First generation process for the preparation of the DPP-IV inhibitor sitagliptin

Article abstract of DOI:10.1021/op0500786

A new synthesis of sitagliptin (MK-0431), a DPP-IV inhibitor and potential new treatment for type II diabetes, suitable for the preparation of multi-kilogram quantities is presented. The triazolopyrazine fragment of sitagliptin was prepared in 26% yield over four chemical steps using a synthetic strategy similar to the medicinal chemistry synthesis. Key process developments were made in the first step of this sequence, the addition of hydrazine to chloropyrazine, to ensure its safe operation on a large scale. The beta-amino acid fragment of sitagliptin was prepared by asymmetric reduction of the corresponding beta-ketoester followed by a two-step elaboration to an N-benzyloxy beta-lactam. Hydrolysis of the lactam followed by direct coupling to the triazolopiperazine afforded sitagliptin after cleavage of the N-benzyloxy group and salt formation. The overall yield was 52% over eight steps.

Full text of DOI:10.1021/op0500786

Organic Process Research & Development 2005, 9, 634−639
First Generation Process for the Preparation of the DPP-IV Inhibitor Sitagliptin
Karl B. Hansen,* Jaume Balsells, Spencer Dreher, Yi Hsiao, Michele Kubryk, Michael Palucki, Nelo Rivera,
Dietrich Steinhuebel, Joseph D. Armstrong III, David Askin, and Edward J. J. Grabowski
Department of Process Research, Merck Research Laboratories, Rahway, New Jersey 07065, U.S.A.
Abstract:
Discussion
The synthesis of triazole 3 began with the alkylation of
A new synthesis of sitagliptin (MK-0431), a DPP-IV inhibitor
and potential new treatment for type II diabetes, suitable for
the preparation of multi-kilogram quantities is presented. The
triazolopyrazine fragment of sitagliptin was prepared in 26%
yield over four chemical steps using a synthetic strategy similar
to the medicinal chemistry synthesis. Key process developments
were made in the first step of this sequence, the addition of
hydrazine to chloropyrazine, to ensure its safe operation on a
large scale. The beta-amino acid fragment of sitagliptin was
prepared by asymmetric reduction of the corresponding beta-
ketoester followed by a two-step elaboration to an N-benzyloxy
beta-lactam. Hydrolysis of the lactam followed by direct
coupling to the triazolopiperazine afforded sitagliptin after
cleavage of the N-benzyloxy group and salt formation. The
overall yield was 52% over eight steps.
chloropyrazine 6 with hydrazine (Scheme 2). Refluxing 6
with 5 equiv of hydrazine monohydrate in 2-propanol
resulted in complete conversion to hydrazine adduct 7. Upon
cooling to room temperature, 7 crystallized and was isolated
directly from the reaction mixture. While this procedure was
successful on a laboratory scale, preparation of larger
quantities was not possible due to problems with the handling
of hydrazine.
Heating of hydrazine-hydrate in organic solvents, espe-
cially solvents which form an azeotrope with water, leads
to explosive anhydrous hydrazine in the vapor phase.6 To
ensure that anhydrous hydrazine would not be present, the
reaction would need to be carried out on a large scale under
aqueous conditions. Unfortunately, performing the reaction
with larger concentrations of water present required heating
to temperatures over 60 °C to ensure complete reaction. The
reaction of 6 with hydrazine was also found to be uncontrol-
lably exothermic at temperatures above 85 °C, so the reaction
procedure would need to be modified to ensure that large
concentrations of 6 would not build up in the vessel.7
A procedure for the preparation of 7 was developed which
could be performed safely on large scale. By adding
chloropyrazine 6 slowly to a 35 wt % aqueous solution of
hydrazine8 at 60-65 °C, its concentration in the reaction
was kept low. This change in the reaction procedure, coupled
with careful monitoring of the reaction temperature allowed
for it to be performed safely on scale-up.9 Upon complete
consumption of 6, the reaction mixture was cooled to ambient
temperature and extracted with 10% 2-propanol/dichlo-
romethane. By using 10% 2-propanol/dichloromethane as the
extraction solvent, the organic extracts contained sufficiently
low levels of hydrazine to allow their concentration by
distillation, while still obtaining a 79% assay yield of 7.10
The organic extracts were distilled in vacuo, replacing them
with isopropyl acetate (IPAC) in preparation for the subse-
quent reaction.
Inhibitors of the enzyme DPP-IV are a promising new
treatment for type II diabetes.1 Sitagliptin (MK-0431, 1) is
a potent inhibitor of DPP-IV which is currently being
evaluated in clinical trials.2 Compound 1 was originally
synthesized by coupling beta-amino acid 2a to triazole 3,
which was prepared by hydrogenation of the unsaturated
triazolopyrazine 4 (Scheme 1).3 The synthesis of beta-amino
acid 2a in optically enriched form was carried out using the
Scho¨llkopf chiral auxiliary.4 The original approach to the
preparation of triazolopiperazine 3 could potentially be scaled
up for our first large-scale preparation of 1; however, the
use of a chiral auxiliary to prepare the amounts of enantio-
merically enriched 2a we required was not tenable.
A synthesis of the 2,5-difluorophenyl analogue of 2a has
been recently reported by others from these laboratories.5
Using this method, 2a could be prepared from lactam 5,
which is derived from an optically enriched beta-hydroxy
acid. This method also requires introduction of the amino
group as an O-benzylhydroxylamine, which we hoped would
sufficiently protect the amino group of 2b during amide
formation with triazole 3. Herein we describe the route used
for the first large-scale preparation of 1 using this strategy.
(6) Information about hydrazine and its safe use can be obtained at http//
www.hydrazine.com.
(7) Calorimetry studies of the reaction of hydrazine with 6 display a strong
exotherm at temperatures above 85 °C.
(1) (a) Weber, A. J. Med. Chem. 2004, 48, 4135-4141. (b) Drucker, D. J.
Exp. Opin. InVestig. Drugs 2003, 12, 87-100. (b) Wiedeman, P. E.;
Trevillyan, J. M. Curr. Opin. InVestig. Drugs 2003, 4, 412-420.
(2) Kim, D.; Wang, L.; Beconi, M.; et al. J. Med. Chem. 2004, 48, 141-151.
(3) Nelson, P. J.; Potts, K. T. J. Org. Chem. 1962, 27, 3243-3248.
(4) (a) Xu, J.; Ok, H. O.; Gonzalez, E. J.; et al. Biorg. Med. Chem. Lett. 2004,
14, 4759-4762. (b) Scho¨llkopf, U.; Groth, U.; Deng, C. Angew. Chem.,
Int. Ed. Engl. 1981, 20, 798-799.
(8) 35 wt % hydrazine aqueous solution has no flashpoint.
(9) As a safety precaution, the reaction temperature was carefully monitored
upon scale-up. If the temperature of the reaction had exceeded 70 °C, a
quench vessel of water was connected to the vessel to quickly stop the
reaction. However there was no issue with maintaining the temperature of
reaction in the 60-65 °C range.
(10) The hydrazine levels were measured by formation of 3,5-dimethylpyrazole
by treating with 2,4-pentanedione and then analyzing by capillary gas
chromatography. Levels in the combined extracts were typically ∼2000 ppm.
(5) Angelaud, R.; Zhong, Y.-L.; Maligres, P.; Lee, J. Askin, D. J. Org. Chem.
2005, 70, 1949-1952.
634
•
Vol. 9, No. 5, 2005 / Organic Process Research & Development
10.1021/op0500786 CCC: $30.25 © 2005 American Chemical Society
Published on Web 08/31/2005

Scheme 1
Scheme 2
Scheme 3
Acylation of 7 with trifluoroacetic anhydride afforded the
plex in methanol at 90 psi H2 and 80 °C.13 The use of a
catalytic amount of HBr allowed the loading of the catalyst
to be reduced to <0.1 mol % without affecting the ee or
yield of beta-hydroxy ester 10.14 Following the reduction,
the ester was hydrolyzed and the carboxylic acid 11 was
isolated in 83% yield and 94% ee. Lactam 5 was then
prepared in a two-step sequence from 11. First, hydroxamate
12 was formed by coupling the carboxylic acid with
BnONH2-HCl using N-(3-dimethylaminopropyl)-N′-ethyl-
carbodiimide hydrochloride (EDC). Upon complete con-
sumption of 11, amine byproducts were removed by aqueous
workup and the organic extracts were dried azeotropically
bis-trifluorohydrazide 8 which was isolated directly from the
reaction by adding heptane and filtering. Heating the isolated
solids in polyphosphoric acid (PPA) resulted in the cycliza-
tion of both compounds to afford triazolopyrazine 4. Due to
the high viscosity of the PPA as well as the strong exotherm
associated with quenching PPA with water, superphos,11
a
more hydrated and less viscous form of PPA was used with
identical results. Finally, pyrazine 4 was hydrogenated with
Pd on carbon to the desired triazolopyrazine which, after
filtration to remove the catalyst, was isolated as its HCl salt
in 51% yield from 8 (26% yield from chloropyrazine 6).
The synthesis of 2b started with the asymmetric hydro-
genation of beta-ketoester 9 (Scheme 3).12 The reduction of
9 was carried out using (S)-BinapRuCl2-triethylamine com-
(12) Brooks, D. W.; Lu, L. D.-L.; Masamune, S. Angew. Chem., Int. Ed. Engl.
1979, 18, 72-73.
(13) Noyori, R.; Ohkuma, T.; Kitamura, M.; Takaya, H.; Sayo, N.; Kumobayashi,
H.; Kkutagawa, S. J. Am. Chem. Soc. 1987, 109, 5856.
(14) King, S. A.; Thompson, A. S.; King, A. O. Verhoeven, T. R. J. Org. Chem.
1992, 55, 6689-6691.
(11) Superphos contains 105% equivalent H3PO4, whereas polyphosphoric acid
is 115% H3PO4.
Vol. 9, No. 5, 2005 / Organic Process Research & Development
•
635

Scheme 4
for direct use in the subsequent reaction. The cyclization to
form 5 was carried out directly in the solution of 12 using
di-isopropyl azodicarboxylate (DIAD) and PPh3.15 Optimum
results were obtained when the azeotropically dried THF
solution of 12 was added to a mixture of PPh3 and DIAD
cooled to 0 °C. The reaction was assayed after complete
addition of 12, and typically less than one HPLC area % of
starting material was observed. The product was isolated in
81% yield by crystallizing from methanol/water. An upgrade
of the optical purity of 5 to >99% ee was observed in the
crystallization when the ee of 11 used in the two-step
sequence was at least 94% ee.
isolated in identical purity and yield as when it was not
present.
Despite having a solution to the problem of methanolysis
of 5, we also needed to process the methyl ester 13 which
had formed to 1. The optical purity of the lactam had not
eroded during its transformation to 13; however a new
isolation needed to be developed to upgrade the ee prior to
elaboration to 1. In addition to the undesired enantiomer,
the solution of 13 contained triphenylphosphine oxide and
reduced DIAD byproducts from the cyclization.17 Methyl
ester 13 exists as an oil and could not be isolated directly;
however its HCl salt was found to be a crystalline solid.
Crystallization of the HCl salt 13 from IPAC/heptane resulted
in an upgrade of the optical purity to >99%, with an overall
yield which was nearly identical to the lactam isolation.
The synthesis of 1 was completed using a four-step
through-process (Scheme 4). Lactam 5 or ester 13 was
hydrolyzed to amino acid 2b with LiOH18 in THF/water by
either stirring at room temperature or, in the case of 13,
heating to 40 °C. While the benzyloxy group of 2b could
be cleaved by hydrogenation and then protected with Boc2O
to prevent side reactions during the coupling to triazole 3,
the benzyloxy group of 2b was found to sufficiently protect
the amino group to allow the desired amide to be formed.
Thus, triazole 3 was coupled to 2b at 0 °C using EDC-HCl
and N-methylmorpholine (NMM) as base to afford 14 in
>99% assay yield. Following an aqueous workup, the
organic extracts were distilled into ethanol and the solution
was subjected to hydrogenation with 10% Pd on carbon. The
presence of water in the hydrogenation was found to be
crucial to the reaction success; anhydrous solutions of 14
hydrogenated with dry Pd on carbon proceeded only to low
levels of conversion to 1, and addition of water to these
reductions resulted in restored performance of the catalyst.
Following hydrogenation, the catalyst was removed by
filtration to provide an ethanol solution of 1. Sitagliptin was
isolated in >99.5% purity as its anhydrous phosphoric acid
salt by crystallizing from aqueous ethanol.
Control of the optical purity of 5 was deemed to be crucial
at this point of the synthesis. Crystallization of 1 as its desired
phosphoric acid salt with ee below the final desired
specification did not upgrade its optical purity, and we
intended to process 5 to 1 in a through-process with only a
final isolation. The isolation of 5 was performed several times
on a kilo scale with excellent results; however upon piloting,
a large batch of 5 was observed to completely form methyl
ester 13 during a solvent switch by distillation to the
methanol crystallization solvent. To process the remaining
batches of lactam, conditions needed to be developed which
would ensure its stability during the distillation of the solvent.
Crude solutions of 5, which had been tested to be slightly
acidic following the reaction, had been stressed by heating
in methanol, and none of methyl ester 13 was observed to
form. The addition of as little as 0.1 mol % NaOH resulted
in the formation of 13 at the temperature the distillation was
carried out. The addition of strong acid resulted only in
formation of amounts of methyl ester corresponding to the
charge of acid. However, solutions charged with acetic acid
were found to be stable for days at elevated temperatures.
The exact cause of the methanolysis was not determined;
however these stability experiments suggest that lactam 5 is
most likely methanolyzed due to adventitious base in the
reaction.16 To protect future batches of 5 from methanolysis,
a small amount of acetic acid was added to the distillation
of solvent to ensure that the pH of the solution would remain
low. This modification to the procedure resulted in complete
prevention of methanolysis, even when experiments were
challenged with aliquots of NaOH prior to and during
distillation. The presence of HOAc did not affect the
performance of the crystallization, and the lactam was
(15) Miller, M. J.; Mattingly, P. G.; Morrison, M. A.; Kerwin, J. F. J. Am. Chem.
Soc. 1980, 102, 7026-7032.
(16) Detergents used to clean vessels and transfer lines can have a very high
pH. A small amount of basic soap catalyzes the formation of 13 from 5.
(17) Solutions of 13 were measured to be only ∼25 HPLC area % purity at 210
nm.
(18) NaOH or KOH could be used instead of LiOH with no effect on reaction
performance.
636
•
Vol. 9, No. 5, 2005 / Organic Process Research & Development

The large-scale synthesis of sitagliptin, which prepares
this important new drug candidate in 60% yield over eight
steps from keto-ester 9 and triazole 3, has been described.
Triazole 3 was prepared in 26% yield by transforming the
route used for the discovery of 1 into a process which can
be run safely on multi-kilogram scale. While this route
allowed for the preparation of large quantities of 3, the
difficulty with the alkylation of chloropyrazine with hydra-
zine coupled with the low overall yield has resulted in the
development of a more efficient process for its preparation.19
Beta-amino acid 2b, prepared in five steps, was coupled
directly to 3 to afford sitagliptin after deprotection. While
this route allowed for the preparation of multi-kilogram
quantities of 1, the number of steps required to form the
beta-amino acid moiety has led us to develop alternative
methodology for its preparation.20 The application of this
methodology to the synthesis of 1 will be published shortly.
1H), 9.30 (s, 1H), 8.66 (d, J ) 2 Hz, 1H), 8.64 (m, 1H). 13
C
NMR (DMSO-d6, 100 MHz) δ 157.6 (q, JC-F ) 37.8 Hz),
157.3 (q, JC-F ) 37.6 Hz), 146.1, 143.6, 143.1, 139.3, 115.4
(q, JC-F ) 288 Hz). 19F NMR (DMSO-d6, 376 MHz) δ
-74.5, -71.3.
3-Trifluoromethyl-[1,2,4]triazolo[4,3-a]piperazine (3).
4.8 kg (15.8 mol, 1 equiv) of 8 were charged to 19.2 L of
superphosphoric acid. The viscous slurry was heated to an
internal temperature of 75 °C at which point the color became
dark red and the mixture became homogeneous. After 5 h,
the reaction was cooled to 30 °C and sampled for conversion
by HPLC. Water (19.2 L) was added over 1.5 h to the
reaction mixture while maintaining the temperature below
40 °C. NH4OH (28.8 L) was carefully added to the mixture
(gas evolution and strong exotherm) while keeping the
temperature below 40 °C, until a pH of 8-9 was achieved.
The slurry was extracted with 2 × 29 L and then 1 × 14 L
of IPAC. The solids present were partitioned with the
aqueous phase during each separation. The combined organic
layers were treated with 0.5 kg of DARCO KB for 1 h and
then filtered through cellulose. The waste cake was washed
with 9.6 L of IPAC. The combined filtrate and wash were
assayed to contain 2.11 kg of 4 (71% HPLC assay yield).
The combined organic extracts were then solvent switched
into EtOH by vacuum distillation, and upon complete
removal of IPAC, the volume was adjusted to 36 L with
EtOH. The solution was then hydrogenated in three 12 L
batches using 200 g of 10% Pd/C (each batch) at 50 psig H2
and 45 °C. The combined hydrogenation mixtures were then
treated with 200 g of DARCO G-60 and then filtered through
cellulose. The cake was washed with ethanol, and the
combined filtrate and wash was assayed to contain 1.87 kg
of piperazine 3 (9.76 mol, 87% yield from 4).
The solvent was then switched into IPA by vacuum
distillation until <2 vol % ethanol remained. The volume
was adjusted to 8 L, and then 2.4 L of a 3.7 M solution of
HCl in IPA (8.8 mol, 1.01 equiv, prepared by diluting
concentrated HCl with IPA, charge based on assay yield of
3) were charged with cooling. IPAC (29 L) was added to
the slurry, and the mixture was aged for 1 h at 20 °C and
then filtered. The cake was washed with 4 L of 13 vol %
IPA in IPAC and then dried under vacuum with a nitrogen
sweep. 1.8 kg (8.01 mol, 51% yield from 8) of 3-HCl as a
white solid were obtained.21
4-(2,4,5-Trifluorophenyl)-3(S)-hydroxybutanoic Acid
(11). A solution of ketoester 9 (2.5 kg, 10.2 mol) and HBr
(115 mL, 48 wt % solution in water, 1.01 mol, 0.1 equiv) in
11.5 L of methanol was degassed by bubbling N2 through
the solution for 20 min and then transferred to a stirred
autoclave. 21.5 g of (S)-BinapRuCl2 triethylamine complex
was dissolved in 1 L of methanol which had been degassed
as above and transferred to the autoclave. Following addition
of the catalyst solution, the entire reaction solution was
further degassed by five vacuum purge/N2 back-fill cycles.
The mixture was then subjected to 90 psig H2 at 80 °C for
10 h at which point <1% ketoester remained by HPLC assay.
The solution was removed from the vessel with a methanol
Experimental Section
Melting points were determined on an open capillary
apparatus and are uncorrected. HPLC assays were carried
out using a C-18 reversed-phase column eluted with aceto-
nitrile and 0.1% H3PO4 (aq). GC assays were carried out
using a 30 m RTX-5 capillary column (1.0 µM thickness).
Assay yields were obtained by HPLC or by GC using pure
compounds as standards. All reagents and solvents were used
as received without further purification.
(1,2)-Bis-trifluoroacetyl-1-pyrazinylhydrazide(8).2-Chlo-
ropyrazine (3.9 L, 5.0 kg, 43.7 mol, 1 equiv) was added
dropwise to 22.2 L of 35 wt % aqueous solution of hydrazine
(245 mol, 5.6 equiv) at 63-65 °C over 4 h. The addition
rate was carefully monitored to ensure that the reaction
temperature did not exceed 65 °C. Following the addition,
the reaction mixture was aged for 11-13 h at 65 °C, at which
point <1% chloropyrazine was observed by GC area percent
assay. The reaction mixture was cooled to 30 °C and then
extracted 5 times with 10% (v/v) 2-propanol/CH2Cl2. The
combined extracts were assayed to contain 3.88 kg (35.3 mol,
81% assay yield) of 6. The extracts were then concentrated
by distillation and flushed with IPAc until <1 vol % of IPA
remained as judged by GC assay. The solution was diluted
with IPAC to a total volume of 51 L at which point the
mixture was a slurry of white solids. The solution was cooled
to 5 °C, and trifluoroacetic anhydride (19.6 L, 29.2 kg, 139.1
mol, 3.94 equiv) was added while maintaining the reaction
temperature below 20 °C by controlling the addition rate
(∼1 h total addition time). Upon complete addition, the
reaction was assayed by HPLC and contained <1 area
percent 7. Heptane (22 L) was added to the slurry, and the
product was isolated by filtration. The solids were washed
with 9 L of 10% (v/v) IPAC/heptane. After drying at ambient
temperature under a vacuum/nitrogen sweep, 6.44 kg of a
yellow solid (61% yield from 7, 49% yield over both steps)
1
was isolated. H NMR (DMSO-d6, 400 MHz) δ 13.30 (s,
(19) Balsells, J.; DiMichele, L.; Kubryk, M.; Hansen, K.; Armstrong, J. D. Org.
Lett. 2005, 7, 1039-1042.
(20) (a) Hsiao, Y.; Krska, S.; Rivera, N.; Rosner, T. et al. J. Am. Chem. Soc.
2004, 126, 9918-9919. (b) Ikemoto, N.; Tellers, D.; Rivera, N. et al. J.
Am. Chem. Soc. 2004, 126, 3047-3048.
(21) Complete characterization of 3-HCl can be found in ref 19.
Vol. 9, No. 5, 2005 / Organic Process Research & Development
•
637

rinse. The solution was assayed to contain 2.51 kg of 10
(99.3% yield) which was 90.8% ee by chiral HPLC analysis
(Chiracel AD-H, 95% hexanes, 5% IPA, 1% TFA, 1 mL/
min, 35 °C, tr ) 12.3 min (S-10), 15.7 (R-10).
slurry keeping the reaction temperature below 10 °C. Upon
completion of the addition, the reaction was warmed to 20
°C and then aged overnight (18 h). The reaction was assayed
by HPLC at which point <1% area percent starting material
remained. Acetic acid (51.6 g, 0.86 mol) was charged to the
solution which was then solvent switched to methanol.
Crystals formed upon complete removal of THF. The final
volume of the mixture was adjusted to 43 L with methanol,
and then water (2.8 L) was added. The mixture was heated
to 35 °C to redissolve all of the solids and then slowly cooled
to -20 °C. After 1 h at -20 °C, the slurry was filtered and
then washed with 2 × 4.5 L of 10% v/v water/methanol
cooled to -20 °C. 4.7 kg of 5 (93 wt % purity by HPLC)
were isolated as a crystalline solid after drying at room
temperature in vacuo (82% yield over two steps). The
material was assayed by chiral HPLC to be 99.7% ee.
(Chiracel AD-RH, 60% CH3CN/40% H2O, 0.6 mL/min, 20
A solution of 10 (5.00 kg, 20.15 mol) in 25 L of methanol
was charged to a round-bottom flask equipped with an
overhead stirrer, steam bath, and thermocouple. A solution
of aqueous NaOH (0.89 kg, 22.16 mol, 1.1 equiv) dissolved
in 25 L of water was charged to the methanol solution. The
mixture was aged for 1 h at which point HPLC assay
indicated complete consumption of ester starting material.
The methanol was removed by distillation in vacuo, and the
resulting solution was transferred to an extractor. 12 N HCl
(3.8 L, 34.3 mol, 1.7 equiv) and 15 L of MTBE were added
to the solution with cooling. The layers were separated, and
the aqueous layer was back-extracted with 15 L of MTBE.
The combined MTBE layers were switched to toluene by
distillation at 50-60 °C, and upon complete removal of
MTBE, the volume was adjusted to 50 L. The solution was
then allowed to cool to room temperature. At ∼37 °C the
product began to crystallize from solution. The mixture was
aged overnight at room temperature (∼18 h). The crystals
were isolated by filtration and washed with 5 L of toluene.
Following drying at 40 °C, 4.08 kg of hydroxy acid 11 which
was 94% ee and 96.7 wt % pure were isolated in 83% yield.
(Chiracel AD-H, 95% hexanes, 5% IPA, 1% TFA, 1 mL/
min, 35 °C, tr ) 20.3 min (S-11), 22.9 (R-11)). Mp: 84 °C.
1H NMR: (CDCl3, 400 MHz) δ 7.11 (m, 1H), 6.92 (m, 1H),
4.27 (m, 1H), 2.82 (d, J ) 6.1 Hz, 2H), 2.61 (dd, J ) 3.4,
16.8 Hz, 1H), 2.52 (dd, J ) 8.7, 16.8 Hz, 1H). 13C NMR:
(CD3OD, 100 MHz) δ 173.7, 156.3 (dd, JC-F ) 9.3, 243.1
Hz), 148.7 (ddd, JC-F ) 12.8, 27.3, 248.0 Hz), 146.2 (ddd,
JC-F ) 3.6, 12.4, 242.5 Hz), 122.1 (ddd, JC-F ) 4.4, 5.4,
18.3 Hz), 119.2 (dd, JC-F ) 6.2, 19.3 Hz), 104.7 (dd, JC-F
) 21.1, 29.3), 67.6, 41.1, 35.1. 19F NMR: (CDCl3, 376 MHz)
-119.7 (dd, JF-F ) 3.2, 15.5 Hz), -136.0 (dd, JF-F ) 3.2,
21.0 Hz), -133.3 (dd, JF-F ) 15.5, 21.0 Hz). [R]D ) +16.3°
(c ) 1.0, CHCl3). Anal. Calcd for C10H9F3O3: C, 51.29; H,
3.87. Found: C, 50.59; H, 3.62.
N-Benzyloxy-4(R)-[1-methyl-(2,4,5-trifluorophenyl)]-2-
oxoazetidine (5). To a slurry of hydroxyacid (4.0 kg, 17.1
mol, 1.0 equiv), O-benzylhydroxy amine hydrochloride (3.0
kg, 18.8 mol, 1.1 equiv), and LiOH (0.72 kg, 17.1 mol, 1.0
equiv) in 10 L of THF and 30 L of water at 20 °C was added
EDC-HCl (4.26 kg, 22.2 mol, 1.3 equiv). The solution was
aged for 1 h at which point <1% hydroxy acid was present
by HPLC assay. 32 L of MTBE were added, and the phases
were separated. The organic layer was then vacuum distilled
and dried by flushing first with MTBE and then with THF
until all of the MTBE had been removed and the water
content was <2000 ppm as judged by Karl Fisher titration.
The final volume was adjusted to 17 L, and the solution was
then used directly in the next step. In a separate vessel, DIAD
(3.70 L, 18.8 mol, 1.1 equiv) was charged slowly via addition
funnel to a solution of PPh3 (4.93 kg, 18.8 mol, 1.1 equiv)
at 0 °C such that the temperature did not rise above 10 °C.
A slurry of white solids formed during the addition. The THF
solution of hydroxamate 12 was then added slowly to the
1
°C, tr ) 9 min (R-5), 11 min (S-5). Mp: 80 °C. H NMR
(CDCl3, 400 MHz): δ 7.37 (m, 5H), 6.90 (m, 2H), 4.92 (m,
2H), 3.63 (m, 1H), 2.86 (dd, J ) 5.0, 14.3 Hz, 1H), 2.67
(m, 2H), 2.33 (dd, J ) 2.2, 13.8 Hz, 1H). 13C NMR (CDCl3,
400 MHz): δ 163.6, 155.9 (dd, JC-F ) 10.6, 244.2 Hz),
149.1 (dd, JC-F ) 12.5, 250.9 Hz), 146.6 (ddd, JC-F ) 3.7,
12.5, 231.3 Hz), 135.3, 129.3, 129.1, 128.6, 119.7 (ddd, JC-F
) 4.4, 5.1, 18.4 Hz), 118.8 (dd, JC-F ) 6.5, 19.2 Hz), 105.5
(dd, JC-F ) 20.8, 28.5 Hz), 78.2, 57.0, 37.6, 31.0. 19F NMR
(CDCl3, 377 Hz): δ -119.4 (dd, JF-F ) 3.2, 15.1 Hz),
-135.5 (dd, JF-F ) 3.2, 21.6 Hz), -147.7 (dd, JF-F ) 14.9,
21.5 Hz). [R]D ) + 26.2° (c ) 1.04, CHCl3). Anal. Calcd
for C17H14F3NO2: C, 63.55; H, 4.39; N, 4.36. Found: C,
63.55; H, 4.27; N, 4.33.
Methyl N-Benzyloxy-4-(2,4,5-trifluorophenyl)-3(R)-
aminobutanoate Hydrochloride (13-HCl). A solution of
3.34 kg of methyl ester 13 (9.44 mol, 1.0 equiv) was charged
to a 100 L of RBF equipped with an overhead stirrer,
thermocouple, and distillation condenser. Hydrochloric acid
(0.89 L, 12 M, 10.7 mol, 1.2 equiv) was added to the solution
with cooling. The solution was concentrated in vacuo to 40%
of its original volume and then flushed continuously with
50 L of IPAC, while distilling at a constant volume and
maintaining the temperature below 45 °C. The solution was
checked for methanol content by GC and found to contain
<1%. The solution was cooled to room temperature and aged
for 1 h at which point a slurry had formed. Heptane (50 L)
was added over 0.5 h, and then the mixture was cooled to 0
°C and aged until the HPLC analysis of the supernatant
showed a phosphine oxide level of <12% relative to methyl
ester at 268 nM (∼12 wt %). The slurry was filtered and
washed with 7 L of 90% heptane/IPAC cooled to 0 °C. The
solids were dried at ambient temperature in a vacuum oven
with a nitrogen sweep to afford 3.02 kg of 13-HCl. The ee
of the methyl ester was assayed to be >99% by hydrolyzing
to 2b using KOH/H2O (Chiracel OD-H, 95% (0.1% TFA in
hexanes)/5% (0.1% TFA in EtOH), 1.0 mL/min, 10 °C, tr )
1
11 min (S-), 13 min (R-). Mp ) 123 to 125 °C. H NMR
(CDCl3, 400 MHZ): δ 7.39 (m, 5H), 7.17 (m, 1H), 6.88
(m, 1H), 5.41 (s, 3H), 4.24 (m, 1H), 3.46 (dd, J ) 14.1, 5.0
Hz, 1H), 3.12 (m, 2H), 2.69 (dd, J ) 17.1, 6.4 Hz, 1H). 13
C
638
•
Vol. 9, No. 5, 2005 / Organic Process Research & Development

NMR (CDCl3, 100 MHz): δ 169.7, 156.4 (dd, JC-F ) 7.0,
245 Hz), 149.5 (ddd, JC-F ) 12.9, 14.1, 252.0 Hz), 146.8
(ddd, JC-F ) 3.5, 12.4, 249.1 Hz), 132.2, 129.6, 128.7, 119.4
(dd, J ) 5.3, 19.4 Hz), 118.3 (ddd, JC-F ) 5.0, 9.7, 18.1
Hz), 105.7 (dd, JC-F ) 20.1, 28.1 Hz), 76.9, 57.2, 52.2, 33.2,
the mixture was heated to 50°C. Phosphoric acid (85 wt %,
1.33 kg, 11.5 mol, 1.0 equiv) was then added in one portion,
and the temperature of the solution raised to 70-74 °C. The
solution temperature was lowered to 65 °C, and it was then
seeded with sitagliptin H3PO4 salt. The slurry was aged for
1 h and then slowly cooled to rt. 75 L of EtOH were added
slowly to the slurry, which was then aged for 18 h at room
temperature. The crystals were isolated by filtration and
washed with 2 × 10 L of ethanol. The solids were dried in
a 40 °C vacuum oven with a nitrogen sweep to afford 5.30
kg of 1-H3PO4 (78% yield from 5). The optical purity was
assayed to be >99.5% ee. Chiralpak AD-H, 60% (2% H2O,
2% diethylamine, 96% hexanes), 40% (2% H2O, 2% diethyl-
amine, 96% EtOH), 0.8 mL/min, 35 °C, tr ) 13 min (S-1),
15 min (R-1). Mp ) 215-217 °C. 1H NMR (D2O, 600 MHz)
δ 7.06 (m, 1H, major and minor), 6.91 (m, 1H, major and
minor), 4.76 (s, 2H, minor), 4.75 (d, J ) 17.6 Hz, 1H, major),
4.70 (d, J ) 17.6 Hz, 1H, major), 4.11 (m, 2H, major), 4.07
(m, 1H, minor), 4.02 (m, 1H, minor), 3.83 (m, 1H minor
and 2H major), 3.79 (m, 2H, major), 2.91 (m, 1H, both),
2.83 (m, 2H, major and minor), 2.79 (dd, J ) 17.0, 4.9 Hz,
1H, major), 2.68 (m, 1H, major and minor). 13C NMR: (D2O,
151 MHz) δ 170.4 (minor), 170.3 (major), 156.5 (m, major
and minor), 151.2 (major), 150.7 (minor), 149.4 (m, major
28.5. 19F NMR (CDCl3, 376 MHz) δ -118.3 (dd, JF-F
)
3.2, 15.5 Hz), -134.1 (dd, JF-F ) 3.2, 22.1 Hz), 142.2 (dd,
JF-F ) 15.5, 22.1 Hz). [R]D ) + 7.3° (c ) 1.0, CHCl3).
Anal. Calcd for C18H19ClF3NO3: C, 55.46; H, 4.91; N, 3.59.
Found: C, 55.54; H, 4.69; N, 3.56.
Sitagliptin (1). A solution of lactam (4.65 kg, 93 wt %
purity, 13.4 mol, 1.0 equiv) and LiOH monohydrate (0.84
kg, 20 mol, 1.5 equiv) in 14 L of THF and 14 L of water
was aged for 1.5 h at 20 °C at which point <1% starting
material was present by HPLC assay. Methanesulfonic acid
(1.30 L, 20 mol, 1.5 equiv) was slowly added to the reaction
with cooling such that the temperature stayed below 20 °C.
MTBE (28 L) was charged, and the phases were mixed well
and then separated. The organic layer was then concentrated
by vacuum distillation and then distilled at constant volume
while the solvent was replaced with acetonitrile. The volume
was then adjusted to 43 L with acetonitrile. Triazole HCl
salt (3.81 kg, 16.7 mol, 1.25 equiv) was charged, and the
mixture was then cooled to 0 °C. N-Methyl morpholine (1.35
kg, 13.4 mol, 1.0 equiv) was then charged to the slurry
followed by EDC-HCl (3.85 kg, 20.07 mol, 1.5 equiv). The
reaction was aged for 2 h at 0 °C at which point <1% 2b
remained by HPLC assay. The mixture was quenched with
20 L of water and 40 L of MTBE. After warming to 15 °C,
the layers were separated and the organic layer was washed
with 1 × 20 L of 10% KHCO3 and 1 × 20 L of 20% NaCl
solution. The organic layer was then concentrated by vacuum
distillation, and the solvent was switched to ethanol. The
volume was adjusted to 36 L with ethanol, and 4 L of water
were charged. The solution was hydrogenated at 40 psi and
50 °C with 1.00 kg of 10% Pd on carbon as catalyst. The
reactions were complete after 16-18 h as judged by complete
consumption of starting material by HPLC assay. The catalyst
was removed by filtration of the reaction through cellulose
which was washed with ethanol. The combined filtrate and
wash was combined and assayed to contain 4.71 kg of
MK-431 free base (86% yield).
and minor), 146.6 (m, major and minor), 144.0 (q, JC-F
)
40.6 Hz, minor), 143.9 (q, JC-F ) 40.7 Hz, major), 119.3
(m, major and minor), 118.6 (m, major and minor), 117.8
(q, JC-F ) 270.1 Hz, major), 117.8 (q, JC-F ) 270.2 Hz,
minor), 106.0 (dd, JC-F ) 28.6, 21.2, major and minor), 48.4
(major and minor), 43.6 (major), 43.3 (minor), 42.0 (minor),
41.3 (major), 38.9 (major), 38.2 (minor), 33.9 (major), 33.8
(minor). [R]D ) -74.4° (c ) 1.0, H2O). Anal. Calcd for
C16H18F6N5O5P: C, 38.03; H, 3.59; O, 13.86. Found: C,
38.08; H, 3.30; N, 13.77.
Acknowledgment
The authors would like to thank Mr. A. Houck, Mr. C.
Bazaral, and Mr. A. Newell of the Merck High Pressure
Laboratory for experimental assistance. We also thank Dr.
R. Reamer, Dr. P. Dormer, and Ms. L. DiMichele for aid in
NMR structure determination.
Received for review May 24, 2005.
OP0500786
The ethanol solution was concentrated by vacuum distil-
lation to a volume of 25 L. 5.5 L of water were added, and
Vol. 9, No. 5, 2005 / Organic Process Research & Development
•
639