Identification and structure-activity studies of novel ultrashort-acting benzodiazepine receptor agonists

Article abstract of DOI:10.1016/S0960-894X(02)00512-7

The synthesis and evaluation of novel ultrashort-acting benzodiazepine (USA BZD) agonists are described. A BZD scaffold was modified by incorporation of amino acids and derivatives. The propionate side chain of glutamic acid tethers an enzymatically labile functionality where the metabolite carboxylic acid displays markedly reduced BZD receptor affinity. The USA BZDs were characterized by full agonism profiles.

Full text of DOI:10.1016/S0960-894X(02)00512-7

Bioorganic & Medicinal Chemistry Letters 12 (2002) 3215–3218
Identification and Structure–Activity Studies of Novel Ultrashort-
Acting Benzodiazepine Receptor Agonists
Jeffrey A. Stafford,y Gregory J. Pacofsky,{ Richard F. Cox, Jill R. Cowan,
George F. Dorsey, Jr., Stephen S. Gonzales, David K. Jung, George W. Koszalka,
Maggie S. McIntyre, Jeffrey H. Tidwell, Robert P. Wiard and Paul L. Feldman*
GlaxoSmithKline, Five Moore Drive, Research Triangle Park, NC 27709, USA
Received 4 February 2002; accepted 7 June 2002
Abstract—The synthesis and evaluation of novel ultrashort-acting benzodiazepine (USA BZD) agonists is described. A BZD scaf-
fold was modified by incorporation of amino acids and derivatives. The propionate side chain of glutamic acid tethers an enzy-
matically labile functionality where the metabolite carboxylic acid displays markedly reduced BZD receptor affinity. The USA
BZDs were characterized by full agonism profiles.
#2000 Elsevier Science Ltd. All rights reserved.
Benzodiazepines (BZDs) have been widely used to pro-
vide anxiolysis and sedation in numerous clinical set-
tings since their discovery over four decades ago.1 BZDs
are known to exert their broad pharmacological effects
through specific binding to a subunit on the g-amino-
butyric acid (GABAA) receptor, the native target for the
primary inhibitory neurotransmitter GABA, that is
referred to as the BZD receptor. Despite their utility,
oversedation and prolonged recovery periods often
confound the therapeutic effects of currently marketed
drugs; this latter property can be particularly proble-
matic with agents that are administered by continuous
intravenous (iv) infusion in intensive care units.2 These
observations together with growing trends towards
ambulatory care both in the United States and in Eur-
ope have demanded shorter-acting sedating agents that
provide for rapid onset, deep sedation, and full, rapid
emergence from the effects of anaesthesia. Thus, our goal
became the discovery of an iv-deliverable, ultrashort-
acting BZD (USA BZD) that retains the broad
pharmacology of a BZD full agonist, such as midazolam
(1)3 or diazepam (2) (Fig. 1).4 Such an agent would be
expected to provide useful alternatives in both con-
temporary anaesthesiology and ambulatory care.5
The strategy for providing a short-acting disposition to
a BZD emanated from our development of the ultra-
short-acting opioid analgesic, remifentanil.6 The phar-
macologically active form of remifentanil is a carboxylic
ester which, by virtue of its lability to nonspecific
plasma and tissue esterases, undergoes rapid metabo-
lism to a pharmacologically null carboxylic acid.7
Translating this model to BZD agonism required: (a)
incorporation of a carboxylic ester into the BZD scaf-
fold without compromising receptor binding affinity, (b)
demonstration that the corresponding carboxylic acid
displays significantly reduced binding affinity for the
BZD receptor and, (c) retention of a full agonism pro-
file. Such pharmacology together with an organ-inde-
pendent elimination mechanism is expected to provide
more predictable and reproducible pharmacodynamic
and pharmacokinetic profiles than BZDs in current
clinical practice.
The synthesis of benzodiazepinone 11a shown in
Scheme 1 exemplifies preparation of the compounds
described in this paper. Condensation between the glu-
tamate-derived acid chloride 38 and an appropriate
2-aminobenzophenone (4) yielded anilide 5 in excellent
yield without detectable racemization.9 Removal of the
*Corresponding author. Fax: +1-919-483-3762; e-mail: plf4602@gsk.
com
yPresent address: Syrrx, Inc., 10410 Science Center Drive, San Diego,
CA 92121, USA.
{Present address: Chrysalon Molecular Research, 11 Avenue Albert
Einstein, 69100 Villeurbanne, France.
0960-894X/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(02)00512-7

3216
J. A. Stafford et al. / Bioorg. Med. Chem. Lett. 12 (2002) 3215–3218
assay.14 Attachment of the ester functionality at C3
provided entities with potency/selectivity and the
opportunity to map SAR as a function of changes in the
tether (Table 1). Based on tether length alone, potent
compounds were identified from each group of ana-
logues, and heteroatom-containing linkers delivered
compounds equipotent to carbon analogues. Interest-
ingly, the configuration at the C3-stereogenic center
proved important regardless of the tether length; nat-
ural l-amino acids conferred greater potency than their
unnatural d-counterparts, and this trend continued with
heteroatom linkers as well. In the end, the potency of
the glutamate-based analogues (11a, 11d, 13, and 15)
versus the single atom-spaced versions (8b–d), the cost-
liness of the aminoadipate analogue (16), and the rela-
tively cumbersome synthesis of the amine-linked
analogues (18b and 19b) afforded advantages to this
glutamate-based series.
Figure 1. Structures of midazolam (1) and diazepam (2).
The strategy called for esters that bind with high affinity
to the BZD receptor whereas their corresponding acids
would display reduced affinity. Thus, subsequent com-
parisons of ester-acid pairs were performed and are
represented by Ki (acid)/Ki (ester) ratio (Table 1). A
ratio of ꢁ100-fold in relative activities was targeted.
This filter confirmed the utility of the glutamate-based
analogues and prompted an investigation of selectivity
as a function of the identity of the ester. Because of the
lipophilic character of the BZD receptor,15 aliphatic and
simple aralkyl esters were evaluated. The methyl ester
11a emerged as a particularly interesting compound
owing to its excellent receptor binding affinity and the
remarkable separation in binding affinity relative to its
corresponding acid 11f, which binds the human BZD
receptor with a Ki value of 6.2 mM.
Scheme 1. Reagents and conditions: (a) CHCl3, reflux; (b) Et3N,
CH2Cl2, reflux; (c) 5% HOAc in (CH2Cl)2, 60 ꢂC; (d) [(n-Bu2Sn
(Cl)]2O, ROH, PhCH3, reflux; (e) LiOH, THF, MeOH, H2O.
FMOC protecting group was achieved by treatment
with triethylamine in dichloromethane. The solvent was
removed and replaced with 5% acetic acid in dichloro-
ethane; the resulting solution was warmed to effect
cyclodehydration and provide 11a in 55% yield.10
Transesterifications were executed using the protocol
reported by Otera11 to deliver 11b–e. Saponification
provided acid 11f.
Neither high affinity binding of a ligand to the benzo-
diazepine receptor nor selectivity within an ester–acid
pair characterizes the intrinsic efficacy (full agonist,
inverse agonist, antagonist) of a BZD receptor ligand.
Therefore, determination of intrinsic efficacy was asses-
sed by the ability of a compound to cause loss of the
righting reflex (LRR) in rats, an effect associated with
benzodiazepine full agonism.16 Compounds producing
LRR were evaluated by three measures: (1) time to
onset of LRR, (2) initial recovery from LRR (animal
rights itself three consecutive times after losing its
righting reflex), and (3) total recovery time (the animal
walks without ataxia and pulls itself up three con-
secutive times when suspended from a horizontal wire).
Several compounds were evaluated in the LRR assay
and the recovery behaviors were compared to the clin-
ical standards 1 and 2. Results are tabulated in Table 1.
Compound 11a presented the desired pharmacody-
namic profile offering rapid, uneventful sedation and
significantly shorter recovery times (initial=9 min,
total=24 min) relative to 1 and 2. The finding that the
acid 11f is the only identifiable metabolite from with-
drawn plasma samples bolsters the hypothesis that the
observed pharmacodynamic profile is attributed to the
plasma and/or tissue degradation of the BZD ester. In
addition, in vitro incubation of 11a in rat plasma gives
rise to 11f with a half-life <1 min.
Analogues 18a–c and 19a–c were prepared according to
Scheme 2. In the event, 3-amino-BZD (17)12 was reacted
with methyl bromoacetate or methyl acrylate after
which chiral HPLC was utilized to supply the adducts in
enantiomerically pure form. Subsequently (3S)-amino-
BZD13 was used to synthesize analogues of interest. As
before, saponification of the esters afforded the corre-
sponding acids.
Our project required esters with levels of potency com-
parable to clinically relevant BZDs, targeted as Ki ꢀ 50
nM as measured in our BZD radioligand binding
Scheme 2. Reagents and conditions: (a) methyl bromoacetate, Et3N,
THF; (b) methyl acrylate, EtOH; (c) LiOH, THF, MeOH, H2O.

J. A. Stafford et al. / Bioorg. Med. Chem. Lett. 12 (2002) 3215–3218
Table 1. Comparative data for USA BZDs
3217
Compd
X
Y
Z
n
R
C3
Kia
Ratiob
Dosec
LRR recoveryd
Initial
Total
1
2
6
7
8a
8b
8c
8d
9a
9b
10
11a
11b
11c
11d
11e
12
2
30
15
25
23
5
4
121
82
14
14
1447
100,000
289
81
45
176
92
163
200
7
39
242
29
34
741
3
118
124
7
3
53
23
480
17
3
189
Cl
Cl
Cl
Cl
Cl
Cl
H
H
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Br
O2N
O2N
O2N
Cl
Cl
Cl
Cl
Cl
H
H
F
F
F
F
F
F
H
F
F
F
F
F
F
F
H
H
F
0
0
0
0
0
0
1
1
1
1
1
1
1
1
1
1
1
1
1
2
1
1
1
1
1
1
CH2Ph
CH2Ph
CH2(4-pyridyl)
CH3
S
R
S
S
S
S
S
S
S
S
S
S
S
S
R
S
S
S
165
297
76
CH2CH3
CH2Ph
CH3
CH2Ph
CH2Ph
CH3
25
13
17
CH2
CH2
CH2
CH2
CH2
CH2
CH2
CH2
CH2
CH2
CH2
CH2
CH2
CH2
NH
70
25
15
9
22
24
984
108
17
143
124
CH(CH3)2
(CH2)3CH3
CH2Ph
CH2(4-pyridyl)
CH3
25
25
15
8e
25
19
25
25
25
25
2
22
2
4
30
14
3
13
CH3
CH3
CH2Ph
CH3
CH3
CH3
CH3
CH3
CH3
431
95
91
14a
14b
15
2
S
S
16
F
186
35
172
9
17
130
1
18a
18b
18c
19a
19b
19c
Cl
Cl
Cl
Cl
Cl
Cl
RS
S
R
RS
S
NH
NH
NHCH2
NHCH2
NHCH2
Cl
Cl
CH3
CH3
R
aAffinity for rat BZD receptor, nM.
bKi (acid)/Ki (ester).
cBolus injection, mg/kg; vehicle=50% PEG/25% EtOH/25% saline.
dRecovery time in min. A compound was identified as inactive in this model if LRR was not observed within 5 min following injection.
eDelayed onset of sedation.
5. Feldman, P. L.; Jung, D. K.; Kaldor, I.; Pacofsky, G. J.;
Stafford, J. A.; Tidwell, J. H. WO 0069836, 2000. CAN
134:4954.
The work described herein had incorporated an enzy-
matically labile ester functionality into a BZD scaffold
to identify 11a as our lead USA BZD compound. Con-
sistent with our strategy, the corresponding acid 11f was
considerably less potent in our receptor binding assay
and lacked the ability to provoke a pharmacological
response in the LRR assay, even at a dose of 100 mg/kg.
Albeit possessing all of the desired in vitro properties, 11a
lacked sufficient aqueous solubility for optimum use as an
injectable in clinical settings. The succeeding manuscript
describes our efforts to address this shortcoming.
6. (a) Remifentanil is marketed under the tradename Ultiva1.
(b) James, M. K.; Feldman, P. L.; Schuster, S. V.; Bilotta, J.
M.; Brackeen, M. F.; Leighton, H. J. J. Pharmacol. Exp. Ther.
1991, 259, 712. (c) Feldman, P. L.; James, M. K.; Brackeen,
M. F.; Bilotta, J. M.; Schuster, S. V.; Lahey, A. P.; Lutz, M.
W.; Johnson, M. R.; Leighton, H. J. J. Med. Chem. 1991, 34,
2202.
7. Lutz, M. W.; Morgan, P. H.; James, M. K.; Feldman, P. L.;
Brackeen, M. F.; Lahey, A. P.; James, S. V.; Bilotta, J. M.;
Pressley, J. C. J. Pharmacol. Exp. Ther. 1994, 271, 795.
8. Bolin, D. R.; Sytwu, I. I.; Humiec, F.; Meienhofer, J. Int.
J. Peptide Protein Res. 1989, 33, 353.
9. Carpino, L. A.; Cohen, B. J.; Stephens, K. J., Jr.; Sadat-
Alaee, S. Y.; Tien, J. -H.; Langridge, D. C. J. Org. Chem.
1986, 51, 3732.
10. For a synthesis of similar ester-containing BZDS from die-
thyl aspartate, see: Manghisi, E.; Salimbeni, A. Boll. Chim. Farm.
1974, 113, 642. These analogues were reported to be inactive.
11. Otera, J.; Ioka, S.; Nozaki, H. J. Org. Chem. 1989, 54,
4013.
References and Notes
1. (a) Marshall, B. E.; Longnecker, D. E. General Anesthetics.
In The Pharmacological Basis of Therapeutics, 8th ed.; Gilman,
A. G., Rall, T. W., Nies, A. S., Taylor, P., Eds.; Pergamon:
New York, 1990; p 303. (b) Rall, T. W. Ibid, p 346.
2. Shafer, A. Crit. Care Med. 1998, 26, 947.
3. Kanto, J. H. Pharmacotherapy 1985, 5, 138.
4. Hudson, R. D.; Wolpert, M. K. Arch Int. Pharmacodyn.
Ther. 1970, 186, 388.
12. Bell, S. C.; McCaully, R. J.; Gochman, C.; Childress, S. J.;
Gluckman, M. I. J. Med. Chem. 1968, 11, 457.

3218
J. A. Stafford et al. / Bioorg. Med. Chem. Lett. 12 (2002) 3215–3218
13. Sherrill, R. G.; Sugg, E. E. J. Org. Chem. 1995, 60, 730.
14. Membrane homogenates were prepared from male Spra-
gue–Dawley rat brain according to the methods described
(Marangos, P. J.; Martino, A. M. Mol. Pharmacol. 1981, 20,
16.) Test compounds were prepared in 100% DMSO at a
concentration of 20–55 mM. Compounds were diluted in assay
buffer such that the first well contained 100 mM (final con-
centration). Eleven 3-fold serial dilutions were prepared in
buffer to complete a 12-point concentration–response curve
for each test compound. Each concentration was tested in tri-
plicate and compounds of interest were tested on at least three
separate occasions. The final concentration of DMSO in each
well did not exceed 0.4%. Nonspecific binding was defined in
the presence of 10 mM 20-chlorodiazepam (Ki=0.5 nM). The
final concentrations of 3H-flunitrazepam were 2 nM. Con-
centration–response curves for midazolam, chlordiazepoxide
or flumazenil were conducted as controls with each assay run.
Radioligand, test compounds and membrane homogenates
were incubated for 90 min at 4 ꢂC in buffer consisting of 50
mM Tris–HCl, pH 7.4, containing 150 mM NaCl. All assays
were conducted in 96-well plates in a total assay volume of 200
mL. Protein concentrations were 12 mg/well. The reaction was
terminated by rapid filtration (Packard Filtermate-196)
through 96-well GF/B filter plates (Packard # 6005177). The
filters were washed, 8 times with 200 mL/well, with ice-cold
Tris 50 mM, pH 7.4 (ꢃ1.6 mL total). Plates were counted
using a Packard TopCount microtiter plate scintillation coun-
ter. Ki values were calculated, see: Cheng, Y.-C.; Prusoff, W.
H. Biochem. Pharmacol. 1973, 22, 3099.
15. Ananthan, S.; Clayton, S. D.; Ealick, S. E.; Wong,
G.; Evoniuk, G. E.; Skolnick, P. J. Med. Chem. 1993, 36,
479.
16. Hester, J. B., Jr.; Rudzik, A. D.; Kamdar, B. V. J. Med.
Chem. 1971, 14, 1078.