Synthesis and biological evaluation of new biphalin analogues with non-hydrazine linkers

Article abstract of DOI:10.1016/j.bmcl.2005.03.067

Biphalin is a potent opioid peptide agonist, with a palandromic structure, composed of two enkephalin-like active fragments connected tail to tail by a hydrazine linker (Tyr-D-Ala-Gly-Phe-NH-NH<-Phe<-Gly<-D-Ala<-Tyr). This study presents the synthesis and

Full text of DOI:10.1016/j.bmcl.2005.03.067

Bioorganic & Medicinal Chemistry Letters 15 (2005) 2471–2475
Synthesis and biological evaluation of new biphalin analogues
with non-hydrazine linkers
Adriano Mollica,a,b Peg Davis,c Shou-Wu Ma,c Josephine Lai,c
Frank Porrecac and Victor J. Hrubya,*
aDepartment of Chemistry, University of Arizona, Tucson, Arizona 85721, USA
bDipartimento di Studi Farmaceutici, Universita’ di Roma ‘‘La Sapienza’’, P. le A. Moro 5, Roma 00185, Italy
cDepartment of Pharmacology, University of Arizona, Tucson, Arizona 85721, USA
Received 3 February 2005; revised 17 March 2005; accepted 17 March 2005
Available online 12 April 2005
Abstract—Biphalin is a potent opioid peptide agonist, with a palandromic structure, composed of two enkephalin-like active frag-
ments connected tail to tail by a hydrazine linker (Tyr-D-Ala-Gly-Phe-NH-NH<-Phe<-Gly<-D-Ala<-Tyr). This study presents the
synthesis and in vitro bioassays of six new biphalin analogues with three different non-hydrazine linkers, some of which have higher
binding affinity and bioactivity than biphalin.
ꢀ 2005 Elsevier Ltd. All rights reserved.
1. Introduction
Biphalin (Fig. 1) is an opioid octapeptide with a dimeric
structure based on two identical portions derived from
enkephalins joined tail to tail by a hydrazide bridge.
This particular structure enhances the antinociceptive
activity of the enkephalins with an unknown mecha-
nism, probably based on a cooperative binding.1 Bipha-
Figure 1. Biphalin.
lin has excellent binding affinity for l and d receptors
with a EC50 of about 1–5 nM at both l and d receptors.
It is a highly potent analgesic (250 times that of mor-
phine; i.c.v. administration; tail flick test), as potent or
tion, is still not available; it has been suggested,
more potent than etorphine.2 A definitive explanation
of the extraordinary in vivo potency shown by this com-
pound, which has pronounced efficacy in pain modula-
potency than other analgesics with novel biological pro-
however, that its high agonist activity at both l and d
receptor may be related.3 Biphalin has significantly higher
files;4 in particular, most recent data show that biphalin
is unlikely to produce dependency in chronic use.5
Abbreviations: 3H-DAMGO, [D-Ala(2),N-Me-Phe(4),Gly-ol(5)]enkep-
halin; 3H-DPDPE, [3H]-c[2-D-penicillamine,5-D-penicillamine]enkeph-
In the past 10 years, there have been many attempts to
modify its structure to obtain products unaffected by
the action of enkephalinases, to enhance its antinocicep-
tive activity, and to modify the BBB penetration.3,4,6
Structure–activity relationship studies (SAR) also were
made in order to understand the elements responsible
for biphalinꢀs high activity.
alin; BBB, Blood Brain Barrier; DCM, Dichloromethane; DMF, N,N-
Dimethyl Formamide; DMSO, Dimethylsulfoxide; EDC, 1-Ethyl-(3-
dimethylaminopropyl)carbodiimide; Et2O, Diethyl Ether; EtOAc,
Ethyl acetate; GPI/LMMP, Guinea pig ileum/longitudinal muscle
myenteric plexus (l opioid receptors); HOBT, 1-hydroxybenzotriazole;
I.C.V., intracerebroventricular; MVD, mouse vas deferens (d opioid
receptors); TEA, Triethylamine; TFA, Trifluoroacetic Acid; TOCSY,
Total Correlation Spectroscopy
Keywords: Biphalin analogues; Opioid agonists; Linkers; Structure–
activity relationship.
SAR studies have shown that the Tyr1 moiety is re-
quired for interaction with the opioid receptors; the 4
and 40 positions can be modified with rigid constrained
*
Corresponding author. Tel.: +1 520 621 6332; fax: +1 520 621
8407; e-mail: hruby@u.arizona.edu
0960-894X/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.03.067

2472
A. Mollica et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2471–2475
aqueous solution was acidified with 1 N HCl to pH 3
and extracted with EtOAc. The solvent was removed
under vacuum to give the crude product as an oil. All
the other coupling reactions were performed with the
standard method9 of HOBtÆH2O/EDC/TEA in
DMF (Schemes 2 and 3).11
Figure 2. Opioid ligand with two overlapping pharmacophores.
Relative affinities (Ki; nM) for l and d binding sites in guinea pig
membranes: 31 2 (l); 187 15 (d). See Ref. 8.
The deprotection of the Na-tert-butyloxycarbonyl group
(Boc) was performed by dissolving each Na-Boc pro-
tected product into a mixture of TFA 50% in DCM
for 30 min at rt, under nitrogen atmosphere. The
DCM and the TFA were removed under vacuum and
the resultant intermediate products were used in the next
step without further purification. With the exception of
1 and 2, all the Na-Boc protected products and the final
products as TFA salts 6–8 and 12–14 were purified by
reverse-phase HPLC using a Vydac (C18-bonded,
Phe analogues or with electron rich Phe analogues
increasing the activity.7 Modification of the hydrazide
bridge by using alkyl diamines led to reduced activity,
probably because of the higher degree of freedom
around the diamide bridge.1a
Figure 2 shows how the hydrazide bridge was used as an
alternative scaffold to design opioid ligands with two
overlapping opioid pharmacophores. It was found in
Lipkowskiꢀs laboratory that this compound possesses
potent analgesic properties after intraperitoneal injec-
tion.8 This compound can be viewed as an analogue of
a truncated/modified biphalin.
˚
300 A; 10 mm · 25 cm) column and a gradient of 10–
90% acetonitrile in 0.1% aqueous TFA over 40 min at
a flow rate of 3 mL/min the extent of purity was moni-
tored at 220, 254, 280, and 350 nm. Approximately
10 mg of crude peptide was injected each time, and the
fractions containing the purified peptide were collected
and lyophilized to dryness.
This analogue still maintained opioid like activity and
represents a good example of molecular simplification.
3. NMR spectroscopy
In this letter we describe the synthesis and biological
evaluation of six new biphalin analogues in which the
hydrazide bridge has been replaced with three different
diamines containing an aromatic or aliphatic cyclic
structure: 1,4-phenylenediamine, 1,2-phenylenediamine,
and piperazine. Both the native biphalin pharmaco-
phore Tyr-D-Ala-Gly-Phe and the shorter pharmaco-
phore Tyr-D-Phe were used in the new derivatives.
All NMR spectra were acquired on a Bruker DRX-600
spectrometer at 25 ꢁC using a Nalorac triple-resonance
single-axis gradient 5 mm probe, and the data processed
with the Bruker software XWINNMR.
NMR analysis of the intermediate Na-Boc protected
products were performed in DMSO-d6, although in the
case of product 2 a mixture of CDCl3 90%/DMSO-d6
10% was used. In the case of the final products 6–8
and 12–14, the samples were dissolved as TFA salts in
90% H2O/10% D2O solution buffered at pH 4.5 contain-
ing 50 mM sodium acetate-d3/HCl buffer and 1.0 mM
NaN3 as preservative, at a peptide concentration of ca.
2. Chemistry
The syntheses of the new analogues were performed in
the solution phase following an Na-Boc strategy. The
peptides Na-Boc-D-Ala-Gly-OEt (1) and Na-Boc-Tyr-
D-Ala-Gly-OEt (2) were synthesized by the asymmetric
anhydride method9 obtained with isobutyl chlorofor-
mate and TEA at ꢀ15 ꢁC (Scheme 1)10. Dipeptide 1
was used for the next step without further purification.
Tripeptide ester 2 was purified by silica gel chromato-
graphy with EtOAc/DCM 1:4 as eluant, and then crys-
tallized in EtOAc/Et2O. The ethyl ester group was
hydrolyzed with NaOH 1 N (6 equiv) in methanol at
rt. The solvent was removed under vacuum; the basic
1
10 mM. 1D H spectra were acquired using the Water-
gate sequence12a using z-axis gradients, with a s delay
of 240 ls and 3–9–19 pulses at full power. 2D TOCSY
spectra12b were acquired in TPPI mode12c with 512 t1
increments and a mixing time of 73 ms, including
2.5 ms trim pulses before and after the MLEV-17 se-
quence. Water suppression was achieved with Watergate
using z-axis gradients, with a s delay of 330 ls. The
power level for spin-lock, trim pulses, and 3–9–19 pulses
was 8.3 kHz.
Scheme 1. Reagents and conditions: (a) H-Gly-OEtÆHCl (1 equiv), i-BuOCOCl (1.1 equiv), TEA (1.1 equiv), DCM (20 mL), ꢀ15 ꢁC for 15 min then
rt for 12 h; (b) TFA 50%, DCM, rt for 30 min under N2 atmosphere; (c) Na-Boc-Tyr-OH (1.1 equiv), i-BuOCOCl (1.1), TEA (1.1 equiv), DCM
(20 mL), ꢀ15 ꢁC for 15 min then rt for 12 h . (Overall yield 85%).

A. Mollica et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2471–2475
2473
Scheme 2. Reagents and conditions: (a) Boc-D-Phe-OH (2.1 equiv), EDC (2.1), HOBtÆH2O (2.1 equiv), TEA (3.2 equiv), 1,4-phenylenediamineÆ2HCl
(1 equiv), DMF (20 mL), 30 min in 0 ꢁC, then rt for 6 h; (b) Boc-D-Phe-OH (2.1 equiv), EDC (2.1 equiv), HOBtÆH2O (2.1 equiv), TEA (2.2 equiv),
piperazineÆHCl (1 equiv), DMF (20 mL), 30 min in 0 ꢁC, then rt for 6 h; (c) Boc-D-Phe-OH (2.1 equiv), EDC (2.1 equiv), HOBtÆH2O (2.1 equiv), TEA
(3.2 equiv), 1,2-phenylenediamineÆ2HCl (1 equiv), DMF (20 mL), 30 min in 0 ꢁC, then rt for 6 h; (d) TFA 50% in DCM 30 min in rt under N2
atmosphere, then Boc-Tyr-OH (2.1 equiv), EDC (2.1 equiv), HOBtÆH2O (2.1 equiv), TEA (3.2 equiv), DMF (20 mL), 30 min in 0 ꢁC, then rt for 6 h;
(e) TFA 50% in DCM, 30 min rt under N2 atmosphere. (Overall yield 30–50%).
3.1. Radioligand labeled binding assays
by fitting the data to the Hill equation by a computer-
ized non-linear least-square method.
Receptor binding affinities to the d and l opioid recep-
tors were evaluated as previously described.13a,b The
ligands used were [3H]DPDPE and [3H]DAMGO for
d and l receptors, respectively.
4. Results and discussion
As reported in Table 1, compounds 6–8 showed weak
binding affinity and in vitro activity at the opioid recep-
tors, probably because the Tyr and Phe moieties are not
in a favorable position to accomplish the overlapping of
the previously postulated pharmacophore. Compounds
12–14 showed exceptionally good binding affinity and
bioactivity. Analogue 12 was comparable with biphalin,
and compound 13 binds the receptors with three to five
times higher affinity than biphalin, with the in vitro
3.2. GPI and MVD in vitro bioassays
The in vitro tissue bioassays were performed as
described previously14 (see Table 1).
IC50 values represent the mean of no less than four
experiments. IC50 values, relative potency estimates,
and their associated standard errors were determined

2474
A. Mollica et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2471–2475
Scheme 3. Reagents and conditions: (a) Boc-L-Phe-OH (2.1 equiv), EDC (2.1 equiv), HOBtÆH2O (2.1 equiv), TEA (3.2 equiv), 1,4-phenylenedi-
amineÆ2HCl (1 equiv), DMF (20 mL), 30 min in 0 ꢁC, then rt for 6 h; (b) Boc-L-Phe-OH (2.1 equiv), EDC (2.1 equiv), HOBtÆH2O (2.1 equiv), TEA
(2.2 equiv), piperazineÆHCl (1 equiv), DMF (20 mL), 30 min 0 ꢁC, then 6 h in rt; (c) Boc-L-Phe-OH (2.1 equiv), EDC (2.1 equiv), HOBtÆH2O
(2.1 equiv), TEA (3.2 equiv), 1,2-phenylenediamineÆ2HCl (1 equiv), DMF (20 mL), 30 min in 0 ꢁC, then rt for 6 h; (d) TFA 50% in DCM 30 min rt
under N2 atmosphere, then Boc-Tyr-D-Ala-Gly-OH (2.1 equiv), EDC (2.1 equiv), HOBtÆH2O (2.1 equiv), TEA (3.2 equiv), DMF (20 mL), 30 min,
0 ꢁC, then rt for 6 h; (e) TFA 50% in DCM, 30 min rt under N2 atmosphere. (Overall yield 30–50%).
Table 1. Binding affinities and in vitro activity
a
Binding IC50 (nM)
a
Bioassay IC50 (nM)
Drugs
d
l
MVD
27 1.5
GPI/LMMP
Biphalinb
2.6 0.4
1.4 0.2
8.8 0.3
6
2400 1000
400 98
640 44
8200 1800
2700 370
3010 1300
1.27 0.08
0.48 0.06
1.93 0.20
17% at 1 lM
8.1% at 1 lM
47% at 10 lM
35.6 6.4
43% at 20 lM
25.5% at 20 lM
61% at 20 lM
40 16.0
7
8
12
13
14
3.17 0.6
0.65 0.30
0.19 0.04
9.3 0.30
0.72 0.20
2.5 0.6
40 13.0
a
SEM.
b Data according to Ref. 7c.
bioassay potency reflecting the same pattern. Analogue
14 shows a 1:10 selectivity for the d versus l opioid
receptors in binding. This preference is more evident in
the bioassays where the bioactivity for the d receptors
is 50 times higher than at the l receptor. As compared
with the activity of biphalin, all the above reported
results show how a reduced degree of freedom between
the two pharmacophore moieties and their consequent
relative position can influence the binding affinity and
selectivity toward different receptors. In the case of the

A. Mollica et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2471–2475
2475
9. Han, So-Y.; Kim, Y.-A. Tetrahedron 2004, 60, 2447–
2467.
10. Selected data: 1: H NMR (600 MHz, DMSO-d6) d 1.18
shorter pharmacophore (compounds 6–8) the affinity
for the receptors is significantly reduced. The presence
of the classic pharmacophore (12–14) maintains or
improves the binding affinity. Whereas the CO–NH of
the linker fragment in compounds 6, 8, 12, and 14
should adopt the usually more favorable trans confor-
mation, this limitation is not present in compounds 7
and 13 which contain tertiary amide bonds at the two
piperazine nitrogen atoms and thus are free to choose
between equivalent conformers. The remarkable activity
of compound 13 leads to the hypothesis that the NH
moiety of hydrazine in biphalin is not related to the
binding at the opioids receptors. We can conclude that
the hydrazine linker is not fundamental for activity or
binding, and it can be conveniently substituted by differ-
ent conformationally constrained cycloaliphatic diamine
linkers. Future in vivo studies will give us more informa-
tion about the biological effects of this type of
modification.
1
(6H, m, D-Ala CH3 and CH3–CH2–O), 1.38 (9H, s, t-
butyl), 3.72–3.88 (2H, m, Gly CH2), 4.0 (1H, m, D-Ala a-
CH), 4.08 (2H, m, CH3ꢀCH2–), 6.90 (1H, d, D-Ala NH),
8.15 (1H, t, Gly NH). FAB-MS 296.93 m/e (M++Na). 2:
1H NMR (600 MHz, CDCl3 90% DMSO-d6 10%) d 1,26–
1,29 (6H, m, CH3CH2O– and D-Ala CH3), 1.49 (9H, s, t-
butyl), 2.89–2.98 (2H, m, Tyr b-CH2), 3.8–4.02 (2H, m,
Gly a-CH2), 4.16 (2H, m, CH3CH2O–), 4.23 (1H, m, Tyr
a-CH), 4.46 (1H, m, a-CH D-Ala), 5.69 (1H, br, Tyr NH),
6.75 and 7.01 (4H, dd, aromatic), 7.34 (1H, br, D-Ala
NH), 7.7 (1H, br, Gly NH), 8.71 (1H, s, Tyr OH).
FAB-MS m/e 460.01 (M++Na); MP 155–158 (EtOAc/
Et2O).
11. Selected data: 6: 1H NMR (600 MHz, H2O/D2O) d 2.83
and 2.95 (4H, m, Phe b-CH2), 2.90 (4H, m, Tyr b-CH2),
4.12 (2H, m, Tyr a-CH), 4.50 (2H, under water signal,
Phe a-CH), 6.75–7.32 (22H, m, aromatic), 8.58 (2H, m,
Phe NH), 9.73 (2H, s, NH-aromatic linker). FAB-MS m/e
729.3437 (M+). 7: 1H NMR (600 MHz, H2O/D2O) d
2.55–3.62 (8H, m, piperazine 4 · CH2), 2.68–78 (4H, m,
Phe b-CH2), 2.96–3.08 (4H, m, Tyr b-CH2), 4.09 (2H, m,
Tyr a-CH), 4.70 (2H, superimposed on H2O signal, Phe
a-CH), 6.78–7.34 (18H, m, aromatic), 8.48 (2H, m, Phe
NH). FAB-MS m/e 706.3557 (M+). 8: 1H NMR
(600 MHz, H2O/D2O) d 2.89–3.12 (8H, m, Phe b-CH2
and Tyr b-CH2), 4.09 (2H, m, Tyr a-CH), 4.75 (2H,
superimposed on H2O signal, Phe a-CH), 6.7–7.38 (22H,
m, aromatics), 8.60 (2H, m, Phe NH), 9.40 (2H, s,
Acknowledgements
This work was supported by a grant from the U.S, Pub-
lic Health Service, National Institute of Drug Abuse,
DA 06284.
References and notes
1
aromatic liker NH). FAB-MS m/e 729.3407 (M+). 12: H
1. (a) Lipkowski, A. W.; Koneka, A. M.; Sroczynska, I.
Peptides 1982, 3, 697–700; (b) Lipkowski, A. W.; Kon-
opka, M.; Osipiak, B.; Gumulka, W. S. In Peptides 1982;
NMR (600 MHz, H2O/D2O) d 1.05 (6H, d, D-Ala CH3),
2.9–3.12 (8H, m, Phe b-CH2 and Tyr b-CH2), 3.85 (4H,
m, Gly CH2), 4.08 (2H, m, Tyr a-CH), 4.15 (2H, m, D-
Ala a-CH), 4.65 (2H, superimposed on H2O signal, Phe a-
CH2), 6.75–7.40 (22H, m, aromatic), 8.02 (2H, m, Phe
NH), 8.32 (2H, m, Gly NH), 8.38 (2H, m, D-Ala NH),
9.58 (2H, s, aromatic linker NH). FAB-MS m/e 985.45
(M+). 13: 1H NMR (600 MHz, H2O/D2O) d 1.08 (6H, dd,
D-Ala CH3), 2.55–3.54 (8H, m, piperazine 4 · CH2), 2.85
and 2.99 (4H, m, Phe b-CH2), 3.01 and 3.10 (4H, m, Tyr
b-CH2), 3.8 (4H, m, Gly CH2), 4.09 (4H, m, Tyr a-CH
and D-Ala a-CH), 4,75 (2H, superimposed on H2O
signal, Phe a-CH), 6.82–7.35 (18H, m, aromatic), 8.03
and 8.07 (2H, 2d, Phe NH, two conformers), 8.30 (2H,
m, Gly NH), 8.4 (2H, m, D-Ala NH). FAB-MS m/e
´
Blaha, K., Malon, P., Eds.; de Gruyter: Berlin, 1983, pp
481–486.
ˇ
2. Horan, P. J.; Mattia, A.; Bilsky, E. J.; Weber, S. J.; Davis,
T. P.; Yamamura, H. I.; Malatynska, E.; Appleyard, S.
M.; Slaninova, J.; Misicka, A.; Lipkowski, A. W.; Hruby,
V. J.; Porreca, F. J. Pharmacol. Exp. Ther. 1993, 265,
1446–1454.
3. Abbruscato, J.; Thomas, S. A.; Hruby, V. J.; Davis, T. P.
J. Neurochem. 1997, 69, 1236–1245.
4. Abbruscato, T. J.; Williams, S. A.; Misicka, A.; Lipkow-
ski, A. W.; Hruby, V. J.; Davis, T. P. J. Pharmacol. Exp.
Ther. 1996, 276, 1049–1057.
5. Yamazaki, M.; Suzuki, T.; Nanta, M.; Lipkowski, A. W.
Life Sci. 2001, 69, 1023–1028.
1
963.48 (M+). 14: H NMR (600 MHz, H2O/D2O) d 1.05
(6H, d, D-Ala CH3), 2.9–3.12 (8H, m, Phe b-CH2 and Tyr
b-CH2), 3.75 (4H, m, a-CH2 Gly), 4.04 (4H, m, D-Ala a-
CH and Tyr a-CH), 4.75 (2H, superimposed on H2O
signal, Phe a-CH2), 6.79–7.33 (18H, m, aromatic), 8.02
(2H, m, Phe NH), 8.18 (2H, m, Gly NH), 8.32 (2H, m, D-
Ala NH), 9.28 (2H, s, aromatic linker NH). FAB-MS m/e
985.4546 (M+).
6. Hruby, V. J.; Davis, T. P.; Polt, R.; Bartosz-Bechowski,
H.; Misicka, A.; Lipkowski, A.; Sharma, S. D.; Li, G.;
Bonner, G., et al. In Peptides: Chemistry, Structure and
Biology; Kaumaya, P. T. P., Hodges, R. S., Eds.;
Mayflower Scientific: Kingswinford, 1996, pp 154–156.
7. (a) Misicka, A.; Lipkowski, A. W.; Horvath, R.; Davis, P.;
Porreca, F.; Yamamura, H. I.; Hruby, V. J. Life Sci. 1997,
60, 1263–1269; (b) Li, G.; Haq, W.; Xiang, L.; Lou, B.-S.;
Hughes, R.; De Leon, I. A.; Davis, P.; Gillespie, T. J.;
Romanowski, M.; Zhu, X.; Misicka, A.; Lipkowski, A.
W.; Porreca, F.; Davis, T. P.; Yamamura, H. I.; OꢀBrien,
D. F.; Hruby, V. J. Bioorg. Med. Chem. Lett. 1998, 8, 555–
560; (c) Lipkowski, A. W.; Misicka, A.; Davis, P.;
Stropova, D.; Janders, J.; Lachwa, M.; Porreca, F.;
Yamamura, H. I.; Hruby, V. J. Bioorg. Med. Chem. Lett.
1999, 9, 2763–2766; (d) Misicka, A.; Lipkowski, A. W.;
Stropova, D.; Yamamura, H. I.; Davis, P.; Porreca, F.;
Hruby, V. J. Peptide Sci. 1999, 726–727.
12. (a) Sklenar, V.; Piotto, M.; Leppik, R.; Saudek, V. J.
Magn. Reson., Ser A 1993, 102, 241–245; (b) Bax, A.;
Davis, D. G. J. Magn. Reson. 1985, 65, 355–360; (c)
Marion, D.; Ikura, M.; Bax, A. J. Magn. Reson. 1989, 84,
425–430.
13. (a) Polt, R.; Porreca, F.; Szabo, L.; Bilsky, E. J.; Davis, T.;
Horvath, R.; Abbruscato, T. J.; Yamamura, H. J.; Hruby,
V. J. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 7114–7118;
(b) Misicka, A.; Lipkowski, A. W.; Horvath, R.; Davis, P.;
Kramer, T. H.; Yamamura, H. I.; Hruby, V. J. Life Sci.
1992, 51, 1025–1032.
14. Kramer, T. H.; Davis, P.; Hruby, V. J.; Burks, T. F.;
Porreca, F. J. Pharmacol. Exp. Ther. 1993, 266, 577–
584.
8. Lipkowski, A. W.; Konecka, A. M.; Sroczynska, I.;
Przewlocki, R.; Stala, L.; Tam, S. W. Life Sci. 1987, 40,
2283–2388.