Synthesis of the GPR40 partial agonist MK-8666 through a kinetically controlled dynamic enzymatic ketone reduction

Article abstract of DOI:10.1021/acs.orglett.6b02910

A scalable and efficient synthesis of the GPR40 agonist MK-8666 was developed from a simple pyridine building block. The key step to set the stereochemistry at two centers relied on an enzymatic dynamic kinetic reduction of an unactivated ketone. Directed

Full text of DOI:10.1021/acs.orglett.6b02910

Letter
pubs.acs.org/OrgLett
Synthesis of the GPR40 Partial Agonist MK-8666 through a
Kinetically Controlled Dynamic Enzymatic Ketone Reduction
Alan M. Hyde,*,† Zhijian Liu,*,† Birgit Kosjek,† Lushi Tan,† Artis Klapars,† Eric R. Ashley,†
Yong-Li Zhong,† Oscar Alvizo,‡ Nicholas J. Agard,‡ Guiquan Liu,§ Xiuyan Gu,§ Nobuyoshi Yasuda,†
John Limanto,† Mark A. Huffman,† and David M. Tschaen†
†Process R&D Department, MRL, Merck & Co., Inc., Rahway, New Jersey 07065, United States
‡Codexis Inc., Redwood City, California 94063, United States
§Shanghai SynTheAll Pharmaceutical Co., Ltd., Jinshan District, Shanghai 201507, China
S
* Supporting Information
ABSTRACT: A scalable and efficient synthesis of the GPR40 agonist MK-8666 was developed from a simple pyridine building
block. The key step to set the stereochemistry at two centers relied on an enzymatic dynamic kinetic reduction of an unactivated
ketone. Directed evolution was leveraged to generate an optimized ketoreductase that provided the desired trans alcohol in >30:1
dr and >99% ee. Further, it was demonstrated that all four diastereomers of this hydroxy-ester could be prepared in high yield
and selectivity. Subsequently, a challenging intramolecular displacement was carried out to form the cyclopropane ring system
with perfect control of endo/exo selectivity. The endgame coupling strategy relied on a Pd-catalyzed C−O coupling to join the
headpiece chloropyridine with the benzylic alcohol tailpiece.
ndividuals afflicted with type II diabetes have long benefited
Scheme 1. Discovery Route to the Tricyclic Headpiece
I
from pharmaceutical intervention targeting insulin and/or
glucose modulation. However, drug-induced hypoglycemia
remains an ever-present risk to patients. The overarching
theme of next-generation therapies is to provide feedback
control that would prevent large disruptions to blood sugar
homeostasis. A relatively new class of druggable targets being
investigated to provide more stable glucose levels are free fatty
acid receptors1 (FFAs), with FFA12 (GPR40) garnering the
most attention. In fact, a small molecule therapeutic was recently
advanced to phase III clinical trials based on this mechanism.3
Given the aforementioned benefits of this modality coupled with
our strong commitment to the diabetes field, MK-8666 was
developed as a GPR40 partial agonist.4
Given the difficulties encountered building the tricyclic ring
system, subsequent efforts were spent on developing a practical
and efficient approach for the production of clinical supply. We
envisioned building the tricyclic ring system through an
intramolecular enolate alkylation (Retrosynthesis in Scheme
2). This approach would require a trans alcohol or halide that
ideally would be generated from a dynamic kinetic resolution.
MK-8666 features an unusual pyridine-fused bicyclo[3.1.0]-
hexane with three contiguous stereocenters. Retrosynthetically,
the molecule can be roughly divided in half to give a biaryl
portion referred to as the tailpiece and the tricyclic pyridine
portion referred to as the headpiece (see abstract graphic).
Synthesis of the tailpiece is described elsewhere5 and will not be
the focus of this discussion. Scheme 1 depicts the early route to
the headpiece4a that commences with construction of indene 2 as
a mixture of regioisomers from 5-bromo-2- methoxypyridine (1).
The crux of this route was the cyclopropanation of 2 with ethyl
diazoacetate that was poorly selective and gave 8 isomers.
Although some aspects of this sequence were improved upon, it
was deemed inadequate for further development and scale up.
Scheme 2. Retrosynthesis of the Tricyclic Headpiece
Received: October 3, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/acs.orglett.6b02910
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
We anticipated building the ketone from a triester and a
difunctionalized-5-halo-4-picoline.
Table 1. Ketone Reduction Optimization, Enzyme Evolution
To investigate this synthetic strategy, we first sought to
establish an efficient synthesis of keto-ester 10 (Scheme 3). First,
Scheme 3. Synthesis of Ketone 10
entry
catalyst
% conv trans/cis % 11a % 11b % 11c % 11
a
a
1
2
3
4
5
KRED P1C1
95
95
<0.1:1
<0.1:1
0.2:1
97
95
0
1
0
2
0
0
5
0
0
0
KRED P1H4
a
KRED 131
100
100
100
82
100
97
18
0
a
KRED 136
<0.1:1
<0.1:1
0
(R,R)-RuCl-
3
0
b
TsDpen
a
6
7
KRED 124
95
93
7.3:1
0
1
12
1
88
94
0
4
KRED
>50:1
a
exp-B1T
a
8
KRED 208
97
93
2.4:1
30
2
0
0
0
0
70
98
c
9
KRED 264
>30:1
optimized
enzyme
lithiation of 2-chloro-5-bromopyridine and quenching with
DMF provided a 4-formylated pyridine that was used without
isolation. Reduction with NaBH4 gave alcohol 4, which was then
chlorinated with SOCl2 to provide the chloromethylpyridine 5.
Alkylation of commercially available triester 6 with 5 was
accomplished using NaOMe in DMAc to give 7. The conversion
of 7 to 10 was first demonstrated by decarboxylation with AcOH-
buffered LiCl in NMP6 to generate diester 8 in 90% yield
followed by lithiation/ring closure to provide ketone 10 in 67%
yield. However, a higher yield of the lithiation/cyclization step
was achieved when the sequence was reversed. We attribute this
to triester 7 having one fewer enolizable center and being more
electrophilic than diester 8. In the event, lithium−bromide
exchange and cyclization of compound 7 at −90 °C gave keto-
diester 9 in 93% yield. Subsequent decarboxylation under similar
reaction conditions proceeded smoothly to provide 10 in 87%
yield.
a
2 g/L 10, 100 wt % KRED, 1 g/L NADP, 0.1 M phosphate buffer pH
b
8.0, 30 °C. 0.075 M in ethyl acetate, 15 mol % catalyst, 5 equiv of
c
formic acid, 2 equiv of DABCO. 50 g/L 10, 2 wt % KRED, 0.1 g/L
NADP, 0.1 M phosphate buffer pH 9.0, 50 °C, active nitrogen sweep.
revealed that the cis diastereomer 11a was favored early (0.8:1
trans/cis at 17% conversion), but as the reaction proceeded, the
trans isomer began forming preferentially, ultimately reaching a
final ratio of ∼2.4:1 trans/cis (Figure 1). Additional insight came
At the outset, our strategy to access trans alcohol 11 rested on
the premise that a dynamic asymmetric reduction of keto-ester
10 would set both adjacent stereocenters. We chose to focus on
biocatalysis technology7 based on its proven reliability coupled
with the diversity of commercially available ketoreductases
(KREDs). Moreover, several enzymatic examples have been
published, albeit mostly with α-keto esters and other activated
systems having pKa’s ranging from 7 to 15.8 Even though ketone
10 was calculated to have a pKa of 18.5 for the α-hydrogen (see
Supporting Information (SI) pp S39−S40 for details), we were
optimistic that epimerization through an enol pathway could take
place at near-neutral pH.9
Initial screening of our KRED library under standard aqueous
conditions at pH 7.0−8.0 revealed enzymes capable of providing
three of the four possible diastereomers with high selectivity
(Table 1). This, coupled with high conversions and rates,
demonstrated the feasibility of the proposed dynamic kinetic
reduction. Most enzymes favored the formation of the cis
diastereomer (Table 1, entries 1−4). Similarly, the application of
ruthenium-catalyzed transfer hydrogenation10 provided only the
cis isomer 11b in 94% ee (Table 1, entry 5). Unfortunately, only a
small subset of enzymes were found that were selective in
forming trans alcohols (Table 1, entries 6−8). The highest
selectivity for the desired isomer 11 was obtained using KRED-
208, achieving >99% ee, but a modest 2.4:1 trans/cis ratio (Table
1, entry 8).
Figure 1. Kinetic profiling of KRED-208 activity.
from monitoring the ee of starting material which remained
racemic throughout the reaction, indicating epimerization was
fast with respect to ketone reduction.
A free energy diagram in Figure 2 illustrates how these
observations can be rationalized. The initial slight intrinsic bias
for the cis product can be traced to the difference in barrier
‡
heights on the red curve (ΔG1 ). As the reaction proceeds,
acetone builds up through 2-propanol oxidation and the
equilibrium between alcohol products and ketone will become
less favorable (reflected by ΔG2) that results in a thermodynami-
cally determined cis/trans ratio (ΔG3). This is supported by
calculations that predict the trans isomer to be between 0.4 and
1.4 kcal/mol more stable than the cis isomer (see SI p S37 for
details). Entry to the blue curve is kinetically prevented, reflected
With the aim to further improve selectivity, an investigation
into the reaction kinetics of KRED-208 was undertaken. This
B
DOI: 10.1021/acs.orglett.6b02910
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
(Table 2, entry 6). Notably, the undesired endo-12 product was
never observed.18
With the headpiece in hand, installation of the tailpiece alcohol
through SNAr chemistry was nontrivial, failing with a wide variety
of bases and solvents examined. In some cases, dimeric or ring-
opened fragmentation products were identified. We attributed
this outcome to the sensitive nature of the strained ring system
with an acidic methylene position adjacent to the cyclopropane.
Ultimately, it was found that chloropyridine 12 and alcohol 14
could be coupled to generate 15 in high yield (>90%) with a
combination of Pd(OAc)2, BrettPhos, and Cs2CO3 in toluene
(Scheme 4).19 The rate of this reaction was sensitive to the
Scheme 4. C−O Coupling, Hydrolysis, and Salt Formation
Figure 2. Free energy diagram of reduction with KRED-208.
‡
by a large ΔG4 , because the enzyme active site is completely
selective for hydride transfer to the Si-face of the ketone.
It became apparent that further progress toward developing a
highly selective reaction would require two advances: (1) An
enzyme with a significant intrinsic bias to generate the trans
isomer and (2) preventing reversibility. The first would require
an evolved enzyme11 that provides improved diastereoselectivity.
Reversibility could be addressed by removal of acetone with a
nitrogen sweep.12,13 In partnership with Codexis Inc., six rounds
of enzyme evolution delivered KRED-264, a novel variant able to
produce 11 with >30:1 dr when a nitrogen sweep is utilized
(Table 1, entry 9). This variant is also more active, requiring only
2 wt % loading with respect to ketone 10. The optimized enzyme
also exhibits high selectivity from the beginning and maintains it
throughout the reaction course. With this development, we
demonstrated the ability to access all four diastereomers14 of
11.15
20
source of Cs2CO3 which was problematic in light of the
potential for decomposition. To address this, the base was milled
to consistently achieve high reactivity. Cleavage of the t-Bu ester
in 15 was accomplished with gentle heating in the presence of
NaOH that avoided decomposition observed with acid-
promoted conditions. Finally, the tris salt was isolated by
crystallization from EtOAc/EtOH/water.
In summary, we have developed an efficient sequence to MK-
8666 that utilized a novel enzymatic DKR reduction to set the
absolute configuration and requisite trans configuration of a key
intermediate. Mechanistic understanding of the kinetic and
thermodynamic factors helped guide our development. The
cyclopropane-forming step was only possible by carefully
balancing the reactivity of the enolate/electrophile pair. The
penultimate step made use of a Pd-catalyzed C−O coupling in
which milling of the Cs2CO3 was deemed critical to obtain
reproducible and robust results.
Having a robust methodology in place for the preparation of
11, we sought to affect the proposed cyclization (Table 2).16 The
a
Table 2. Optimization of Enolate Cyclization
ASSOCIATED CONTENT
* Supporting Information
■
S
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.orglett.6b02910.
entry
activating group
base
Et3N
solvent
yield of 12
b
1
2
3
4
5
6
TsCl
THF
THF
THF
THF
THF
0
Experimental procedures and characterization of new
b
1
compounds, including their H and 13C NMR spectra;
TsCl
LiHMDS
LiHMDS
LiHMDS
LiHMDS
LiHMDS
5
c
(Me2N)2POCl
(PhO)2POCl
(EtO)2POCl
(EtO)2POCl
30
details of DFT calculations with xyz coordinates and
energies, for all stationary points (PDF)
60
70
94
MTBE
b
AUTHOR INFORMATION
Corresponding Authors
■
a
All reactions were run for 16−24 h at −10 °C. Significant quantities
of lactone 13 were generated. Degradation of 12 was observed.
c
*E-mail: alan_hyde@merck.com.
*E-mail: zhijian_liu@merck.com.
nature of the activating group had the greatest influence on the
outcome of this transformation. For example, the sulfonate
derived from TsCl in combination with a base primarily gave rise
to lactone 13 (Table 2, entries 1, 2) arising from O-alkylation of
the enolate with the strong leaving group. The use of a weaker
leaving group such as a phosphate ester or phosphorodiamidate
was critical to obtain the desired product (Table 2, entries 3−
6).17 MTBE was a superior solvent for this transformation, and
the desired cyclopropanation product was isolated in a 94% yield
when (EtO)2POCl was used in conjunction with LiHMDS
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Jon Wilson, Steve Miller, and Kevin Belyk for catalyst
screening; John Chung and Ed Sherer for calculating the
energetics of endo vs exo ring closure; Bryon Simmons for
contributing to finding conditions for the cyclopropane ring-
closure; Jake Song for suggesting the use of phosphate ester
C
DOI: 10.1021/acs.orglett.6b02910
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
2009, 351, 253. For a two-step synthesis of four diastereomers, see:
leaving groups; Shane Grosser and Zachary Dance for particle
characterization of the Cs2CO3; Bob Reamer for providing NMR
analytical assistance; and Wes Schafer and Becky Arvary for
providing other analytical assistance (Merck & Co., Inc., Rahway,
New Jersey, U.S.A.). We also thank Jin She, Yuhua Shi, Weifeng
Hu, Yunchun Dai, Honglin Ye, and Xianliang He for additional
experimental assistance (Shanghai SynTheAll Pharmaceutical
Co., Ltd.).
(b) Wolberg, M.; Hummel, W.; Muller, M. Chem. - Eur. J. 2001, 7, 4562.
̈
(15) For a review discussing enantiocomplementary enzymes, see:
Mugford, P. F.; Wagner, U. G.; Jiang, Y.; Faber, K.; Kazlauskas, R. J.
Angew. Chem., Int. Ed. 2008, 47, 8782.
(16) For a similar cyclization, see: Tan, L.; Yasuda, N.; Yoshikawa, N.;
Hartner, F. W.; Eng, K. K.; Leonard, W. R.; Tsay, F.-R.; Volante, R. P.;
Tillyer, R. D. J. Org. Chem. 2005, 70, 8027.
(17) Song, Z. J.; Zhao, M.; Frey, L.; Li, J.; Tan, L.; Chen, C. Y.;
Tschaen, D. M.; Tillyer, R.; Grabowski, E. J. J.; Volante, R.; Reider, P. J.;
Kato, Y.; Okada, S.; Nemoto, T.; Sato, H.; Akao, A.; Mase, T. Org. Lett.
2001, 3, 3357.
(18) This was rationalized with the aid of DFT calculations which
shows a substantially higher kinetic barrier (4−5 kcal/mol) to form
endo-12 than 12. See SI for transition state geometries and energies.
(19) (a) Wu, X.; Fors, B.; Buchwald, S. L. Angew. Chem., Int. Ed. 2011,
50, 9943. (b) Kataoka, N.; Shelby, Q.; Stambuli, J. P.; Hartwig, J. F. J.
Org. Chem. 2002, 67, 5553. (c) Maligres, P. E.; Li, J.; Krska, S. W.;
Schreier, J. D.; Raheem, I. T. Angew. Chem., Int. Ed. 2012, 51, 9071.
(20) For exmples in which Cs2CO3 particle size is critical to cross-
coupling reaction performance, see: (a) Betti, M.; Genesio, E.; Marconi,
G.; Coccone, S. S.; Wiedenau, P. Org. Process Res. Dev. 2014, 18, 699.
REFERENCES
■
(1) For a review on FFA targets, see: Ichimura, A.; Hasegawa, S.;
Kasubuchi, M.; Kimura, I. Front. Pharmacol. 2014, 5, 1.
(2) (a) Mancini, A. D.; Poitout, V. Trends Endocrinol. Metab. 2013, 24,
398. (b) Defossa, E.; Wagner, M. Bioorg. Med. Chem. Lett. 2014, 24,
2991.
(3) (a) Srivastava, A.; Yano, J.; Hirozane, Y.; Kefala, G.; Gruswitz, F.;
Snell, G.; Lane, W.; Ivetac, A.; Aertgeerts, K.; Nguyen, J.; Jennings, A.;
Okada, K. Nature 2014, 513, 124. (b) Kaku, K.; Enya, K.; Nakaya, R.;
Ohira, T.; Matsuno, R. Diabetes, Obes. Metab. 2015, 17, 675.
(4) For the Med-Chem patent on GPR40 preclinical candidates
including MK-8666, see: (a) Hagmann, W. K.; Nargund, R. P.; Blizzard,
T. A.; Josien, H.; Biju, P.; Plummer, C. W.; Dang, Q.; Li, B.; Lin, L. S.;
Cui, M., Antidiabetic tricyclic compounds. WO2014022528A1,
February 6, 2014. For previous GPR40 preclinical candidates developed
at Merck & Co., Inc., Kenilworth, NJ USA, see: (b) Tan, C. P.; Feng, Y.;
Zhou, Y.-P.; Eiermann, G. J.; Petrov, A.; Zhou, C.; Lin, S.; Salituro, G.;
Meinke, P.; Mosley, R.; Akiyama, T. E.; Einstein, M.; Kumar, S.; Berger,
J. P.; Mills, S. G.; Thornberry, N. A.; Yang, L.; Howard, A. D. Diabetes
2008, 57, 2211. (c) Zhou, C.; Tang, C.; Chang, E.; Ge, M.; Lin, S.; Cline,
E.; Tan, C. P.; Feng, Y.; Zhou, Y.-P.; Eiermann, G. J.; Petrov, A.; Salituro,
G.; Meinke, P.; Mosley, R.; Akiyama, T. E.; Einstein, M.; Kumar, S.;
Berger, J.; Howard, A. D.; Thornberry, N.; Mills, S. G.; Yang, L. Bioorg.
Med. Chem. Lett. 2010, 20, 1298.
(b) Meyers, C.; Maes, B. U. W.; Loones, K. T. J.; Bal, G.; Lemiere, G. L.
̀
F.; Dommisse, R. A. J. Org. Chem. 2004, 69, 6010.
(5) Goto, M.; Kajiwara, T.; Kondo, Y.; Konishi, T.; Maeda, H.; Sera,
M.; Yamano, M.; Yamasaki, S. Production Method of Optically Active
Dihydrobenzofuran Derivatives U.S. Patent 8,952,185, August 23, 2012.
(6) The use of AcOH was not required, but it was used to buffer
chloride, thereby attenuating its reactivity to prevent unsafe exotherms.
(7) For reviews on biocatalytic reductions, see: (a) Moore, J. C.;
Pollard, D. J.; Kosjek, B.; Devine, P. N. Acc. Chem. Res. 2007, 40, 1412.
(b) Groger, H.; Hummel, W.; Metzner, R. In Comprehensive Chirality;
̈
Carreira, E. M., Yamamoto, H., Eds.; 2012, 7, 181−215. (c) Hall, M.;
Bommarius, A. S. Chem. Rev. 2011, 111, 4088.
(8) For a general review on enzymatics DKR reactions, see:
(a) Applegate, G. A.; Berkowitz, D. B. Adv. Synth. Catal. 2015, 357,
1619. For examples of biocatalytic DKR reductions of diketones and
ketoesters, see: (b) Kalaitzakis, D.; Smonou, I. J. Org. Chem. 2010, 75,
8658. (c) Kalaitzakis, D.; Rozzell, J. D.; Kambourakis, S.; Smonou, I. Org.
Lett. 2005, 7, 4799. (d) Zhu, D.; Mukherjee, C.; Rozzell, J. D.;
Kambourakis, S.; Hua, L. Tetrahedron 2006, 62, 901. (e) Kosjek, B.;
Tellers, D. M.; Biba, M.; Farr, R.; Moore, J. C. Tetrahedron: Asymmetry
2006, 17, 2798.
(9) A recent publication has described a DKR reduction of an
unactivated ketone at near-neutral pH: Hanson, R. L.; Guo, Z.;
́
Gonzalez-Bobes, F.; Fenster, M. D. B.; Goswami, A. J. Mol. Catal. B:
Enzym. 2016, 133, 20.
(10) Akashi, M.; Arai, N.; Inoue, T.; Ohkuma, T. Adv. Synth. Catal.
2011, 353, 1955.
(11) Bornscheuer, U. T.; Huisman, G. W.; Kazlauskas, R. J.; Lutz, S.;
Moore, J. C.; Robins, K. Nature 2012, 485, 185.
(12) Calvin, S. J.; Mangan, D.; Miskelly, I.; Moody, T. S.; Stevenson, P.
J. Org. Process Res. Dev. 2012, 16, 82.
(13) For a study on reversibility with various ketones, see: Tewari, Y.
B.; Phinney, K. W.; Liebman, J. F. J. Chem. Thermodyn. 2006, 38, 388.
(14) To the best of our knowledge, this is the first example of the feat.
For an example with three out of four diastereomers using three different
KREDs, see: (a) Ludeke, S.; Richter, M.; Muller, M. Adv. Synth. Catal.
̈
̈
D
DOI: 10.1021/acs.orglett.6b02910
Org. Lett. XXXX, XXX, XXX−XXX