Efficient and scalable enantioselective synthesis of a CGRP antagonist

Article abstract of DOI:10.1021/ol302262q

An enantioselective synthesis of the CGRP antagonist BMS-846372, amenable to large scale preparation, is presented. This new synthesis showcases a chemo- and enantioselective reduction of a cyclohepta[b]pyridine-5,9-dione as well as a Pd-catalyzed alpha-arylation reaction to form the key carbon-carbon bond and set the absolute and relative stereochemistry.

Full text of DOI:10.1021/ol302262q

ORGANIC
LETTERS
XXXX
Vol. XX, No. XX
000–000
Efficient and Scalable Enantioselective
Synthesis of a CGRP Antagonist
David K. Leahy,*,† Yu Fan,*,† Lopa V. Desai,† Collin Chan,† Jason Zhu,† Guanglin Luo,‡
Ling Chen,‡ Ronald L. Hanson,† Masano Sugiyama,† Thorsten Rosner,†
Nicolas Cuniere,† Zhiwei Guo,† Yi Hsiao,† and Qi Gao§
Chemical Development and Drug Product Science and Technology, Bristol-Myers
Squibb Company, One Squibb Drive, New Brunswick, New Jersey 08903, United States,
and Molecular Sciences and Candidate Optimization, Bristol-Myers Squibb Company,
5 Research Parkway, Wallingford, Connecticut 06492, United States
david.leahy@bms.com; yu.fan@bms.com
Received August 14, 2012
ABSTRACT
An enantioselective synthesis of the CGRP antagonist BMS-846372, amenable to large scale preparation, is presented. This new synthesis
showcases a chemo- and enantioselective reduction of a cyclohepta[b]pyridine-5,9-dione as well as a Pd-catalyzed alpha-arylation reaction to
form the key carbonꢀcarbon bond and set the absolute and relative stereochemistry.
Calcitonin gene-related peptide (CGRP) is a 37-amino
acid peptide associated with the pathophysiology of
migraines.1 Inhibition of the CGRP receptor with a small
molecule has the potential to be an effective and safe treat-
ment for people who suffer from migraines.2 The current
standard of care for migraines belongs to the triptan
class of drugs, which are vascular constrictors and suffer
from mechanism-based cardiovascular risks.3 In contrast,
a CGRP antagonist should prevent vasodilation thereby
circumventing cardiovascular risks.4 During the course of
a program directed at using CGRP antagonists for the
treatment of acute migraine, BMS-846372 (1) emerged as a
potential clinical candidate (Figure 1).5
† Chemical Development, Bristol-Myers Squibb Company
‡ Molecular Sciences and Candidate Optimization, Bristol-Myers Squibb
Company
§ Drug Product Science and Technology, Bristol-Myers Squibb Company
(1) (a) Goadsby, P. J. Drugs 2005, 65, 2557. (b) Edvinsson, L.
Cephalalgia 2004, 24, 611. (c) Williamson, D. J.; Hargreaves, R. J.
Microsc. Res. Tech. 2001, 53, 167. (d) Goadsby, P. J.; Edvinsson, L.;
Ekman, R. Ann. Neurol. 1990, 28, 183. (e) Goadsby, P. J.; Edvinsson, L.
Ann. Neurol. 1993, 33, 48.
Figure 1. CGRP antagonist BMS-846372 (1).
(2) Bell, I. M.; Bednar, R. A.; Fay, J. F.; Gallicchio, S. N.; Hochman,
J. H.; McMasters, D. R.; Miller-Stein, C.; Moore, E. L.; Mosser, S. D.;
Pudvah, N. T.; Quigley, A. G.; Salvatore, C. A.; Stump, C. A.; Theberge,
C. R.; Wong, B. K.; Zartman, C. B.; Zhang, X.-F.; Kane, S. A.; Graham,
S. L.; Vacca, J. P.; Williams, T. M. Bioorg. Med. Chem. Lett. 2006, 16,
6165.
Compound 1 is a cyclohepta[b]pyridine with a 2,3-
difluorophenyl and an O-bound carbamate situated in a
1,4-trans disposition. The relative and absolute control of
(3) Goadsby, P. J.; Lipton, R. B.; Ferrari, M. D. N. Eng. J. Med.
2002, 346, 257.
(4) Olesen, J.; Diener, H. C.; Husstedt, I. W.; Goadsby, P. J.; Hall, D.;
(5) Luo, G.; Chen, L.; Conway, C. M.; Denton, R.; Keavy, D.;
Gulianello, M.; Huang, Y.; Kostich, W.; Lentz, K. A.; Mercer, S. E.;
Schartman, R.; Signor, L.; Browning, M.; Macor, J. E.; Dubowchik,
G. M. ACS Med. Chem. Lett. 2012, 3, 337.
Meier, U.; Pollentier, S.; Lesko, L. M. N. Eng. J. Med. 2004, 350, 1104.
r
10.1021/ol302262q
XXXX American Chemical Society

Scheme 1. Retrosynthetic Analysis for 1
stereochemistry in this unusual cyclohepta[b]pyridine ring
system was expected to be the paramount challenge in
designing an efficient and scalable synthetic route to this
molecule.
Table 1. Chemo- and Enantioselective Reduction of 6
In a retrosynthetic direction, cleavage of the carbamate
bond of compound 1 simplifies the target to chiral alcohol
3 and the known amine 2 (Scheme 1).6 The addition of a
carbonyl group to compound 3 reveals a key intermediate
4 containing a synthetic handle that allows for the installa-
tion of the difluoroaryl moiety. In the forward direction,
compound 4 would be synthesized via a transition metal
catalyzed R-arylation reaction between 1-bromo-2,3-di-
fluorobenzene and ketone 5.7 The aryl group was antici-
pated to be installed in a trans disposition to the judiciously
protected alkoxy group, thus relying on substrate control
for the basis of relative stereoselectivity. In contrast, cata-
lyst control would be employed to set the target molecule’s
first chiral center, as compound 5 would be produced via a
chemo- and enantioselective reduction of the known 7,8-
dihydro-5-H-cyclohepta[b]pyridine-5,9-dione, 6.8 Finally,
the known dione 6 should be readily accessible on large
scale via a Dieckmann cyclizationꢀdecarboxylation se-
quence starting from the readily available dimethyl 2,3-
pyridinedicarboxylate.9
Following our synthetic strategy, we expected that an
enantioselective ketone reduction would be complicated
by the necessity to differentiate the two ketones present in
compound 6. As such, initial efforts were focused on an
enzymatic approach, and high throughput screening ef-
forts identified multiple enzymes that reduced diketone 6
tothe corresponding alcohol withencouragingchemo-and
enantioselectivity (Table1, entry 1). Gratifyingly, whenthe
reaction using the reductase enzyme ES-KRED-119 was
run at 2 °C, enhanced enantioselectivity (98.0% ee) was
realized (entry 2).10 While these results were very promis-
ing, the formation of diol 8 and 5-hydroxy alcohol 7 could
not be completely suppressed. An isolation protocol was
developed to purge these impurities while upgrading the
chiral purity. The chiral alcohol, 5, was isolated as its HCl
entry catalyst/enzyme conversion (%) ee (%) ratio 5:7:8
1
2
3
4
5
ES-KRED-119a
99
97
92.5
98.0
84:3:13
90:3:7
ES-KRED-119b
9c,d
9e
9f
100
100
100
g99.9
g99.9
g99.9
100:0:0
100:0:0
100:0:0
a Reaction at rt. b Reaction at 2 °C. c 9 = Rh(R-binapine)(COD)BF4.
d Reaction in MeOH at 0.1 mol % catalyst loading. e Reaction in DCM
at 0.1 mol % catalyst loading. f Reaction in DCM at 0.02 mol % catalyst
loading.
salt in 99.2% ee in 81% yield with near complete rejection
of impurities 7 and 8. While the enzymatic approach gave
us a path forward, a more direct and selective method was
desired. Since the carbonyl at the 9-position is intrinsically
more electrophilic than the one at the 5-position, differ-
entiation of diketone 6 via an enantioselective hydrogena-
tion was hypothesized to be feasible.11 Hence, a screen
of chiral rhodium complexes revealed Rh(R-binapine)-
(COD)BF4 9 as a highly chemo- and enantioselective
catalyst for this transformation (entry 3).12 This reaction
could also be performed in DCM, which aids in down-
stream processing (vide infra), and could be carried out
with as little as 0.02 mol % Rh while maintaining complete
conversion and extremely high chemo- and enantioselec-
tivity (g99.9% ee; entries 4, 5).
Having accomplished the strategic goal of differentiat-
ing the two ketones of compound 6 during the establish-
ment of the first stereogenic center, we next set out to form
the key CꢀC bond using the configuration of the substrate
to impart the desired stereocontrol. We realized that the
diastereoselectivity achieved in the key Pd-catalyzed
R-arylation would be limited by the relative thermodynamic
(6) Leahy, D. K.; Desai, L. V.; Deshpande, R. P; Mariadass, A. V.;
Rangaswamy, S.; Rajagopal, S. K.; Madhavan, L.; Illenddula, S. J. Org.
Process Res. Dev. 2012, 16, 224.
(7) For a recent review of transition metal-catalyzed ketone aryla-
tions, see: Bellina, F.; Rossi, R. Chem. Rev. 2010, 110, 1082.
(8) Jones, G.; Jones, R. K. J. Chem. Soc., Perkin. Trans. 1 1973, 27.
(9) Upon scale-up KOtBu was used instead of sodium metal for
safety considerations.
(11) For a discussion of reactivity of the substituents on the pyridine
ring, see:Uff, R. C. In Comprehensive Heterocyclic Chemistry: Katritzky,
A., Ed.; Pergamon Press: Oxford, 1997; Vol VII, Chapter 2.06.
(12) Tang, W.; Wang, W.; Chi, Y.; Zhang, X. Angew. Chem., Int. Ed.
2003, 42, 3509.
(10) ES-KRED-119 is a ketoreductase enzyme available from
SyncoZymes.
B
Org. Lett., Vol. XX, No. XX, XXXX

stability of the two diastereomers. However, because of the
readily epimerizable nature of the newly formed stereo-
center, an opportunity existed to convert the undesired
stereoisomer to the desired product. Hence, a two stage
approach could overcome this hurdle: in the first stage, the
requisite CꢀC bond would be formed, and in the second,
the stereochemistry could be controlled.
The choice of protecting group was expected to be of
paramount importance for the success of both stages of
R-arylation reaction. The triisopropylsilyl group was ulti-
mately chosen, as we hypothesized that a bulky group
would maximize the stereochemical control obtained. For
its formation, the use of TIPSCl led to incomplete conver-
sion; hence, the more reactive TIPSOTf with Et3N was
employed. Since both the asymmetric hydrogenation and
the TIPS protection could be carried out in DCM (vide
supra), a telescoped processwas developed where the crude
alcohol 5 was directly transformed to the TIPS protected
10 in 74% isolated yield. (Scheme 2).
conditions outlined by Hartwig utilizing the relatively
inexpensive and widely available PtBu3 ligand with NaOtBu
as the base, which resulted in ∼60% in-process yield of
compounds 11a and 11b.13a While these results were
encouraging, the pyrophoric nature of this ligand was a
concern to scale-up. Hence, the air stable and nonpyro-
phoric tetrafluoroborate salt of this ligand was examined
and gave comparable results (entry 4).14 Additional
gains were realized by running the reaction more con-
centrated and at a lower temperature, leading to an in-
process yield of up to ∼85% (entries 5ꢀ6). Isolation of
this compound proved to be challenging, and while
losses to the workup and mother liquor were higher than
desired, the product was isolated as a crystalline solid
from NMP/water in 63% yield with improved diastereo-
selectivity of 12:1.
With the key CꢀC bond formed, stage two of the
R-arylation sequence was examined in an effort to enhance
the dr of compound 11a. Since the desired trans-isomer 11a
was more readily crystallized than the cis-isomer 11b, and
the R-keto stereocenter was easily epimerized, we hypothe-
sized that adding a base during crystallization may enable
a crystallization-induced dynamic resolution.15 Hence,
subjection of the 12:1 mixture of compounds 11a and
11b to a catalytic amount of DBU in IPA/water for 48 h
provided compound 11a in 90% yield and 40:1 dr after
isolation. The long crystallization age was utilized to
maximizethestereochemicalpurityenhancementavailable
via equilibration.
Table 2. Optimization of Pd-Catalyzed R-Arylation Reactiona
ligand
conc
(M)
temp
yield of
entry
(4 mol %)
(°C)
11a þ 11b (%)
1
2
3
4
5
6
MePhos
0.43
120
55
56
60
60
71
80
SPhos
“
“
“
PtBu3
“
PtBu3HBF4
“
“
“
“
1.0
“
“
95
a Reaction Conditions: 1 equiv of 10, 1.2 equiv of 2,3-difluoro-1-
bromobenzene, 1.3 equiv of NaOtBu, 4 mol % Pd(OAc)2 in toluene.
With compound 10 in hand, the first stage of the key
R-arylation reaction was examined. Screening of various
literature reaction conditions revealed the following
trends.7 Typically, strong bases and sterically demanding
electron rich ligands (MePhos, SPhos, PtBu3, etc.) per-
formed best for this reaction giving high conversion and
superior impurity profile (Table 2, entries 1ꢀ3).13 Addi-
tionally, the diastereoselectivity achieved was similar for
all successful reactions (∼6:1). Further experiments ver-
ified that resubjection of a pure sample of either the trans
diastereomer 11a or the cis diastereomer 11b to the reac-
tion conditions (or any basic conditions) did indeed epi-
merize the aryl stereocenter, resulting in the observed ∼6:1
diastereoselectivity. We chose to focus our attention on the
Figure 2. ORTEP drawing of ent-11a.
Having served its purpose as a handle for installing the
aryl group, the carbonyl group needed to be removed.
Surprisingly, many common methods for direct deoxygena-
tion were unsuccessful.16 The desired product 14 could only
be obtained via a stepwise reduction: first ketone 11a was
reduced to the alcohol 12, which was then converted to the
corresponding mesylate 13 and subsequently reduced by
lithium triethylborohydride to compound 14 (Scheme 2).17
(14) Netherton, M. R.; Fu, G. C. Org. Lett. 2001, 3, 4295.
(15) Caddick, S.; Jenkins, K. Chem. Soc. Rev. 1996, 25, 447.
(16) Only intractable byproducts were observed under Wolffꢀ
Kishner conditions. No reaction was observed with silane reducing
agents. Lewis acid promoted thioketal formation was unsucessful.
(17) Holder, R. W.; Matturro, M. G. J. Org. Chem. 1977, 42, 2166.
(13) (a) Kawatsura, M.; Hartwig, J. F. J. Am. Chem. Soc. 1999, 121,
1473. (b) Fox, J. M.; Huang, X.; Chieffi, A.; Buchwald, S. L. J. Am.
Chem. Soc. 2000, 122, 1360.
Org. Lett., Vol. XX, No. XX, XXXX
C

Scheme 2. Synthesis of 1
Interestingly, it was observed that the minor mesylate
stereoisomer (cis to the aryl group) is reduced only at the
OꢀS bond, and the corresponding alcohol was recovered
from the reaction mixture. As such, it was necessary to
develop a highly diastereoselective reduction of compound
11a. The diastereoselectivity was found to be highly depen-
dent on the choice of the hydride reducing agent. While
sodium borohydride provided a disappointing dr of 3:1, the
bulky lithium tri(t-butoxyalminum)hydride reduced ketone
11a to alcohol 12 in 45:1 dr.18
The enhanced diastereoselectivity is likely due to syner-
gistic steric effects exerted by the TIPS group. The crystal
structural of ketone ent-11a provides substantial evidence
for this hypothesis (Figure 2). In solid state, the TIPS
group exerts a predominant influence on the overall struc-
tural features: the carbonyl is puckered up toward the
TIPS group with a dihedral angle (OꢀC5ꢀC11ꢀC4) of
36 degrees, and as a result, the aryl group is pointed away
from the carbonyl with a dihedral angle of (CꢀCꢀCꢀO)
of 38 degrees. In this disposition, Si face attack on the
carbonyl would be disfavored because of developing tor-
sional strain between the emerging CꢀO bond and the
C-aryl bond.19 The accentuated steric repulsion between
the bulky TIPS protecting group and the incoming hydride
reagent would further discourage the Si face attack. There-
fore when a stericly encumbered hydride reagent is em-
ployed, much enhanced diastereoselectivity is obtained.
The dr of alcohol 12 wasfurtherimprovedtogreaterthan
99:1 when it was isolated in 80% yield as hydrogen chloride
salt (Scheme 2). The mesylate 13 was formed cleanly with
mesyl anhydride, Et3N and catalytic DMAP and was
subsequently reduced using lithium triethylborohydride
at 0 °C to provide compound 14. The deprotection of
compound 14 using TBAF provided penultimate 3 cleanly.
Since compounds 13 and 14 were noncrystalline, a tele-
scoped, three-step-one-pot-isolation protocol was devel-
oped for the mesylation/reduction/deprotection sequence.
Thus 3 was produced in 76% overall yield from 12 in
>99.7% purity with one isolation.
The final step of the synthesis of compound 1 involves
the late-stage coupling of alcohol 3 with amine 2 using a
phosgene equivalent to form the final carbamate bond
(Scheme 2).6 We expected that CDI could be used as the
phosgene equivalent, which reacted smoothly with amine 2
€
and Hunig’s base to provide imidazyl-urea 15 in 86%
isolated yield. Alcohol 3 was then treated with 1.1 equiv
of imidazyl-urea 15 and KOtBu to provide the coupled
product 1 in 97% yield and >99.9 HPLC area percent
purity.
In conclusion we have developed an efficient and scalable
enantioselective synthesis of the CGRP antagonist BMS-
846372 (1). This synthesis featured a novel chemo- and
enantioselective reduction of dione 6, where electrophilicity
differences between the two ketones were exploited. Also
featured is a practical palladium catalyzed R-arylation,
which in combination with a crystallization-induced dy-
namic resolution installed the final carbonꢀcarbon bond
with excellent stereocontrol. This route provided the target
CGRP antagonist 1 in excellent chemical purity for use in
preclinical and clinical trials.
Supporting Information Available. Experimental pro-
cedures and compound characterization are available.
This material is available free of charge via the Internet
at http://pubs.acs.org.
(18) Haubenstock, H.; Eliel, E. J. Am. Chem. Soc. 1962, 84, 2362.
(19) Ashby, E. C.; Noding, S. A. J. Am. Chem. Soc. 1976, 98, 2010.
The authors declare no competing financial interest.
D
Org. Lett., Vol. XX, No. XX, XXXX