3-Phenyl-substituted imidazo[1,5-a]quinoxalin-4-ones and imidazo[1,5- a]quinoxaline ureas that have high affinity at the GABA(A)/benzodiazepine receptor complex

3820

J . Med. Chem. 1996, 39, 3820-3836

3-P h en yl-Su bstitu ted Im id a zo[1,5-a ]qu in oxa lin -4-on es a n d

Im id a zo[1,5-a ]qu in oxa lin e Ur ea s Th a t Ha ve High Affin ity a t th e GABA_A/

Ben zod ia zep in e Recep tor Com p lex^†

E. J on J acobsen,*^,‡Lindsay S. Stelzer,^‡Kenneth L. Belonga,^‡Donald B. Carter,^§Wha Bin Im,^§

Vimala H. Sethy,^§Andrew H. Tang,^§Philip F. VonVoigtlander,^§and J ames D. Petke

Departments of Structural and Medicinal Chemistry, Central Nervous System Diseases Research, and Computational

Chemistry, Pharmacia & Upjohn, Kalamazoo, Michigan 49001

Received J anuary 22, 1996^X

A series of imidazo[1,5-a]quinoxalin-4-ones and imidazo[1,5-a]quinoxaline ureas containing

substituted phenyl groups at the 3-position was developed. Compounds within the imidazo-

[1,5-a]quinoxaline urea series had high affinity for the GABA_A/benzodiazepine receptor complex

with varying in vitro efficacy, although most analogs were partial agonists as indicated by

[³⁵S]TBPS and Cl^-current ratios. Interestingly, a subseries of piperazine ureas was identified

which had biphasic efficacy, becoming more antagonistic with increasing concentration. Analogs

within the imidazo[1,5-a]quinoxalin-4-one series had substantially decreased binding affinity

as compared to the quinoxaline urea series. These compounds ranged from antagonists to full

agonists by in vitro analysis, with several derivatives having roughly 4-fold greater intrinsic

activity than diazepam as indicated by Cl^-current measurement. Numerous compounds from

both series were effective in antagonizing metrazole-induced seizures, consistent with anti-

convulsant properties and possible anxiolytic activity. Most of the quinoxaline ureas and

quinoxalin-4-ones were active in an acute electroshock physical dependence side effect assay

in mice precluding further development.

In tr od u ction

The excitability of many central nervous system

(CNS) pathways is controlled by the action of γ-ami-

nobutyric acid (GABA) on the GABA_Achloride ion

channel complex.¹Of the chemical classes which have

binding sites on this macromolecular ionophore, the

benzodiazepines are the most widely studied. Ligands

which interact with the benzodiazepine receptor (BzR)

and allosterically modulate the action of GABA on

neuronal chloride ion flux have a continuum of intrinsic

activity,²ranging from full agonists (anxiolytic, hyp-

notic, and anticonvulsant agents) through antagonists

to inverse agonists (proconvulsant and anxiogenic

agents). Partial agonists exist within this continuum

and may be devoid of typical benzodiazepine-mediated

side effects such as physical dependence, ethanol po-

tentiation, amnesia, oversedation (for anxiolytic agents),

and muscle relaxation or, in the case of inverse agonists,

convulsive activity.³Furthermore, several different

F igu r e 1. Quinoxalines 1-4.

side effects.^6bThis series, in part, arose from a struc-

receptor subunits (R, â, γ, δ) combine to form the GABA_A

receptor complex,⁴with at least three to four native

ture-activity relationship (SAR) evaluation of a class

of imidazo[1,5-a]quinoxalin-4-ones, of which 2 (U-78875)

was the lead compound.⁷Unfortunately 2 was removed

from clinical trials for the treatment of anxiety due to

increases in liver enzymes. Detailed toxicological evalu-

ation of 2 subsequently indicated that the increase in

serum triglycerides observed is due to cyclopropanecar-

boxylic acid, which results from species-dependent

metabolism of the oxadiazole group.⁸Compounds con-

taining other groups substituted at the 5′-oxadiazole

position, such as tert-butyl 4 (U-94000) or the “reverse”

O-N oxadiazole 3 (U-95963) as shown in Figure 1, also

promoted enzyme induction, whereas methyl, ethyl, and

isopropyl substituents did not.⁹Disappointingly, these

derivatives containing simple C1-C3 alkyl groups were

rapidly metabolized in contrast to 2. Although 1 had a

receptor subtypes identified thus far.⁵The possibility

that subtype selective ligands may discriminate between

useful anxiolytic or hypnotic activity and overt side

effects may also allow for improved medications.

We recently reported on a series of high-affinity

partial agonist imidazo[1,5-a]quinoxalines which were

appended with amide, urea, carbamate, and thiocar-

bamate side chains.⁶Selected compounds such as 1 (U-

91571; Figure 1) displayed excellent antianxiety activity

in animal models yet had limited benzodiazepine-type

^†A portion of this material was presented at the 207th American

Chemical Society National Meeting, March 13-17, 1994.

^‡Department of Structural and Medicinal Chemistry.

^§Department of Central Nervous System Diseases Research.

Department of Computational Chemistry.

^XAbstract published in Advance ACS Abstracts, August 15, 1996.

3-Phenylimidazo[1,5-a]quinoxalines

J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19 3821

Sch em e 1^a

a

Reagents: (a) tBuOK, THF, diethyl chlorophosphate; (b) tBuOK, 7.

Sch em e 2^a

Sch em e 3^a

a

Reagents: (a) EtOCHO; (b) POCl₃, Et₃N, CH₂Cl₂; (c) 50%

NaOH, CHCl₃, BnEt₃NCl, CH₂Cl₂.

a

Reagents: (a) iPrNH₂or tBuNH₂, 150 °C; (b) veratrylamine,

EtNiPr₂, 75 °C; (c) H₂, 5% Pd/C, EtOH; (d) H₂, Pt/S/C, EtOH; (e)

EtOCOCOCl, EtNiPr₂, toluene.

significantly different metabolic profile than 2, concerns

over possible degradation of the oxadiazole group to

release cyclopropanecarboxylic acid discouraged further

interest in this compound.

following the general procedure of Weber,¹²substituted

benzylamine 9 was converted to isocyanide 7 directly.

The various 2,3-dioxoquinoxaline templates 5 were

prepared as illustrated in Scheme 3 following proce-

dures similar to that developed by Watjen.¹³Reaction

of the appropriate 2-halonitrobenzene 11 with isopro-

pylamine provided 12 in excellent yield. Reduction of

the nitro group in 12 was accomplished either via

catalytic hydrogenation or through the use of Raney

nickel or titanium(III) trichloride to provide relatively

unstable diamine 13. Reaction of 13 with ethyl oxalyl

chloride at -78 °C in toluene followed by heating the

reaction mixture at reflux gave the desired diones 5a -

e. Substitution of isopropylamine for either tert-butyl-

amine or veratrylamine provided the appropriately

substituted 2,3-dioxoquinoxaline templates 5f-i, uti-

lized as starting material in Scheme 5. Table 1 high-

lights the physical data of the imidazo[1,5-a]quinoxalin-

4-ones prepared by these methods.

Within the imidazole benzodiazepine template, nu-

merous functional groups have been appended at the

3-position to control affinity and modulate efficacy.^10,11

Most notable are esters (from which the oxadiazoles

were derived), amides, oxadiazoles, thiazoles, furans,

thiophenes, and phenyl rings. Typically compounds

with ester, oxadiazole, or furan substituents have

greater affinity to the benzodiazepine/GABA_Areceptor

complex than the other 3-aryl analogs.¹⁰With the need

to reevaluate other non-oxadiazole substituents at the

3-position in the series derived from both 1 and 2,

analogs containing esters, isoxazoles, and substituted

phenyl groups were synthesized. In this manuscript we

disclose our work directed toward 3-phenylimidazo[1,5-

a]quinoxalin-4-ones and 3-phenylimidazo[1,5-a]quinoxa-

line ureas, as well as the dramatic effects that this

substitution pattern has on affinity and in vitro efficacy.

The 3-aryl-substituted imidazo[1,5-a]quinoxaline ure-

as were prepared by two different routes. As shown in

Scheme 4, exposure of quinoxaline template⁶14 to

phosgene (or triphosgene) provided a carbamoyl chloride

which was reacted directly with an amine to provide

urea 15. Cyclization of 15 using diethyl chlorophos-

phate, potassium tert-butoxide, and isocyanide 7 as

detailed above provided imidazo[1,5-a]quinoxaline 16.

This sequence was satisfactory when R⁴was methyl,

providing 16 in yields of 23-74% (15 f 16). However,

when R⁴was hydrogen, only a trace amount of 16 was

isolated. Presumably in this case, the intermediate enol

phosphonate of 17 (R⁴) H) can be deprotonated by the

relatively basic isocyanide anion leading to decomposi-

tion. In contrast with the gem-dimethyl analogs 17 (R⁴

Ch em istr y

The desired imidazo[1,5-a]quinoxalin-4-ones 8 were

synthesized as shown in Scheme 1. Reaction of 5 with

potassium tert-butoxide and diethyl chlorophosphate

provided enol phosphate ester 6. This intermediate 6,

which was usually not isolated, was reacted with the

desired isocyanide 7 in the presence of additional

potassium tert-butoxide to provide 8.

The aryl isocyanides were prepared by one of two

routes as shown in Scheme 2, starting with the desired

substituted benzylamine. Reaction of 9 with ethyl

formate provided formamide 10 which was dehydrated

with phosphorous oxychloride in the presence of tri-

ethylamine to provide the isocyanide 7. Alternatively

3822 J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19

Ta ble 1. Physical Data for Imidazo[1,5-a]quinoxalin-4-ones

J acobsen et al.

compd

R^3′

R⁶

R⁷

mp (°C)

method

yield (%)

formula

C₁₉H₁₇N₃O

C₂₀H₁₉N₃O₂‚(H₂O)_1/2

C₂₀H₁₉N₃O

anal.

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

H

Cl

CH₃

H

F

207-208

187-189

209-210

211-212

201-202

184-185

180-181

215-217

210-211

151-152

223-224

>300

229-230

286-288

164-165

157-158

179-180

165-166

A

B

A

B

52

41

20

42

31

62

78

74

83

37

15

30

51

43

44

43

50

65

C, H, N

C, H, N, Cl

C, H, N

C, H, N, Cl

C, H, N

4′-OCH₃

4′-CH₃

4′-Cl

C

₁₉H₁₆N₃OCl

₁₉H₁₆N₃OF

4′-F

4′-CF₃

3′-OCH₃

3′-F

3′-CF₃

2′-OCH₃

H

4′-OCH₃

4′-F

4′-OCH₃

4′-F

4′-OCH₃

4′-F

C₂₀H₁₆N₃OF₃

C₂₀H₁₉N₃O₂

C

₁₉H₁₆N₃OF‚(H₂O)_1/8

C₂₀H₁₆N₃OF₃

C₂₀H₁₉N₃O₂

C

₁₉H₁₆N₃OF

C₂₀H₁₈N₃O₂F

F

C

₁₉H₁₅N₃OF₂‚(H₂O)_3/8

Cl

H

₁₉H₁₅N₃OClF‚H₂O

C₂₀H₁₈N₃O₂Cl

₁₉H₁₅N₃OClF

C

C₂₁H₂₁N₃O₂

C₂₀H₁₈N₃OF

C, H, N

Sch em e 4^a

hydride (LAH) provided amines 18a ,b,e-g which were

deprotected with trifluoroacetic acid to give the desired

template 19. Occasionally a small amount of the imine

of 19 was formed during this step which was reduced

with sodium borohydride. An alternative route to 19

was also developed. Deprotection of 8c,d (R¹) tert-

butyl) provided amides 20c,d which were reduced with

either LAH or aluminum hydride to give the imidazo-

[1,5-a]quinoxaline 19. Acylation of 19 with a substi-

tuted benzoyl chloride provided amide 21. Ureas and

carbamates were formed by exposure of 19 to phosgene

or triphosgene to provide carbamoyl chloride 22, which,

typically without isolation, was reacted with the desired

nucleophile (amines, alcohols) to provide 16. The car-

bamoyl chloride could also be isolated and stored

indefinitely before further reaction. The 3,4,5-trimeth-

ylpiperazine analogs 49, 68, 70, and 72 were prepared

from their cis-3,5-dimethylpiperazine precursors by the

reductive amination procedure of Borch.¹⁴Table 2

highlights the physical data and methods used for the

synthesis of the imidazo[1,5-a]quinoxaline ureas, amides,

and carbamates.

Reagents: (a) phosgene in toluene, CH₂Cl₂, EtNiPr₂, R⁵H,

a

EtNiPr₂; (b) tBuOK, THF, diethyl chlorophosphate, tBuOK, 7.

Resu lts a n d Discu ssion

) Me), the enol phosphonate does not have an acidic

hydrogen and reacts cleanly with the isocyanide anion.

Additionally the use of the less basic oxadiazole isocya-

nide anion, which was used successfully (47-70%) in

our previous work,⁶prevents this side reaction.

The binding affinity of the imidazo[1,5-a]quinoxalines

at the benzodiazepine receptor in rat cortical mem-

branes was determined by competition experiments

with radiolabeled [³H]flunitrazepam (Fnz).¹⁵The in

vitro efficacy of these compounds was accessed by two

different methods. The TBPS ratio¹⁶was determined

for each compound by measuring its effect on tert-butyl

bicyclophosphorothionate (TBPS) binding to the picro-

toxin convulsant site on the GABA_Achloride complex.

Differences in TBPS binding presumably occur due to

conformational changes in the chloride ionophore, al-

losterically modulated by the binding of the test com-

pound to the benzodiazepine receptor. The resultant

value, expressed as a ratio of that for the test drug to

that of diazepam, is 1 for a full agonist and 0 for an

antagonist, with negative values for inverse agonists.

A second and more direct measure of in vitro efficacy

was determined by a Cl^-current assay.^16-18The

The use of a blocking group at the 4-position (imidazo-

[1,5-a]quinoxaline numbering) allowed for a successful

resolution to this problem as shown in Scheme 5.

Protected quinoxaline-2,3-diones 5f-i (R¹) tert-butyl

or 3,4-dimethoxybenzyl) were prepared as shown in

Scheme 3. The use of veratrylamine as a protecting

group at this position was necessary for the halogenated

analogs as derivatives of diamine 13f (R⁶or R⁷) halo)

would not cyclize with ethyl oxalyl chloride to provide

5f-i (R⁶or R⁷) halo). The standard imidazole ring-

forming reaction was carried out with 5 to provide

imidazo[1,5-a]quinoxalinones 8a -g. Reduction with

borane-methyl sulfide complex or lithium aluminum

3-Phenylimidazo[1,5-a]quinoxalines

J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19 3823

Sch em e 5^a

a

Reagents: (a) tBuOK, THF, diethyl chlorophosphate, tBuOK, 7; (b) BH₃DMS, THF; (c) TFA, CH₂Cl₂; (d) LAH, THF; (e) AlH₃, THF;

(f) phosgene in toluene, EtNiPr₂, CH₂Cl₂; (g) R⁵H, EtNiPr₂, THF; (h) R⁵Na, R⁵H; (i) ArCOCl, EtNiPr₂, THF.

synaptic chloride conductance effected by GABA acti-

vating the GABA_Areceptor complex is modulated by

ligands acting at the benzodiazepine receptor. Full

agonists increase GABA-mediated current, and antago-

nists have no effect while inverse agonists decrease ion

flow. The test compounds were again compared to

diazepam and thus, like the TBPS assay, had a similar

numerical scale. All compounds were evaluated in the

R₁â₂γ₂subtype in this assay. To provide a quick

measure of in vivo efficacy, most analogs were evaluated

for their ability to antagonize metrazole-induced sei-

zures (clonic and tonic).¹⁹While a direct measure of

anticonvulsant activity, this assay is also predictive of

possible anxiolytic properties as standard full agonist

benzodiazepine (BZD) agents such as diazepam, alpra-

zolam, and zolpidem are extremely effective, as are

partial agonists such as 2, bretazenil, and abercarnil,²⁰

although to a lesser degree.

analogs 29-31 with low affinity (188-480 nM), while

the lone ortho-substituted derivative 32 (o-OMe) was

completely inactive. The substituent on the 3-phenyl

ring also had significant effects on in vitro efficacy. In

the TBPS assay, 23, 29, and 30 were nearly antagonists

and the 4′-fluoro and 4′-methoxy analogs partial ago-

nists, while p-methyl 25 was a full agonist. These

compounds were all significantly more agonistic than 2

in this assay. In the chloride current screen, 23 and

26 were antagonists like 2, while 24, 25, and 27 were

partial agonists. In the meta-series, all three analogs

were weak partial agonists or antagonists as indicated

by both assays except for 29, which had similar efficacy

to that of diazepam in the chloride current assay. In

the metrazole assay, only 24 had reasonable activity

(ED₅₀) 12.5 mg/kg), which is consistent with its affinity

and chloride current ratio. The p-fluoro analog 27 was

also (weakly) active as a metrazole antagonist, consis-

tent with its relatively low affinity and partial agonist

properties in the TBPS and chloride current assays.

Both of these analogs were significantly less potent in

this in vivo assay than 2 (ED₅₀) 1.6 mg/kg), which is

remarkably active given that it is an antagonist by in

vitro determinants.

For the simple quinoxalin-4-one template (R⁶, R⁷)

H), replacement of the oxadiazole of 2 with a phenyl

ring resulted, for the most part, in dramatically de-

creased binding affinity (Table 3). The most direct

comparison (23 vs 2) indicates a 24-fold decrease in

receptor affinity. Relatively small groups such as

p-fluoro and p-methyl resulted in an additional 4-fold

decrease in affinity, while the larger and more elec-

tronegative chloro (26) and trifluoromethyl (28) deriva-

tives were relatively inactive. However, substitution on

the 3-phenyl ring with an electron rich p-OMe group

did provide slightly enhanced affinity with 24 having a

K_iof 32 nM. Substitution at the meta-position provided

Further exploration of this series with substituents

at the 6- and 7-positions provided some intriguing

results. Substitution at the 6-position had profound

effects on affinity and in vitro efficacy. In particular,

6-chloro 38 had an 8-fold improvement in affinity (K_i)

14.6 nM) as compared to 27 (K_iof 115 nM). By the

TBPS assay this compound was 2 times as efficacious

3824 J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19

J acobsen et al.

Ta ble 2. Physical Data for Imidazo[1,5-a]quinoxaline Ureas, Carbamates, and Amides

yield

(%)

compd

R³′

R⁴

R⁵

R⁶

R⁷

mp (°C)

method

formula

anal.

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

H

F

H

NMe₂

H

Cl

H

Cl

H

F

225-226

192-193.5

204-205

205-207

217-218

238.5-239.5

193-194

207-209

167-169

177-180

222-223

196-200

241-243.5

224-225.5

248-249

232-233

215-216

156-160

160-160.5

230-231

176-177

185-189

180-181.5

142-144

280-281

139-141

112-114

203-205

213-214

184-185

177.5-180

C

E

C

E

G

C

E

C

B

D

F

39

74

52

37

72

81

76

87

57

35

63

86

40

74

55

23

48

31

35

42

72

75

78

75

48

42

37

78

64

81

47

49

C₁₉H₁₈N₄O‚(H₂O)_1/5

C₁₉H₁₇N₄OF

C₂₀H₂₀N₄O

C, H, N

C,^aH, N

C, H, N

C, H, N, Cl

Me

H

F

Me

F

H

F

H

F

Me

F

Me

F

OMe

pyrrolidine

morpholine

3,5-dimethylpiperazine

3,4,5-trimethylpiperazine

piperidine

NH₂

NHEt

NHtBu

NMe₂

pyrrolidine

OMe

OiPr

p-F-Ph

m-F-Ph

o-F-Ph

morpholine

3,5-dimethylpiperazine

3,4,5-trimethylpiperazine

3,5-dimethylpiperazine

3,4,5-trimethylpiperazine

3,5-dimethylpiperazine

3,4,5-trimethylpiperazine

C

₂₁H₂₀N₄O

₂₁H₁₉N₄OF

₂₂H₂₂N₄O

₂₁H₁₉N₄O₂F

₂₃H₂₄N₅OF‚(H₂O)_1/4

₂₄H₂₆N₅OF‚(H₂O)_1/4

₂₂H₂₂N₄O‚(H₂O)_1/4

C₁₇H₁₃N₄OF

C

₁₉H₁₇N₄OF

₂₁H₂₁N₄OF

H

Me

H

C₂₁H₂₂N₄O

C₂₁H₂₁N₄OF

C₂₂H₂₄N₄O

C

₂₃H₂₃N₄OF

₂₄H₂₆N₄O

₁₈H₁₄N₃O₂F

₂₀H₁₈N₃O₂F

₂₃H₁₅N₃OF₂

₂₁H₁₈N₄F₂O₂

₂₁H₁₈N₄O₂ClF

F

E

C

E

G

C

G

E

G

Cl

H

F

₂₃H₂₃N₅F₂O‚(H₂O)_1/4C, H, N

F

₂₄H₂₅N₅F₂O‚(H₂O)_1/2C, H, N

H

₂₃H₂₃N₅OClF

₂₄H₂₅N₅OClF

C, H, N, Cl

C,^bH, N, Cl

C, H, N

₂₄H₂₇N₅O₂‚(H₂O)_1/2

₂₅H₂₉N₅O₂‚(H₂O)_1/4

H

C, H, N

a

b

C: calcd, 72.81; found, 73.22. C: calcd, 63.50; found, 62.42.

as diazepam, while the close analog 37 (p-OMe) was a

full agonist (TBPS shift of 1.0). Most fascinating was

that 38 had 7-fold greater efficacy than diazepam in the

chloride current assay (R₁â₂γ₂subtype), while 37 was

also a potent agonist. This degree of efficacy has not

been previously observed in any of the other classes of

BzR ligands that we have studied. Both of these

compounds were extremely effective in the metrazole

antagonism assay. Interestingly, 37 and the two 6-

methyl-analogs, 39 and 40, did not have enhanced

binding affinity over the unsubstituted analogs 24 and

27. Nonetheless, both of the 6-methylquinoxalin-4-ones

(like the 6-chloro derivatives) were full agonists, having

up to 3-fold greater efficacy than diazepam in the

chloride current assay (R₁â₂γ₂subtype), and were potent

metrazole antagonists. The dramatic improvement in

affinity and efficacy through 6-chloro substitution (i.e.,

38) is unique and in clear contrast to classical benzo-

diazepine SAR.^10,11In contrast, substitution at the

7-position with a chloro group (36) resulted in dramati-

cally decreased affinity (15-fold lower than 27) yet had

little effect on in vitro efficacy. The 7-fluoro analogs as

a group were also quite different. While 33 had similar

affinity to 23, 34 suffered a 2-fold decrease in affinity

over unsubstituted 24. The contrast was true for the

35 and 27 pair, with 35 having 2-fold greater affinity.

While all three of the 7-fluoro analogs had greater in

vitro efficacy as compared to the unsubstituted analogs,

none were active in the metrazole antagonism assay.

The disparate values between the TBPS shift and Cl^-

current assays for several of these compounds may, in

part, be attributed to heterogeneity of the GABA_A

receptor population. Classical benzodiazepines interact

with the R₁â₂γ₂, R₂â₂γ₂, R₃â₂γ₂, and R₅â₂γ₂subtypes with

nearly equal affinity,²¹while atypical BzR ligands such

as CL 218872 and Zolpidem have moderate selectivity

for the R₁â₂γ₂subtype.^4a,21However, classical benzo-

diazepines do not interact with the R₆â₂γ₂subtype

located in cerebellar granule cells, whereas ligands such

as Ro 15-4513 have high affinity for this subtype.²²

Furthermore, these BzR ligands often display efficacy

selectivity where Cl^-current is modulated to varying

degrees in these receptor subtypes.²³The search for

selective ligands for these subtypes as well as the

elucidation of their function remains an important goal

in this area.

A few analogs from this series were examined in two

different benzodiazepine receptor subtypes.^24,25Binding

affinity for a number of compounds from this series was

measured in the R₁â₂γ₂and R₃â₂γ₂receptor subtypes

with little selectivity noted. The only compound dis-

playing modest selectivity (4-fold) was 38, which had a

K_iof 17 nM in R₁â₂γ₂and a K_iof 72 nM in R₃â₂γ₂. The

rest of the analogs evaluated had essentially equivalent

3-Phenylimidazo[1,5-a]quinoxalines

J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19 3825

Ta ble 3. [³H]Fnz Binding, TBPS Shift, Cl^-Current Changes, and Metrazole Antagonism Data for Imidazo[1,5-a]quinoxalin-4-ones

Cl^-current^b,c

metrazole^dED₅₀

compd

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

R^3′

R⁶

R⁷

K_i(nM)^a

[

³⁵S]TBPS shift^b,c

R₁â₂γ₂

R₃â₂γ₂

(mg/kg, ip)

>50

12.5

>50

30

H

Cl

CH₃

H

F

38.5

31.7

174

2890

115

0.31

0.75

1.09

0.74

0.58

0.36

0.24

0.31

0.45

0.07

0.45

0.38

0.04

0.43

1.02

1.0

4′-OCH₃

4′-CH₃

4′-Cl

4′-F

0.45

0.36

0.12

4′-CF₃

3′-OCH₃

3′-F

3′-CF₃

2′-OCH₃

H

4′-OCH₃

4′-F

4′-OCH₃

4′-F

4′-OCH₃

4′-F

>10000

207

480

188

>50

0.11

0.44

0.03

>10000

39.6

81.5

59.4

1692

27.9

14.6

49.7

126

>50

0.77

0.97

0.80

0.14

1.04

2.04

0.92

1.83

0.50

1.05

0.72

0.36

2.23

7.0

1.8

3.2

1.0 ( 0.2

0.04

>50

F

Cl

H

0.03

1.0

3.4

1.3

1.7

1.3 (1.0-1.9)

38

39

40

0.9 (0.7-1.3

3.1 (1.7-5.8)

2.6 (1.9-3.7)

0.50 (0.38-1.2)

1.6

diazepam

2

4.9

1.6

1.0 ( 0.2

0.06

1.0 ( 0.2

a

Mean binding affinity against [³H]flunitrazepam; see ref 15 and the Experimental Section for methods. The standard error was

b

<(10% of the mean. Diazepam is defined as a full agonist which gives a value of 1. Antagonists are defined as having a shift value of

0; partial agonists are intermediate. ^cSee the Experimental Section. The standard error was <(10% of the mean. Antagonism of

d

metrazole-induced clonic convulsions in the rat after ip injection; see the Experimental Section.

affinity for each subtype. However, significant differ-

ences in efficacy in the Cl^-current assay were noted

for most of the 3-phenyl quinoxalin-4-ones evaluated in

these two subtypes. All but one of the analogs evaluated

(27) had enhanced efficacy in the R₁â₂γ₂subtype (data

provided in Table 3). Most notably, 29 was a full agonist

in R₁â₂γ₂while an antagonist in the R₃â₂γ₂subtype. For

some of the more agonistic analogs, efficacy was up to

2-fold greater in the R₁â₂γ₂subtype.

the 4′-fluorophenyl substituent provided the most desir-

able profile (high affinity/partial agonist) and was used

for further analog development.

Quite a few additional substituents were explored at

the 5-position. Primary and monoalkyl ureas 51-53

had surprisingly low affinity (49-283 nM) as did the

two carbamate analogs 59 and 60, in marked contrast

to the oxadiazole series.^6bThese analogs were partial

to full agonists as determined by the TBPS shift and

chloride current assays but, other than the potent

agonist 60, were inactive in the metrazole antagonism

assay. The isopropoxy derivative 60, like other car-

bamates in the oxadiazole series,⁶had 2-fold greater

efficacy than diazepam in the Cl^-current screen. Of

the three benzamide analogs, only 62 had reasonable

affinity with a K_iof 5.9 nM. The fluoro groups on the

benzamide substituent had a fairly dramatic and un-

expected effect on both in vitro and in vivo efficacy, with

62 having almost a 3-fold greater enhancement of

chloride current than diazepam. This compound was

also extremely effective in the metrazole antagonism

assay in contrast to 61, presumably due to increased in

vitro efficacy and affinity.

In light of the desirable profile of selected compounds

such as 42 and 45, other dialkyl urea substituents were

examined at the 5-position. A piperidine ring (50) was

also tolerated having similar affinity and efficacy (TBPS)

to 47. However, by chloride current measurement, the

piperidine analog was a partial agonist having about

one-half the efficacy of 47 and was about 10-fold less

potent in the metrazole antagonism assay. Quite

surprisingly, the relatively bulky cis-3,5-dimethylpip-

erazine and cis-3,4,5-trimethylpiperazine groups could

also be incorporated into this template with high affinity

maintained. These piperazines were incorporated into

In contrast to the quinoxalin-4-one series and the

imidazobenzodiazepine literature, the replacement of

the oxadiazole group with a phenyl substituent in the

quinoxaline urea class (dimethyl, pyrrolidino, mor-

pholino) maintained excellent binding affinity as indi-

cated for analogs 41-47 in Table 4. Comparison of 41

to 1 indicates only a 3-fold loss of affinity (3.4 vs 1.0

nM), while pyrrolidine urea analogs with a 3-phenyl (44)

or 3-oxadiazole^6bsubstituent have nearly identical af-

finity. Fluoro and methyl substituents at the para-

position of the 3-aryl ring were also well tolerated with

only 43 suffering a slight loss of affinity as compared

to 41. Like 1, most of these urea analogs were partial

agonists as indicated by their TBPS shift ratios. The

only exception was full agonist 46 which contained a

4′-methyl substituent. Most of these analogs were also

partial agonists in the chloride current assay which was

run only in the R₁â₂γ₂subtype. Again p-methyl sub-

stitution provided for enhanced efficacy, in the general

order Me > F > H for the pyrrolidine analogs. The

morpholine analog 47 was also a full agonist by Cl^-

current measurement, consistent with its exceptional

activity in the metrazole antagonism assay. The other

analogs examined in this assay were also quite effective

at antagonizing metrazole-induced seizures, substan-

tially more potent than 1. From this brief SAR study,

3826 J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19

J acobsen et al.

Ta ble 4. [³H]Fnz Binding, TBPS Shift, Cl^-Current Changes, and Metrazole Antagonism Data for Imidazo[1,5-a]quinoxaline Ureas,

Carbamates, and Amides

[

³⁵S]TBPS

shift^b,c

Cl^-current,^b,c

metrazole^d

ED₅₀(mg/kg, ip)

compd

R^3′

R⁵

R⁶

R⁷

K_i(nM)^a

R₁â₂γ₂

41

42

43

44

45

46

47

48

49

50

51

52

53

54^e

55^e

56^e

57^e

58^e

59

60

61

62

63

64

65

66

67

68

69

70

71

72

1

H

F

NMe₂

H

Cl

H

Cl

H

F

3.41

2.96

11.1

0.37

0.97

1.54

1.61

5.85

13.2

1.65

216

48.6

283

478

163

334

12.6

35.0

47.2

50.8

20.5

5.91

42.4

2.16

15.2

2.97

2.49

5.06

33.4

47.4

1.39

4.16

1.0

0.40

0.53

0.36

0.68

4.4 (2.7-7.2)

4.4 (2.6-7.7)

Me

H

F

Me

F

H

F

H

F

Me

F

Me

F

OMe

pyrrolidine

morpholine

3,5-dimethylpiperazine

3,4,5-trimethylpiperazine

piperidine

NH₂

NHEt

NHtBu

NMe₂

pyrrolidine

OMe

OiPr

p-F-Ph

m-F-Ph

o-F-Ph

morpholine

3,5-dimethylpiperazine

3,4,5-trimethylpiperazine

3,5-dimethylpiperazine

3,4,5-trimethylpiperazine

3,5-dimethylpiperazine

3,4,5-trimethylpiperazine

0.69

0.64

1.19

0.54

0.75

0.31

0.52

0.29

0.42

1.02

0.45

0.49

0.71

0.60

1.05

0.53

0.93

0.82

1.06

0.84

0.70

1.05

0.44

0.72

0.64

0.31

0.62

0.31

0.27

0.69

0.52

0.85

1.08

1.07

0.89

0.56

0.52

0.69

1.00

0.27

0.38

0.80

0.54

0.81

0.21

2.0

0.48

2.75

0.80

0.58

0.78

1.09

1.17

1.25

0.93

1.12

1.09

0.87

0.68

14.8 (8.8-25)

0.30

1.1 (1.1-1.1)

13 (8.4-18.5)

4.5 (2.3-8.8)

>50

17.7 (17.7-17.7)

>50

6.3 (3.9-10.1)

29.7 (17.7-50)

1.3

0.14 (0.08-0.26)

0.76 (0.44-1.30)

3.7 (1.9-7.6)

0.5 (0.35-0.91)

1.3 (1.0-1.9)

1.8 (1.13-3.14)

0.8 (0.5-1.2)

0.8 (0.4-1.5)

21

Cl

H

F

H

a

Mean binding affinity against [³H]flunitrazepam; see ref 15 and the Experimental Section for methods. The standard error was

b

<(10% of the mean. Diazepam is defined as a full agonist which gives a value of 1. Antagonists are defined as having a shift value of

0; partial agonists are intermediate. ^cSee the Experimental Section. The standard error was <(10% of the mean. Antagonism of

d

metrazole-induced clonic convulsions in the rat after ip injection; see the Experimental Section. ^egem-4,4-Dimethyl substitution.

this template as work in the related 3-tert-butyl ester

series revealed that considerably improved ex vivo

binding and pharmacokinetic properties result for this

substitution (vide infra).²⁶Both 48 and 49 suffered only

a 5-10-fold decrease in affinity (K_i) 5.8 and 13.2 nM,

respectively) as compared to 47 or 50. Trimethylpip-

erazine 49 was a partial agonist by both TBPS shift and

chloride current measurements with corresponding

activity in the metrazole antagonism assay. The di-

methyl analog had greater in vitro efficacy (TBPS 0.75;

Cl^-current 0.89) and was extremely effective as a

metrazole antagonist (ED₅₀) 1.1 mg/kg). To modulate

efficacy, the para-substituent on the 3-aryl ring was also

varied to include a methoxy group within the piperazine

urea framework. Both analogs 71 and 72 had high

affinity and were weak partial agonists by TBPS shift

determination. Interestingly, by chloride current mea-

surement, both urea analogs were full agonists with

exceptional potency in the metrazole assay.

pyrrolidine urea led to an improvement in affinity with

57 having a K_iof 12.6 nM. The low affinity observed

for many of these analogs is quite surprising as the

corresponding oxadiazole derivatives^6bhad K_i’s of roughly

3 nM. These urea analogs were all partial agonists

except for 58, which was a full agonist as indicated by

its TBPS shift ratio and, nearly so, as determined by

chloride current measurement. Only 58 was active as

a metrazole antagonist.

The role of chloro and fluoro groups at the 6- and

7-positions on in vitro and in vivo activity was also

examined in the quinoxaline series. The urea groups

were maintained as either a morpholine, cis-3,5-di-

methylpiperazine, or cis-3,4,5-trimethylpiperazine. In

the 6-chloro series the morpholine analog 66 had similar

affinity and in vitro efficacy to 47. The metrazole

activity, however, was decreased by roughly 12-fold. The

two piperazine analogs 69 and 70 were less tolerant of

this substitution with affinity notably decreased. None-

theless, both analogs had similar or enhanced efficacy

over 48 and 49 in the TBPS shift and chloride current

assays. Both of the 6-chloro analogs were also ex-

tremely effective in the metrazole antagonism assay

(ED₅₀) 1.8 mg/kg), which is surprising given their

affinity and the decreased activity of 66 as compared

In contrast to the high binding affinity observed for

most analogs within this series, ureas containing gem-

dimethyl substitution at the 4-position were disappoint-

ing. For the dimethylamino ureas, the gem-dimethyl

analogs suffered between a 30- and 140-fold drop in

affinity. This substitution pattern combined with a

3-Phenylimidazo[1,5-a]quinoxalines

J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19 3827

F igu r e 3. Piperazinyl quinoxaline 73.

low concentrations 49 is nearly a full agonist (-90).

However with increasing concentration, the Cl^-current

decreases to 0% of diazepam, consistent with antago-

nistic properties. At even higher concentrations, 49

behaves as an inverse agonist (>+100). The TBPS shift

ratio was also found to have a dose-dependent relation-

ship (data not shown). The different values reported

for the two screening assays (Table 4) for analogs in the

piperazine class may thus be a reflection of different

concentrations used for the test substance as the TBPS

assay was run at 5.0 µM whereas Cl^-current was

measured at 0.5 µM.

Biphasic effects in the in vitro assays were only

observed for the piperazine ureas. Compounds with

other substituents at the 5-position, or the imidazo[1,5-

a]quinoxalin-4-ones, did not express biphasic properties.

Interestingly, the substituent at the 3-position may be

modified within the quinoxaline piperazine framework

to include oxadiazoles, tert-butyl esters,²⁶and isoxazoles

with the biphasic dose-response curve maintained. As

reported for tert-butyl ester analog 73 (U-97775; Figure

3), the unique biphasic effects observed for these pip-

erazines is most likely the result of these compounds

having two distinct binding sites on the GABA_Acom-

plex, a high-affinity site and a low-affinity site.²⁷To

our knowledge, the biphasic properties observed for

these piperazines are unique and have not been ob-

served for other classes of BzR ligands. Frequently, the

reverse situation does occur with the degree of agonism

increasing with drug concentration. Therapeutically,

compounds such as 49 could be quite unique among BzR

ligands as their agonistic properties would presumably

be limited across a wide dose range, if the in vivo

activity parallels the in vitro properties. In particular,

drug abuse problems, which are common for BzR

ligands, may be lessened for these agents.

On the basis of the results of the in vitro (binding,

TBPS shift, Cl^-current) and metrazole assays, selected

compounds from both series were evaluated for typical

benzodiazepine-type side effects. As with the oxadiazole

series,⁶two of the most important side effects promoted

by BZD full agonists are ethanol potentiation and

physical dependence. The potentiation of the effects of

alcohol by the test compounds was determined by

measuring the drug ED₅₀s for traction and loss of

righting after coadministration. In this assay,^6bdiaz-

epam had ED₅₀’s of 0.54 and 0.86 mg/kg for traction and

loss of righting reflex, respectively. The degree of acute

physical dependence was assessed through an elec-

troshock seizure assay.^28aWithdrawal from chronic

high-dose treatment of benzodiazepines may in some

cases result in signs and symptoms related to hyper-

excitability of the CNS. Clinically, these manifestations

of physical dependence are undesirable as they are

F igu r e 2. Cl^-current dose-response curve of 49. Full agonist

) -100, antagonist ) 0, full inverse agonist ) +100.

to 47. In the 7-chloro series only the morpholine analog

65 was synthesized. This derivative had decreased

affinity as compared to 47. The TBPS shift ratio was

notably increased for 65, while the chloride current was

lower. Even with decreased affinity, 65 was quite

effective in the metrazole antagonism assay. Substitu-

tion with a 7-fluoro group maintained (for morpholine)

or enhanced (for 67 and 68) binding affinity as compared

to the unsubstituted analogs. The TBPS shift ratio is

consistent with the three fluoro analogs as being partial

agonists, while by chloride current determination, the

piperazine analogs 67 and 68 were full agonists. All

three urea analogs were quite potent in the metrazole

assay. Overall, substitution with a chloro group at the

6- or 7-position or with a 7-fluoro often resulted in

decreased affinity with mixed effects on in vitro efficacy.

Frequently the chloride current was increased for the

halogenated analogs, while metrazole activity was usu-

ally maintained or enhanced. In marked contrast,

halogenated analogs in the oxadiazole series were

devoid of metrazole activity.^6bThe dramatic effects of

6-Cl substitution observed in the 3-phenyl quinoxalin-

4-one series were absent in this class. Evaluation of

selected members of this class for affinity and efficacy

in the R₁â₂γ₂and R₃â₂γ₂subtypes revealed little selec-

tivity. Of the compounds screened, only 66 displayed

modest selectivity (6-fold) for the R₁â₂γ₂subtype (K_i)

1.6 vs 9.5 nM), while no analogs exhibited differential

efficacy between these subtypes.

Within the quinoxaline urea series, the values for Cl^-

current and the TBPS shift ratio were consistently quite

different, with greater efficacy usually indicated by the

chloride current screen. While this is often the case as

the chloride current assay was usually determined in

only a single receptor subtype (R₁â₂γ₂), whereas the

TBPS shift ratio was measured across all native recep-

tors as mentioned previously, the frequent discrepancy

between these values (particularly for the piperazines)

was intriguing. Further investigation of this subclass

revealed that the differences between these assays may,

in part, be explained due to a biphasic dose-response

curve observed for the piperazine analogs in both of

these assays. The dose-response curve for chloride

current (R₁â₂γ₂subtype) of 49, which is typical of the

piperazine class, is shown in Figure 2. As apparent, at

3828 J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19

J acobsen et al.

Ta ble 5. Benzodiazepine Antagonism, Ethanol Potentiation, and Acute Physical Dependence Data for Imidazo[1,5-a]quinoxalines

ethanol potentiation ED₅₀

acute electroshock physical

(95% confidence interval) (mg/kg)

BZD antag traction^aED₅₀(95%

confidence interval) (mg/kg)

dependence activity^b,c(drug MA₅₀

/

compd

24

27

37

traction^a

LRR

V122 MA₅₀)

>30

14 (7-28)

>30

IA (20.9/20.9)

IA (19.3/21.2)

A (15.0/21.9)

A (17.7/19.5)

A (15.4/19.7)

A (15.9/18.1)

A (16.9/19.4)

A (17.9/20.3)

A (13.9/20.7)

A (12.5/17.7)

A (16.5/20.5)

A (14.9/18.1)

A (17.7/21.1)

A (18.1/20.7)

A (14.8/21.1)

A (12.1/19.4)

A (17.0/19.7)

38

39

40

42

43

45

47

48

49

50

60

62

64

65

66

>30

4.9 (2-9)

11 (6-19)

>30

0.34 (0.19-0.61)

4.0 (2-9)

>30

0.5 (0.2-1.0)

>30

5 (3-10)

11 (6-19)

17 (17-17)

27 (15-49)

67

68

>30

A (12.4/21.0)

69

70

71

72

>30

A (15.7/19.5)

A (9.4/19.7)

A (11.9/19.7)

A^d(18.2/25.4)

diazepam

0.54

0.86

(0.54-0.54)

(0.5-1.4)

2

1.7 (0.8-3.8)

6.6

>30

IA (21.2/22.2)

(3.3-13)

27

1

>30

IA (23.2/24.9)

a

b

d

See the Experimental Section. p < 0.05; IA no significant activity. ^cSee the Experimental Section. Tested at 15 mg/kg.

similar to the disorder being treated.^28bFurthermore,

the abuse potential for drugs inducing physical depen-

dence generally leads to scheduling of the compound by

the regulatory agencies. With this in mind, it is highly

desirable that new benzodiazepine-related therapeutic

agents under development have minimized physical

dependence-inducing properties. In animals, changes

in hyperexcitability can be quantified by determining

the current seizure threshold (MA₅₀) to elicit elec-

troshock-induced seizures. In this assay, the lowering

of electroshock thresholds precipitated by a benzodiaz-

epine antagonist after an acute regimen of test com-

pound was used to quantify the development of physical

dependence. Mice having received either drug or vehicle

were assessed for electroshock seizure thresholds (drug

MA₅₀compared to vehicle MA₅₀) 5 min after flumazenil-

precipitated withdrawal. Diazepam was active at doses

as low as 15 mg/kg/day in this assay. In contrast, both

1 and 2 did not produce physical dependence at doses

up to 150 mg/kg/day. In the alcohol enhancement

screen, only 2 promoted traction (ED₅₀) 6.6 mg/kg)

with no effects on loss of righting observed for either

compound. Finally, while not a direct measure of a

benzodiazepine-type side effect, many of the analogs

were evaluated for their ability to antagonize the muscle

relaxation effects of triazolam, a full agonist BzR ligand.

Diazepam and related full agonist benzodiazepines were

inactive in this assay, whereas partial agonists such as

1 and 2 were quite effective, with ED₅₀’s of 1.0 and 1.7

mg/kg, respectively.

corresponding oxadiazole analogs, which had similar in

vitro activity, were effective triazolam antagonists.^6bIn

the ethanol potentiation screen, most analogs did not

have significant effects on the loss of righting reflex,

with only 64 and 65 weakly active. Like 2 and several

compounds from the series derived from 1, a number of

analogs had significant effects on traction. Quinoxaline

ureas 45 and 50 were quite potent with ED₅₀’s similar

to that of diazepam, although neither of these deriva-

tives were full agonists by the TBPS shift and Cl^-

current screens. Of the five quinoxalin-4-ones tested

in the acute physical dependence screen, only 27 and

37 were inactive. The lack of activity of 27 in this side

effect screen is consistent with its in vitro behavior (low-

efficacy partial agonist) in contrast to 37, which would

be expected to be active due to its full agonist properties.

The other three analogs within this series were active

in the electroshock assay, consistent with their “super”

agonist in vitro profile. Unfortunately, all of the 3-phen-

ylquinoxaline urea analogs of 1 were active in the acute

physical dependence screen.

The potent activity observed in the acute physical

dependence assay across the quinoxaline urea series is

completely unexpected considering that many of these

phenyl derivatives have in vitro properties consistent

with that of a partial agonist, like the corresponding

oxadiazole analogs, which often did not produce physical

dependence. Furthermore, several analogs had exceed-

ingly low electroshock seizure thresholds, lower than

that produced by diazepam, albeit at 10 times the dose

(150 mg/kg for the test compound as compared to 15

mg/kg diazepam). The most efficacious compounds in

this side effect assay contained either piperazines or

morpholine urea groups at the 5-position. Additionally,

an electron-donating substituent (OMe) at the para-

position of the 3-phenyl ring had significant effects on

Even though many of the quinoxalines had low

efficacy in the TBPS and Cl^-current assays, consistent

with antagonist or partial agonist profiles, only one

compound 42 (ED₅₀) 4.0 mg/kg) was active as a

benzodiazepine antagonist in the in vivo assay (Table

5). These results are quite surprising as many of the

3-Phenylimidazo[1,5-a]quinoxalines

J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19 3829

F igu r e 4. Molecular structures of the two lowest-energy

conformers of 23.

F igu r e 5. Molecular structures of the two lowest-energy

conformers of 41.

activity with 71 and 72 having the lowest electroshock

potentials after drug withdrawal. The oxadiazole ana-

log of 47 was active in this assay^6bbut had a signifi-

cantly greater seizure threshold: 86% of control vs 67%

of control for 47. Similarly, the oxadiazole analogs

containing piperazine ureas were also active but again

had significantly higher seizure threshold values (83%

of control). These results suggest that the high degree

of activity of these piperazines (and morpholines) in the

acute physical dependence assay is not directly related

to their biphasic properties. Instead, the potent effects

of the 3-phenylquinoxaline ureas in the physical depen-

dence assay as compared to the oxadiazole analogs are

consistent with the 3-phenyl group promoting this side

effect. The reasons for the remarkable differences

between the phenyl and oxadiazole series in this assay

are unclear but may relate to changes in the binding

interactions of the 3-substituent, subtype selectivity,

metabolism, or CNS penetration.

energy orientations corresponding to out-of-plane rota-

tions of (50°.

For the imidazoquinoxaline urea template, calcula-

tions using the AMBER* and MM2* force fields yielded

very similar structures but differing energetics. Figure

5 shows the two lowest-energy conformers of 41 ob-

tained using the AMBER* force field. Structures A and

B are identical except for the orientation of the 5-sub-

stituent. The 3-phenyl group is rotated 30° out-of-plane,

with the edge of the phenyl ring nearer to N₂located

below the plane. With the exception of the 3-substitu-

ent, structures A and B of 41 are identical with those

of the oxadiazole analog 1.^6bFor the gem-dimethyl

analog 54, two low-energy conformers with imidazoqui-

noxaline rings and 5-substituents similar to those shown

in Figure 5 were obtained. However, the 3-phenyl group

in both conformers of 54 was constrained to lie ap-

proximately perpendicular to the plane.

Molecu la r Con for m a tion s. The low-energy con-

formers of a number of the present analogs were

determined by molecular mechanics methods. Calcula-

tions were performed with both the AMBER* and MM2*

force fields, which were developed for the MacroModel²⁹

system of programs as extensions of the original AM-

BER³⁰and MM2³¹parameter sets. For each analog, 200

initial structures were generated using an internal

coordinate Monte Carlo procedure³²in which torsional

angles around all rotatable bonds were allowed to vary.

Subsequently, each initial structure was subjected to

energy minimization, and duplicate minimized struc-

tures were discarded in compiling the final set of

conformers. It should be noted that at each relative

energy minimum, a pair of mirror image conformers was

obtained, only one of which is described below.

In studies of the quinoxalin-4-one series of com-

pounds, both the AMBER* and MM2* force fields gave

essentially the same structures and relative energies,

and thus only data obtained using the AMBER* force

field are discussed below. Figure 4 shows the two

lowest-energy conformers of 23, structure A being more

stable than B by only 0.67 kcal/mol. The two structures,

which are typical of most analogs in this series, are

identical except for the orientation of the isopropyl

group, which in structure B is rotated 150° relative to

that in structure A, as shown. The 3-phenyl group is

not coplanar with the imidazoquinoxaline substructure;

instead it may assume one of two possible minimum-

Relative energies calculated for 41 and 54 followed

the same trends obtained previously for other imidazo-

quinoxalines.^6bSpecifically for 41, structure B was

more stable than A by 0.74 kcal/mol according to the

AMBER* calculations, but the reverse order and a

larger energy difference of 3.38 kcal/mol was obtained

using MM2*. Therefore, the preferred conformation of

41 and related analogs could not be determined from

the present methods. In the case of 54, both approaches

predicted structure A to be more stable than B, with

energy differences of 2.89 and 6.29 kcal/mol from

AMBER* and MM2*, respectively.

P h a r m a cop h or e Mod els. As discussed in a previ-

ous report,^6bthe imidazoquinoxaline series of BzR

ligands exhibit a number of features consistent with

general models of BzR recognition and activation pro-

posed by Wermuth et al.³³and Gardner.³⁴In particular,

the fused phenyl component of the quinoxaline ring and

the N₂atom act as the aromatic region and hydrogen-

acceptor site (δ₁), respectively, regarded as two impor-

tant features necessary for receptor recognition.^33,34

Furthermore, in the quinoxalin-4-one series, the C₄-

carbonyl group may correspond to a second hydrogen-

acceptor site (δ₂) described in models developed by

Villar³⁵and Cook,³⁶and also included in the Wermuth

model. However, the δ₁-δ₂distance of 4.2-4.4 Å in 23

and related analogs is somewhat longer than the range

of 3.0-3.5 Å specified in the Villar model. In the

3830 J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19

J acobsen et al.

quinoxaline urea series, there is no substituent that

correlates to the δ₂site in these models.^6b

The R⁵substituent corresponds to the “out-of-plane”

region described by Wermuth, which is believed to

correlate with full agonist activity.³³In the present

study, this region appears to be particularly effective

in enhancing efficacy when R⁵is an isopropyl group

forced into an out-of-plane position as in the 6-substi-

tuted analogs 38 and 40. Similar effects have been

reported previously for the 2 series of analogs as well.^7b

Furthermore, bulky R⁵substituents in this out-of-plane

region in the quinoxaline urea series were generally

found to increase efficacy.^6b

In the present imidazoquinoxalines, the 3-phenyl

group correlates with both the “freely rotating aromatic”

group in the Wermuth model and the ester group cited

in the Gardner model. As shown by the present data,

the choice of 3-substituent can have a major influence

on binding affinity. The general trends in receptor

affinity can be rationalized by examining the energy

computed as a function of the rotation angle of the

3-substituent, as illustrated by the following discussion.

Energies were calculated by means of constrained

molecular mechanics energy minimizations using the

AMBER* force field. Each initial structure was con-

structed from the minimum-energy conformer by rotat-

ing the 3-substituent through a specific angle, φ; sub-

sequently each structure was subjected to energy

minimization using an additional torsional potential of

sufficient strength to hold the 3-substituent in its initial

orientation.

F igu r e 6. Relative energies (∆E) of 2, 23, and 32 as a function

of the rotation angle of the 3-substituent (φ). The value φ ) 0

corresponds to the 3-substituent positioned coplanar with the

imidazole ring. Solid lines denote ranges of φ corresponding

to free rotation of the 3-substituent with no additional distor-

tions of the structure. Dashed lines indicate ranges of φ where

restricted rotation and distortions of the imidazoquinoxaline

ring were observed.

Figure 6 compares the energy of 2 as a function of φ

with that of 23 and 32. In 2, the oxadiazole substituent

is positioned 30° out of the imidazoquinoxaline plane

in the minimum-energy structure and can easily be

rotated into a planar position with the expenditure of

only 0.65 kcal/mol. In contrast, the 3-phenyl substituent

in 23 is seen to be confined to an out-of plane orientation

(45-50°) and is not free to assume coplanarity with the

imidazoquinoxaline ring. The phenyl ring though can

be rotated to within 25-30° of planarity with only a

modest energy expenditure (1.7-1.0 kcal/mol). More-

over, for 32, an inactive compound, the 3-phenyl group

may not be rotated to within 45° of planarity without

both a substantial increase in energy and a significant

distortion of the imidazoquinoxaline ring system.

ment of the conditions which must be placed on the

3-substituent to achieve high affinity at the BzR.

Con clu sion

Substitution of the oxadiazole group for a phenyl ring

at the 3-position in a series of imidazo[1,5-a]quinoxalin-

4-ones invariably resulted in notably decreased binding

affinity for the BzR. While in vitro efficacy was mod-

erately enhanced for the phenyl analogs, anti-metrazole

activity was dramatically decreased in comparison to

the oxadiazole analogs. Uniquely, substitution at the

6-position with a chloro group resulted in significantly

enhanced affinity and in vitro efficacy. Up to 7-fold

greater efficacy than diazepam was noted for compound

38 in the Cl^-current assay, measured in cells express-

ing the R₁â₂γ₂subtype of GABA_Areceptors. Similar

substitution with a methyl group at this position also

led to highly efficacious compounds, although affinity

was not increased. Most compounds from this series

had greater efficacy in the R₁â₂γ₂subtype as compared

to R₃â₂γ₂, although little binding selectivity was ob-

served.

Figure 7 shows a similar comparison for 1 and analogs

41 and 54. The preferred orientation of the oxadiazole

substituent in 1 is seen to be approximately planar, and

out-of-plane geometries are increasingly unfavorable. In

the minimum-energy structure of 41, the 3-phenyl group

is found with φ ) 33°, but it may be rotated to within

10° of planarity with a penalty of only 1 kcal/mol.

Moreover, the 3-phenyl group in 41 may be freely

rotated through a full 360° at a cost of no more than

2.70 kcal/mol. In contrast, the 3-phenyl group in the

low-affinity gem-dimethyl analog 54 is clearly confined

to a nonplanar conformation.

In a related class of imidazo[1,5-a]quinoxaline ureas,

replacement of the oxadiazole group by a phenyl ring

usually maintained the high affinity typically observed

in this series. The major exceptions proved to be

compounds containing gem-dimethyl substituents at the

4-position. In both of these series, binding affinity

correlated quite well with the ability of the 3-phenyl ring

The preceding analysis shows that there is a clear

correlation between binding affinity and the ability of

the 3-substituent to assume a planar orientation at

minimum-energy cost. This observation is consistent

with the general tenets of the Wermuth and Gardner

models and, in addition, provides a more precise state-

3-Phenylimidazo[1,5-a]quinoxalines

J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19 3831

the indicated organic solvent, separate the organic layer,

extract the aqueous layer several times with the organic

solvent, dry the combined organic layers with the indicated

drying agent, filter off the drying agent, and remove solvent

using a rotary evaporator at reduced pressure. When “basic

workup (organic solvent, aqueous basic solvent, drying agent)”

is indicated, the procedure was similar to aqueous workup,

except the indicated aqueous base was used instead of H₂O.

When “acidic workup (organic solvent, organic solvent, drying

agent)” is indicated, the procedure was to dilute the reaction

mixture with the first indicated solvent, extract the organic

solution several times with 10% HCl, basify the combined

acidic layers with solid KOH, extract the basic solution with

the second indicated organic solvent several times, dry the

organic layers with the indicated drying agent, filter off the

drying agent, and remove solvent using a rotary evaporator

under reduced pressure. Tetrahydrofuran (THF) and ether

were distilled from sodium and benzophenone. Dichlo-

romethane was distilled from calcium hydride, and DMF was

dried over 3 Å molecular sieves. All other solvents were EM

Science HPLC grade, distilled in glass. Diethyl chlorophos-

phate and potassium tert-butoxide (1.0 M in THF) were

purchased from the Aldrich Chemical Co., Milwaukee, WI.

Phosgene in toluene (CAUTION: phosgene is highly toxic and

should be used with extreme care) was purchased from Fluka

Chemie AG or Columbia. All reactions were run under

nitrogen or argon.

P r ep a r a t ion of 2,3-Dioxoq u in oxa lin e Tem p la t es 5.

N-Isop r op yl-2-m eth yl-6-n itr oa n ilin e (12e). A mixture of

2-bromo-3-nitrotoluene (11e, Y ) Br; 5.30 g, 24.5 mmol) and

isopropylamine (30.0 mL) was heated at 150 °C in a bomb for

36 h. After cooling to room temperature the solution was

concentrated. Basic workup (CH₂Cl₂, NaHCO₃, MgSO₄) gave

4.75 g (100%) of the product as an orange oil: IR (neat) 1605,

F igu r e 7. Relative energies (∆E) of 1, 41, and 54 as a function

of the rotation angle of the 3-substituent (φ). The conventions

used in this figure are the same as those in Figure 6.

1

1533, 1485, 1451, 1339, 1325, 1264, 742 cm^-1; H NMR (300

MHz, CDCl₃) δ 7.91 (d, J ) 8.5 Hz, ArH), 7.32 (d, J ) 7.3 Hz,

ArH), 6.82 (apparent t, J ) 7.9 Hz, ArH), 6.5-6.7 (m, NH),

3.55-3.75 (m, NHCH), 2.37 (s, ArCH₃), 1.15 (d, J ) 6.4 Hz,

CH(CH₃)₂); MS (EI) m/ e 194, 179, 132.

to rotate to near planarity to the imidazoquinoxaline

ring system. Most of the urea analogs retained a partial

agonist in vitro profile, regardless of the 3-phenyl

substituent. The 5-position was widely tolerant of a

number of functional groups (ureas, amides, carbam-

ates) with relatively bulky substituents providing de-

rivatives with good binding affinity. Compounds from

this series did not display selectivity for the R₁â₂γ₂and

R₃â₂γ₂subtypes (affinity, efficacy). Even with a partial

agonist in vitro profile, many of the urea analogs

displayed excellent activity in the metrazole antagonism

assay. An unprecedented biphasic dose-efficacy rela-

tionship was noted for the piperazine urea subseries

where, by in vitro measurement, the analogs became

increasingly antagonistic with increasing drug concen-

tration, theoretically limiting agonist-type properties at

high dose levels. Unfortunately, almost all of the

analogs from both series, which displayed anxiolytic-

like activity in the in vitro and in vivo screening assays,

produced an unacceptably high level of physical depen-

dence in an acute electroshock assay, precluding further

development.

1-Isop r op yl-6-m eth yl-1,2-p h en ylen ed ia m in e (13e).

A

mixture of 12e (5.19 g, 26.7 mmol), ethanol (350 mL), and 5%

Pd/C (1.3 g) was hydrogenated at room temperature for 4 h.

The mixture was filtered, the solids were washed with ethanol,

CH₂Cl₂, and ethanol, and the combined filtrates were concen-

trated to give 4.21 g (96%) of 13e as an oil: ¹H NMR (300 MHz,

CDCl₃) δ 6.80 (apparent t, J ) 7.7 Hz, ArH), 6.55-6.65 (m,

ArH, 2 H), 3.6-4.0 (m, NH), 3.3-3.5 (m, NHCH(CH₃)₂), 2.23

(s, ArCH₃), 1.13 (d, J ) 6.4 Hz, CH(CH₃)₂).

1,2,3,4-Tetr a h yd r o-1-isop r op yl-8-m eth yl-2,3-d ioxoqu i-

n oxa lin e (5e). Ethyl oxalyl chloride (3.0 mL, 27 mmol) was

added to a solution of 13e (4.21 g, 25.6 mmol), toluene (200

mL), and diisopropylethylamine (6.0 mL, 34 mmol) at -78 °C.

The solution was stirred at -78 °C for 1 h and allowed to warm

to room temperature over 16 h. As starting material was still

present, the solution was recooled to -78 °C, and additional

ethyl oxalyl chloride (1.0 mL, 9.0 mmol) and diisopropylethyl-

amine (2.0 mL, 11 mmol) were added. After gradually warm-

ing to room temperature the solution was heated at reflux for

24 h. Upon cooling to room temperature, the residue was

diluted with water (200 mL) and ether (200 mL). The solids

were filtered, washed with water and ether, and dried to give

2.49 g (45%) of dione 5e as a tan powder (mp 248-249 °C):

IR (mineral oil) 1691, 1676, 1405, 778 cm^-1

;

¹H NMR (300

MHz, CDCl₃) δ 7.0-7.15 (m, ArH, 3 H), 4.35-4.5 (m,

NCH(CH₃)₂), 2.60 (s, ArCH₃), 1.69 (d, J ) 6.7 Hz, CH(CH₃)₂);

MS (EI) m/ e 218, 176, 148.

Following this general procedure 1,2,3,4-tetrahydro-1-iso-

propyl-2,3-dioxoquinoxaline (5a ), 1-tert-butyl-1,2,3,4-tetrahy-

dro-2,3-dioxoquinoxaline (5f), 7-fluoro-1,2,3,4-tetrahydro-1-

isopropyl-2,3-dioxoquinoxaline (5c), 7-chloro-1,2,3,4-tetrahydro-

1-isopropyl-2,3-dioxoquinoxaline (5b), and 8-chloro-1,2,3,4-

tetrahydro-1-isopropyl-2,3-dioxoquinoxaline (5d ) were pre-

pared.¹³

Exp er im en ta l Section

Ch em istr y. Thin-layer and flash chromatography utilized

E. Merck silica gel (230-400 mesh). Melting points were

taken on a Thomas-Hoover capillary melting point apparatus

and are uncorrected. Mass spectra, infrared spectra, and

combustion analysis were obtained by the Physical and

Analytical Chemistry Department of The Upjohn Co. ¹H NMR

spectra were recorded at 300 MHz with a Bruker Model AM-

300 spectrometer.

In cases where synthetic intermediates or products were

isolated by “aqueous workup (organic solvent, drying agent),”

the procedure was to quench the reaction with H₂O, dilute with

Exa m p le P r oced u r e for th e Syn th esis of Ben zylfor -

m am ides 10. [4-(Tr iflu or om eth yl)ben zyl]for m am ide (10f).

Ethyl formate (11.1 mL, 137 mmol) was added dropwise to

3832 J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19

J acobsen et al.

4-(trifluoromethyl)benzylamine (9f; 20.0 g, 114 mmol) at 0 °C.

The mixture was stirred for 2 h at 0 °C and for 2 h at room

temperature. Hexane was added to the solution; the resultant

precipitate was filtered and dried to give 15.1 g (65%) of 10f

as a white solid (mp 70-72 °C): ¹H NMR (300 MHz, CDCl₃)

δ 8.31 (s, CHO), 7.60 (d, J ) 8.2 Hz, ArH, 2 H), 7.41 (d, J )

8.1 Hz, ArH, 2 H), 6.05 (br s, NH), 4.55 (d, J ) 6.2 Hz, NHCH₂).

Also prepared by this method were the following compounds.

[3-(Trifluoromethyl)benzyl]formamide (10i; 95%): ¹H NMR

(300 MHz, CDCl₃) δ 8.27 (s, CHO), 7.4-7.65 (m, ArH, 4 H),

6.37 (br s, NH), 4.52 (d, J ) 6.2 Hz, NHCH₂). (3-Methoxy-

benzyl)formamide (10g; 90%): ¹H NMR (300 MHz, CDCl₃) δ

8.27 (s, CHO), 7.2-7.3 (m, ArH), 6.75-6.9 (m, ArH, 3 H), 5.91

(br s, NH), 4.46 (d, J ) 5.9 Hz, NHCH₂), 3.80 (s, OCH₃). (3-

Fluorobenzyl)formamide (10h ; 89%): ¹H NMR (300 MHz,

CDCl₃) δ 8.29 (s, CHO), 7.2-7.45 (m, ArH), 6.85-7.2 (m, ArH,

3 H), 5.7-6.2 (br s, NH), 4.49 (d, J ) 6.1 Hz, NHCH₂). (4-

Chlorobenzyl)formamide (10d ; 83%): ¹H NMR (300 MHz,

CDCl₃) δ 8.27 (s, CHO), 7.15-7.4 (m, ArH, 4 H), 5.96 (br s,

NH), 4.45 (d, J ) 6.1 Hz, ArCH₂). (4-Fluorobenzyl)formamide

(10e; 88%): ¹H NMR (300 MHz, CDCl₃) 8.21 (s, CHO), 7.15-

7.3 (m, ArH, 2 H), 6.95-7.05 (m, ArH, 2 H), 6.37 (br s, NH),

4.41 (d, J ) 6.0 Hz, ArCH₂).

Exa m p le P r oced u r e for th e Syn th esis of Ben zyl Iso-

cya n id es 7. 4-(Tr iflu or om eth yl)ben zyl Isocya n id e (7f).

Phosphorus oxychloride (6.9 mL, 74 mmol) was added dropwise

to a solution of formamide 10f (15.1 g, 74.3 mmol), triethyl-

amine (31.0 mL, 222 mmol), and CH₂Cl₂(200 mL) at 0 °C.

The solution was stirred at 0 °C for 1 h, and aqueous sodium

carbonate (7.85 g in 100 mL of H₂O, 74.1 mmol) was added.

Stirring was continued for 1 h. Aqueous workup (CH₂Cl₂,

MgSO₄) and purification by flash chromatography (5 f 20%

ethyl acetate/hexane) provided 10.4 g (76%) of 7f as an off-

white solid: ¹H NMR (300 MHz, CDCl₃) δ 7.68 (d, J ) 8.2 Hz,

ArH, 2 H), 7.49 (d, J ) 8.1 Hz, ArH, 2 H), 4.73 (s, NCH₂).

Also prepared by this procedure were the following com-

pounds. 3-(Trifluoromethyl)benzyl isocyanide (7i; 86%): ¹H

NMR (300 MHz, CDCl₃) δ 7.5-7.7 (m, ArH, 4 H), 4.73 (s,

NCH₂). 3-Methoxybenzyl isocyanide (7g; 86%): ¹H NMR (300

MHz, CDCl₃) δ 7.25-7.35 (m, ArH), 6.85-6.95 (m, ArH, 3 H),

4.61 (s, NCH₂), 3.82 (s, OCH₃). 3-Fluorobenzyl isocyanide (7h ;

80%): ¹H NMR (300 MHz, CDCl₃) δ 7.3-7.45 (m, ArH), 7.0-

7.2 (m, ArH, 3 H), 4.66 (s, ArCH₂N). 4-Chlorobenzyl isocyanide

(7d ; 72%): ¹H NMR (300 MHz, CDCl₃) δ 7.35-7.4 (m, ArH, 2

H), 7.25-7.35 (m, ArH, 2 H), 4.61 (s, ArCH₂). 4-Fluorobenzyl

isocyanide (7e; 84%): ¹H NMR (300 MHz, CDCl₃) δ 7.3-7.4

(m, ArH, 2 H), 7.0-7.15 (m, ArH, 2 H), 4.62 (s, CH₂).

6.34 mmol) and DMF (1.5 mL) was cooled to -42 °C. Isocya-

nide 7a (0.77 mL, 6.3 mmol) was added and the mixture

allowed to warm to 0 °C. This mixture was transferred to the

solution of the quinoxaline which was then allowed to warm

to room temperature. After 1 h the reaction was quenched

with 1.0 mL of acetic acid, and the mixture was concentrated.

The residue was triturated with ethyl acetate to provide 580

mg of 23 as a white powder. The filtrate was concentrated

and purified by flash chromatography (20:1 CH₂Cl₂:acetone)

to provide an additional 189 mg (0.769 g total, 52%) of the

desired product. An analytical sample was prepared by

recrystallization from CH₂Cl₂/hexane (mp 207-208 °C): IR

(mineral oil) 3089, 1662, 1656, 1515, 1501, 1478, 1304, 1298,

1221, 745, 692 cm^-^{1 1}H NMR (300 MHz, CDCl₃) δ 8.46 (s,

;

ArH), 8.17 (dd, J ) 8.5, 1.5 Hz, ArH, 2 H), 7.80 (d, J ) 8.1 Hz,

ArH), 7.54 (d, J ) 8.5 Hz, ArH), 7.15-7.5 (m, ArH, 5 H), 5.2-

5.6 (br s, CH(CH₃)₂), 1.65 (d, J ) 7.0 Hz, CH(CH₃)₂); MS (EI)

m/ e 303, 261. Anal. (C₁₉H₁₇N₃O) C, H, N.

Exa m p le P r oced u r e for Cycliza tion of 5 w ith Isocya -

n id e 7. Meth od B. 3-(3-F lu or op h en yl)-5-(1-m eth yleth yl)-

im id a zo[1,5-a ]qu in oxa lin -4(5H)-on e (30). Potassium tert-

butoxide (5.20 mL, 5.20 mmol, 1.0 M in THF) was added to a

mixture of 5a (1.00 g, 4.90 mmol) and THF (20.0 mL) at -20

°C. The mixture was allowed to warm to 0 °C over 45 min

and was cooled to -40 °C. Diethyl chlorophosphate (0.75 mL,

5.2 mmol) was added, and the mixture was allowed to warm

to room temperature over 30 min. The resultant solution was

cooled to -78 °C, and a solution of isocyanide 7h (723 mg, 5.35

mmol) and THF (1.0 mL) was added. Potassium tert-butoxide

(5.20 mL, 5.20 mmol) was then added dropwise over several

minutes. The solution was stirred at -78 °C for 30 min and

allowed to warm to room temperature over 2.5 h. After

stirring at room temperature for 2 h, water was added. The

solids were collected, washed with water (3 × 30 mL) and ether

(3 × 30 mL), and dried in vacuo to give 1.04 g of 30.

Recrystallization from hot EtOAc/MeOH/hexane gave 866 mg

of 30 as a white wool-like material (mp 215-217 °C). Aqueous

workup (ethyl acetate, MgSO₄) of the combined filtrates,

purification by flash chromatography (3:1 hexane:ethyl ac-

etate), and recrystallization as above provided an additional

306 mg (1.17 g total, 74%) of product: IR (mineral oil) 3092,

1663, 1656, 1613, 1515, 1501, 1323, 1301, 1223, 750 cm^-1; ¹H

NMR (300 MHz, CDCl₃) δ 8.47 (s, ArH), 7.95-8.1 (m, ArH, 2

H), 7.83 (d, J ) 8.4 Hz, ArH), 7.57 (d, J ) 8.4 Hz, ArH), 7.35-

7.5 (m, ArH, 2 H), 7.2-7.35 (m, ArH), 7.0-7.15 (m, ArH), 5.2-

5.5 (m, CH(CH₃)₂), 1.67 (d, J ) 7.1 Hz, CH(CH₃)₂); MS (EI)

m/ e 321, 279. Anal. (C₁₉H₁₆N₃OF‚(H₂O)_1/8) C, H, N.

5-ter t-Bu tyl-3-(4-flu or op h en yl)im id a zo[1,5-a ]qu in oxa -

lin -4(5H)-on e (8a , R¹) tBu ). Following procedure B, but

using 5.00 g (22.9 mmol) of 5f and purification by flash

chromatography (1:2 f 1:1 ethyl acetate:hexane), gave 4.53 g

(59%) of the desired product as a light yellow powder (mp 135-

137 °C): IR (mineral oil) 1668, 1498, 1299, 751 cm^-1; ¹H NMR

(300 MHz, CDCl₃) δ 8.31 (s, ArH), 8.22 (dd, J ) 8.9, 5.6 Hz,

ArH, 2 H), 7.63 (d, J ) 7.9 Hz, ArH), 7.55 (d, J ) 7.1 Hz, ArH),

7.05-7.35 (m, ArH, 4 H), 1.76 (s, tBu); MS (EI) m/ e 335, 279.

5-ter t-Bu tyl-3-(4-flu or oph en yl)-4,5-dih ydr oim idazo[1,5-

Exa m p le P r oced u r e for th e P r ep a r a tion of Isocya -

n id es 7 Dir ectly fr om 9. 4-Meth ylben zyl Isocya n id e (7c).

Aqueous 50% NaOH (12 mL) was added to a mixture of

4-methylbenzylamine (9c; 5.00 g, 41.3 mmol), benzyltriethy-

lammonium chloride (114 mg, 0.502 mmol), chloroform (4.93

g, 41.3 mmol), and CH₂Cl₂(12 mL). The mixture was heated

at 50 °C for 3.5 h. After cooling to room temperature, aqueous

workup (CH₂Cl₂, brine, Na₂SO₄) and purification by flash

chromatography (30% ethyl acetate/hexane) provided 2.63 g

(49%) of 7c as a yellow oil: ¹H NMR (300 MHz, CDCl₃) δ 7.1-

7.3 (m, ArH, 4 H), 4.59 (s, ArCH₂), 2.37 (s, ArCH₃).

a ]qu in oxa lin e (18a , R¹

) tBu ). A solution of 8a (R¹) tBu;

1.32 g, 3.94 mmol), THF (50.0 mL), and borane-methyl sulfide

complex (2.6 mL, 26 mmol, 10.0 M) was heated at reflux for

24 h. After cooling to room temperature, excess hydride was

quenched with 10% HCl. Basic workup (ethyl acetate, NaH-

CO₃, MgSO₄) and purification by flash chromatography (1.5:1

hexane:ethyl acetate) gave 730 mg (58%) of the quinoxaline

as a white solid (mp 129-130 °C): IR (mineral oil) 1504, 1215,

Also prepared by this procedure were the following com-

pounds. 2-Methoxybenzyl isocyanide (7j; 43%): ¹H NMR (300

MHz, CDCl₃) 7.35 (d, J ) 7.4 Hz, ArH), 7.2-7.35 (m, ArH),

6.94 (dd, J ) 8.2, 7.6 Hz, ArH), 6.82 (d, J ) 8.3 Hz, ArH), 4.56

(s, ArCH₂), 3.78 (s, OCH₃). 4-Methoxybenzyl isocyanide (7b;

46%): ¹H NMR (300 MHz, CDCl₃) δ 7.2-7.3 (m, ArH, 2 H),

6.85-6.95 (m, ArH, 2 H), 4.56 (narrow m, ArCH₂), 3.81 (s,

OCH₃).

1203, 842, 751 cm^-^{1 1}H NMR (300 MHz, CDCl₃) δ 8.06 (s,

;

ArH), 7.6-7.75 (m, ArH, 2 H), 7.45-7.55 (m, ArH), 7.1-7.4

(m, ArH, 5 H), 4.46 (s, NCH₂), 1.00 (s, tBuN); MS (EI) m/ e

321, 306, 264.

3-(4-F lu or op h en yl)-4,5-d ih yd r oim id a zo[1,5-a ]qu in oxa -

lin e (19a ). A solution of amine 18a (R¹) tBu; 728 mg, 2.27

mmol), CH₂Cl₂(20.0 mL), and trifluoroacetic acid (20.0 mL)

was stirred at room temperature for 72 h. Concentration and

basic workup (CH₂Cl₂, NaHCO₃, MgSO₄) gave 574 mg (95%)

of the desired product as a white foam, homogeneous by TLC

analysis: IR (mineral oil) 3381, 1511, 1492, 1290, 1231, 838,

Exa m p le P r oced u r e for Cycliza tion of 5 w ith Isocya -

n id e 7. Meth od A. 5-(1-Meth yleth yl)-3-p h en ylim id a zo-

[1,5-a ]qu in oxa lin -4(5H)-on e (23). Quinoxaline 5a (1.00 g,

4.90 mmol) was added to a mixture of hexane-washed sodium

hydride (254 mg, 6.35 mmol, 60% mineral oil dispersion) and

DMF (10.8 mL). The solution was stirred for 30 min at room

temperature and cooled to -10 °C. Diethyl chlorophosphate

(0.92 mL, 6.4 mmol) was added, and the solution was allowed

to warm to room temperature. After 1 h the solution was

cooled to 0 °C. A mixture of potassium tert-butoxide (711 mg,

3-Phenylimidazo[1,5-a]quinoxalines

741 cm^{-1 1}H NMR (300 MHz, CDCl₃) δ 8.04 (s, ArH), 7.58

(dd, J ) 8.8, 5.4 Hz, ArH, 2 H), 7.42 (d, J ) 6.8 Hz, ArH),

7.0-7.2 (m, ArH, 3 H), 6.8-6.95 (m, ArH, 2 H), 4.66 (d, J )

1.7 Hz, NHCH₂), 4.07 (br s, NH); MS (EI) m/ e 265, 264, 237,

144, 118.

J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19 3833

;

washed with MeOH and CH₂Cl₂, and the filtrate was concen-

trated to provide 14.3 g (96%) of 13g as a dark solid which

was carried on crude: ¹H NMR (300 MHz, CDCl₃) δ 6.8-7.0

(m, ArH, 3 H), 6.65 (narrow m, ArH, 3 H), 4.21 (s, NCH₂), 3.89

(s, OCH₃, 6 H).

Also prepared by this procedure was 4,5-dihydro-3-(4-

methoxyphenyl)imidazo[1,5-a]quinoxaline (19b) (mp 177-180

°C): IR (mineral oil) 3379, 1511, 1497, 1297, 1249, 1240, 739

cm^-1; ¹H NMR (300 MHz, CDCl₃) δ 8.04 (s, ArH), 7.5-7.6 (m,

ArH, 2 H), 7.42 (d, J ) 8.0 Hz, ArH), 7.07 (apparent t, J ) 7.7

Hz, ArH), 6.9-7.05 (m, ArH, 2 H), 6.75-6.9 (m, ArH, 2 H),

4.66 (d, J ) 1.6 Hz, NCH₂), 4.08 (s, NH), 3.85 (s, OCH₃); MS

(EI) m/ e 277, 276, 144.

5-ter t-Bu t yl-3-p -t olylim id a zo[1,5-a ]q u in oxa lin -4(5H )-

on e (8c, R¹) tBu ). Following procedure B using 5.00 g (22.9

mmol) of 5f and purification by flash chromatography (20%

ethyl acetate/hexane) provided 3.32 g (44%) of the product as

a light yellow solid (mp 160-165 °C): IR (mineral oil) 1668,

1502, 1295, 1186, 748 cm^-1; ¹H NMR (300 MHz, CDCl₃) δ 8.31

(s, ArH), 8.10 (d, J ) 8.2 Hz, ArH, 2 H), 7.62 (d, J ) 7.9 Hz,

ArH), 7.53 (d, J ) 8.3 Hz, ArH), 7.15-7.35 (m, ArH, 4 H), 2.40

(s, ArCH₃), 1.75 (s, C(CH₃)₃); MS (EI) m/ e 331, 275.

7-Ch lor o-1,2,3,4-t et r a h yd r o-1-(3,4-d im et h oxyb en zyl)-

2,3-d ioxoqu in oxa lin e (5g). Ethyl oxalyl chloride (1.70 mL,

15.2 mmol) was added to a solution of 13g (4.21 g, 14.4 mmol),

toluene (105 mL), THF (100 mL), and diisopropylethylamine

(3.26 mL, 18.7 mmol) at -78 °C. The mixture was stirred at

-78 °C for 1 h and allowed to warm to room temperature

overnight. The reaction mixture was heated at reflux for 24

h and allowed to cool to room temperature. The mixture was

filtered; the solids were washed extensively with water and

ether and dried to provide 4.40 g (88%) of 5g as a gray powder

(mp 277-280 °C): IR (mineral oil) 1699, 1682, 1522, 1470,

1

1462, 1444, 1265, 1250, 1140, 1020, 845 cm^-1; H NMR (300

MHz, DMSO-d₆) δ 12.17 (s, NH), 7.27 (s, ArH, 2 H), 7.1-7.25

(m, ArH, 2 H), 7.02 (s, ArH), 6.87 (d, J ) 8.4 Hz, ArH), 6.79

(d, J ) 7.7 Hz, ArH), 5.28 (s, NCH₂), 3.72 (s, OCH₃), 3.70 (s,

OCH₃); MS (EI) m/ e 346, 151.

7-Ch lor o-3-(4-flu or op h en yl)-5-(3,4-d im eth oxyben zyl)-

im id a zo[1,5-a ]qu in oxa lin -4(5H)-on e (8e). Following pro-

cedure B, but starting with 5g (9.35 g, 27.0 mmol), gave 7.86

g (63%) of the quinoxaline as a gray powder (mp 237-239

3-p-Tolylim idazo[1,5-a ]qu in oxalin -4(5H)-on e (20c). Tri-

fluoroacetic acid (59 mL) was added to a solution of amide 8c

(R¹) tBu; 2.95 g, 8.90 mmol) in CH₂Cl₂(59 mL) at 0 °C. The

mixture was allowed to warm to room temperature over 2 h

and concentrated. The solids were filtered, washed with water,

and dried to give 2.30 g (94%) of the deprotected amide as a

white solid (mp >300 °C): IR (mineral oil) 1686, 1654, 1494,

1376, 823 cm^-1; ¹H NMR (300 MHz, DMSO-d₆) δ 11.39 (s, NH),

9.16 (s, ArH), 8.20 (apparent d, J ) 8.1 Hz, ArH, 3 H), 7.2-

7.45 (m, ArH, 5 H), 2.36 (s, CH₃); MS (EI) m/ e 275, 247, 219.

4,5-Dih yd r o-3-p-tolylim id a zo[1,5-a ]qu in oxa lin e (19c).

Aluminum hydride³⁷(50 mL, 32 mmol, 0.63 M in THF) was

added to 20c (1.60 g, 5.81 mmol) and the mixture stirred at

room temperature for 2 days and heated at reflux for 16 h.

After cooling to room temperature, excess aluminum hydride

was decomposed by adding methanol (20 mL) and 6 N NaOH

(50 mL). Aqueous workup (CH₂Cl₂, MgSO₄) and trituration

in 75% ether/hexane gave 1.15 g (75%) of a beige powder

containing product and starting material³⁸(10:1) which was

carried on crude. Spectral data for a pure sample³⁹of 19c (mp

189-194 °C): IR (mineral oil) 1522, 1507, 1288, 959, 742, 735

°C): IR (mineral oil) 1656, 1521, 1501, 1287, 1239, 1135 cm^-1

;

¹H NMR (300 MHz, CDCl₃) δ 8.53 (s, ArH), 8.27 (dd, J ) 9.2,

5.4 Hz, ArH, 2 H), 7.75 (d, J ) 8.7 Hz, ArH), 7.31 (narrow m,

ArH), 7.1-7.3 (m, ArH, 3 H), 6.8-6.9 (m, 3 H), 5.38 (s, NCH₂),

3.86 (s, OCH₃), 3.85 (s, OCH₃); MS (EI) m/ e 463, 151, 107.

7-Ch lor o-3-(4-flu or op h en yl)-4,5-d ih yd r oim id a zo[1,5-a ]-

qu in oxa lin e (19e). A mixture of 8e (2.00 g, 4.31 mmol),

lithium aluminum hydride (720 mg, 19.0 mmol), and THF (75

mL) was stirred at room temperature for 4 days. Water (0.7

mL), 10% KOH (1.0 mL), and water (2.0 mL) were added

successively. The mixture was stirred for 1 h and filtered. The

solids were washed with EtOAc and CH₂Cl₂several times. The

combined filtrates were dried (MgSO₄), filtered, and concen-

trated to give 1.84 g of a mixture of crude 18e and starting

material.

The residue was combined with CH₂Cl₂(30 mL) and anisole

(1.1 mL, 10 mmol) and the solution cooled to 0 °C. Trifluoro-

acetic acid (30 mL) was added, and the solution was stirred

for 1 h at 0 °C and for 16 h at room temperature. Concentra-

tion and basic workup (CH₂Cl₂, NaHCO₃, MgSO₄) gave 19e,

contaminated by trace amounts of the corresponding imine.

Sodium borohydride (1.25 g, 33.0 mmol) was added to a

solution of the crude material and ethanol (100 mL). The

solution was stirred at room temperature for 3 h and concen-

trated. Aqueous workup (CH₂Cl₂, MgSO₄) and purification by

flash chromatography (1:1 hexane:ethyl acetate; 1% CH₂Cl₂)

gave 623 mg (48% overall) of the product as a white solid (mp

186-189 °C): IR (mineral oil) 1609, 1516, 1506, 1290, 1263,

1

cm^-1; H NMR (300 MHz, DMSO-d₆) δ 8.43 (s, ArH), 7.66 (d,

J ) 8.4 Hz, ArH), 7.50 (d, J ) 8.1 Hz, ArH, 2 H), 7.23 (d, J )

8.0 Hz, ArH, 2 H), 6.95-7.1 (m, ArH), 6.87 (d, J ) 7.7 Hz,

ArH), 6.74 (apparent t, J ) 7.9 Hz, ArH), 6.31 (s, NH), 4.57

(s, NCH₂), 2.33 (s, ArCH₃); MS (EI) m/ e 261, 260, 231, 219,

144, 116.

Also prepared by this procedure was 4,5-dihydro-3-phen-

ylimidazo[1,5-a]quinoxaline (19d ) (mp 187-189 °C): IR (min-

1

eral oil) 3382, 1616, 1523, 766, 736, 692 cm^-1; H NMR (300

MHz, CDCl₃) δ 8.12 (s, ArH), 7.62 (dd, J ) 8.5, 1.3 Hz, ArH,

2 H), 7.35-7.5 (m, ArH, 3 H), 7.25-7.35 (m, ArH), 7.0-7.15

(m, ArH), 6.75-6.95 (m, ArH, 2 H), 4.68 (s, NCH₂); MS (EI)

m/ e 247, 246, 219, 144, 109.

1

1237, 1158, 840, 832 cm^-1; H NMR (300 MHz, CDCl₃) 8.02

(s, ArH), 7.5-7.7 (m, ArH, 2 H), 7.35 (d, J ) 7.7 Hz, ArH),

7.12 (apparent t, J ) 9.2 Hz, ArH, 2 H), 6.8-7.0 (m, ArH, 2

H), 4.67 (s, NCH₂), 4.17 (s, NH); MS (EI) m/ e 299, 298, 178,

151.

3-Ch lor o-N-(3,4-d im eth oxyben zyl)-6-n itr oa n ilin e (12g).

A mixture of veratrylamine (12.2 g, 73.0 mmol), 2,4-dichlo-

ronitrobenzene (11, R⁶) H, R⁷) Y ) Cl; 12.9 g, 67.2 mmol),

and diisopropylethylamine (27.9 mL, 160 mmol) was stirred

at room temperature for 6 days and at 75 °C for 7 days.

Additional veratrylamine (10.0 g, 59.8 mmol) and diisopropyl-

ethylamine (10.0 mL, 57.3 mmol) were added at 8 and 2 days,

respectively. After cooling to room temperature, aqueous

workup (ethyl acetate, 10% HCl, NaHCO₃, brine washes,

MgSO₄), and purification by flash chromatography (5:1 hexane:

ethyl acetate) gave 16.4 g (76%) of 12g as an orange solid (mp

106-107 °C): IR (mineral oil) 1565, 1524, 1491, 1309, 1259,

1211 cm^-1; ¹H NMR (300 MHz, CDCl₃) δ 8.3-8.4 (m, NH), 8.15

(d, J ) 9.1 Hz, ArH), 6.8-6.95 (m, ArH, 4 H), 6.64 (dd, J )

9.2, 2.2 Hz, ArH), 4.44 (d, J ) 5.4 Hz, NHCH₂), 3.90 (s, OCH₃),

3.89 (s, OCH₃); MS (EI) m/ e 322, 304, 165, 151.

Also prepared by this procedure were the following com-

pounds. 6-Chloro-3-(4-fluorophenyl)-4,5-dihydroimidazo[1,5-

a]quinoxaline (19f) (mp 146-149 °C): IR (mineral oil) 1505,

1222, 837, 757 cm^-^{1 1}H NMR (300 MHz, CDCl₃) δ 8.04 (s,

;

ArH), 7.58 (dd, J ) 8.8, 5.4 Hz, ArH, 2 H), 7.35 (d, J ) 8.1 Hz,

ArH), 7.05-7.25 (m, ArH, 3 H), 6.80 (apparent t, J ) 8.1 Hz,

ArH), 4.74 (d, J ) 1.7 Hz, HNCH₂), 4.68 (s, NH); MS (EI) m/ e

299, 298, 178, 118. 7-Fluoro-3-(4-fluorophenyl)-4,5-dihydroim-

idazo[1,5-a]quinoxaline (19g) (mp 175 °C): IR (mineral oil)

1526, 1507, 1272, 1158, 836 cm^-1; ¹H NMR (300 MHz, DMSO-

d₆) δ 9.06 (s, HCdN), 7.7-7.85 (m, ArH), 7.6-7.75 (m, ArH, 2

H), 7.34 (apparent t, J ) 12.0 Hz, ArH, 2 H), 6.55-6.8 (m,

ArH, 2 H), 4.64 (s, NCH₂); MS (EI) m/ e 283, 282, 255, 162,

127.

4-Ch lor o-2-[(3,4-d im eth oxyben zyl)a m in o]a n ilin e (13g).

A mixture of 12g (16.4 g, 50.8 mmol), ethanol (900 mL), and

platinum on sulfide carbon (1.5 g) was hydrogenated at 30 psi

for 16 h. The mixture was filtered, the residual solids were

E xa m p le P r oced u r e for Ca r b a m oyl Ch lor id es 22.

6-Ch lor o-5-(ch lor oca r bon yl)-3-(4-flu or op h en yl)-4,5-d ih y-

d r oim id a zo[1,5-a ]qu in oxa lin e (22f). Phosgene (9.3 mL, 18

mmol, 1.93 M in toluene) was added dropwise to a solution of

3834 J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19

J acobsen et al.

19f (1.80 g, 6.01 mmol), diisopropylethylamine (3.14 mL, 18.0

mmol), THF (17 mL), and CH₂CH₂(17 mL) at room temper-

ature. The mixture was stirred at room temperature over-

night. Aqueous workup (ethyl acetate, MgSO₄) gave 2.16 g

(99%) of 22f as a yellow oil which was carried on crude.

(m, ArH, 12 H), 5.0-5.5 (m, NCH₂); MS (EI) m/ e 387, 123.

Anal. (C₂₃H₁₅N₃OF₂) C, H, N.

Gen er a l Red u ctive Am in a tion P r oced u r e for 3,4,5-

Tr im eth ylp ip er a zin es. Meth od G. 6-Ch lor o-3-(4-flu o-

r op h en yl)-4,5-d ih yd r o-5-[[1-(3,4,5-tr im eth ylp ip er a zin o]-

ca r bon yl]im id a zo[1,5-a ]qu in oxa lin e (70). Sodium cyano-

borohydride (236 mg, 3.76 mmol) was added to a solution of

69 (550 mg, 1.25 mmol) and formaldehyde (37%, 1.9 mL, 25

mmol) in methanol (8 mL) at room temperature. The mixture

was stirred overnight at room temperature. Aqueous workup

(ethyl acetate, MgSO₄) and recrystallization from ethyl acetate/

hexane provided 460 mg (81%) of the desired compound as a

yellow solid (mp 213-214 °C): IR (mineral oil) 1657, 1500,

1448, 1416, 1232, 1222, 836 cm^-1; ¹H NMR (300 MHz, CDCl₃)

δ 8.08 (s, ArH), 7.58 (dd, J ) 8.8, 5.4 Hz, ArH, 2 H), 7.50 (dd,

J ) 7.9, 1.4 Hz, ArH), 7.35 (dd, J ) 8.2, 1.3 Hz, ArH), 7.2-7.3

(m, ArH, 1 H), 7.16 (apparent t, J ) 8.7 Hz, ArH, 2 H), 4.85

(s, ArNCH₂), 3.66 (d, J ) 15.5 Hz, 2 H), 2.77 (apparent t, J )

11.5 Hz, 2 H), 2.26 (s, NCH₃), 2.15-2.35 (m, 2 H), 1.07 (d, J )

6.1 Hz, CHCH₃, 6 H); MS (EI) m/ e 453, 418, 382, 361, 297,

278, 155, 141, 113, 98, 84, 72. Anal. (C₂₄H₂₅N₅OClF) C, H,

N, Cl.

GABA_ARecep tor Exp r ession a n d Mem br a n e P r ep a r a -

tion . DNA manipulations and general baculovirus methods

(Sf-9 cell cultivation, infection, and isolation; purification of

recombinant viruses) were performed as described elsewhere.²⁴

The Sf-9 cells were infected at a multiplicity of infection of 3

PFU of viruses: AcNPV-R₁, -R₃or -R₆, AcNPV-â₂, and AcNPV-

γ₂. Infected cells were used for electrophysiological measure-

ments at 48 h postinfection or for membrane preparations at

60 h postinfection. The stable cell lines expressing R₁, R₃, or

R₆, â₂, and γ₂subunits of GABA_Awere derived by transfection

of plasmids containing cDNA and a plasmid encoding G418

resistance into human kidney cells (A293 cells) as described

elsewhere.²⁵After 2 weeks of selection in 1 mg/mL G418, cells

positive for all three GABA_Areceptor mRNAs by Northern

blotting were used for electrophysiology to measure GABA-

induced Cl^-currents.

For equilibrium binding measurements, Sf-9 cells infected

with baculovirus-carrying cDNAs for R₁, R₃, or R₆, â₂, and γ₂

subunits were harvested in 2 L batches at 60 h postinfection.

The membranes were prepared following the procedure de-

scribed previously.²⁴Briefly, the membranes were prepared

in normal saline after homogenization with a Polytron PT 3000

homogenizer (Brinkman) for 4 min. Unbroken cells and large

nuclei aggregates were removed by centrifugation at 1000g

for 10 min. Then the membranes were recovered with the

second centrifugation of the supernatant at 40000g for 50 min.

The membranes were resuspended to a final concentration of

5 mg/mL in a solution containing 300 mM sucrose, 5 mM Tris/

HCl, pH 7.5, and glycerol to a final concentration of 20% and

stored at -80 °C. Equilibrium binding of [³H]flunitrazepam

or [³H]Ro-4513 to the cloned GABA_Areceptors was measured

in a 500 mL volume of normal saline containing 6 nM [³H]-

flunitrazepam or [³H]Ro-4513, varying concentrations of test

ligands, and 50 mg of membrane protein. The mixture was

incubated for 60 min at 4 °C, filtered over a Whatman glass

fiber filter, and washed four times with cold normal saline.

The filter was then counted for radioactivity in the presence

of a scintillation cocktail (Insta Gel).

GABA_ARecep tor Bin d in g. The in vitro binding affinity

of the imidazo[1,5-a]quinoxalines for GABA_Awas determined

as previously described with minor modification.¹⁵Freshly

prepared rat cerebellar membranes were suspended in 300 mM

sucrose and 10 nM Hepes/Tris, pH 7.4. Typically, the reaction

medium contained 6 mM [³H]flunitrazepam, 50 mg of mem-

brane protein, test drugs at various concentrations or vehicle

in 200 mL, 118 mM NaCl, 10 mM Hepes/Tris, pH 7.4, and 1

mM MgCl₂. The mixtures were incubated for 60 min at 4 °C.

The amount of binding was determined with rapid filtration

techniques using Whatman GF/B filters.

Exa m p le P r oced u r e for th e Rea ction of Ca r ba m oyl

Ch lor id es 22 w ith Am in es. Meth od C. 6-Ch lor o-5-[[1-

(3,5-d im eth ylp ip er a zin o)]ca r bon yl]-3-(4-flu or op h en yl)-

4,5-d ih yd r oim id a zo[1,5-a ]qu in oxa lin e (69). To a solution

of 22f (1.44 g, 3.98 mmol) and diisopropylethylamine (0.92 mL,

5.3 mmoL) in THF (20 mL) at 0 °C was added cis-2,6-

dimethylpiperazine (543 mg, 4.76 mmol). The mixture was

allowed to warm slowly and stirred at room temperature for

4 days. Aqueous workup (CH₂Cl₂, MgSO₄) and purification

by flash chromatography (4% methanol/CH₂Cl₂) provided 1.12

g (64%) of 69 as a yellow solid. An analytical sample was

prepared by recrystallization from ethyl acetate/hexane to give

a light yellow solid (mp 203-205 °C): IR (mineral oil) 1659,

1507, 1494, 1410, 1258, 1225, 831, 780 cm^-^{1 1}H NMR (300

;

MHz, CDCl₃) δ 8.08 (s, ArH), 7.55-7.65 (m, ArH, 2 H), 7.50

(d, J ) 7.9 Hz, ArH), 7.35 (d, J ) 7.0 Hz, ArH), 7.2-7.3 (m,

ArH), 7.16 (apparent t, J ) 8.7 Hz, ArH, 2 H), 4.85 (s,

ArNCH₂), 3.73 (d, J ) 13.8 Hz, 2 H), 2.8-3.0 (m, 2 H), 2.4-

2.6 (m, 2 H), 1.04 (d, J ) 7.6 Hz, CHCH₃, 6 H); MS (EI) m/ e

439, 404, 382, 361, 321, 298, 141. Anal. (C₂₃H₂₃N₅OClF) C,

H, N, Cl.

Exa m p le P r oced u r e for th e Rea ction of Ca r ba m oyl

Ch lor id es 22 w ith Alcoh ols. Meth od D. Meth yl 3-(4-

F lu or op h en yl)-4,5-d ih yd r oim id a zo[1,5-a ]qu in oxa lin e-5-

ca r boxyla te (59). Sodium methoxide (180 mg of sodium in

6.0 mL of MeOH, 7.8 mmol) was added to a mixture of

carbamoyl chloride 22a (2.30 mmol) and MeOH (10.0 mL) at

0 °C. The mixture was stirred for 1 h at 0 °C and for 16 h at

room temperature. Acetic acid (0.50 mL) was added and the

mixture concentrated. Aqueous workup (CH₂Cl₂, MgSO₄) and

purification by flash chromatography (1:1 hexane:ethyl ac-

etate) gave 447 mg of an oil which was crystallized from

hexane/ether to provide 258 mg (35%) of 59 as a yellow powder

(mp 156-160 °C): IR (mineral oil) 1709, 1505, 1435, 1368,

1

1223 cm^-1; H NMR (300 MHz, CDCl₃) δ 8.07 (s, ArH), 7.6-

7.8 (m, ArH), 7.65 (dd, J ) 8.8, 5.4 Hz, ArH, 2 H), 7.5-7.6 (m,

ArH), 7.25-7.35 (m, ArH, 2 H), 7.16 (apparent t, J ) 8.8 Hz,

ArH, 2 H), 5.11 (s, NCH₂), 3.81 (s, OCH₃); MS (EI) m/ e 323,

308, 264, 236. Anal. (C₁₈H₁₄N₃O₂F) C, H, N.

Syn t h esis of Ur ea s 16 Dir ect ly fr om Am in es 19.

Met h od E . 4,5-Dih yd r o-5-(p yr r olid in oca r b on yl)-3-p -

tolylim id a zo[1,5-a ]qu in oxa lin e (46). A mixture of 19c (450

mg, 1.72 mmol), diisopropylethylamine (0.35 mL, 2.0 mmol),

and CH₂Cl₂(25 mL) was cooled to 0 °C. Triphosgene (276 mg,

0.930 mmol) was added and the solution allowed to warm to

room temperature. After stirring at room temperature for 1

h, the solution was cooled to 0 °C and pyrrolidine (0.55 mL,

6.6 mmol) was added. The solution was allowed to warm and

stirred at room temperature for 2 h. Aqueous workup (CH₂Cl₂,

MgSO₄) and purification by flash chromatography (20 f 50%

ethyl acetate/hexane) provided, after drying in vacuo, 498 mg

(81%) of 46 as a white solid (mp 238.5-239.5 °C): IR (mineral

oil) 1636, 1506, 1413, 752 cm^-1; ¹H NMR (300 MHz, CDCl₃) δ

8.07 (s, ArH), 7.45-7.65 (m, ArH, 3 H), 7.1-7.3 (m, ArH, 5

H), 4.91 (s, ArNCH₂), 3.25 (br s, CH₂NCH₂), 2.38 (s, ArCH₃),

1.7-1.85 (m, NCH₂CH₂CH₂); MS (EI) m/ e 358, 260, 98. Anal.

(C₂₂H₂₂N₄O) C, H, N.

Gen er a l P r oced u r e for th e Acyla tion of 19 To P r ovid e

21. Meth od F . 3-(4-F lu or op h en yl)-5-(2-flu or oben zoyl)-

4,5-d ih yd r oim id a zo[1,5-a ]qu in oxa lin e (63). 2-Fluoroben-

zoyl chloride (0.29 mL, 2.4 mmol) was added to a solution of

amine 19a (600 mg, 2.26 mmol), diisopropylethylamine (0.47

mL, 2.7 mmol), and THF (20 mL) at 0 °C. The solution was

stirred at 0 °C for 1.5 h and allowed to warm to room

temperature. After stirring for 16 h, aqueous workup (ethyl

acetate, MgSO₄), purification by flash chromatography (1:1

hexane:ethyl acetate), and crystallization from ethyl acetate/

hexane gave 680 mg (78%) of 63 as light yellow crystals (mp

185-189 °C): IR (mineral oil) 1656, 1507, 1284, 1219, 751

[

³⁵S]-ter t-Bu tyl Bicyclop h osp h or oth ion a te ([³⁵S]TBP S)

Bin d in g. Binding of [³⁵S]TBPS in the rat brain membranes

was measured in the medium containing 2 nM [³⁵S]TBPS,

unless specified otherwise, 50 mg of membrane proteins, 1 M

NaCl, and 10 mM Tris/HCl, pH 7.4, in a total volume of 500

cm^-^{1 1}H NMR (300 MHz, CDCl₃) δ 8.15 (s, ArH), 6.6-7.75

;

3-Phenylimidazo[1,5-a]quinoxalines

J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19 3835

mL. Drugs were added in concentrated methanolic solutions,

and the level of methanol did not exceed 0.2% and was

maintained constant in all tubes. The mixtures were incu-

bated for 120 min at 24 °C. The reaction mixtures were

filtered over a Whatman GF/B filter under vacuum. The filters

were washed three times with 4 mL of the reaction buffer

without radioisotope and counted for radioactivity. Nonspecific

binding was estimated in the presence of 2 mM unlabeled

TBPS and subtracted to compute specific binding.¹⁶

Electr op h ysiology. The whole cell configuration of the

patch clamp technique was used to record the GABA-mediated

Cl^-currents in the A293 cells expressing the R₁â₂γ₂subtype,

as described earlier.¹⁸Briefly, patch pipets made of borosilicate

glass tubes were fire-polished and showed a tip resistance of

0.5 -2 MΩ when filled with a solution containing (mM) 140

CsCl, 11 EGTA, 4 MgCl₂, 2 ATP, and 10 Hepes, pH 7.3. The

cell-bathing external solution contained (mM) 135 NaCl, 5 KCl,

1 MgCl₂, 1.8 CaCl₂, and 5 Hepes, pH 7.2 (normal saline).

GABA at the concentration of 5 mM in the external solution

with or without indicated drugs was applied through a U-tube

placed within 100 mm of the cells for 10 s, unless indicated

otherwise. The current was recorded with an Axopatch 1D

(muscle relaxation) as indicated above. If the expected loss

of traction in response to triazolam was not observed, the test

compound was assumed to be an antagonist. Lower doses (at

0.5 log intervals) of the test compound were then evaluated to

establish the antagonist dose₅₀

(ED₅₀).

Refer en ces

(1) (a) Squires, R., Ed. GABA and Benzodiazepine Receptors; CRC

Press: Boca Raton, FL, 1988; Vol. I, II. (b) Haefely, W. In

Benzodiazepines; Beaumont, G., Brandon, S., Leonard, B. E.,

Eds.; J ohn Wiley and Sons: New York, 1990; pp 1-18. (c)

Haefely, W. Partial Agonists of the Benzodiazepine Receptor:

From Animal Data to Results in Patients. In Chloride Channels

and Their Modulation by Neurotransmitters and Drugs; Biggio,

G., Costa, E., Eds.; Raven Press: New York, 1988; pp 275-292.

(2) (a) Gardner, C. R. Functional In Vivo Correlates of the Benzo-

diazepine Agonist-Inverse Agonist Continuum. Prog. Neurobiol.

1988, 31, 425-476. (b) Braestrup, C.; Nielsen, M. In Receptor

Biochemistry and Methodology; Olsen, R. W., Venter, J . C., Eds.;

A. R. Liss, Inc.: New York, 1986; Vol. 5, p 167. (c) Bare, T. M.;

Resch, J . F.; Patel, J . B. Anxiolytics and Sedative-Hypnotics. In

Annual Reports in Medicinal Chemistry; Bailey, D. M., Ed.;

Academic Press: New York, 1987; Vol. 22, pp 11-20. (d) Haefely,

W.; Kyburz, E.; Gerecke, M.; Mohler, H. Recent Advances in the

Molecular Pharmacology of Benzodiazepine Receptors and in the

Structure-Activity Relationships of Their Agonists and Antago-

nists. In Advances in Drug Research; Testa, B., Ed.; Academic

Press: New York, 1985; Vol. 14, pp 165-322. (e) Braestrup, C.;

Honore, T.; Nielsen, M.; Petersen, E. N.; J ensen, L. H. Ligands

for Benzodiazepine Receptors With Positive and Negative Ef-

ficacy. Biochem. Pharmacol. 1984, 33, 859-862.

(3) (a) Petersen, E. N.; J ensen, L. H.; Diejer, J .; Honore, T. New

Perspectives in Benzodiazepine Receptor Pharmacology. Phar-

macopsychiatry 1986, 19, 4-6. (b) Haefely, W. Psychopharma-

cology of Anxiety. Eur. Neuropsychopharmacol. 1991, 1, 89-95.

(4) (a) Pritchett, D. B.; Sontheimer, H.; Shivers, B. D.; Ymer, S.;

Kettenmann, H.; Schonfield, P. R.; Seeburg, P. H. Importance

of a Novel GABA_AReceptor Subunit for Benzodiazepine Phar-

macology. Nature 1989, 338, 582-585. (b) Pritchett, D. B.;

Seeburg, P. H. γ-Aminobutyric Acid Type A Receptor Point

Mutation Increases the Affinity of Compounds for the Benzodi-

azepine Site. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 1421-1425.

(5) (a) Skolnick, P.; Wong, G. In Imidazopyridines in Anxiety

Disorders: A Novel Experimental and Therapeutic Approach;

Zivkovic, B., Langer, S., Bartholini, G., Eds.; Raven Press: New

York, 1992. (b) For a recent review, see: Yeh, H. H.; Grigorenko,

E. V. Deciphering the Native GABA_AReceptor: Is There Hope?

J . Neurosci. Res. 1995, 41, 567-571.

amplifier and a CV-4 headstage (Axon Instrument Co.).

A

Bh-1 bath headstage was used to compensate for changes in

bath potentials. The currents were recorded with a Gould

Recorder 220. GABA currents were measured at the holding

potential of -60 mV at room temperature (21-24 °C).

Metr a zole An ta gon ism . Compounds were tested for their

ability to antagonize metrazole-induced convulsions in mice

after ip injection as described elsewhere.¹⁹Briefly, male CF-1

mice were injected with metrazole (85 mg/kg, sc), and convul-

sive seizure was elicited 15 min later with an auditory

stimulation (5 dB for 10 s). Drugs tested for metrazole

antagonism were injected ip 30 min before the metrazole

challenge, 4 mice/dose at a 0.3 log dose interval. ED₅₀’s for

protection against tonic seizure were calculated by the method

of Spearman-Karber (Finney, D. J . Statistical Methods in

Biological Assay).

Eth a n ol P oten tia tion . Male CF-1 Charles River mice

were injected orally with 7.5 mL/kg 50% aqueous ethanol and

simultaneously received the test compound by the subcutane-

ous route (10 mL/kg in 0.25% aqueous methyl cellulose).

Thirty minutes later, they were tested for the loss of traction

response (muscle relaxation) and loss of righting reflex (an-

esthesia). The former test consists of determining if the mouse

after suspension by the forepaws on an 18 g wire can place

one of its hindlimbs on the wire within 10 s. The latter test

determines whether the same animal will roll over onto its

feet within 10 s of release on its back. Six mice were tested

at each dose, and active compounds were retested at several

(6) (a) TenBrink, R. E.; Im, W. B.; Sethy, V. H.; Tang, A. H.; Carter,

D. B. Antagonist, Partial Agonist, and Full Agonist Imidazo-

[1,5-a]quinoxaline Amides and Carbamates Acting through the

GABA_A/Benzodiazepine Receptor. J . Med. Chem. 1994, 37, 758-

768. (b) J acobsen, E. J .; TenBrink, R. E.; Stelzer, L. S.; Belonga,

K. L.; Carter, D. B.; Im, H. K.; Im, W. B.; Sethy, V. H.; Tang, A.

H.; VonVoigtlander, P. F.; Petke, J . D. High Affinity Partial

Agonist Imidazo[1,5-a]quinoxaline Amides, Carbamates and

Ureas at the GABA_A/Benzodiazepine Receptor Complex. J . Med.

Chem. 1996, 39, 158-175.

doses to establish an ED₅₀

.

(7) (a) Tang, A. H.; Franklin, S. R.; Himes, C. S.; Ho, P. M.

Behavioral Effects of U-78875, a Quinoxalinone Anxiolytic with

Potent Benzodiazepine Antagonist Activity. J . Pharmacol. Exp.

Ther. 1991, 259, 248-254. (b) Petke, J . D.; Im, H. K.; Im, W. B.;

Blakeman, D. P.; Pregenzer, J . F.; J acobsen, E. J .; Hamilton, B.

J .; Carter, D. B. Characterization of Functional Interactions of

Imidazoquinoxaline Derivatives with Benzodiazepine-γ-Ami-

nobutyric Acid_AReceptors. Mol. Pharmacol. 1992, 42, 294-301.

(8) Ulrich, R. G.; Cramer, C. T.; Bacon, J . A.; Funk, G. M.; Hunt,

C. E.; Kholmukhamedov, E.; Petrella, D. K.; Sun, E. L. Interfer-

ence with Hepatocelluar Metabolism by a Quinoxalinone Anxi-

olytic and Its Carboxylic Acid Metabolite: What Factors Con-

tribute Towards Cell Death? 1994 Annual Society of Toxicology

Meeting, Dallas, TX, Mar. 14-17, 1994.

Acu te P h ysica l Dep en d en ce.²⁸Male CF-1 mice (10/

group) were dosed sc twice daily with the test compound, once

at 0800 and again at 1600 (with 2 times the 0800 dose) for 3

days. Twenty-four hours after the last dose, they received an

intravenous injection of the benzodiazepine antagonist fluma-

zenil (2.5 mg/kg in 5% aqueous N,N-dimethylacetamide). Five

minutes later, the electroshock seizure threshold was esti-

mated by the up-down method in which the stimulus current

was lowered or elevated by 0.05 log interval if the preceding

animal did or did not convulse, respectively. From the data

thus generated, a threshold (effective current of 50) was

calculated. The other parameters of the transpinnal (e.g.,

delivered across the ears via saline-soaked ear clip electrodes)

square wave stimulation were held constant (0.6 ms pulses at

100 Hz for 0.2 s). Test compounds, the precipitated with-

drawal from which significantly lowered (below 95% confidence

interval of parallel control group treated for 3 days with saline

and injected intravenously with flumazenil) the seizure thresh-

old, were considered to have caused physical dependence.

Ben zod ia zep in e An ta gon ism . Male CF-1 Charles River

mice were injected ip simultaneously with 1 mg/kg triazolam

and test compound starting at 100 mg/kg. A dose-response

was run on a 0.5 log scale. All drugs were made up in no. 122

sterile vehicle (0.25% aqueous methyl cellulose). Fifteen

minutes later, they were tested for loss of traction response

(9) Ulrich, R. G. Personal communication.

(10) Kyburz, E. Serendipity in Drug Discovery. The Case of BzR

Ligands. Farmaco 1989, 44, 345-382.

(11) Watjen, F.; Baker, R.; Engelstoff, M.; Herbert, R.; MacLeod, A.;

Knight, A.; Merchant, K.; Moseley, J .; Saunders, J .; Swain, C.

J .; Wong, E.; Springer, J . P. Novel Benzodiazepine Receptor

Partial Agonists: Oxadiazolylimidazobenzodiazepines. J . Med.

Chem. 1989, 32, 2282-2291.

(12) Weber, W. P.; Gokel, G. W. An Improved Procedure for the

Hofmann Carbylamine Synthesis of Isonitriles. Tetrahedron Lett.

1972, 17, 1637-1640.

(13) Watjen, F.; Engelstoft, M. U.S. Pat. 4,774,245, 1988. Watjen,

F.; Engelstoft, M. U.S. Pat. 4,795,749, 1989.

(14) Borch, R. F.; Bernstein, M. D.; Durst, H. D. The Cyanoborohy-

dride Anion as a Selective Reducing Agent. J . Am. Chem. Soc.

1971, 93, 2897-2904.

3836 J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 19

J acobsen et al.

(15) Sethy, V. H.; Harris, D. W. Determination of Biological Activity

of Alprazolam, Triazolam and Their Metabolites. J . Pharm.

Pharmacol. 1982, 34, 115-116.

(28) (a) VonVoigtlander, P. F.; Lewis, R. A. A Rapid Screening Method

for the Assessment of Benzodiazepine Receptor-Related Physical

Dependence in Mice. J . Pharmacol. Methods. 1991, 26, 1-5. (b)

Fialip, J .; Aumaitre, O.; Eschalier, A.; Maradeix, B.; Dordain,

G.; Lavarenne, J . Benzodiazepine Withdrawal Seizures: Analy-

sis of 48 Case Reports. Clin. Neuropharmacol. 1987, 10, 538-

544.

(29) Mohamadi, F.; Richards, N. G. J .; Guida, W. C.; Liskamp, R.;

Lipton, M.; Caufield, C.; Chang, G.; Hendrickson, T.; Still, W.

C. MacroModel-An Integrated Software System for Modeling

Organic and Bioorganic Molecules Using Molecular Mechanics.

J . Comput. Chem. 1990, 11, 440-467.

(30) Weiner, S. J .; Kollman, P. A.; Nguyen, D. T.; Case, D. A. An All

Atom Force Field for Simulations of Proteins and Nucleic Acids.

J . Comput. Chem. 1986, 7, 230-252.

(31) Allinger, N. L. Conformational Analysis. 130. MM2: A Hydro-

carbon Force Field Utilizing V₁and V₂Torsional Terms. J . Am.

Chem. Soc. 1977, 99, 8127-8134.

(32) Chang, G.; Guida, W. C.; Still, W. C. An Internal Coordinate

Monte Carlo Method for Searching Conformational Space. J . Am.

Chem. Soc. 1989, 111, 4379-4386.

(33) Tebib, S.; Bourguignon, J .-J .; Wermuth, C.-G. The Active Analog

Approach Applied to the Pharmacophore Identification of Ben-

zodiazepine Receptor Ligands. J . Comput.-Aided Mol. Des. 1987,

1, 153-170.

(34) Gardner, C. R. A Review of Recently-Developed Ligands for

Neuronal Benzodiazepine Receptors and Their Pharmacological

Activities. Prog. Neuro.-Psychopharmacol. Biol. Psychiatry 1992,

16, 755-781.

(35) Villar, H. O.; Uyeno, E. T.; Toll, L.; Polgar, W.; Davies, M. F.;

Loew, G. H. Molecular Determinants of Benzodiazepine Receptor

Affinities and Anticonvulsant Activities. Mol. Pharmacol. 1989,

36, 589-600.

(16) Im, W. B.; Blakeman, D. P. Correlation Between γ-Aminobutyric

Acid_AReceptor Ligand-Induced Changes in t-Butylbicyclo-

phosphoro[³⁵S]thionate Binding and ³⁶Cl^-Uptake in Rat Cere-

brocortical Membranes. Mol. Pharmacol. 1991, 39, 394-398.

(17) Schwartz, R. D.; Suzdak, P. D.; Paul, S. M. γ-Aminobutyric Acid

(GABA)- and Barbiturate-Mediated ³⁶Cl^-Uptake in Rat Brain

Synaptoneurosomes: Evidence for Rapid Desensitization of the

GABA Receptor-Coupled Chloride Ion Channel. Mol. Pharmacol.

1986, 30, 419-426.

(18) Im, H. K.; Im, W. B.; Hamilton, B. J .; Carter, D. B.; VonVoigt-

lander, P. F. Potentiation of GABA-Induced Chloride Currents

by Various Benzodiazepine Site Agonists with R₁γ₂, â₂γ₂and

R₁â₂γ₂Subtypes of Cloned GABA_AReceptors. Mol. Pharmacol.

1993, 44, 866-870.

(19) Rudzik, A. D.; Hester, J . B.; Tang, A. H.; Straw, R. N.; Friis, W.

Triazolobenzodiazepines,

A New Class of Central Nervous

System-Depressant Compounds. In The Benzodiazepines; Ga-

ratini, S., Mussini, E., Randall, L. O., Eds.; Raven Press: New

York, 1973; pp 285-297.

(20) These standards had ED₅₀’s in the range of 0.17-4.5 mg/kg.

(21) (a) Pritchett, D. B.; Seeburg, P. H. γ-Aminobutyric Acid_A

Receptor R₅-Subunit Creates Novel Type II Benzodiazepine

Receptor Pharmacology. J . Neurochem. 1990, 54, 1802-1804.

(b) Pritchett, D. B.; Luddens, H.; Seeburg, P. H. Type I and Type

II GABA_A-Benzodiazepine Receptors Produced in Transfected

Cells. Science 1989, 245, 1389-1392.

(22) Luddens, H.; Pritchett, D. B.; Kohler, M.; Killisch, I.; Keinanen,

K.; Monyer, H.; Sprengel, R.; Seeburg, P. H. Cerebellar GABA_A

Receptor Selective for a Behavioral Alcohol Antagonist. Nature

1990, 346, 648-651.

(23) Puia, G.; Vicini, S.; Seeburg, P. H.; Costa, E. Influence of

Recombinant γ-Aminobutyric Acid_AReceptor Subunit Composi-

tion on the Action of Allosteric Modulators of γ-Aminobutyric

Acid_A-Gated Cl^-Currents. Mol. Pharmacol. 1991, 39, 691-696.

(24) Carter, D. B.; Thomsen, D. R.; Im, W. B.; Lennon, D. J .; Ngo, D.

M.; Gale, W.; Im, H. K.; Seeburg, P. H.; Smith, M. W. Functional

Expression of GABA_AChloride^-Channels and Benzodiazepine

Binding Sites in Baculovirus Infected Insect Cells. Bio/ Technol-

ogy 1992, 10, 679-681.

(25) Hamilton, B. J .; Lennon, D. J .; Im, H. K.; Im, W. B.; Seeburg,

P. H.; Carter, D. B. Stable Expression of Cloned Rat GABA_A

Receptor Subunits in a Human Kidney Cell Line. Neurosci. Lett.

1993, 153, 206-209.

(26) J acobsen, E. J .; TenBrink, R. E.; Stelzer, L. S.; Belonga, K. B.;

Carter, D. B.; Im, H. K.; Im, W. B.; Sethy, V. H.; Tang, A. H.;

VonVoigtlander, P. F.; Petke, J . D. Unpublished observations.

(27) Im, H. K.; Im, W. B.; Pregenzer, J . F.; Carter, D. B.; J acobsen,

E. J .; Hamilton, B. J . Characterization of U-97775 as a GABA_A

Receptor Ligand of Dual Functionality in Cloned Rat GABA_A

Receptor Subtypes. Br. J . Pharmacol. 1995, 115, 19-24.

(36) Zhang, W.; Koehler, K. F.; Harris, B.; Skolnick, P.; Cook, J . M.

Synthesis of Benzo-Fused Benzodiazepines Employed as Probes

of the Agonist Pharmacophore of Benzodiazepine Receptors. J .

Med. Chem. 1994, 37, 745-757.

(37) Brown, H. C.; Yoon, N. M. Selective Reductions. X. Reaction of

Aluminum Hydride with Selected Organic Compounds Contain-

ing Representative Functional Groups. Comparison of the

Reducing Characteristics of Lithium Aluminum Hydride and Its

Derivatives. J . Am. Chem. Soc. 1966, 88, 1464-1472.

(38) Determined by ¹H NMR integration of the imidazole methine

protons at δ 8.43 and 9.13 for the product and starting material,

respectively.

(39) Pure 19c was prepared by resubmitting a portion of the crude

product mixture to the reaction conditions.

J M960070+

Article Doi

3-Phenyl-substituted imidazo[1,5-a]quinoxalin-4-ones and imidazo[1,5- a]quinoxaline ureas that have high affinity at the GABA(A)/benzodiazepine receptor complex

DOI: 10.1021/jm960070+

Source and publish data:

Authors:

Article abstract of DOI:10.1021/jm960070+

Full text of DOI:10.1021/jm960070+

Products guided by the article

R&D Labs maybe for 124423-98-9

Relevant to this article

Hot Product