An efficient enantiopure synthesis of a pivotal precursor to substance P antagonists

Article abstract of DOI:10.1021/op050213e

Many substance P antagonists have a core structure based on the quinuclidine skeleton. Manufacture of these drug antagonists proceeds through the advanced intermediate (2S,3S)-cis-2-benzhydryl-3-aminoquinuclidine 1, and all previous syntheses of 2S,3S-1 p

Full text of DOI:10.1021/op050213e

Organic Process Research & Development 2006, 10, 142−148
An Efficient Enantiopure Synthesis of a Pivotal Precursor to Substance P
Antagonists1
Thomas C. Nugent*,† and Robert Seemayer‡
Department of Chemistry, School of Engineering & Science, International UniVersity Bremen, Campus Ring 1,
28759 Bremen, Germany, and CV Therapeutics, Inc., Pharmaceutical DeVelopment and Manufacturing,
3172 Porter DriVe, Palo Alto, California 94304, U.S.A.
Abstract:
drug candidates, larger quantities of (2S,3S)-1 were required
(Schemes 1 and 2). A review of the literature showed the
most efficient route originating from the hydrochloride salt
of quinuclidinone 2.2-4 To appreciate the chemistry involved
we repeated the racemic synthesis of primary amine 1
following a literature preparation, which began with an aldol
condensation of benzaldehyde and quinuclidinone 2 (Scheme
1).4e,c The resulting enone 3 was readily isolated from the
Many substance P antagonists have a core structure based on
the quinuclidine skeleton. Manufacture of these drug antago-
nists proceeds through the advanced intermediate (2S,3S)-cis-
2-benzhydryl-3-aminoquinuclidine 1, and all previous syntheses
of 2S,3S-1 proceed through quinuclidinone 2. The synthesis
described herein provides a 40% improved synthetic yield of
(2S,3S)-1 from quinuclidinone 2, when compared to all previ-
ously reported syntheses. The key process improvements
originate from: (1) dynamic kinetic resolution of ketone rac-4,
producing ketone (2S)-4 and (2) the subsequent reductive
amination of ketone (2S)-4 without epimerization. The former
dynamic kinetic resolution, the first demonstrated for this
ketone (quinuclidinone) architecture, uses inexpensive (natural)
Scheme 1. Previous racemic synthesis of primary amine 1
L-tartaric acid to provide ketone (2S)-4 in high yield (90%) and
enantiopurity (95% ee). The latter demonstrates the first use
of Ti(OiPr)4/Pt-C/H2 for reductive amination and is especially
noteworthy for its ability to preserve the r-labile C2 stereo-
center of ketone (2S)-4. The new reductive amination method
is general in nature and should find broad applicability.
aqueous reaction medium by simple Bu¨chner funnel filtra-
tion. Reaction of this geometrically pure enone with phen-
ylmagnesium chloride yielded the Michael addition product,
benzhydryl ketone rac-4, as a well-behaved solid. Further
treatment with benzylamine and camphorsulfonic acid (cat.)
over 24 h enabled imine formation under the azeotropic
conditions of refluxing toluene (Dean-Stark apparatus). The
reaction is high yielding, but the resultant imine 5 is sensitive
Introduction
Substance P is a naturally occurring undecapeptide
belonging to the tachykinin family of peptides.2 Quinuclidine
derivatives are prepared for use as substance P receptor
antagonists and treat many ailments ranging from gas-
trointestinal and central nervous system (psychotic) disorders
to inflammatory diseases, pain, and migraine.3 Their curative
effects are perhaps unsurprising, since quinuclidine deriva-
tives, both naturally occurring and synthetic, have long been
associated with physiological properties.4e A brief review of
the medical literature confirms the intense research effort
invested in the use of these compounds and further reveals
that many of these drugs contain the 3-amino-2-benzhydryl
quinuclidine scaffold 1.4 A vast amount of literature regarding
their synthesis is contained within pharmaceutical patents.5,7
(3) A process for preparing a quinuclidine derivative. Lowe, J. A. (Pfizer Inc.,
India). Patent number: IN173570, 1994.
(4) (a) Swain, C. J.; Seward, E. M.; Cascieri, M. A.; Fong, T. M.; Herbert, R.;
MacIntyre, D. E.; Merchant, K. J.; Owen, S. N.; Owens, A. P.; Sabin, V.;
Teall, M.; VanNiel, M. B.; Williams, B. J.; Sadowski, S.; Strader, C.; Ball,
R. G.; Baker, R. J. Med. Chem. 1995, 38, 4793. (b) Lowe, J. A., III; Drozda,
S. E.; Snider, R. M.; Longo, K. P.; Zorn, S. H.; Morrone, J.; Jackson, E.
R.; McLean, S.; Bryce, D. K.; Bordner, J.; Nagahisa, A.; Kanai, Y.; Suga,
O.; Tsuchiya, M. J. Med. Chem. 1992, 35, 2591. (c) Warawa, E. J.; Mueller,
N. J.; Fleming, J. S. J. Med. Chem. 1975, 18, 587. (d) Warawa, E. J.;
Mueller, N. J.; Gylys, J. A. J. Med. Chem. 1975, 18, 71. (e) Warawa, E. J.;
Mueller, N. J.; Jules, R. J. Med. Chem. 1974, 17, 497.
(5) (a) Norris, T.; Santafianos, D.; Bordner, J. J. Chem. Soc., Perkin Trans. 1
1997, 3679 and patent citations therein. (b) Quinuclidine compound and
medicinal use thereof. Shu, M.; Hiroshi, K.; Masahiko, K.; Hiroshi, Y.
(Yoshitomi Pharmaceutical). Patent number: WO9309116, 1993.
(6) In our hands the stereoselectivity (cis/trans ratios) ranged from 20:1 to 1:1,
depending on the reductant employed: NaB(OAc)3H > BH3 = Pt/C >
NaBH4. The use of 9-BBN has been reported to provide excellent cis product
ratios, see ref 4b.
Discussion and Results
Assessment of Prior Syntheses and Resolutions. To
advance the clinical development programs of several lead
(7) For a classical resolution of rac-1 see: (a) Production of optically active
3-amino-2-benzhydrylquinuclidine compound. Osamu, M.; Hiroshi, K.
(Yoshitomi Pharmaceutical). Patent number: JP7025874, 1995. For a
classical resolution of the o-methoxybenzyl derivatives of rac-6, see: (b)
Process and intermediates for preparing azabicyclo[2.2.2]octane-3-imines.
Godek, D. M.; Murtiashaw, C. W. (Pfizer. Inc.). U.S. Patent 5,138,060,
1992. (c) Resolution of 1-azabicyclo[2.2.2]octane-3-amine,2-(diphenyl-
methyl)-N-[[2-methoxy-5-(1-methylethyl)phenyl]methyl]. Tickner, D. L.;
Meltz, M. (Pfizer, Inc.). Patent number: WO9703984, 1997.
* To whom correspondence should be addressed. E-mail: t.nugent@
iu-bremen.de.
† International University Bremen.
‡ CV Therapeutics, Inc., Pharmaceutical Development and Manufacturing.
(1) This research was performed at Catalytica Pharmaceuticals, 430 Ferguson
Drive, Mountain View, CA 94043, U.S.A.
(2) Quinuclidine derivatives. Ito, F.; Kondo, H.; Nakane, M.; Shimada, K.;
Lowe, J. A.; Rosen, T. J. (Pfizer Inc.). U.S. Patent 5,807,867, 1998.
142
•
Vol. 10, No. 1, 2006 / Organic Process Research & Development
10.1021/op050213e CCC: $33.50 © 2006 American Chemical Society
Published on Web 12/23/2005

to moisture, thus we found it beneficial to reduce it without
further purification.5 A mixture of cis and trans rac-6
resulted, and depending on the reductant employed, high
diastereoselectivity (>20:1, cis/trans) could be achieved for
racemic cis-6.6 This level of diastereoselectivity was en-
couraging, but our experimental results and knowledge of
the reported literature made it evident that no one reductant
was ideal when considering the combined outcome of
diastereoselectivity, yield, and rate of reaction. Finally, the
title compound, primary amine rac-1, was cleanly produced
upon hydrogenolysis of rac-6 with Pd/C in methanolic
hydrogen chloride (Scheme 1).4a
Identification of Weak Synthetic Points. Our improved
route for the production of primary amine (2S,3S)-1 would
ultimately come from identifying the strengths and weak-
nesses of the known syntheses. These points can be sum-
marized as follow. Alternatives exist for the incorporation
of a benzhydryl moiety, but we found the demonstrated two-
step process satisfactory both in ease of preparation and
purification (Scheme 1). Next we questioned the need for
stepwise imine formation, isolation, and then reduction. A
significant decrease in processing time and a possible
increase in yield might be realized if ketone 4 could be
reductively aminated.
Two literature methods in particular are representative and
descriptive for detailing the removal of the minor trans
enantiomers (2R,3S and 2S,3R of 6 or 1) and simultaneously
resolving the cis enantiomers (2S,3S and 2R,3R) (Scheme
2). The first is a two-step process, requiring derivatization
Although the last points concerned us, we identified the
choice of resolving an advanced intermediate, e.g. rac-6 or
-1, as the most critical factor contributing to the low overall
yield of all previous syntheses. Resolutions of these late-
stage intermediates, while respectable in yield, were per-
formed when a large amount of labor and expense had
already been invested. As a consequence the overall yield
of all previous syntheses were significantly compromised,
since the undesired enantiomer could not be recycled and
thus became part of the waste stream.
Scheme 2. Literature methods for the isolation of primary
amine (2S,3S)-1
An early stage resolution could impart a significant
economic benefit, but no literature precedent could be
established. Thus we were intrigued by the possibility of
performing a static or better yet a dynamic kinetic resolution
(DKR) of racemic benzhydryl ketone 4. An immediate
beneficial consequence of a dynamic kinetic resolution would
be a theoretical doubling of the yield, a benefit that could
not be ignored. In the event such a resolution could be
realized, we faced one more significant hurdle, preservation
of the R-keto stereocenter upon introduction of nitrogen, i.e.
could enantiopure benzhydryl quinuclidinone (2S)-4 [ketone
(2S)-4] be further elaborated to (2S,3S)-6 without epimer-
ization. In our final analysis we deemed the benefits of such
an approach outweighed the downside of lost research time
in the event of failure.
Finally, the source of nitrogen, benzylamine, was ques-
tioned. Reductive amination with ammonia would provide
the product directly from ketone (2S)-4, obviating the need
for a debenzylation step. While it was an interesting idea,
we were uncertain of the diastereoselectivity (cis/trans)
associated with the reduction of a non-N-alkylated imine
intermediate and our ability to mitigate further alkylation of
the desired reaction product primary amine-1.8 In the end,
benzylamine would prove to be a convenient source of
nitrogen.
of secondary amine rac-6 with (S)-(+)-1-(1-naphthyl)ethyl
isocyanate.4b The desired diastereomer is insoluble in the
reaction solvent toluene and filtration allows unfettered
isolation. The resulting urea derivative (S,S,S)-7 is then
hydrolyzed with concentrated H2SO4, yielding amine (2S,3S)-
6. This product is assumed to be of high enantiopurity
because further recrystallization did not improve the mea-
sured optical rotation of the material ([R]D +67.4 (c ) 1.0,
DMSO)).4b
The second method is of greater utility, requiring only a
static (classical) resolution of racemic amine 1 with unnatural
D-(-)-tartaric acid (1.0 equiv) in hot methanol as described
in a Japanese patent.7 Filtration at room temperature provided
a 44% yield of the desired diastereomeric salt (98.5% pure,
97% de), further treatment with aqueous alkali and extraction
with methylene chloride provide the free amine(2S,3S)-1
(97% ee). We did not examine either of these methods of
resolution.
Despite the good resolution properties of rac-1, the most
efficient synthesis remained too costly to be economically
viable for large-scale production. From a process perspective
the synthesis of (2S,3S)-1 had several shortcomings,
namely: (1) resolution of an advanced intermediate; (2) use
of expensive resolving agents; (3) stepwise imine formation,
isolation, and reduction thereof; and (4) mediocre, when
considered in total, cis/trans diastereoselectivity, yield, and
reaction time for the imine reduction step. Additionally, any
potential solution would have to be amenable to scale-up
and avoid expensive reagents and chemistries requiring low
temperatures, and preferably have solid intermediates for ease
of purification. This manuscript describes our approach to
the aforementioned challenges, which were in large part met.
Dynamic Kinetic Resolution of Benzhydryl Ketone 4.
Overcoming the need for a late-stage static resolution for
the production of (2S,3S)-1 was the key and tactical point
we focused on. A literature search revealed that no attempts
at early-stage resolutions or early- or late-stage dynamic
kinetic resolutions had been reported for the synthesis of
quinuclidine derivatives. Benzhydryl quinuclidinone rac-4
met our requirements for an early-stage dynamic kinetic
resolution study, and to investigate this possibility we
(8) Over alkylation is probably not a concern, see: (a) Miriyala, B.; Bhatta-
charyya, S.; Williamson, J. S. Tetrahedron 2004, 60, 1463. (b) Hirayama,
Y.; Ikunaka, M.; Matsumoto, J. Org. Process Res. DeV. 2005, 9, 30.
Vol. 10, No. 1, 2006 / Organic Process Research & Development
•
143

prepared kilogram quantities of rac-4 without event.4c By
doing so, we were able to liberally examine the dynamic
kinetic resolution properties of rac-4 with a variety of
enantiopure, inexpensive, and commercially available car-
boxylic acids on a tens of gram scale. An initial screen
of economically viable chiral carboxylic acids soon led
to the observation that natural L-tartaric acid allowed (2S)-4
to precipitate as its tartrate acid salt in 80-85% yield
(92-94% ee) from hot 2-propanol. A static (classical)
resolution provides a maximum 50% yield of one enan-
tiomer; thus, our yield and ee clearly demonstrated that a
dynamic kinetic resolution had been achieved. The dynamic
kinetic nature of this resolution was further supported by
analysis of the mother liquor of the resolution reaction after
isolation of the tartrate salt. It showed that the remaining
ketone-4, in the solvent, is indeed a racemate. Analysis of
the precipitated tartrate salt showed the molar ratio between
ketone-4 and L-tartaric acid to be 1:1.
As initially observed, L-tartaric acid is efficient at
selectively precipitating only the (2S)-4 L-tartrate salt, but it
is rather slow at converting the undesired (2R)-4 enantiomer
into the (2S)-4 enantiomer even in refluxing 2-propanol (>24
h). Adding more equivalents of an appropriate acid, to
accelerate the rate of racemization, could mitigate this
problem. Keeping the cost of the process in mind, we then
studied the effect of adding a commodity carboxylic acid,
e.g., acetic acid.9 This and other 1-alkanoic acids served the
stated purpose and additionally had the desired property of
not precipitating salts of ketone-4.
After further experimentation and optimization we found
that L-tartaric acid (1.0 equiv) with acetic acid (1.0 equiv)
in refluxing ethanol efficiently converted ketone rac-4 to the
(2S)-4-L-tartrate salt in 85-90% yield (94-96% ee) within
12 h. These reaction conditions allowed timely removal of
the desired enantiomer, i.e. precipitation of the (2S)-4
L-tartrate salt, from the solution-phase racemization equi-
librium.
The enantioenriched ketone (2S)-4 was obtained from its
salt form by suspending it in an organic solvent (e.g. toluene,
ethyl acetate, or methyl tert-butyl ether) and then dropwise
adding an alkaline aqueous base (NaHCO3, Na2CO3 or
NaOHspotassium salts are also acceptable) until the pH of
the aqueous layer was approximately 9 ((0.5). Generally,
the temperature was maintained below 25 °C during the
neutralization. This is particularly important if a hydroxide
base is used because elevated temperatures lead to deteriora-
tion of the optical purity of ketone (2S)-4. The free base of
(2S)-4 is recovered as a solid after concentration of the
organic extract. Using this procedure and utilizing sodium
hydroxide as the base we produced several 0.5 kg batches
of benzhydryl ketone (2S)-4 in near quantitative yield and
with uncompromised enantiopurity (free ketone ee ) 95%)
without event.
we were baffled to find a total lack of precedent for reductive
amination in the quinuclidine literature. Due to the time
restraints imposed, we needed to examine the dynamic kinetic
resolution of ketone rac-4 and its reductive amination
chemistries in parallel. To do so, we found it useful to use
the readily prepared R-deuterated ketone 4, as a surrogate
for (2S)-4, at the beginning of our reductive amination
research effort (Scheme 3). All previous researchers chose
2
Scheme 3. Use of H-rac-4 as a probe for assessing the
propensity of (2S)-4 to epimerize during reductive amination
to incorporate the exoskeletal nitrogen in a stepwise fashion
(i.e., imine formation, isolation, then reduction), thus we
initiated a one-pot reductive amination feasibility study using
R-deuterated benzhydryl ketone 4 (2H-rac-4). Using standard
1H NMR experiments we were able to measure the level of
deuterium remaining in the subsequent product (secondary
amine 2H-6) and thus drew immediate conclusions concern-
ing the propensity of (2S)-4 to epimerize during reductive
amination.
A variety of reductive amination protocols exist in the
literature and are procedurally similar in that a Brønsted acid
is used to facilitate “imine” formation, while a coexisting
reductant reduces the imine. Using the standard literature
conditions, we examined the reductive amination of 2H-rac-4
in THF with benzylamine (1.1 equiv), NaB(OAc)3H (1.4
equiv), and AcOH (1.0 equiv).10 After 11 h the desired
product was obtained in 40% yield after purification by
column chromatography (Scheme 3). Extending the reac-
tion time (6 d) and/or adding additional AcOH and/or
NaB(OAc)3H provided only marginal improvements in yield.
That said, a significant improvement in yield occurred when
the initial conditions were as follows: BnNH2 (1.5 equiv),
NaB(OAc)3H (2.5 equiv), and AcOH (1.7 equiv) in THF
(0.4 M), with heating at 35 °C for 25 h (Table 1, entry 1).
Unfortunately, these Brønsted acid conditions allowed gross
exchange of the deuterium label that we were trying to
preserve.
Next we investigated the ability of a Lewis acid,
Ti(OiPr)4, in combination with a borohydride-based reducing
agent, to reductively aminate 2H-rac-ketone 4. This method
of reductive amination was first demonstrated by Mattson
et al. and calls for the neat reaction of a ketone, an amine
and Ti(OiPr)4 in approximately equal molar quantities.11,12
After 1 h of stirring, EtOH and NaBH3CN are added,
allowing amine product formation. The method has since
been championed by Bhattacharyya, using the reductant
NaBH4,13,14 and is noted for its ketone/amine substrate
breadth and tolerance of several spectator functional groups.15
Introduction of Nitrogen to Benzhydryl Quinuclidi-
none (2S)-4. Intent on reductively aminating ketone (2S)-4,
(10) 1,2-Dichloroethane was the solvent of choice for most of their reactions,
although they mention that THF, CH2Cl2, or CH3CN may be also be used,
see: Abdel-Magid, A. F.; Carson, K. G.; Harris, B. D.; Maryanoff, C. A.;
Shah, R. D. J. Org. Chem. 1996, 61, 3849.
(11) Mattson, R. J.; Pham, K. M.; Leuck, D. J.; Cowen, K. A. J. Org. Chem.
1990, 55, 2552.
(9) We have recently become aware that this general strategy has been employed
before: see specifically p 803 of Breuer, M.; Ditrich, K.; Habicher, T.;
Hauer, B.; Kesseler, M.; Stu¨rmer, R.; Zelinski, T. Angew. Chem., Int. Ed.
2004, 43, 788.
144
•
Vol. 10, No. 1, 2006 / Organic Process Research & Development

Table 1. Ti(OiPr)4 vs AcOH-catalyzed reductive amination
with NaB(OAc)3H
Table 2. Ti(OiPr)4-mediated reductive aminations with
various borohydrides
temp time product 6
temp
(°C)
% loss
yield
(%)
catalyst (°C) (h)
(%)a
% 2H in 6b cis/trans of 6c
borohydride
of 2H in 6
cis/trans
AcOH
35
25
25
11
70
85
28
51
16:1
17:1
NaB(OAc)3H
NaBH4
NaBH4
22
22
0
41
0
0
17:1
6:4
1:1
6:5
85
a
a
Ti(OiPr)4
NaBH3CN
22
1
a
a
b
c
Isolated yields after chromatography. Starting ketone 94% 2H. 1H NMR
data.
a The chromatographic yield was not determined for this reaction, but the
mass recovery was good and the 1H NMR of the crude product showed only
product.
Aware of the aforementioned reductants, in combination
with Ti(OiPr)4, we first decided to investigate the use of
NaB(OAc)3H as a reductant.16 Our initial experiments
employed Ti(OiPr)4 (1.2 equiv), BnNH2 (1.1 equiv), and
NaB(OAc)3H (2.3 equiv), producing the secondary amine 6
in <11 h at room temperature. Again the reductive amination
product showed gross loss of deuterium at C2 (Table 1). It
is valuable to note that both methods provided very respect-
able and very similar cis/trans diastereoselectivity. While
we found the procedure using Ti(OiPr)4 more reliable and
convenient, the goal of enantiopreservation (or retention of
deuterium in this case) was clearly not feasible using this
combination of reagents.
none of the borohydrides we examined were able to
simultaneously provide nonepimerizing conditions, adequate
cis/trans diastereoselectivity, and sufficient rate of reaction.
We then turned our attention to the use of hydrogen as a
reductant. Our first experiment was as follows: ketone (2S)-4
(1.0 mmol, 94% ee), THF, BnNH2, and Ti(OiPr)4 were stirred
for 2.5 h and subsequently hydrogenated at 50 psi with Pt/C
(25 wt %). The reaction yielded the product in 75% yield,
91% ee, and with a respectable cis/trans ratio of 7:1.17,18
Further refinements (see Experimental Section) increased
the diastereomeric yield (85%), improved the diastereose-
lectivity (9:1, cis/trans), and maintained a high ee (g92%)
for the desired cis enantiomer.19,20 This was possible while
also reducing the amount of heterogeneous hydrogenation
catalyst required. A large number of experiments were
performed and can be summarized by stating that: (1) Pt/C
(10 wt %, 60% water content) is a superior hydrogenation
catalyst21 for this ketone substrate regarding the yield and
enantiomeric excess of (2S,3S)-6; (2) Pd/C provides excellent
diastereoselectivity (cis/trans ratios as high as 40:1), but the
reactions failed to go to completion,22 and lower optical
purities (50-75% ee) were observed with increasing reaction
Next we examined the use of NaBH4 and NaBH3CN as
reductants for the titanium (IV)-mediated method. Thus
2
reaction of H-rac-4 with Ti(OiPr)4 (1.2 equiv) and BnNH2
(1.1 equiv) in THF (1.0 M) for 2.5 h, followed by removal
of THF (rotary evaporator), addition of EtOH (0.25 M) and
NaBH4 or NaBH3CN (2.0 equiv) allowed the smooth
formation of amine 6 (Table 2). No byproducts were formed
(1H NMR of crude product), and significantly the deuterium
label was fully preserved when using either of these
reductants. Unfortunately, the cis/trans diastereoselectivity
was poor at best, and in the case of NaBH4 decreased on
cooling (Table 2, entry 3). These results were edifying and
clearly established that Ti(OiPr)4 was not responsible for the
earlier observed deuterium/hydrogen exchange (epimeriza-
(17) We simultaneously found that a stepwise imine formation/isolation/reduction
pathway was possible. Thus, the 2H-rac-4 was condensed with benzylamine
under the conditions of (1) catalytic CSA acid with azeotropic water removal,
(2) neat at 145 °C under N2, (3) alumina (basic) or anhydrous MgSO4 in a
dry THF. The first two conditions provided the imine, albeit with gross
loss of deuterium, and the last two reactions failed to provide imine, even
under forcing conditions, i.e., higher temperature. Despite these setbacks
we did find that imine formation was possible without loss of deuterium
using Ti(OiPr)4 at room temperature (2.5 h), followed by workup with sat.
NaHCO3 or H2O. Yields varied from 60 to 95%; this likely reflects the
short workup times required to avoid imine hydrolysis but which are
suspected of being insufficient to free all of the product from the titanium
salts. Workup using NaOH (1.0 M) led to a 29% decrease in the percent of
deuterium.
2
tion) at C2 of ketone H-rac-4.
A New Method for Reductive Amination: Synthesis
of Primary Amine (2S,3S)-1. Our earlier research results
revealed that reductive amination catalyzed by Ti(OiPr)4 is
strongly dependent upon the nature of the reductant. Yet,
(12) Similar Ti(IV) methods using TiCl4 have been described; in our hands these
methods proved unhelpful, see: (a) Barney, C. L.; Huber, E. W.; McCarthy,
J. R. Tetrahedron Lett. 1990, 31, 5547. (b) Johansson, A.; Lindstedt, E.-L.;
Olsson, T. Acta Chem. Scand. 1997, 51, 351.
(18) All crucial previous experiments with 2H-rac-4 were reexamined using (2S)-
4; no inconsistencies were observed.
(13) For examples of reductive amination using Ti(OiPr)4 and NaBH4 see: ref
8a and (a) Neidigh, K. A.; Avery, M. A.; Williamson, J. S.; Bhattacharyya,
S. J. Chem. Soc., Perkin Trans. 1 1998, 2527. (b) Bhattacharyya, S. J. Org.
Chem. 1995, 60, 4928. (c) Bhattacharyya, S. J. Chem. Soc., Perkin Trans.
1 1995, 1845. (d) Bhattacharyya, S. Synth. Commun. 1995, 25 (1), 9. (e)
Bhattacharyya, S. Tetrahedron Lett. 1994, 35, 2401. (f) Bhattacharyya, S.;
Chatterjee, A.; Williamson, J. S. Synlett 1995, 1079.
(19) The new reductive amination protocol described herein, Ti(OiPr)4/H2/Pt-
C, has been used internally at Catalytica Pharmaceuticals (and later at DSM
and Pfizer) since 1998 as a trade secret. The patent application process
took its own course as mergers and acquisitions occurred. Process for the
preparation of (S,S)-cis-2-benzhydryl-3-benzylaminoquinuclidine: Nugent,
T. C.; Seemayer, R. (Pfizer Products, Inc. and DSM Pharmaceuticals, Inc.).
Patent number: WO2004035575, 2004.
(20) Related research has since been published, see: (a) Nugent, T. C.;
Wakchaure, V. N.; Ghosh, A. K.; Mohanty, R. R. Org. Lett. 2005, 7, 4967.
(b) Alexakis, A.; Gille, S.; Prian, F., Rosset, S.; Ditrick, K. Tetrahedron
Lett. 2004, 45, 1449.
(21) Further attempts to improve the cis/trans selectivity proved to be futile,
e.g. (1) hydrogenation at 0 °C, (2) use of Pt on alternative supports, e.g.
alumina powder or sulfide carbon, and (3) prereduction of the Pt/C catalyst.
(22) Even with high Pd catalyst loadings and higher pressure (e.g., 125 psi) and/
or heating, yields in the range of 10-35% were observed. Nevertheless,
since the reason for incomplete reaction using Pd/C is not understood, further
catalyst screening should be considered.
(14) For an example of reductive amination using Ti(OiPr)4 and polymethylhy-
drosiloxane (PMHS), see: Chandrasekhar, S.; Reddy, R. C.; Ahmed, M.
Synlett 2000, 1655.
(15) For substrate breadth and compatible functional groups, e.g. carbamates,
urethanes, tertiary amides, acetonides, silyl ethers, esters, etc., see: refs 8a,
13a,f, and Seebach, D.; Hungerbuehler, E.; Naef, R.; Schnurrenberger, P.;
Weidmann, B.; Zueger, M. Synthesis 1982, 138.
(16) At the time of the following research, 1998, the use of NaB(OAc)3H in
combination with Ti(OiPr)4 was unreported. It has since been reported, but
the researchers additionally added AcOH, see: Breitenbucher, J. G.; Hui,
H. C. Tetrahedron Lett. 1998, 39, 8207.
Vol. 10, No. 1, 2006 / Organic Process Research & Development
•
145

Table 3. Effect of water on the product yield profile (HPLCa
Table 5. Effects of solvent and cosolvent on reaction
data)b
profilea
H2O
(equiv)
(2S,3S)-6
(2S,3R)-6
(2R,3R)-6
(2S,3S)-6 (2S,3R)-6 (2R,3R)-6
entry
(%)
(%)
(%)
ee
entry solvent additiveb
(%)
(%)
(%)
ee
1
2
3
4
5
6
none
0.4
0.6
1.2
1.8
3.0
75.7
77.7
80.6
80.2
79.7
77.7
14.3
9.0
8.8
8.4
8.2
8.0
1.58
3.37
3.20
4.57
4.49
4.87
96
92
92
89
89
88
1
THF
MeOH
77.9
12.1
1.68
96
(25 equiv)
2
3
toluene none
toluene H2O
(3 equiv)
toluene HOiPr
(6 equiv)
67.2
72.2
8.6
6.3
1.69
3.00
95
92
4
68.9
8.6
1.68
95
a Ultron ES OVM HPLC column (chiral) and Waters Symmetry C8 HPLC
column (nonchiral) data. a 1.0 mmol scale, 5% Pt/C (10 wt %, 1-4% water
content)
a Ultron ES OVM HPLC column (chiral) and Waters Symmetry C8 HPLC
column (nonchiral) data. b Additives added at start of hydrogenation.
Table 6. Debenzylation of (2S,3S)-6a
entry
solvent
acid/equivalents
time (h)
conversionb
1
2
3
4
5
CH3OH
CH3OH
CH3OH
H2O
p-TsOH/1.05
CH3SO3H/1.05
HCl/1.05
HCl/1.05
HCl/2.4
9.0
12
12
14
6.5
99
100
100
52
Figure 1. Possible diastereomeric amine products from reduc-
tive amination of ketone (2S)-4 with benzylamine.
Table 4. Reaction yield and optical purity versus Ti(OiPr)4
H2O
100
stoichiometrya
a All reactions 1 mmol scale, 1/10 wt % of 5% Pd/C (50% wet), 50 ( 5 °C,
50 ( 5 Psi H2, 0.25 M. b First aliquot showing 100% conversion or last aliquot
examined.
Ti(OiPr)4
(equiv)
yield
(%)
entry
ee
1
2
3
4
5
1.05
1.1
78
80
79
83
36
91
90
89
93
79
with Pd(OH)2 or Pd/C in the presence of an acid resulted in
incomplete reaction. Next we examined the hydrogenolysis
at elevated temperature. Regardless of the acid employed
the debenzylations were quantitative.23 From a manufacturing
point of view, methanesulfonic acid was more attractive than
HCl because it allows more flexibility in the potential
equipment trains to be used.
1.2
1.3
0.34
a Ultron ES OVM HPLC column (chiral) and Waters Symmetry C8 HPLC
column (nonchiral) data; 60 psi, all reactions stopped at 10 h.
time; (3) control of the water content allows the overall yield
of (2S,3S)-6 to be maximized (Table 3, Figure 1). Finally,
(4) lower grade Ti(OiPr)4 (Aldrich, TYZOR 97%, 2.0 L )
$109.00, neat) works equally well as highly purified
Ti(OiPr)4 (99.999%). Stoichiometric quantities are required
for complete reaction (Table 4).
Conclusion
The synthesis outlined in Scheme 4 has enabled us to
increase the yield of primary amine (2S,3S)-1 by 40% when
compared to the previous best synthetic methods reported
for its synthesis. Paramount to achieving this overall yield
increase was the identification of racemic benzhydryl qui-
nuclidinone 4 for dynamic kinetic resolution studies, that
culminated in an efficient method for the conversion of
ketone rac-4 to (2S)-4. All previous resolution strategies used
advanced intermediates that were not capable of dynamic
kinetic resolution.
Another avenue to change the diastereoselectivity of
the reduction could potentially arise from the use of alter-
native solvents and/or cosolvents. A very brief study of
these factors follows. Addition of 25 equiv (based on the
limiting reagent: ketone (2S)-4) of anhydrous CH3OH, a
protic solvent, essentially left the product profile unchanged,
compare entry 1 (Table 5) with entry 1 (Table 3), the same
result as if no water had been added. Replacing THF with
toluene slowed the reaction down, and the addition of further
cosolvents did not improve the reaction profile (Table 5).
The crude product 6 could be recrystallized from a
toluene/n-hexane mixture once and consistently provided the
product (2S,3S)-6 in 61% yield with >99.5% chemical purity
(HPLC) and g99% de and g99% ee (chiral HPLC, see
Experimental Section). Interestingly, the use of hexanes or
heptanes, in place of n-hexane, proved deleterious, resulting
in a slightly poorer impurity profile that did not meet our
target specifications for (2S,3S)-6.
Scheme 4. Synthesis of (2S,3S)-1 from racemic ketone 4
Secure in our ability to produce reliable batches of
enantiopure (2S,3S)-6, we next examined its debenzylation
(Table 6). Reaction of amine (2S,3S)-6, at room temperature,
146
•
Vol. 10, No. 1, 2006 / Organic Process Research & Development

Introduction of the exoskeletal nitrogen was not trival
because it required preservation of the existing R-labile (2S)
stereocenter of ketone (2S)-4, good cis diastereoselectivity,
and good yield for the amine product (2S,3S)-6. To maximize
these desired attributes, we developed a new method for
diastereoselective reductive amination [Ti(OiPr)4/Pt-C/H2].19
This reagent combination is a powerful tool for reductively
aminating R-chiral ketones with retention of stereochemical
information. To our knowledge this is the first example of
such a feat, albeit no negative examples were found in the
literature. Finally, this new method is noteworthy because it
demonstrates the first reported use of Ti(OiPr)4/Pt-C/H2 for
reductive amination. Previously, chemists had only examined
the use of hydride reagents11,13,14 in combination with
Ti(OiPr)4 for reductive amination.20
in 2-propanol (20 mL), dilute with mobile phase to 100 mL.
Retention time (min). (2S)-ketone-4: 7.45; (2R)-ketone-4:
8.02.
Chiral HPLC analysis for racemic cis- and trans-6: Ultron
ES-OVM column; dimensions: 4.6 mm × 150 mm (add
guard column); mobile phase: pH 6.2 buffer/MeCN, 82:18
(v/v) [buffer prepared as follows: dissolve KH2PO4 (1.36
g) in H2O (950 mL), adjust pH to 6.2 with H3PO4, dilute to
1000 mL with H2O]; flow rate: 1.0 mL/min; detection
wavelength: 210 nm; temperature: 25 °C; injection vol-
ume: 20 µL; sample preparation: sonicate 20 mg of sample
in MeCN (20 mL), then dilute to 100 mL with mobile phase.
Retention time (min). (2S,3S)-6 (cis desired): 16.61; (2R,3R)-6
(cis undesired): 23.44.
(2S)-Benzhydryl-3-quinuclidinone 4 L-Tartaric Acid
Salt. Racemic 2-benzhydryl-3-quinuclidinone 4 (52.45 g, 180
mmol) was dissolved in denatured ethanol (5% methanol,
5% 2-propanol) (525 mL, 0.35 M) with acetic acid (10.4
mL, 180 mmol), and L-tartaric acid (27 g, 180 mmol) was
added. The mixture was heated to reflux for 12 h and then
allowed to cool to room temperature and held for 1 h. The
solids were filtered, collected, and dried under vacuum at
40 °C for 12 h. The desired 1:1 L-tartaric acid salt of (2S)-
ketone 4 weighed 69.9 g (88% yield).
The aforementioned combined advances, as outlined
above, now allow the efficient large-scale synthesis of amine
(2S,3S)-1 for the synthesis of quinuclidine derivatives.
Experimental Section
General Remarks. The NMR spectra were recorded on
1
a Brueker instrument at 300 MHz for H and 75 MHz for
13C, with tetramethylsilane as an internal standard. All
melting points were measured on a capillary melting point
apparatus and are uncorrected. All reactions were performed
under a positive pressure of nitrogen. All chemicals and
reagents were purchased and used without further purification
unless otherwise mentioned. Dry solvents were used as
purchased.
(2S)-Benzhydryl-3-quinuclidinone 4. The L-tartaric acid
salt from the previous example (69.9 g, 158 mmol) was
suspended in toluene (700 mL) and cooled with an ice-
water bath while a saturated solution of sodium bicarbonate
(500 mL) was added dropwise. Note the use of NaOH avoids
the often noted foaming associated with the use of bicarbon-
ate. The temperature was maintained below a maximum of
25 °C. The clear, biphasic mixture was stirred for 20 min at
25 °C, and the layers were separated. The organic layer was
washed with water (100 mL), the layers were separated, and
the organics were dried over sodium sulfate. The organics
were filtered and evaporated in vacuo to provide the desired,
optically active ketone as a colorless solid (45.66 g; 99%
HPLC Methods. Reaction progress was monitored by
HPLC with purities being determined by peak area percent.
A Hewlett-Packard (Agilent) 1100 with variable UV/Vis
detector was used for all analyses.
HPLC analysis for compounds racemic-4 and racemic
cis- and trans-6: Waters Symmetry C8 column; dimen-
sions: 3.9 mm × 150 mm; mobile phase: pH 5.0 buffer/
MeCN ) 65:35 (v/v) [buffer prepared as follows: dissolve
NH4OAc (1.93 g) in H2O (950 mL), add Et3N (1.0 mL) and
adjust pH to 5 with HOAc, dilute to 1000 mL with H2O];
flow: 1.0 mL/min; detection wavelength: 220 nm (band-
width 8 nm), ref 360 nm (bandwidth 100 nm); temperature:
23 °C; injection volume: 20 µL/mL; sample preparation:
sonicate 20 mg of sample in MeCN (20 mL), dilute with
mobile phase to 100 mL. Retention time (min). rac-4: 8.5
min; (2S,3R)- and (2R,3S)-6 (enantiomeric trans products):
15.3; (2S,3S)- or (2R,3R)-6 (enantiomeric cis products): 16.4.
Chiral HPLC analysis for (2R)- or (2S)-ketone-4: Chiral-
Pak-AD column; dimensions: 4.6 mm × 250 mm (and guard
column); mobile phase: 95 vol % hexane, 5 vol % 2-pro-
panol, 0.1 vol % diethylamine; flow: 1.0 mL/min; detection
wavelength: 220 nm; temperature: 20 °C; injection vol-
ume: 10 µL/mL; sample diluent: 20 vol % 2-propanol, 80%
mobile phase. Sample preparation: sonicate 20 mg of sample
1
yield). Mp ) 145-146 °C. H NMR (300 MHz, CDCl3) δ
1.86-2.00 (m, 4H), 2.41-2.43 (m, 1H), 2.54-2.59 (m, 2H),
3.08 (t, 2H), 3.98 (d, 1H), 4.55 (d, 1H), 7.17 (m, 8H), 7.38-
7.41 (m, 2H).
(2S)Benzhydryl-(3S)-benzylamino-quinuclidine
6
[(2S,3S)-6]. With Aluminum Tri-isoproxide. Under nitrogen,
(2S)-benzhydryl-3-quinuclidinone 4 (0.50 g, 1.0 equiv, 1.72
mmol) was dissolved in anhydrous THF (2.0 mL). Benzyl-
amine (0.21 mL, 1.1 equiv, 1.89 mmol) was then added
followed by a solution of aluminum isopropoxide (0.42 g,
1.2 equiv, 2.06 mmol) in anhydrous THF (2.0 mL). The
solution was stirred for 3 h. To this colorless solution was
then added a slurry of 5% Pt/C (0.063 g, Degussa F101RA/
W, ∼60% wet) in anhydrous THF (1.0 mL). The reaction
was placed in a Parr reactor, pressurized to 75 psi H2, and
allowed to react at room temperature for 15 h. The reaction
mixture was poured into 15 mL of 2.0 M HCl, followed by
filtration, basification with 1 M NaOH and extraction with
50 mL of methyl tertiary butyl ether (MTBE). The MTBE
layer was dried with MgSO4, followed by removal of solvent
in vacuo, leaving a white crystalline solid. This was analyzed
(23) Transfer hydrogenation conditions were briefly examined for hydrogenolysis.
The most useful reaction was one in which a methanolic solution (0.30 M)
of (2S,3S)-6 had ammonium formate (4.0 equiv) and 5% Pd/C (1/6 wt %,
50% wet) added to it. When this solution was heated at 50 °C for 2.5 h, a
70% conversion was noted. An additional 3 h at the increased temperature
of reflux was required for full conversion.
Vol. 10, No. 1, 2006 / Organic Process Research & Development
•
147

as all cis isomer (<2% trans isomer), >99% ee (none of
other enantiomer observed).
hydrochloric acid. The mixture was agitated for 30 min at
900 rpm and again pressure transferred into an Erlenmeyer
flask. The combined biphasic heterogeneous solution was
filtered through a 1-cm Celite pad under vacuum to remove
the Pt/C catalyst. The filter cake was further rinsed with
aqueous 10% HCl (100 mL). The clear filtrate phase
separated immediately, and the organic layer was removed
and discarded. Under stirring and cooling, 50 mL of toluene
was added and the pH adjusted to approximately 13 by slow
addition of 50% NaOH (30 mL). The biphasic slurry was
filtered through a 1-cm Celite pad to remove titanium salts.
The filter cake was washed with toluene (2 × 50 mL), the
layers were then separated, and the toluene layer was
concentrated at 80 °C until the volume of toluene was
reduced to 20 mL. Then, 40 mL of n-hexane was added,
and the mixture was slowly cooled to 10 °C over 2-3 h
(0.5 g of seeds (∼ 5%) was added at 55 °C). The precipitate
was filtered, washed with 40 mL toluene/n-hexane 1/6
(v/v), and dried in a vacuum at 40 °C. The yield of colorless
solid was 7.3 g, 61% of theory.4a,c This material matched
the reported spectroscopic data (1H and 13C NMR) and was
further analyzed by achiral HPLC (>99% de) and chiral
HPLC (>99% ee).
With Titanium Tetra-isoproxide. (2S)-Benzhydryl-3-qui-
nuclidinone 4 (9.00 g, 30.9 mmol) was dissolved in
anhydrous THF (75 mL, 0.40 M). The solution was
transferred through a port to a 300-mL autoclave with the
hydrogenation head secured while a positive flow of nitrogen
was maintained. Through the same port on the hydrogenator
head and under 300 rpm stirring was added benzylamine (3.7
mL, 33.9 mmol) followed by titanium (IV) isopropoxide
(10.9 mL, 36.9 mmol). The port was closed, and the
autoclave was pressure tested (90 psi nitrogen) while the
reaction mixture was stirred at 300 rpm. After 3.0 h at 25
°C the N2 pressure was released, and under a positive flow
of nitrogen was added a slurry of 5% Pt/C (1.13 g; 59.4%
wet) in THF (3.0 mL) via syringe (14-gauge needle) through
the port. Additional THF (2 mL) was used to slurry the
residual catalyst and was added to the reaction. The port was
closed and the autoclave pressurized to 75 psi with hydrogen
and then slowly vented. This was repeated three times. The
final hydrogen pressure was adjusted to 75 psi and the
reaction mixture hydrogenated overnight (12 h) with stirring
maintained at 600 rpm. The vessel was then vented and
subsequently pressurized with nitrogen (90 psi) and vented.
The reactor was pressurized with nitrogen and vented three
more times.
Acknowledgment
In this period of fast development timelines, we greatly
appreciate the time supportive environment encouraged at
Catalytica, and later at DSM and Pfizer, Inc. for creative
thinking.
Under positive nitrogen flow 42 mL of ice-cold 12.4%
hydrochloric acid (28 mL water + 14 mL 37% HCl) was
added slowly and the reaction mixture stirred under nitrogen
for 1 h at 25 °C and 900 rpm and subsequently pressure
transferred into a 250-mL Erlenmeyer flask. The hydroge-
nator was charged with toluene (50 mL) and 30 mL of 10%
Received for review October 23, 2005.
OP050213E
148
•
Vol. 10, No. 1, 2006 / Organic Process Research & Development