Discovery and synthesis of a new class of opioid ligand having a 3-azabicyclo[3.1.0]hexane core. An example of a 'magic methyl' giving a 35-fold improvement in binding

Article abstract of DOI:10.1016/j.bmcl.2011.05.132

In looking for a novel achiral μ opioid receptor antagonist for the treatment of pruritus, we designed and synthesised azabicyclo[3.1.0]hexane compounds as a new class of opioid ligand. During optimisation, an addition of a single methyl resulted in a 35-

Full text of DOI:10.1016/j.bmcl.2011.05.132

Bioorganic & Medicinal Chemistry Letters 21 (2011) 4608–4611
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery and synthesis of a new class of opioid ligand having
a 3-azabicyclo[3.1.0]hexane core. An example of a ‘magic methyl’
giving a 35-fold improvement in binding
⇑
Graham Lunn , Bernard J. Banks, Robert Crook, Neil Feeder, Alan Pettman, Yogesh Sabnis
Pfizer Global Research and Development, Sandwich Laboratories, Ramsgate Road, Sandwich CT13 9NJ, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
In looking for a novel achiral
synthesised azabicyclo[3.1.0]hexane compounds as a new class of opioid ligand. During optimisation, an
addition of a single methyl resulted in a 35-fold improvement in binding. An early example from the ser-
l opioid receptor antagonist for the treatment of pruritus, we designed and
Received 7 April 2011
Revised 24 May 2011
Accepted 25 May 2011
Available online 16 June 2011
ies had excellent
ritus study.
l
opioid receptor antagonist antagonist activity and was very effective in an in vivo pru-
Ó 2011 Elsevier Ltd. All rights reserved.
Keywords:
Opioid
Opiate
Pruritis
l
Opioid ligand antagonist
Rotational energy barrier
Pruritus is intense itch, often associated with flea allergy
dermatitis and atopic dermatitis in dogs. Excessive scratching in
response to itch can result in wounds then liable to secondary
bacterial or fungal infection, causing further inflammation and itch
sensation. Pruritus in dogs is a major cause for veterinary referral.
Current therapies include corticosteroids, antihistamines, essential
fatty acid dietary supplements and various emollients. However
there is a continuing need for alternative improved treatments of
pruritus. We desired a safe nonsteroidal agent which rapidly
blocked the itch sensation, suitable for once or twice a day oral
dosing.
Our design was influenced by literature which discussed the
4-phenylpiperidine
opioid ligand series pharmacology.4,5 The
l
evidence presented suggested that if the phenyl ring adopted an
axial position in a piperidine chair conformer then agonist pharma-
cology was observed, however if the phenyl ring adopted and pre-
ferred an equatorial position, then the compound acted as an
antagonist. The trans-3,4-dimethyl-4-(3-hydroxyphenyl)piperi-
dine template is described as having this phenyl equatorial pre-
ferred conformation. This lower energy equatorial conformation
was driven by the trans dimethyl substituents at the 3 and 4 posi-
tions. Deletion of the ‘3 methyl’ resulted in compounds with a
weak preference for phenyl axial conformation and some agonist
pharmacology.
We designed 3-azabicyclo[3.1.0]hexane targets (Fig. 2) with a
‘trans’ relationship across the cyclopropyl ring between the phenyl
ring and the fused five-membered ring to lock the molecule in the
proposed ‘equatorial antagonist’ geometry. This enabled deletion
of the methyl corresponding to the 3 substituent in the piperidine
series to produce a symmetrical achiral compound.
l
-Opiate antagonists have been reported to be antipruritic.1,2
We had demonstrated that examples from our lead 4-phenylpi-
peridine opioid series such as 1 (Fig. 1) produce rapid and
dramatic reduction in pruritic behaviour in dogs with flea allergy
dermatitis.3
As the treatment was indicated for the veterinary market, low
cost of compound manufacture was highly desirable. Due to a
lengthy synthesis of 1 the projected cost of manufacture was high.
Therefore we desired a novel more potent
l
-selective ligand, pref-
Two synthetic strategies were adopted to access the
substituted 3-azabicyclo[3.1.0]hexane templates; both paying
careful attention to the desired fused ring geometry. The first
was intended to enable variation on the basic nitrogen as late as
possible (Scheme 1).
erably achiral with a simpler more cost effective synthesis. The
compound would need to have no opiate agonism pharmacology
to avoid any abuse potential.
The key step (i) forming the cyclopropane ring was a rho-
dium(II)acetate catalysed reaction of ethyldiazoacetate with the
appropriate trans cinnamyl halide. Yields with the chlorides were
⇑
Corresponding author.
E-mail addresses: graham.lunn@pfizer.com, grahamlunn40@yahoo.co.uk
(G. Lunn).
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.05.132

G. Lunn et al. / Bioorg. Med. Chem. Lett. 21 (2011) 4608–4611
4609
O
NO2
NO2
S
NO2
NO2
N
N
N
O
ii.b
N
i
ii.a
H
N
1
N+
O
O
O
O
N
O
NNH2
NO2
N
Figure 1. 4-Phenyl piperidine series lead
l
opioid receptor antagonist.
NH2
.HCl
NHSO2Me
ii.c
iii.a,b,c
iv
CH3
CH3
3
O
O
MeO2SHN
N
N
N
MeO2SHN
N
N
R
R
CH3
CHIRAL
FUSED ACHIRAL
Scheme 2. Second synthetic strategy to key 3-azabicyclo[3.1.0] target 3. Reagents
and conditions: (i) H2NNH2, IMS, water, reflux, 82%, (ii.a) MnO2, dioxan, 20 °C, (ii.b)
dioxan 20 °C, (ii.c) reflux, 65%; (iii.a) NaBH4, BF3ÁO(Et)2, EtOAc, 0–65 °C, (iii.b) 5% Pd/
C, hydrogen (1 atm), EtOAc, 20 °C, (iii.c) HCl, 90%; (iv) MsCl.
Figure 2. Design of locked geometry in 3-azabicyclo[3.1.0]hexane core.
significantly higher than the bromides.6 This carbenoid addition
gave the (racemic) major product cyclopropane with desired ste-
reochemistry7: that is, with ‘retention’ of the starting cinnamyl ole-
fin geometry, and the ethylcarboxylate trans to the bulkier
aromatic ring. Ring closure with 3-phenylpropylamine formed
the racemic lactam. Reduction then sulfonylation afforded the final
target compounds 2 and 3.
This route was not particularly amenable to large scale up due
to the potentially hazardous ethyldiazoacetate step so the second
synthetic strategy (Scheme 2) was adopted and developed for bulk
scale up. This route forms the key cyclopropane ring of the bicycle
Table 1
Human and dog
l receptor binding for compounds 2 and 3, with binding for dog j
and d opiate receptor
Compds Human
l
Dog
l
Dog
j
Dog d binding
Kia (nM)
binding Kia
binding Kia
binding Kia
(nM)
(nM)
(nM)
2
3
168 ( 57.9)
4.79 ( 1.92)
379b
4.1 ( 2.3)
1576b
94 ( 35)
>10000b
183 ( 103)
a
Values are means of five or more experiments, standard deviation is given in
parentheses.
by the thermal extrusion of nitrogen from
intermediate.
a pyrazoline
b
Mean of two experiments.
1-(3-Nitrophenyl)-1-ethanone hydrazone is oxidised with man-
ganese dioxide to the diazo intermediate; this undergoes a 3+2
cycloaddition with 1-(3-phenylpropyl)-1H-pyrrole-2,4-dione to
form a pyrazoline intermediate; thermolysis results in an diradical
closure to form the cyclopropane ring with excellent selectivity for
the desired geometry—the bulkier phenyl trans to the succinimide
anti-nociception. This indeed demonstrated that this novel achiral
series could offer potent antagonist pharmacology.
What was quite remarkable in the comparison between com-
pounds 2 and 3, was that in making a single change from a hydro-
gen to a methyl, a binding affinity improvement of 35 was
l
ring. Reduction then sulfonylation afforded
quantities.
3 in multigram
Using this route the projected cost of bulk synthesis of 3 was a
quarter of that projected for 1. Furthermore later this route was
developed further enabling several versatile late stage intermedi-
ates suitable for making analogues in parallel.8 Compound 2 has
also been resynthesised using this methodology.
observed in the human
greater than 35-fold difference was observed in the
l
receptor—a truly magic methyl, and a
and d dog
l
receptors (note n = 2 for 2 in the dog binding assays). A methyl
group with its lipophilic surface well buried into a protein hydro-
phobic pocket is ‘expected’ to provide up to 1.5 kcal molÀ1 binding
energy and so give up to a 10-fold improvement in binding en-
ergy.13 In this case, comparing 2 with 3 something significantly
extra was occurring. We postulated that the added methyl
impacted the rotational freedom of the phenyl ring at the 6 posi-
tion on the 3-azabicylo[3.1.0] core, giving rise to a preferred lower
energy tortional angle, thus reducing rotational entropic loss on
binding. To understand this further we modelled the energy pro-
files of these systems with the rotation of this phenyl ring.14 The
flexible phenpropyl chain could adopt a variety of conformations.
In the modelling this was truncated to methyl so as to focus on
the energy differences due to the rotation of the phenyl ring. The
results were very interesting, and showed a distinct energy profile
difference between the compounds (Fig. 3).
Both 2 and 3 were potent l
-selective ligands (Table 1).9,10 2 was
not potent enough at the time to warrant any further investigation.
Pleasingly 3 confirmed as a full opiate receptor antagonist with a
pA2 of 7.97 against
l in an isolated guinea-pig myenteric plexus-
longitudinal experiment preparation.11 In the mouse tail flick mod-
el12 at 10 mg/kg s.c. 3 showed antagonism of morphine induced
NO2
NO2
ii
i
R
R
CO2Et
Br/Cl
NO2
Br/Cl
NH2
As
s varies the difference between the maximum and minimum
H
O
N
energy where R = H is 1.1 kcal molÀ1, suggesting unrestricted rota-
tion of this aromatic ring. However for R = Me this range is
3.99 kcal molÀ1, giving a higher barrier to rotation. For the range
of angles 45°–135° and 210°–315° the relative difference of
R = Me to R = H is up to 2.5 kcal molÀ1. If the tortional angle re-
quired for receptor binding exists in one of these regions, then with
such relative energy differences it seems reasonable to propose
that the methyl group does contribute to binding by influencing
the population of rotational isomers as well as lipophilic binding.
S
O
iii, iv
v
R
R
R
2 R= H
3 R=Me
O
N
N
N
Scheme 1. First synthetic strategy to key 3-azabicyclo[3.1.0] targets 2 and 3.
Reagents and conditions: (i) N2CH2CO2Et, DCM, Rh2(OAc)4, 20 °C; (ii) 3-phenylpro-
pylamine DMF, NaHCO3, 150 °C; (iii) LiAlH4 in THF, reflux; (iv) Raney Ni, hydrogen
(1 atm), EtOH, 20 °C; (v) MsCl, DCM, pyridine, 0–20 °C.
While tempting to postulate that the angle of
s in binding at the

4610
G. Lunn et al. / Bioorg. Med. Chem. Lett. 21 (2011) 4608–4611
pruritic behaviour in dogs with flea allergy dermatitis which lasted
NHSO2Me
for 12 hours (Fig. 5).16
τ
R
Compound 3 was mainly metabolized by the dog CYP2D15 en-
zyme acting as both substrate but unfortunately inhibitor. On dose
elevation 3 showed nonlinear pharmacokinetics and did not show
N
a
sufficient safety margin over side effects to warrant full
development.
Normalised energy profile vs tortional angle
τ
In summary a novel achiral 3-azabicyclo[3.1.0]hexane series of
2
1.5
1
0.5
0
-0.5
-1
-1.5
-2
l
opioid receptor ligands has been described. This series was de-
signed to be locked in a defined geometry, so that chiral substitu-
ents were not necessary to provide a clean opiate antagonist
profile. The difference in binding between analogues with hydro-
gen and methyl at the 6 position was 35-fold. We propose this is
due to lipophilic binding and the influence over the rotation of
the 6 position phenyl ring. The SAR in this series will be detailed
further.8
-2.5
-3
Tortional angle tau
Acknowledgments
R = H
R = Me
The authors thank Adrian Foster, Steve Gibson, Bill Howson,
David Johnson, Jinee Patel and Torren Peakman for their contribu-
tions to this work.
Figure 3. Normalised comparison of lowest energy state profile with variation in
between model compounds.
s
References and notes
1. Bergasa, N. V.; Talbot, T. L.; Alling, D. W.; Schmitt, J. M.; Walker, E. C.; Baker, B.
L.; Korenman, J. C.; Park, Y.; Hoofnagle, J. H.; Jones, E. A. Gastroenterology 1992,
102, 544.
2. Metze, D.; Reimann, S.; Beissert, S.; Luger, T. J. Am. Acad. Dermatol. 1999, 41,
533.
3. Armer, R.E.; Dutton, C. J.; Gethin, D. M.; Gibson, S. P.; Smith, J.D.; Tommasini, I.
PCT. Int. Appl., WO199959971, 1999.
4. Loew, G.; Lawson, J.; Toll, L.; Uyeno, E.; Frenking, G.; Polgar, W. NIDA Res.
Monogr. 1986, 75(Prog. Opioid Res.), 49.
5. Zimmerman, D. M.; Leander, J. D.; Cantrell, B. E.; Reel, J. K.; Snoddy, J.;
Mendelsohn, L. G.; Johnson, B. G.; Mitch, C. H. J. Med. Chem. 1993, 36, 2833.
6. First attempts with cinnamyl bromides gave poor isolated yields (11%) of
desired product (A) and an equal amount of side product olefin (C).
Figure 4. X-ray crystal structure of HCl salt monohydrate of 3.
Ar
Ar
receptor is close to that of the phenol ring in classical ‘rigid’ ligands
such as morphine or naltrexone, one truly cannot be sure.
The structure of 3 was further confirmed by X-ray crystallogra-
phy of the hydrochloride salt.15 (Fig. 4) It is interesting to note that
Ar
X = Br, Cl
+
LnRh CHCO2Et
+
X
EtO2C
X
EtO2C
X
A
B
( X = Br, 11%)
( X = Cl, 30 - 40%)
( X = Br, 1%)
( X = Cl, 3%)
for the crystal structure
s = 103°.
Further NMR studies on various salts were unable to reveal any
clear stereo isomeric differences at the charged nitrogen atom be-
tween compounds 2 and 3.
On further profiling 3 was negative in the Ames and CHO micro-
nucleus genetic toxicology in vitro assays. Its pharmacokinetic
profile in rat and dog showed it suitable for oral dosing. With a
dose of 2 mg/kg po 3 produced a rapid and dramatic reduction in
Ar
Ar
C
+
X
C
( X = Br, 11%)
( X = Cl, 2%)
EtO2C
EtO2C
X
This arose from a side reaction where the carbenoid species reacted on the
bromine to form an unwanted ylide intermediate (see Doyle, M. P.; Tamblyn,
W. H.; Bagheri, V. J. Org. Chem. 1981, 46, 5094), which underwent
a
2,3-sigmatropic rearrangement giving (C). This problem was circumvented by
using the cinnamyl chloride analogue. Chlorine being less nucleophilic than
bromine did not form the ylide intermediate so readily, resulting in higher
yield, 40% of (A) with only 3% (C). Only traces of the non-desired isomer (B)
were detected and easily removed by column chromatography.
7. For all intermediates in both synthetic routes, the coupling constants of the
protons in the cyclopropyl ring were entirely consistent with the desired
stereochemistry. For example, where R = H the coupling constant for this with
the neighbouring ‘trans’ protons trans in the cyclopropyl ring were always
below 4.5 Hz; In 3 J = 3.2 Hz. Furthermore, 2D ROESY data further confirms
relative stereochemistry.
Effect of 3 at 4mg/kg & 2mg/kg p.o.
on elapsed scratching time in 12 FAD dogs
600.0
Placebo
4mg/kg
500.0
2mg/kg
400.0
300.0
200.0
100.0
8. Manuscript in preparation.
9. Dog receptor binding protocols for l, j and d described in Banks, B. J.; Crook, R.
J.; Gibson, S. P.; Lunn, G.; Pettman, A. J. PCT. Int. Appl., WO20039089, 1999.
10. Human binding Ki determined from [3H]DAMGO filter binding assay.
11. Guinea-pig myenteric plexus-longitudinal muscle strips were prepared
following the procedure from Henderson, G.; Hughes, J.; Kosterlitz, H.W. Br. J.
Pharmacol. 1975, 53, 505, attached to an isometric transducer and an electrode.
The tissues were then placed under a 1 g tension and left to equilibrate for 1 h,
washed with Kreb’s solution every 15 min, maintained at 37 °C throughout
with constant supply of 95% O2/5% CO2. After equilibration, the tissues were
rebalanced and then electrically stimulated using frequency of 0.1 Hz and
1msec pulse width. When all the tissues were stabilised, supramaximal voltage
was determined. This voltage was used for the rest of the experiment. The
0.0
0-240 61-300 121-360 181-420 241-480 301-540 361-600 421-660 481-720
Time (min)
Figure 5. Antipruritic activity of 3 in dogs with flea allergy dermatitis. Showing
mean of 12 FAD dogs SEM.

G. Lunn et al. / Bioorg. Med. Chem. Lett. 21 (2011) 4608–4611
4611
tissues were then washed and left for 20 min. 1
nor-Binaltorphimine was then added and left for 10 min. 1
l
M of the kappa antagonist
13. Kuntz, I. D.; Chen, K.; Sharp, K. A.; Kollmann, P. A. Proc. Natl. Acad. Sci. U.S.A.
1999, 96, 9997.
lM of captopril, a
peptidase inhibitor, was added to prevent the breakdown of DAMGO. When
the twitches plateaued, a dose range of compound 3 was added. After a further
14. Energies were calculated using implicit models for solvent and quantum
mechanics; the self-consistent reaction field (SCRF) model using B3LYP/6–
31G⁄⁄ as implemented in Jaguar7.5 release 207. The SCRF method is able to
model both aqueous and nonaqueous solvents, which permits the calculation
of both gas to water and gas to octanol solvation energies. Jaguar 7.5 release
207 Schrodinger: Portland, OR, 2008. See: http://www.schrodinger.com/.
15. Cambridge Crystallographic Data Centre deposition number CCDC 819853.
16. A colony of mixed breed flea allergic dogs was challenged with fleas. Pruritic
behaviour was recorded on video tape and evaluated by trained observers.
Duration of pruritic activity is determined by measuring time spent in actual
pruritic behaviour (scratching, rubbing, and licking) and is recorded as elapsed
time.
10 min, a cumulative dose response was constructed for the
l agonist DAMGO
(each concentration 3 min contact time with 1 min recording). pA2 for 3 = 7.97
( 0.26 std. dev) based on n = 3.
12. Seven male ICR mice (17–19 g) were injected with vehicle (5% DMSO:5%
Emulphur :90% Saline) or test compound (s.c.). Fifteen minutes later, each was
injected with vehicle (saline) or morphine sulfate (3.2 mg/kg: s.c.). Thirty
minutes later analgesia was assessed via tail flick. Latency to remove tail from a
heated light source was recorded in seconds (12 s max). Data are presented as
a
%
reversal from Veh/Morp [(Test compound mean À Vehicle/Morphine)/
(Vehicle/Morphine À Vehicle/Vehicle)]. Compound 3 gave 89% at 10 mg/kg s.c.