Potent nonpeptide vasopressin receptor antagonists based on oxazino- and thiazinobenzodiazepine templates

Article abstract of DOI:10.1016/j.bmcl.2004.03.083

Vasopressin receptor antagonists can elicit ion-sparing diuretic effects (i.e., aquaresis) in vivo by blunting the action of the circulating hypophyseal hormone arginine vasopressin. We have identified two new series of basic tricyclic benzodiazepines, represented by general structure 1, which contain compounds that bind with high affinity to human V₂ receptors. For example, (S)-(+)-8 and 5 are potent and selective V₂ receptor antagonists with pronounced aquaretic activity in rats on oral administration.

Full text of DOI:10.1016/j.bmcl.2004.03.083

Bioorganic & Medicinal Chemistry Letters 14 (2004) 2747–2752
Potent nonpeptide vasopressin receptor antagonists
based on oxazino- and thiazinobenzodiazepine templates
Jay M. Matthews,a,* William J. Hoekstra,a,ꢀ Alexey B. Dyatkin,a Leonard R. Hecker,a
Dennis J. Hlasta,a Brenda L. Poulter,a Patricia Andrade-Gordon,a Lawrence de Garavilla,a
Keith T. Demarest,b Eric Ericson,b Joseph W. Gunnet,b William Hageman,a
Richard Look,b John B. Moore,b Charles H. Reynoldsa and Bruce E. Maryanoffa,*
aDrug Discovery, Johnson & Johnson Pharmaceutical Research & Development, Spring House, PA 19477-0776, USA
bDrug Discovery, Johnson & Johnson Pharmaceutical Research & Development, Raritan, NJ 08869-0602, USA
Received 10 February 2004; accepted 25 March 2004
Abstract—Vasopressin receptor antagonists can elicit ion-sparing diuretic effects (i.e., aquaresis) in vivo by blunting the action of the
circulating hypophyseal hormone arginine vasopressin. We have identified two new series of basic tricyclic benzodiazepines, rep-
resented by general structure 1, which contain compounds that bind with high affinity to human V2 receptors. For example, (S)-(+)-
8 and 5 are potent and selective V2 receptor antagonists with pronounced aquaretic activity in rats on oral administration.
ꢀ 2004 Elsevier Ltd. All rights reserved.
1. Introduction
receptor antagonists have received attention for their
potential use in treating diseases of excessive renal
reabsorption of water.2–4 For instance, lixivaptan (VPA-
The nonapeptide arginine vasopressin (AVP), which is
principally secreted from the posterior pituitary gland, is
responsible for numerous biological actions as both a
hormone and a neurotransmitter.1 Three G-protein-
coupled receptors, denoted as V1a, V1b, and V2, are
involved in AVP binding and cellular activation,
resulting in important physiological responses such as
reabsorption of water in the kidneys (V2), contraction of
bladder, uterine, and vascular smooth muscle (V1a),
breakdown of glycogen in the liver (V1a), aggregation of
platelets (V1a), and release of corticotropin from the
anterior pituitary gland (V1b).1 Additionally, in the
central nervous system, AVP modulates aggressive,
social, and sexual behavior, stress response, and mem-
ory.
985),
a pyrrolobenzodiazepine that exhibits low-
nanomolar binding to the V2 receptor and 230-fold
selectivity over V1a receptor binding, exerts an aquaretic
effect on oral administration to rats and dogs (Fig. 1).2a;c
Tricyclic vasopressin receptor antagonists of this struc-
tural type2–4 have physical properties that can present
difficulties for oral drug development, such as elevated
molecular weight (450–600 Da), high hydrophobicity
(log P > 4), and limited aqueous solubility (<1 mg/mL).5
Thus, we became interested in introducing a basic amine
center into the tricyclic molecular framework of lixi-
vaptan in the context of thiomorpholine and morpholine
N
The V2 receptors on renal epithelial cells mediate AVP-
induced antidiuresis to preserve normal plasma osmo-
lality. Thus, selective, nonpeptide vasopressin V2
N
Z
R
1
N
N
Me
F
O
O
R
O
O
3
N
H
Cl
N
H
R
2
* Corresponding authors. Tel.: +1-215-628-5530; fax: +1-215-628-4985
(B.E.M.); tel.: +1-215-628-5214; fax: +1-215-628-3297 (J.M.M.);
e-mail addresses: bmaryano@prdus.jnj.com; jmatthew@prdus.jnj.com
R
4
Lixivaptan (VPA-985)
1 (X = O or S)
ꢀ
Present address: GlaxoSmithKline, R&D, 5 Moore Drive, PO Box
13398, Research Triangle Park, NC 27709, USA.
Figure 1. Structures of lixivaptan and benzodiazepine 1.
0960-894X/$ - see front matter ꢀ 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.03.083

2748
J. M. Matthews et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2747–2752
rings fused to the benzodiazepine framework, as in
formula 1 (Fig. 1). It was hoped that these novel
chemical entities would give rise to receptor antagonists
with good in vitro and in vivo potency, and favorable
oral bioavailability. In this paper, we report on the
biological properties of compounds of general formula
1, some of which represent advanced V2 receptor
antagonist leads, such as 5 and (S)-(+)-8.
N
S
a, b, c
N
N
S
O
N
H
3
O
Ph
Cl
N
H
2
Scheme 1. Reagents and conditions: (a) 2-Cl–4-NO2C6H4C(O)Cl,
Et3N, CH2Cl2 (90%); (b) Zn dust, NH4Cl, MeOH (90%); (c) 2-
PhC6H4C(O)Cl, Et3N, CH2Cl2 (55%).
2. Synthetic chemistry
obtained by fractional crystallization of the (S)-())-
binaphthyl-2,20-diyl hydrogen phosphate salt in etha-
nol,6 and the free base was carried through the
remainder of the route in Scheme 1. We generated des-
halogen analogues (R)-())-8 and (S)-(+)-8 (Table 1) in
an analogous fashion.8 The other analogues in Table 1
were prepared from tricycle 3, or its congeners, using
similar chemistry. The oxazinobenzodiazepines (Table
2) were accessed from appropriate tricycles, as illus-
The thiazinobenzodiazepines (Table 1) were obtained by
reacting the appropriate tricyclic secondary amine,6 as
illustrated for target 2 in Scheme 1. Racemic 3 was
acylated with 2-chloro-4-nitrobenzoyl chloride and the
intermediate amide was reduced with zinc dust and
ammonium chloride. Acylation of the resulting aniline
with 2-phenylbenzoyl chloride yielded 2. To prepare the
(R)-()) enantiomer of 2, (R)-())-37 (>97% ee) was
Table 1. Vasopressin V1a and V2 binding and functional data for thiazinobenzodiazepines
N
S
R
1
N
2
O
R
3
O
R
N
R
4
H
Compda
R1
R2
R3
R4
V1a bndgb Ki, nM V2 bndgb Ki, nM
V1a functc Ki, lM V2 functc Ki, lM
6
H
Cl
Cl
H
Me
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
5-F
5-F
H
IA
IA
60
37
14
14
0.3
7
H
0.07
0.02
0.015
0.04
0.04
8
H
49
5.0
3.2
25
8.0
(+)-8d;e
H
H
H
84
290
IA
1.4
1.3
1.4
())-8e;f
H
H
H
9g
H
H
H
65
11
2
H
Cl
Cl
Cl
F
H
490
250
410
IA
(+)-2e;h
10
H
H
3.7
3.7
10
14
0.017
H
4-F
H
11
12
H
H
Me
OMe
OH
H
H
IA
IA
10
15
13
14
H
H
H
H
110
IA
3.2
10
8.0
9.0
18
2.3
15
16
9-Cl
8-Me
8-F
8,9-F2
H
H
H
18
14
0.023
0.023
17
18
H
H
IA
IA
H
H
VPA-985i
44
6.0
a Target compounds were purified by reverse-phase semi-prep HPLC and isolated as trifluoroacetate salts, unless noted otherwise. Purities were
judged by reverse-phase HPLC/MS at 215 and 254 nm (YMC J’Sphere C-18 column, 0.4 · 5 cm; mobile phase: MeCN–H2O). All compounds were
characterized by ESI-MS; selected compounds were analyzed by 300-MHz 1H NMR. Compounds are racemates unless noted otherwise.
b Inhibition of [3H]-AVP binding to recombinant human vasopressin V1a or V2 receptors (N ¼ 1–3). IA ¼ inactive, that is, <30% inhibition of
radioligand binding at a concentration of 100 nM.
c Inhibition of AVP-induced effects on cells expressing either human V1a or human V2 receptors. Ki values were determined by using cells treated with
1 nM AVP to stimulate calcium mobilization for V1a receptors or cAMP accumulation for V2 receptors (N ¼ 1–4).
25
d (S)-(+) enantiomer; ½aŠ +583.3 (c 0.03, MeOH).
D
e HCl salt.
25
D
f (R)-()) enantiomer; ½aŠ )1045 (c 0.06, MeOH).
g S,S-Dioxide (i.e., a sulfone).
25
D
h (S)-(+) enantiomer; ½aŠ +173.4 (c 0.15, MeOH).
i Reference standard. IC50 values (left to right): 150, 5.0 nM; 12.5, 0.091 lM.

J. M. Matthews et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2747–2752
Table 2. Vasopressin V1a and V2 binding and functional data for oxazinobenzodiazepines
2749
N
O
H
N
2
O
O
R
3
R
N
R
H
4
Compda
R2
R3
R4
V1a bndgb Ki, nM
V2 bndgb Ki, nM
V1a functc Ki, lM
V2 functc Ki, lM
19
Cl
Cl
H
Me
Ph
5-F
5-F
H
100
ca. 300d
ca. 30f
24
2.8
11
2.0
6.4
0.012
0.012
0.016
0.004
0.17
20
21e
5g
Ph
Ph
3.7
0.90
15
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
H
0.60
14
22h
Ph
H
640
IA
23
24
4-MeOPh
3-MeOPh
4-OH–Ph
3-OH–Ph
Ph
H
2.8
3.7
4.2
5.0
1.9
4.2
17
1.9
3.4
4.2
11
0.003
0.002
0.011
0.02
H
IA
IA
25
26
H
H
IA
IA
27
28
4-F
4-OMe
5-OMe
4-OH
5-OH
0.70
1.3
0.002
0.005
Ph
Ph
IA
IA
29
30
Ph
Ph
ca. 30
IA
44
1.4
13
1.6
1.9
6.0
0.006
0.07
0.023
31
VPA-985i
2.3
a Same as in Table 1, except compounds are (S)-(+) enantiomers (as depicted in the structural formula), unless noted otherwise.
b Same as in Table 1.
c Same as in Table 1, except N ¼ 1–9.
d 62% inhibition @ 1000 nM.
e Racemic mixture.
f 69% inhibition @ 100 nM.
25
D
g C32H28ClN3O3ÆHClÆ1.3H2O (correct microanalysis for C/H/N/H2O); mp 210 ꢁC (dec), ½aŠ +215.5 (c 0.28, MeOH); 98.7% enantiomeric purity by
chiral HPLC (Chiralcel AS column, 0.46 · 5 cm; mobile phase: 90:10 hexanes/i-PrOH, with 0.1% Et2NH).
25
h HCl salt; (R)-()) enantiomer related to 5, ½aŠ )200.7 (c 0.28, MeOH).
D
i Reference standard. IC50 values (left to right): 150, 5.0 nM; 12.5, 0.091 lM.
analogues (Table 2) were prepared from (S)-(+)-4 in a
similar manner.
N
N
O
O
a
H
N
H
N
H
4
(S)-(+)-4
3. Results and discussion
N
O
b, c, d
H
The thiazinobenzodiazepines were evaluated for their
binding to human V1a and V2 receptors (Table 1).12 The
direct analogue of VPA-985, 6, exhibited weaker binding
to both receptors than the reference agent, although the
V2 Ki value of 60 nM is still respectable. Introduction of
an o-phenyl group, as in 7, enhanced V2 receptor affinity
(Ki ¼ 37 nM), with good V2 selectivity. Thus, we adop-
ted the o-phenyl substituent for other analogues.13 Des-
N
O
O
Ph
Cl
N
H
5
Scheme 2. Reagents and conditions: (a) Ref. 6 (35%); (b) 2-Cl–4-
NO2C6H4C(O)Cl, Et3N, CH2Cl2 (52%); (c) Zn dust, NH4Cl, MeOH
(97%); (d) 2-PhC6H4C(O)Cl, Et3N, CH2Cl2 (52%).
halogen parent
8
showed notable V2 affinity
(Ki ¼ 5:0 nM) with modest selectivity (V1a/V2 ¼ 10).
Each of its enantiomers, (S)-(+)-8 and (R)-())-8, sur-
prisingly has significant V2 receptor affinity (Ki ¼ 3:2
and 25 nM, respectively). The mere 8-fold difference in
V2 affinity for these two enantiomers suggests that the
geometry around this portion of the ligand is not very
critical for V2 binding interactions. The excellent V2
affinity of (S)-(+)-8, coupled with the 26-fold selectivity
for V2 over V1a, was viewed as encouraging. Subsequent
variation of R2 on the 4-aminobenzamide ring provided
trated in Scheme 2 for the (S)-(+) enantiomeric series.
We prepared tricycle 4 according to literature proce-
dures.6;9 To obtain 5, and the corresponding (R)-())
enantiomer, 22 (Table 2), we resolved racemate 4 into its
enantiomers by fractional crystallization through acid-
addition salts involving the opposite enantiomers of di-
p-toluoyltartaric acid (methanol-ether).6;10 Other chiral

2750
J. M. Matthews et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2747–2752
potent, reasonably V2-selective analogues (cf. 8 with 2
and 10–14). Overall, notable V2 affinities were obtained
with (S)-(+)-2 (Ki ¼ 3:7 nM), (S)-(+)-8 (Ki ¼ 3:2 nM), 10
(Ki ¼ 3:7 nM), and 14 (Ki ¼ 3:2 nM). S,S-Dioxide 9 had
a 13-fold lower affinity than that for sulfide 8. The
results for 15–18 indicate that R1 substitution is well
tolerated, although this modification can enhance V1a
affinity, as observed for 16 (V1a Ki ¼ 18 nM).
A dose of just 1 mg/kg, po, caused a 700% increase of
urine output (N ¼ 10) over untreated controls (N ¼ 18),
with a 60% reduction of urine osmolality.15
The structure of 5, as a tosylate salt, was determined by
X-ray diffraction (Crystalytics).16 The tricycle adopts a
chair-like conformation for the seven-membered ring
with its pendant amide carbonyl (C13) in an axial ori-
entation. The carbonyl oxygen of this amide, O2, is anti
to the fused benzene ring (Fig. 2). This arrangement is
analogous to that observed for related N-acyl-
tetrahydrobenzazepines in solution.17 We carried out a
Monte Carlo conformational search on protonated 5
with the OPLS-AA force field and GB/SA water
model.18 The global energy minimum contains a trans-
fused oxazinobenzodiazepine with a chair-like seven-
membered ring, an axial pendant amide, and an amide
carbonyl anti to the fused benzene ring (Fig. 3). The
To follow up on these results, we studied several com-
pounds in cell-based functional assays involving human
V1a and V2 receptors.14 Compounds (+)-2, (S)-(+)-8,
(R)-())-8, 9, 16, and VPA-985 potently antagonized the
effects of AVP on human V2 receptors (Ki ¼
0:015–0:04 lM), whereas they had just weak potency
against human V1a receptors (Ki ¼ 1:3–14 lM).14 The
reduced V2 functional potency for 6 (Ki ¼ 300 nM)
versus VPA-985 is consistent with the V2 binding result.
Enantiomer (S)-(+)-28 gave functional V1a and V2 Ki
values of 1400 and 15 nM, for ca. 90-fold V2 selectivity,
and (S)-(+)-30 gave functional V1a and V2 Ki values of
14,000 and 17 nM, for an impressive V2 selectivity of ca.
820-fold. The potency of each enantiomer of 28 in the V2
receptor functional assay (Ki ¼ 15–40 nM) is reasonably
consistent with the binding data.
For the oxazinobenzodiazepines, we generally studied
the (S)-(+) enantiomeric series because it has better V2
receptor binding (Table 2). In comparing 5 and 22, each
enantiomer has a significant V2 affinity (Ki ¼ 0:9 and
15 nM, respectively), but there is a 17-fold preference for
5. Interestingly, 5 also has high affinity for the V1a
receptor (Ki ¼ 24 nM), which accounts for a 26-fold V2
binding selectivity. The direct VPA-985 analogue, 19,
exhibited excellent V2 receptor affinity, as well as modest
V1a affinity, in contradistinction to related thi-
azinobenzodiazepine 6. Given our success with the
o-phenyl group for the thiazinobenzodiazepine series,
we incorporated it here, as well.13 In general, this sub-
stitution resulted in good-to-excellent V2 affinity. In-
deed, several oxazino analogues had single-digit
nanomolar V2 receptor binding with good selectivity
versus V1a, such as 5, 21, 23–25, 27, 28, and 30. The most
potent compounds possessed 2-phenyl (5, V2
Ki ¼ 0:9 nM), 2-phenyl-4-fluoro (27, V2 Ki ¼ 1:9 nM),
and 2-phenyl-4-hydroxy (30, V2 Ki ¼ 1:4 nM) groups. In
the V2 functional assay, 5, 19, 20, 23–28, and 30 were
very potent in antagonizing the effects of AVP
(Ki ¼ 0:002–0.02 lM), whereas they were weak in the
V1a functional assay (Ki ¼ 0:6–14 lM).14 In the cell-
based assays, 5 exhibited Ki values of 600 and 4 nM for
V1a and V2, reflecting 150-fold V2 selectivity. The (R)-())
enantiomer 22 showed comparatively less functional
potency than did 5 (Ki ¼ 14; 000 and 170 nM), in con-
trast to (S)-(+)-8 and (R)-())-8.
Figure 2. Perspective drawing of the solid-state structure of 5ÁTsOH
[(S)-(+) isomer], showing the cationic subunit, with its atom-number-
ing scheme (standard atom color code; H ¼ cyan).
Figure 3. Structure of the global minimum-energy conformation of
protonated 5 (standard atom color code).
Oral administration of (S)-(+)-8 to Sprague-Dawley rats
elicited a dose-dependent aquaretic effect. At a dose of
10 mg/kg, po, urine output (N ¼ 8) was increased 300%
over untreated controls (N ¼ 8), with a reduction in
urine osmolality of 70%.15 Oral administration of 5 to
Sprague-Dawley rats produced
aquaretic effect with remarkable potency.
a
dose-dependent
Figure 4. Structure of the next-higher-energy conformation of pro-
tonated 5 (+1.7 kcal/mol) (standard atom color code).

J. M. Matthews et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2747–2752
2751
Trybulski, E. J.; Williams, D. K. Bioorg. Med. Chem. Lett.
2000, 10, 783.
tricyclic nucleus in this structure was closely super-
imposable on that of the X-ray structure. The next-
higher-energy structure is 1.7 kcal/mol (7.1 kJ/mol) less
stable than the global minimum. It has a cis-fused tri-
cycle, resulting from inversion of the ammonium center
(NHþ), a chair-like seven-membered ring, an axial
amide, and an amide carbonyl anti to the fused benzene
(Fig. 4). This structural information could be helpful in
developing a pharmacophore model for V2 receptor
binding.19
3. (a) Ogawa, H.; Yamashita, H.; Kondo, K.; Yamamura,
Y.; Miyamoto, H.; Kan, K.; Kitano, K.; Tanaka, M.;
Nakaya, K.; Nakamura, S.; Mori, T.; Tominaga, M.;
Yabuuchi, Y. J. Med. Chem. 1996, 39, 3547; (b) Matsuh-
isa, A.; Taniguchi, N.; Koshio, H.; Yatsu, T.; Tanaka, A.
Chem. Pharm. Bull. 2000, 48, 21.
4. Trybulski, E. J. Annu. Rep. Med. Chem. 2001, 36, 159.
5. (a) Oprea, T. I. J. Comput.-Aided Mol. Design 2000, 14,
251; (b) Walters, W.; Patrick, A.; Murcko, M. A. Curr.
Opin. Chem. Biol. 1999, 3, 384.
6. Matthews, J. M.; Dyatkin, A. B.; Evangelisto, M.;
Gauthier, D. A.; Hecker, L. R.; Hoekstra, W. J.; Poulter,
B. L.; Maryanoff, B. E. Tetrahedron: Asymmetry 2004, 15,
in press.
4. Conclusion
7. The absolute configuration of intermediate (R)-())-3 was
assigned by X-ray crystallography of its (S)-())-binaph-
thyl-2,20-diyl hydrogen phosphate salt.6
8. (S)-(+)-8 was derived from (S)-(+)-3 (95% ee), which was
obtained by fractional crystallization of its (R)-(+)-bi-
naphthyl-2,20-diyl hydrogen phosphate salt (from EtOH).6
9. Kogami, Y.; Okawa, K. Bull. Chem. Soc. Jpn. 1987, 60,
2963.
10. The absolute configuration of intermediate (R)-())-4 was
initially assigned by comparing the sign of its optical
rotation to that of (R)-())-3. This assignment was
confirmed by a vibrational circular dichroism (VCD)
study on (S)-(+)-4.11
11. Dyatkin, A. B.; Freedman, T. B.; Cao, X.; Dukor, R. K.;
Maryanoff, B. E.; Maryanoff, C. A.; Matthews, J. M.;
Shah, R. D.; Nafie, L. A. Chirality 2002, 14, 215.
12. Receptor binding studies were performed by using
recombinant human V1a or V2 receptor preparations
derived from the membranes of transfected HEK-293
cells. Compounds were evaluated for their ability to
displace [3H]-AVP.
We have identified two noteworthy series of nonpeptide
vasopressin receptor antagonists containing thiazino-
and oxazinobenzodiazepine core structures. There were
several analogues with low-nanomolar V2 receptor
affinity and at least 20-fold selectivity for V2 over V1a
receptors. In the thiazino class, (S)-(+)-8 has excellent
V2 affinity (Ki ¼ 3:2 nM), moderate binding selectivity
(V1a/V2 ¼ 26), good functional selectivity (V1a/V2 ¼ 93),
and oral efficacy as an aquaretic agent in rats. Also, (S)-
(+)-2 has excellent V2 affinity (Ki ¼ 3:7 nM) and notable
V2 selectivity (binding V1a/V2 ¼ 68; functional V1a/
V2 ¼ 823). In the oxazino class, (S)-(+) enantiomer 5 has
excellent V2 receptor affinity (Ki ¼ 0:9 nM), moderate
binding selectivity (V1a/V2 ¼ 27), good functional selec-
tivity (V1a/V2 ¼ 150), and impressive oral potency as an
aquaretic agent in rats. Specific compounds from these
two novel series have potential for the treatment of
edematous conditions in patients. On the basis of an
extensive array of pre-clinical data, oxazinobenzo-
diazepine 5 was advanced into human clinical studies.20
13. An o-phenylbenzoyl group is present in the vasopressin
receptor antagonist conivaptan.3b The o-phenylbenzoyl
analogue of lixivaptan was reported to have potent human
V2 receptor binding (IC50 ¼ 2.7 nM), with 140-fold selec-
tivity relative to human V1a binding.2d
14. Inhibition of receptor activation caused by AVP was
quantitated in HEK-293 cells expressing human V1a or V2
receptors; changes in intracellular Ca2þ (V1a) or cAMP
(V2) concentrations were measured.
Acknowledgements
We are grateful to Diane Gauthier and Patrick Sasso for
chemical and biological support, and thank Ralph A.
Rivero for helpful discussions and support.
15. For a comparison, at doses of 10 and 1 mg/kg (po), OPC-
31260 is reported to give the following values for urine
output/osmolality, respectively: +500%/)65% and 0%/
)10%;3a lixivaptan is reported to give the following values
for urine output/osmolality: +450%/)70% and +200%/
)50%.2a
References and notes
16. Single crystals of 5ÆTsOH, [C32H29N3O3Cl][C7H7O3S],
1. (a) Albright, J. D.; Chan, P. S. Curr. Pharm. Design 1997,
3, 615; (b) Thibonnier, M.; Conarty, D. M.; Preston, J. A.;
Wilkins, P. L.; Berti-Mattera, L. N.; Mattera, R. Adv.
Exp. Med. Biol. 1998, 449, 251.
from EtOAc/MeOH (mp > 240 ꢁC), are (at )80 2 ꢁC)
2
monoclinic [space group P21)C2
(No. 4)] with
ꢀ
ꢀ
ꢀ
3
a ¼ 8:001ð1Þ A,
b ¼ 18:231ð2Þ A,
c ¼ 12:067ð1Þ A,
ꢀ
b ¼ 91:395ð2Þꢁ, V ¼ 1759:7ð3Þ A , and Z ¼ 2 formula
units [dcalcd ¼ 1:340 g cmꢀ3; la(MoKa ¼ 0:220 mmꢀ1]. A
full hemisphere of diffracted intensities (x-scan
width ¼ 0.30ꢁ) was measured by using graphite-mono-
2. (a) Albright, J. D.; Reich, M. F.; Santos, E. G. D.; Dusza,
J. P.; Sum, F. W.; Venkatesan, A. M.; Coupet, J.; Chan, P.
S.; Ru, X.; Mazandarani, H.; Bailey, T. J. Med. Chem.
1998, 41, 2442; (b) Aranapakam, V.; Albright, J. D.;
Grosu, G. T.; Chan, P. S.; Coupet, J.; Saunders, T.; Ru,
X.; Mazandarani, H. Bioorg. Med. Chem. Lett. 1999, 9,
1733; (c) Martinez-Castelao, A. Curr. Opin. Invest. Drugs
2001, 2, 525; (d) Albright, J. D.; Santos, E. G. D.; Dusza,
J. P.; Chan, P. S.; Coupet, J.; Ru, X.; Mazandarani, H.
Bioorg. Med. Chem. Lett. 2000, 10, 695; (e) Ashwell, M.
A.; Bagli, J. F.; Caggiano, T. J.; Chan, P. S.; Molinari, A.
J.; Palka, C.; Park, C. H.; Rogers, J. F.; Sherman, M.;
ꢀ
chromated MoKa radiation (k ¼ 0:71073 A) on a Bruker
SMART CCD Single Crystal Diffraction System. Lattice
constants were determined with the Bruker SAINT
software by using the peak centers for 2093 reflections.
A total of 18,635 integrated intensities with 2h(MoKa)
<61.01ꢁ were produced by using SAINT, 9886 of which
were independent and gave Rint ¼ 0:056. The Bruker
SHELXTL-PC software package (Version 5) was used to
solve the structure via ‘direct methods’ techniques. All

2752
J. M. Matthews et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2747–2752
stages of weighted full-matrix least-squares refinement
parameters of the remaining hydrogens were fixed at
values 1.2 (nonmethyl) or 1.5 (methyl) times the equivalent
isotropic thermal parameter of the attached carbon atom.
Atomic coordinates for this structure are available from
the authors on request.
were conducted by using Fo2 data. The final agreement
factors at convergence are: R1(unweighted, based on
F ) ¼ 0.054 for 5039 independent reflections with
2h(MoKa) <61.01ꢁ and I > 2rðIÞ; R1(unweighted, based
on F ) ¼ 0.109 and wR2 (weighted, based on F 2) ¼ 0.078 for
all 9886 independent reflections with 2h(MoKa) <61.01ꢁ.
The absolute configuration of the cation was confirmed as
S by using the anomalous dispersion technique,10 with the
‘Flack’ absolute structure parameter for this solution
refined to a final value of )0.09(5). The structural model
incorporated anisotropic thermal parameters for all non-
hydrogen atoms and isotropic thermal parameters for all
hydrogen atoms. Amine and amide hydrogen atoms (H2N
and H3N) were located from a difference Fourier synthesis
and refined as independent isotropic atoms. The tosylate
methyl group (C47 + 3H’s; not shown) was refined as a
rigid rotor (idealized sp3-hybridization and C–H bond
17. Hassner, A.; Amit, B.; Marks, V.; Gottlieb, H. E. J. Org.
Chem. 2003, 68, 6853.
18. (a) Maestro, 5.1 ed., Schrodinger Inc. (Portland, OR),
2003; (b) Jorgensen, W. L.; Tirado-Rives, J. J. Am. Chem.
Soc. 1988, 110, 1657; (c) Qui, D.; Shenkin, P. S.; Hollinger,
F. P.; Still, W. C. J. Phys. Chem. A 1997, 101, 3005; (d)
Chang, G.; Guida, W. C.; Still, W. C. J. Am. Chem. Soc.
1989, 111, 4379.
19. For some molecular modeling studies, see: Gieldon, A.;
Kazmierkiewicz, R.; Slusarz, R.; Ciarkowski, J. J. Com-
put.-Aided Mol. Design 2001, 15, 1085.
20. (a) For comparative purposes, here are some molecular
parameters for
5
and lixivaptan. Compound 5:
ꢀ
length of 0.98 A) that was allowed to rotate about its C–C
bond in least-squares cycles. The remaining hydrogen
atoms were included in the structure factor calculations as
MW ¼ 538 Da;
log P ¼ 3:63;
log D
(pH 3) ¼ 2.01;
pKa ¼ 4:75=12:7. Lixivaptan: MW ¼ 474 Da; calcd
log P ¼ 6:1; calcd pKa ¼ 11.9 (SciFinder, Advanced Chem-
istry Development software, Solaris v4.67); (b) Compound
5 showed good pharmacokinetics in rats and dogs. In rats
(30 mg/kg, po, 3 mg/kg, iv; N ¼ 3), the oral bioavailability
was very good (F ¼ 68%), with an oral t1=2 of 3.7 h.
3
idealized atoms (sp2-or sp -hybridized carbon atoms and
ꢀ
C–H bond lengths of 0.95–1.00 A). The isotropic thermal
parameters for H2N and H3N refined to final Uiso values of
2
ꢀ
0.03(1) and 0.02(1) A , respectively. The isotropic thermal