51;N, 6.05.
ꢀ-Hydroxy-ꢀ-(naphthalen-1-ylethynyl)caprolactam (14).
To a mixture of 10 (0.80 g, 0.005 mol), 1-iodonaphtha-
lene (1.80 g, 0.007 mol), DMF (0.50 mL), and CH3CN
(1.00 mL) in a 25-mL one-neck round-bottom flask were
added triethylamine (8.88 g, 0.088 mol), PPh3 (0.21 g,
0.0008 mol), Pd(PPh3)4 (0.92 g, 0.0008 mol), and cop-
Benzyl butyl ketone (20). Crushed magnesium turnings
(1.94 g, 0.025 mol) and a few crystals of iodine were
placed in a dry 250-mL two-neck round-bottom flask.
Benzyl bromide (1.0 mL, 0.008 mol) was added. Addi-
tional benzyl bromide (7.9 mL, 0.066 mol) was dissolved
in dry ether (40.0 mL) and added dropwise. Valeroni-

J. B. Grimm et al. / Bioorg. Med. Chem. 11 (2003) 4133–4141
4141
trile (19, 2.6 mL, 0.025 mol) was similarly dissolved in
Acknowledgements
dry ether (15.0 mL) and slowly added over 15 min. The
reaction was stirred at reflux under nitrogen for 8 h.
After removing the oil bath, the mixture was stirred
overnight at room temperature. The reaction was quen-
ched by the slow addition of 5% H2SO4 (75.0 mL) and
stirred at reflux for 90 min. After extracting with ether
(3 ꢃ 100 mL) and drying the combined organic layers
over magnesium sulfate, the solvent was removed to
afford a dark yellow oil. Flash chromatography on silica
gel (8 cm ꢃ 25 cm) with 5% ethyl acetate/hexanes yiel-
ded 2.52 g (57%) of the desired product as a yellow oil
(Rf 0.38, 10% ethyl acetate/hexanes).
J.B.G. would like to thank David A. Harrison III, the
Harrison family, the Faculty Senate of the University of
Virginia, Pfizer Global Research and Development, the
University of Virginia Department of Chemistry, and
the University of Virginia Research Experience for
Undergraduates Program for their financial support of
this work.
References and Notes
1. Hadjipavlou-Litina, D. Med. Res. Rev. 1998, 18, 91.
2. Brouillette, W. J.;Brown, G. B.;DeLorey, T. M.;Shirali,
S. S.;Grunewald, G. L. J. Med. Chem. 1988, 31, 2218.
3. Catterall, W. A. Science 1988, 242, 50.
4. Smissman, E. E.;Matuszack, A. J. B.;Corder, C. N. J.
Pharm. Sci. 1964, 54, 1541.
5. Willow, M.;Catterall, W. A. Mol. Pharmacol. 1982, 22,
627.
6. Brown, M. L.;Liang, G.;Brouillette, W. J. Personal
communication.
7. Brouillette, W. J.;Einspahr, H. M. J. Org. Chem. 1984, 49,
5113.
8. Rieger, J.;Brown, M. L.;Sullivan, G. W.;Linden, J.;
Macdonald, T. L. J. Med. Chem. 2001, 44, 531.
9. Ghose, A. K.;Crippen, G. M. J. Chem. Inf. Comput. Sci.
1987, 27, 21.
2-Benzyl-2-hydroxyhexanenitrile (22). In a dry 10-mL
one-neck round-bottom flask were combined 20 (1.23 g,
0.007 mol), trimethylsilyl cyanide (1.87 mL, 0.014 mol),
and a few crystals of zinc iodide. The reaction was stir-
red under nitrogen at room temperature for 24 h;the
reaction was monitored by IR for the disappearance of
the C¼O stretch. The TMS ether (21) was not isolated
but was instead converted directly to the cyanohydrin
by combining the solution with 15% aq HCl (15.0 mL)
and ether (15.0 mL). After stirring vigorously for 4 h at
room temperature, the mixture was extracted with ether
(3 ꢃ 50 mL). Evaporation of the solvent yielded 1.10 g
(77%) of 22 as a red oil that was carried on without
further purification.