Design and synthesis of poly(ADP-ribose) polymerase-1 (PARP-1) inhibitors. Part 4: Biological evaluation of imidazobenzodiazepines as potent PARP-1 inhibitors for treatment of ischemic injuries

Article abstract of DOI:10.1016/S0968-0896(03)00333-X

A class of poly(ADP-ribose) polymerase (PARP-1) inhibitors, the imidazobenzodiazepines, are presented in this text. Several derivatives were designed and synthesized with ionizable groups (i.e., tertiary amines) in order to promote the desired pharmaceuti

Full text of DOI:10.1016/S0968-0896(03)00333-X

Bioorganic & Medicinal Chemistry 11 (2003) 3695–3707
Design and Synthesis of Poly(ADP-Ribose) Polymerase-1
(PARP-1) Inhibitors. Part 4: Biological Evaluation of
Imidazobenzodiazepines as Potent PARP-1 Inhibitors for
Treatment of Ischemic Injuries
Dana Ferraris,* Rica Pargas Ficco, David Dain, Mark Ginski, Susan Lautar,
Kathy Lee-Wisdom, Shi Liang, Qian Lin, May X.-C. Lu, Lisa Morgan, Bert Thomas,
Lawrence R. Williams, Jie Zhang, Yinong Zhou and Vincent J. Kalish
Guilford Pharmaceuticals Inc., 6611 Tributary Street, Baltimore, MD 21224, USA
Received 14 May 2003; accepted 16 May 2003
Abstract—A class of poly(ADP-ribose) polymerase (PARP-1) inhibitors, the imidazobenzodiazepines, are presented in this text.
Several derivatives were designed and synthesized with ionizable groups (i.e., tertiary amines) in order to promote the desired
pharmaceutical characteristics for administration in ischemic injury. Within this series, several compounds have excellent in vitro
potency and our computational models accurately justify the structure–activity relationships (SARs) and highlight essential
hydrogen bonding residues and hydrophobic pockets within the catalytic domain of PARP-1. Administration of these compounds
(5q, 17a and 17e) in the mouse model of streptozotocin-induced diabetes results in maintainance of glucose levels. Furthermore, one
such inhibitor (5g, IC50=26 nM) demonstrated significant reduction of infarct volume in the rat model of permanent focal cerebral
ischemia.
# 2003 Elsevier Ltd. All rights reserved.
Introduction
depletion of NAD+ and, subsequently, a decrease in
the level of intracellular ATP. This depletion of ATP
results in cell death through a necrotic pathway.8,9 Ear-
lier studies done with PARP-1 ‘knockout’ mice demon-
strated that a lack of PARP-1 offers protection in
animal models of stroke10 and heart ischemia.11,12 More
recently, PARP-1 has also been implicated in the
destruction of b-islet cells in diabetic mice.13ꢀ15 These
data indicate that PARP-1 may be a clinically relevant
target for ischemia/reperfusion injuries, and the pro-
gression of type 1 diabetes mellitus, two areas of medi-
cine with unmet therapeutic needs.
Poly(ADP-ribose) polymerase-1 (PARP-1, EC 2.4.2.30)
also known as poly(ADP-ribose) synthetase is an
abundant nuclear enzyme with an important role in the
cellular life cycle.1ꢀ3 The PARP-1 enzyme has three
important structural domains: the DNA binding
domain containing two zinc fingers, the automodification
domain and the catalytic domain.4,5 The catalytic
domain is responsible for converting nicotinamide
adenine dinucleotide (NAD+) into nicotinamide and
an ADP-ribose unit that is attached to either PARP-1
itself or other proteins.6 During normal cell division,
PARP-1 is involved with repairing damaged DNA and
for this reason inhibitors of this enzyme have been uti-
lized as an adjunct for cancer chemotherapy treatment.7
When overactivated, however, PARP-1 can cause
extensive polymerization of ADP-ribose, leading to the
Most of the published inhibitors of PARP-1 have a
common structural motif, namely the benzamido moiety
(outlined in bold, Fig. 1) as shown in substituted benza-
mides 1,16 isoquinolones 2,17 and 5[H]phenanthridin-6-
ones 3.18,19 Structural studies indicate that this amide
forms three important hydrogen bonds with two amino
acid residues in the catalytic domain of PARP-1,
namely Ser904 and Gly863. Removal or alteration of
this amide moiety and disruption of these hydrogen
*Corresponding author. Tel.: +1-410-631-8126; fax: +1-410-631-
6797; e-mail: ferrarisd@guilfordpharm.com
0968-0896/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0968-0896(03)00333-X

3696
D. Ferraris et al. / Bioorg. Med. Chem. 11 (2003) 3695–3707
we decided to derivatize at this position in order to
improve the pharmaceutical characteristics of our com-
pounds without the loss of in vitro potency. In this paper,
in vitro potencies and permeability analysis were gathered
for representative members of this series. Most impor-
tantly, selected imidazobenzodiazepines were found to be
efficacious in animal models of diabetes and stroke and
the results from these studies will be reported herein.
Chemistry
The initial synthesis of imidazobenzodiazepines is out-
lined in Scheme 1 (Method A). The commercially avail-
able benzoic acid 6 was esterified to methyl ester 7 via
the acid chloride in 95% yield.25 The formation of the
seven-membered diazepine ring was accomplished by
refluxing with ethylene diamine in n-butanol to afford nitro
diazepine 8 in 84% yield. Derivatization of compound 8
was accomplished by acylation with the requisite acid
chlorides leading to amides 9b–e. Although nitro amides
9b–e were reduced quantitatively, the subsequent cycli-
zation step required refluxing in toluene over extended
periods of time affording compounds 5b–e in low yield.
Figure 1. Classes of PARP-1 inhibitors.
bonds results in compounds with negligible activity in
almost every instance.20 In addition, the conformation
of this amide is important to potency. That is, there are
two equally populated conformations for this amide in
this ‘first-generation’ benzamide series (1, A and B, Fig.
1). The preferred conformation of this amide in the
active site, however, is illustrated by conformation
A.21,22 Hence, constraining this amide within an adjacent
ring (mimicking conformation A) led to a ‘second
generation’ of compounds that were over an order of
magnitude more potent than the ‘first-generation’ ben-
zamides as shown in the isoquinolone and 5[H]-phen-
anthridin-6-one series.
This prompted us to explore an alternative synthetic
route from 8 to 5 as outlined in Scheme 1 (Method B).
The nitro diazepine 8 was reduced in the presence of
Raney nickel and hydrazine to yield amine 10 in 94%
yield. Various aldehydes reacted with this diamine in the
presence of palladium on carbon to form the benzimi-
dazole ring in moderate to good yield (30–82%). All of
the final products were easily recrystallizable from ethyl
acetate or acetonitrile. The higher yields, relative ease of
purification and abundance of commercially available
aldehydes make Method B desirable for large-scale pre-
paration and rapid analysis of structure–activity rela-
tionships of 2-substituted analogues. The parent
compound (5a, R=H) was initially synthesized as well
as several aryl- and heteroaryl analogues (5f–q). A
similar transformation from phenylene diamines to
benzimidazoles was undertaken in previous publications
using Cu(OAc)2.20,26 This method, while affording the
desired benzimidazoles in moderate yields, the workup
requires subjecting the compounds to strongly acidic
and basic conditions. The workup for method B is
much milder and requires just a simple filtration and
recrystallization in many instances.
The benzimidazoles 4a–b were concurrently designed
with the idea of elegantly forcing this amide into the
desired conformation through an intramolecular
hydrogen bond (4a and 4b, X=N, Y=H, Fig. 1). Using
various methods to synthesize benzimidazoles, a wide
variety of 2-substituted analogues were made which
demonstrated good PARP-1 inhibition.20 In addition,
crystallographic studies confirmed the importance of
conformation A (Fig. 1), the size of the nicotinamide
binding pocket and the tolerance for large substituents
at the 2-position. Formation of a third ring (seven
membered) in place of this intramolecular hydrogen
bond has also been done recently, leading to a new ser-
ies of tricyclic PARP-1 inhibitors (4c, X=C,
Y=–CH2CH2–Fig. 1) with similar SAR to the benzi-
midazoles and in vitro potency in models of cellular
growth inhibition.23,24
Substituted aryl derivatives were synthesized to deter-
mine simple structure–activity relationships around the
phenyl ring of compound 5c. Phenyl substituted analo-
gues 5m–q were prepared by the route outlined in
Scheme 1 (Method B) in yields ranging from 32 to 82%.
The ortho-, meta- and para-methyl derivatives 5m, 5n
and 5o were synthesized to determine the effect of sim-
ple aryl substitution on enzyme activity. Further deri-
vatization from the 3- and 4-positions was performed to
incorporate tertiary amines in the compounds for salt
formation and improved physiologic solubility. Other
amine salts 11a–b were synthesized by the coupling of
carboxylic acid 5p with N-methylpiperazine and N,N,N-
trimethylpropylenediamine in dichloromethane in
We report herein the synthesis and in vitro activity of
imidazobenzodiazepines 5 as a related series of tricyclic
PARP-1 inhibitors (Fig. 1). This class of compounds
was initially designed to restrict the conformation of the
amide pharmacophore within a seven-membered ring
analogous to the recent tricyclic indoles.24 Under-
standing the relative tolerance that PARP-1 has for
bulky substituents at the 2-position of benzimidazoles,

D. Ferraris et al. / Bioorg. Med. Chem. 11 (2003) 3695–3707
3697
Scheme 1. (a) Toluene/SOCl2, MeOH, 95%; (b) ethylenediamine, NaHCO3, nBuOH, 84%; method A: (c) THF, RC(O)Cl, Et3N, 30–87%; (d) H2,
50 psi, Pd/C, MeOH, quant; (e) toluene reflux, 3 days, 30–55%; method B: (f) Raney Ni, H2NNH2, MeOH, 94%; (g) RCHO, Pd/C, MeOH, 30–
82%.
5r, 5s and 5t were formed by the addition of diamine 10
to the requisite aldehydes as illustrated in Scheme 1
(Method B) in yields ranging from 35 to 82%. The
alcohol 12 was synthesized from the ester 5t by reduc-
tion with lithium aluminum hydride in 96% yield. Fur-
ther chlorination of the alcohol 12 with thionyl chloride
afforded benzylic chloride 13 in 72% yield. Amination
with pyrrolidine directly formed the water-soluble
amine hydrochloride salt 14ꢁ HCl.
Using tertiary amines as a means to make physiologi-
cally soluble salts led to the synthesis of derivatives 17a–e
Scheme 2. Method C: (a) EDC/DMAP, DCM/NMP, HNR2, 51–91%
yield; (b) 2.0 M HCl/Et2O (xs), quant.
(Scheme 4). For the preparation of amines 17a–e, the
route outlined in Scheme 4 was utilized. The diamine 10
reacted with the commercially available t-butyldi-
methylsilyloxyacetaldehyde in THF to yield the alcohol
15 after deprotection with TBAF. The chlorination of
alcohol 15 with thionyl chloride led to the chloride 16 in
74% yield. Amination of chloride 16 in acetonitrile
resulted in amines 17a–e in yields ranging from 69–89%.
The salt forms of these amine derivatives (17a–eꢁ HCl)
were synthesized by simple protonation with hydro-
excellent yields (Method C, Scheme 2). The salt forma-
tion of 5q and 11a–b was accomplished by suspending
the free base in ethyl acetate followed by the addition of
hydrochloric acid (2.0 M in diethyl ether) in order to
quantitatively precipitate the HCl salt out of solution.
Similarly, the phenethyl substituted derivatives of 5g
were synthesized as outlined in Scheme 3. Compounds
Scheme 3. (a) LAH, THF, 96%; (b) SOCl2, 72%; (c) pyrrolidine, CH3CN, 92%.

3698
D. Ferraris et al. / Bioorg. Med. Chem. 11 (2003) 3695–3707
Scheme 4. (a) TBDMSOCH2CHO, Pd/C, MeOH, 76%; (b) TBAF, 89%; (c) SOCl2, 74%; (d) HNR2, K2CO3, CH3CN; (e) HCl/Et2O (2.0 M),
EtOAc or CH3CN, 69–89%; (f) 2.0 M HCl/Et2O, quant.
chloric acid and precipitation from ethyl acetate or ace-
tonitrile.
benzamido moiety (2) the planar, fused ring system
(benzimidazole) and (3) the N1 nitrogen that can form a
water-mediated H-bond to the enzyme as discussed in
the molecular modeling section. To reiterate, the result-
ing imidazobenzodiazepine core 5a is an order of mag-
nitude more potent than benzamide 1, isoquinolone 2a,
and benzodiazepines 8 and 10. Compound 5a also dis-
plays similar potency to the compounds in other series
of PARP-1 inhibitors, namely isoquinolones 2b, 5[H]-
phenanthiridin-6-one 3, benzimidazole 4a, and tricyclic
indole 4c (Table 1, entries 7–9).
Biological Evaluation and SAR Discussion
Table 1 summarizes the basic differences in potency
between the benzodiazepines, imidazobenzodiazepines
and other classes of PARP-1 inhibitors. As stated in the
introduction, the ‘first-generation’ PARP-1 inhibitor, 3-
aminobenzamide (3-AB, R=NH2, 1a) displays a mod-
est potency against the enzyme (IC50=9.8 mM). One
member of the ‘second-generation’ PARP-1 inhibitors,
5-nitroisoquinolin-1-one 2a (IC50=3.4 mM) displays
enzymatic activity 3 times better than 3-AB. 5-Amino-
isoquinolin-1-one 2b (IC50=300 nM)27 is one of the
most potent members of this series emphasizing the
importance of the amine in the 5-position. Based on
crystallographic data, this amine is in a position to form
a water-mediated H-bond with PARP-1, accounting for
enhanced potency.20
Outlined in Table 2 are the activities of several 2-sub-
stituted derivatives of the imidazobenzodiazepine core,
such as methyl derivative 5b (entry 2)28 as well as several
aryl- and heteroaryl compounds. The addition of aryl
groups at the 2-position maintained or improved activity
over 5a in almost every instance. The 2-phenyl (5c, entry 3),
2-(m-pyridyl) (5j, entry 10), and 2-furyl (5k, entry 11), deri-
vatives had slightly better activity than the core 5a, while
the pyrrolyl (5l, entry 12) and b-naphthyl derivatives
(5e, entry 5) were 5 and 9 times as potent, respectively.
The potency of the seven-membered nitro benzodiaze-
pine 8 (entry 4) was similar to the 5-nitroisoquinolin-1-
one 2a. The aminobenzodiazepine 10, however, did not
improve to the same degree as 2b underscoring the
necessity of both a planar, fused ring system (iso-
quinolone) and the amine in the 5-position for inhibi-
tion. Improvement to the benzodiazepine core did
occur, however, by fusing a five-membered ring to this
moiety leading to the imidazobenzodiazepine 5a
(IC50=299 nM). Similar to 2b, the imidazobenzodiaze-
pine core has three key elements that contribute to the
overall potency: (1) the conformationally constrained
Table 2. PARP-1 inhibition of 2-substituted imidazobenzodiazepines
Entry Compd
R group
Methoda Yield (%) IC50 (nM)
1
2
3
4
5
6
7
8
5a
5b
5c
5d
5e
5f
5g
5h
5i
H
CH3
Ph
a-Np
b-Np
CH2Ph
B
A
A
A
A
B
B
B
B
B
B
B
32
299
141
90
382
32
108
26
71
44
Table 1. PARP-1 inhibition for several classes of PARP-1 inhibitors
49b
40b
43b
55b
52
Entry
Compd
IC50 (mM)a
1
2
3
4
5
6
7
8
9
1a
2a
2b
8
10
5a
4a
4c
3
9.8
3.4
0.3
5.1
CH2CH2Ph
(E)-CH¼CHPh
Cꢂ;CPh
3-Pyridyl
2-Furyl
76
75
75
54
75
48
42c
2.6
9
0.299
0.750b
0.203c
0.340
10
11
12
13
5j
5k
5l
168
108
61
2-Pyrrolyl
–CH2N(CH3)2
17a
66
aSee Experimental for details.
bCompound synthesized as indicated in ref 21.
cCompound synthesized as indicated in ref 24.
aSee Experimental for details.
bYield from compound 8.
cSee Scheme 4, overall yield from 10.

D. Ferraris et al. / Bioorg. Med. Chem. 11 (2003) 3695–3707
3699
This result concurs with findings by other groups indi-
cating that the most potent (<100 nM) PARP-1 inhibi-
tors have multiple aromatic rings within their structures,
similar to these derivatives.23,24 In fact, the only aro-
matic moiety that decreased the core activity was the
a-naphthyl group (5d, IC50=382 nM). In compound 5d,
however, the naphthyl ring is more orthogonal to the
core than in any other aryl derivative, thus disrupting
the planarity of the inhibitor.
their synthetic precursor 5p (Table 3, entry 5) demon-
strated a slightly weaker potency than 5c, but they have
the added benefit of a water-solubilizing group (tertiary
amine and carboxylic acid). The para-substituted com-
pound 5q (Table 3, entry 8, IC50=63 nM), however,
was more potent than any phenyl substituted derivative.
The potency of 5q and the potential to form a water-
solubilizing salt prompted further in vivo studies for this
compound as discussed below. Compounds 11a–b and
5q further demonstrate that there is minimal loss in
potency even with large, hydrophobic groups at the 2-
position.
Remarkable improvements in activity also occurred by
the addition of two atoms between the 2-position of the
imidazobenzodiazepine core and aryl group. While the
extension of the aryl moiety by one methylene unit
showed some improvement, as compared to the core
activity (5f, entry 6), the addition of two carbon atoms
between the core and the aryl ring led to three of the
most active compounds in the series (5g–i, entries 7–9)
as discussed vide infra. Also, the addition of a tertiary
amine to the 2-position resulted in compound 17a
whose activity, despite the lack of an aryl substituent is
similar to the best compounds within the series (Table 2,
entry 13), necessitating further derivatization.
The in vitro PARP-1 inhibition results from the deriva-
tives of 5g are outlined in Table 4. There was a notable
decline in activity upon derivatization of the two-carbon
linker. The addition of a methyl group reduced the
activity several fold (Table 4, entry 2) as did the addi-
tion of an amine (Table 4, entry 3). Derivitization of the
aryl ring of 5g also resulted in a moderate reduction in
activity in almost every instance. Hydrophilic groups in
the para position, however, were more deleterious to
activity than hydrophobic ones. For example, the car-
boxylic acid 5u, and the alcohol 12 were an order of
magnitude worse than 5g (Table 4, entries 5 and 6) while
the more hydrophobic ester 5t and amine 14ꢁ HCl
showed comparable activity to 5g (Table 4 entries 4 and 7).
The inhibitory activities of aryl-substituted derivatives
of 5c are outlined in Table 3. In general, substitution at
the ortho-position of the aryl ring, as foreshadowed by
the poor activity of a-naphthyl analogue 5d, was more
deleterious to activity than either the meta- or para-
substitution. Compound 5m was approximately three
times less active than the parent phenyl derivative 5c,
while the isomeric tolyl derivatives 5n and 5o (Table 3,
entries 3 and 4) displayed similar potencies to this core.
With these data in mind, solubilizing groups were
appended to the 3- and 4-positions of the phenyl ring.
The amides 11a and 11b (Table 3, entries 6–7) as well as
The PARP-1 in vitro potencies of amine derivatives
17a–e are outlined in Table 5. Once again, this series of
compounds illustrates the tolerance that the enzyme has
for large, hydrophobic groups. Minimal loss in activity
from 17a is noted with the N-methyl-N-benzyl and
bicyclic derivatives 17b and 17e (Table 5, entries 2 and
5). Even the bulky t-butyl ester derivative 17d (Table 5,
entry 4) is only 3 times worse than 17a.
Table 4. PARP-1 inhibition of analogues of 5gd
Table 3. PARP-1 inhibition of analogues of 5c
Entry Compd
R group
Methoda Yield (%) IC50 (nM)
1
2
3
4
5
5c
5m
5n
5o
5p
H
A
B
B
B
B
40
32
45
69
82
90
329
97
111
155
Entry Compd
R1
R2
R3
Yield (%) IC50 (nM)
2-CH3
3-CH3
4-CH3
3-COOH
1
2
3
4
5
6
5g
5r(ꢃ)
5s(S)
5t
12
5u
H
H
NH2
H
H
H
H
CH3
H
H
H
H
H
H
76a
35a
47b
82a
96c
82a
26
130
183
97
284
259
COOEt
CH2OH
COOH
6
11a
C
91b
161
H
51b
65
244
63
.
14 HCl
7
H
H
92
68
7
8
11b
5q
C
B
aMethod B was used.
4-O(CH2)3N(CH3)2
bYield after deprotection.
cYield after reduction of 5t.
dYield from amination of 13.
aSee Experimental for details.
bYield from coupling reaction in Scheme 2.

3700
D. Ferraris et al. / Bioorg. Med. Chem. 11 (2003) 3695–3707
Table 5. PARP-1 inhibition of analogues of 17a
Entry
1
Compd
NR2
Yield (%)a
82
IC50 (nM)b
66
17a
2
3
4
17b
17c
17d
74
72
70
88
109
187
5
17e
82
74
aRepresents yield from chloride 16.
bSee Experimental for details.
Molecular Modeling Data
Most of the SAR obtained from the compounds in
Tables 1–5 can be explained with the aid of computer
modeling and the published crystal structure data.20,24
We modeled compound 5g with benzimidazole 4b
(R=3-methoxyphenyl) to illustrate the key interaction
between the substituent on the 2-position and the
enzyme. The nicotinamide binding pocket presumably
binds the amide moiety in compound 5g in an analogous
manner to the primary amide in the benzimidazole 4b
(fig. 2). This portion of the pocket contains one hydro-
gen bond from Ser904 and two hydrogen bonds from
Gly863 to the amide oxygen and nitrogen. This binding
conformation is widely accepted and provides the pri-
mary ‘signature’ element for PARP-1 inhibitors as
defined by previous studies where the amide is altered or
deleted.20,24 There is also a propensity for planar com-
pounds to bind in the active site of PARP-1.29 This fact
can be explained by the position of two parallel tyrosine
residues (Tyr896 and Tyr907) that are within the active
site. These two tyrosine residues interact with the planar
aryl moieties of many PARP-1 inhibitors by forming a
p–electron ‘sandwich’. This pi stacking is beneficial for
binding planar aromatic compounds.30,31 This planar
pocket is utilized by 2-aryl and 2-heteroaryl substituted
imidazobenzodiazepines (e.g., 5c, 5e and 5j–l), account-
ing for their potency. Compounds 5g–i can utilize a sec-
ondary interaction with the hydrophobic pocket formed
by residues Tyr889, Gly888, and Val762. Any substitu-
tion on the aryl ring of 5g (R3) presumably forces this
aryl ring out of this secondary binding pocket, resulting
in a higher IC50 (e.g., 5t, 5u, 12 and 14, Table 4). In
addition, this pocket could explain the increased activity
of naphthyl derivative 5e over its isomer 5d, whose
naphthyl side chain is unable to access this pocket.
Figure 2. Molecular model of 5g and benzimidazole 4b in the PARP-1
catalytic site.
Pharmacology Data
Caco-2 data
In order to further probe the drug-like characteristics of
this series of PARP-1 inhibitors, representative com-
pounds from various subseries (5g, 5k, 5q and 17a) were
tested for permeability using the caco-2 cell model.32 In
this model, the concentration of compound is measured
both apically and basolaterally in caco-2 intestinal cells
to determine the propensity for passive diffusion and/or
active transport. The permeability ratio is determined
for diffusion/transport in each direction (basolateral to
apical, B–A cm/sꢄ106 and apical to basolateral, A–B
cm/sꢄ106) and the ratio of these two results is indicated
in Table 6 as the efflux ratio. As shown in Table 6,
compounds 5g, 5k, 5q and 17a are all highly permeable
(>10 cm/sꢄ10ꢀ6).33 In addition, none of the imidazo-
benzodiazepines tested were high-risk efflux substrates
as indicated by the ratio of permeability measurements
(high-risk substrates: efflux ratio >10).
These Caco-2 results predict that imidazobenzodiaze-
pines are highly permeable and at low risk for efflux.
The hydrophobic nature of this class of compounds is
similar to that of most PARP-1 inhibitors, thus one
would predict comparable permeability profiles with
other classes as indicated by 5[H]-phenanthridin-6-one 3
(entry 5).

D. Ferraris et al. / Bioorg. Med. Chem. 11 (2003) 3695–3707
3701
Table 6. Caco-2 permeability resultsa
of 5g ip 30 min pre occlusion. Then, 30 min after ische-
mia, 20 mg/kg 5g was given as a bolus dose ip After
sacrifice at 24 h, the resulting tissue slices indicated a
59% reduction in infarct volume (Table 8, entry 1). This
reduction compares favorably with the best known
PARP-1 inhibitor reported in the literature.19 In a some-
what more stringent model of stroke, the permanent cer-
ebral ischemic model, 5g also demonstrated a remarkable
neuroprotective effect. Compound 5g reduced infarct
volume by 39% (entry 2, Table 8) when administered in
rats where the middle cerebral artery was permanently
occluded.
Entry
Compd
Permeability
B–Ab
Permeability
A–Bb
Efflux
Ratioc
(cm/sꢄ10ꢀ6
)
(cm/sꢄ10ꢀ6
)
1
2
3
4
5
5g
5k
5q
17a
3
46.2
34.4
20.7
10.3
36.3
65.4
64.0
46.9
18.0
72.9
1.42
1.86
2.27
1.75
2.01
aSee ref 32 for experimental details.
bHigh permeability: >10 cm/sꢄ10ꢀ6; medium permeability: 1–10 cm/
sꢄ10ꢀ6; low permeability: <1 cm/sꢄ10ꢀ6. See ref 33.
cHigh-risk efflux substrate has ratio >10. See ref 33.
As stated above, ischemic diseases should ideally be
treated by iv administration of the drug and results
from iv administered PARP-1 inhibitors in animal
models will be reported in due course.
Streptozotocin-induced diabetic mice data
Previous studies indicate that PARP-1 is involved with
the regeneration of b-islet cells in the pancreas and
consequently delay or even reverse the progression of
type 1 Diabetes mellitus.7 Streptozotocin (STZ) is a
chemical that specifically damages islet b-islet cells in
the pancreas ultimately leading to the development of
type 1 diabetes in mice. This model was utilized to
demonstrate the in vivo efficacy of PARP-1 inhibitors to
attenuate the hyperglycemia. In this model, the com-
pounds were administered ip (or orally for 17a) prior to
insult with streptozotocin. Compounds 5q, 17a and 17e
were pre-administered to STZ induced diabetic mice
and the blood glucose levels were monitored over sev-
eral hours. The administration of 5q, 17a, and 17e at 10
mg/kg resulted in a 67–75% reduction of the hypergly-
cemia (Table 7, entries 3–5) compared to the vehicle
treated animals (no STZ, Table 7, entry 1). That is,
diabetic mice treated with PARP-1 inhibitors recovered
67–75% of their normal glucose levels. The higher dose
(30 mg/kg) of these compounds resulted in a compar-
able reduction while the lowest dose (3 mg/kg) was
ineffective. Perhaps the most interesting result from this
study is that compound 17a even showed moderate effi-
cacy when administered orally (Table 7, entry 6).
Conclusions
We have designed and synthesized a series of 2-sub-
stituted imidazobenzodiazepines as potent PARP-1
inhibitors with clearly defined structure–activity rela-
tionships. This new series includes several examples of
highly potent PARP-1 inhibitors (10 examples, 26–100
nM). Molecular modeling accounts for the potencies of
the most active compounds in vitro as illustrated by an
additional binding pocket for the side chains of com-
pounds 5e and 5g–i. Representative examples from each
subseries were determined to be highly permeable.
Finally and most importantly, this series of PARP-1
inhibitors demonstrates activity in vivo as compounds
5q, 17a and 17e showed prominent reduction of hyper-
glycemia in the mouse streptozotocin induced diabetes
model and compound 5g displayed a significant neuro-
protective effect in the rat transient and permanent
MCAO model of brain ischemia.
Experimental
General
Focal cerebral ischemic stroke data
As PARP-1 activation is implicated in the pathogenesis
of stroke, preliminary data was obtained on PARP-1
inhibitors in two models of stroke, a transient34 and a
permanent middle cerebral areterial occlusion model.35
In the transient model, rats were injected with 20 mg/kg
Melting points were obtained on a MEL-TEMP II
(Meltemp Laboratory Devices, Inc.). Proton nuclear
magnetic resonance were recorded at 400 MHz on a
Varian 400 using deuterated solvent as an internal
standard. Chemical shift values are indicated in
parts per million. Mass spectra were recorded using
a Micromass LCS Platform LC/MS. Elemental ana-
lyses were obtained from Atlantic Microlabs, Inc.
Table 7. Percent reduction of blood glucose levels in STZ induced
micea
Entry
Compd
3 mg/kg
10 mg/kg
30 mg/kg
Table 8. Cerebral ischemic stroke results
1
2
3
4
5
6
Vehichle (no STZ)
STZ control
5qb
100
0
26.6
12.5
46.1
29.2
100
0
67.2
72.7
74.8
62.4
100
0
72.8
78.1
83.2
69.3
Entry
Compound
Dose
(mg/kg)
Reduction in
infarct volume
MCAO
model
17eb
1
2
5g
5g
40a
40a
59 (r=0.009)
39 (r=0.02)
Transientb
Permanentc
17ab
17ac
aCompound was administered 20 mg/kg, 30 min pre-occlusion and 20
mg/kg 30 min post-ischemia.
bSee ref 34.
cSee ref 35.
aSee ref 40 for experimental details.
bDrug was administered ip.
cDrug was administered orally.

3702
D. Ferraris et al. / Bioorg. Med. Chem. 11 (2003) 3695–3707
(Norcross, GA, USA). All reagents were purchased
from Aldrich Chemical (Milwaukee, WI, USA) unless
otherwise stated.
ing Parr Bomb with 10% Pd/C (25 mg). This solution
was hydrogenated at 30 psi for 4 h. The mixture was
filtered through a plug of Celite1 and concentrated. The
crude material was dissolved in boiling toluene and
refluxed for 12 h to induce cyclization. After cooling,
the solution was concentrated and the product was
recrystallized from EtOAc to yield 35 mg (44%) of the
final product.
Synthesis of 9-nitro-1,2,3,4-tetrahydro-benzo[e][1,4]dia-
zepin-5-one (8) 2-Chloro-3-nitro-benzoic acid methyl
ester 7 (1.0 g, 4.7 mmol), prepared from 2-chloro-3-
nitro-benzoic acid 6 according to literature,36 was dis-
solved in n-butanol (5 mL). Sodium carbonate (0.50 g,
4.7 mmol) was added to the solution followed by ethyl-
ene diamine (282 mg, 4.7 mmol). After several h of
heating at 80 ꢅC, an orange solid started to precipitate
out of solution. The reaction was stopped after 16 h and
the orange solid was filtered off and washed with water
several times and recrystallized from EtOAc. The dry
yield of the final orange crystals was 810 mg (84%).
1-Methyl-8,9-dihydro-7H-2,7,9a-triaza-benzo[cd]azulen-
6-one (5b). mp=>300 ꢅC (dec); H NMR (DMSO-d6)
1
d 8.35 (bt, 1H), 7.79 (d, 1H), 7.73 (d, 1H), 7.25 (t, 1H),
4.28 (m, 2H), 3.58 (m, 2H), 2.53 (s, 3H). Anal. calcd for
C11H11N3O: C, 62.84; H, 5.75; N, 19.99. Found: C,
62.43; H, 5.54; N, 19.43.
mp=187–190 ꢅC (dec). H NMR (DMSO-d6) d 8.72 (t,
1-Phenyl-8,9-dihydro-7H-2,7,9a-triaza-benzo[cd]azulen-
6-one (5c). Yield=40%; mp=253–257 ꢅC; 1H NMR
(DMSO-d6) d 8.48 (bt, 1H), 7.89 (m, 4H), 7.59 (m, 3H),
7.37 (t, 1H), 4.47 (m, 2H), 3.54 (m, 2H). Anal. calcd for
C16H13N3O: C, 72.97; H, 4.98; N, 15.96. Found: C,
72.32; H, 5.06; N, 15.88.
1
1H), 8.42 (t, 1H), 8.23 (d, 1H), 8.15 (d, 1H), 6.75 (t, 1H),
3.64 (m, 2H), 3.35 (m, 2H). Anal. calcd for C9H9N3O3:
C, 52.17; H, 4.38; N, 20.28. Found: C, 52.25; H, 4.59; N,
20.19.
Method A
1-Naphthalen-1-yl-8,9-dihydro-7H-2,7,9a-triaza-benzo[cd]-
azulen-6-one (5d). Yield=43%; mp=226–229 ꢅC; 1H
NMR (DMSO-d6) d 8.43 (t, 1H), 8.18 (d, 1H), 8.09 (d,
1H), 7.98 (m, 2H), 7.91 (d, 1H), 7.83 (d, 1H), 7.71 (t,
1H), 7.59 (m, 2H), 7.43 (t, 1H), 4.13 (m, 2H), 3.53 (m,
2H). Anal. calcd for C20H15N3O (0.25H2O): C, 75.57;
H, 4.92; N, 13.22. Found: C, 75.46; H, 4.87; N, 13.19.
General procedure for the acylation of 8. A solution of
compound 8 (460 mg, 2.22 mmol) was dissolved in THF
(10 mL). Triethyl amine (340 mL, 2.44 mmol) was added
to this solution followed by acetyl chloride (175 mL, 2.44
mmol). The solution was stirred overnight with gentle
warming (50 ꢅC). The reaction was quenched with water
(10 mL) followed by extraction with EtOAc (3ꢄ10 mL).
The combined organics were dried, concentrated and
chromatographed on silica gel to yield compound 9b in
30% yield (160 mg).
1-Naphthalen-2-yl-8,9-dihydro-7H-2,7,9a-triaza-benzo
[cd]azulen-6-one (5e). Yield=55%; mp=261–265 ꢅC;
1H NMR (DMSO-d6) d 8.52 (t, 1H), 8.47 (s, 1H), 8.11
(m, 2H), 8.03 (m, 2H), 7.95 (d, 1H), 7.91 (d, 1H), 7.65
(m, 1H), 7.40 (t, 1H), 4.59 (m, 2H), 3.57 (m, 2H). Anal.
calcd for C20H15N3O (1H2O): C, 72.49; H, 5.14; N,
12.68. Found: C, 72.23; H, 5.17; N, 12.65.
1-Acetyl-9-nitro-1,2,3,4-tetrahydro-benzo[e][1,4]diazepin-
1
5-one (9b) H NMR (CDCl3) d 8.13 (d, 1H), 8.06 (d,
1H), 7.72 (t, 1H), 6.96 (t, 1H), 4.97 (m, 2H), 3.36 (m,
2H), 2.25 (s, 3H). This material was used without fur-
ther purification.
Synthesis of 9-amino-1,2,3,4-tetrahydro-benzo[e][1,4]di-
azepin-5-one (10). The nitro compound 8 (1.8 g, 8.7
mmol) was dissolved in MeOH (125 mL) and heated to
40 ꢅC. Raney nickel (200 mg) was added to the solution
followed by dropwise addition of hydrazine mono-
hydrate (5 mL, xs). The reaction was heated to reflux
for 30 min or until all of the starting material was gone.
Then the mixture was filtered hot through a plug of
Celite to remove the residual nickel. The Celite was
washed with boiling MeOH (100 mL) and the filtrate
was concentrated and dried in vacuo. The resulting air
sensitive solid (1.45 g, 94%) was stored under nitrogen.
An analytical sample of 10 could be obtained from tri-
turation with diethyl ether. 1H NMR (DMSO-d6) d 7.91
(t, 1H), 7.03 (d, 1H), 6.69 (d, 1H), 6.47 (t, 1H), 4.97 (s,
2H), 4.64 (m, 2H), 3.20 (m, 2H). Anal. calcd for
C9H11N3O (0.25H2O): C, 61.00; H, 6.26; N, 23.71.
Found: C, 60.68; H, 6.29; N, 23.39.
1-Benzoyl-9-nitro-1,2,3,4-tetrahydro-benzo[e][1,4]diazepin-
1
5-one (9c) Yield=80%. H NMR (CDCl3) d 8.63 (bs,
1H), 8.40 (d, 1H), 8.06 (d, 1H), 7.60 (d, 2H), 7.52 (t,
1H), 7.42 (t, 2H), 6.85 (t, 1H), 4.29 (m, 2H), 3.84 (m,
2H).
1-(Naphthalene-1-carbonyl)-9-nitro-1,2,3,4-tetrahydro-
1
benzo[e][1,4]diazepin-5-one (9d) Yield=87%. H NMR
(CDCl3) d 8.60 (bs, 1H), 8.36 (d, 1H), 8.04 (m, 1H), 7.91
(d, 1H), 7.85 (m, 2H), 7.50 (m, 2H), 7.44 (m, 2H), 6.77
(t, 1H), 4.45 (m, 2H), 3.93 (m, 2H).
1-(Naphthalene-2-carbonyl)-9-nitro-1,2,3,4-tetrahydro-
1
benzo[e][1,4]diazepin-5-one (9e) Yield=85%. H NMR
(CDCl3) d 8.69 (bs, 1H), 8.41 (d, 1H), 8.17 (s, 1H), 8.03
(d, 1H), 7.87 (m, 3H), 7.54 (m, 3H), 6.85 (t, 1H), 4.34
(m, 2H), 3.89 (m, 2H).
Method B
General procedure for the cyclization of amides 9b–e
(5b). The acetamide 9b (100 mg, 0.40 mmol) was dis-
solved in a 1:1 mixture of EtOAc/MeOH (10 mL). This
solution was degassed and added to a nitrogen contain-
General procedure for synthesis of benzimidazoles 5a, f–q.
The diamine 10 (200 mg, 1.13 mmol), paraformaldehyde
or desired aldehyde (1.1 equiv) and palladium on car-

D. Ferraris et al. / Bioorg. Med. Chem. 11 (2003) 3695–3707
3703
bon (50 mg) were all mixed in MeOH (10 mL) and
refluxed overnight. The reaction was filtered hot
through a Celite plug and the filtrate was concentrated
in vacuo. The resulting solid was triturated with diethyl
ether or EtOAc (5 mL) and filtered. Analytical samples
of the final products could be obtained by recrystalliza-
tion in EtOAc or EtOAc/MeOH.
2H), 7.31 (t, 1H), 7.04 (s, 1H), 6.77 (s, 1H), 6.29 (m,
1H), 4.53 (m, 2H), 3.59 (m, 2H); MS (ES+=253.21).
Anal. calcd for C14H12N4O (0.4H2O): C, 64.80; H, 4.97;
N, 21.59. Found: C, 64.78; H, 4.82; N, 21.75.
1-o-Tolyl-8,9-dihydro-7H-2,7,9a-triaza-benzo[cd]azulen-
6-one (5m). Yield=32%; mp=224–229 ꢅC; 1H NMR
(DMSO-d6) d 8.47 (bt, 1H), 7.89 (m, 2H), 7.70 (s, 1H),
7.65 (d, 1H), 7.47 (t, 1H), 7.36 (m, 2H), 4.47 (m, 2H),
3.54 (m, 2H), 2.43 (s, 3H). Anal. calcd for C17H15N3O
(0.5H2O): C, 72.68; H, 5.53; N, 14.96. Found: C, 72.88;
H, 5.58; N, 14.91.
8,9-Dihydro-7H-2,7,9a-triaza-benzo[cd]azulen-6-one (5a).
Yield=32%; mp=225–235 ꢅC (dec); 1H NMR (DMSO-
d6) d 8.45 (bt, 1H), 8.37 (s, 1H), 7.95 (m, 2H), 7.39 (t,
1H), 4.50 (m, 2H), 3.65 (m, 2H). Anal. calcd for
C10H9N3O (0.75H2O) C, 59.84; H, 5.27; N, 20.94.
Found: C, 59.69; H, 5.16; N, 20.73.
1-m-Tolyl-8,9-dihydro-7H-2,7,9a-triaza-benzo[cd]azulen-
6-one (5n). Yield=45%; mp=234–238 ꢅC; 1NMR
(DMSO-d6) d 8.47 (bt, 1H), 7.89 (m, 2H), 7.70 (s, 1H),
7.65 (d, 1H), 7.47 (t, 1H), 7.36 (m, 2H), 4.47 (m, 2H),
3.54 (m, 2H), 2.43 (s, 3H). Anal. calcd for C17H15N3O
(0.5H2O): C, 72.68; H, 5.53; N, 14.96. Found: C, 72.88;
H, 5.58; N, 14.91.
1-Benzyl-8,9-dihydro-7H-2,7,9a-triaza-benzo[cd]azulen-
6-one (5f). Yield=52%; mp=224–227 ꢅC; 1H NMR
(DMSO-d6) d 8.53 (t, 1H), 8.02 (m, 2H), 7.47 9m, 6H),
4.51 (m, 2H), 3.75 (m, 2H), 2.70 (m, 2H). Anal. calcd for
C17H15N3O (0.25H2O) C, 72.45; H, 5.54; N, 14.91.
Found: C, 72.89; H, 5.55; N, 14.86.
1-p-Tolyl-8,9-dihydro-7H-2,7,9a-triaza-benzo[cd]azulen-
6-one (5o). Yield=69%; mp=258–263 ꢅC; 1H NMR
(DMSO-d6) d 8.44 (t, 1H), 7.88 (m, 2H), 7.76 (d, 1H),
7.37 (m, 3H), 4.45 (m, 2H), 3.53 (m, 2H), 2.42 (s, 3H).
Anal. calcd for C17H15N3O: C, 73.63; H, 5.45; N, 15.15.
Found: C, 73.13; H, 5.49; N, 15.10.
1-Phenethyl-8,9-dihydro-7H-2,7,9a-triaza-benzo[cd]azulen-
6-one (5g). Yield=76%; mp=195–200 ꢅC; 1H NMR
(DMSO-d6) d 8.37 (t, 1H), 7.86 (t, 2H), 7.34 (m, 3H),
7.25 (m, 3H), 4.22 (m, 2H), 3.55 (m, 6H). Anal. calcd for
C18H17N3O: C, 74.20; H, 5.88; N, 14.42. Found: C,
73.05; H, 6.04; N, 14.49.
3-(6-Oxo-6,7,8,9-tetrahydro-2,7,9a-triaza-benzo[cd]azulen-
1-yl)-benzoic acid (5p). Yield=82%; mp=300–320 ꢅC;
1H NMR (DMSO-d6) d 13.28 (s, 1H), 8.48 (t, 1H), 8.41
(s, 1H), 8.11 (m, 2H), 7.90 (m, 2H), 7.72 (t, 1H), 7.37 (t,
1H), 4.47 (m, 2H), 3.54 (m, 2H). Anal. calcd for
C17H13N3O3 (0.25H2O) C, 65.48; H, 4.36; N, 13.48.
Found: C, 65.46; H, 4.51; N, 13.43.
(E)-1-Styryl-8,9-dihydro-7H-2,7,9a-triaza-benzo[cd]azu-
len-6-one (5h). Yield=75%; mp=300–305 ꢅC; H NMR
(DMSO-d6) d 8.40 (t, 1H), 7.85 (m, 5H), 7.35 (m, 5H),
4.58 (m, 2H), 3.64 (m, 2H). Anal. calcd for C18H15N3O
(0.1H2O) C, 74.26; H, 5.26; N, 14.43. Found: C, 74.17;
H, 5.36; N, 14.38.
1
1-Phenylethynyl-8,9-dihydro-7H-2,7,9a-triaza-benzo[cd]-
azulen-6-one (5i). Yield=75%; mp=261–263 ꢅC; 1H
NMR (DMSO-d6) d 8.46 (t, 1H), 7.96 (d, 1H), 7.90 (d,
1H), 7.76 (d, 2H), 7.55 (m, 3H), 7.41 (t, 1H), 4.56 (m,
2H), 3.68 (m, 2H). Anal. calcd for C18H13N3O (0.3H2O)
C, 73.86; H, 4.68; N, 14.35. Found: C, 73.92; H, 4.67; N,
14.27.
General procedure for the coupling of acid 5p to amines
(Method C). Carboxylic acid 5p (80 mg, 0.26 mmol),
EDC (94 mg, 0.52 mmol), DMAP (5 mg) and the
requisite amine (0.52 mmol) were mixed together in a
solution of DCM/NMP (10:1, 5 mL). The reactions
were agitated overnight. Workup consisted of washing
with water (3 mL) and drying the organic phase through
a plug of sodium sulfate. The crude amides 11a–b were
all isolated by concentrating the organic phase.
1-Pyridin-3-yl-8,9-dihydro-7H-2,7,9a-triaza-benzo[cd]-
azulen-6-one (5j). Yield=54%; mp=276–279 ꢅC; 1H
NMR (DMSO-d6) d 9.06 (s, 1H), 8.77 (d, 1H), 8.49 (t,
1H), 8.29 (d, 1H), 7.95 (d, 1H), 7.91 (d, 1H), 7.64 (t,
1H), 7.39 (t, 1H), 4.49 (m, 2H), 3.56 (m, 2H). Anal.
calcd for C15H12N4O (0.25H2O): C, 67.03; H, 4.69; N,
20.84. Found: C, 67.07; H, 4.70; N, 20.71.
1-[3-(4-Methyl-piperazine-1-carbonyl)-phenyl]-8,9-dihydro-
7H - 2,7,9a - triaza - benzo[cd]azulen - 6 - one (11a).
ꢅ
1
Yield=91%; mp=184–187 C; H NMR (DMSO-d6) d
8.49 (bt, 1H), 7.90 (m, 4H), 7.67 (t, 1H), 7.58 (d, 1H),
7.38 (t, 1H), 4.49 (m, 2H), 3.65 (m, 2H), 3.55 (m, 2H),
3.35 (m, 2H), 2.35 (m, 4H), 2.20 (s, 3H). MS
(ES+)=389.98. Anal. calcd for C22H23N5O2
(0.25H2O): C, 67.10; H, 6.05; N, 17.80. Found: C, 67.01;
H, 6.04; N, 17.84. Salt formation of 11aꢁ HCl was done
by suspending the free base in EtOAc and adding HCl/
Et2O. The precipitate was filtered and washed with
EtOAc to yield the purified salt. 1H NMR (D2O) d
7.75 (m, 3H), 7.60 (m, 3H), 7.32 (t, 1H, J=8 Hz), 4.67
(m, 2H), 4.32 (m, 2H), 3.62 (m, 4H), 3.20 (m, 4H), 2.71
(s, 3H). Anal. calcd for C22H26ClN5O2 (2H2O): C,
57.2; H, 6.10; N, 15.20. Found: C, 57.61; H, 6.03; N,
15.02.
1-Furan-2-yl-8,9-dihydro-7H-2,7,9a-triaza-benzo[cd]-
azulen-6-one (5k). Yield=75%; mp=271–275 ꢅC; 1H
NMR (DMSO-d6) d 8.46 (t, 1H), 8.02 (s, 1H), 7.86 (d,
2H), 7.36 (t, 1H), 7.25 (d, 1H), 6.79 (d, 1H), 4.60 (m,
2H), 3.61 (m, 2H). Anal. calcd for C14H11N3O2
(0.6H2O): C, 63.68; H, 4.66; N, 15.91. Found: C, 63.76;
H, 4.54; N, 15.93.
1-(1H-Pyrrol-2-yl)-8,9-dihydro-7H-2,7,9a-triaza-benzo[-
ꢅ
1
cd]azulen-6-one (5l). Yield=48%; mp=323–330 C; H
NMR (DMSO-d6) d 11.93 (s, 1H), 8.45 (t, 1H), 7.79 (t,

3704
D. Ferraris et al. / Bioorg. Med. Chem. 11 (2003) 3695–3707
N-(2-Dimethylamino-ethyl)-N-methyl-3-(6-oxo-6,7,8,9-
tetrahydro-2,7,9a-triaza-benzo[cd]azulen-1-yl)-benzamide
(11b). Yield=50%; 1H NMR (DMSO-d6) d 8.13 (d,
1H), 8.02 (d, 1H), 7.81 (m, 2H), 7.61 (m, 2H), 7.44 (t,
1H), 7.13 (t, 1H), 4.52 (m, 2H), 3.73 (m, 2H), 3.60 (m,
1H), 3.33 (m, 1H), 3.24 (m, 1H), 3.12 (s, 1.5H), 3.01 (s,
1.5H), 2.42 (m, 2H), 2.27 (s, 6H), 1.67 (m, 2H); MS
(ES+)=406.02. Anal. calcd for C23H27N5O2: C, 57.2;
H, 6.10; N, 15.20. Found: C, 57.61; H, 6.03; N, 15.02.
4-[2-(6-Oxo-6,7,8,9-tetrahydro-2,7,9a-triaza-benzo[cd]-
azulen - 1 - yl) - ethyl] - benzoic acid (5u). Yield=82%;
mp=300–320 ꢅC; 1H NMR (DMSO-d6) d 12.86 (bs,
1H), 8.35 (bt, 1H), 7.87 (d, 2H), 7.81 (m, 2H), 7.43
(d, 2H), 7.28 (t, 1H), 4.27 (m, 2H), 3.53 (m, 2H),
3.21 (m, 4H). Anal. calcd for C19H17N3O (3H2O): C,
3
58.76; H, 4.71; N, 10.82. Found: C, 58.74; H, 4.85;
N, 10.81.
1
HCl salt of 11bꢁ HCl: H NMR (D2O) d 8.14 (t, 1H,
Reduction of ester 5w to form 1-[2-(4-hydroxymethyl-
phenyl)-ethyl]-8,9-dihydro-7H-2,7,9a-triaza-benzo[cd]-
azulen-6-one (12). Ethyl ester 5t (500 mg, 1.38 mmol)
was suspended in THF (20 mL) and cooled to 0 ꢅC.
Lithium aluminum hydride (100 mg, 2.74 mmol) was
added portionwise over the next 30 min. The reaction
mixture was stirred at room temperature overnight. The
reaction was quenched with 10 mL EtOAc and washed
with water (10 mL). The organic layer was partitioned
and the aqueous layer was repeatedly extracted with
EtOAc (4ꢄ10 mL). The combined organics were dried
with Na2SO4 and concentrated in vacuo. The resulting
crude solid was triturated with diethyl ether (10 mL)
and filtered. The resulting solid was characterized as
the alcohol 12. Yield=400 mg (96%); 1H NMR
(DMSO-d6) d 8.32 (bt, 1H), 7.78 (t, 2H), 7.25 (t, 1H),
7.22 (m, 4H), 5.11 (t, 1H), 4.45 (m, 2H), 4.44 (d, 2H),
3.49 (m, 2H), 3.12 (m, 2H), 3.08 (m, 2H); MS
(ES+=322.40). Anal. calcd for C19H19N5O2 (0.75H2O):
C, 68.14; H, 6.08; N, 12.55. Found: C, 68.59; H, 6.08; N,
12.27.
J=8 Hz), 8.07 (t, 1H, J=8 Hz), 8.05 (s, 1H), 7.87 (m,
3H), 7.69 (t, 1H, J=8 Hz), 4.66 (m, 2H), 3.77 (m, 4H),
3.37 (t, 2H, J=8 Hz), 3.17 (s, 3H), 3.05 (s, 3H), 2.90 (s,
3H), 2.27 (m, 2H). Anal. calcd for C23H28ClN5O2
(2.5H2O): C, 56.70; H, 6.81; N, 14.44. Found: C, 56.29;
H, 6.78; N, 14.48.
1-[4-(3-Dimethylamino-propoxy)-phenyl]-8,9-dihydro-7H-
2,7,9a-triaza-benzo[cd]azulen-6-one (5q). Yield=65%;
mp=105–108 ꢅC; 1H NMR (CDCl3) d 8.09 (d, 1H), 8.00
(d, 1H), 7.70 (d, 2H), 7.41 (t, 1H), 7.06 (d, 2H), 6.95 (bt,
1H), 4.50 (m, 2H), 4.11 (t, 2H), 3.73 (m, 2H), 2.52 (t,
2H), 2.30 (s, 6H), 2.02 (m, 2H). Anal. calcd for
C21H24N4O2 (1.5H2O): C, 64.43; H, 6.95; N, 14.31.
1
Found: C, 64.74; H, 6.91; N, 14.15. 5qꢁ HCl: H NMR
(D2O) d 7.69 (d, 1H), 7.65 (d, 1H), 7.29 (t, 1H), 7.25 (d,
2H), 6.68 (d, 2H), 4.50 (m, 2H), 4.05 (m, 2H), 3.64 (m,
2H), 3.24 (m, 2H), 2.88 (s, 6H), 2.12 (m, 2H). Anal.
calcd for C21H24N4O2 (1.0H2O, 1.0HCl): C, 60.73; H,
6.46; N, 13.49. Found: C, 60.80; H, 6.38; N, 13.37.
1-(2-Phenyl-propyl)-8,9-dihydro-7H-2,7,9a-triaza-benzo
[cd]azulen-6-one (5r). Yield=35%; mp=163–167 ꢅC; 1H
NMR (CDCl3) d 8.04 (d, 1H), 7.93 (d, 1H), 7.36 (t, 1H),
7.23 (m, 4H), 7.11 (d, 1H), 6.91 (s, 1H), 3.90 (m, 1H), 3.60
(m, 1H), 3.42 (m, 1H), 3.01 (m, 1H) 1.75 (m, 2H), 1.48 (d,
3H). Anal. calcd for C19H19N3O (0.5 EtOAc): C, 72.18;
H, 6.63; N, 12.03. Found: C, 71.73; H, 6.84; N, 12.15.
1-[2-(4-Chloromethyl-phenyl)-ethyl]-8,9-dihydro-7H-2,7,9a-
triaza-benzo[cd]azulen-6-one (13). The alcohol 12 (350
mg, 1.09 mmol) was added portionwise to a cooled
(0 ꢅC), stirred solution of thionyl chloride (3 mL). After
3 h of stirring, the thionyl chloride was removed in
vacuo and the crude chloride was triturated with diethyl
ether and filtered to yield 295 mg of the crude chloride
(est purity >95%). This material was aminated without
1
(S)-1-(1-Amino-2-phenyl-ethyl)-8,9-dihydro-7H-2,7,9a-
triaza-benzo[cd]azulen-6-one (5s). The Boc protected
amine was made by the general protocol of coupling N-
Boc-l-phenylalanine carboxaldehyde with diamine 10.
Crude yield=70%. The Boc group was deprotected by
refluxing the crude carbamate in HCl/Et2O overnight.
After stirring for 16 h, the solvent was removed and the
residue was triturated with diethyl ether (1.0 M, 5 mL)
and filtered. The crystals were washed several times with
diethyl ether and dried to yield 405 mg (47%).
Mp=155–160 ꢅC (dec.); 1H NMR (DMSO-d6) d 7.93 (d,
1H), 7.85 (d, 1H), 7.40 (t, 2H), 7.13 (m, 3H), 6.89 (d,
1H), 5.01 (m, 1H), 4.15 (m, 2H), 3.48 (m, 2H), 3.21 (t,
2H). Anal. calcd for C18H18N4O HCl (2H2O): C, 51.64;
H, 5.86; N, 13.02. Found: C, 52.09; H, 5.74; N, 12.94.
further purification. H NMR (DMSO-d6) d 8.46 (bt,
1H), 7.88 (m, 2H), 7.35 (m, 4H), 4.74 (s, 2H), 4.30 (bs,
2H), 3.53 (bs, 2H), 3.27 (m, 2H), 3.17 (m, 2H); MS
(ES+=340.30).
Amination of chloride 13 to form 1-[2-(4-pyrrolidin-1-yl-
methyl-phenyl)-ethyl]-8,9-dihydro-7H-2,7,9a-triaza-ben-
zo[cd]azulen-6-one hydrochloride (14ꢁ HCl). The benzyl
chloride (850 mg, 2.5 mmol) was suspended in CH3CN
with pyrrolidine (183 mg, 2.5 mmol) and heated to 60 ꢅC
overnight. After cooling the reaction mixture, the
hydrochloride salt precipitated out and was filtered off
1
to yield 950 mg (92%); H NMR (D2O) d 7.93 (d, 1H),
7.79 (d, 1H), 7.49 (t, 1H), 7.22 (d, 2H), 7.11 (d, 2H),
4.50 (m, 5H), 4.17 (s, 2H), 3.40 (m, 2H), 3.25 (m, 4H),
3.13 (m, 2H), 2.99 (m, 2H), 2.00 (m, 2H), 1.83 (m, 2H);
MS (ES+=375.38). Anal. calcd for C23H27ClN4O1
(4H2O): C, 57.19; H, 7.30; N, 11.60. Found: C, 56.93;
H, 7.46; N, 11.74.
4-[2-(6-Oxo-6,7,8,9-tetrahydro-2,7,9a-triaza-benzo[cd]-
azulen - 1 - yl) - ethyl] - benzoic acid ethyl ester (5t3).
Yield=82%; mp=230–233 ꢅC; H NMR (DMSO-d6) d
1
8.34 (bt, 1H), 7.89 (d, 2H), 7.80 (m, 2H), 7.46 (d, 2H),
7.27 (t, 1H), 4.50 (m, 2H), 4.30 (q, 2H), 3.53 (m, 2H),
3.21 (m, 4H), 1.31 (t, 3H). MS (ES+=364.23). Anal.
calcd for C21H21N3O3 (0.25H2O): C, 68.56; H, 5.89; N,
11.42. Found: C, 68.30; H, 5.84; N, 11.52.
Synthesis of 1-chloromethyl-8,9-dihydro-7H-2,7,9a-triaza-
benzo[cd]azulen-6-one (16). The amine 10 (12.5 g, 70.6
mmol), palladium on carbon (500 mg) and t-butyldi-
methysilyloxyacetaldehyde (15.0 g, 84.7 mmol) were

D. Ferraris et al. / Bioorg. Med. Chem. 11 (2003) 3695–3707
3705
suspended in 500 mL THF and refluxed overnight. The
reaction was monitored by TLC (EtOAc) and after
consumption of the amine (16 h), the palladium was
filtered off and the filtrate was treated with tetra-
butylammonium fluoride (75 mL, 1.0 M in THF). The
solvent was removed and the resulting residue was tri-
turated with 75 mL diethyl ether and 75 mL MeOH
and filtered. The solid was dried and characterized as
the intermediate alcohol 15 (13.6 g, 89%, >95% pur-
[Methyl-(6-oxo-6,7,8,9-tetrahydro-2,7,9a-triaza-benzo
[cd]azulen-1-ylmethyl)-amino]-acetic acid tert-butyl ester
(17d). Yield=70%; mp=180–185 ꢅC; 1H NMR
(DMSO-d6) d 8.37 (t, 1H), 7.82 (m, 2H), 7.29 (t, 1H),
4.48 (m, 2H), 3.94 (s, 2H), 3.58 (m, 2H), 3.26 (s, 2H),
2.30 (s, 3H), 1.38 (s, 9H). Anal. calcd for C18H24N4O3:
C, 62.77; H, 7.02; N, 16.27. Found: C, 62.73; H, 7.05; N,
16.22.
1
ity) H NMR (DMSO-d6) d 8.37 (t, 1H), 7.82 (m, 2H),
(S,S)-1-(5-Benzyl-2,5-diaza-bicyclo[2.2.1]hept-2-ylmethyl)-
8,9-dihydro-7H-2,7,9a-triaza-benzo[cd]azulen-6-one (17e).
Yield=82%; mp=192–197 ꢅC; 1H NMR (CDCl3) d
8.38 (t, 1H), 7.85 (d, 1H), 7.80 (d, 1H), 7.32 (m, 5H),
7.23 (m, 1H), 4.45 (m, 2H), 3.95 (dd, 2H), 3.65 (dd, 2H),
3.64 (m, 2H), 3.25 (m, 2H), 2.80 (m, 2H), 2.59 (m, 4H),
1.67 (m, 2H); MS (ES+)=388.10. Anal. calcd for
C23H25N5O: C, 71.29; H, 6.50; N, 18.07. Found: C,
7.29 (t, 1H), 5.64 (bs, 1H), 4.71 (s, 2H), 4.40 (m, 2H),
3.59 (m, 2H). This compound was chlorinated without
further purification. The alcohol was added portion-
wise to thionyl chloride (25 mL) and stirred overnight.
The thionyl chloride was then removed in vacuo and
the residue was triturated several times with diethyl
ether. The crude solid was recrystallized from acetoni-
1
1
trile (12.2 g, 74% overall yield). H NMR (DMSO-d6)
70.93; H, 6.40; N, 18.03. 17eꢁ 2HCl: H NMR (D2O) d
d 8.44 (t, 1H), 7.92 (d, 1H), 7.87 (d, 1H), 7.36 (t, 1H),
5.08 (s, 2H), 4.42 (m, 2H), 3.63 (m, 2H); MS
(ES+=392.34). Anal. calcd for C11H10ClN3O: C,
56.06; H, 4.28; N, 17.83. Found: C, 56.06; H, 4.27; N,
17.83. The amines 17a–e were made from this com-
pound.
7.98 (d, 1H), 7.89 (d, 1H), 7.52 (t, 1H), 7.45 (m, 5H),
4.43–4.18 (m, 6H), 3.83 (m, 2H), 3.72 (m, 2H), 3.63 (m,
2H), 3.32 (m, 2H), 3.03 (m, 2H), 2.25 (m, 2H). Anal.
calcd for C23H25N5O: C, 60.00; H, 5.91; N, 15.21.
Found: C, 60.04; H, 6.06; N, 15.16.
PARP-1 inhibition assay. Purified recombinant human
PARP from Trevigan (Gaithersburg, MD, USA) was
used to determine the IC50 values of a PARP inhibitor.
The PARP enzyme assay is set up on ice in a volume of
100 mL consisting of 50 mM Tris–HCl (pH 8.0), 2 mM
MgCl2, 30 mg/mL of DNase activated herring sperm
DNA (Sigma, MO, USA), 30 mM [3H]nicotinamide
adenine dinucleotide (67 mCi/mmol), 75 mg/mL PARP
enzyme, and various concentrations of the compounds
to be tested. The reaction was initiated by incubating
the mixture at 25 ꢅC. After 15 min of incubation, the
reaction was terminated by adding 500 mL of ice cold
20% (w/v) trichloroacetic acid. The precipitate formed
is transferred onto a glass fiber filter (Packard Unifilter-
GF/B) and washed three times with ethanol. After the
filter is dried, the radioactivity is determined by scintil-
lation counting.
General procedure for amination of chloride 16. 1-Di-
methylaminomethyl-8,9-dihydro-7H-2,7,9a-triaza-benzo[-
cd]azulen-6-one (17a). The chloride 16 (775 mg, 3.3
mmol) was suspended in THF (20 mL) and dimethyl-
amine (3.3 mL 2.0 M in THF, 6.6 mmol) was added
followed by refluxing for 12 h. The solvent was removed
in vacuo and the resulting solid was triturated with H2O
(10 mL) and filtered to yield the desired amine 17a in
1
84% yield (675 mg). H NMR (DMSO-d6) d 8.37 (t,
1H), 7.85 (d, 1H), 7.81 (d, 1H), 7.29 (d, 1H), 3.69 (s,
2H), 3.59 (m, 2H), 3.34 (s, 6H), 3.32 (m, 2H). Anal.
calcd for C13H16N4O: C, 63.92; H, 6.60; N, 22.93.
Found: C, 63.55; H, 6.62; N, 22.42. The hydrochloride
salt was prepared by suspending the amine in EtOAc
(15 mL) followed by addition of 1.1 equiv HCl/Et2O.
The reaction was stirred for 1 h followed by filtration
and characterization of the HCl salt. 17aꢁ HCl: 1H
NMR (D2O) d 8.02 (m, 2H), 7.51 (t, 1H), 4.82 (s, 6H),
4.76 (s, 2H), 4.50 (m, 2H), 3.81 (m, 2H). Anal. calcd for
C13H16N4O (0.8H2O): C, 52.90; H, 6.35; N, 18.98.
Found: C, 53.11; H, 6.30; N, 19.03.
Methods for molecular modeling
Three PARP-1 crystal structures (PDB designations:
1PAW,37 4PAX,30 1EFY20) were used as templates for
molecular modeling. All modeling was performed using
the Sybyl molecular modeling software (SYBYL 6.7.1
Tripos, Inc., St. Louis, MO 63144, USA). The
MMFF94s forcefield was used for both the protein and
1-[(Benzyl-methyl-amino)-methyl]-8,9-dihydro-7H-2,7,9a-
triaza-benzo[cd]azulen-6-one
(17b).
Yield=72%;
mp=211–215 ꢅC; H NMR (DMSO-d6) d 8.44 (t, 1H),
7.89 (m, 2H), 7.35 (m, 6H), 4.45 (m, 2H), 3.90 (s, 2H),
3.65 (m, 2H), 3.62 (s, 2H), 2.17 (s, 3H); MS
(ES+=321.01). Anal. calcd for C19H20N4O (0.8H2O):
C, 68.3; H, 6.45; N, 16.80. Found: C, 68.77; H, 6.23; N,
16.66.
1
ligand.38 A distance dependent dielectric and 15 A non-
˚
bonded cutoff were employed. The protein structure was
held rigid during all calculations. Initial complex struc-
tures were obtained by overlaying the putative ligand
onto the inhibitor from the 4PAXstructure, requiring
the crucial nicotinamide moiety to be conserved. Multi-
ple conformations of each ligand were evaluated with
each PARP crystal structure. From the initial complex
structure, the ligand was minimized in the presence of
the enzyme. Final structures were evaluated based on
the number and quality of interactions made by the
ligand with the protein. In most cases, the structures
obtained using the 1PAW structure were deemed to be
1-Pyrrolidin-1-ylmethyl-8,9-dihydro-7H-2,7,9a-triaza-
benzo[cd]azulen - 6 - one (17c). Yield=74%; mp=210–
ꢅ
1
215 C; H NMR (DMSO-d6) d 8.37 (bt, 1H), 7.81 (m,
2H), 7.28 (t, 1H), 4.39 (bs, 2H), 3.88 (s, 2H), 3.58 (m,
2H), 3.35 (m, 4H), 1.70 (m, 4H). Anal. calcd for
C15H18N4O (1.0H2O): C, 62.5; H, 7.02; N, 19.49.
Found: C, 62.22; H, 6.95; N, 19.60.

3706
D. Ferraris et al. / Bioorg. Med. Chem. 11 (2003) 3695–3707
superior. The location of the Gln763 sidechain in the
4PAXand 1EFY structures resulted in steric clashes
between the inhibitor and the side chain of Gln763. In
the 1PAW structure, the Gln763 sidechain is in a unique
location resulting from a different conformation than
observed in 4PAXand 1EFY. A small hydrophobic
pocket is formed by the side chains of Val762, Gln763,
Gln759 and Tyr889 in the 1PAW structure.
Student t test with a confidence interval accepted when
pꢆ0.05.
Methods for testing 5g in focal cerebral ischemic stroke
Transient MCAO model. Transient focal ischemic
stroke was modeled in rats using the intraluminal thread
procedure as modified by Belayev.34 Male Sprague–
Dawley rats (280–320 g, Charles River, Wilmington,
MA, USA) were anesthetized with isofluorane. A poly-
lysinized 3–0 suture was inserted through the internal
carotid artery into the origin of the middle cerebral
artery (MCA). The suture is removed after 2 h of MCA
occlusion.
Methods for caco-2 permeability assay39
Caco-2 cells were grown at 37 ꢅC in T-75 flasks in an
atmosphere of 5% CO2 and 95% relative humidity
using Dulbecco’s Modified Eagle’s Medium (pH 7.4)
supplemented with 10% fetal bovine serum, 1% non
essential amino acids, and 0.05% penicillin–streptomy-
cin. Cells were passaged at 80–90% confluency using a
0.25% trypsin/0.20% ethylene diamine tetraacetic acid
(EDTA) solution. Media was changed approximately
every 48 h. Caco-2 cells (passages 30–70) were seeded at
6.3ꢄ104 cells/cm2 on polycarbonate 12-well Transwell1
filters (Corning Costar Corporation, Cambridge, MA,
USA) (3.0 mm mean pore size). Caco-2 cells were cul-
tured on Transwells1 under the same incubation con-
ditions and used for transport experiments 21–28 days
after seeding.
Permanent MCAO model. Permanent focal cerebral
ischemic stroke was modeled in rats using a modifi-
cation of the method described by Chen et al.35 Male
Long Evans rats (280–350 g, Harlan–Sprague–Dawley,
Indianapolis, IN, USA) were anesthetized with iso-
fluorane. The right and left common carotid arteries
and the right distal MCA were exposed and isolated.
The distal MCA was cauterized and the carotids were
transiently occluded for 90 min.41
All drugs were administered 30 min after the start of
MCA occlusion. For both models, the brains were col-
lected into cold PBS at 24 h after ischemia. The brains
were sliced, processed for TTC histochemistry, and the
infarcts were measured using computer-assisted plani-
metry. In the transient paradigm, total infarcts average
350 mm3 including both cortical and subcortical tissue.
In the permanent model, the infarcts were exclusively
cortical and averaged 180 mm3. The effect of drug
treatment is expressed as the percent reduction in infarct
volume compared to coincident vehicle controls. Statis-
tical significance was determined using the Student t-test
with a confidence interval accepted when pꢆ0.05.
The permeability of PARP-1 inhibitors across Caco-2
cell monolayers was investigated (n=3) in both apical to
basolateral (AB) and basolateral to apical (BA) direc-
tions at donor concentrations of 100 mM. Permeability
experiments were conducted at pH 7.0, 37 ꢅC, 5% CO2,
95% relative humidity, and shaking at 50 revolutions
per min (RPM) while maintaining sink conditions.
Samples were collected at 30, 60 and 90 min and ana-
lyzed by HPLC/UV.
Permeability coefficients (Peff) of G0–G4 were calcu-
lated as previously reported. Control transport experi-
ments were also conducted across Transwell1 filters
without Caco-2 cells to determine the filter permeability
(Pfilter). The permeability of Caco-2 cell monolayers
(Pm) was estimated by correcting the effective perme-
ability (Peff) for filter permeability (Pfilter) according to:
Acknowledgements
The authors thank Dr. Takashi Tsukamoto and Mr.
Yao-Sen Ko for helpful comments regarding the manu-
script.
ꢀ1
Pꢀeff1= Pꢀm1+P
.
filter
Streptozotocin-induced diabetes assay
References and Notes
Male mice, (25–30 g, Charles River) were pre-treated
with vehicle (DMSO, 2 mL/kg) or drug at 3, 10, 30, and
100 mg/kg ip, n=5 in each group. After 30 min, acute
hyperglycemia was induced by ip administration of
streptozoticin (250 mg/kg, SIGMA-Aldrich, St. Louis,
MO, USA).40 After 48 h, blood glucose measurements
were taken with a glucose test strip (Accu Check
Advantage, Roche Diagnostics, New York, NY, USA).
Two days after the STZ treatment, the blood glucose
level in vehicle treated rats increases from a normal
average of 140 – 460 ng/dl, a greater than 3-fold
increase. The effect of drug pre-treatment is expressed as
the percentage reduction in glucose level towards nor-
mal from the level measured in the vehicle-treated
group. Statistical significance was determined using the
1. Decker, P.; Isenberg, D.; Muller, S. J. Biol. Chem. 2000,
275, 9043.
2. Herceg, Z.; Wang, Z. Q. Mol. Cell. Biol. 1999, 19, 5124.
3. Yu, S.-W.; Wang, H.; Poitras, M. F.; Coombs, C.; Bowers,
W. J.; Federoff, H. J.; Poirer, G. G.; Dawson, T. M.; Dawson,
V. L. Science 2002, 297, 259.
4. Menissier-de Murcia, J.; Molinete, M.; Gradwohl, G.;
Simonin, F.; de Murcia, G. J. Mol. Biol. 1989, 210, 229.
5. Rolli, V.; O’Farrell, M.; Menissier-de Murcia, J.; de Mur-
cia, G. Biochemistry 1997, 36, 12147.
6. Jeggo, P. A. Curr. Biol. 1998, 8, R49.
7. Griffin, R. J.; Pemberton, L. C.; Rhodes, D.; Bleasdale, C.;
Bowman, K.; Calvert, A. H.; Curtin, N. J.; Durkacz, B. W.;
Newell, D. R.; Porteous, J. K. Anti-Cancer Drug Des. 1995,
10, 507.

D. Ferraris et al. / Bioorg. Med. Chem. 11 (2003) 3695–3707
3707
8. Nagayama, T.; Simon, R. P.; Chen, D.; Henshall, D. C.;
Pei, W.; Stetler, R. A.; Chen, J. J. Neurochem. 2000, 74, 1636.
9. Takahashi, K.; Pieper, A. A.; Croul, S. E.; Zhang, J.; Sny-
der, S. H.; Greenberg, J. H. Brain Res. 1999, 829, 46.
10. Eliasson, M. J.; Sampei, K.; Mandir, A. S.; Hurn, P. D.;
Traystman, R. J.; Bao, J.; Pieper, A.; Wang, Z. Q.; Dawson,
T. M.; Snyder, S. H.; Dawson, V. L. Nat. Med. 1997, 3, 1089.
11. Pieper, A. A.; Walles, T.; Wei, G.; Clements, E. E.; Verma,
A.; Snyder, S. H.; Zweier, J. L. Mol. Med. 2000, 6, 271.
12. Szabo, C.; Dawson, V. L. Trends Pharmacol. Sci. 1998,
19, 287.
Zhang, K. E.; Ekker, A.; Calabrese, C. R.; Curtin, N. J.; Kyle,
S.; Thomas, H. D.; Wang, L.-Z.; Calvert, A. H.; Golding,
B. T.; Griffin, R. J.; Newell, D. R.; Webber, S. E.; Hom-
stomsky, Z. J. Med. Chem. 2002, 45, 4961.
25. Kukla, M. J.; Breslin, H. J.; Pauwels, R.; Fedde, C. L.;
Miranda, M.; Scott, M. K.; Sherrill, R. G.; Raeymaekers, A.;
Van Gelder, J.; Andries, K.; Janssen, M. A.; De Clercq, E.;
Janssen, P. A. J. J. Med. Chem. 1991, 34, 746.
26. Denny, W. A.; Rewcastle, G. W.; Baguley, B. C. J. Med.
Chem. 1990, 33, 814.
27. McDonald, M. C.; Mota-Filipe, H.; Wright, J. A.;
Abdelrahman, M.; Threadgill, M. D.; Thompson, A. S.;
Thiemermann, C. Br. J. Pharmacol. 2000, 130, 843.
28. Geneste, P.; Kamenka, J.-M.; Vidal, Y.; Fauquet, J.-P.;
Morel, M. E.; Muller, P.; Warolin, C. Eur. J. Med. Chem.
1978, 13, 53.
13. Soriano, F. G.; Pacher, P.; Mabley, J.; Liaudet, L.; Szabo,
C. Circ. Res. 2001, 89, 684.
14. Burkart, V.; Wang, Z. Q.; Radons, J.; Heller, B.; Herceg,
Z.; Stingl, L.; Wagner, E. F.; Kolb, H. Nat Med. 1999, 5, 314.
15. Akiyama, T.; Takasawa, S.; Nata, K.; Kobayashi, S.; Abe,
M.; Shervani, N. J.; Ikeda, T.; Nakagawa, K.; Unno, M.; Mat-
suno, S.; Okamoto, H. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 48.
16. Watson, C. Y.; Whish, W. J. D.; Threadgill, M. D. Bioorg.
Med. Chem. 1998, 6, 721.
29. Ruf, A.; Rolli, V.; de Murcia, G.; Schulz, G. E. J. Mol.
Biol. 1998, 278, 57.
30. Ruf, A.; de Murcia, G.; Schulz, G. E. Biochemistry 1998,
37, 3893.
17. Suto, M. J.; Turner, W. R.; Arundel-Suto, C. M.; Werbel,
L. M.; Sebolt-Leopold, J. S. Anti-Cancer Drug Des. 1991, 7,
107.
18. Li, J. H.; Serdyuk, L.; Ferraris, D. V.; Xiao, G.; Tays, K.;
Kletzly, P. W.; Li, W.; Lautar, S.; Zhang, J.; Kalish, V.
Bioorg. Med. Chem. Lett. 2001, 11, 1687.
19. Abdelkarim, G. E.; Gertz, K.; Harms, C.; Katchanov, J.;
Dirnagl, U.; Szabo, C.; Endres, M. Int. J. Mol. Med. 2001, 7,
255.
20. White, A. W.; Almassy, R.; Calvert, A. H.; Curtin, N. J.;
Griffin, R. J.; Hostomsky, Z.; Maegley, K.; Newell, D. R.;
Srinivasan, S.; Golding, B. T. J. Med. Chem. 2000, 43, 4084.
21. Banasik, M.; Komura, H.; Shimoyama, M.; Ueda, K. J.
Biol. Chem. 1992, 267, 1569.
22. Li, H.; Goldstein, B. M. J. Med. Chem. 1992, 35, 3560.
23. Skalitzky, D. J.; Marakovits, J. T.; Maegley, K. A.; Ekker,
A.; Yu, X.-H.; Hostomsky, Z.; Webber, S. E.; Eastman, B. W.;
Almassy, R.; Li, J.; Curtin, N. J.; Newell, D. R.; Calvert, A. H.;
Griffin, R. J.; Golding, B. T. J. Med. Chem. 2003, 46, 210.
24. Canan Koch, S. S.; Thoresen, L. H.; Jayashree, G. T.;
Maegley, K. A.; Almassy, R. J.; Li, J.; Yu, X.-H.; Zook, S. E.;
Kumpf, R. A.; Zhang, C.; Boritzki, T. J.; Mansour, R. N.;
31. Constantino, G.; Macchiarulo, A.; Camaioni, E.; Pelli-
ciari, R. J. Med. Chem. 2001, 44, 3786.
32. Mandagere, A. K.; Thompson, T. N.; Hwang, K.-K. J.
Med. Chem. 2002, 45, 304.
33. Arturrson, P. J. Pharm. Sci. 1990, 79, 476.
34. Belayev, L. Stroke 1996, 27, 1616.
35. Chen, S. T. Stroke 1986, 17, 738.
36. Breslin, H. J.; Kukla, M. J.; Ludovici, D. W.; Mohrba-
cher, R.; Ho, W.; Miranda, M.; Rodgers, J. D.; Hitchens,
T. K.; Leo, G.; Gauthier, D. A.; Ho, C. Y.; Scott, M. K.; De
Clercq, E.; Rauwels, R.; Andries, K.; Janssen, M. A.; Janssen,
P. A. J. J. Med. Chem. 1995, 38, 771.
37. Ruf, A.; de Murcia, J. M.; de Murcia, G. M.; Schultz,
G. E. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 7481.
38. Halgren, T. J. Comp. Chem. 1999, 20, 720.
39. Fleisher, D. Peptide-based Drug Design: Controlling
Transport and Metabolism; American Chemical Society:
Washington, DC, 1995.
40. Tokunaga, H.; Yoneda, Y.; Kuriyama, K. Eur. J. Phar-
macol. 1983, 87, 237.
41. Takahashi, K.; Greenberg, J. H.; Jackson, P.; Maclin, K.;
Zhang, J. J. Cereb. Blood Flow Metab. 1997, 17, 1137.