SAR development of a series of 8-azabicyclo[3.2.1]octan-3-yloxy-benzamides as kappa opioid receptor antagonists. Part 2

Article abstract of DOI:10.1016/j.bmcl.2010.07.112

Further structure activity relationship studies on a previously reported 8-azabicyclo[3.2.1]octan-3-yloxy-benzamide series of potent and selective kappa opioid receptor antagonists is discussed. Modification of the pendant N-substitution to include a cycl

Full text of DOI:10.1016/j.bmcl.2010.07.112

Bioorganic & Medicinal Chemistry Letters 20 (2010) 5405–5410
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
SAR development of a series of 8-azabicyclo[3.2.1]octan-3-yloxy-benzamides
as kappa opioid receptor antagonists. Part 2
*
Todd A. Brugel , Reed W. Smith, Michael Balestra, Christopher Becker, Thalia Daniels, Gerard M. Koether,
Scott R. Throner, Laura M. Panko, Dean G. Brown, Ruifeng Liu, John Gordon, Matthew F. Peters
CNS Discovery Research, AstraZeneca Pharmaceuticals, 1800 Concord Pike, Wilmington, DE 19850, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Further structure activity relationship studies on a previously reported 8-azabicyclo[3.2.1]octan-3-yloxy-
benzamide series of potent and selective kappa opioid receptor antagonists is discussed. Modification of
the pendant N-substitution to include a cyclohexylurea moiety produced analogs with greater in vitro opi-
oid and hERG selectivity such as 12 (j IC50 = 172 nM, l:j ratio = 93, d:j ratio = >174, hERG IC50 = >33 lM).
Changes to the linker conformation and identity as well as to the benzamide ring moiety were also
Received 18 June 2010
Revised 21 July 2010
Accepted 26 July 2010
Available online 3 August 2010
investigated.
Keywords:
Ó 2010 Published by Elsevier Ltd.
Kappa opioid antagonist
Depression
The opioid family of G-protein coupled receptors (GPCRs) is com-
prised of kappa ( ), mu ( ), delta (d), and opioid-like receptor (ORL-
1). The chemistry and biology of the natural and synthetic ligands of
tency. Starting with 8-azabicyclo[3.2.1]octan-3-yloxy-benzonitrile
4,7 hydration of the nitrile to the primary amide was followed by
reductive amination with either trans-Boc-4-aminocyclohexane-
carbaldehyde 5 or trans-Boc-2-(4-aminocyclohexyl)acetaldehyde
6 to provide the corresponding trans-Boc-4-aminocyclohexyl ana-
logs 7 and 8 (Scheme 1). Deprotection of the carbamate group with
saturated ethanolic hydrogen chloride followed by reaction of the
primary amines with dimethylcarbamoyl chloride gave the desired
trans-ureas 9 and 10.8
j
l
these receptors have been extensively studied by the scientific and
medical community.1 In particular, the role of
j opioid receptor li-
gands in the modulation of mood has attracted great interest in re-
cent years.2 Originally developed as ‘molecular probes’ to study
the behavioral effects of
j agonism, selective j antagonists such as
nor-BNI (1,Fig. 1) have been shown to possess potential anti-depres-
sant activity in preclinical animal studies.3
Final compounds were tested for their ability to inhibit agonist-
Our interest in developing potential new treatments for major
depressive disorder (MDD) led to the recent identification of a series
stimulatedactivation of G-proteins using a [35S]GTP
cS bindingassay
in HEK membranes containing each of the cloned human opioid
of potent and selective
j antagonists based on an 8-azabicy-
receptor subtypes.4 No measureable levels of agonism were ob-
clo[3.2.1]octan-3-yloxy-benzamide chemotype.4 The low molecular
served at either j or l for analogs described below.
weight, highly CNS penetrant thiophene analog 2 (Figure 1) showed
potent in vitro
in vivo activity through reversal of
InthisLetter, wewishto reportonadditionalstructure–activity rela-
tionship studies of this series of antagonists.
j
antagonism ([35S]GTP
cS IC50 = 20 nM) as well as
j
agonist induced rat diuresis.
j
An issue that was recognized early on with analog 2 and other
related derivatives was their potent in vitro inhibition of the hERG
ion channel.5 Methylthiophene 2 was found to inhibit hERG with
an IC50 of 260 nM (Table 1). Non-aromatic cyclohexylmethyl ana-
log 3 showed reduced hERG activity (IC50 = 2.0
lM), but with a
similar loss in antagonism. We were aware from another internal
j
project that urea substitution of cyclohexyl moieties was a plausi-
ble way to improve hERG selectivity while maintaining target po-
Figure 1. Selective kappa opioid antagonists nor-BNI
clo[3.2.1]octan-3-yloxy-benzamide 2 (in-house data).
1
and 8-azabicy-
* Corresponding author. Tel.: +1 302 885 5759; fax: +1 302 885 1806.
E-mail address: todd.brugel@astrazeneca.com (T.A. Brugel).
0960-894X/$ - see front matter Ó 2010 Published by Elsevier Ltd.
doi:10.1016/j.bmcl.2010.07.112

5406
T. A. Brugel et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5405–5410
As summarized in Table 1, the methylene linked trans-cyclo-
antagonism relative to 3,
hexylurea 9 showed a ꢀ6-fold loss in
j
while the ethylene linked trans-cyclohexylurea 10 was more simi-
lar in potency to 3 (451 nM and 204 nM, respectively). Both urea
derivatives were inactive at hERG and showed high
l and d selec-
tivity. The corresponding cis-4-cyclohexylurea analogs 11 and 12
were synthesized according to the procedure outlined in Scheme
1. Once again, the methylene-linked derivative 11 showed reduced
j
antagonism (20-fold relative to 3). Conversely, ethylene linked
cis-4-cyclohexylurea analog 12 maintained the same level of po-
tency as 3, but with improved selectivity relative to (93-fold)
and hERG (IC50 >33 M). Unfortunately, this also resulted in re-
j
l
l
duced in vitro permeability (vide infra). The improved opioid selec-
tivity observed with this extended N-substitution could be
attributed to a better fit into the ‘message’ and ‘address’ sites of
the
j
binding pocket.9
We next sought to understand the effects of modification to the
linker atom between the benzamide ring and azabicyclooctane
core. Initially, we investigated a change in the linker stereochemis-
try at C(3) of the azabicyclooctane core from endo to exo. Compared
to the parent endo analog 13, the exo derivative 14 (Table 2)
showed a 34-fold decrease in
j antagonism, suggesting that this
conformation change impaired the ability of exo-14 to properly
associate with the receptor.10
In addition to differences brought about by endo and exo stereo-
chemistry changes at the C(3) position, we were also interested in
modifying the linker group itself. The initial linker modification ex-
plored was a methylene replacement. Synthesis of the endo and exo
isomers of the methylene linked azabicyclooctane is described in
Scheme 1. Reagents and conditions: (a) 4, KOH, t-BuOH, 85 °C; (b)
macroporous polystyrene bound triacetoxyborohydride (MP-BH(OAc)3), DCM; (c)
HCl/EtOH (satd), rt; (d) dimethylcarbamoyl chloride, Et3N, DCM.
5 or 6,
Table 1
SAR comparison of unsubstituted analogs 2–3 with trans and cis-4-cyclohexylurea derivatives 9–12a
d
Compd
R
n
[
35S]GTP
c
S IC50 (nM)
hERG IC50 (lM)
ja
lb
dc
2
3
—
—
1
20
3
721 47
8306 1635
>30,000
0.26
2.0
H
204 70
1826 877
O
N
N
N
N
trans-9
trans-10
cis-11
1
2
1
2
1301 394
>30,000
>30,000
>30,000
>30,000
>30,000
>33
>33
>33
>33
HN
O
451 39
4197 382
172 10
18,862 1840
18,513 4610
16,334 1060
HN
O
HN
O
cis-12
HN
a
Kappa (
j
) antagonism was determined in a human [35S]GTP
c
S assay using Dynorphin A(1-13) as the agonist stimulus. IC50’s are reported as the mean SEM (n P 2) and
S assay using DAMGO as the agonist stimulus. IC50’s are reported as the mean SEM (n P 2) and without SEM
cS assay using SNC-80 as the agonist stimulus. IC50’s are reported as the mean SEM (n P 2) and without SEM
without SEM where n = 1.
b
Mu (
where n = 1.
l c
) antagonism was determined in a human [35S]GTP
c
Delta (d) antagonism was determined in a human [35S]GTP
where n = 1.
d
Inhibition of the human ether-a-go-go (hERG) voltage-gated ion-channel was determined using an IonWorks™ electrophysiology assay.9 IC50’s are reported as the mean
of n P 2 experiments.

T. A. Brugel et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5405–5410
5407
Table 2
Comparison of 3-endo and 3-exo linked 3-oxo-meta-benzamide-8-azabicyclo-[3.2.1]-octane analogs 13–14, 18–19, and 23–25
a
Compd
A
[
35S]GTP
c
S IC50 (nM)
hERG IC50 (lM)
ja
la
da
O
O
endo-13
exo-14
286 20
10,682 1425
NDb
27,442 255
>30,000
0.61
3.0
9680 960
endo-18
exo-19
596 26
2143 172
338 11
3941 13
>30,000
1.5
11,910 1310
3077 407
24,290 700
>30,000
0.25
1.1
S
endo-23
O
S
endo-24
endo-25
8242 1966
1489 78
NDa
>30,000
20
H
N
17,039 3440
14,697 1250
1.1
a
See footnotes in Table 1.
ND = not determined
b
Scheme 2. Reagents and conditions: (a) diethyl 3-cyanobenzylphosphonate, NaH,
THF, 91%; (b) 10% Pd/C, H2 (1 atm), MeOH, 50%; (c) (i) ACE-Cl, DCM, reflux, (ii)
MeOH, reflux, (6:1, 16:17); (d) PhCHO; DCM, NaBH(OAc)3, rt; (e) [PtH(PMe2OH)(P-
Me2O)2H], EtOH/H2O (6:1), 80 °C.
Scheme 2. Tropanone 15 underwent Horner–Wadsworth–Emmons
reaction with diethyl 3-cyanobenzylphosphonate to provide the
corresponding exocyclic C(3) alkene. Reduction of the olefin fol-
lowed by demethylation with 1-chloroethylchloroformate (ACE-
Cl) provided a 6:1 mixture of endo:exo carbon-linked products 16
and 17. The sterically favored addition of hydrogen across the ole-
fin from the face opposite the ethylene bridge of the azabicyclooc-
tane core influences the level of selectivity observed for this
reaction. Isomers 16 and 17 could be separated via chiral supercrit-
ical fluid chromatography (SFC) prior to reductive amination and
nitrile hydration11 to provide 18 and 19.
Scheme 3. Reagents and conditions: (a) MsCl, Et3N, DCM; (b) 3-bromothiophenol,
NaH, THF, reflux; (c) Zn(CN)2, Pd(PPh3)4, DMF, 100 °C; (d) 4 M HCl/dioxane, DCM;
(e) PhCHO, MP-BH(OAc)3, DCM; (f) [PtH(PMe2OH)(PMe2O)2H], EtOH/H2O (6:1),
80 °C; (g) NaIO4, MeOH/H2O.
was determined to be reduced versus the oxygen linked analog
(seven-fold vs 37-fold, respectively).
We decided to explore additional linker-atom changes in the
endo configuration. Hence, synthesis of the sulfur-linked deriva-
tives was undertaken beginning with Boc-pseudotropine 20
(Scheme 3). Mesylation of the C(3) hydroxyl followed by nucleo-
philic displacement with 3-bromothiophenol produced exclusively
the endo sulfur-linked intermediate 21.12 To incorporate the pri-
mary amide functionality, 21 was subjected to palladium catalyzed
It was observed that endo-18 produced only a slight reduction
(two-fold) in
j antagonism relative to its oxygen counterpart
endo-13 (Table 2). As was seen with 13, exo-19 was less potent
than endo-18, but only by 3.5-fold. Although the endo isomer re-
mained preferred for the methylene linked analog, the difference
was markedly less. Additionally, the
l/j selectivity for endo-18

5408
T. A. Brugel et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5405–5410
to be an adequate replacement for oxygen as endo-23 showed near
equal potency to endo-10, although with somewhat reduced
selectivity. However, conversion to the sulfoxide endo-24 was not
well tolerated leading to reduced antagonism.
l/j
j
Finally, synthesis of the NH-linked analog endo-25 commenced
with reductive amination between tropanone 15 and 3-aminoben-
zonitrile to give exclusively the endo intermediate (Scheme 4).
Standard manipulations as described previously produced endo-
Scheme 4. Reagents and conditions: (a) 3-aminobenzonitrile, Na(OAc)3BH, DCE; (b)
(i) ACE-Cl, DCM, reflux; (ii) MeOH, reflux; (c) PhCHO, MP-BH(OAc)3, DCM; (d)
[PtH(PMe2OH)(PMe2O)2H], EtOH/H2O (6:1), 80 °C.
25. Subsequent testing in the
j functional assay showed that the
NH-linker was the least tolerated modification, producing a five-
fold decrease in antagonism.
With N-substitution and linker modifications adequately
scoped, we turned our attention to the benzamide region of this
cyanation13 to afford the corresponding meta-benzonitrile product.
Boc deprotection followed by reductive amination gave 22. Hydra-
tion of the nitrile produced endo-23. Additionally, selective oxida-
tion of 22 to the corresponding sulfoxide followed by nitrile
hydration gave endo-24. As seen in Table 5, the thio linker proved
series of
j antagonists. Our initial desire was to incorporate het-
eroatoms into the benzamide ring to lower the lipophilicity of this
region in anticipation that this structural change would assist in
limiting hERG activity. All four pyridyl analogs of the 3-substituted
benzamide 13 were synthesized (26–29, Table 3).14 A minimum of
Table 3
SAR of 3-oxo-meta-benzamide-8-azabicyclo-[3.2.1]-octane analogs 26–34 containing
a modified meta-benzamide group
a 30-fold loss in
Additionally, pyridazine analog 30 was also inactive at both
and suggesting a strict criteria for a lipophilic aromatic group
j antagonism was observed with these changes.
j
l
at this position in the receptor binding site. Furthermore, none of
these modifications had a positive impact on reducing hERG
inhibition.
Addition of small substitutions on the benzamide ring was ex-
plored and good tolerance for a 5-fluoro-substituent with meta-
Compd
R
[
35S]GTP
c
S IC50 (nM)
hERG IC50
M)
(
l
j
l
amide analog 31 (
the fluoro for a methyl at the 5-position in 32 resulted in some
reduction in potency. Repositioning this methyl to either the
6- (33) or 2-position (34) led to even further loss of potency.
We also explored repositioning of the primary amide function-
ality on the aromatic ring. The ortho (35) and para (36) isomers of
j IC50 = 222 nM) was observed. Replacement of
N
O
26
27
8624 705
>30,000
ND
1.0
1.6
j
NH2
N
O
14,161 385
13 were not well tolerated in the
However, upon addition of a 2-methyl group to para-benzamide
36, we observed a recovery of potency (37, IC50 = 462 nM) al-
beit with complete loss of selectivity. We then looked to place
j antagonist assay (Table 4).
NH2
j
j
N
N
l/j
O
28
29
30
17,802 1600 ND
1.5
1.0
15
other substituents at the 2-position including fluoro (38) and nitro-
gen (39), as well as nitrogen at the 3-position (40). Unfortunately,
these modifications only resulted in a loss of activity.
NH2
In order to better understand the role of the primary amide in
this series of antagonists, we explored several derivatives of this
group. As seen in Table 4, neither N-methyl nor N,N-dimethyl sub-
stitution to give secondary amide 41 or tertiary amide 42, respec-
tively, were tolerated. This underscores the specific requirement
for a primary amide to obtain any level of opioid antagonist activ-
ity. Several research groups have observed that the primary car-
boxamide moiety serves as an effective isostere for the phenolic
group in many known peptidic and non-peptidic opioid ligands.15
We therefore synthesized phenol analog 43 which did show equiv-
O
>30,000
>30,000
ND
ND
NH2
N
N
O
NH2
F
31
32
222
2
8084
0.68
alent
to the parent benzamide 13.
In Table 5, the in vitro ADME profile for a pair of our novel
j antagonism, and slightly improved l/j selectivity, relative
O
NH2
j
antagonists is summarized. Overall, these analogs possessed low
metabolism by human liver microsomes with a moderate increase
observed when tested in rat microsomes. However, these results
represent an attenuation of rat metabolism relative to what was
observed in compounds described in our previous communica-
tion.4 Permeability (MDCK-MDR1 Papp A > B) was seen as very good
for carbon-linked analog 18, while the more extended analog 12
showed limited permeability. In both cases, P-gp efflux ratios were
moderate. These analogs also showed high unbound levels of com-
pound as measures in rat plasma protein binding (PPB).
1733 51
17,429 2495 0.66
O
NH2
O
33
34
3500 120
4176 790
>23,040
>26,704
ND
1.1
NH2
O
We next undertook an in vivo rat pharmacokinetic study to fur-
ther understand how the in vitro properties of these analogs would
translate. As summarized in Table 6, analog 18 showed high levels
of plasma clearance (CL) and a moderate volume of distribution
NH2

T. A. Brugel et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5405–5410
5409
Table 4
SAR of 3-oxo-benzamide-8-azabicyclo-[3.2.1]-octane analogs 35–43 containing ortho and para-benzamides as well as amide isosteres
Compd
R
[
35S]GTP
c
S IC50 (nM)
hERG IC50 (lM)
j
l
35
>30,000
ND
0.34
H2N
O
O
O
36
37
38
39
40
41
10,729 515
462 29
ND
0.68
1.0
1.4
1.3
ND
ND
NH2
NH2
NH2
419 28
8188 188
ND
O
F
2698 117
19,287 1670
>30,000
O
O
N
NH2
N
ND
NH2
O
>30,000
ND
HN
N
O
42
43
>30,000
347 75
ND
ND
1.6
22,230 4235
OH
Table 5
In vitro DMPK properties for azabicyclo[3.2.1]octane derivatives 12 and 18
Efflux ratioc
Rat PPBd (%free)
a
a
b
Compd
j
IC50 (nM)
hCLint
rCLint
Papp A > B
12
18
172
596
<4
<4
30
65
19
140
2.9
3.1
92
81
a
Human and rat liver microsomal intrinsic clearance (hCLint) was measured as
protocol.
ll/min/mg protein according to the standard human liver microsomal stability assay
b
Papp A > B is the rate (nm/s) of transport in the apical to basolateral (A > B) direction as measured with MDR1 transfected MDCK1 canine kidney cells.
Efflux is measured as a ratio of transport from apical to basolateral (A > B) versus basolateral to apical (B > A) directions using MDR1 P-glycoprotein (P-gp) expressed
c
MDCK cells.
d
Rat plasma protein binding (PPB) was determined as percent free (%free) compound available after incubation in male Sprague–Dawley rat plasma.
(VDss) when dosed intravenously (iv). This combination contrib-
uted to the short iv half-life (t1/2 = 1.3 h) and low bioavailability
(%F = 5%) observed for 18. A good brain/plasma ratio of 1.9 sug-
gested sufficient levels of CNS exposure for this compound and
could be predicted from the favorable in vitro permeability data.
Conversely, cyclohexylmethyl urea analog 12 possessed poor brain
exposure (total brain/plasma ratio = 0.22, predicted unbound
brain/plasma ratio <0.2), a result consistent with the moderate
P-gp efflux liability and poor permeability identified in vitro.
In conclusion, we have followed our initial disclosure of potent
and selective
j antagonists based on an 8-azabicyclo[3.2.1]octan-
3-yloxy-benzamide structure with additional SAR analysis at three
key regions. High tolerance for extended substitutions of the basic
nitrogen was observed leading to analogs showing improved hERG

5410
T. A. Brugel et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5405–5410
Table 6
References and notes
In vivo rat pharmacokinetic profile for azabicyclo[3.2.1]octane derivatives 12 and 18
a
a
1. For a review see: Aldrich, J. V.; Vigil-Cruz, S. C. Burger’s Medicinal Chemistry and
Drug Discovery, 6th ed.; John Wiley & Sons: New York, 2003. pp 329–481.
2. Carlezon, W. A.; Béguin, C.; Knoll, A. T.; Cohen, B. M. Pharm. Therap. 2009, 123,
334.
Compd
CLa (ml/min/kg)
t1/2 (h)
VDss (L/kg)
[Br]/[Pl]a
%Fb
12
18
NDc
124
ND
1.3
ND
6.6
0.22
1.9
ND
5
3. (a) Portoghese, P. S.; Lipkowski, A. W.; Takemore, S. E. Life Sci. 1987, 40, 1287;
(b) McLaughlin, J. P.; Maron-Popovici, M.; Chavkin, C. J. Neurosci. 2003, 23,
5674.
4. Brugel, T. A.; Smith, R. W.; Balestra, M.; Becker, C.; Daniels, T.; Hoerter, T. N.;
Koether, G. M.; Throner, S. R.; Panko, L. M.; Folmer, J. J.; Cacciola, J.; Hunter, A.
m.; Liu, R.; Edwards, P. D.; Brown, D. G.; Gordon, J.; Ledonne, N. C.; Pietras, M.;
Schroeder, P.; Sygowski, L. A.; Hirata, L. T.; Zacco, A.; Peters, M. F. Bioorg. Med.
Chem. Lett. 2010. doi:10.1016/j.bmcl.2010.07.113.
a
Plasma clearances (CL), half-lives (t1/2), volume of distributions (VDss) and
brain:plasma concentration ratios ([Br]/[Pl]) were measured after a 5 mol/kg (12)
or 2 mol/kg (18) intravenous (iv) infusion dose in Sprague–Dawley rats.
Percent bioavailability (%F) was measured using the above iv data along with
data from a 2 mol/kg (18) oral (po) dose in Sprague–Dawley rats.
ND = not determined; these parameters were not calculated from the iv infu-
sion data.
l
l
b
l
c
5. Inhibition of the human ether-a-go-go (hERG) voltage-gated ion-channel was
determined using an IonWorks™ electrophysiology assay.6 IC50’s are reported
as the mean of n P 2 experiments.
6. Schroeder, K.; Neagle, B.; Trezise, D. J.; Worley, J. J. Biomol. Screening 2003, 8, 50.
7. See Ref. 4 above for details on the preparation of 4.
and opioid selectivity. Modifications to the 3-substituent of the
azabicyclic core revealed a preference for the endo over exo confor-
mation and moderate tolerance for methylene and sulfur linkers.
Finally, changes to the benzamide region of this series, including
increasing hydrophilicity and repositioning the primary amide,
8. All new analogs were synthesized in >95% purity as determined by 1H NMR and
LC–MS analysis.
9. (a) Schwyzer, R.; Eberle, A. Front. Horm. Res. 1977, 4, 18; (b) Metzger, T. G.;
Paterlini, G.; Portoghese, P. S.; Ferguson, D. M. Neurochem. Res. 1996, 21, 1287.
10. Endo versus exo stereochemistry of the linker substitution was determined
using 1D-NOESY NMR experiments.
generally resulted in loss of
j antagonism and/or opioid selectivity.
11. Gaffar, T.; Parkins, A. W. Tetrahedron Lett. 1995, 36, 8657.
12. Nagase, T.; Takahashi, T.; Sasaki, T.; Nagumo, A.; Shimamura, K.; Miyamoto, Y.;
Kitazawa, H.; Kanesaka, M.; Yoshimoto, R.; Aragane, K.; Tokita, S.; Sato, N. J.
Med. Chem. 2009, 52, 4111.
This knowledge has provided greater scope about this new chem-
otype that can be further utilized to optimize the properties of
these novel
j antagonists.
13. Tschaen, D. M.; Desmond, R.; King, A. O.; Fortin, M. C.; Pipik, B.; King, S.;
Verhoeven, T. R. Synth. Commun. 1994, 24, 887.
Acknowledgments
14. Analogs from Tables
3 and 4 were synthesized according to the general
procedure described in Ref. 4.
We thank Jennifer Van Anda, David Coomber, Frances McLaren,
and Xiaomei Ye for purification support; James Hall for NMR sup-
port; Timothy Blake for assistance with LC–MS analysis; and Nor-
man Ledonne and Patricia Schroeder for in vitro and in vivo ADME
support.
15. (a) Dolle, R. E.; Machaut, M.; Martinez-Teipel, B.; Belanger, S.; Cassel, J. A.;
Stabley, G. J.; Gracyzk, T. M.; De Haven, R. N. Bioorg. Med. Chem. Lett. 2004, 14,
3545; (b) Wentland, M. P.; Lou, R.; Denhardt, C.; Duan, W.; Cohen, D. J.; Bidlack,
J. M. Bioorg. Med. Chem. Lett. 2001, 11, 1717; (c) Wentland, M. P.; Lou, R.; Ye, Y.;
Cohen, D. J.; Richardson, G. P.; Bidlack, J. M. Bioorg. Med. Chem. Lett. 2001, 11,
623.