Design and synthesis of tricyclic derivatives as high density lipoprotein cholesterol enhancers

Article abstract of DOI:10.1016/S0960-894X(00)00668-5

A pharmacophore for increasing HDLC was proposed based on common structural features of non-thio-containing compounds with HDLC enhancing properties. A search of the compound database identified various series of these non-thio-containing compounds, including a novel tricyclic imidazoisoquinolone. Preparation of 1-aryl-3-oxo-1,3-dihydro-2-benzofuran-1-carboxamides using a novel and widely applicable one-step process from 2-acyl benzoic acids is reported. Reaction of diamines with 1-aryl-3-oxo-1,3-dihydro-2-benzofuran-1-carboxamides and related aza-analogues proceeded with regio-control to furnish imidazoisoquinolones, pyrimidoisoquinolones, and imidazonaphthyridines. NMR studies and X-ray crystallography confirmed the regio-chemistry of the products. Compounds of these series increased concentrations of HDLC in test animals following oral administration.

Full text of DOI:10.1016/S0960-894X(00)00668-5

Bioorganic & Medicinal Chemistry Letters 11 (2001) 339±342
Design and Synthesis of Tricyclic Derivatives as High Density
Lipoprotein Cholesterol Enhancers
Hassan Elokdah,a,* Sie-Yearl Chai,a Douglas Hob and Theodore Sulkowskia
aChemical Sciences, Wyeth-Ayerst Research, CN 8000, Princeton, NJ 08543, USA
bDepartment of Chemistry, Princeton University, Princeton, NJ 08544, USA
Received 4 October 2000; accepted 15 November 2000
AbstractÐA pharmacophore for increasing HDLC was proposed based on common structural features of non-thio-containing
compounds with HDLC enhancing properties. A search of the compound database identi®ed various series of these non-thio-con-
taining compounds, including a novel tricyclic imidazoisoquinolone. Preparation of l-aryl-3-oxo-1,3-dihydro-2-benzofuran-1-car-
boxamides using a novel and widely applicable one-step process from 2-acyl benzoic acids is reported. Reaction of diamines with 1-
aryl-3-oxo-1,3-dihydro-2-benzofuran-1-carboxamides and related aza-analogues proceeded with regio-control to furnish imidazo-
isoquinolones, pyrimidoisoquinolones, and imidazonaphthyridines. NMR studies and X-ray crystallography con®rmed the regio-
chemistry of the products. Compounds of these series increased concentrations of HDLC in test animals following oral adminis-
tration. # 2001 Elsevier Science Ltd. All rights reserved.
Serum high density lipoprotein cholesterol (HDLC)
concentrations have been shown to inversely correlate
to both the risk of coronary heart disease (CHD) in
humans and the severity of experimental atherosclerosis
in animals.1 Atherosclerosis is the process of accumula-
tion of cholesterol within the arterial wall that results in
the occlusion, or stenosis, of coronary and cerebral
arterial vessels and subsequent myocardial infarction
and stroke. Angiographical studies have shown that
elevated levels of some HDL particles appear to be cor-
related to a decreased number of sites of stenosis in the
coronary arteries of humans.2 HDL may protect against
the progression of atherosclerosis through several
mechanisms. In vitro studies have shown that HDL is
capable of removing cholesterol from cells.3 Data of this
nature suggest that one antiatherogenic property of
HDL may lie in its ability to deplete tissues of excess
free cholesterol and eventually lead to the delivery of
this cholesterol to the liver.4 This has been supported by
experiments showing ecient transfer of cholesterol
from HDL to the liver.5 In addition. HDL may serve as
a reservoir in the circulation for apoproteins necessary
for the rapid metabolism of triglyceride-rich lipopro-
teins.6 Accordingly, agents that increase serum HDLC
concentrations would be of utility as anti-athero-
sclerotic agents useful in the treatment of dyslipopro-
teinemias and related coronary heart disease. An
increase in total serum cholesterol may result from an
increase of the HDLC fraction. Data mining for com-
pounds that increased total serum cholesterol in test
animals generated a diverse list of compounds. Testing
of representative compounds in animal models identi-
®ed several hits that increase the HDLC without alter-
ing other lipid fractions. However, the majority of these
agents fell into various thio-containing classes of com-
pounds such as thiohydantoins, thiouracils, thiosemi-
carbazones, and sulfanyl imidazolidinones.7 The
benzamide series 2 and taclamine 3 were amongst the
non-thio-containing hits. These were of interest to our
research group (Fig. 1).
These two apparently di€erent classes of compounds
carry common structural elements that can be super-
imposed within a three-dimensional matrix; the tertiary
*Corresponding author. Tel.: +1-732-274-4504; fax: +1-732-274-
4505; e-mail: elokdah@war.wyeth.com
Figure 1. Structures of non-thio-containing hits with HDLC
enhancing properties.
0960-894X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(00)00668-5

340
H. Elokdah et al. / Bioorg. Med. Chem. Lett. 11 (2001) 339±342
amine and the two phenyl rings. Due to these com-
monalities, the three moieties were postulated to de®ne
a pharmacophore associated with the HDLC enhance-
ment. Compounds of structure 2 can adopt the con-
formations of the more rigid taclamine 3. Taclamine
was thus selected as a model for further structure based
searching. Both diastereomers (R,R and R,S) of tacla-
mine were energy minimized (Tripos force ®eld). Aver-
age distances between the centroids of the phenyl rings
(a=5.0 A) and the distances between the nitrogen atom
and either phenyl ring centroid (b=3.8 A and c=5.2 A),
were determined based on the geometry of both isomers
(Fig. 2). Using these three points and retaining the dis-
tance requirement, a 3-D search of the corporate data-
base was performed. The search generated a number of
compounds that clustered into several structural series.
Of these, the imidazoisoquinolone series 4 was identi®ed
as most interesting. In this paper, the synthesis and
evaluation of imidazoisoquinolones and the related
pyrimidoisoquinolones and imidazonaphthyridines as
HDLC enhancing compounds will be discussed.
acyl benzoic acids 5 to the corresponding 3-carboxa-
mido phthalides 9 (Scheme 1). In this process, 2-acyl
benzoic acids and potassium cyanide are heated in ace-
tic acid at atmospheric pressure or in a stoppered pres-
sure bottle to yield the desired compounds 9. Using this
one-step process, various 3-carboxamido phthalides
were prepared. Representative examples are shown in
Table 1.
Reaction of 9 with diamines in re¯uxing toluene a€or-
ded the tricyclic derivatives 1 (Scheme 2). Reaction of 9
with ethylenediamine provided the imidazoisoquino-
lones, and with 1,3-propanediamine a€orded the pyr-
imidoisoquinolones. Reaction of 9b with 2-methyl-1,2-
propane diamine furnished dimethylimidazoisoquino-
lone as a single regioisomer 1d (Scheme 3).
Attempts to elucidate the regiochemical outcome of the
reaction using 1H±13C NMR correlation experiments
were unsuccessful. Alternatively, hydride reduction of
compound 1d provided the hexahydroimidazoisoquino-
linol 12 allowing for structural determination by NOE
experiments. The induction of NOE intensity increases
between the C-3 and C-5 proton resonances of 12
supported the formation of regioisomer 1d rather than
11 (Scheme 3). Compound 1d is the expected product
from the attack of the less hindered amine at the lactone
carbonyl of 9b.
Synthesis of imidazoisoquinolones 4 required the pre-
paration of 3-carboxamido phthalides 9. Synthesis of 9
from 2-acyl benzoic acids 5 in a ®ve-step process has
been described in the literature.8 This procedure, as out-
lined in Scheme 1, involves reduction of the 2-acyl benzoic
acid 5 to the lactone 6. Metalation of the 3-position of 6
and subsequent quenching with carbon dioxide a€orded
the lactone acid 7. Conversion of 7 to the corresponding
acid chloride 8 followed by treatment with ammonia
provided 9.
To further con®rm the generality of the regiocontrol of
the reaction, 1,2-propanediamine was reacted with 1-(4-
¯uorophenyl)-3-oxo-1,3-dihydro-2-benzofuran-1-carbox-
amide 9b.9 The reaction a€orded a mixture of two
diastereomers 1f and 1g that were separated and crys-
tallized. X-ray crystal structures of both isomers were
obtained (Figs. 3 and 4) and their relative stereo-
chemistry was unambiguously assigned.10
In the course of our studies, we developed a simpli®ed,
novel, and widely applicable one-step conversion of 2-
Table 1. One-step conversion of o-acyl benzoic acids (5) to 3-car-
boxamido phthalides (9)
Compd
A
R
Reaction time (h)
Yieldc (%)
9a
9b
9c
9d
9e
9f
CH
CH
CH
N
CH
CH
CH
Phenyl
70a
48a
9b
48a
68a
48a
l7a
54
50
36
71
80
70
41
Figure 2. Proposed three-point pharmacophore and the discovery of
imidazoisoquinolones as HDLC enhancers.
4-Fluoro-phenyl
4-Chloro-phenyl
4-Chloro-phenyl
3-Bromo-phenyl
Methyl
9g
Hydrogen
aReaction run in a pressure bottle.
bReaction run at atmospheric pressure.
cUnoptimized isolated yield.
Scheme 1. Conversion of 2-acyl benzoic acids (5) to 3-carboxamido
phthalides (9): (a) Zn, AcOH; (b) (Na/K/Li) NH2, NH3; (c) CO2; (d)
SOCl2; (e) NH3; (f) KCN, AcOH, heat.
Scheme 2. Synthesis of imidazoisoquinolin-5-ones, pyrimido-
isoquinolin-6-ones and imidazonaphthyridin-5-ones (1). (a) Diamine,
toluene, heat.

H. Elokdah et al. / Bioorg. Med. Chem. Lett. 11 (2001) 339±342
341
Target compounds were evaluated in an in vivo assay
for their e€ects on the lipid pro®le.11 HDL cholesterol
concentrations in serum were determined by separating
the lipoprotein classes using a modi®cation of Kieft's12
method.
other lipid fractions such as LDLC, VLDLC, triglyceride,
or total cholesterol.
In conclusion, database mining led to the discovery of
numerous thio-containing and non-thio-containing
compounds with HDLC enhancing properties. Based on
the common structural elements of the non-thio-con-
The changes in HDLC concentration following
treatment with imidazoisoquinolin-5-one derivatives,
pyrimidoisoquinolin-6-one derivatives and imidazo-
naphthyridin-5-one derivatives are presented in Table 2.
Generally, compounds of this series increase HDLC
concentrations. Signi®cant increases were found for
compounds with the 2-methyl or 2,2-dimethyl substitu-
tion 1a, 1d, 1f, and 1g. Substitution of a ¯uorine at R1
signi®cantly enhanced activity in compounds with R2
and R3 substitution (1d vs 1a); however, in the absence
of R2 and R3 substitution, a ¯uorine substitution exhi-
bits no e€ect (1b). Signi®cant di€erences in the activities
of the two mono-methyl diastereomers 1f and 1g were
observed. The 2R,10S isomer 1f was more potent than
the 2R,10R isomer 1g. Importantly, none of the com-
pounds tested had any signi®cant e€ect on the levels of
taining compounds,
a
proposed pharmacophore
was generated. Using this pharmacophore model, a 3-D
search of the database identi®ed the novel tricyclic
Figure 4. X-ray crystal structure of Æ-(2R,10R)-10-(4-¯uorophenyl)-
10-hydroxy-2-methyl-2,10-dihydroimidazo[1,2-b]isoquinolin-5(3H)-one
(1g), triclinic, P1/C, a=13.789(1) A, b=15.554(2) A, c=16.163(2) A,
a=114.088(7)ꢀ, b=104.515(6)ꢀ, g=93.864(7)ꢀ, V=3006.6(5) A3, Z=8.
Table 2. E€ect of imidazoisoquinolin-5-one derivatives, pyrimido-
isoquinolin-6-one derivatives and imidazonaphthyridin-5-one deriv-
atives on HDLC in cholesterol cholic acid-fed male rats
Scheme 3. Regiochemistry of addition of diamines to 9b. Hydride
reduction of 1d to 12.
Compd
A
n
R1
R2
R3
R4
HDLCa
1a
1b
1c
1d
1e
1f
1g
1h
1i
CH
CH
CH
CH
CH
CH
CH
N
0
0
1
0
0
0
0
0
0
H
F
F
F
F
F
F
Cl
Cl
Me
H
H
Me
H
H
H
H
H
Me
H
H
Me
H
H
H
H
55b
0
22
90b
21
-(CH2)4-
Me(2R,10S)
Me(2R,10R)
H
H
H
H
H
105b
38b
49b
c
CH
H
aAverage % HDLC change in six animals (vs control) after treatment
for 8 days at a dose of 100 mg/kg/day.
bStatistically signi®cant with P<0.05.
cNot tested.
Figure 3. X-ray crystal structure of Æ-(2R,10S)-10-(4-¯uorophenyl)-10-
hydroxy-2-methyl-2,10-dihydroimidazo[1,2-b]isoquinolin-5(3H)-one (1f),
monoclinic, P21/C, a=17.5165(6) A, b=10.7125(3) A, c=7.8425(3) A,
b=92.026(2)ꢀ, V=1470.69(9) A3, Z=4.

342
H. Elokdah et al. / Bioorg. Med. Chem. Lett. 11 (2001) 339±342
imidazoisoquinolone series. Compounds of this series
were prepared from the corresponding 3-carboxamido
phthalides and diamines with regiochemical control as
con®rmed by NMR and X-ray crystallography. A novel
and widely applicable one-step synthesis of 3-carboxa-
mido phthalides was discovered. Finally, several of
these derivatives exhibited potent HDLC enhancing
activity in an animal model.
cyanide (28 g, 0.431 mol), and glacial acetic acid (160 mL) was
heated at 115±125 ꢀC in a sealed pressure bottle for 48 h.
After cooling to ambient temperature, the mixture was
poured into ice water (1.5 L). The solid was collected by
®ltration and crystallized from ethanol to give 42 g of 1-(4-
¯uorophenyl)-3-oxo-1,3-dihydro-2-benzofuran-1-carboxamide
(9, R=4-¯uorophenyl, R0=H) as a white solid, mp l70±
173 ꢀC. Mass spectrum (EI, M+) m/z 271. Anal. for
C15H10NO3F. Calcd: C, 66.42; H, 3.72; N, 5.16. Found: C,
66.06; H, 3.83; N, 5.14. A mixture of 1-(4-¯uorophenyl)-3-oxo-
1,3-dihydro-2-benzofuran-1-carboxamide (12.5 g, 0.046 mol),
1.2-diamino-propane (13.5 g, 0.182 mol), and toluene (150 mL)
was heated at re¯ux for 18 h in a ¯ask equipped with a water
separator. The mixture was evaporated to dryness. The residue
was treated with hot ethanol (200 mL). The solids were col-
lected by ®ltration and dried to give 4.3 g of (2R,10S)-10-(4-
¯uorophenyl)-10-hydroxy-2-methyl-2,10-dihydroimidazo[1,2-b]-
isoquinolin-5(3H)-one (1f) as a white solid, mp 229±232 ꢀC.
Mass spectrum (EI, M+) m/z 310. 1H NMR (400 MHz,
DMSO-d6) d 8.04 (d, 1H, J=7.7 Hz), 7.62 (t, 1H, J=7.3 Hz),
7.51 (t, 1H, J=7.3 Hz), 7.35±7.29 (m, 3H), 7.10 (t, 2H,
J=8.8 Hz), 7.04 (s, 1H), 4.16±4.10 (m, 1H), 4.02 (dd, 1H,
J=11.2, 11.0 Hz), 3.52 (dd, 1H, J=11.2, 5.93 Hz) 1.20 (d, 3H,
J=6.6 Hz). Anal. for C18H15N2O2F Calcd: C, 69.67; H, 4.87;
N, 9.03. Found: C, 69.86; H, 4.77; N, 9.14. The ethanolic ®l-
trate was evaporated to dryness. The residue was puri®ed by
¯ash chromatography on silica gel using methanol/methylene
chloride (8:92). Crystallization from ethyl acetate/hexane gave
2.9 g of (2R,10R)-10-(4-¯uorophenyl)-10-hydroxy-2-methyl-2,10-
dihydroimidazo[1,2 - b]-isoquinolin-5(3H)-one (1g) as a white
Acknowledgements
We would like to thank Mr. Bruce Hofmann and Mr.
James Mattes for performing 1H NMR experiments,
Dr. Reinhardt Stein for molecular modeling, Ms.
Christina Kraml for assistance with X-ray crystal-
lography and Dr. Elaine Quinet for animal testing.
References and Notes
1. (a) For a review article see: Barter, P. J.; Rye, K.-A.
Atherosclerosis 1996, 121, 1. (b) Barr, D. P.; Russ, E. M.;
Eder, H. A. Am. J. Med. 1951, 11, 480. (c) Gofman, J. W.;
Young, W.; Tandy, R. Circulation 1966, 34, 679. (d) Miller,
G. J.; Miller, N. E. Lancet 1975, 1, 16. (e) Gordon, D. J.;
Probst®eld, J. L.; Garrison, R. J.; Neaton, J. D.; Castelli,
W. P.; Knoke, J. D.; Jacobs, D. R., Jr.; Bangdiwala, S.;
Tyroler, H. A. Circulation 1989, 79, 8. (f) Stampfer, M. J.;
Sacks, F. M.; Salvini, S.; Willett, W. C.; Hennekens, C. H. N.
Engl. J. Med. 1991, 325, 373. (g) Badimon, J. J.; Badimon, L.;
Galuz, A.; Dische, R.; Fuster, V. Lab. Invest. 1989, 60, 455.
2. Miller, N. E.; Hammett, F.; Saltissi, S.; Rao, S.; Zeller,
H. V.; Coltant, J.; Lewis, B. Br. Med. J. 1981, 282, 1741.
3. Picardo, M.; Massey, J. B.; Kuhn, D. E.; Gotto, A. M., Jr.;
Gianturco, S. H.; Pownall, H. J. Arteriosclerosis 1986, 6, 434.
4. Glomset, J. A. J. Lipid Res. 1968, 9, 155.
5. (a) Glass, C. K.; Pittman, R. C.; Keller, G. A.; Steinberg,
D. J. Biol. Chem. 1983, 258, 7161. (b) Mackinnon, M.; Savage,
J.; Wishant, R.; Barter, P. J. Biol. Chem. 1986, 261, 2548.
6. (a) Grow, T. E.; Fried, M. J. Biol. Chem. 1978, 253, 8034.
(b) Lagocki, P. A.; Scanu, A. M. J. Biol. Chem. 1980, 255, 3701.
(c) Schaefer, E. J.; Wetzel, M. G.; Bengtsson, G.; Scow, R. O.;
Brewer, H. B., Jr.; Olivecrons, T. J. Lipid Res. 1982, 23, 1259.
7. (a) Elokdah, H. M.; Chai, S.; Sulkowski, T. S.; Strike, D. P.
U.S. Patent 5,554,607, 1996; Chem. Abstr. 1996, 125, 275874.
(b) Chai, S.; Elokdah, H. M.; Sulkowski, T. S. U.S. Patent
5,807,864. 1998; Chem. Abstr. 1998, 129, 245160. (c) Com-
mons, T. J.; Christman, S. U.S. Patent 5,968,975, 1999; Chem.
Abstr. 1999, 131, 286270. (d) Elokdah, H.; Sulkowski, T.;
Cochran, D.; McKean, M.-L.; Quinet, E. Bioorg. Med. Chem.
Lett. 2000, 10, 1791.
solid, mp 142±146 ꢀC. Mass spectrum (EI, M+) m/z 310. H
1
NMR (400 MHz, DMSO-d6) d 8.04 (d, 1H, J=7.91 Hz), 7.61
(t, 1H, J=7.47 Hz), 7.49 (t, 1H, J=6.69 Hz), 7.39 (d, 1H,
J=7.91 Hz), 7.36±7.31 (m, 2H), 7.10 (t, 2H, J=9.0 Hz), 6.96
(s, 1H), 4.24±4.38 (m, 1H), 4.14 (dd, 1H, J=11.2, 11.0 Hz),
3.45 (dd, 1H, J=11.0, 6.6 Hz), 1.13 (d, 3H, J=6.6 Hz). Anal.
for C18H15N2O2F. Calcd: C, 69.67; H, 4.87; N, 9.03. Found:
C, 69.44; H, 4.84; N, 9.19.
10. Data have been deposited with the Cambridge Crystal-
lographic Data Center as supplementary publication nos.
CCDC 149545 and CCDC 149546.
11. Male Sprague±Dawley rats weighing 200±225 g are housed
two per cage and fed Purina Rodent Chow Special Mix 5001-S
supplemented with 0.25% cholic acid and 1.0% cholesterol
and water ad libitum for 8 days. Each test substance is admi-
nistered to a group of six rats fed the same diet with the test
diet mixed in as 0.005±0.1% of the total diet. Body weight and
food consumption are recorded prior to diet administration
and at termination. Typical doses of the test substances are 5±
100 mg/kg/day. At termination, blood is collected from anes-
thetized rats and the serum is separated by centrifugation.
Total serum cholesterol is assayed using the Sigma Diag-
nostics enzymatic kit for the determination of cholesterol,
Procedure No. 352, modi®ed for use with 96-well microtiter
plates.
8. Hauser, C. R.; Tetenbaum, M. T.; Ho€enberg, D. S. J.
Org. Chem. 1958, 23, 861.
9. Preparation and spectral data of (2R,10S)- and (2R,10R)-
10-(4-¯uorophenyl)-10-hydroxy-2-methyl-2,10-dihydroimid-
azo[1,2-b]isoquinolin-5(3H)-one (1f and 1g): A mixture of 2-
(4-¯uorobenzoyl)benzoic acid (75 g, 0.307 mol), potassium
12. (a) Kieft, K. A.; Bocan, T. M. A.; Krause, B. R. I. Lipid
Res. 1991, 32, 859. (b) The relative concentration of each
lipoprotein class is calculated as the percent of total absor-
bance. HDLC concentration, in serum, is calculated as the
percent of total cholesterol as determined by FPLC multiplied
by the total serum cholesterol concentration.