Process research and development of an NK-1 receptor antagonist. Enantioselective trifluoromethyl addition to a ketone in the preparation of a chiral isochroman

Article abstract of DOI:10.1021/op7001886

CJ-17,493 (4) is a chiral NK-1 receptor antagonist. It was first prepared through a diastereoselective crystallization, then through chiral chromatography of a key intermediate, and ultimately via asymmetric synthesis. Multiple routes for the preparation of a key isochroman were demonstrated, and conditions for improved regioselectivity of a Friedel-Crafts acylation were identified. Cesium fluoride was found to be an acceptable initiator for the generation of a nucleophilic trifluoromethyl anion from CF₃TMS. A cinchonine-derived catalyst was identified for the enantioselective addition of the trifluoromethyl group to the ketone, and it was found that the product of the addition would be converted directly to the isochroman by treatment with t-BuOK. A Duff reaction was used for the formylation, and the resulting aldehyde was coupled to amine 5 to afford CJ-17,493 (4).

Full text of DOI:10.1021/op7001886

Organic Process Research & Development 2007, 11, 1015–1024
Process Research and Development of an NK-1 Receptor Antagonist.
Enantioselective Trifluoromethyl Addition to a Ketone in the Preparation of a Chiral
Isochroman
Stéphane Caron,* Nga M. Do, Janice E. Sieser, Patrice Arpin, and Enrique Vazquez†
Chemical R&D, Pfizer Global R&D, MS-8118D/4002, Eastern Point Road, Groton, Connecticut 06340, U.S.A.
Abstract:
CJ-17,493 (4) is a chiral NK-1 receptor antagonist. It was first
prepared through a diastereoselective crystallization, then through
chiral chromatography of a key intermediate, and ultimately via
asymmetric synthesis. Multiple routes for the preparation of a key
isochroman were demonstrated, and conditions for improved
regioselectivity of a Friedel–Crafts acylation were identified.
Cesium fluoride was found to be an acceptable initiator for the
generation of a nucleophilic trifluoromethyl anion from CF3TMS.
A cinchonine-derived catalyst was identified for the enantioselective
addition of the trifluoromethyl group to the ketone, and it was
found that the product of the addition would be converted directly
Figure 1
to the isochroman by treatment with t-BuOK. A Duff reaction
was used for the formylation, and the resulting aldehyde was
coupled to amine 5 to afford CJ-17,493 (4).
Introduction
Substance-P (Neurokinin 1 or NK-1) (1; Figure 1) is a
naturally occurring peptide composed of 11 amino acids in the
Figure 2
thetic standpoint, it is clear that the compound can be assembled
through coupling of the piperidine fragment (5) with a precursor
(6), either through a reductive amination (X ) O) or an
alkylation (X ) H, halogen).
tachykinin family of peptides. It has been shown to be
upregulated in several conditions, such as arthritis, migraine,
inflammatory bowel disease, asthma, pain, and depression. As
such, NK-1 receptor antagonists are recognized as a new class
of potential therapeutic agents for the treatment of these
diseases.1 Recently, aprepitant (2)2 was approved as the first
NK-1 receptor antagonist in a new class of antiemetics for
control of chemotherapy-induced nausea and vomiting, while
maropitant (3)3 was approved as a veterinary medication to
prevent and treat acute vomiting in dogs.
CJ-17,493 (4; Figure 2) is another NK-1 receptor antagonist
which is structurally distinct from aprepitant and maropitant.
The drug candidate contains two major fragments: a chiral
2-phenyl-3-amino-substituted piperidine and an isochroman.
Structural complexity is added due to the presence of a
trifluromethyl-bearing quaternary stereocenter. From a retrosyn-
The medicinal chemistry synthesis is depicted in Scheme
1.4 Anisole derivative 7 was converted to bromo compound 8
in two straightforward steps. Halogen–metal exchange of the
bromide was accomplished with n-BuLi, followed by a capri-
cious addition to trifluoroacetone. This reaction required
cryogenic conditions and was plagued by competitive proton
transfer. The alkoxide generated displaced the primary chloride
and provided isochroman 9. Cleavage of the methyl ether was
followed by acetylation, to yield acetate 10, which underwent
a surprisingly efficient lipase hydrolysis, providing the desired
enantiomer 11 in high yield (45%, 90% of theory) and high ee
(94%). The phenol of the undesired enantiomer was removed
by extractive workup under basic conditions, while the desired
acetate was retained in the organic layer. Hydrolysis of the
acetate, followed by alkylation of the phenoxide, provided the
methyl ether 12. A reductive amination approach was pursued
for coupling of the piperidine. Thus, formylation of the arene
allowed access to aldehyde 13, which reacted with aminopi-
peridine 5 to generate the desired target. Overall, this synthetic
strategy was reasonable and allowed for the preparation of
* To whom correspondence should be addressed. E-mail: stephane.caron@
pfizer.com. Tel: (860) 441-3103. Fax: (860) 715 8584.
† Current address: Merck & Co. Inc., RY-800-C367, P.O. Box 2000, Rahway,
NJ 07065-0900.
(1) Chahl, L. A. Curr. Drug Targets 2006, 7, 993–1003.
(2) Hale, J. J.; Mills, S. G.; MacCoss, M.; Finke, P. E.; Cascieri, M. A.;
Sadowski, S.; Ber, E.; Chicchi, G. G.; Kurtz, M.; Metzger, J.;
Eiermann, G.; Tsou, N. N.; Tattersall, F. D.; Rupniak, N. M. J.;
Williams, A. R.; Rycroft, W.; Hargreaves, R.; MacIntyre, D. E. J. Med.
Chem. 1998, 41, 4607–4614.
(3) Ito, F.; Kondo, H.; Shimada, K.; Nakane, M.; Lowe, J. A., III; Rosen,
T. J.; Yang, B. V. Preparation of 2-diphenylmethyl-3-benzylamino-
quinuclidines as substance P antagonists. 92-US33179221677, 19920428,
1999.
(4) Satake, K. Preparation of N-(isobenzofuranylmethyl)piperidinamines
and analogs as substance P receptor antagonists. 98-IB17049925714,
19981026, 1999.
10.1021/op7001886 CCC: $37.00
Published on Web 10/30/2007
2007 American Chemical Society
Vol. 11, No. 6, 2007 / Organic Process Research & Development
•
1015

Scheme 1
Scheme 2
Table 1
several derivatives and access to both enantiomers of the
isochroman. However, from a process standpoint, several steps
lacked robustness. Condensation of the aryllithium species with
trifluoroacetone would be very difficult to scale because of the
narrow reaction temperature range. It is also unfortunate that
the methyl ether of 9 had to be cleaved and subsequently
reinstalled using conditions which must be avoided on a large
scale because of safety concerns (NaH in the presence of DMF).
Finally, an alternative reagent was desirable for the formylation
because of the carcinogenic nature of the dichloromethyl ether
selected.
ratio
Lewis acid
entry (amt, equiv) solvent
T, °C
AlCl3 (2.03) CH2Cl2 -40
AlCl3 (3.38) CH2Cl2 room temp
AlCl3 (3.39) CH3NO2 room temp 1.1
AlBr3 (3.74) CH2Cl2 room temp 11
TiCl4 (2.17) CH2Cl2 room temp
17
18
19
1
2
3
4
5
1
4
1
minor
minor
1
1
0
1
0
minor
1.3
1
In light of these concerns, an alternative route was sought.
Because of the short timelines desired for the preparation of a
few kilograms of the product and the availability of a large
quantity of enantiomerically pure piperidine 5, the strategy of
coupling 5 with a racemic isochroman and generation of two
potentially separable diastereoisomers was adopted. It was
known that the enantiomeric purity of product 4 could be
enhanced from 94% to >99% ee by preparation of the HCl
salt. As such, the racemate of aldehyde 13 became the
immediate synthetic target. A potential retrosynthesis is depicted
in Scheme 2. The isochroman should be generated through an
intramolecular alkylation of tertiary alcohol 14. This approach
had a precedent under mild acidic conditions for simple benzylic
alcohols5 and had been demonstrated in Discovery through the
alkoxide intermediate. The desired substrate (14) would arise
either from the addition of MeLi or MeMgX to a trifluoromethyl
ketone (R ) CF3) or addition of a trifluoromethyl anion to an
acetophenone derivative (R ) CH3). Either intermediate could
theoretically be accessed by Friedel–Crafts acylation of anis-
ole 7.
Results and Discussion
It was rapidly found that trifluoroacetylation of anisole
derivative 15 (3) under Lewis acid activation could not be
achieved using reagents such as TFAA,6 trifluoroacetyl chlo-
ride,7 2-trifluoroacetoxypyridine,8 or 4-(dimethylamino)-1-(tri-
fluoroacetyl)pyridinium trifluoroacetate.9 Thus, a Friedel–Crafts
aromatic electrophilic substitution of 15 with an acetylating
agent was pursued.
The first experiments led to surprising results (Table 1). Upon
treatment of 1510,11 with a slight excess of AcCl in the presence
of 2 equiv of AlCl3 at -40 °C, three products (desired 17,
regioisomer 18, and demethylated regioisomer 19) were ob-
(6) Stewart, R.; Teo, K. C. Can. J. Chem. 1980, 58, 2491–6.
(7) Hull, W. E.; Seeholzer, K.; Baumeister, M.; Ugi, I. Tetrahedron 1986,
42, 547–552.
(8) Keumi, T.; Shimada, M.; Takahashi, M.; Kitajima, H. Chem. Lett.
1990, 783–786.
(9) Simchen, G.; Schmidt, A. Synthesis 1996, 1093–1094.
(10) Bergmann, E.; Weizmann, A. J. Org. Chem. 1939, 4, 266–269.
(11) Natelson, S.; Gottfried, S. P. J. Am. Chem. Soc. 1939, 61, 1001–1002.
(12) Prakash, G. K. S.; Krishnamurti, R.; Olah, G. A. J. Am. Chem. Soc.
1989, 111, 393–395.
(5) Almena, J.; Foubelo, F.; Yus, M. Tetrahedron 1995, 51, 3351–3364.
1016
•
Vol. 11, No. 6, 2007 / Organic Process Research & Development

Scheme 3
Figure 3
regioselectivity in CS2, a highly undesirable solvent.13 Using
AlBr3 in conjunction with AcBr, a 10:1 ratio of regioisomers
favoring 21 was obtained in high yield (Scheme 5). The
synthetic sequence described in Scheme 4 worked equally well
on 21, and the acetate was cleaved using 1 N NaOH. All these
operations were carried out in a single reaction and afforded
the crystalline diol 22 in 71% overall yield. To complete the
synthesis of 9, several leaving groups were evaluated, and
mesylation of the primary alcohol proceeded with concomitant
cyclization in near -quantitative yield. Overall, the preparation
of racemic 9 was high-yielding and provided a crystalline solid
for purification. Because of these attributes, it was implemented
on a multikilogram scale.
A second approach was also evaluated (Scheme 6). The route
begins from carboxylic acid 23, which is the precursor to alcohol
7 in the previous route. The Friedel–Crafts acylation proceeded
in acceptable yield as a single regioisomer (24), potentially from
generation of the anhydride followed by intramolecular acyla-
tion. Introduction of the trifluoromethyl group to acid 24 could
not be accomplished. Esterification afforded the keto ester 25
as a crystalline solid in 91% yield. This substrate proved to be
very prone to enolization under basic conditions. In the presence
of CsF in DMF, a red color was observed, presumably from
the generation of the enolate. Upon addition of CF3TMS,
nucleophilic addition proceeded, although the starting material
was never fully consumed. TBAF was introduced to induce
cleavage of the silyl ether and lactonization to provide 26 in
54% overall yield.
It had been hoped that conversion of 26 to the hydroxy acid
27 would provide a handle for a classical resolution. While 26
could be hydrolyzed with NaOH, it relactonized upon treatment
with a number of chiral amines. Treatment of the sodium
carboxylate salt with a chiral amine followed by slow addition
of HCl also proved to be unsuccessful. Alternatively, 26 could
be reduced to lactol 28 using in situ generated borane, and 9
was obtained by reduction with Et3SiH under acidic conditions.14
Since the classical resolution approach failed, this route offered
no advantages over that described previously.
A Duff reaction was adopted for the formylation of 9
(Scheme 7). Trifluoroacetic acid was premixed with hexa-
methylenetetramine (HMTA) for 90 min at 70 °C prior to
addition of 9. Upon completion of the reaction, the resulting
imine was hydrolyzed under aqueous acidic conditions, afford-
ing aldehyde 13 with 12:1 regioselectivity. In order to achieve
high-purity (+/-)-13 in the penultimate step, the tosyl hydra-
zone 29 was prepared and provided a highly crystalline
intermediate (52% yield from 9 after purification). The hydra-
zone could be hydrolyzed in 95% yield with CuCl2 in t-BuOH/
served. Using a larger excess of the Lewis acid at higher
temperature (entry 2) increased the ratio of 17, while the
undesired regioisomer was largely demethylated. TiCl4 mini-
mized demethylation but did not improve the selectivity (entry
5), while the use of a polar Lewis basic solvent, such as
nitromethane, worsened the ratio (entry 3). Fortunately, using
a larger Lewis acid such as AlBr3 maximized the formation of
17 (entry 4).
We speculated that in order to achieve high regioselectivity,
it was beneficial to have a bulky Lewis acid coordinate with
the aryl ether, which is probably not feasible in a solvent such
as nitromethane. This hypothesis would also explain why only
the demethylated product of the undesired regioisomer is
observed. It is also possible that elevated temperatures and
excess Lewis acid might help dissociation of the acylium ion
with the Lewis acid and disfavor intramolecular acylation, as
shown in Scheme 3.
Having these principles in mind, the optimal conditions were
identified (Scheme 4 ). Arene 15 was added to a 0 °C solution
of AlBr3 and AcCl, therefore always assuring a high excess of
the Lewis acid. This reaction was dose-controlled, which is also
advantageous from a process safety perspective, and provided
high regioselectivity (>20:1). Impurity 19 was easily purged
in a basic extractive workup.
For the addition of the trifluoromethyl anion to 17, TBAF-
promoted addition of (trifluoromethyl)trimethylsilane to alde-
hydes and ketones is well precedented.12 However, elimination
of the primary chloride under basic conditions was a concern,
and milder conditions to promote the reaction and avoid
generation of the styrene were sought. CsF (8 mol %) was
identified as a preferred catalyst for the conversion of 17 to 20.
Furthermore, TBAF could be added upon completion of the
trifluoromethylation and the resulting tertiary alkoxide under-
went nucleophilic displacement of the primary halide, providing
the desired isochroman 9 in high overall yield. Unfortunately,
while this synthetic sequence is attractive chemically, the
physical properties of the intermediates, which were all oils,
combined with the very poor stability of acetophenone 17
rendered this route unattractive for scale-up, and a substitute
for the primary halide had to be identified.
Performing the acylation directly on 7 was an attractive
strategy, since it eliminated the chlorination step and could
conceivably provide a functional group more stable toward
elimination. The reaction is known to proceed with 3:1
(13) Sternberg, E. D.; Vollhardt, K. P. C. J. Org. Chem. 1984, 49, 1574–
1583.
(14) Kursanov, D. N.; Parnes, Z. N.; Loim, N. M. Synthesis 1974, 633–
651.
Vol. 11, No. 6, 2007 / Organic Process Research & Development
•
1017

Scheme 4
Scheme 5
Scheme 6
Scheme 7
Scheme 8
H2O to provide 13 as a low-melting solid. This method
accelerated the rate of hydrolysis to approximately 3 h, as
opposed to several days when using CuSO4.15 On scale, 13
proved to be of sufficient chemical purity to crystallize directly
out of heptane after the formylation, which avoided the need
for the preparation of the tosyl hydrazone. Attempts to perform
a classical resolution through oxazoline 3016 met with no
success, which meant that a diastereoselective crystallization
of the final product would be necessary.
reductive amination with sodium triacetoxyborohydride.20
A
basic wash in the workup removed the mandelic and acetic
acids, providing a 1:1 mixture of diastereoisomers of 4. It was
known from the work in Discovery Chemistry that a 97:3
mixture of diastereoisomers could be purified through formation
of the HCl salt, but this proved to be unsuccessful on the 1:1
mixture. The phase diagram of the HCl salt was generated, and
it was determined that any mixture above a 3:1 ratio would be
above the eutectic point and should crystallize to high diaster-
eomeric purity.21 Over a dozen different salts were evaluated
As shown in Scheme 8, racemic 13 was coupled to the
resolved dimandelate salt of aminopiperidine (5)17–19 through a
(15) Enders, D.; Wortmann, L.; Peters, R. Acc. Chem. Res. 2000, 33, 157–
169.
(16) Urban, F. J.; Moore, B. S. Tetrahedron: Asymmetry 1992, 3, 731–
734.
(19) Snyder, W. M.; Watson, H. A., Jr.; Wilcox, G. E. Preparation of chiral
2-phenyl-3-benzylaminopiperidines as substance P antagonists. 94-
IB59427966, 19940406, 1994.
(17) Miller, J. A.; Farrell, R. P. Tetrahedron Lett. 1998, 39, 6441–6444.
(18) Caron, S.; Massett, S. S.; Bogle, D. E.; Castaldi, M. J.; Braish, T. F.
Org. Process Res. DeV. 2001, 5, 254–256.
(20) Abdel-Magid, A. F.; Mehrman, S. J. Org. Process Res. DeV. 2006,
10, 971–1031.
1018
•
Vol. 11, No. 6, 2007 / Organic Process Research & Development

Scheme 9
Scheme 10
Scheme 11
in a number of solvents to identify a diastereoselective crystal-
lization. It was discovered that (+)-mandelic acid (32) reliably
provided a 4:1 mixture, while its enantiomer provided little to
no enrichment. The crystallization was optimized with EtOH
(20 volumes) using 2.0 equiv of the acid to obtain a 4:1 ratio
with 45% recovery (representing 36% of the desired diastere-
omer out of a possible 50% over 2 steps). Another advantage
of the mandelate salt was that it provided an excellent purge
point of impurities carried through the synthesis, including
the products derived from the undesired regioisomer in the Duff
reaction.
The completion of the synthesis proved to be operationally
simple. The mandelate salt of 4 was filtered and free based in
diisopropyl ether with aqueous NaOH. The ether layer was
displaced by MeOH, and 2 equiv of aqueous HCl was added.
In a final solvent mixture of approximately 3:1 MeOH/H2O,
the dihydrochloride 4·2HCl) crystallized and was isolated with
a dr >98:2. An additional recrystallization provided the final
compound in 99% ee, and an excess of 2 kg of the candidate
was prepared using this protocol.
While the route described above was acceptable for the first
campaign, it clearly lacked efficiency due to intermediacy of
racemic 13 taken into the reductive amination. An easy way to
lessen the loss of material associated with this approach was to
separate the enantiomers of an earlier intermediate. While this
could not be achieved through a classical resolution, chiral
chromatography appeared to be an attractive option, since many
of the intermediates were oils with very high organic solubility
(Scheme 9).22 Diol 22 was rapidly identified as an ideal
candidate for simulated moving-bed chromatography,23 which
provided the desired enantiomer (22) in 96% ee. One drawback
was that, unlike racemic diol 22, the scalemic product was no
longer crystalline. The remainder of the synthesis proceeded
as previously described without the need for the mandelate salt
formation or recrystallization at the final step. One interesting
aspect of this sequence was that it had been demonstrated that
(R)-22 could be racemized when treated with a strong acid such
as TFA, presumably through the quinone–methide intermediate
33. Fortunately, this was not the case with the isochroman,
which conserved its enantiopurity during the Duff reaction. This
observation opened the door for a potential recycle of the
undesired enantiomer of the diol.
Ultimately, an enantioselective synthesis of isochroman 9
was wanted to avoid the loss of 50% of a costly intermediate.
An attractive approach would be the enantioselective addition
of a nucleophilic trifluoromethyl anion to an acetophenone
derivative. Much progress has been made in recent years in
this area either through the use of chiral auxiliaries24 or through
the use of organocatalysis on a cinchonine-derived framework.
25–27
Since the conditions described in the literature were not
appropriate for our original substrate, a thorough evaluation of
the factors influencing the reaction was undertaken (Scheme
10).28 It was quickly determined that CH2Cl2 was critical for
the reaction to proceed with a useful level of enantioselectivity.
The original optimization was conducted with a cinchonine-
derived catalyst (35), although quinine, quinidine, and cinchoni-
dine were also studied. It was also determined that the free
secondary alcohol was necessary and that the 1-naphthyl
derivative on the tertiary nitrogen was optimal for a high level
of conversion and enantioselectivity. Finally, the preferred
substrate proved to have the primary alcohol protected as the
3,4-dimethoxybenzoate. As previously reported,28 the absolute
configuration of the product was secured through elucidation
of the X-ray crystal structure of a bromobenzoate derivative.
The final synthesis proceeds as shown in Scheme 11.
Alcohol 7 was acylated using acid chloride 36 to provide ester
37, which underwent a smooth Friedel–Crafts acylation to afford
crystalline acetophenone 38 in 97% yield over two steps. It is
interesting to note that, by having an electron-rich ester as the
(24) Roussel, S.; Billard, T.; Langlois, B. R.; Saint-James, L. Chem. Eur.
J. 2005, 11, 939–944.
(25) Iseki, K.; Nagai, T.; Kobayashi, Y. Tetrahedron Lett. 1994, 35, 3137–
3138.
(26) Nagao, H.; Yamane, Y.; Mukaiyama, T. Chem. Lett. 2007, 36, 666–
667.
(21) Collet, A. Angew. Chem., Int. Ed. 1998, 37, 3239–3241.
(22) Welch, C. J. Prep. Enantiosel. Chromatogr. 2005, 1–18.
(23) Strube, J.; Haumreisser, S.; Schmidt-Traub, H.; Schulte, M.; Ditz, R.
Org. Process Res. DeV. 1998, 2, 305–319.
(27) Mizuta, S.; Shibata, N.; Akiti, S.; Fujimoto, H.; Nakamura, S.; Toru,
T. Org. Lett. 2007, 9, 3707–3710.
(28) Caron, S.; Do, N. M.; Arpin, P.; Larivee, A. Synthesis 2003, 1693–
1698.
Vol. 11, No. 6, 2007 / Organic Process Research & Development
•
1019

Scheme 12
Scheme 13
Figure 4
preparation of the candidate. A chiral chromatography permitted
the evaluation of the properties of scalemic intermediates
and minimized the utilization of the chiral aminopiperidine.
Finally, an enantioselective trifluoromethylation was achieved,
followed by a direct base-mediated isochroman formation. This
sequence allowed for a seven-step synthesis of CJ-17,493 (4)
from the commercially available alcohol 7 in 49% overall yield.
Experimental Section
1-(2- (2-Chloroethyl)-4-methoxyphenyl)ethanone (17). To
a solution of AlBr3 (19.8 g, 74.2 mmol) in CH2Cl2 (40 mL) at
0 °C was slowly added AcCl (3.27 mL, 46 mmol). The reaction
mixture was warmed to 15 °C, and 1-(2-chloroethyl)-3-
methoxybenzene (15; 5.61 g, 32.9 mmol) in CH2Cl2 (75 mL)
was added over 20 min. The reaction mixture was stirred for
1 h and then poured over ice (100 mL). To the mixture was
added 1 N aqueous HCl (100 mL). The aqueous layer was
removed, and the organic layer was washed with 1 N aqueous
NaOH (50 mL), dried over MgSO4, filtered, and concentrated
to afford 1-(2-(2-chloroethyl)-4-methoxyphenyl)ethanone (17;
6.18 g, 88%). 1H NMR (300 MHz, CDCl3): δ 2.60 (s, 3), 3.38
(t, 2, J ) 6.9), 3.82 (t, 2, J ) 6.9), 3.90 (s, 3), 6.85 (m, 1),
6.85–7.29 (m, 1), 7.85 (d, 1, J ) 8.1). 13C NMR (75 MHz,
CDCl3): δ 30.15, 39.60, 46.38, 56.69, 112.95, 119.68, 130.60,
134.47, 143.01, 163.30, 200.54.
protecting group, regioselectivity increased to the level of
20:1. The trifluoromethylation reaction was conducted at -50
°C in the presence of 2 equiv of CF3TMS and 4 mol % of the
catalyst, affording the crude 39 in 92% ee and 97% yield.
Cleavage of the trimethylsilyl ether of 39 proceeded smoothly,
and the enantiomeric excess could be increased by precipitation
and filtration of the racemic alcohol, which had very poor
solubility in CH2Cl2. For example, when starting from 39 of
only 76% ee, the alcohol (R)-40 could be isolated as an oil in
94% ee and 68% yield (Scheme 12). Furthermore, the 3,4-
dimethoxybenzoate was identified as an acceptable leaving
group for the direct conversion of ester (R)-40 to isochroman
(R)-9 using either KHMDS or t-BuOK.
Surprisingly, it was found that the tertiary silyl ether of 39
could be cleaved when treated with t-BuOK, which resulted in
direct conversion to (R)-9 (Scheme 13). The byproduct, tert-
butyl ester 41, can be explained if the reaction proceeds through
intramolecular silyl transfer, as depicted in Figure 4. Although
migration of a tertiary TMS to a secondary alcohol is prece-
dented under basic conditions,29 we believe that this is the first
example of the deprotection of a tertiary silyl ether using
t-BuOK in THF. The crude mixture could be taken directly in
the Duff reaction, where the acidic conditions cleaved the tert-
butyl ester and simplified removal of the carboxylic acid as
part of the extractive workup. The chiral aldehyde (R)-13 was
obtained in 60% yield over the two steps and was progressed
through the reductive amination and salt formation sequence,
which proceeded in 87% yield.
Acetic Acid 2-(2-Acetyl-5-methoxyphenyl)ethyl Ester
(21). This compound was prepared by modification of a known
procedure.13 To a solution of AlBr3 (43.8 g, 164 mmol) in
CH2Cl2 (70 mL) at 0 °C was slowly added AcBr (14.6 mL,
197 mmol). The reaction mixture was warmed to 15 °C, and
2-(3-methoxyphenyl)ethanol (7; 10.0 g, 65.7 mmol) in CH2Cl2
(20.0 mL) was added over 45 min. The reaction mixture was
stirred for 1 h and then poured over ice (100 mL). To the
mixture was added 1 N aqueous HCl (100 mL). The organic
layer was separated and the aqueous layer was extracted with
CH2Cl2 (100 mL). The combined organic extracts were washed
with 1 N aqueous NaOH (100 mL), dried over MgSO4, filtered
through Celite and concentrated to afford acetic acid 2-
(2-acetyl-5-methoxyphenyl)ethyl ester (21) as an oil (14.8 g,
1
95%). H NMR (300 MHz, CDCl3): δ 2.05(s, 3), 2.59 (s, 3),
3.29 (t, 2, J ) 6.9), 3.89 (s, 3), 4.33 (t, 2, J ) 6.9), 6.81 (d, 1,
J ) 2.5), 8.85 (dd, 1, J ) 8.6, 2.6), 7.82 (d, 1, J ) 8.6). 13C
NMR (75 MHz, CDCl3): δ 22.28, 30.37, 35.36, 56.63, 66.11,
101.21, 112.63, 119.17, 131.00, 134.23, 142.90, 163.24, 172.32.
IR: 1737, 1674, 1604, 1567, 1358, 1239, 1037 cm-1. Anal.
Calcd for C13H16O4: C, 66.09; H, 6.83. Found: C, 65.71; H,
7.21.
1,1,1-Trifluoro-2-[2-(2-hydroxyethyl)-4-methoxyphenyl]
propan-2-ol ((+/-)-22). To a solution of acetic acid 2-(2-
acetyl-5-methoxyphenyl)ethyl ester (21; 7.21 g, 30.5 mmol) and
cesium fluoride (0.550 g, 3.62 mmol) in DMF (40 mL) at 0 °C
was slowly added (trifluoromethyl)trimethylsilane (5.90 mL,
In summary, several syntheses of CJ-17,493 were investi-
gated in the course of the development of this new NK1
candidate. A diastereoselective crystallization approach allowed
for the identification of a regioselective acylation and rapid
(29) Kuramochi, K.; Saito, F.; Takeuchi, R.; Era, T.; Takemura, M.;
Kobayashi, J. i.; Sakaguchi, K.; Kobayashi, S.; Sugawara, F. Tetra-
hedron 2006, 62, 8006–8015.
1020
•
Vol. 11, No. 6, 2007 / Organic Process Research & Development

39.9 mmol). The reaction mixture was stirred for 40 min, after
which GC/MS and HPLC analysis showed no starting material.
For characterization purposes, the reaction mixture was poured
into water and extracted with MTBE (100 mL). The organic
layer was washed with H2O (2 × 75 mL) and brine (50 mL),
dried over MgSO4, filtered, and concentrated to provide acetic
acid 2-[5-methoxy-2-(2,2,2-trifluoro-1-methyl-1-((trimethylsi-
lanyl)oxy)ethyl)phenyl]ethyl ester as a crude oil. 1H NMR (300
MHz, CDCl3): δ 0.19 (s, 9), 1.93 (s, 3), 2.10 (s, 3), 3.23–3.33
(m, 1), 3.42–3.52 (m, 1), 3.83 (s, 3), 4.26–4.32 (m, 2), 6.77
(dd, 1, J ) 8.9, 2.8), 6.86 (d, 1, J ) 2.9), 7.32 (d, 1, J ) 8.9).
13C NMR (100 MHz, CDCl3): δ 2.03, 21.03, 24.64, 32.86,
55.11, 65.54, 78.90 (q, J ) 30.3), 111.26, 117.44, 125.70 (q, J
) 287), 129.56, 129.79, 139.77, 159.17, 171.09. IR: 2961, 1741,
1610, 1383, 1286, 1255, 1165, 1140, 1039, 864, 846 cm-1.
Anal. Calcd for C17H25F3O4Si: C, 53.95; H, 6.66. Found: C,
53.72; H, 6.53.
To the reaction mixture containing a solution of acetic acid
2-[5-methoxy-2-(2,2,2-trifluoro-1-methyl-1-((trimethylsilanyl)
oxy)ethyl)phenyl]ethyl ester was added TBAF (31.0 mL of a
1.0 M solution in tetrahydrofuran, 31.0 mmol). The reaction
mixture was stirred for 1 h, after which GC/MS and HPLC
analysis showed no starting material. For characterization
purposes, the reaction mixture was poured into water and
extracted with MTBE (75 mL). The organic layer was washed
with H2O (75 mL) and brine (50 mL), dried over MgSO4,
filtered, and concentrated to provide a crude oil. 1H NMR (400
MHz, CDCl3): δ 1.82 (s, 3), 2.01 (s, 3), 2.98–3.06 (m, 2), 3.55
(dt, 1, J ) 13.7, 6.8), 3.79 (s, 3), 4.27–4.34 (m, 2), 6.73–6.77
(m, 2), 7.28 (d, 1, J ) 8.5). 13C NMR (100 MHz, CDCl3): δ
20.92, 25.50, 34.16, 55.10, 66.49, 76.67 (q, J ) 30.3), 111.55,
118.25, 126.02 (q, J ) 286), 128.67, 129.56, 139.70, 159.20,
171.32. IR: 3453, 1720, 1610, 1249, 1161, 1134, 1038 cm-1.
Anal. Calcd for C14H17F3O4: C, 54.90; H, 5.59. Found: C, 55.03;
H, 5.85.
6-Methoxy-1-methyl-1-(trifluoromethyl)isochroman ((+/
-)-9). To a solution of 1,1,1-trifluoro-2-[2-(2-hydroxyethyl)-
4-methoxyphenyl]-propan-2-ol (22; 3.12 g, 11.8 mmol) in
CH2Cl2 (18 mL) was added Et3N (5.70 mL, 41.1 mmol). The
solution was cooled to 0 °C, and MsCl (1.00 mL, 12.9 mmol)
was added dropwise. The reaction mixture was warmed to room
temperature and was stirred for 20 h. The formation of
methanesulfonic acid 2-[5-methoxy-2-(2,2,2-trifluoro-1-hydroxy-
1-methylethyl)phenyl]ethyl ester was rapid and its disappearance
monitored by HPLC (retention time 4.5 min, Zorbax Rx-C6
column 4.6 × 150 mm, 40 °C, 50% CH3CN/50% (0.2% Et3N,
0.1% H3PO4 aqueous pH 3.2 buffer, 1 mL/min). At the end of
the reaction, the mixture was poured into 1 N aqueous HCl
(20 mL) and was extracted with MTBE (30 mL). The organic
extracts were washed with H2O (20 mL), dried over MgSO4,
filtered, and concentrated to afford 6-methoxy-1-methyl-1-
(trifluoromethyl)isochroman ((+/-)-9) as an oil (2.89 g, 99%).
1H NMR (300 MHz, CDCl3): δ 1.69 (s, 3), 2.85–2.90 (m, 2),
3.85 (s, 3), 3.90–3.98 (m, 1), 4.14–4.21 (m, 1), 6.72 (d, 1, J )
2.6), 6.85 (dd, 1, J ) 8.7, 2.6), 7.31 (d, 1, J ) 8.7). 13C NMR
(100 MHz, CDCl3): δ 23.25, 29.42, 55.19, 61.37, 76.10 (q, J
) 27.4), 112.84, 113.43, 124.85, 125.96 (q, J ) 289), 127.86,
136.49, 158.98. IR: 2946, 2839, 1738, 1611, 1505, 1162, 1137,
1101 cm-1. Anal. Calcd for C12H13F3O2: C, 58.54; H, 5.32.
Found: C, 58.27; H, 5.35.
(2-Acetyl-5-methoxyphenyl)acetic Acid (24). To a solution
of AlBr3 (57.6 g, 216 mmol) in CH2Cl2 (90 mL) at 0 °C was
slowly added AcCl (11.5 mL, 162 mmol). To the reaction
mixture was added (3-methoxyphenyl)acetic acid (23; 17.9 g,
108 mmol) in CH2Cl2 (20.0 mL). The reaction mixture was
stirred for 1 h and then poured over ice (100 mL). The organic
layer was separated, and 1 N aqueous NaOH (100 mL) was
added. The biphasic mixture was stirred vigorously for 90 min,
and the layers were separated. The organic layer was discarded,
and concentrated HCl was added to the aqueous layer until the
pH reached 1. A solid precipitated and was filtered and air-
dried to afford (2-acetyl-5-methoxyphenyl)acetic acid (24;
16.8 g, 75%): mp 153–155 °C. 1H NMR (300 MHz, CDCl3):
δ 2.68 (s, 3), 3.91 (s, 3), 3.92 (s, 2), 6.92–6.95 (m, 2), 7.88 (d,
1, J ) 9.5). 13C NMR (100 MHz, CDCl3): δ 28.33, 41.43, 55.46,
112.54, 118.26, 129.17, 133.08, 136.94, 162.65, 174.80, 200.96.
IR: 3435, 1704, 1663, 1609, 1568, 1258 cm-1. Anal. Calcd for
C11H12O4: C, 63.45; H, 5.81. Found: C, 63.35; H, 5.46.
(2-Acetyl-5-methoxyphenyl)acetic Acid Methyl Ester (25).
To a solution of (2-acetyl-5-methoxyphenyl)acetic acid (24;
11.0 g, 52.8 mmol) in MeOH (55 mL) was added concentrated
H2SO4 (2.2 mL). The reaction mixture was stirred at room
temperature for 3 h, after which it was concentrated to a low
volume, EtOAc (50 mL) was added and the mixture concen-
trated to a low volume. EtOAc (50 mL) was added, and the
solution was washed with 1 N aqueous NaOH (50 mL). The
layers were separated, and the organic layer was dried over
MgSO4, filtered, and concentrated to an oil. Hexanes (70 mL)
was added, a precipitate formed, and the mixture was stirred
for 2 h. The product was filtered to afford (2-acetyl-5-
methoxyphenyl)acetic acid methyl ester (25; 10.6 g, 91%): mp
74–76 °C. 1H NMR (300 MHz, CDCl3): δ 2.58 (s, 3), 3.74 (s,
3), 3.89 (s, 3), 3.95 (s, 2), 6.78 (d, 1, J ) 2.6), 6.89 (dd, 1, J )
To the reaction mixture containing acetic acid 2-[5-methoxy-
2-(2,2,2-trifluoro-1-hydroxy-1-methylethyl)phenyl]ethyl ester
was added 1 N aqueous NaOH (35 mL, 35 mmol). The reaction
mixture was warmed to room temperature and was stirred until
reaction completion. The mixture was poured into H2O (100
mL) and extracted with MTBE (150 mL). The organic layer
was washed with H2O (50 mL) and brine (50 mL), dried over
MgSO4, and concentrated. To the crude solid was added
hexanes (25 mL) and MTBE (5 mL), and the solid was
triturated, filtered, and dried to provide 1,1,1-trifluoro-2-[2-(2-
hydroxyethyl)-4-methoxyphenyl]propan-2-ol (5.70 g, 71% over-
all yield from acetic acid 2-(2-acetyl-5-methoxyphenyl)ethyl
1
ester (21)): mp 110–111 °C. H NMR (300 MHz, CDCl3): δ
1.83 (s, 3), 2.91 (dt, 1, J ) 13.7, 3.9), 3.76 (ddd, 1, J ) 13.7,
9.3, 4.4), 3.85 (s, 3), 3.85–3.93 (m, 1), 4.08 (dt, 1, J ) 9.3,
3.7), 6.80–6.83 (m, 2), 7.38 (d, 1, J ) 8.4). 13C NMR (100
MHz, CDCl3): δ 26.01, 36.12, 55.19, 64.13, 76.52 (q, J ) 28.9),
111.47, 117.43, 125.99 (q, J ) 287), 129.69, 129.94, 140.86,
159.55. IR: 3395, 3162, 1610, 1513, 1467, 1248, 1157, 1087,
1046 cm-1. Anal. Calcd for C12H15F3O3: C, 54.54; H, 5.72.
Found: C, 54.65; H, 5.70.
Vol. 11, No. 6, 2007 / Organic Process Research & Development
•
1021

8.7, 2.6), 7.89 (d, 1, J ) 8.6). 13C NMR (75 MHz, CDCl3): δ
29.65, 42.35, 53.11, 56.69, 113.17, 120.00, 130.52, 134.39,
138.90, 163.54, 173.23, 200.35. IR: 1739, 1665, 1605, 1568,
1321, 1247, 1165 cm-1. Anal. Calcd for C12H14O4: C, 65.85;
H, 6.35. Found: C, 64.87; H, 6.44.
3.7), 6.69 (d, 1, J ) 2.7), 6.82 (dd, 1, J ) 8.7, 2.7), 7.22–7.27
(m, 1). 13C NMR (100 MHz, CDCl3,data reported for identifi-
able signals of the major diastereoisomer): δ 24.52, 35.46, 55.16,
90.71, 113.11, 113.98, 125.22, 127.57, 132.98, 159.59. IR: 3439,
2949, 1735, 1613, 1506, 1166, 1141, 1070 cm-1
6-Methoxy-1-methyl-1-(trifluoromethyl)isochroman-3-
one (26). To a solution of (2-acetyl-5-methoxyphenyl)acetic acid
methyl ester (25; 2.00 g, 9.00 mmol) and CsF (96.0 mg, 0.632
mmol) in DMF (12 mL) at 0 °C was slowly added (trifluo-
romethyl)trimethylsilane (1.73 mL, 11.7 mmol). The reaction
mixture was stirred at 0 °C for 7 h. For characterization
purposes, the reaction mixture was poured into water and
extracted with MTBE (50 mL). The organic layer was washed
with H2O (2 × 75 mL) and brine (50 mL), dried over MgSO4,
filtered, and concentrated to provide [5-methoxy-2-(2,2,2-
trifluoro-1-methyl-1-((trimethylsilanyl)oxy)ethyl)pheny-
l]acetic acid methyl ester as an oil. 1H NMR (400 MHz, CDCl3):
δ 0.11 (s, 9), 1.89 (s, 2), 3.68 (s, 3), 3.77 (s, 3), 3.98 (d, 1, J )
17.2), 4.28 (d, 1, J ) 17.0), 6.74–6.77 (m, 2), 7.29 (d, 1, J )
9.1). 13C NMR (100 MHz, CDCl3): δ 1.87, 24.25, 39.32, 51.75,
55.12, 78.67 (q, J ) 30.3), 111.97, 118.30, 125.70 (q, J ) 286),
129.50, 129.57, 136.10, 159.17, 172.81. IR: 2956, 1745, 1611,
1577, 1467, 1436, 1290, 1256, 1166, 1092, 989, 863, 847 cm-1.
Anal. Calcd for C16H23F3O4Si: C, 52.73; H, 6.36. Found: C,
52.84; H, 6.36.
To the reaction mixture containing a solution of [5-methoxy-
2-(2,2,2-trifluoro-1-methyl-1-((trimethylsilanyl)oxy)ethyl)phe-
nyl]acetic acid methyl ester was added TBAF (9.00 mL of a
1.0 M solution in THF, 9.00 mmol). The reaction mixture was
stirred for 1 h, after which it was poured into H2O (50 mL)
and extracted with MTBE (50 mL). The organic layer was
washed with H2O (50 mL) and brine (30 mL), dried over
MgSO4, filtered, and concentrated to afford 6-methoxy-1-
methyl-1-(trifluoromethyl)isochroman-3-one (26) as an oil (1.26
g, 54%). 1H NMR (400 MHz, CDCl3): δ 1.89 (s, 3), 3.71 (d,
1, J ) 20.6), 3.79 (s, 3), 3.89 (d, 20.8), 6.65 (d, 1, J ) 1.5),
6.85–6.89 (m, 1), 7.29 (d, 1, J ) 8.7). 13C NMR (100 MHz,
CDCl3): δ 21.45, 34.32, 55.33, 83.01 (q, J ) 30.3), 112.21,
113.88, 120.57, 124.68 (q, J ) 285.7), 127.73, 132.18, 160.75,
167.45. IR: 1765, 1614, 1509, 1322, 1301, 1274, 1259, 1183,
1101, 997, 813 cm-1. Anal. Calcd for C12H11F3O3: C, 55.39;
H, 4.26. Found: C, 55.03; H, 4.54.
6-Methoxy-1-methyl-1-(trifluoromethyl)isochroman ((+/
-)-9). To a solution of 6-methoxy-1-methyl-1-(trifluoro-
methyl)isochroman-3-ol (28; 8.36 g, 31.9 mmol) in CH2Cl2 (84
mL) was added triethylsilane (15.3 mL, 95.8 mmol) followed
by TFA (14.7 mL, 191 mmol). The reaction was stirred at room
temperature for 2 h and was poured into 1 N aqueous NaOH
(250 mL). The organic layer was separated and washed with 1
N aqueous NaOH (100 mL). The organic layer was dried over
MgSO4, filtered, and concentrated to afford 6-methoxy-1-
methyl-1-(trifluoromethyl)isochroman (9) as an oil (6.88 g,
88%). Spectroscopic data were identical with those of the
material prepared by the previous method.
6-Methoxy-1-methyl-1-(trifluoromethyl)isochroman-7-
carbaldehyde ((+/-)-13). Method A. To HMTA (31.3 g, 223
mmol) was added TFA (400 mL), and the mixture was heated
to 70 °C for 90 min. A solution of 6-methoxy-1-methyl-1-
(trifluoromethyl)isochroman (9; 50.0 g, 203 mmol) in TFA (100
mL) was then added to the reaction mixture over 40 min. The
solution was stirred for 3 h, and H2O was added (450 mL).
The reaction mixture was stirred for 16 h, cooled to room
temperature, and poured into MTBE (500 mL). The organic
layer was separated and washed with H2O (3 × 300 mL). The
organic layer was poured into a round-bottom flask and cooled
to 0 °C. NaOH (6 N) was added in portions until the pH rose
to 10 (∼500 mL). The organic layer was separated, washed
with H2O (200 mL), dried over MgSO4, filtered, and concen-
trated to afford 6-methoxy-1-methyl-1-(trifluoromethyl)isoch-
roman-7-carbaldehyde (13) as an oil (54.2 g of a 12:1 mixture
1
of regioisomers, 97%): mp 82–93 °C. H NMR (400 MHz,
CDCl3): δ 1.71 (s, 3), 2.95 (dt, 2, J ) 2.6, 5.3), 3.90–3.97 (m,
1), 3.97 (s, 3), 4.19 (dt, 1, J ) 11.2, 5.6), 6.81 (d, 1, J ) 1.2),
10.4 (s, 1). 13C NMR (75 MHz, CDCl3): δ 23.07, 29.98, 55.73,
60.83, 76.03 (q, J ) 27.4), 111.81, 112.50, 123.65, 125.32,
125.64 (q, J ) 287), 127.06, 160.89, 188.92. IR: 1683, 1616,
1498, 1296, 1271, 1163, 1149, 1120, 1096, 874 cm-1. Anal.
Calcd for C13H13F3O3: C, 57.13; H, 5.05. Found: C, 56.94; H,
4.78.
6-Methoxy-1-methyl-1-(trifluoromethyl)isochroman-
3-ol (28). To a solution of 6-methoxy-1-methyl-1-(trifluorom-
ethyl)isochroman-3-one (26; 1.50 g, 5.76 mmol) in THF (30
mL) at 0 °C was added NaBH4 (0.240 g, 6.34 mmol) followed
by BF3 ·Et2O complex (0.992 g, 8.07 mmol). The reaction
mixture was warmed to room temperature and was stirred
overnight. The reaction mixture was added to H2O (75 mL)
and extracted with MTBE (75 mL). The layers were separated,
and the organic layer was washed with 1 N aqueous HCl (50
mL), dried over MgSO4, filtered, and concentrated to afford
6-methoxy-1-methyl-1-(trifluoromethyl)isochroman-3-ol (28) as
an oil and a mixture of R and ꢀ anomers (1.19 g, 79%). Data
are reported for the major diastereoisomer. 1H NMR (400 MHz,
CDCl3): δ 1.74 (s, 3), 2.85 (dd, 1, J ) 15.7, 4.3), 2.88–2.99
(m, 1), 3.11 (dd, 1, J ) 15.7, 3.2), 3.80 (s, 3), 5.63 (t, 1, J )
Method B. A mixture of CuCl2 (52.7 g, 309 mmol) and N′-
1-[(E)-1-(6-methoxy-1,1-dimethyl-3,4-dihydro-1H-isochromen-
7-yl)methylidene]-4-methyl-1-benzenesulfonohydrazide (29;
45.5 g, 103 mmol) in t-BuOH (910 mL) and H2O (228 mL)
was heated to 70 °C for 2.5 h. The reaction mixture was cooled
to room temperature, concentrated to about 300 mL, and poured
into MTBE (500 mL) and H2O (500 mL). The mixture was
stirred for 15 min and filtered. The filtrate was poured into
MTBE (200 mL), and the layers were separated. The organic
layer was washed with water (4 × 250 mL), dried over MgSO4,
filtered, and concentrated to provide 6-methoxy-1-methyl-1-
(trifluoromethyl)isochroman-7-carbaldehyde (13) as an oil
which solidified on standing (26.8 g, 95%). Spectroscopic data
were identical with those of the material reported above.
1022
•
Vol. 11, No. 6, 2007 / Organic Process Research & Development

(E)-N′-((6-Methoxy-1-methyl-1-(trifluoromethyl)-3,4-di-
hydro-1H-isochromen-7-yl)methylene)-4-methylbenzene-
sulfonohydrazide (29). To a solution of the crude 6-methoxy-
1-methyl-1-(trifluoromethyl)isochroman-7-carbaldehyde (13)
(8.97 g, 32.7 mmol) in methanol (90 mL) was added p-
toluenesulfonhydrazide (6.09 g, 32.7 mmol) followed by 2%
aq. AcOH (14.0 mL). The reaction mixture was heated to reflux
for 1 h and cooled to room temperature. H2O (90 mL) was
added dropwise, and the mixture was stirred for 2 h. The solid
was isolated, stirred in MTBE (62 mL) at reflux for 2 h, stirred
overnight at room temperature, and filtered to provide (E)-N′-
((6-methoxy-1-methyl-1-(trifluoromethyl)-3,4-dihydro-1H-iso-
chromen-7-yl)methylene)-4-methylbenzenesulfonohydrazide
(2S,3S)[((1R)-6-Methoxy-1-methyl-1-(trifluoromethyl)iso-
chroman-7-yl)methyl](2-phenylpiperidin-3-yl)amine Dihy-
drochloride (4 ·2HCl). ((6-Methoxy-1-methyl-1-(trifluorome-
thyl)isochroman-7-yl)methyl)(2-phenylpiperidin-3-yl)amine (S)-
(+)-mandelate(4·32;6.25kgofa76:24mixtureofdiastereoisomers,
10.7 mol) was stirred overnight in MTBE (62.5 L) and 0.5 N
aqueous NaOH (62.5 L). The layers were separated, and the
organic layer was washed with 0.5 N aqueous NaOH (50 L)
and H2O (2 × 50 L). The organic layer was concentrated to a
low volume, and methanol (20 L) was added. The solution was
stirred at room temperature, and a solution of 1.89 L of
concentrated HCl in H2O (9.77 L) was slowly added. The
mixture was stirred for 6 h, and the solids were filtered. The
solids were stirred for 20 h in 3:1 MeOH/H2O (31.2 L), filtered,
and dried to afford ((6-methoxy-1-methyl-1-(trifluoromethyl)
isochroman-7-yl)methyl)(2-phenylpiperidin-3-yl)amine dihy-
drochloride (4·2HCl; 3.02 kg, 56%) as a 98:2 mixture of
diastereoisomers. The diastereomeric ratio could be further
1
(29) (7.87 g, 54%): mp 181–183 °C. H NMR (300 MHz,
CDCl3): δ 1.71 (d, 3, J ) 0.7), 2.44 (s, 3), 2.85–2.89 (m, 2),
3.84 (s, 3), 3.93 (dt, 1, J ) 11.2, 5.6), 4.16 (dt, 1, J ) 11.2,
5.6), 6.65 (s, 1), 7.33 (d, 2, J ) 8.1), 7.79 (d, 1, J ) 1.2), 7.89
(d, 2, J ) 8.4), 8.13 (s, 1). 13C NMR (75 MHz, CDCl3): δ
21.48, 23.07, 29.50, 55.47, 60.99, 76.02 (q, J ) 27.4), 110.91,
120.45, 124.74, 125.04, 125.72 (q, J ) 287), 127.95, 129.45,
134.97, 138.97, 143.34, 144.22, 157.04. IR: 3223, 1623, 1505,
1417, 1325, 1289, 1275, 1172, 1157, 1123, 1098, 918, 658
cm-1. Anal. Calcd for C20H21F3N2O4S: C, 54.29; H, 4.78; N,
6.33. Found: C, 54.34; H, 4.73; N, 6.37.
1
increased by crystallization from methanol/water (75:25). H
NMR (400 MHz, D2O, data reported for major diastereoisomer):
δ 1.52 (s, 3), 1.80–1.92 (m, 2), 1.95–2.50 (m 1), 2.21–2.26 (m,
1), 2.63–2.71 (m, 2), 3.04–3.11 (m, 1), 3.36 (s, 3), 3.45–3.49
(m, 1), 3.65–3.81 (m, 3), 3.90–3.96 (m, 1), 4.09 (d, 1, J ) 13.5),
6.46 (s, 1), 6.98–7.07 (m, 3), 7.23–7.25 (m, 2), 7.30 (t, 1, J )
7.5). IR: 2958, 1457, 1377, 1143, 749, 692 cm-1. Anal. Calcd
for C24H31Cl2F3N2O2: C, 56.81; H, 6.16; Cl, 13.97; N, 5.52.
Found: C, 56.69; H, 6.31; Cl, 14.13; N, 5.55.
(2S,3S)-[((1R)-6-Methoxy-1-methyl-1-(trifluoromethyl)is-
ochroman-7-yl)methyl](2-phenylpiperidin-3-yl)amine (S)-
(+)-mandelate (4 ·32). Sodium triacetoxyborohydride (10.28
kg, 55.0 mol) was added in one portion to a slurry of
6-methoxy-1-methyl-1-(trifluoromethyl)isochroman-7-carbalde-
hyde (13; 6.59 kg, 24.0 mol) and (2S-3S)-2-phenylpiperidin-
3-ylamine (5) dimandelate (32) (11.66 kg, 24.2 mol) in CH2Cl2
(230 L). Within 15 min most starting material was dissolved
and slow precipitation of product began shortly after. The
reaction mixture was stirred for 2.5 h at room temperature and
cooled to 0 °C, and 2 N NaOH (60 L) was added slowly. The
layers were separated. The organic extracts were stirred for 15
min with 2 N aqueous NaOH (120 L), the layers were separated,
and the organic layer was washed with H2O (2 × 30 L). EtOH
2B (187 L) was added, and the CH2Cl2 was distilled. The
mixture was cooled to 18–22 °C, and (S)-(+)-mandelic acid
(32) (7.35 kg, 48.3 mol) was added. The mixture was stirred,
and crystallization began to proceed. After it was stirred
overnight, the mixture was filtered to yield 6.38 kg (45%) of
((6-methoxy-1-methyl-1-(trifluoromethyl)isochroman-7-yl)m-
ethyl)(2-phenylpiperidin-3-yl)amine (S)-(+)-mandelate (4·32)
as a mixture of diastereomers (76:24 ratio by HPLC analysis).
1H NMR (400 MHz, CDCl3, data reported for major diastereo-
isomer): δ 1.42–1.64 (m, 2), 1.53 (s, 3), 1.72–1.79 (m, 1),
1.94–1.98 (m, 1), 2.46–2.89 (m, 4), 3.15–3.28 (m, 3), 3.45 (s,
3), 3.47–3.78 (m, 1), 3.92–3.97 (m, 2), 4.27 (bs, 1), 4.52 (s, 1),
6.66 (s, 1), 7.04–7.19 (m, 4), 7.27–7.36 (m, 7). 13C NMR (100
MHz, CDCl3): δ 16.99, 22.53, 25.96, 28.58, 45.05, 45.46, 53.52,
53.95, 55.08, 60.61, 61.86, 73.25, 75.54 (q, J ) 28.2), 110.36,
126.02, 126.27, 126.32, 126.42, 126.55, 127.01, 127.43, 127.57,
128.27, 135.04, 137.83, 143.16, 156.51, 174.59. IR: 3441, 1576,
1358, 1160, 1136, 1098, 1038, 775, 756, 698 cm-1. Anal. Calcd
for C32H37F3N2O5: C, 65.52; H, 6.36; N, 4.78. Found: C, 65.55;
H, 6.03; N, 4.84.
3,4-Dimethoxybenzoic Acid 2-(3-Methoxyphenyl)ethyl
Ester (37). To a solution of 3-methoxyphenethyl alcohol (7;
21.24 g, 139.6 mmol) in CH2Cl2 (200 mL) at 0 °C was added
Et3N (23.3 mL, 167 mmol), DMAP (3.22 g, 26.4 mmol), and
3,4-dimethoxybenzoyl chloride (33.57 g, 167.3 mmol). The
reaction mixture was warmed slowly to room temperature and
stirred for 16 h. To the reaction mixture was added H2O (200
mL). The aqueous layer was removed, and the organic layer
was washed with H2O (100 mL) and brine (100 mL), dried
over Na2SO4, filtered, and concentrated to afford 3,4-dimethoxy-
benzoic acid 2-(3-methoxyphenyl)ethyl ester (37) as a crude
brown oil (44.16 g, 100%). This material was used without
1
further purification. H NMR: δ 3.08 (t, 2, J ) 7.0), 3.82 (s,
3), 3.95 (s, 3), 3.97 (s, 3), 4.54 (t, 2, J ) 7.0), 6.80 (dd, 1, J )
2.6, 0.8), 6.83 (dd, 1, J ) 2.6, 0.8), 6.87 (t, 1, J ) 1.9), 6.90 (d,
1, J ) 8.6), 7.24 (d, 1, J ) 7.8), 7.54 (d, 1, J ) 1.9), 7.69 (dd,
1, J ) 8.4, 2.0).
3,4-Dimethoxybenzoic Acid 2-(2-Acetyl-5-methoxyphe-
nyl)ethyl Ester (38). To a solution of the crude 3,4-dimethoxy-
benzoic acid 2-(3-methoxyphenyl)ethyl ester (37; 44.16 g, 139.6
mmol) in CH2Cl2 (400 mL) at 0 °C was added TiCl4 (23.0 mL,
209 mmol) and AcCl (19.8 mL, 279 mmol). After 2 h, the
reaction mixture was poured over ice (100 mL). To the mixture
was added 1 N aqueous HCl (200 mL), and the resulting
mixture was stirred for 2 h. The organic layer was separated
and washed with H2O (100 mL) and brine (100 mL), dried over
Na2SO4, filtered, and concentrated to afford 3,4-dimethoxyben-
zoic acid 2-(2-acetyl-5-methoxyphenyl)ethyl ester (38) as a pale
peach solid (48.4 g, 97%): mp 110–111 °C (6:1 i-PrOH/
hexanes). 1H NMR: δ 2.57 (s, 3), 3.39 (t, 2, J ) 6.6), 3.79 (s,
3), 3.90 (s, 3), 3.92 (s, 3), 4.54 (t, 2, J ) 6.6), 6.81 (dd, 1, J )
Vol. 11, No. 6, 2007 / Organic Process Research & Development
•
1023

8.3, 2.5), 6.84 (d, 1, J ) 2.5), 6.86 (d, 1, J ) 8.3), 7.49 (d, 1,
J ) 2.1), 7.64 (dd, 1, J ) 8.7, 2.1), 7.80 (d, 1, J ) 8.7). 13C
NMR: δ 29.33, 34.56, 55.52, 56.13, 56.20, 65.37, 110.37,
111.62, 112.10, 118.37, 123.17, 123.72, 129.90, 133.30, 142.13,
148.71, 153.04, 162.16, 166.47, 199.53. IR: ν 1709, 1672, 1602,
1270, 1223, 1031, 533 cm-1. Anal. Calcd for C20H22O6: C,
67.03; H, 6.19. Found: C, 67.27; H, 6.27.
was added to the reaction mixture at 70 °C. After 4 h at 70 °C,
the resulting reaction mixture was cooled to room temperature
and stirred overnight. The reaction mixture was partitioned
between H2O (10 mL) and toluene (30 mL). The organic layer
was separated and washed with H2O (3 × 10 mL). The organic
layer was then transferred into a flask and cooled to 0 °C, and
1 N NaOH was added to raise the pH to 12. The organic layer
was separated, washed with H2O (10 mL) and brine (10 mL),
dried over Na2SO4, filtered, and concentrated to provide a crude
oil. Purification via filtration through SiO2 with 40% EtOAc/
hexanes as eluent afforded 6-methoxy-1-methyl-1-(trifluoro-
methyl)isochroman-7-carbaldehyde ((R)-13) as an amorphous
solid (0.709 g, 60% from 39). Spectroscopic data were identical
with those of material prepared from the previous method.
(2S,3S)[((1R)-6-Methoxy-1-methyl-1-(trifluoromethyl)iso-
chroman-7-yl)methyl](2-phenylpiperidin-3-yl)amine Dihy-
drochloride (4 ·2HCl). To a solution of 6-methoxy-1-methyl-
1-(trifluoromethyl)isochroman-7-carbaldehyde (13; 50.0 g, 182
mmol) and (2S,3S)-(2-phenylpiperidin-3-yl)amine dimandelate
(5·32; 80.6 g, 219 mmol) in CH2Cl2 (1.0 L) was added TEA
(61.0 mL, 437 mmol). The mixture was stirred for 15 min,
sodium triacetoxyborohydride (77.0 g, 363.0 mmol) was added,
and the reaction mixture was stirred for 2 h. The reaction was
quenched by addition of 4 N NaOH (456 mL), and the layers
were separated. The organic layer was washed with 1 N HCl
(500 mL to pH 6). The CH2Cl2 was displaced with MeOH until
absence of CH2Cl2. To the mixture (375 mL) was added a
solution of concentrated HCl (33 mL) in H2O (90 mL), and
the mixture was stirred overnight, cooled to 10 °C, and filtered.
The crude solid was triturated in 3:1 MeOH/H2O (2 × 500 mL)
and filtered to afford ((6-methoxy-1-methyl-1-(trifluorometh-
yl)isochroman-7-yl)methyl)(2-phenylpiperidin-3-yl)amine di-
hydrochloride (4·2HCl; 80.8 g, 87%) as a >99:1 mixture of
diastereoisomers. The spectroscopic and other analytical data
were identical with those of the material prepared by the
previous route.
(1R)-3,4-Dimethoxybenzoic Acid 2-[5-Methoxy-2-(2,2,2-
trifluoro-1-methyl-1-((trimethylsilanyl)oxy)ethyl)phe-
nyl]ethyl Ester ((R)-39). To a solution of 3,4-dimethoxybenzoic
acid 2-(2-acetyl-5-methoxyphenyl)ethyl ester (38; 200 mg, 0.558
mmol) and catalyst 35 (10.1 mg, 4 mol %) in CH2Cl2 (2.0 mL)
at -50 °C was added CF3TMS (0.12 mL, 8.1 mmol). The
solution was stirred at -50 °C for 3 h and warmed to room
temperature overnight. HPLC analysis showed a 97% conver-
sion and 92% ee. The mixture was poured into H2O and
extracted with CH2Cl2. The organic extract was concentrated
to provide an oil. For analytical purposes, the material was
purified by chromatography on SiO2 (EtOAc/Hex, 20:80). 1H
NMR: δ 0.22 (s, 9), 1.96 (s, 3), 3.42 (ddd, 1, J ) 13.4, 8.4,
6.5), 3.56–3.65 (m, 1), 3.81 (s, 3), 3.97 (s, 3), 3.98 (s, 3),
4.44–4.59 (m, 2), 6.78 (dd, 1, J ) 9.0, 2.8), 6.94 (d, 1, J )
8.4), 6.96 (d, 1, J ) 2.8), 7.34 (d, 1, J ) 9.0), 7.60 (d, 1, J )
1.9), 7.74 (dd, 1, J ) 8.4, 1.9). 13C NMR: δ 2.31, 24.84, 33.40,
55.29, 56.13, 56.20, 66.09, 79.24 (q, J ) 30.1), 110.45, 111.53,
112.13, 117.92, 123.10, 123.79, 124.18 (q, J ) 286), 129.76,
130.05, 140.09, 148.80, 153.15, 159.43, 166.56. IR: 1711, 1604,
1515, 1270, 1223, 1174, 1134, 1027, 861, 846, 763 cm-1. Anal.
Calcd for C24H31F3O6Si: C, 57.58; H, 6.24. Found: C, 57.16;
H, 6.30.
6-Methoxy-1-methyl-1-(trifluoromethyl)isochroman-7-
carbaldehyde (R)-(13). To a solution of (1R)-3,4-dimethoxy-
benzoic acid 2-[5-methoxy-2-(2,2,2-trifluoro-1-methyl-1-((tri-
methylsilanyl)oxy)ethyl)phenyl]ethyl ester ((R)-39; 2.16 g, 4.31
mmol) in THF (20 mL) at 0 °C was added t-BuOK (0.734 g,
6.54 mmol). The reaction mixture was warmed slowly to room
temperature and was stirred overnight. To the reaction mixture
was added 1 N NaOH (10 mL), and the mixture was stirred
vigorously for 30 min and was partitioned between H2O (10
mL) and toluene (20 mL). The organic layer was washed with
1 N HCl (10 mL) and brine (10 mL), dried over Na2SO4,
filtered, and concentrated to a brown oil, which was taken
directly into the next step. To HMTA (0.738 g, 5.26 mmol)
was added TFA (9.0 mL), and the mixture was heated to 70
°C for 2 h. A solution of the crude mixture of 9 and 41 in TFA
(5 mL) was then added to the reaction mixture. The solution
was stirred for 4 h at 70 °C, after which GC/MS and HPLC
analyses showed no remaining starting material. Water (10 mL)
Acknowledgment
We thank Paul D. Hill, Jerry A. Murry (currently at Amgen),
John A. Ragan, Kunio Satake (Discovery Chemistry), and Cliff
Meltz for their contributions and helpful discussions. We also
acknowledge Ernest B. Pollard and Raymond J. Bemish for
analytical support and Ronald Bates (currently at Bristol-Myers
Squibb) and Teri L. Hochdorfer for their work in the chiral
chromatography separation of diol 22.
Received for review August 16, 2007.
OP7001886
1024
•
Vol. 11, No. 6, 2007 / Organic Process Research & Development