Asymmetric synthesis of cis-2,5-disubstituted pyrrolidine, the core scaffold of β3-AR agonists

Article abstract of DOI:10.1021/ol400252p

A practical, enantioselective synthesis of cis-2,5-disubstituted pyrrolidine is described. Application of an enzymatic DKR reduction of a keto ester, which is easily accessed through a novel intramolecular N→C benzoyl migration, yields syn-1,2-amino alcoh

Full text of DOI:10.1021/ol400252p

ORGANIC
LETTERS
XXXX
Vol. XX, No. XX
000–000
Asymmetric Synthesis of
cis-2,5-Disubstituted Pyrrolidine,
the Core Scaffold of β3‑AR Agonists
Feng Xu,*,† John Y. L. Chung,*,† Jeffery C. Moore,† Zhuqing Liu,† Naoki Yoshikawa,†
R. Scott Hoerrner,† Jaemoon Lee,† Maksim Royzen,† Ed Cleator,‡ Andrew G. Gibson,‡
Robert Dunn,† Kevin M. Maloney,† Mahbub Alam,‡ Adrian Goodyear,‡ Joseph Lynch,†
Nobuyashi Yasuda,† and Paul N. Devine†
Process Chemistry, Merck Research Laboratories, Rahway, New Jersey 07065,
United States, and Process Research Preparative Laboratory, Merck Sharp & Dohme
Research Laboratories, Hoddesdon, Hertfordshire, EN11 9BU, U.K.
feng_xu@merck.com; john_chung@merck.com
Received January 31, 2013
ABSTRACT
A practical, enantioselective synthesis of cis-2,5-disubstituted pyrrolidine is described. Application of an enzymatic DKR reduction of a keto ester,
which is easily accessed through a novel intramolecular NfC benzoyl migration, yields syn-1,2-amino alcohol in >99% ee and >99:1 dr.
Subsequent hydrogenation of cyclic imine affords the cis-pyrrolidine in high diastereoselectivity. By integrating biotechnology into organic
synthesis and isolating only three intermediates over 11 steps, the core scaffold of β3-AR agonists is synthesized in 38% overall yield.
β3 adrenergic receptor (β3-AR) agonists represent a new
class of agents that may provide advantages over current
therapies for the treatment of overactive bladders. Pyrro-
lidine 1, recently reported by Merck laboratories, is the
core scaffold of a novel class of potent and selective human β3-
AR agonists.1 In comparison with these acyclic β-hydroxyl-
amine β3-AR agonists,1 a pyrrolidine moiety incorporated to
link the C2,C5 substituents significantly improves the selec-
tivity and potentially the metabolic stability of drug candi-
dates but raises the chemical complexity of accessing the
evolved generation of β3-AR agonist drug candidates.
To support the drug development program, an efficient
synthesis of 1 suitable for large-scale preparation was
required. The key synthetic challenge in preparing pyrro-
lidine scaffold 1 is the effective and practical establishment
of three contiguous stereogenic centers. In particular, the
unique structure of 1 possesses an R,R (C10, C2) stereo-
chemistry setup,2 while the (C2, C5) substituted groups in
the pyrrolidine ring are in a cis relationship. Several
(2) For example, to prepare the less potent C10(R), C2(S) epimers
through a Pd-catalyzed cycloamination, see: Lemen, G. S.; Wolfe, J. P.
Org. Lett. 2010, 12, 2322–2325. Applying a similar protocol for the
preparation of 1 afforded unsatisfactory results.
(3) For recent review, see: (a) Schultz, D. M.; Wolfe, J. P. Synthesis
2012, 44, 351–361.
(4) Campos, K. R.; Klapars, A.; Waldman, J. H.; Dormer, P. G.;
† Merck Research Laboratories.
‡ Merck Sharp & Dohme Research Laboratories.
(1) (a) Morriello, G. J.; Wendt, H. R.; Bansal, A.; Salvo, J. D.;
Feighner, S.; He, J.; Hurley, A. L.; Hreniuk, D. L.; Salituro, G. M.;
Reddy, M. V.; Galloway, S. M.; McGettigan, K. K.; Laws, G.;
McKnight, C.; Doss, G. A.; Tsou, N. N.; Black, R. M.; Morris, J.; Ball,
R. G.; Sanfiz, A. T.; Streckfuss, E.; Struthers, M.; Edmondson, S. D.
Bioorg. Med. Chem. Lett. 2011, 21, 1865–1870. (b) Edmonson, S. D.;
Berger, R.; Chang, L.; Colandrea, V. J.; Hale, J. J.; Harper, B.; Kar,
N. F.; Kopka, I. E.; Li, B.; Morriello, G. J.; Moyes, C. R.; Sha, D.; Shen,
D.-M.; Wang, L.; Wendt, H.; Zhu, C. PCT Int. Appl. WO025690, 2011.
(c) Berger, R.; Chang, L.; Edmonson, S. D.; Goble, S. D.; Harper, B.; Kar,
N. F.; Kopka, I. E.; Li, B.; Morriello, G. J.; Moyes, C. R.; Shen, D.-M.; Wang,
L.; Wendt, H.; Zhu, C. PCT Int. Appl. WO123870, 2009.
Chen, C. J. Am. Chem. Soc. 2006, 128, 3538–3539.
(5) For recent examples to prepare cis-2,5-disubstituted pyrrolidines,
see: (a) Babij, N. R.; Wolfe, J. P. Angew. Chem., Int. Ed. 2012, 51, 4128–
4130. (b) Hong, Z.; Liu, L.; Sugiyama, M.; Fu, Y.; Wong, C.-H. J. Am.
Chem. Soc. 2009, 131, 8352–8353. (c) Kuwano, R.; Kashiwabara, M.;
Ohsumi, M.; Kusano, H. J. Am. Chem. Soc. 2008, 130, 808–810. (d)
Zhang, S.; Xu, L.; Miao, L.; Shu, H.; Trudell, M. L. J. Org. Chem. 2007,
72, 3133–3136. (e) Weihofen, R.; Dahnz, A.; Tverskoy, O.; Helmchen,
G. Chem. Commun. 2005, 3541–3543. (f) Aggarwal, V. K.; Astle, C. J.;
Rogers-Evans, M. Org. Lett. 2004, 6, 1469–1471. (g) Brenneman, J. B.;
Martin, S. F. Org. Lett. 2004, 6, 1329–1331.
r
10.1021/ol400252p
XXXX American Chemical Society

protocols2ꢀ5 for the asymmetric synthesis of 2,5-disubsti-
tuted pyrrolidines have been developed recently. However,
practical preparation of optically active cis-2,5-disubsti-
tuted pyrrolidines such as 1 still remains a challenging task
to date. In fact, the central problem of the initial syntheses1
of 1 essentially lies in the tortuous nature of setting the
desired stereochemistry via chiral auxiliaries1a,b or not
readily available chiral starting materials.1a,b We envi-
sioned that a straightforward approach to prepare cis-
pyrrolidine 1 (Scheme 1) could be accomplished through
an intramolecularly induced asymmetric reduction of the
corresponding pyrrolidine imine intermediate derived
from amino ketoneprecursor 2 with the C10,C2 stereogenic
centers established already. On the basis of stereofacial
bias of the pyrrolidine, in principal, high cis C2,C5 selec-
tivity should be achievable by optimizing proper reduction
conditions and/or by modifying the R group of 2 to
amplify the capability of asymmetric induction.
quantities. Alternatively, we envisioned that an effective
preparation of syn 1,2-amino alcohol could be possible via
an selective asymmetric ketone reduction of the correspond-
ing enantiomer ent-(S)-5 as a fast epimerization of 5 can be
achieved under dynamic kinetic resolution (DKR) con-
ditions (Scheme 1).
For this synthetic strategy, a one-pot through-process
was first developed to prepare the desired R-amino β-keto
ester 9 (Scheme 2). Treatment of glycine ester 7 with benzoyl
chloride under SchottenꢀBaumann reaction conditions at
0 °C, in the presence of Na2CO3 or Et3N, gave the corre-
sponding benzamide in >95% assay yield. The aqueous
phase was discarded; the organic layer was solvent switched
to dry MeCN and treated with (Boc)2O in the presence of a
catalytic amount of DMAP to afford 8. Without workup,
the reaction stream in the same pot was directly treated with
a solution of t-BuOK in THF at 0ꢀ5 °C. As such, 8 was
rearranged to 9 through an intramolecular nucleophilic
attack on the amide carbonyl group, a Chan-type NfC
benzoyl migration,10 effectively affording the desired keto
ester 9. Thus, ester 9 was directly isolated from aqueous
i-PrOH in 90% yield over three steps.
Scheme 1. Retrosynthetic Analysis of Pyrrolidine 1
Scheme 2. Through-Process to Keto Ester 9
With 9 in hand, we started to explore the opportunities
to prepare the desired syn-1,2-amino alcohol 10. Studies
showed that a fast epimerization of ketone 9 could beeasily
achieved with a weak base such as Et3N. However, DKR
hydrogenation of racemic 9 in the presence of Noyori’s
Ru-BINAP catalyst, which, to our knowledge,7 is the only
reported method for the asymmetric preparation of syn-1,2-
amino alcohol analogues of 10 through a reduction of
R-amino-β-keto esters at the time,11 resulted in an extremely
slow reaction with low conversion under various conditions.7
In parallel, we also investigated the preparation of 10 via
an enzymatic DKR reduction. If the desired enzymatic
reactivity converting 9 to 10 (even with low conversion)
could be realized through screening, we were confident
that we could evolve/develop the initial proof-of-concept
result to a practical process through enzyme evolution, given
the recent development and success on enzyme engineering
technology.12
Further, a retro-synthetic disconnection of ketone 2
leads to a syn 1,2-amino alcohol aldehyde 3 and phospho-
nate 4.6 In comparison with the previous synthesis,1 the
application of a HornerꢀWadsworthꢀEmmons reaction
can offer a significant convergence to access pyrrolidine 1.
However, a literaturesurveyshowed thatmethodologiesto
establish a syn stereochemistry relationship of 1,2-amino
alcohols in an open-chain system, such as 2 (R = H), have
been lacking, while the corresponding undesired anti 1,2-
amino alcohols can be easily prepared enantioselectively.7
Several approaches to prepare precursor 3 were explored.
The use of amino diol 6, which has the desired C10 and C2
stereochemistry, is not feasible for large-scale preparation8,9
because 6 is not commercially readily available in large
(6) Maloney, K. M.; Chung, J. Y. L. J. Org. Chem. 2009, 74, 7574–
7576.
(10) (a) Farran, D.; Parrot, I.; Toupet, L.; Martinez, J.; Dewynter, G.
Org. Biomol. Chem. 2008, 6, 3989–3996. (b) Hara, O.; Ito, M.; Hamada,
Y. Tetrahedron Lett. 1998, 39, 5537–5540.
(11) Noyori, R.; Ikeda, T.; Ohkuma, T.; Widhalm, M.; Kitamura,
M.; Takaya, T.; Akutagawa, S.; Sayo, N.; Saito, T. J. Am. Chem. Soc.
1989, 111, 9134–9135.
(7) (a) Liu, Z.; Schultz, C. S.; Sherwood, C. A.; Krska, S.; Dormer,
P. G.; Desmond, R.; Lee, C.; Cherer, E. C.; Shpungin, J.; Cuff, J.; Xu, F.
Tetrahedron Lett. 2011, 52, 1685–1688. (b) Seashore-Ludlow, B.; Villo,
€
P.; Hacker, C.; Somfai, P. Org. Lett. 2010, 12, 5274–5277and references
cited therein.
(8) To support the early development work, a process to prepare
(12) For an excellent example, see: Savile, K. C.; Janey, J. M.;
Mundorff, E. C.; Moore, J. C.; Tam, S.; Jarvis, W. R.; Colbeck, J. C.;
Krebber, A.; Fleitz, F. J.; Brands, J.; Devine, P. N.; Huisman, G. W.;
Hughes, G. J. Sciences 2010, 329, 305–309.
aldehyde 12 from 6 was also developed.9
(9) For more detailed discussion, see the Supporting Information.
B
Org. Lett., Vol. XX, No. XX, XXXX

formed during the reaction and to maintain the reaction
pH at 7.5 to accomplish a fast epimerization of 9, thereby
achieving >98% conversion within 2 days. It is also worth
noting that the use of i-PrOHꢀt-BuOMe for extractive
workup resulted in a clear phase separation that overcame
the potential emulsion caused by enzyme protein. As such,
10 was obtained in 92% assay yield with >99% ee and
>99:1 dr selectivity. Thus, the syn stereochemistry of 10,
which is generally not readily prepared through a chemical
reduction, was established through a powerful enzymatic
DKR reduction of racemic substrate 9. The integration of
biotechnology into classical organic synthesis is an alter-
native and useful tool to solve synthetic problems.
Scheme 3. Through-Process to Enone 15 via Enzymatic DKR
To convert syn-amino alcohol 10 to aldehyde 13, which
thereby allowed us to extend the desired functionality for the
preparation of 1 through a HornerꢀWadsworthꢀEmmons
reaction (Scheme 1), acetonide 11 was prepared in the pre-
sence of 10 mol % of BF3 OEt2 and dimethoxypropane in
3
toluene/acetone (Scheme 2). The use of a small amount of
toluene streamlined the process, such that the conversion
from 9 to 14 was performed without isolating any intermedi-
ates. Reduction of ester 11 with either 1.3 equiv of LiBH4 or
0.7 equiv of LiAlH4 gave 12 in 95% assay yield. Subsequent
TEMPO/bleach oxidation was carried out in a biphasic mix-
ture of aqueous MeCN/toluene in the presence of NaHCO3
and 15 mol % of KBr at <10 °C to afford aldehyde 13
without any ee loss in 88% assay yield. The aqueous phase
was discarded, and the wet crude reaction stream of alde-
hyde 13 was directly used “as is” in the subsequent step.
Appropriately mild conditions14 were identified to pro-
mote the desired HornerꢀWadsworthꢀEmmons reaction.
In the presence of 3 equiv of LiBr and 3 equiv of Hunig’s
base, the coupling of phosphonate 56 and aldehyde 12 pro-
ceeded smoothly in MeCN to give 14 in 89% assay yield.
The use of either LiCl or <3 equiv of LiBr led to a lower
conversion and yield. Interestingly, byproduct 15, which
competitively formed up to ∼12% through a reversible
aldol condensation of 14 and aldehyde 1315 within initial
several hours, was eventually converted to the desired 14
in >99% conversion16 upon aging the reaction at 20 °C
overnight (Scheme 3). After aqueous workup, enone
14 was crystallized from aqueous i-PrOH in 80% yield
and >98% purity. Thus, starting from keto ester 9, enone
14 was prepared in 60% yield over five steps without
isolating any intermediates.
To this end, screening of enzymatic DKR reduction
identified that several enzymes promoted the desired reaction
to produce 10. Preliminary results showed that methyl ester 9
gave higher dr selectivity than the corresponding ethyl ester.
The attainment of the catalytic cycle for the desired
enzymatic DKR reduction required the use of coenzyme
NADP and CDX-901, both of which, by consuming glu-
cose to gluconic acid, deliver the hydride source to selec-
tively reduce ent-(S)-9 to give the desired 10. Enzymes13
KRED-130 and CDX-901 were initially selected for fur-
ther optimization. Although promising results were ob-
tained on a laboratory scale to give 10 in 99% ee and 90%
assay yield, the stability and reactivity of the enzymes on
large-scale runs were not ideal and resulted in incomplete
conversion. However, this issue was overcome by employ-
ing the evolved enzymes GDH-105 and CDX-018,13 which
were also evolved to withstand higher concentration of
organic solvent desired for improving the solubility of 9
and therefore the volume productivity. After optimization,
the highly enantioselective enzymatic DKR reduction was
carried out by treating 9 in an aqueous DMSO phosphate
buffer in the presence of glucose, a catalytic amount of
NADP, and enzymes GDH-105 and CDX-018 at 30 °C.
NaOH (2 N) was added to neutralize the gluconic acid
Hydrogenation of 14 (5 wt % of 10% PdꢀC, 20 psi H2,
THF, 20 °C, 2 h) gave ketone 16 cleanly (Scheme 4). The
use of THF avoided hydrogenolysis cleavage of the
benzylic CꢀO bond. With compound 16 in hand, we set
forward for global deprotection. However, initial
(14) (a) Blanchette, M. A.; Choy, W.; Davis, J. T.; Essenfeld, A. P.;
Masamune, S.; Roush, W. R.; Sakai, T. Tetrahedron Lett. 1984, 25,
2183–2186. (b) Rathke, M. W.; Nowak, M. J. Org. Chem. 1985, 50,
2624–2626.
(15) Since HornerꢀWadsworthꢀEmmons coupling is irreversible,
13 liberated via a retro-aldo reaction can be fully consumed to yield 14.
(16) For examples, see: (a) Lygo, B.; Beynon, C.; McLeod, M. C.;
Roy, C. E.; Wade, C. E. Tetrahedron 2010, 66, 8832–8836. (b) Ahn,
K. H.; Kwon, G. J.; Choi, W.; Lee, S. J.; Yoo, D. J. Bull. Korean Chem.
Soc. 1994, 15, 825–827.
(13) In collaboration with Codexis Inc., USA.
Org. Lett., Vol. XX, No. XX, XXXX
C

Table 1. Selected Initial Results of Hydrogenation of Iminea
Scheme 4. Deprotection and Hydrogenation
entry substrates catalysts solvents convb (%) cis/transb
1
17
Pd/C
i-PrOH
EtOH
64
93
60:40
56:44
61:39
84:16
66:33
84:16
96:4c
96:4
2
Pd/C
3
Rh/Al2O3
Ru/C
MeOH
i-PrOH
EtOH
94
4
61
5
PtO2
69
6
Pt/Al2O3
Raney Nic MeOH
i-PrOH
>99
>99
>99
72
7
8
18
Pt/Al2O3
Rh/Al
Pd/C
THF
THF
THF
THF
9
98:2
10
11
78
91:9
Pd/Al
37
95:5
a Unless otherwise mentioned, all reactions were carried out at 25 °C,
15ꢀ40 psi H2 with 10ꢀ25 wt % of catalyst loading. b Determined by
HPLC analysis.9 c 75 °C, 40 psi H2, 100 wt % of Raney Ni.
TMS-protected imine 18 quantitatively, which set a new
stage for screening hydrogenation conditions in an at-
tempt to improve the cis selectivity. Interestingly, under
the basic conditions, imine 17 was readily oxidized upon
exposure to air to give ketone 21; notwithstanding, main-
taining an inert atmosphere was straightforward on the
plant scale. Finally, several suitable catalysts/conditions
were identified to prepare 1 in high diastereoselectivity
(Table 1, entry 8). Given overall considerations in terms
of reaction rate, conversion, and impurity profile, Pt/Al2O3
was selected for further optimization. In practice, the in situ
TMS-silylated imine 18 was directly subjected to hydro-
genation in the presence of 5 wt % of 5% Pt/Al2O3 at
20ꢀ25 °C/40 psi H2 for 12 h to give the corresponding
amine 19 (cis/trans = 96:4). After aqueous HCl workup,
the desired cis-pyrrolidine 1 hemihydrate was crystallized
from aqueous i-PrOH in 80% yield and >99% purity. The
undesired trans diastereomer was cleared to <0.2%.
In summary, an efficient asymmetric synthesis of cis-2,5-
disubstituted pyrrolidine 1 has been developed. This prac-
tical synthesis features the application of enzymatic DKR
reduction to establish the two stereogenic centers of syn
phenyl 1,2-amino alcohol in >99% ee and >99:1 dr in one
step. Effective hydrogenation of chiral cyclic imine affords
pyrrolidine 1 in high cis diastereoselectivity. Starting from
inexpensive glycine ester and only isolating 3 intermedi-
ates, the core scaffold of β3-AR agonists is prepared in
38% yield over 11 steps. This synthesis is amenable to the
preparation of various analogues of 1.
treatment of 16 with TFA resulted in the desired imine 17
along with 20ꢀ25% of cyclic carbamate byproduct 20.
Further studies showed that complete global deprotection
followed by spontaneous dehydrationꢀcyclization was
best carried out in i-PrOHꢀHCl. After 24 h at ambient
temperature, imine 17 bis-HCl salt monohydrate was
directly isolated from i-PrOHꢀi-PrOAc in 95% yield and
>97% purity.
To complete the construction of the core skeleton de-
sired for a novel class of β3 adrenergic receptor agonists,
diastereoselective reduction of imine 17 was highly desired.
Preliminary results showed the reduction applying hydride
methods gave unsatisfactory selectivity; we therefore focused
our efforts on hydrogenation of 17. Among the various
catalysts/conditions9 examined, Raney Ni gave 96:4 dr se-
lectivity (cis:trans) in MeOH at 75 °C/40 psi H2 (Table 1,
entry 7). However, 100 wt % of Raney Ni was required in
order to achieve >95% conversion in 20 h. In addition, the
use of aqueous MeOH instead of dry MeOH resulted in
lower dr selectivity.9
Acknowledgment. We gratefully acknowledge R.
Reamer, P. Dormer, and L. DiMichele (Merck & Co.,
Inc.) for NMR assistance.
A survey of the literature1,16 suggested that increasing
the bulkiness of the substitute group of the C2 stereogenic
center in the pyrroline ring, which can be achieved by mask-
ing the OH group in 18, could improve the cis-selectivity.
Thus, bis-HCl salt monohydrate 17 was treated with
2.1 equiv of (Me3Si)2NH in THF to give the corresponding
Supporting Information Available. Experimental pro-
cedure/data and discussion including an alternative
synthesis of 13. This material is available free of charge
via the Internet at http://pubs.acs.org.
The authors declare no competing financial interest.
D
Org. Lett., Vol. XX, No. XX, XXXX