Asymmetric synthesis of highly functionalized tetrahydropyran DPP-4 inhibitor

Article abstract of DOI:10.1021/ol502661g

A practical synthesis of a highly functionalized tetrahydropyran DPP-4 inhibitor is described. The asymmetric synthesis relies on three back-to-back Ru-catalyzed reactions. A Ru-catalyzed dynamic kinetic resolution (DKR) reduction establishes two contiguous stereogenic centers in one operation. A unique dihydropyran ring is efficiently constructed through a preferred Ru-catalyzed cycloisomerization. Hydroboration followed by a Ru-catalyzed oxidation affords the desired functionalized pyranone core scaffold. Finally, stereoselective reductive amination and subsequent acidic deprotection afford the desired, potent DPP-4 inhibitor in 25% overall yield. (Chemical Equation Presented).

Full text of DOI:10.1021/ol502661g

Letter
pubs.acs.org/OrgLett
Asymmetric Synthesis of Highly Functionalized Tetrahydropyran
DPP‑4 Inhibitor
Feng Xu,*,† Michael J. Zacuto,*,† Yoshinori Kohmura,† Jon Rosen,† Andrew Gibb,‡ Mahbub Alam,‡
Jeremy Scott,‡ and David Tschaen†
†Process Chemistry, Merck Research Laboratories, Rahway, New Jersey 07065, United States
‡Process Research Preparative Laboratory, Merck Sharp & Dohme Research Laboratories, Hoddesdon, Hertfordshire, EN11 9BU,
U.K.
S
* Supporting Information
ABSTRACT: A practical synthesis of a highly functionalized tetrahy-
dropyran DPP-4 inhibitor is described. The asymmetric synthesis relies
on three back-to-back Ru-catalyzed reactions. A Ru-catalyzed dynamic
kinetic resolution (DKR) reduction establishes two contiguous
stereogenic centers in one operation. A unique dihydropyran ring is
efficiently constructed through a preferred Ru-catalyzed cycloisomeriza-
tion. Hydroboration followed by a Ru-catalyzed oxidation affords the
desired functionalized pyranone core scaffold. Finally, stereoselective reductive amination and subsequent acidic deprotection
afford the desired, potent DPP-4 inhibitor in 25% overall yield.
Scheme 1. Racemic Synthesis of 1
ype 2 diabetes mellitus is a growing worldwide epidemic
1
Taffecting more than 347 million people.
The clinical
application of dipeptidyl peptidase-4 (DPP-4) inhibitors has
recently proved to be an effective new therapy for the treatment
of type 2 diabetes.2 Due to the clinical success of DPP-4
inhibitors, interest in this area has grown. As a result of the efforts
to discover structurally diversified potent drug candidates with
additional benefits over current DPP-4 inhibitors, Merck
laboratories recently discovered highly functionalized tetrahy-
dropyran 1, which represents a new class of structurally
differentiated DPP-4 inhibitors.3 Tetrahydropyran 1 possesses
a unique core scaffold required for achieving the desired
selectivity and efficacy in the tested diabetes model, but it raises
the chemical complexity of accessing the evolved new generation
of DPP-4 inhibitor drug candidates.
To support the drug development program, an efficient
synthesis of 1 suitable for large scale preparation was required.
The main synthetic challenge in preparing 1 was the effective and
practical construction of three stereogenic centers. In particular,
the unique structure of 1 possesses a contiguous R,S (C2, C3)
stereochemical array with an R (C5) functionalized amino group
in the tetrahydropyran ring. In fact, the central problem of the
initial racemic synthesis3 of 1 essentially was the arduous nature
of establishing the desired relative C2,C3 stereochemistry
(Scheme 1). Also, reductive amination between 6 and 7 suffered
from low diastereoselectivity in establishing the C-5 stereogenic
center. The overall yield of the synthesis was only ∼1.9%.
Although initial results showed that the dr of the reductive
amination was low, the convergent endgame strategy was
logically sound. Based on the stereofacial bias of the reduction
of the corresponding iminium species derived from 6, in
principle, an improvement in C5 (R) selectivity could be
achieved by optimizing proper reduction conditions, including
modification of reduction reagents. Therefore, a straightforward
approach to prepare ketone 8 became the main focus of our
efforts (Scheme 2).
We envisioned 8 to arise from dihydropyran 9 through a
hydroboration4 followed by oxidation, as 9 could be prepared via
a cycloisomerization of 10. In particular, several metal catalyzed
cycloisomerization protocols5 recently developed for the
preparation of 2,3-disubstituted dihydropyrans were promising,
although the competitive cyclization between N vs O selectivity
had not been well studied at the time of our initial research.6
Received: September 8, 2014
Published: September 30, 2014
© 2014 American Chemical Society
5422
dx.doi.org/10.1021/ol502661g | Org. Lett. 2014, 16, 5422−5425

Organic Letters
Letter
one-pot, was treated with excess NaOH (2.5 equiv) followed by
(Boc)2O in biphasic aqueous MTBE at ambient temperature to
give the desired Boc protected acid 14. The crude 14 was treated
with CDI followed by Weinreb’s amine in DMF to yield the
desired amide 15, which was directly crystallized upon addition
of water to the reaction mixture. This practical through process
afforded 15 in 83% yield over four steps. Grignard reagent 16 was
then prepared through a halogen−metal exchange upon
treatment of 1,4-difluoro-2-bromobenzene with i-PrMgCl or
turbo Grignard (i-PrMgCl/LiCl) in THF or toluene.14 Treat-
ment of amide 15 with 16 gave the desired ketone 17, which was
isolated from heptane in 78% yield.
Scheme 2. Retrosynthetic Analysis of Tetrahydropyran 1
With 17 in hand, we explored opportunities to prepare the
desired anti 1,2-amino alcohol 10 through a DKR reduction. The
facial selectivity of ketone reduction is controlled by a chiral
catalyst while the diastereoselectivity of the process is controlled
by the relative ratio of the epimerization rate vs reduction rate of
the desired enantiomer S-17. Attempts to apply DAIPEN type
ligands and Ru-catalyzed DKR hydrogenation were unsuccess-
ful.9 Noyori’s Ru-transfer hydrogenation system15 gave low
diastereoselectivity initially. However, the major diastereomer
was desired compound 18, and no over-reduction of the alkyne
group was observed. After ligands available at the time were
screened, pentafluorophenyl-DPEN was identified as a promis-
ing lead. Screening of bases showed that carbonates gave poor
selectivity due to a slow epimerization, whereas amine bases such
as DBU, DABCO, and morpholine gave improved results. Of
these bases, DABCO was the best with no inhibition of the Ru
catalyst. In the cases that either the reduction rate was accelerated
faster than the epimerization rate or the epimerization rate was
not competitive with the reduction rate, a higher catalyst loading
or higher reaction temperature led to lower diastereoselectivity.
Slow addition of formic acid did not improve the diastereomeric
ratio. Solvent choice had a profound effect with THF and
CH2Cl2 giving higher yields and diastereomeric ratios. In the
presence of 0.5 mol % of 20 and 3 equiv of DABCO in THF at 35
°C, 9:1 dr and 95% ee were realized with near-quantitative assay
yield. Attempts to upgrade stereochemical purity through direct
crystallization of 18 were unsuccessful. Therefore, the crude
DKR stream was directly used for subsequent cycloisomeriza-
tion.
Furthermore, with the recent progress in dynamic kinetic
resolution (DKR) reduction,7 we envisioned that the two O and
N bearing contiguous stereogenic centers in 10 could be
established in one operation by applying an asymmetric DKR
reduction. Thus, the challenge to achieve an asymmetric
synthesis of 1 was retrosynthetically bridged to a racemic
preparation of amino ketone 11.
A key intermediate of our synthetic approach to amino ketone
11 was Boc propargylglycine 14, which is commercially available,
but expensive. After evaluating the reported preparations,8−10 we
quickly settled on a strategy for the preparation of 14 through
alkylation of glycine benzophenone imine with propargyl
besylate.11 The use of phase transfer catalyst Bu4NBr in the
presence of Cs2CO3 in MTBE was crucial to achieve a
reproducible conversion and reaction rate for this heterogeneous
alkylation. Interestingly, the addition order of the reagents had a
significant impact on yield, as we noticed that the formation of
byproducts could be effectively suppressed when Cs2CO3 was
charged to the reaction mixture last.12 Thus, under this optimal
addition order >99% conversion and >95% yield were obtained.
A through process was then developed to carry the crude
alkylation stream to Weinreb amide 15 (Scheme 3). Upon the
completion of the alkylation, the reaction was quenched with
water directly.13 The crude stream was washed with 1 N HCl to
afford hydrolyzed amine HCl salt 13 in aqueous phase, which, in
Scheme 3. Preparation of Ketone 17 and DKR Reduction
The investigation of the desired cycloisomerization began with
studies on competitive cyclization of a vinylidene species 21
between N vs O selectivity (Scheme 4). To this end, purified
Scheme 4. Metal Catalyzed Cycloisomerization of 18
alcohol 18 was used for initial studies to evaluate several catalyst
systems.5,6,16 It was found that the use of (CO)4WC(OMe)-
Me/Et3N in THF at 40 °C under McDonald’s conditions5f,g led
to exclusive N-cyclization. Surprisingly, a 2-pyrroline carbene 23
with Boc group migration was obtained.9
5423
dx.doi.org/10.1021/ol502661g | Org. Lett. 2014, 16, 5422−5425

Organic Letters
Letter
Trost’s Rh based catalytic system5c led to the desired O-
cycloisomerization exclusively. The use of a preprepared
fluorinated analog of Wilkinson’s catalyst [(3-F-Ph)3P]3RhCl
showed a remarkable improvement of the reaction rate and yield
over the catalyst prepared in situ from [Rh(COD)Cl]2/P(3-F-
Ph)3. Following optimization, a 93% assay yield was obtained by
heating 18 to 80 °C in DMF in the presence of 1.5 mol % of [(3-
F-Ph)3P]3RhCl. Attempts to further reduce the catalyst loading
resulted in incomplete conversion. When these optimized
conditions were applied to the crude mixture of diastereomers
(18:19 = 9:1) produced through the DKR reduction, diminished
performance of the catalyst was observed, characterized by
incomplete conversion of 18 and 19. A successful solution to
overcome this issue was to preserve the effective catalyst
concentration through slow addition of the substrates. As such,
the intermolecular side reactions were suppressed. Under
optimized conditions a solution of a 9:1 mixture (18:19) in
DMF was added over 2−3 h to a solution of 2 mol % Rh catalyst
at 80 °C to achieve >98% conversion.
Scheme 5. Through-Process to Pyranone 27 and Endgame
We further extended our cycloisomeration studies to
inexpensive, commercially available Ru catalysts. We were also
interested in expanding upon the identification of O-selectivity
observed with Rh vinylidenes.5c Our preliminary results showed
that carbamate NH capture of Ru vinylidene was disfavored and
therefore inefficient compared to the desired O-cyclization to
form dihydropyran under Trost’s conditions.6 After examining
several conditions, promising results were obtained with
RuClCp(PPh3)2 (Table 1, entry 6). However, the reaction was
rejection of the undesired product 26 (<0.5%). In addition, the
introduction of Bu3P (20 mol %) during the crystallization
allowed for an effective rejection of residual Rh and Ru to lower
the burden of controlling the level of heavy metals in final
product 1, as residual heavy metals are strictly regulated for active
pharmaceutical ingredients. Thus, starting from ketone 17,
pyranol 25 was isolated in 64% yield and >99% ee over three
steps without isolating any intermediates.
Table 1. Selected Results of Ru-Catalyzed Cycloisomerization
a
of 18
With a practical route to pyranol 25 in place, we turned our
attention to an oxidation to afford pyranone 6. Attempts to
oxidize 25 with catalytic TEMPO under various conditions were
plagued by incomplete conversion; only one of the diastereomers
was oxidized.9 Fortunately, we discovered that Ru-catalyzed
oxidations converted both diastereomers of 25 with equal
efficiency. In the presence of 0.2 mol % RuCl319 and 0.55 equiv of
NaBrO3, the oxidation proceeded smoothly in aqueous HOAc−
MeCN at 0 °C.20 Upon completion of the oxidation, i-PrOH was
added to quench the excess oxidants, because other reducing
reagents such as Na2SO3 and NaS2O3 could cause the reaction
mixture to turn to a gel during the aqueous workup. With the
appropriate ratio of MeCN−water determined, the desired
ketone 6 was crystallized in >98% purity by adding water to the
reaction mixture directly.
To complete the construction of the skeleton of 1, a highly
diastereoselective reductive amination of 6 with 7 was desired,
since the dr with decaborane was low in the initial synthesis.3
Surprisingly, a breakthrough in improving diastereoselectivity
came from an unexpected, significant salt/acid buffer effect
(Table 2). Among the various catalysts/conditions examined, the
reductive amination with NaBH(OAc)3 was best carried out in
the presence of a weak acid such as HOAc.21 In particular, when
MsOH or pTSA salt 7 was neutralized with a tertiary amine base
followed by pH buffering with HOAc, the dr selectivity was
dramatically improved in amide solvents (Table 2, entries 5−8).
With a combination of Et3N and HOAc in DMA, reductive
amination of bis pTSA salt 7 afforded 27 in 19:1 selectivity
(Table 2, entry 7). The desired crystalline product 27 was
directly isolated in 88% yield simply by adding aqueous ammonia
to the crude reaction mixture. It is important to note that the
filtration rate was significantly improved when the reaction slurry
time
(h)
conv
b
yield
b
entry
Ru complex
R3P
(%)
(%)
1
2
3
4
5
6
[RuCl2(C10H14)]2
[RuCl2(C10H14)]2
[RuCl2(CO)3]2
RuCl3
BINAP
30
30
40
40
16
24
62
73
97
98
99
98
−
−
38
41
20
89
(3-FPh)3P
(3-FPh)3P
(3-FPh)3P
PPh3
RuCl3
RuClCp(PPh3)2
None
a
Unless otherwise mentioned, all reactions were carried out at 85 °C
in DMF (0.4 M) in the presence of a Ru complex (5 mol %), Bu4NPF6
(50 mol %), NaHCO3 (50 mol %), and R3P (20 mol %). Determined
b
by HPLC analysis.9
very sensitive with poor reproducibility. After various studies, we
finally found that the attainment of a high catalytic Ru cycle for
the desired cycloisomerization could be achieved by simply
introducing 6.6 mol % PPh3 into the reaction system (Scheme
5); presumably, the active Ru species was further stabilized with
more phosphine ligand available in the reaction mixture.
Without isolation of dihydropyran 22, the crude stream after
aqueous workup was directly subjected to hydroboration. In
order to achieve full conversion, 2.5 equiv of BH3·SMe2 were
used (Scheme 5).17 Presumably, the NHBoc functionality
consumed/deactivated 1 equiv of borane. The hydroboration
proceeded efficiently between −10 and 0 °C. Following an
oxidative workup (NaBO3),18 the assay yield of 25 was >95%. At
this point, development of an effective crystallization process was
desired to isolate the desired 25 from the crude mixture of
diastereomers. After several experiments the desired crystalline
25 with >98% purity was successfully isolated as a 3:2 mixture of
diastereomers from toluene/heptane, along with nearly complete
5424
dx.doi.org/10.1021/ol502661g | Org. Lett. 2014, 16, 5422−5425

Organic Letters
Letter
a
Table 2. Selected Results of Reductive Amination of 6
REFERENCES
■
(1) Danaei, G.; Finucane, M. M.; Lu, Y.; Singh, G. M.; Cowan, M. J.;
Paciorek, C. J.; Lin, J. K.; Farzadfer, F.; Khang, Y.-H.; Stenes, G. A.; Rao,
M.; Ali, M. K.; Riley, L. M.; Robinson, C. A.; Ezzati, M. Lancet 2011, 378,
31.
(2) Since sitagliptin, the first DPP-4 inhibitor was approved by the
FDA in 2006, several additional DPP-4 inhibitors, including valdagliptin,
saxagliptin, linagliptin, and alogliptin are on the market.
(3) Biftu, T.; Qian, X.; Chen, P.; Feng, D.; Scapin, G.; Gao, Y.; Cox, J.;
Roy, R. S.; Eiermann, G.; He, H.; Lyons, K.; Salituro, G.; Patel, S.;
Petrov, A.; Xu, F.; Xu, S. S.; Zhang, B.; Calwell, C.; Wu, J. K.; Weber, A.
Bioorg. Med. Chem. Lett. 2013, 23, 5361.
b
entry
7
additives
5 equiv HOAc
solvents
DMA
dr
1
2
3
4
5
free base
6.4:1
9.5:1
3.6:1
8:1
5 equiv HOAc
2.5 equiv HOAc
5 equiv HOAc
NMP
i-PrOH
HCl salt
10% H2O/DMF
DMF
c
MsOH salt 3 equiv Et3N, 5 equiv
HOAc
16:1
6
7
8
3 equiv Et3N, 5 equiv
HOAc
DMF
DMA
NMP
14:1
19:1
12:1
pTSA salt
3 equiv Et3N, 5 equiv
HOAc
(4) For selectivity of hydroboration, see: Brown, H. C.; Prasad, J. V. N.
V. J. Am. Chem. Soc. 1986, 108, 2049.
(5) For example, see: (a) Verela-Fernan
C.; Varela, J. A.; Saa, C. Org. Lett. 2009, 11, 5350. (b) Koo, B.;
3 equiv Et3N, 5 equiv
HOAc
́ ́
dez, A.; Gonzalez-Rodríguez,
a
Unless otherwise noted, all reactions were carried out at 20 °C with
́
NaBH(OAc)3. Determined by HPLC analysis.9 Hygroscopic.
b
c
McDonald, F. E. Org. Lett. 2007, 9, 1737. (c) Trost, B. M.; Rhee, Y. H. J.
Am. Chem. Soc. 2003, 125, 7482. (d) Trost, B. M.; Rhee, Y. H. J. Am.
Chem. Soc. 2002, 124, 2528. (e) McDonald, F. E.; Chatterjee, A. K.
Tetrahedron Lett. 1997, 38, 7687. (f) McDonald, F. E.; Zhu, Y. H.
Tetrahedron 1997, 53, 11061. (g) McDonald, F. E.; Gleason, M. M. J.
Am. Chem. Soc. 1996, 118, 6648.
(6) For a preliminary communication of studies on selective
cycloisomerization, see: Zacuto, M. J.; Tomita, D.; Pirzada, Z.; Xu, F.
Org. Lett. 2010, 12, 684.
(7) For recent examples, see: (a) Xu, F.; Chung, J. Y. L.; Moore, J. C.;
Liu, Z.; Yoshikawa, N.; Hoerrner, R. S.; Lee, J.; Royzen, M.; Cleator, E.;
Gibson, A. G.; Dunn, R.; Maloney, K. M.; Alam, M.; Goodyear, A.;
Lynch, J.; Yasuda, N.; Devine, P. N. Org. Lett. 2013, 15, 1342.
(b) Limanto, J.; Krska, S. W.; Dorner, B. T.; Vazquez, E.; Yoshikawa, N.;
Tan, L. Org. Lett. 2010, 12, 512. (c) Arai, N.; Ooka, H.; Yabuuchi, T.;
Kurono, N.; Inoue, T.; Ohkuma, T. Org. Lett. 2007, 9, 939.
was heated to 70 °C to dissolve/digest fine particle solids before
the batch was cooled to ambient temperature for filtration.
The initial rejection of the diastereomer of 27 from the
reaction mixture was inefficient, as the isolated 27 contained
about 4% of the diastereomer. However, the rejection of the
corresponding diastereomer was excellent in the endgame. Thus,
treatment of 27 with HCl in aqueous EtOH yielded 1 near-
quantitatively. The bis HCl salt dihydrate 1 was directly isolated
from the reaction stream in 90% yield and >98.9% purity. The
corresponding minor diastereomer carried from the previous
step was easily cleared to <0.5%.
In summary, an efficient asymmetric synthesis of tetrahy-
dropyran DPP-4 inhibitor 1 has been developed. This practical
synthesis features an application of Ru-catalyzed DKR reduction
to establish two contiguous stereogenic centers of an anti aryl
1,2-amino alcohol in >99% ee in one step. A Ru-promoted O-
selective cycloisomerization followed by hydroboration and a
Ru-catalyzed oxidation prepares the desired functionalized
pyranone 6. Finally, stereoselective reductive amination and
subsequent acidic deprotection complete the synthesis of 1.
Starting from inexpensive glycine ester 12, the overall yield of this
synthesis is 25%. This synthesis is also amenable to the
preparation of various analogs of the title compound.
́
(8) For leading references, see: (a) Brea, R. J.; Lopez-Deber, P.;
Castedo, L.; Granja, J. J. Org. Chem. 2006, 71, 7871. (b) Park, K.-H.;
Kurth, M. J. Tetrahedron Lett. 1999, 40, 5841. (c) Sauvagnat, B.; Lamaty,
F.; Lazaro, R.; Martinez. Tetrahedron Lett. 1998, 39, 821. (d) Lop
́
ez, A.;
Moreno-Manas, M.; Pleixats, R.; Roglans, A. Tetrahedron 1996, 24, 8365
̃
and references cited therein.
(9) For a detailed discussion, see the Supporting Information.
(10) For example, the procedure reported in ref 8a was cumbersome to
run and the yields obtained in our lab were not as high as quoted.
(11) It is not recommended to use propargyl bromide on large scale
due to safety concerns.
(12) The addition mode has been studied in-depth, but its mechanism
to suppress the formation of byproducts remains unclear.
(13) Alternatively, the inorganic salts could be removed by filtration.
(14) Scott, J. P.; Brewer, S. E.; Davies, A. J.; Brands, K. J. M. Synlett
2004, 9, 1646.
ASSOCIATED CONTENT
■
(15) (a) Hayes, A.; Clarkson, G.; Wills, M. Tetrahedron: Asymmetry
2004, 15, 2079. (b) Sandoval, C.; Ohkuma, T.; Muniz, K.; Noyori, R. J.
Am. Chem. Soc. 2003, 125, 13490. (c) Noyori, R.; Yamakawa, M.;
Hashiguchi, S. J. Org. Chem. 2001, 66, 7931.
S
* Supporting Information
̃
Experimental procedure/data and discussion. This material is
available free of charge via the Internet at http://pubs.acs.org.
(16) Preliminary results showed that Au-catalyzed cyclization led to an
undesired 5-exo dig cyclization/isomerization.
AUTHOR INFORMATION
■
(17) With <1 equiv BH3·SMe2, no desired product was observed.
(18) The use of H2O2 was unsafe due to its exothermic and catalytic
decomposition by residual Ru or Rh present in solution from the
previous step. The use of NaBO3, by contrast, was free of these concerns.
For application of NaBO3, see: Kabalka, G. W.; Shoup, T. M.;
Goudgaon, N. M. J. Org. Chem. 1989, 54, 5930.
Corresponding Authors
*E-mail: feng_xu@merck.com.
*E-mail: michael_zacuto@merck.com.
Notes
The authors declare no competing financial interest.
(19) The two polymorph forms of RuCl3 could dramatically affect the
oxidation rate due to different dissolution rates.9
(20) Overoxidation of 6 was suppressed at 0 °C.
ACKNOWLEDGMENTS
■
(21) The reductive amination could be carried out in the presence of a
small amount of water in DMF. However, competitive reduction of 6 to
the corresponding alcohol became significant if more water was
introduced. Attempts to use strong acid salts of 7 directly for reductive
amination were unsuccessful.
We gratefully acknowledge R. Reamer, P. Dormer, and L.
DiMichele (Merck & Co., Inc.) for NMR assistance, Margaret
Figus (Merck & Co., Inc.) for HRMS assistance, and T. Alorati
and S. Brewer (Merck & Co., Inc.) for helpful discussions and
experimentation.
5425
dx.doi.org/10.1021/ol502661g | Org. Lett. 2014, 16, 5422−5425