Asymmetric synthesis of telcagepant, a CGRP receptor antagonist for the treatment of migraine

Article abstract of DOI:10.1021/jo101704b

A highly efficient, asymmetric synthesis of telcagepant (1), a CGRP receptor antagonist for the treatment of migraine, is described. This synthesis features the first application of iminium organocatalysis on an industrial scale. The key to the success of

Full text of DOI:10.1021/jo101704b

pubs.acs.org/joc
Asymmetric Synthesis of Telcagepant, a CGRP Receptor Antagonist for
the Treatment of Migraine
Feng Xu,* Michael Zacuto, Naoki Yoshikawa, Richard Desmond, Scott Hoerrner,
Tetsuji Itoh, Michel Journet, Guy R. Humphrey, Cameron Cowden,
Neil Strotman, and Paul Devine
Department of Process Research, Merck Research Laboratories, Rahway, New Jersey 07065, United States
feng_xu@merck.com
Received September 1, 2010
A highly efficient, asymmetric synthesis of telcagepant (1), a CGRP receptor antagonist for the
treatment of migraine, is described. This synthesis features the first application of iminium
organocatalysis on an industrial scale. The key to the success of this organocatalytic transformation
was the identification of a dual acid cocatalyst system, which allowed striking a balance of the
reaction efficiency and product stability effectively. As such, via an iminium species, the necessnary
C-6 stereogenicity was practically established in one operation in >95% ee. Furthermore, we enlisted
an unprecedented Doebner-Knoevenagel coupling, which was also via an iminium species, to
efficiently construct the C3-C4 bond with desired functionality. In order to prepare telcagepant (1)
in high quality, a practical new protocol was discovered to suppress the formation of desfluoro
impurities formed under hydrogenation conditions to <0.2%. An efficient lactamization facilitated
by t-BuCOCl followed by a dynamic epimerization-crystallization resulted in the isolation of
caprolactam acetamide with the desired C3 (R) and C6 (S) configuration cleanly. Isolating only three
intermediates, the overall yield of this cost-effective synthesis was up to 27%. This environmentally
responsible synthesis contains all of the elements required for a manufacturing process and prepares
telcagepant (1) with the high quality required for pharmaceutical use.
Introduction
plays a crucial role in migraine.2 Antagonism of CGRP
receptors, as a potential new therapy for the treatment of
migraine, could offer the advantage of avoiding the cardio-
vascular liabilities associated with other existing antimi-
Migraine is a neurovascular disorder characterized by
severe, debilitating, and throbbing unilateral headache.1
Though a leading cause of disability, it is a highly prevalent
disease with a clear unmet medical need.1 With the signifi-
cant progress achieved in the field of pathophysiology in the
past decades, to date, it is well recognized that the neuropep-
tide calcitonin gene-related peptide (CGRP), which is ex-
pressed mainly in the central and peripheral nervous system,
graine therapies.3,4 Indeed, telcagepant (1, Scheme 1),5
a
(3) For reviews, see: (a) Poyner, D. R.; Hay, D. L.; Conner, A. C. Expert
ꢀ
Opin. Drug Discovery 2009, 4, 1253. (b) Villalon, C. M.; Olesen, J. Pharmacol.
Ther. 2009, 124, 309.
(4) For lead references, see: (a) Lynch, J. J., Jr.; Regan, C. P.; Edvinsson,
L.; Hargreaves, R. J.; Kane, S. A. J. Cardiovasc. Pharmacol. 2010, 55, 518. (b)
Petersen, K. A.; Birk, S.; Lassen, L. H.; Kruuse, C.; Jonassen, O.; Lesko, L.;
Olesen, J. Cephalalgia 2005, 25, 139. (c) Durham, P. L.; Russo, A. F.
Pharmacol. Ther. 2002, 94, 77 and references therein.
(5) Paone, D. V.; Shaw, A. W.; Nguyen, D. N.; Burgey, C. S.; Deng, J. Z.;
Kane, S. A.; Koblan, K. S.; Salvatore, C. A.; Mosser, S. D.; Johnston, V. K.;
Wong, B. K.; Miller-Stein, C. M.; Hershey, J. C.; Graham, S. L.; Vacca, J. P.;
Williams, T. M. J. Med. Chem. 2007, 50, 5564.
(1) (a) Goadsby, P. J.; Lipton, R. B.; Ferrari, M. D. N. Engl. J. Med. 2002,
346, 257. (b) Stovner, L.; Hagen, K.; Jensen, R.; Katsarava, Z; Lipton, R.;
Scher, A.; Steiner, T.; Zwart, J.-A. Cephalalgia 2007, 27, 193.
(2) For lead references, see: (a) Durham, P. L. Headache 2006, 46 (suppl 1),
S3. (b) Goadsby, P. J. Drugs 2005, 65, 2557. (c) Edvinsson, L. Cephalalgia 2004,
24, 611.
DOI: 10.1021/jo101704b
r
Published on Web 10/18/2010
J. Org. Chem. 2010, 75, 7829–7841 7829
2010 American Chemical Society

Xu et al.
JOCArticle
SCHEME 1. Retrosynthetic Analysis of Telcagepant (1)
resolution (DKR)11 crystallization in the presence of a cata-
lytic amount of 8.
This route provided a means to prepare large quantities
(>500 kg) of talcagepant to support the early safety and
clinical studies. Despite the significant improvement over the
early syntheses, the overall yield was still not satisfactory
upon scale-up.12 The central problem of the synthesis essen-
tially arose from the tortuous nature of creating the C-3 and
C-6 stereogenic centers in the caprolactam ring. The C-3 (R)
configuration was established through double DKR opera-
tions. The substrate 9 with undesired C-3 (S) configuration,
created through DKR resolution first in order to build up the
desired C-6 (S) stereochemistry, must be corrected to the
desired (R) configuration upon diastereoselective hydroge-
nation via laborious operation.
As depicted in Scheme 3, we envisioned that a more
efficient synthesis could arise from direct construction of
the desired C-6 (S) stereochemistry enantioselectively. An
asymmetric addition of MeNO2 to an R,β-unsaturated iminium
intermediate formed by activating cinnamaldehyde 14 with an
organocatalyst would provide the precursor 12 in one opera-
tion.13 The field of organocatalysis has blossomed dramatically
over the past decade. Its impact on synthetic chemistry is
revolutionary, with numerous novel reactions being disco-
vered. However, to date, their industrial applications have been
mainly focused on medicinal chemistry on the small laboratory
scale.14a,15 An important advance in this field could be achieved
if we could overcome the development challenges to implement/
demonstrate the utility of organocatalysis on an industrial scale.
The next synthetic challenge was to transform the desired
substrate 11 to the seven-membered lactam. Therefore, con-
densation of aldehyde 12 with a glycine enolate equivalent 13
was required. This transformation is typically achieved via a
Horner-Wadsworth-Emmons reaction. However, the phos-
phorus waste produced on industrial scale and lack of atom-
economy were of concern to develop an environmentally
responsible synthesis. To overcome this issue, it became clear
that the application of a Knoevenagel-type condensation would
be ideal, although the literature precedents for this type of
transformation with glycine enolate equivalent 13 have been
lacking.16 Nevertheless, global reduction of nitro enamine 11
followed by alkylation and cyclization would deliver the
caprolactam 2 and/or its C-3 epimer. Epimerization9 of the
C-3 amino stereogenic center would thus deliver the desired
selective and potent CGRP receptor antagonist, was recently
demonstrated to be effective as an acute treatment for
migraine in phase III clinical trials.6
Retrosynthetic disconnection of telcagepant (1) leads to
two heterocyclic species, caprolactam 2 and piperidine 3. As
a practical synthesis of piperidine 37 was realized recently,
the key challenges to synthesize 1, in particular, became two
practical strategic aspects: (1) asymmetric setup of the C-3
and C-6 stereogenic centers in 2 and (2) effective construction
of the caprolactam moiety with desired functionalities. To
support the development of telcagepant, several syntheses of
2 have been realized.5,8,9 Scheme 2 depicts the first generation
of synthesis suitable for large scale preparation.9
In line with an early synthetic approach,8a the above suc-
cessful synthesis also relies on a diastereoselective hydrogena-
tion of dehydrocaprolactam to establish the C-6 stereogenic
center in the caprolactam ring. After careful evaluation of the
substrate of choice, a dramatic improvement in diastereos-
electivity was achieved by hydrogenating dehydrocaprolac-
tam 9 in the presence of Pd-BaSO4. However, the hydrogena-
tion gave a 13:1 mixture of cis/trans isomers in favor of the
undesired cis isomer 10. Fortunately, epimerization of the C-3
amino group of10could be achieved via anepimerizable imine
intermediate by treating 10 with a catalytic amount of alde-
hyde 8 so that the desired trans isomer 2 could be obtained. To
access the optimal hydrogenation substrate 9, a racemic
synthesis of 6 was achieved via Pd(0)-catalyzed epoxide open-
ing of 410 with acetamidomanolate 5. The racemic substrate 7
was smoothly converted to the desired chiral dehydrocapro-
lactam 9 as its ditoluoyl L-tartaric acid salt via dynamic kinetic
(12) Approximately 12% overall yield to caprolactam 2 was obtained in
hundred kilogram scale runs.
(6) (a) Ho, T. W.; Ferrari, M. D.; Dodick, D. W.; Galet, V.; Kost, J.; Fan,
X.; Leibensperger, H.; Froman, S.; Asaid, C.; Lines, C.; Koppen, H.; Winner,
P. K. Lancet 2008, 372, 2115. (b) Ho, T. W.; Mannix, L.; Fan, X.; Assaid, C.;
Furtek, C.; Jones, C.; Lines, C.; Rapoport, A. Neurology 2008, 70, 1305.
(7) (a) McLaughlin, M.; Palucki, M.; Davies, I. W. Org. Lett. 2006, 8, 3311.
(b) McLaughlin, M.; Palucki, M.; Davies, I. W. Org. Lett. 2006, 8, 3307.
(8) For early syntheses, see ref 5 and (a) Burgey, C. S.; Paone, D. V.; Shaw,
A. W.; Deng, J. Z.;Nguyen, D. N.; Potteiger, S. L.; Graham, S. L.; Vacca, J. P.;
Williams, T. M. Org. Lett. 2008, 10, 3235. (b) Steinhuebel, D. P.; Krska, S. W.;
Alorati, A.; Baxter, J. M.; Belyk, K.; Bishop, B.; Palucki, M.; Sun, Y.; Davies,
I. W. Org. Lett. 2010, 12, 4201. (c) Baxter, J. M.; Steinhuebel, D. P.; Palucki,
M.; Davies, I. W. Org. Lett. 2005, 7, 215. (d) Janey, J. M.; Orella, C. J.; Njolito,
E.; Baxter, J. M.; Rosen, J. D.; Palucki, M.; Sidler, R. R.; Li, W.; Kowal, J. J.;
Davies, I. W. J. Org. Chem. 2008, 73, 3212.
(9) Palucki, M.; Davies, I.; Steinhuebel, D.; Rosen, J. PCT Int. Appl.
WO2007120589, 2007.
(10) Intermediate 4 could be prepared from 1,2-difluorobenzene in multi-
ple steps; cf. ref 9.
(11) For a review, see: Brands, K. M.; Davies, A. J. Chem. Rev. 2006, 106,
2711.
(13) (a) Gotoh, H.; Ishikawa, H.; Hayashi, Y. Org. Lett. 2007, 9, 5307. (b)
Palomo, C.; landa, A.; Mielgo, A.; Oiarbide, M.; Pugel, A.; Vera, S. Angew.
Chem., Int. Ed. 2007, 46, 8431. (c) Zu, L.; Xie, H.; Li, H.; Wang, J.; Wang, W.
Adv. Synth. Catal. 2007, 349, 2660. (d) Hojabri, L.;Hartikka, A.;Moghaddam,
F. M.; Arvidsson, P. I. Adv. Synth. Catal. 2007, 349, 740. (e) Wang, Y.; Li, P.;
Liang, X.; Zhang, T. Y.; Ye, J. Chem. Commun. 2008, 1232.
(14) For recent reviews, see: (a) MacMillan, D. W. C. Nature 2008, 455,
304. (b) Walji, A. M.; MacMillan, D. W. C. Synlett 2007, 1477. (c) Gaunt,
M. J.; Johansson, C. C. C.; McNally, A.; Vo, N. T. Drug Discovery Today
2007, 12, 8. (d) Ley, S. V. Asymmetric Synth. 2007, 201. (e) Bernardi, L.; Fini,
F.; Fochi, M.; Ricci, A. Chimia 2007, 61, 224. (f) Lelais, G.; MacMillan,
D. W. C. Aldrichimica Acta 2006, 39, 79. (g) Palomo, C.; Mielgo, A. Angew.
Chem., Int. Ed. 2006, 45, 7876.
(15) For example, see: Ishikawa, H.; Suzuki, T.; Hayashi, Y. Angew.
Chem., Int. Ed. 2009, 48, 1304.
(16) For recent examples, see: (a) Bogen, S.; Arasappan, A.; Pan, W.;
Ruan, S.; Padilla, A.; Saksena, A. K.; Girijavallabhan, V.; Njoroge, F. G.
ꢀ
Bioorg. Med. Chem. Lett. 2008, 18, 4219. (b) Velazquez, F.; Venkatraman, S.;
Wu, W.; Blackman, M.; Prongay, A.; Girijavallabhan, V.; Shih, N.-Y.;
Njoroge, F. G. Org. Lett. 2007, 9, 3061.
7830 J. Org. Chem. Vol. 75, No. 22, 2010

Xu et al.
JOCArticle
SCHEME 2. First Generation of Large Scale Synthesis of Caprolactam 2
SCHEME 3. Retrosynthetic Analysis of Caprolactam 2
promising result (50-75% yield) and more importantly
confirmed that high enantioselectivity (95% ee) could be
achieved by using the Jørgensen-Hayashi catalyst 15.17
However, after further scrutinizing the reaction profile and
conditions, we quickly realized that the use of alcohol
solvents led to the formation of undesired byproducts in
the presence of the acidic cocatalyst PhCO2H.18 In particu-
lar, both 14 and product 12 could be trapped by alcohol
solvents to form the corresponding acetal byproducts 16 and
17, respectively. Although the reaction rate to form acetals
was slower in i-PrOH than in MeOH, the acetal formation
was inevitable and varied from batch to batch during devel-
opment work. A quick solvent survey, however, confirmed
that in Hayashi’s system the use of alcohol solvents was
crucial and optimal to obtain relatively higher yields and
reasonable reaction rates. Unless we could overcome this
limitation by developing a new system to include nonalco-
holic solvents and to harness reaction pathways without
sacrificing the yield, ee, and the reaction rate, this approach
would not be suitable for further development.
Our preliminary results in dry aprotic solvents revealed that
the reaction proceeded in very low conversion (<30%) or gave a
complicated reaction profile (Table 1, entry 3; see also Support-
ing Information). However, the use of aqueous aprotic solvents
did improve the conversion. Among the various solvents
screened, aqueous THF was identified as the most promising
solvent system. To further understand the reaction pathways
and to restore the reactivity as well as the yield in nonalcohol
solvents, the structures of several byproducts19 were unambigu-
ously elucidated (Figure 1). For mechanistic discussion about
their formation, see Supporting Information.
caprolactam 2, although a preferred, direct epimerization of
caprolactam substrates without involving an imine interme-
diate9 was unprecedented at the time of this work. Finally, to
meet the high purity specifications (no new impurities >0.1%)
required for acceptable use as an administered drug, it was
crucial for our new long-term synthesis to be developed in such
a way that the formation of impurities could be suppressed in
each step without sacrificing yield and productivity. This article
describes the successful realization of all of these goals, resulting
in a highly efficient, environmentally responsible, asymmetric
synthesis of telcagepant (1) via caprolactam 2.
Results and discussion
Enantioselective 1,4-Addition of Nitromethane. Over the
past decade, the advent of organocatalysis has brought
revolutionary impact on synthetic chemistry at an extraor-
dinary level. Rather than discovered, asymmetric iminium
catalysis was the first organocatalytic activation mode de-
signed for organic synthesis.14a By applying small organic
molecules tailored to mimic biological processes, various
novel chemical transformations have been achieved more
effectively.14 However, to the best of our knowledge, appli-
cation of iminium organocatalysis on large scale has not been
realized.14 With the fast-growing interest in this relatively
new field, asymmetric 1,4-addition of nitromethane to R,β-
unsaturated aldehydes via a chiral iminium-ion species in the
presence of chiral amine catalysts appeared in 2007.13
After several experiments, a unique, effective “cocktail”
catalyst system (t-BuCO2H/B(OH)3/5 mol % 15) was dis-
covered. The use of t-BuCO2H was essential to achieve a
higher reaction rate and conversion; however, overcharging
t-BuCO2H also had an adverse effect on the stability of
product 12, giving rise to a higher level of impurities includ-
ing 20 (Table 1, entry 5).20 In contrast, the use of B(OH)3 as a
(17) (a) Marigo, M.; Wabnitz, T. C.; Fielenbach, D.; Jørgensen, K. A.
Angew. Chem., Int. Ed. 2005, 44, 794. (b) Hayashi, Y.; Gotoh, H.; Hayashi,
T.; Shoji, M. Angew. Chem., Int. Ed. 2005, 44, 4212.
(18) PhCO2H was the optimal acid of choice in Hayashi’s system; cf. ref
13a.
(19) Many low level impurities were formed during the reaction.
(20) Preliminary results revealed the use of a combination of B(OH)3/t-
BuCO2H/15 in aq THF effectively promoted enantioselective addition of
MeNO2 on cinammaldehydes.
Our first attempt at an asymmetric addition of MeNO2 to
14 under Hayashi’s conditions13a (Table 1, entry 1) gave a
J. Org. Chem. Vol. 75, No. 22, 2010 7831

Xu et al.
JOCArticle
a
TABLE 1. Selected Results of Asymmetric 1,4-Addition of MeNO2
MeNO2
(equiv)
conversion
(%)
yield
(%)
ee
entry
1
cocatalysts
solvents
MeOH
conditions
(%)b
byproducts (%)
20 mol % PhCO2H
6
10 °C, 17 h
99
98
∼50-75
∼55-70
95
94
16a (varied from 2 to 20%),
18 (<0.5%), 19 (1%), 20 (2%)
16b (varied from 1 to 15%),
18 (<1%), 19 (1%), 20 (7.9%)
18 (<1%), 19 (3.9%), 20 (72%)
18 (<1%), 20 (2.9%)
18 (<1%), 19 (1.0%), 20 (7.2%)
18 (<1%), 19 (1%), 20 (0.7%)
18 (<0.1%), 19 (1.0%), 20 (2.8%)
2
10 mol % PhCO2H
10 mol % PhCO2H
6
i-PrOH
20 °C, 25 h
3
4
5
6
7
6
3
6
5
6
MeCN
20 °C, 14 h
99
96
94
85
94
∼20
58
65
61
10% H2O-THF 20 °C, 63 h
25% H2O-THF 20 °C, 14 h
25% H2O-THF 20 °C, 67 h
17% H2O-THF 20 °C, 29 h
95
96
95
95
15 mol % t-BuCO2H
100 mol % B(OH)3
5 mol % t-BuCO2H,
50 mol % B(OH)3
73
aAll reactions were carried out in the presence of 5 mol % of 15. bThe enantiomeric excess (ee) was determined by high performance liquid
chromatography (HPLC) analysis: ChiralPak AD-RH column, 4.6 mm ꢀ 150 mm, 5 μm particle size, 45 °C, isocratic 0.1% aq H3PO4/MeCN/MeOH
(30:45:25); flow rate 0.5 mL/min. HPLC sample preparation: 12 was converted to the corresponding hydrazone by treatment with 2,4-dinitrophenylhy-
drazine in aq H3PO4-MeCN at 40 °C for 5 min.
SCHEME 4. Process To Prepare Aldehyde 12 without Isolation
of Any Intermediates22
FIGURE 1. Identified byproducts formed during the asymmetric
addition of MeNO2 to 14.
cocatalyst led to fewer impurities, albeit at the expense of a
much slower reaction rate (Table 1, entry 6). Interestingly,
the use21 of a combination of the weak acids t-BuCO2H
(5 mol %)/B(OH)3 (50 mol %) allowed for the balance of
the reaction efficiency and product stability. As a result, the
t-BuCO2H charge was reduced from 15 to 5 mol % without
sacrificing the reaction rate or conversion (Table 1, entry 7),
commercially available 1,2-difluorobenzene (21),22 14 was
while the formation of impurities, especially 20, was success-
fully suppressed to improve the yield significantly. It was
believed that both the formation of the iminium-ion inter-
mediate (from the condensation of 14 and 15) and hydrolysis
of the subsequent enamine intermediate (from the addition
of MeNO2 to the iminium-ion intermediate) to liberate the
catalyst 15 were accelerated with the help of the acidic
cocatalysts.14 (For plausible reaction pathways, see Support-
ing Information.)
efficiently prepared in a two-pot without isolating any in-
termediates (Scheme 4). 2,3-Difluorophenyllithium, gen-
erated in situ at e55 °C with n-butyl or n-hexyllithium to
suppress its decomposition via 1,2-elimination,23 was
captured by acrolein to afford the 1,2-addition product
22. The resulting alcohol 22 was then treated with sul-
furic acid as a part of an aqueous workup so that the
corresponding rearranged cinnamyl alcohol 23 was ob-
tained in 86% assay yield.24 TEMPO (1 mol %) catalyzed
oxidation in MTBE/aq phosphate buffer at pH 10-11
gave the desired cinnamaldehyde 14 in 95% assay yield,25
With the “cocktail” conditions for the preparation of 12
in hand, we then turned our attention to developing a cost-
effective process to prepare 14. Starting from inexpensive,
(21) The use of weak acids such as t-BuCO2H and B(OH)3 does not
interfere with the proper function of amine catalyst 15. For an example of the
acidity of t-BuCO2H and its interaction with amine bases, see: Xu, F.;
Armstrong, J. D., III; Zhou, G. X.; Simmons, B.; Hughes, D.; Ge, Z.;
Grabowski, E. J. J. J. Am. Chem. Soc. 2004, 126, 13002.
(23) Coe, P. L.; Waring, A. J.; Yarwood, T. D. J. Chem. Soc., Perkin
Trans. 1 1995, 2729.
(24) The acid-promoted rearrangement had to be carried out at 60 °C.
The conditions reported in the literature for a similar rearrangement did not
ꢁ
work well for this substrate. Leleti, R. R.; Hu, B.; Prashad, M.; Repie, L.
Tetrahedron Lett. 2007, 48, 8505.
(22) Unless otherwise mentioned, the processes described here were
successfully carried out in a pilot plant on >100 kg scale.
7832 J. Org. Chem. Vol. 75, No. 22, 2010

Xu et al.
JOCArticle
SCHEME 5. Preparation of 25 via Horner-Wadsworth-Emmons
Condensation
SCHEME 6. Condensation of Aldehyde 12 with 27 in
Ac2O-Pyridine and Its Plausible Pathways
which was then directly subjected to the subsequent asym-
metric MeNO2-Michael addition.
Further exhaustive screening of various chiral amines con-
firmed that the use of catalyst 15 was deemed optimal for the
preparation of 12. It is worthwhile to point out that the prolinol
24 had a much lower reactivity than that of the corresponding
TMS-protected catalyst 15. As the reaction pathways were
harnessed under our cocatalyst system, the iminium organoca-
talysis on large scale performed extremely robustly. The crude
catalyst 15, prepared in the presence of TMSCl (1.3 equiv) and
imidazole (1.7 equiv) in THF at 50 °C, was directly used after
aqueous workup (Scheme 4). The final optimal conditions in-
volved the use of nitromethane26 (6 equiv) in 17% H2O/THF in
thepresenceof15 (5mol%),t-BuCO2H (5 mol %), and B(OH)3
(50 mol %). Typically, after 30 h at ambient temperature, 12
could be reproducibly prepared in 73% assay yield and 95% ee.
To minimize the degradation of the relatively unstable nitro
aldehyde 12, in practice, the crude solution of 12 was directly
used for the next step without further purification.
Amine-Catalyzed Doebner-Knoevenagel Condensation.
With the nitro aldehyde 12 secured, we turned our attention
to a formal aldol condensation between 12 and a glycine anion
equivalent 13. A viable solution to this synthetic challenge
would need to address the sensitive aldehyde and nitro func-
tionalities of 12, which are labile under basic conditions or at
elevated temperature.
Application of a Horner-Wadsworth-Emmons conden-
sation to install a glycine moiety was first explored. Appro-
priately mild conditions27 were quickly found to accommo-
date the use of a glycine anion equivalent 2528 with aldehyde
12. A slow addition of i-Pr2NEt to a solution of 12 and 25 in
the presence of LiCl in MeCN at 10-20 °C afforded the
desired product 26 as a mixture of olefin isomers in 70-80%
yield (Scheme 5).
The coupling reaction was effective; however, this approach
was not atom-economical. In addition, the environmental
burden for the disposal of phosphate byproducts and LiCl (3
equiv) was of concern for a route for the potential lifetime of the
drug supply. This observation inspired us to develop an more
environmentally responsible process by using readily available
acetamidomalonate acid 27 as the source of a glycine anion
equivalent, via a formal Doebner-Knoevenagel reaction,29 to
prepare enamine 26 (Scheme 6).14 While efficient with aromatic
aldehydes, this transformation usually gives low yields with
enolizable aldehydes.29
Not surprisingly, treatment of 12 with Ac2O/pyridine
under similar conditions29 afforded the desired product 26
in only 44% assay yield. The formation of azlactone 28,
generated in situ through the Ac2O-mediated cyclodehydra-
tion of 27, was crucial for the initial condensation with 12.
After further examination of the reaction mechanism, we
noticed that intermediate 30, key for decarboxylation, be-
came available via azlactone ring opening of 29 only upon
aqueous quench. At this point, it became clear that this
transformation could be ultimately efficient if a desirable
decarboxylation-elimination of 33 (Scheme 7), similar to 30,
could be achieved directly. We envisioned that 33 could arise
from direct nucleophilic addition of glycine anion equivalent 27
or 32a to iminium ion 31, which could be easily accessed by
condensation of an aldehyde with an amine.30 Ideally, under
these circumstances, only a catalytic amount of amine would be
required since turnover of the amine is achieved through
elimination of β-amino acid 33. Indeed, the decarboxylation
of 33 is reminiscent of a Doebner-Knoevenagel condensation
of an aldehyde with EtO2CCH2CO2H.29 However, the use of
(29) For lead references, see: (a) List, B.; Doehring, A.; Fonseca,
M. T. H.; Jobb, A.; Torresa, R. R. Tetrahedron 2006, 62, 476. (b) Narsaiah,
A. V.; Basak, A. K.; Visali, B.; Nagaiah, K. Synth. Commun. 2004, 34, 2893.
(c) Ragoussis, N.; Ragoussis, V. J. Chem. Soc., Perkin Trans. 1 1998, 3529
and references cited therein. For mechanistic discussion, see: (d) Corey, E. J.
J. Am. Chem. Soc. 1952, 74, 5897. (e) Corey, E. J. J. Am. Chem. Soc. 1953, 75,
1163. The decarboxylation was believed to proceed via intermediate I.
(25) Aldehyde 14 is a volatile solid and hence was not suitable for
isolation via drying by suction or vacuum oven; however, its loss through
concentration of solvents under reduced pressure was negligible. Although
the same oxidation could be carried out in the presence of 0.1 mol% of
AZADO, industrial utilization of AZADO is not feasible because of its
availability and cost. Shibuya, M.; Tomizawa, M.; Suzuki, I.; Iwabuchi, Y.
J. Am. Chem. Soc. 2006, 128, 8412.
(26) For more discussion about optimization, see Supporting Information.
(27) (a) Blanchette, M. A.; Choy, W.; Davis, J. T.; Essenfeld, A. P.;
Masamune, S.; Roush, W. P.; Sakai, T. Tetrahedron Lett. 1984, 25, 2183. (b)
Rathke, M. W.; Nowak, M. J. Org. Chem. 1984, 50, 2624.
(28) For a cost-effective preparation of 25, see: Xu, F.; Devine, P. Org.
Process Res. Dev. 2010, 14, 666.
(30) For example, see: Xu, F.; Corley, E.; Murry, J. A.; Tschaen, D. M.
Org. Lett. 2007, 9, 2669.
J. Org. Chem. Vol. 75, No. 22, 2010 7833

Xu et al.
JOCArticle
SCHEME 7. Proposed Pathways for Amine-Catalyzed Doebner-
Knoevenagel Condensation via Iminium Intermediate 31
TABLE 3. Selected Results of Doebner-Knoevenagel Condensation of
12 with Di-acid 32a in AcNMe2 at 20 °C
amine catalyst
(15-20 mol %)
assay yielda
of 34 (%)
entry
Z:E
1
2
3
4
5
6
(Bn)2NH
(Bn)MeNH
(n-Bu)MeNH
piperidine
pyrrolidine
15
64
60
60
60
85
0
>98:2
97:3
97:3
97:3
91:9
aThe yield was determined by high performance liquid chromatog-
raphy (HPLC) analysis: Ascentis Express C18 column, 4.6 mm ꢀ 100
mm, 2.7 μm particle size, 40 °C, mobile phase 0.1% aq H3PO4/MeCN;
flow rate 2.0 mL/min.
was used instead of a stoichiometric charge, as the decomposi-
tion of 12 was minimized (Table 2, entries 3 and 4).31 In addi-
tion, introducing additional H2O suppressed the desired reac-
tivity, which is understandable as the formation of iminium
species 31 became disfavored in the presence of extra water,
although 1 equiv of water was formed during the reaction.
Spontaneous decarboxylation, evidenced with CO2 evolution
observed as the reaction proceeded, confirmed the catalytic
turnover of the amine.
TABLE 2. Selected Results of Doebner-Knoevenagel Condensation of
12 and Mono-acid 27 at 20 °C
yield of
The reaction was further extended to another unprece-
dented scope by using acid 32a. We were pleased to find that
free acid 32a was a competent reagent for the desired
condensation reaction in the presence of 15 mol % of
pyrrolidine at ambient temperature, if the reaction was
carried out in solvents such as DMF or AcNMe2 that
improved the limited solubility of 32a (Table 3, entry 5).
Ultimately, pyrrolidine was confirmed to be the most effec-
tive catalyst after the screening of various amines (Table 3).
The steric bulk of the amine catalysts could dominate the
efficiency of the reaction, as evidenced by the inactivity of the
TMS-protected diphenylprolinol 15, which was used in the
preceding nitromethane addition step (Table 3, entry 6).
The successful extension of the reaction scope to the use of
diacid 32a offered a significant advantage. As such, pure acid
34a could be isolated as its crystalline n-Bu3N salt prior to its
subsequent hydrogenation as outlined in the retrosynthetic
plan (Scheme 3). This was deemed important because, in
general, the performance of a hydrogenation reaction in terms
of reaction rate and impurity profile is typically affected by the
purity of the substrate. In addition, it was also a proper choice
to conduct purification/isolation at this stage32 especially since
five steps had been carried out without any isolation/purifica-
tion starting from difluorobenzene (21).
entry
amines
mol %
solvents
26a (%)
Z:Ea
1
2
3
4
DMAP
Et3N
pyrrolidine
pyrrolidine
10
100
10
DMF
THF
THF
THF
40
0
85
50
3:1
2:1
2:1
100
aThe yield and Z:E ratio were determined by high performance liquid
chromatography (HPLC) analysis: Agilent Eclipse Plus C18 column,
4.6 mm ꢀ 50 mm, 1.6 μm particle size, 25 °C, mobile phase 0.1% aq
H3PO4/MeCN; flow rate 1.5 mL/min.
acetamidomalonic acid for a Doebner-Knoevenagel reaction
promoted via an iminium species as proposed in Scheme 7 was
unprecedented at the time of this work. In addition, extension of
the use of amine catalysis concept to the Doebner-Knoevenagel
variant with aliphatic aldehydes for selective preparation of
R,β-unsaturated esters/acids is rare and leads predominantly to
the formation of the corresponding the β,γ-unsaturated isomers
under these conditions.29 The formation of the β,γ-unsaturated
isomers suggested that the decarboxylation pathway under these
classical conditions did not occur via an intermediate similar to
33, which is fully consistent with Corey’s mechanistic studies.29d
Furthermore, in the case of telcagepant, formation of the β,γ-
unsaturated product would have destroyed the carefully estab-
lished nitromethane-bearing benzylic stereogenic center.
Nevertheless, exploration of the primary/secondary amine-
catalyzed Doebner-Knoevenagel condensation via an iminium
intermediate (Scheme 7) was initiated. A breakthrough was
achieved when it was discovered that pyrrolidine effectively faci-
litated the desired reaction. The use of 10 mol % of pyrrolidine
afforded 26 in ∼85% yield as a mixture of olefin isomers (Z:E=
2:1, Table 2, entry 3). In contrast, catalysis by DMAP afforded
the desired product 26, presumably via a less potent iminium-
like species,29a,b in poor yield (∼40%) along with the formation
of many impurities (Table 1, entry 1).
Although by using acid 32a we had achieved the proof of
concept of the strategy to prepare 34a directly, 32a was not
suitable for large scale preparation due to safety concerns of
its decomposition. In contrast, the corresponding disodium
salt dihydrate 32b33 was quite stable. Unfortunately, the use
of dihydrate salt 32b did not result in the formation of any
desired product 34a under similar conditions. At this stage, it
became clear that we had to develop a process involving
(31) Decomposition pathways including the self-condensation of alde-
hyde 12 catalyzed by amines were amplified with increasing pyrrolidine
charge.
(32) The isomeric purity (olefin geometric isomers) was not required for
the subsequent hydrogenation. However, isolation of the crystalline n-Bu3N
salt resulted in rejection of the E isomer (vide infra). Therefore, it was also
vital to improve the Z:E ratio in order to maximize the isolated yield.
(33) For a practical, one-step preparation of disodium salt 32b, see
Supporting Information.
Consistent with our mechanistic consideration (Scheme 7),
the use of a weak tertiary amine base such as Et3N, which could
not form a corresponding iminium ion, resulted in no desired
product (Table 2, entry 2). Furthermore, the reaction was clea-
ner and more efficient when a catalytic amount of pyrrolidine
7834 J. Org. Chem. Vol. 75, No. 22, 2010

Xu et al.
JOCArticle
SCHEME 8. Preparation of Hydroxyl Impurity 35
SCHEME 9. Five-Step Process To Prepare 34b without Isola-
tion of Any Intermediates
sequential salt break followed by Doebner-Knoevenagel
condensation. To accept this target-orientated synthetic
challenge, we thus needed to evolve the Doebner-Knoevenagel
condensation to the next practical level by using sodium salt
32b directly.
After several experiments, the use of H2SO4 became our
acid of choice to liberate the free acid 32a from 32b, as the use
of AcNMe2 was an optimal solvent. We also noted that the
salt break was most conveniently carried out by adding
H2SO4 and 32b in alternating portions in order to minimize
the viscosity of the thick slurry on scale. From a stoichio-
metric standpoint, the condensation reaction was observed
empirically to be more efficient when a slight excess of H2SO4
was used. In order to accommodate this increase in acid
charge, the optimal pyrrolidine charge was increased to
35 mol %. With this understanding in hand, the Doebner-
Knoevenagel condensation afforded 34a in 82% assay yield.
After aqueous workup,34 34a was then isolated as its n-Bu3N
salt in 75% yield as the pure Z isomer.
However, further scrutiny of the isolated product 34a led
to the discovery of a new impurity 35 (∼0.5-1%) in the
isolated n-Bu3N salt, leading to the formation of correspond-
ing caprolactam impurities that could not be ultimately
rejected in downstream chemistry. As such, we would have
not been able to meet the high purity specifications required
for pharmaceutical use.
The structure of 35 was unambiguously elucidated by
applying NMR and LC/MS techniques. Indeed, treatment
of 12 with 32a in the presence of Et3N did afford 35 in low
yield along with numerous byproducts. By contrast, at-
tempts to prepare 35 from the addition of H2O to the acid
34a were unsuccessful. After several carefully designed ex-
periments, we concluded that an “uncatalyzed” low-yielding,
direct coupling of 12 and 32a, bypassing the pathway via
iminium intermediate 31 (Scheme 8), was responsible for the
formation of 35. The 2 equiv of water brought into the system
by the dihydrate salt 32b could decrease the efficiency or rate
of formation of the iminium species 31, thereby rendering
this background coupling competitive and, in fact, resulting
in the formation of ∼3% of 35. In line with this observation,
the use of the dried salt 32b35 led to a faster reaction rate and
a higher yield (91% of E and Z isomers) with improved
selectivity (Z:E = 94:6). More importantly, the formation of
35 was finally suppressed to <0.4% in the crude reaction
stream. The isolated n-Bu3N salt 34a contained only 0.1% of
35, which met the high purity standards for preparing pure
telcagepant.
Thus, after several evolutions, a final manufacturing process
for the Doebner-Knoevenagel condensation was developed.36
In practice, to maintain the robustness of the process, 1.5 equiv
of anhydrous salt 32b was treated with 1.55 equiv of H2SO4 at
0-10 °C to create a slurry with acceptable viscosity. Further
robustness was met by running the reaction at 15 °C in the
presence of 35 mol % pyrrolidine to increase the stability of the
acid 32a formed in situ. Finally, n-Bu3N salt 34b was isolated in
80% yield with >99% purity as its pure Zisomer and >99% ee
upgrade. Therefore, starting from 1,2-difluorobenzene (21), 34b
was thus obtained in 48% overall yield without isolation of any
intermediates (Scheme 9).
Hydrogenation-Trifluoroethylation. With the preparation
of enamine precursor 34b established, we set forth for reduction
of the nitro group to an amine so that the trifluoroethyl group
could be installed. The use of commercially available trifluor-
oethyl triflate became our trifluoroethylation of choice. At-
tempts to convert the hydrogenated acid 36c to 37 selectively
were not successful as a result of O- versus N-alkylation
competition. Although the O,N-bis-alkylated product 38 could
be converted to desired 37, the use of an excess of trifluoroethyl
triflate to complete the reaction was not favored.
To overcome the O-alkylation issue, hydrogenated esters
36a,b were subjected to trifluoroethylation (Scheme 10).
Interestingly, because of the electronic effect of the
N-trifluoroethyl group, the alkylated product 39 did not
undergo further alkylation to 41. However, the reduced
methyl and ethyl esters 36a,b were not particularly stable37
with lactam 40 formed in up to 12% yield under these
conditions. Fortunately, further studies showed that the
formation of 40 could be eliminated by using the more
sterically hindered isopropyl ester 42.
(34) The extraction of 34a into aqueous phase was inefficient below pH 8,
and 34a decomposed slowly above pH 9. To extract 34a into i-PrOAc, the
aqueous phase was acidified to pH 2-3. Above pH 3 the extraction was
inefficient, and 34a slowly decomposed below pH 2.
(36) Further studies on the reaction scope of this novel Doebner-
Knoevenagel coupling and its application will be reported in due course.
(37) Esters 36a,b could slowly convert 40 upon storage at ambient
temperature after several weeks.
(35) Dehydrated at 80-90 °C in vacuum with dry N2 sweep.
J. Org. Chem. Vol. 75, No. 22, 2010 7835

Xu et al.
JOCArticle
SCHEME 10. Reduction and Trifluoroethylation
reaction. Otherwise, telcagepant produced from this route
would not meet the required high purity specifications.
Indeed, the highly restrictive standards for impurity tol-
erance in the pharmaceutical industry make the hydrogena-
tion more challenging for suppression of desfluoro impuri-
ties. Although dehalogenation is often observed under het-
erogeneous hydrogenation conditions, in general, effective
protocols to suppress dehalogenation to date have been
lacking. Among the reported methods,39 the use of Lewis
acids such as ZnBr2 and ZnCl2 to modulate the Pd, Pt
catalysts and minimize dechlorination/debromination of
aromatics is attractive.40 However, this protocol has not
been extended to suppress defluorination. Interestingly, after
several control experiments, preliminary results showed that
exposure of the free acid 34a, obtained after salt break in aq
HCl/EtOAc containing ∼30 mol % of n-Bu3NþHCl-,41 to
the hydrogenation conditions used for salt 34b resulted in
only ∼0.2% of desfluoro byproduct 43 (Table 4, entry 4).42
Although introducing a salt break of 34b to 34a was not ideal
for a manufacturing route, this result, in line with the
literature regarding dehalogenation, encouraged us to initi-
ate an exhaustive screening with chloride additives to sup-
press defluorination during the hydrogenation of salt 34b.
The screening results are summarized in Table 4. In terms
of robustness and reaction rate, the Pearlman catalyst (Pd-
(OH)2-C) was superior to all other catalysts screened. The use
of ZnCl2 reduced defluorination to 0.6% but was not powerful
enough to meet our requirement for <0.2% defluorination
(Table 4, entry 6). However, introducing soluble, noncovalently
bonded chloride sources (Table 4, entries 2-5) dramatically
suppressed the formation of desfluoro impurities. In contrast,
the use of insoluble NaCl showed minimal improvement to
suppress the defluorination (Table 4, entry 7). After extensively
screening various salts or acids, the use of LiCl became our
additive of choice to effectively suppress defluorination.
SCHEME 11. One-Pot Synthesis of Acid 37
At this point, although we had achieved proof of concept
of the strategy to prepare 37 directly from 34b and to
suppress defluorination, a large amount of optimization
remained in order for the one-pot process to be viable for a
commercial manufacturing process. The chemistry was
further optimized by carefully understanding the effect of
each factor on defluorination as well as on hydrogenation.
Furthermore, we found that conversion and defluorination
were relatively insensitive to reaction temperature. The use
of less LiCl, higher pressure, and relatively higher catalyst
loading led to higher conversion, faster reaction rate, and
relatively higher levels of defluorination. In contrast, intro-
ducing excess LiCl as well as other chloride sources led to a
significant decrease in reaction rate, as the defluorination
was minimized, presumably due to decreased catalyst activ-
ity. Furthermore, the use of higher catalyst loading beyond
the optimal 10-15 wt % range led to a significant increase of
A streamlined synthesis required the formation of isopro-
pyl ester 42 in situ during hydrogenation. Thus, to promote a
one-pot hydrogenation/esterification, 34b was subjected to
hydrogenation in i-PrOH in the presence of H2SO4 and
Pd-C at 50 °C (Scheme 11). Unfortunately, significant levels
of desfluoro diastereomers 43 (up to 2.5%) were observed,
while the other desfluoro isomers 44 were present at
<0.05%.38 In addition, rejection of the desfluoro impurities
was poor in downstream chemistry. With this in mind, we
realized that 43 had to be suppressed to <0.2% in the crude
(39) For a lead reference, see: Mallat, T.; Baiker, A. Appl. Catal., A 2000,
200, 3.
(40) Wu, G.; Huang, M.; Richards, M.; Poirier, M.; Wen, X.; Draper,
R. W. Synthesis 2003, 1657.
(41) Because of its hydrophobicity, n-Bu3NþHCl- tends to remain in the
organic phase.
(42) Studies also showed that 43 was not generated under hydrogen-
starved conditions, which could result in defluorination. For example, see:
Brands, K. M. J.; Krska, W.; Rosner, T.; Conrad, K. M.; Corley, E. G.;
Kaba, M.; Larsen, R. D.; Reamer, R. A.; Sun, Y.; Tsay, F. Org. Process Res.
Dev. 2006, 10, 109.
(38) The structures of the desfluoro impurities were unambiguously
characterized by applying LC/MS techniques and by synthesizing authentic
samples.
7836 J. Org. Chem. Vol. 75, No. 22, 2010

Xu et al.
JOCArticle
TABLE 4. Selected Results of Hydrogenation of 34b and Effects of Additives on Defluorinationa
entry
catalysts (10 wt % loading)
additives
H2 (psig)
temp (°C)
isomers of desfluoro 43b (%)
conversionb (%)
1
2
3
4
5
6
7
8
9
10% Pd(OH)2-C
10% Pd(OH)2-C
10% Pd(OH)2-C
10% Pd(OH)2-C
10% Pd(OH)2-C
10% Pd(OH)2-C
10% Pd(OH)2-C
5% Pd/C shell
Pd/BaSO4
50
50
35
35
35
35
50
50
50
50
1.06-2.5
0.12
0.17
0.12
0.15
0.61
0.78
<0.02
<0.02
>99
98
97
94
92
95
99
65
20
LiCl, 0.3 equiv
75-100
100
100
100
100
100
100
100
35% HCl, 0.3 equiv
n-Bu3NHþCl-, 0.3 equiv
Et3NHþCl-, 0.3 equiv
ZnCl2, 0. Five equiv
NaCl, 0.5 equiv
LiCl, 0.5 equiv
LiCl, 1.0 equiv
aFor more screening results, see the Supporting Information. bUpon completion of esterification, the desfluoro impurities as well as the conversion
were determined by high performance liquid chromatography (HPLC) analysis: Ascentis Express C18, 4.6 mm ꢀ 100 mm, 2.7 μm particle size, 40 °C,
mobile phase 0.1% HClO4/MeCN; flow rate 2.0 mL/min.
43 as the reaction was overexposed to hydrogenation. The
effect of the chloride anion has been studied in depth for
production of consistent hydrogenation results, but its mec-
hanism remains unclear. In conclusion, an effective protocol
to suppress defluorination by introducing soluble Cl- to
hydrogenation was achieved.
SCHEME 12. Process To Prepare 46 without Isolation of Any
Intermediates
Taking into consideration the overall influence of each factor,
optimally balanced conditions were obtained. Hydrogenation of
salt 34b was thus performed in the presence of 0.3 equiv of LiCl,
2.5 equiv of H2SO4, and 10-15 wt % of 10 wt % Pd(OH)2-C
(50% wet) at 50-100 psig of hydrogen between 35 and 50 °C. As
a result, upon achieving completion of the esterification of the
crude hydrogenated mixture of 36c and 42 via azeotropic
distillation at 65-70 °C, the defluorination impurity 43 was
reproducibly suppressed to <0.2% in the crude reaction stream,
thus securing a process to produce telcagepant which met the
quality requirements of the pharmaceutical industry.
Et3N was added to the above crude hydrogenation stream
of 42 to neutralize the acids. Once the trifluoroethylation
was brought to completion with 1.1 equiv of CF3CH2OTf
(Scheme 10),43 the reaction was quenched with aq NaOH
while ester 42 was simultaneously hydrolyzed. The resulting
crude acid 37 was extracted into i-PrOAc at pH 4-5 in >95%
yield and most of n-Bu3N was removed during the aqueous
workup for the ease of the isolation of subsequent product.
Cyclization. Our initial cyclization process called for the
use44 of P(OPh)3 as a dehydration promoter at ∼110 °C in
toluene-AcNMe2, which effectively gave the desired diastereo-
mers 45/46 in 87% assay yield. Despite of the efficiency of this
cyclization,45 the search for an environmentally friendly “green”
process continued due to our concerns regarding the phos-
phorus waste and phenol generated during the cyclization.
Taking advantage of the decreased nucleophilicity46 of the
nitrogen tethered with the trifluoroethyl group, we envi-
sioned that the formation of the caprolactam ring could be
realized via an appropriate anhydride species, selective for-
mation of which could be possible by activating the carbox-
ylate without acylating the secondary amine moiety. Indeed,
after several experiments, we were pleased to find that slowly
introducing 1.2 equiv of t-BuCOCl at 35 °C in the presence of
5 mol % of DMAP and 2.1 equiv of Et3N in i-PrOAc was
extremely efficient in promoting the desired cyclization and
affording a mixture of diastereomer 45/46 in 89% assay
yield47 (Scheme 12).
At this stage, a further streamlined synthesis of talcage-
pant would require the development of a base-catalyzed, dy-
namic crystallization process30 to isolate the desired 46 by
epimerizing the undesired isomer 45. It should be noted that a
selective epimerization of a C-3 amino capolactam without
involving an in situ formation of an imine intermediate9 was
not well studied at the time of this work. After screening several
conditions, we found that treatment of 45 with strong bases such
as NaOH in aq DMSO promoted the epimerization cleanly and
effectively. The desired caprolactam 46 was stable under these
conditions, and neither hydrolysis of the acetamide nor capro-
lactam ring opening was observed. The thermodynamic equi-
librium ratio of 46:45 in homogeneous aq NaOH-DMSO was
determined to be 97:3 at ambient temperature. In practice, after
aqueous workup, the crude reaction mixture of 45 was simply
subjected to aq NaOH-DMSO for crystallization at 50 °C. The
desired 46 was thus isolated in 81% yield and >99% purity,
while >99% of the undesired 45 was converted to 46 under this
preferential crystallization conditions.
-
(43) The resulting weak acid HSO4 from Et3N neutralization did not
affect the alkylation.
(44) For example, see : Mitin, Y. V.; Glinskays, O. V. Tetrahedron Lett.
1969, 60, 5267.
(45) It is interesting to note that intramolecular ring closure of 37 could
also occur just by heating the acid 37 itself in solvents such as AcNMe2,
although the reaction was extremely slow (∼10% conversion after 1 week).
(46) The decreased nucleophilicity was evidenced by unsuccessful bis-
trifluoroethylation of 39 under various conditions (vide supra).
(47) The effect of concentration on reaction yield is not significant.
Typically, the reaction was carried out in ∼0.25 M of 37. Interestingly,
starting from 1:1 mixture of 37, the ratio of 46:45 was increased to ∼3:1-4:1
during the cyclization by using either P(OPh)3 or t-BuCOCl. Resubjecting a
mixture of 45/46 to the cyclization conditions did not result in a ratio change.
J. Org. Chem. Vol. 75, No. 22, 2010 7837

Xu et al.
JOCArticle
SCHEME 13. Endgame of Telcagepant Synthesis
In summary, sequential hydrogenation, alkylation, and cycli-
zation in combination with efficient epimerization/crystalliza-
tion workup made this robust through-process extremely effi-
cient. Starting from n-Bu3N salt 34b, 46 was obtained in 73%
overall yield without isolation of any intermediates.
Endgame. As mentioned above, acetamide 46 is stable against
hydrolysis under basic conditions. After evaluating various con-
ditions, the use of HCl in aq i-PrOH became our hydrolysis of
choice. Treatment of 46 with 6.7 equiv of concd HCl in a solution
of i-PrOH/water (2:1) at 80 °C for ∼17 h gave the desired amine
2 in 85% assay yield. The byproduct acid 47 was reduced from
7% to 0.5% through a simple aqueous workup.
Although the process had produced caprolactam 2 in a
decent yield, even a small improvement in chemical yield at a
late stage of the synthesis would result in a significant benefit to
a manufacturing route. With more experiments carefully de-
signed to understand each factor affecting the hydrolysis, we
found that the formation of the acid 47 could be substantially
reduced by further decreasing the HCl charge, although redu-
cing the reaction rate significantly. To overcome the decrease in
reaction rate caused by lowering the HCl charge, the hydrolysis
was thus carried out at 95 °C under pressure (25 psig of N2).
This simple modification, easily implemented in large scale
preparation, successfully reduced the formation of the acid 47
to ∼3% and improved the yield to 90%. Finally, 2 was isolated
as its HCl salt MTBE solvate in 83% yield in >99% purity with
near perfect ee (Scheme 13).48
Finally, with caprolactam 2 MTBE solvate HCl salt in hand,
the new synthesis merged with the previously reported synthesis
of talcagepant in the last coupling step.5,8,9 Treatment of 2 with
CDI in the presence of Et3N followed by addition of 3 afforded
telcagepant 1, which was then isolated as its potassium salt
EtOH solvate in 92% yield with >99.8% purity and >99.9%
ee. Thus, the new synthesis met all quality attributes in the
pharmaceutical industry.
scale has been described. The entire synthesis is carried out with a
minimum number of isolations and operations. The synthesis
features the first application of iminium organocatalysis on
industrial scale. The key to the success of our application is to
overcome the development challenge by developing a unique
organocatalyst “cocktail” system (5 mol % t-BuCO2H/50 mol
% B(OH)3/5 mol % 15) to harness the reaction pathways. As
such, the C-6 stereogenic center via an iminium species is
practically established in one operation in >95% ee. Further-
more, a subsequent novel Doebner-Knoevenagel reaction cat-
alyzed by pyrrolidine, which is also via an iminium species,
has been thoroughly developed so that the preparation of
34b is achieved efficiently by utilizing the crude 12 and diacid
sodium salt 32b. Guided by a mechanistic understanding of the
Doebner-Knoevenagel reaction, a full control of the suppres-
sion of the hydroxyl impurities 35 has been realized. Starting
from commercially available 1,2-difluorobenzene, a robust
through-process affords the crystalline enamine acid n-Bu3N
salt 34b in 48% isolated overall yield and >99% purity. One-
step global hydrogenation/esterification affords the amine 42 in
high yield. In order to produce pure telcagepant for clinical use, a
practical and effective protocol was developed to suppress the
formation of the desfluoro byproducts during the hydrogena-
tion; as such, defluorination in the presence of LiCl could be
reproducibly controlled to <0.2%. After trifluoroethylation of
42, an efficient lactamization in the presence of t-BuCOCl
followed by a base-catalyzed dynamic crystallization delivers
caprolactam acetamide 46, the second isolated crystalline solid in
this synthesis, in >99% purity, >99% ee, and 73% yield over
three steps. Finally, coupling caprolactam 2 with 3 completes the
synthesis of telcagepant (1, >99.8% purity and >99.9% ee).
Isolating only three intermediates, the overall yield of this cost-
effective synthesis is up to 27%.
The route described in this manuscript is an example of a
synthetic target driving the discovery of new chemistries: an
industrial application of a dual acid cocatalyst system for
enantioselective 1,4-addition of nitromethane, a novel amine-
catalyzed Doebner-Knoevenagel type of coupling as well as a
discovery in suppressing the formation of desfluoro impurities
under hydrogenation conditions. Containing all of the elements
required for a manufacturing process, the route is efficient and
prepares 1 with the high quality required for pharmaceutical
Conclusion
In summary, a highly efficient, asymmetric synthesis of telca-
gepant (1) suitable for implementation on the manufacturing
(48) Alternatively, the hydrolysis could be performed in aq HCl/HOAc to
afford a similar yield. However, robust rejection of process impurities was
obtained through hydrolysis of 2 in aq HCl/i-PrOH.
7838 J. Org. Chem. Vol. 75, No. 22, 2010

Xu et al.
JOCArticle
use. More importantly, practicing the philosophy of “green
chemistry”, the synthesis of telcagepant was developed with a
strong interest to minimize the waste and maximize the mea-
sures taken to protect our environment without sacrificing the
efficiency of the synthesis. The success of our synthesis, as a
result of target-oriented and mechanism-guided development,
clearly demonstrates the importance of developing an environ-
mentally responsible class of chemistry.
(S,Z)-2-Acetamido-5-(2,3-difluorophenyl)-6-nitrohex-2-enoic
Acid Tri-n-butylamine Salt (34b). A concentrated solution of
aldehyde 14 prepared as above (1.189 mol assay, 200 g of 14) was
diluted with THF to a volume of 2.8 L. H2O (400 mL), pivalic acid
(6.08 g, 0.0594 mol), boric acid (36.78 g, 0.594 mol), TMS-prolinol
15 (19.36 g assay, 0.0594 mol, ∼250 mL solution in THF/MTBE;
for its preparation, see Supporting Information) and nitromethane
(436 g, 7.143 mol) were added at ambient temperature. The
resulting homogeneous solution was stirred at ambient tempera-
ture for 30 h (>94% conversion50). The reaction solution was
cooled to 2-7 °C and quenched by addition of i-PrOAc (2 L), 15%
aq NaCl (600 mL), and 1 N HCl (200 mL). The organic layer was
washed with 6% NaHCO3 in 5% aq NaCl (1 L) and 15% aq NaCl
(600 mL ꢀ 2). The organic phase was azeotropically dried and
solvent-switched to i-PrOAc while maintaining the internal tem-
perature at <15 °C. The final solution (∼2 L, ∼10 wt % of 12) was
directly used in the subsequent step or could be stored at 5 °C;
73-75% assay yield of 12; ∼95% ee.
Experimental Section49
(E)-3-(2,3-Difluorophenyl)acrylaldehyde (14). To a solution of
1,2-difluorobenzene (21, 4.80 kg, 42.1 mol) in THF (38.4 L) in a
100 L flask was added n-hexyllithium (2.3 M in hexane, 19.21 L,
44.2 mol) dropwise over 3 h, while the internal temperature was
maintained at e55 °C. The resulting slurry was aged at -60 to
-55 °C for additional 0.5 h. A solution of acrolein (3.26 L, 46.3
mol) in THF (5 L) was then added dropwise over 2.5 h. After
additional 30 min at -60 to -55 °C, the cold reaction solution
was inverse quenched into water (25 L). The aqueous layer was
discarded. To the organic layer was added an aq H2SO4 solution
(6.0 L of 96% H2SO4 þ 13.4 L of water). After the reaction
mixture was agitated at 60 °C overnight under nitrogen and
cooled to ambient temperature, the layers were separated. The
aqueous layer was extracted with toluene (20 L). The combined
organic phase was washed with water (24 L) followed by half-
saturated sodium bicarbonate (12 L). The organic layer was then
concentrated under reduced pressure to ∼10 L, and the crude
was used directly in the next step. HPLC assay: 6.195 kg of the
desired product 23; 86% assay yield.
Sodium bromide (0.74 kg, 7.21 mol), potassium phosphate
dibasic (3.14 kg, 18.02 mol), and water (11.67 L) were charged to
a 100 L vessel. The solution pH was adjusted to 10.5-11.0 with
1 N KOH (∼2.5 L). A solution of the crude alcohol 23 (9.725 kg,
63.05 wt % solution in MTBE/toluene, 36.0 mol), MTBE
(34.0 L), and TEMPO (0.056 kg, 0.360 mol) were added at am-
bient temperature. The reaction solution was cooled to ∼10 °C.
Sodium hypochlorite (Chlorox, 39.7 L, 6.15 wt %, 36.0 mol) was
added dropwise over 2 h, while the internal temperature was
maintained at 10-15 °C. The aqueous layer was removed at
ambient temperature, and the organic phase was washed with
0.1 M Na2S2O3 (10 L) followed by half-saturated aqueous
sodium bicarbonate (10 L). The organic phase was concentrated
to ∼130 L, and solvent-switched to THF at constant volume
under reduced pressure by feeding THF. The residual MTBE
was less than 0.5%, and the residual toluene was less than 10%.
The desired organic layer was used directly in the next step.
HPLC assay: 95% assay yield. 1H NMR (400 MHz, CDCl3): δ
9.66 (d, J = 7.6 Hz, 1 H), 7.55 (d, J = 16.4 Hz, 1 H), 7.27 (m, 1
H), 7.17 (m, 1 H), 7.08 (m, 1 H), 6.71 (dd, J = 16.4, 7.6 Hz, 1 H).
13C NMR (100 MHz, CDCl3): δ 193.6, 151.1 (dd, J = 250.0,
12.8 Hz), 149.4 (dd, J = 256.2, 13.7 Hz), 143.4 (t, J = 3.4 Hz),
131.7 (d, J = 5.6 Hz), 124.8 (dd, J = 7.2, 4.8 Hz), 124.5 (d, J =
8.4 Hz), 123.6 (d, J = 3.4 Hz), 119.7 (d, J = 17.1 Hz). Anal.
Calcd for C9H6F2O: C, 64.29; H, 3.60. Found: C, 63.89; H, 3.26.
(E)-3-(2,3-Difluorophenyl)prop-2-en-1-ol (23). 1H NMR (400
MHz, CDCl3): δ 7.21 (m, 1 H), 7.05 (m, 3 H), 6.77 (d, J = 16.1
Hz, 1 H), 6.49 (dt, J = 16.1, 5.4 Hz, 1 H), 4.37 (dd, J = 5.4, 1.2
Hz, 2 H), 1.65 (s, 1 H). 13C NMR (100 MHz, CDCl3): δ 151.2
(dd, J = 247.7, 12.8 Hz), 148.5 (dt, J = 250.5, 13.5 Hz), 132.9 (d,
J = 4.8 Hz), 127.0 (d, J = 9.6 Hz), 124.0 (dd, J = 7.2, 4.8 Hz),
122.5 (t, J = 3.4 Hz), 122.4 (t, J = 3.1 Hz), 116.1 (d, J = 17.2
Hz), 63.78. Anal. Calcd for C9H8F2O: C, 63.53; H, 4.74. Found:
C, 63.85; H, 4.53.
A three-neck flask equipped with an overhead stirrer, a
thermocouple, and a nitrogen inlet was charged with Me2NAc
(700 mL) and anhydrous di-Na salt 32b (134.2 g, 0.654 mol). To the
resulting slurry at 0-10 °C was added H2SO4 (96%, 75.0 mL,
1.352 mol) dropwise while the internal temperature was main-
tained at <10 °C with external cooling. The slurry was agitated at
0-10 °C for 1 h, and more anhydrous di-Na salt 32b (134.2 g, 0.654
mol) was added in portions over 1-2 h. The resulting slurry was
agitated between 0 and 10 °C for 4-6 h. Pyrrolidine (25.5 mL,
0.306 mol) was then added dropwise. The slurry was aged between
0 and 10 °C for 30 min. Then, nitro aldehyde 12 (0.8727 mol assay,
200 g of 12 in ∼2 L of i-PrOAc) was charged over 30 min. The
reaction mixture was aged between 15 and 20 °C for 20-24 h.
Aqueous Na2CO3 (4 wt %, ∼3.5 L) was added to adjust the pH to
8-9 while the internal temperature was maintained at <25 °C.
The desired aq phase was separated at 20-25 °C. i-PrOAc (1.8 L)
was added. The biphasic solution was adjusted to pH = 2-3 by
adding 5 N HCl (∼700 mL) slowly. The organic phase was
separated, and the aq phase was extracted with i-PrOAc (600
mL). The combined organic phase was washed with brine (15 wt
%, 300 mL); 91% assay yield of E and Z isomers (261 g, Z:E =
∼95:5). The organic phase was azeotropically dried and solvent
switched to i-PrOAc at a volume of ∼1.23 L.
A 5 L flask equipped with a reflux condenser and an overhead
stirrer was charged with n-Bu3N (267 mL, 1.121 mol) and i-PrOAc
(735 mL). After about ∼15% (370 mL) of the above i-PrOAc
solution of the free acid 34a was charged dropwsie at 45 °C, the
batch was seeded with n-Bu3N salt 34b (1.9 g) and agitated at 45 °C
for 1-2 h. The remaining i-PrOAc solution of the free acid 34a was
added at 45 °C over 3-4 h. The resulting slurry was aged at 45 °C
for an additional 1 h, and then heptane (980 mL) was added drop-
wise at 45 °C over 3 h. The resulting slurry was aged at 45 °Cfor1h.
The batch was gradually cooled to 0-5 °C over 3 h and stirred at
0-5 °C for 3-6 h. The solids were collected by filtration and
washed with 40% heptane/i-PrOAc (735 mL ꢀ 2) while maintain-
ing the batch temperature at 0-5 °C. Drying at 45 °C in vacuum
afforded the desired n-Bu3N salt 34b (359 g) in >99% Z diastereo-
1
mer and 99.1% ee; 80% isolated yield from 12. H NMR (400
MHz, d6-DMSO): δ 9.02 (s, 1 H), 7.33 (m, 1 H), 7.29 (m, 1 H), 7.21
(m, 1H), 6.14 (t, J = 6.8 Hz, 1 H), 4.95 (m, 2 H), 3.91 (m, 1 H), 2.42
(m, 6 H), 1.92 (s, 3 H), 1.38 (m, 6 H), 1.27 (m, 6 H), 0.87 (t, J = 7.3
H, 9 H). 13C NMR (100 MHz, d6-DMSO): δ 167.8, 166.1, 149.7
(50) Nitroaledehyde 12 gave a broad peak by HPLC assay, which was
difficult to monitor and assay. However, treatment of 12 with 2,4-dinitro-
phenylhydrazine gave the corresponding hydrazone derivative quantita-
tively, which gave a sharp peak by both chiral and achiral HPLC assays.
HPLC sample preparation: To a solution of 2,4-dinitrophenylhydrazine
(∼10 mg/mL in MeCN, 2 mL) and H3PO4 (25 μL) in a 25 mL volumetric
flask was added the reaction solution (∼100 μL). The resulting solution was
aged at 40 °C for 5 min and diluted to 25 mL with acetonitrile prior to use.
(49) Unless otherwise mentioned, the processes described here in much
larger scale were carried out in pilot plant successfully and reproducibly.
J. Org. Chem. Vol. 75, No. 22, 2010 7839

Xu et al.
JOCArticle
(dd, J = 245.5, 13.0 Hz), 148.2 (dd, J = 246.9, 12.9 Hz), 130.8,
129.4 (d, J = 10.6 Hz), 127.9, 125.1 (dd, J = 6.7, 4.3 Hz), 124.1,
116.3 (d, J = 16.4 Hz), 78.3, 51.7, 35.8, 30.6, 26.0, 22.7, 19.7, 13.6.
Anal. Calcd for C26H41F2N3O5: C, 60.80; H, 8.05; N, 8.18. Found:
C, 60.65; H, 8.09; N, 8.22.
1.99 (s, 3 H of the major isomer), 1.83 (m, 3 H of the major
isomer), 1.83-1.42 (m, 4 H), 1.23-1.15 (m, 6 H). 13C NMR (100
MHz, CDCl3): δ 172.1 (major isomer), 172.0 (minor isomer),
170.0 (major isomer), 169.8 (minor isomer), 150.9 (dd, J =
248.5, 13.5 Hz), 149.6 (dd, J = 246.3, 12.8 Hz, major isomer),
149.5 (dd, J = 245.7, 12.6 Hz, minor isomer), 132. Two (d, J =
11.3 Hz), 124.4 (m), 123.5 (t, J = 3.5 Hz, major isomer), 123.3 (t,
J = 3.4 Hz, minor isomer), 115.5 (d, J = 16.9 Hz), 69.4, 52.4
(minor isomer), 52.0 (major isomer), 47.0 (major isomer), 46.8
(minor isomer), 42.8 (major isomer), 42.7 (minor isomer), 23.35
(major isomer), 23.30 (minor isomer), 21.8 (major isomer), 21.7
(minor isomer). 19F NMR (376 MHz, CDCl3): major isomer δ
-137.87 (d, J = 21.1 Hz), -143.1 (d, J = 21.1 Hz); minor
isomer δ -137.89 (d, J = 21.1 Hz), -143.4 (d, J = 21.1 Hz).
(3R,6S)-3-Amino-6-(2,3-difluorophenyl)-1-(2,2,2-trifluoroethyl)-
azepan-2-one Hydrochloride Methyl tert-Butyl Ether Solvate (2).
To a suspension of acetamide 46 (800 g, 2.2 mol) in i-PrOH (1.2 L)
were added H2O (2.96 L) and 37 wt % hydrochloric acid (758 g,
7.7 mol). The vessel was closed and pressurized with nitrogen
(25 psi). The resulting mixture was heated to 95 °C for 17 h until
g99% conversion was achieved. The reaction mixture was cooled
to 50 °C, diluted with i-PrOH (1.2 L), and further cooled to
ambient temperature. Aqueous NaOH (5 N, 2.2 L, 11 mol) was
added, maintaining the batch temperature below 30 °C. The
reaction mixture was extracted with MTBE (4 L). The organic
layer was washed with 0.2 N NaOH (2.4 L) followed by H2O
(2.4 L). The organic phase was solvent switched to i-PrOH in vacuo
azeotropically to a final volume of 5 L with 1-0.3 wt % water.
MTBE (960 mL) followed by 2.56 N HCl in aq i-PrOH (0.24 mol,
94 mL; prepared by diluting 212 mL of 37 wt % HCl with i-PrOH
to 1 L) was added at 30 °C. The resulting solution was seeded with
caprolactam HCl salt MTBE solvate 2 (26 g) and aged at 30 °C for
1 h. More 2.56 N HCl in aq i-PrOH (2.29 mol, 894 mL; see above
for the preparation) was added dropwise at 30 °C over 4 h. After
addition, the batch was cooled to ambient temperature over 1 h and
aged for 3 h before filtration. The wet cake was washed with 30%
MTBE/i-PrOH (1.8 L, displacement wash) followed by MTBE
wash (2.5 L ꢀ 2). Vacuum drying under nitrogen at <25 °C
afforded 840 g of caprolactam HCl salt MTBE solvate 2 as a white
solid in 83% yield with 99.9% purity and 100% ee. 1H NMR (400
MHz, d6-DMSO): δ 8.57 (s, 3 H), 7.32 (m, 1 H), 7.22 (m, 2 H), 4.69
(d, J = 11.2 Hz, 1 H), 4.46 (m, 1 H), 4.39 (m, 1 H), 4.20 (m, 1 H),
3.39 (d, J = 15.4 Hz, 1 H), 3.07 (s, 3 H),, 3.06 (m, 1 H), 2.18 (m, 1
H), 2.07 (m, 1 H), 1.99 (m, 1 H), 1.74 (m, 1 H), 1.10 (s, 9 H). 13C
NMR (100 MHz, d6-DMSO): δ 171.4, 149.7 (dd, J = 246.2, 13.5
Hz), 146.9 (dd, J = 244.3, 12.8 Hz), 133.1 (d, J = 11.7 Hz), 125.1
(dd, J = 4.7, 7.4 Hz), 124.7 (q, J= 280.5 Hz), 123.4 (t, J= 2.9Hz),
115.5 (d, J = 16.7 Hz), 72.0, 52.6, 51.8, 48.7, 47.8 (q, J = 32.5 Hz),
36.6, 33.0, 28.4, 26.8. Anal. Calcd for C19H28ClF5N2O2: C, 51.07;
H, 6.32; N, 6.27. Found: C, 50.82; H, 6.28; N, 6.27.
N-((3R,6S)-6-Difluorophenyl)-2-oxo-1-(2,2,2-trifluoroethyl)-
azepan-3-yl)acetamide (46). A mixture of n-Bu3N salt 34b (150 g,
0.292 mol), H2SO4 (96%, 40.5 mL, 0.73 mol), LiCl (3.71 g,
0.0876 mol), and Pearlman’s catalyst (20 g, 10 wt % Pd(OH)2 on
carbon contained ∼50% water) in i-PrOH (900 mL) was hydro-
genated under 50 psig H2 at 40 °C for ∼20 h. The catalyst was
removed through filtration and washed with i-PrOH (300 mL).
The combined filtrate was azeotropically dried (KF <1000) under
vacuum with i-PrOH, while the internal temperature could be
maintained at 60 °C. The reaction solution was aged at 60-65 °C
for 8 h to complete esterification (>98% conversion). Then, the
batch was further concentratde to a volume of ∼500 mL and KF<
3000; 95% assay yield of the hydrogenated amine ester 42 (95 g
assay in ∼500 mL i-PrOH, 0.278 mol). Et3N (174 mL, 1.25 mol)
was added at ambient temperature dropwise. The solution was then
warmed to 55 °C, and CF3CH2OTf (50 mL, 0.347 mol) was added
dropwise over 1-2 h while the reaction temperature was main-
tained between 50 and 60 °C. The reaction solution was aged at
60 °C for 8 h until >99% conversion was achieved. The reaction
solution was then cooled to ambient temperature, and 5 N NaOH
solution (306 mL, 1.53 mol) was added dropwise over 0.5-1 h. The
resulting solution was stirred for 1-2 h at 30-35 °C until the
saponification of the ester was deemed complete by HPLC: >99%
conversion. The reaction was cooled to ambient temperature, and
water (300 mL) and heptane (500 mL) were added. To the separated
aq phase was added 6 M HCl (∼150 mL) dropwise to adjust pH =
4-5, while the internal temperature was maintained at <25°Cwith
external cooling. The aq phase was extracted with i-PrOAc (1.2 L).
The organic phase was washed with 15 wt % brine (400 mL) and
azeotropically dried with i-PrOAc in vacuum to a volume of 1.1 L
(KF < 1000, <2 vol % i-PrOH); 95% assay yield of 46 (101 g
assay, 0.264 mol). DMAP (13.2 mmol, 1.61 g) and Et3N (77.3 mL,
0.554 mol) were added. Then, trimethylacetyl chloride (39 mL,
0.317 mol) was added dropwise over 1 h at 35-40 °C. The reaction
mixture was stirred for additional 5-8 h until >99% conversion
was achieved. The reaction mixture was cooled to ambient tem-
perature and quenched with 1 N HCl (350 mL). The organic phase
was washed with 7.5% NaHCO3 (440 mL). The organic layer was
solvent switched to DMSO (500 mL) in vacuo at 60 °C, and 2.5 N
NaOH (260 mL) was added dropwise while the internal tempera-
ture was maintained at <30 °C. After aging 30 min, the batch was
seeded and agitated overnight at ambient temperature. Water (240
mL) was added dropwise over 1-2 h to adjust the DMSO-water
ratio to 1:1. The slurry was aged for 2 h at ambient temperature
before filtration. The wet cake was washed with DMSO-water
(1:1, 160 mL) followed by water (320 mL). Vacuum drying in
vacuum at 55 °C under nitrogen gave 78 g of the desired product 46;
73% overall yield from 34b. 1H NMR (400 MHz, CDCl3): δ 7.05
(m, 2 H), 7.0 (d, J = 6.0 Hz, 1 H), 6.92 (m, 1 H), 4.84 (dd, J = 10.2,
6.2 Hz, 1H), 4.09 (m, 3 H), 3.34 (d, J = 15.5 Hz, 1 H), 3.03 (m, 1 H),
2.18 (m, 2 H), 2.07 (m, 1 H), 2.01 (s, 3 H), 1.65 (m, 1 H). 13C NMR
(100 MHz, CDCl3): δ 174.0, 169.4, 150.8 (dd, J = 249.4, 13.2 Hz),
148.1 (dd, J = 246.7, 12.8 Hz), 132.4 (d, J = 12.0 Hz), 124.7 (dd,
J = 7.2, 4.9 Hz), 124.2 (q, J = 280.7 Hz), 122.4 (t, J = 3.3 Hz),
116.0 (d, J = 17.0 Hz), 54.5, 52.2, 48.9 (q,. J = 33.8 Hz), 38.4, 33.9,
31.9, 23.2. 19F NMR (376 MHz, CDCl3): δ -70.2, -137.2 (d, J =
22.5 Hz), -143.2 (d, J = 22.5 Hz). Anal. Calcd for C16H17F5N2O2:
C, 52.75; H, 4.70; N, 7.69. Found: C, 52.81; H, 4.25; N, 7.69.
(S)-Isopropyl 2-acetamido-6-amino-5-(2,3-difluorophenyl)-
hexanoate (42). The ratio of the two isomers was about 2:1. 1H
NMR (400 MHz, CDCl3): δ 7.03 (m, 2 H), 6.90 (m, 1 H), 6.17 (d,
J = 7.6 Hz, 1 H), 4.98 (m, 1 H), 4.53 (m, 1 H of the major
isomer), 4.47 (m, 1 H of the minor isomer), 2.83-3.05 (m, 3 H),
Telcagepant Potassium Salt Ethanol Solvate (1). Telcagepant
was prepared by treating caprolactam 2 with CDI followed by 3
in the presence of Et3N in THF as reported previously.5,8,9 Upon
workup, the crude reaction stream in i-PrOAc was solvent switched
to EtOH in vacuo and treated with a solution prepared by dissolv-
ing KOBu-t in EtOH (1.3 equiv, 16 wt %) to afford telcagepant as
its crystalline potassium salt ethanol solvate in 92% yield with
>99.9% purity and >99.9% ee. 1H NMR (400 MHz, d4-MeOH):
δ 7.75 (dd, J = 5.3, 1.4 Hz, 1 H), 7.38 (dd, J = 7.6, 1.4 Hz, 1 H),
7.15 (m, 3 H), 6.70 (dd, J=7.6, 5.3 Hz, 1 H), 4.85 (d, J=11.4 Hz,
1 H), 4.55 (m, 1 H), 4.45 (dq, J = 15.4, 9.5 Hz, 1 H), 4.27 (m, 3 H),
4.05 (dq, J = 15.4, 9.0 Hz, 1 H), 3.61 (q, J = 7.1 Hz, 2 H), 3.46 (d,
J=15.4 Hz, 1 H), 3.16 (m, 1 H), 3.0 (m, 2 H), 2.42 (dq, J = 12.7, 4.4
Hz, 1 H), 2.27 (dq, J = 12.7, 4.4 Hz, 1 H), 2.16 (m, 3 H), 1.81 (m, 3
H). 1.18 (t, J = 7.1 Hz, 3 H). 13C NMR (100 MHz, d4-MeOH): δ
176.8, 166.1, 159.3, 157.4, 152.1 (dd, J= 246.8, 13.6 Hz), 149.4 (dd,
J=245.1, 13.1 Hz), 139.2, 134.7 (d, J = 11.9 Hz), 127.7, 126.3 (q,
J=279.7 Hz), 126.2 (dd, J=7.1, 4.8 Hz), 124.3 (t, J=3.4 Hz),
7840 J. Org. Chem. Vol. 75, No. 22, 2010

Xu et al.
JOCArticle
116.8 (d, J = 17.1 Hz), 114.5, 113.8, 58.5, 55.3, 55.2, 51.6, 49.9 (q,
J = 33.6 Hz), 45.4, 45.3, 39.8, 35.9, 32.7, 30.74, 30.72, 18.5.
Bazaral, and K. Emerson for experimental assistance and
helpful discussion.
Acknowledgment. We thank R. Reamer, Dr. P. Dormer,
and L. DiMichele for assistance with NMR studies,
E. Corley, Dr. T. Itoh, Dr. D. Kumke, Dr. J. Xu, Dr. R.
Hartman, Dr. A. Kalinin, Dr. L. Schenck, M. Weisel, C.
Supporting Information Available: Experimental proce-
dure, additional detailed information on mechanism discussion,
reaction condition optimization and screening. This material is
available free of charge via the Internet at http://pubs.acs.org.
J. Org. Chem. Vol. 75, No. 22, 2010 7841