Synthesis of novel triplet drugs with 1,3,5-trioxazatriquinane skeletons and their pharmacologies. Part 2: Synthesis of novel triplet drugs with the epoxymethano structure (capped homotriplet)

Article abstract of DOI:10.1016/j.bmcl.2011.07.065

An improved synthetic method for triplet drugs with the 1,3,5-trioxazatriquinane skeleton was developed that used p- toluenesulfonylmethyl isocyanide (TosMIC) instead of 1,3-dithiane. Using the improved method, we synthesized compounds with two identical

Full text of DOI:10.1016/j.bmcl.2011.07.065

Bioorganic & Medicinal Chemistry Letters 21 (2011) 6198–6202
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Synthesis of novel triplet drugs with 1,3,5-trioxazatriquinane skeletons
and their pharmacologies. Part 2: Synthesis of novel triplet drugs
with the epoxymethano structure (capped homotriplet)
Hiroshi Nagase a, , Koji Koyano a, Naohisa Wada a, Shigeto Hirayama a, Akio Watanabe a, Toru Nemoto a,
⇑
Mayumi Nakajima b, Kaoru Nakao b, Hidenori Mochizuki b, Hideaki Fujii a
a School of Pharmacy, Kitasato University, 5-9-1 Shirokane, Minato-ku, Tokyo 108-8641, Japan
b Pharmaceutical Research Laboratories, Toray Industries, Inc., 6-10-1, Tebiro, Kamakura, Kanagawa 248-8555, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
An improved synthetic method for triplet drugs with the 1,3,5-trioxazatriquinane skeleton was devel-
oped that used p-toluenesulfonylmethyl isocyanide (TosMIC) instead of 1,3-dithiane. Using the improved
method, we synthesized compounds with two identical pharmacophore units and an epoxymethano
group, that is, capped homotriplets. Among the synthesized capped homotriplets, KNT-123 showed high
Received 30 May 2011
Revised 16 July 2011
Accepted 16 July 2011
Available online 26 July 2011
selectivity for the
l
receptor over the
j
receptor, and the
l
selectivity was the highest among the
dose-
reported selective nonpeptide ligands. KNT-123 administered subcutaneously induced
l
a
Keywords:
Cap structure
Capped homotriplet
p-Toluenesulfonylmethyl isocyanide
Opioid
dependent analgesic effect in the acetic acid writhing assay, and its potency was 11-fold more potent
than that of morphine. KNT-123 may serve as a useful tool for the study of the pharmacological actions
mediated specifically via the
l receptor.
Ó 2011 Elsevier Ltd. All rights reserved.
Antinociception
Many twin drugs have been reported.1,2 Symmetrical twin
drugs can simultaneously fit into symmetrical binding sites of a
protein complex to afford increased activity, while nonsymmetri-
cal twin drugs may bind to individual relevant binding sites to pro-
vide dual action.3 However, twin drugs can only play one role,
either an increase of activity or a dual action. The advent of a rigid
triplet drug (trimer drug) containing three pharmacophore units in
a single molecule might provide the ability to deliver both in-
creased activity and dual actions. Recently, we reported a method
for synthesizing the rigid symmetrical and nonsymmetrical triplet
drug 1 with a 1,3,5-trioxazatriquinane skeleton that, depending on
the substituents, carried three identical and nonidentical morphi-
nan units, respectively (Fig. 1).4,5 Among the synthesized triplet
drugs, the symmetrical triplet KNT-93 (R1 = R2 = Me, R3 = R4 = OH)
showed about 56-fold more potent analgesic effects than did mor-
phine in an acetic acid writhing test,5 indicating that triplet drugs
would be useful tools for pharmacological investigation. For the
comparison with pharmacological effects induced by triplet drugs,
we considered the importance of the synthesis of the correspond-
ing compounds 2, which have two identical pharmacophore units
and the epoxymethano structure (we term the structure ‘cap struc-
ture’). Moreover, the compound 2 is expected to be useful for
investigation of the dimerization phenomena of G protein-coupled
receptors.6–11 Herein, we report the synthesis of novel compound 2
with the cap structure (which we named ‘capped homotriplets’)
and their pharmacologies. We also report the improved synthesis
of triplet drugs.
In the previous synthesis of triplet drugs (Scheme 1), air- and
moisture-sensitive n-BuLi and malodorous 1,3-dithiane were used.
R1
R4
N
R3
H
R3
H
O
R2
O
N
R1
N
O
R4
N
O
R3
O
N
O
O O
R4
O
O
R4
R4
O
R3
R3
N
N
R1
2
R1
1
R1, R2 = Me, CPM
R3 = H, OH
R4 = OMe, OH
⇑
Corresponding author.
E-mail address: nagaseh@pharm.kitasato-u.ac.jp (H. Nagase).
Figure 1. Structures of symmetrical or nonsymmetrical triplet drugs 1 and capped
homotriplets 2. The cap structure is depicted in blue. CPM: cyclopropylmethyl.
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.07.065

H. Nagase et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6198–6202
6199
OH
O
OH
OH
OH
MeO
S
Me
Me
Me
N
N
Me
N
N
H
OMe
i)
ii)
iii)
S
O
OH
O
OH
O
OH
O
O
OMe
Me
OMe
OMe
OMe
3
4
5
6
OH
N
MeO
N
iv)
v)
OH
O
O
1 (R1 = R2 = Me, R3 = OH, R4 = OMe)
O
H
OMe
HO
N
Me
7
Scheme 1. Previous synthesis of triplet 1 (R1 = R2 = Me, R3 = OH, R4 = OMe). Reagents and conditions: (i) n-BuLi, 1,3-dithiane, DME, ꢀ78 °C, 65%; (ii) CuCl2ꢂ2H2O, p-TsOHꢂH2O,
CH(OMe)3, MeOH, reflux, 38%; (iii) 2 M HCl, reflux; (iv) NH4Cl, AcONa, MeOH, reflux, 75% from 5; (v) CSA, CHCl3, reflux, 29%.
Me
N
O
O
OH
Me
N
HO
OH
Ts
CHO
Me
O
i)
ii)
iii)
MeO
MeO
MeO
MeO
+
MeO
Me
Me
Me
O
HO
O
H
Me
OMe
OMe
8
9
10
11
12
Scheme 2. New synthesis of the key intermediate oxazoline dimer 12. Reagents and conditions: (i) TosMIC, K2CO3, MeOH, rt; (ii) 2 M HCl, THF, rt; (iii) NH4Cl, AcONa, MeOH,
reflux, 64% from 8.
Moreover, the low reaction temperature (ꢀ78 °C) was required.4,5
O
OH
#
OH
Therefore, we sought reaction conditions that were more practical
N
#
MeO
N
N
O
MeO
and concise. After investigations using model compound 8, we
found a new synthetic method for the key oxazoline dimer inter-
mediates 12 that used p-toluenesulfonylmethyl isocyanide (Tos-
MIC).12,13 Ketone 8 was treated with TosMIC at rt in the presence
of K2CO3 to give tosyloxazoline intermediate 9, followed by hydro-
#
O
N
O
OH
O
O
*
O
i)
O
H
OMe
HO
OMe
N
HO
N
14R
16
lysis of 9 with 2 M HCl to afford a mixture of
a-hydroxyaldehyde
10 and hemiacetal dimer 11. The resulting mixture was treated
with NH4Cl in the presence of AcONa in MeOH to provide the oxaz-
oline dimers 12 in 64% from 8 (Scheme 2).
Scheme 4. Reagents and conditions: (i) glycolaldehyde dimer, TMSOTf, CHCl3, rt,
63%.
With the milder new synthetic method for preparation of the
oxazoline dimer intermediate in hand, we attempted to apply the
method to the synthesis of capped homotriplet 15. The oxazoline
dimer intermediates 14, prepared from naltrexone methyl ether
(13) by the same method as shown in Scheme 2, was treated with
the glycolaldehyde dimer in the presence of camphorsulfonic acid
(CSA) at rt to give capped homotriplet 15 in 62% yield from 13
(Scheme 3). The R-isomer 14R of the oxazoline dimer 14 with the
R-configuration at the ꢁ-position was reported to be the kinetically
controlled product and to be converted into the corresponding tri-
mer with S-configuration at all methine carbons in the triox-
azatriquinan moiety via the thermodynamically controlled
oxazoline dimer 14S with the S-configuration at the ꢁ-position.4
Therefore, the mixture of oxazoline dimers 14R + 14S could be con-
verted into capped homotriplet 15 under the reaction conditions
described in Scheme 3. It is interesting that the treatment of the
R-oxazoline dimer 14R, which was obtained by silica gel column
chromatographic purification from a mixture of 14R and 14S, in
the presence of TMSOTf instead of CSA provided capped homotri-
plet 16 with R-configuration at all the #-positions (Scheme 4).
The configurations of the 1,3,5-trioxazatriquinane skeleton moie-
ties in capped homotrimers 15 and 16 were determined by NOE
and ROESY experiments (Fig. 2).
OH
O
OH
N
N
MeO
OH
N
MeO
O
N
N
O
O
O
OH
O
i) - iii)
iv)
O
*
O
O
H
O
OMe
OMe
HO
N
HO
OMe
N
13
14
15
Scheme 3. Reagents and conditions: (i) TosMIC, K2CO3, MeOH, rt; (ii) 2 M HCl, THF, rt; (iii) NH4Cl, AcONa, MeOH, reflux; (iv) glycolaldehyde dimer, CSA, CHCl3, rt, 62% from
13.

6200
H. Nagase et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6198–6202
R2
2.7 %
H
O
8.2 %
HO
R1
N
HO
N
O
H
O
3.3 %
O
H
i) or ii)
O
X
X
H
15, 18a-e
N
N
OH
O
OMe
R2
O
O
H
N
R1
H
KNT-152: R1 = CPM, R2 = OH, X = -O-
KNT-123: R1 = Me, R2 = OH, X = -O-
SYK-138: R1 = i-Bu, R2 = OH, X = -O-
SYK-139: R1 = CBM, R2 = OH, X = -O-
SYK-219: R1 = CPM, R2 = OH, X = H, H
SYK-220: R1 = CPM, R2 = H, X = -O-
9.9 %
OMe
HO
N
15
NOE
H
O
H
H
OH
H
O
O
HO
N
N
OMe
N
O
HO
N
H H
O
O
O
O
i)
O
O
16
OH
HO
OMe
16
HO
N
N
ROESY
SYK-140
OH
Figure 2. Observed NOEs and ROESYs in compounds 15 and 16, respectively.
N
HO
N
R2
N
OH
i)
O
O
R1
O
*
O
14R, 14S
R2
N
MeO
R1
O
N
N
OH
O
X
X
i) - iv)
HO
O
or
X
v) - ix)
OMe
R2
SYK-43: isomer
R
N
OMe
KNT-75: S isomer
R1
17a-e
18a-e
Scheme 6. Reagents and conditions: (i) 1-PrSH, t-BuOK, DMF, 130 °C, 34–86%; (ii)
BBr3, CH2Cl2, 0 °C, 83%.
a: R1 = Me, R2 = OH, X = -O-
b: R1 = i-Bu, R2 = OH, X = -O-
c: R1 = CBM, R2 = OH, X = -O-
d: R1 = CPM, R2 = OH, X = H, H
e: R1 = CPM, R2 = H, X = -O-
Table 1
Binding affinity of triplet drugs for the opioid receptorsa
b
d
Compound
Ki (l)
(nM)
Ki (d)c (nM)
Ki (
j)
(nM)
Selectivity
d/l
Scheme 5. Reagents and conditions: (i) TosMIC, K2CO3, MeOH, rt; (ii) 2 M HCl, THF,
rt; (iii) NH4Cl, AcONa, MeOH, reflux; (iv) glycolaldehyde dimer, CSA, CHCl3, rt, 12–
19% from 17; (v) n-BuLi, 1,3-dithiane, DME, ꢀ78 °C, 65–90%; (vi) CuCl2ꢂ2H2O, p-
TsOHꢂH2O, CH(OMe)3, MeOH, reflux; (vii) 2 M HCl, reflux; (viii) NH4Cl, AcONa,
MeOH, reflux; (ix) glycolaldehyde dimer, CSA, CHCl3, rt, 9–12% from 17.
j/
l
Naltrexone
KNT-152
KNT-123
SYK-138
SYK-139
SYK-219
SYK-220
SYK-140
SYK-43
0.265
0.480
2.093
25.27
2.600
0.771
0.719
38.22
1.031
1.900
12.27
16.15
52.86
>1000
49.96
5.979
7.797
>1000
39.62
53.74
0.702
2.752
337.4
118.7
176.1
8.670
6.355
247.2
1.656
1.164
2.64
5.73
161.2
4.70
67.73
11.25
8.84
6.47
1.61
0.61
46.23
33.62
25.26
—
19.22
7.76
10.85
—
Capped homotriplets 18 with various N-substituents or without
4,5-epoxy bridges14 or angular hydroxy groups were synthesized
from the corresponding ketones 17 by the new (TosMIC method)
or by the previous synthetic method using dithiane (Scheme 5).
Obtained capped homotriplets 15, 16, and 18 and intermediate
oxazoline dimers 14R and 14S were demethylated with 1-PrSH/t-
BuOK or BBr3 (Scheme 6).
The binding affinities of the prepared capped homotriplets and
oxazoline dimer derivatives for the opioid receptors were evalu-
ated with a competitive binding assay. The results were shown
in Table 1 with the affinities of naltrexone for comparison. The as-
says were performed by a modification of a previously reported
procedure.15 The binding affinities of SYK-138 and SYK-140 for
all types of opioid receptors were much lower than those of the
other test compounds. The low affinities of SYK-138 may arise
from both the weak electron donating ability and steric hindrance
of N-isobutyl groups. Although the methyl group is a weaker elec-
tron donor than the isobutyl group, the affinities of KNT-123 with
N-methyl groups was stronger than those of SYK-138. These results
38.43
28.28
KNT-75
a
Binding assays were carried out in duplicate (j: cerebellum of guinea pig, l and
d: whole brain without cerebellum of mouse).
b
[3H] DAMGO was used.
c
[3H] DPDPE was used.
d
[3H] U-69,593 was used.
may be due to reduced steric hindrance of the methyl group in
comparison to the isobutyl group and/or the high contribution of
methyl group to the affinity for the
l
receptor.5,16 In contrast to
the low affinity of SYK-140, its stereoisomer, KNT-152 showed suf-
ficient affinity for each type of opioid receptor. The difference of
affinities between the two stereoisomers SYK-140 and KNT-152
may stem from the different configurations of the trioxazatriqu-
inane skeletons in the two isomers. The absence of the 4,5-epoxy
bridges or the angular hydroxy groups (as seen in SYK-219 or

H. Nagase et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6198–6202
6201
Figure 5. Effects of naloxone on the antinociception induced by KNT-123 in the
mouse acetic acid writhing test. The statistical significance of differences between
the groups was assessed with Student’s t-test. ⁄⁄p <0.01 versus saline treated mice.
higher selectivity for the
trend for the nonpeptide ligands has been low selectivity for the
receptor type, especially with respect to the
receptor.19–22 To the
best of our knowledge, the selectivity of KNT-123 over the
receptor appears to be the highest of the reported selective non-
peptide ligands. The particular pharmacological effects mediated
through the receptor including side effects have not yet been
clarified because no nonpeptide ligands with sufficient selectivity
for the receptor have been available. Therefore, KNT-123 is
expected to be a useful tool for the investigation of pharmacologies
via the receptor.
l receptor over the j receptor. A general
l
j
l
j
l
l
l
l
We next evaluated the analgesic effects induced by KNT-123 in
the acetic acid writhing test.23,24 Subcutaneous administration of
KNT-123 exhibited dose-dependent antinociception (Fig. 4) and
the effect was significantly reduced by a
(Fig. 5). The analgesic effect of KNT-123 (ED50 = 0.14 mg/kg) was
nearly fourfold more potent than that of morphine, a representa-
l antagonist naloxone
Figure 3. Superimposition of the 3D-alignment of nor-BNI onto that of KNT-75 (A)
or KNT-152 (B). The 170-nitrogen, which was reported to important to exert
j
selectivity,26 and the corresponding nitrogen in nor-BNI and KNT-75 or KNT-152
were indicated by green and yellow circle, respectively.
tive
l agonist (ED50 = 0.6 mg/kg), but about fourfold less potent
than that symmetrical triplet KNT-93 (ED50 = 0.037 mg/kg).5 When
the ED50 values were converted from gram units to molar units,25
SYK-220) afforded only a small effect on their affinities. Interest-
the potency of KNT-123 in antinociception (ED50 = 0.18
was 11-fold higher than that of morphine (ED50 = 1.9
but fivefold lower than that of KNT-93 (ED50 = 0.034
In conclusion, we have developed an improved synthetic meth-
od for oxazoline dimers, which were the key intermediates for the
synthesis of triplet drugs with the 1,3,5-trioxazatriquinane skele-
ton. Using the improved method, we synthesized compounds with
two identical pharmacophore units and an epoxymethano group,
that is, capped homotriplets. Among the synthesized capped
l
mol/kg)
mol/kg),
lmol/kg).
ingly, the capped homotriplet KNT-152 showed
l selectivity,
l
whereas its precursor, the oxazoline dimer KNT-75, exhibited
j
selectivity. The relative spatial locations of the two 4,5-epox-
ymorphinan units in KNT-75 appeared to be similar to that of
the selective
j
antagonist, nor-BNI,17,18 while those of KNT-152
were different (Fig. 3). It was noteworthy that KNT-123 showed
homotriplets, KNT-123 showed high selectivity for the
over the receptor, and the selectivity was the highest among
the reported selective nonpeptide ligands. KNT-123 adminis-
l receptor
j
l
l
tered subcutaneously induced dose-dependent analgesic effects
in the acetic acid writhing assay, and its potency was 11-fold more
potent than that of morphine. KNT-123 is expected to be a useful
tool for the study of the specific pharmacological actions via the
l
receptor.
Acknowledgments
We acknowledge the financial supports from Shorai Foundation
for Science and Uehara Memorial Foundation. We also acknowl-
edge the Institute of Instrumental Analysis of Kitasato University,
School of Pharmacy for its facilities.
Figure 4. Analgesic effects induced by KNT-123 in the mouse acetic acid writhing
test. The statistical significance of differences between the groups was assessed
with one-tailed nonparametric Williams’ test. ⁄p <0.025 versus saline treated mice.

6202
H. Nagase et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6198–6202
oxabicyclo[3.2.1]octane derivative instead. Watanabe, A.; Fujii, H.; Nakajima,
Supplementary data
M.; Hasebe, K.; Mochizuki, H.; Nagase, H. Bioorg. Med. Chem. Lett. 2009, 19,
2416. On the other hand, the treatment of the oxazoline dimer with CSA and
glycolaldehyde dimer at rt afforded only capped homotriplet 18d.
15. Fujii, H.; Ogawa, R.; Ohata, K.; Nemoto, T.; Nakajima, M.; Hasebe, K.;
Mochizuki, H.; Nagase, H. Bioorg. Med. Chem. 2009, 17, 5983.
16. Yamamoto, N.; Fujii, H.; Nemoto, T.; Nakajima, R.; Momen, S.; Izumimoto, N.;
Hasebe, K.; Mochizuki, H.; Nagase, H. Bioorg. Med. Chem. Lett. 2011, 21, 4104.
17. Portoghese, P. S.; Lipkowski, A. W.; Takemori, A. E. J. Med. Chem. 1987, 30, 238.
18. Portoghese, P. S.; Nagase, H.; Lipkowski, A. W.; Larson, D. L.; Takemori, A. E. J.
Med. Chem. 1988, 31, 836.
Supplementary data (new synthesis of the oxazoline dimmers
12) associated with this article can be found, in the online version,
at doi:10.1016/j.bmcl.2011.07.065.
References and notes
1. Fujii, H. Top. Curr. Chem. 2011, 299, 239. and references cited therein.
2. Schiller, P. W. Life Sci. 2010, 86, 598. and references cited therein.
3. Contreras, J.-M.; Sippl, W. In The Practice of Medicinal Chemistry; Wermuth, C.
G., Ed., 3rd ed.; Academic press: Oxford, 2008; pp 380–414.
4. Nagase, H.; Watanabe, A.; Harada, M.; Nakajima, M.; Hasebe, K.; Mochizuki, H.;
Yoza, K.; Fujii, H. Org. Lett. 2009, 11, 539.
19.
A l antagonist, cyprodime: Schmidhammer, H.; Jennewein, H. K.; Krassnig, R.;
Traynor, J. R.; Patel, D.; Bell, K.; Froschauer, G.; Mattersberger, K.;
Jachs-Ewinger, C.; Jura, P.; Fraser, G. L.; Kalinin, V. N. J. Med. Chem. 1995, 38,
3071.
20.
21.
22.
l Antagonists, 6-heterocyclic substituted naltrexamine derivatives: Li, G.;
Aschenbach, L. C.; Chen, J.; Cassidy, M. P.; Stevens, D. L.; Gabra, B. H.; Selley, D.
E.; Dewey, W. L.; Westkaemper, R. B.; Zhang, Y. J. Med. Chem. 2009, 52, 1416.
5. Nagase, H.; Watanabe, A.; Nemoto, T.; Nakajima, M.; Hasebe, K.; Mochizuki, H.;
Fujii, H. Bioorg. Med. Chem. Lett. 2011, 21, 4023.
A
l antagonist, 14-O-heterocyclic-substituted naltrexone derivative: Li, G.;
6. Rios, C. D.; Jordan, B. A.; Gomes, I.; Devi, L. A. Pharmacol. Ther. 2001, 92, 71.
7. Levac, B. A. R.; O’Dowd, B. F.; George, S. R. Curr. Opin. Pharmacol. 2002, 2, 76.
8. Prinster, S. C.; Hague, C.; Hall, R. A. Pharmacol. Rev. 2005, 57, 289.
9. van Rijn, R. M.; Whistler, J. L.; Waldhoer, M. Curr. Opin. Pharmacol. 2010, 10, 73.
10. Palczewski, K. Trends Biochem. Sci. 2010, 35, 595.
11. Some twin drugs that consist of two pharmacophore units linked together by a
linker with an appropriate length are generally applied for the investigation of
receptor dimmers. Such twin drugs can be classified into two categories by the
linker length: those with longer chains (usually ten to twenty atom units) and
those with shorter or no linker. The former are believed to bind two active sites
simultaneously, while the mode of binding of the latter is not clear. However,
the latter show strong binding affinities and/or profound potencies of their
pharmacological effects. See Ref. 1.
Aschenbach, L. C. K.; He, H.; Selley, D. E.; Zhang, Y. Bioorg. Med. Chem. Lett. 2009,
19, 1825.
A
l agonist, morphine: Toll, L.; Berzetei-Gurske, I. P.; Polgar, W. E.; Brandt, S.
R.; Adapa, I. D.; Rodriguez, L.; Schwartz, R. W.; Haggart, D.; O’Brien, A.; White,
A.; Kennedy, J. M.; Craymer, K.; Farrington, L.; Auh, J. S. NIDA Res Monogr. 1998,
178, 440.
23. Kawai, K.; Hayakawa, J.; Miyamoto, T.; Imamura, Y.; Yamane, S.; Wakita, H.;
Fujii, H.; Kawamura, K.; Matsuura, H.; Izumimoto, N.; Kobayashi, R.; Endo, T.;
Nagase, H. Bioorg. Med. Chem. 2008, 16, 9188.
24. Fujii, H.; Osa, Y.; Ishihara, M.; Hanamura, S.; Nemoto, T.; Nakajima, M.; Hasebe,
K.; Mochizuki, H.; Nagase, H. Bioorg. Med. Chem. Lett. 2008, 18, 4978.
25. The calculations for converting the ED50 values from gram units to molar units
were based on the molecular weights of KNT-123 trihydrochloride, morphine
hydrochloride, and KNT-93 tetrahydrochloride.
12. Oldenziel, O. H.; van Leusen, A. M. Tetrahedron Lett. 1974, 15, 163.
13. Oldenziel, O. H.; van Leusen, A. M. Tetrahedron Lett. 1974, 15, 167.
14. The treatment of the oxazoline dimer without a 4,5-epoxy bridge with CSA
under CHCl3 reflux conditions provided not the symmetrical triplet, but the
26. Lin, C.-E.; Takemori, A. E.; Portoghese, P. S. J. Med. Chem. 1993, 36, 2412.