Enantioselective synthesis of nicotinic receptor probe 7,8-difluoro-1,2,3, 4,5,6-hexahydro-1,5-methano-3-benzazocine

Article abstract of DOI:10.1021/ol0623062

(Chemical Equation Presented) The development of a concise enantioselective synthesis of nicotinic alkaloid 1 is presented. The route features the synthesis and use of a "stable" aliphatic triflate 21 in an alkylation step to generate Heck precursor 24 and an enantioselective cyclization to establish a compound with the key [3.2.1]-bicyclic core, 29.

Full text of DOI:10.1021/ol0623062

ORGANIC
LETTERS
2006
Vol. 8, No. 26
5947-5950
Enantioselective Synthesis of Nicotinic
Receptor Probe 7,8-Difluoro-1,2,3,4,5,6-
hexahydro-1,5-methano-3-benzazocine
Crystal G. Bashore, Michael G. Vetelino, Michael C. Wirtz, Paige R. Brooks,
Heather N. Frost, Ruth E. McDermott, David C. Whritenour,* John A. Ragan,
Jennifer L. Rutherford, Teresa W. Makowski, Steven J. Brenek, and
Jotham W. Coe*
Pfizer Global Research and DeVelopment, Groton Laboratories, Pfizer Inc.,
Groton, Connecticut 06340
jwcoe@pfizer.com
Received September 19, 2006
ABSTRACT
The development of a concise enantioselective synthesis of nicotinic alkaloid 1 is presented. The route features the synthesis and use of a
“stable” aliphatic triflate 21 in an alkylation step to generate Heck precursor 24 and an enantioselective cyclization to establish a compound
with the key [3.2.1]-bicyclic core, 29.
Naturally occurring nicotinic receptor ligands such as
nicotine, epibatidine, anatoxin A, cytisine, and their deriva-
tives have been increasingly studied as potential therapeutic
agents for cognition, pain, attention-deficit/hyperactivity
disorder, schizophrenia, Parkinson’s disease, and addiction.1
Our ongoing pursuit of cholinergic probes necessitated the
development of an efficient synthesis of the bicyclic deriva-
tive 7,8-difluoro-1,2,3,4,5,6-hexahydro-1,5-methano-3-ben-
zazocine (1). Our early approaches, designed for generality,
allowed the synthesis of multiple pharmacological targets
and utilized triflate intermediates.2 These compounds were
accessed from alkoxy precursors that were incorporated into
the initial synthetic design because they displayed robust
directing effects in metalation strategies for controlling
substituent regiochemistry. Herein, we describe refinements
that obviated their use and resulted in a brief enantioselective
synthesis of 1.
The original approach to racemic 1 required multiple steps
to prepare Heck precursor 7 (Scheme 1).3 Metalation of 2
with LDA (1.1 equiv) gave an ortho-lithio anisole intermedi-
ate, and subsequent low-temperature reaction with aldehyde
3 successfully established the required benzylic C-C bond
to provide 4.4 Reductive dehydration, anisole deprotection,5
and triflate formation gave Heck cyclization precursor 7.
Although the overall yield for this sequence was acceptable
(∼30%), we believed that the early stages of the synthesis
could be improved by a more judicious choice of starting
material.
We considered 3,4-difluorobromobenzene 12 to be an ideal
alternative. Metalation of 12 gave an anion that reacted
(3) Coe, J. W. Arylfused Azapolycyclic Compounds. PCT Int. Appl.
WO99 55,680, 1999. U.S. Patent 6,462,035, October 8, 2002, and U.S.
Patent 6,706,702, March 16, 2004.
(4) The ortho-lithio anisole intermediate failed to react with Weinreb
amides at low temperature and, upon warming, presumably suffered from
benzyne-mediated decomposition before any addition to the amide.
(5) Brooks, P. R.; Wirtz, M. C.; Vetelino, M. G.; Rescek, D. M.;
Woodworth, G. F.; Morgan, B. P.; Coe, J. W. J. Org. Chem. 1999, 64,
9719-9721.
(1) For leading reviews in this area, see: (a) Bunnelle, W. H.; Dart, M.
J.; Schrimpf, M. R. Curr. Med. Chem. 2004, 4, 299-334. (b) Holladay, M.
W.; Dart, M. J.; Lynch, J. K. J. Med. Chem. 1997, 40, 4169-4194. (c)
Glennon, R. A.; Dukat, M. Med. Chem. Res. 1996, 465-486.
(2) Coe, J. W.; Vetelino, M. G.; Bashore, C. G.; Wirtz, M. C.; Brooks,
P. R.; Arnold, E. P.; Lebel, L. A.; Fox, C. B.; Sands, S. B.; Davis, T. I.;
Rollema, H.; Schaeffer, E.; Schulz, D. W.; Tingley, F. D., III; O’Neill, B.
T. Bioorg. Med. Chem. Lett. 2005, 15, 2974-2979.
10.1021/ol0623062 CCC: $33.50
© 2006 American Chemical Society
Published on Web 11/29/2006

that benzyne formation and subsequent decomposition of
2-metalated 3,4-difluoro-1-bromobenzene species would
compete with the alkylation process. We found that activated
cyclopentene carbinol derivatives of 20 (e.g., Cl, Br, I, OMs)
failed to react with aryllithium 23 at low temperature
(Scheme 3). Warming the lithio species invariably promoted
Scheme 1
Scheme 3
decomposition, presumably as a result of benzyne formation
under the reaction conditions; the alkylating agents were
stable and recovered unchanged. Triflate 21, however, proved
uniquely suited to this alkylation,9 giving advanced Heck
intermediate 24 in one step from 3,4-difluorobromobenzene
12 (Scheme 3). Our emphasis, therefore, shifted to this
promising methodology, which was further developed to
support multikilogram-scale preparations.
with aldehyde 3 to afford benzyl alcohol 13.6 This intermedi-
ate performed well in the Heck reaction to give an 85:15
diastereomeric mixture of products (Scheme 2).7
Although alcohol 20 is known in the literature,10 a number
of modifications were introduced to support pilot plant
processing requirements and to avoid the isolation of oils or
low-melting solids (Scheme 4). The base-mediated alkylation
Scheme 2
Scheme 4
Conversion of 14 to piperidine 16 proceeded uneventfully
as shown in Scheme 2. In the conversion to 1, the removal
of the hindered benzylic alcohol during hydrogenolysis did
not occur without prior activation as the corresponding
mesylate or bromide.
Removing the benzylic hydroxyl group detracted from this
sequence and prompted the study of alkylation methods to
introduce the critical benzylic C-C bond. Although there
are examples of direct alkylation of aryl anion species flanked
by exchangeable halides, as with 12,8 we were still concerned
of 25 kg of dimethyl malonate (25) with cis-1,4-dichloro-
2-butene (26) was accomplished with LiH between 25 and
40 °C in 9:1 (v/v) THF-1,3-dimethyl-3,4,5,6-tetrahydro-2-
(1H)-pyrimidinone (DMPU). Although exothermic, the di-
alkylation was relatively slow and easily controlled. The
resulting diester was saponified directly by the addition of
aqueous LiOH to give diacid 27. After workup, the decar-
(6) Lithiated 3,4-difluorobromobenzene failed to react with the corre-
sponding Weinreb amide.
(7) In theory, with optically enriched alcohol, chiral induction in the Heck
cyclization step should be possible.
(8) Hayan, S. E.; Domagala, J. M.; Heifetz, C. L.; Johnson, J. J. Med.
Chem. 1991, 34, 1155-1161.
(9) Bashore, C. G.; Samardjiev, I. J.; Bordner, J.; Coe, J. W. J. Am.
Chem. Soc. 2003, 125, 3268-3272. Vedejs, E.; Engler, D. A.; Mullins, M.
J. J. Org. Chem. 1977, 42, 3102-3113.
(10) (a) Depre´s, J.-P.; Greene, A. E. J. Org. Chem. 1984, 49, 928-931.
(b) Depre´s, J.-P.; Greene, A. E. Org. Synth. 1997, 75, 195-200. (c) Johnson,
C. R.; Keiser, J. E.; Sharp, J. C. J. Org. Chem. 1969, 34, 860-864.
5948
Org. Lett., Vol. 8, No. 26, 2006

boxylation proceeded smoothly in toluene at 80-120 °C to
give monoacid 28. A 9:1 (v/v) THF-toluene solution of 28
was added to 1 M LiAlH4/THF at 20-25 °C. After an
aqueous workup, alcohol 20 was isolated by fractional
distillation (81 °C/29 mmHg) in 66% overall yield.
Typically, the generation of trifluoromethane sulfonates
involves aqueous workup procedures. We suspected that
under these conditions trace amounts of triflic acid, either
from the workup or from a slower adventitious hydrolysis
of unreacted triflic anhydride and trace levels of water, could
compromise the stability of the resulting trifluoromethane
sulfonate esters. In the Schleyer procedure, for example, the
conversion of alcohol 20 to triflate 2111 invariably led to
solutions with dramatic differences in stability from batch
to batch. Such instability was unsuitable for further scale-
up, and a protocol to eliminate residual triflic anhydride/
acid from the isolated reaction solution was developed
(Scheme 3). The reaction of pyridine and triflic anhydride
in hexanes generated a sparingly soluble slurry of pyridinium
triflate salt 19, effectively consuming the highly organic-
soluble triflic anhydride. This salt is an effective agent for
converting alcohol 20 to triflate 21.
Palladium-mediated Heck cyclization reactions of 24 with
Pd(OAc)2/P(o-tol)3 provided racemic material in 77% yield.
An enantioselective process to prepare olefin 29 was
subsequently developed using chiral ligands. From initial
screenings,
2,2′-bis(diphenylphosphino)-1,1′-binaphthyl
(BINAP), Pd(OAc)2, and DMF emerged as attractive com-
ponents for this reaction (Scheme 5). Under controlled
Scheme 5
On a range of scales, treatment of the slurry of 19 with
alcohol 20 generated hexane-soluble triflate 21 and the
pyridinium hydrogen triflate salt. All salts (19 and 22) were
conveniently removed by filtering the reaction mixture
through a pad of activity I neutral alumina. These conditions
effectively maintained anhydrous processing conditions while
eliminating residual triflic anhydride, the suspected source
of adventitious triflic acid. This process routinely gave
colorless solutions of 21 that were stable at 0 to -15 °C for
8-16 h without observable decomposition. The improved
stability was effectively established using differential scan-
ning calorimetry measurements (see Supporting Information).
Samples prepared by this nonaqueous Tf2O/pyr method had
a much higher decomposition onset temperature (202 °C)
than products derived from the Schleyer methodology (46
°C).
Confident in our ability to prepare acceptably stable
solutions of triflate 21, we began investigating the alkylation
of 2-lithio-3,4-difluoro-1-bromobenzene 23 (Scheme 3).
Anion formation was effected with LDA in THF at -60 to
-78 °C. The lithiation was indirectly monitored by in situ
IR spectroscopy by observing the appearance/disappearance
of absorbance bands attributable to diisopropylamine. On the
basis of the IR data, upon addition of LDA, proton transfer
was immediate, and the resulting solutions showed little
change over the next 2 h (IR). The triflate 21/hexanes
solution was then added to the aryllithium solution at a rate
that maintained the internal temperature below -70 °C
(Scheme 3). This procedure provided good conversion to the
desired product, as indicated by the ratio of product to
4-bromo-1,2-difluorobenzene (from anion protonation): by
NMR, 24/12 ratios were typically 15-35:1 in yields of 70-
92%.
conditions on the gram scale, high levels of enantioselectivity
[enantiomeric excess (ee) > 93%] were seen using 1-10
mol % Pd(OAc)2/BINAP levels.12 Purging and sparging with
N2 were found to be critical to achieve high levels of
enantioselectivity. When the reaction was insufficiently
purged, the rate slowed and the ee was reduced to 60% (see
Supporting Information). This latter result was observed
during a single large-scale run (51% yield, 60% ee on an
18-kg scale) and reflects the importance of effective purging
of process reactors.
Although the catalytic dihydroxylation of olefin 29 could
be effectively accomplished with osmium tetroxide, concern
over the monitoring and control of residual osmium led us
to use an alternate reagent during scale-up. We found that
stoichiometric potassium permanganate performed well for
this conversion, as reported by Ogino and Mochizuki.13An
aqueous NaHSO3 workup followed by a filtration of the
organic phase through two weight equivalents of silica gel
with dichloromethane-ethyl acetate provided diol 30 as an
oil, which crystallized upon standing (42-78% yield). The
structure shown in Scheme 5 was obtained by X-ray crystal
analysis of 30, confirming that dihydroxylation had occurred
from the less-hindered exo face.
(11) (a) Su, T. M.; Sliwinski, W. F.; Schleyer, P. v. R. J. Am. Chem.
Soc. 1969, 91, 5386-5388. (b) Stang, P. J.; Hanack, M.; Subramanian, L.
R. Synthesis 1982, 85-126. (c) Bashore, C. G.; Samardjiev, I. J.; Bordner,
J.; Coe, J. W. J. Am. Chem. Soc. 2003, 125, 3268-3272.
(12) Gas-phase calculations of carbopallidation adducts before â-hydride
elimination were performed, but the results did not align with the observed
sense of enantiofacial induction.
(13) Ogino, T.; Mochzuki, K. Chem. Lett. 1979, 443-446.
Org. Lett., Vol. 8, No. 26, 2006
5949

free base 31, and conversion to D-tartrate salt 33 completed
the synthesis of the desired active pharmaceutical ingredient
(Scheme 6).
Scheme 6
We have described an efficient enantioselective synthesis
of 1 using a six-step sequence. A series of process improve-
ments were introduced that supported an effective large-scale
synthesis. Adopting an alkylation strategy eliminated the need
to remove a pendant hydroxyl group thus streamlining the
process from the original approach. Key observations were
made that enhanced our understanding and supported large-
scale processing. First, we found that lithiation of 12 by LDA
is rapid, and this allowed us to remove any unnecessary
reaction hold times. This, in conjunction with a novel
procedure to produce stable solutions of the highly reactive
triflate 21, provided a consistently reproducible procedure
for the critical alkylation C-C bond-forming step to produce
24. Studies of the BINAP/Pd(OAc)2-mediated Heck cycliza-
tion reaction revealed that greatly improved enantioselec-
tivities to product 29 result from control of adventitious
oxygen. Resolution with dibenzoyl-L-tartaric acid effectively
enhanced the chiral purity of 31, and final cleavage of the
benzyl protecting group provided 1, which was isolated as a
D-tartrate salt.
Conversion of diol 30 to benzyl amine 31 (Scheme 5) was
accomplished by initial cleavage of the diol, with sodium
periodate in buffered methanol, to form the dialdehyde
(which existed as a mixture of hydrated cyclic hemiacetals
1
by H NMR analysis). The dichloromethane extracts from
this reaction were added to a cold dichloromethane slurry
containing benzyl amine (2 equiv) and NaBH(OAc)3 (3.3
equiv) to complete the N-benzyl piperidine formation, 31.
The reaction was quenched after 1-2 h; prolonged reaction
times (>12 h) generated an impurity corresponding to
N-benzylacetamide by MS analysis.
To enrich the desired enantiomer from racemic and
nonracemic mixtures, a dibenzoyl-L-tartaric acid resolution
with 31 was developed. The formation of salt 32 using
isopropyl ether served as an initial purification step. Two
additional reslurries in ethyl acetate drove the chiral purity
to 94% ee. Hydrogenolysis cleaved the benzyl amine to give
Acknowledgment. The authors would like to thank Scot
Mente for gas-phase calculations and Jon Bordner for X-ray
analyses.
Supporting Information Available: Experimental pro-
cedures and spectral data for all new compounds, differential
scanning calorimetry, and enantiomeric ratio experiments.
This material is available free of charge via the Internet at
http://pubs.acs.org.
OL0623062
5950
Org. Lett., Vol. 8, No. 26, 2006