Development of a practical large-scale synthesis of denagliptin tosylate

Article abstract of DOI:10.1021/op900178d

A large-scale synthesis of denagliptin tosylate has been developed. The efficiency of the synthesis has been improved from the initially scaled route by changing the order of steps (performing a dehydration at a late stage). The key step of the synthesis is a single-step peptide coupling/dehydration, mediated by n-propanephosphonic acid cyclic anhydride. The challenges of developing this synthesis into a robust and practical manufacturing route are described.

Full text of DOI:10.1021/op900178d

Organic Process Research & Development 2009, 13, 900–906
Development of a Practical Large-Scale Synthesis of Denagliptin Tosylate
Daniel E. Patterson,* Jeremiah D. Powers, Michael LeBlanc, Tyler Sharkey, Emily Boehler, Erwin Irdam, and
Martin H. Osterhout
Chemical DeVelopment, GlaxoSmithKline, FiVe Moore DriVe, P.O. Box 13398, Research Triangle Park,
North Carolina 27709-3398, U.S.A.
Scheme 1. Initial route to denagliptin tosylate
Abstract:
A large-scale synthesis of denagliptin tosylate has been developed.
The efficiency of the synthesis has been improved from the initially
scaled route by changing the order of steps (performing a
dehydration at a late stage). The key step of the synthesis is a
single-step peptide coupling/dehydration, mediated by n-pro-
panephosphonic acid cyclic anhydride. The challenges of develop-
ing this synthesis into a robust and practical manufacturing route
are described.
Introduction
Denagliptin tosylate1 1 is a dipeptidyl peptidase IV (DPP-
IV) inhibitor2 in development for the treatment of type II
diabetes. For the final supply route to this compound, it was
imperative to develop a synthesis that was efficient and cost
effective. The initial supply route to this compound starting from
N-Boc-(4S)-fluoro-L-proline 2 and (S)-difluorophenyl amino
acid 4 is shown in Scheme 1. These starting materials were
initially synthesized in-house3 but were purchased externally
as part of the final supply strategy. The initial supply route was
fairly straightforward, starting with conversion of 2 to its
primary amide, followed by dehydration to the nitrile and
removal of the Boc-protecting group.1a The resulting proline
nitrile 3 was coupled to acid 4 catalyzed by O-(7-azabenzot-
riazol-1-yl)-N,N,N′,N′-tetramethyluronium hexafluorophosphate
(HATU), and deprotected in the same pot. A final recrystalli-
zation gave the active pharmaceutical ingredient (API) in 42%
overall yield from 2 in >97% purity. This route was successful
in providing the early supplies of the compound, but several
challenges had to be solved in order to develop a cost-effective
and robust route. The redevelopment of the route to meet these
challenges is the subject of this article.
The initial route was undesirable as a manufacturing process
in part because of the cost of the peptide coupling agent HATU
which was a major contributor to the overall cost of goods. In
addition, there has been increasing safety concern related to
the use of triazole-based coupling agents (i.e., HATU, O-ben-
zotriazole-N,N,N′,N′-tetramethyluronium hexafluorophosphate
(HBTU), etc.)4 for peptide coupling on large scale due to the
explosive nature of the reagents and their byproduct (>1500
J/g; maximum operating temperature approximately 50 °C).5
For these reasons, it was a priority to find an alternate coupling
agent for use in the synthesis. This was challenging because
the amine 3 has low reactivity, and in initial screens only HATU
was identified to give complete conversion to the desired
product. A number of common coupling reagents were screened
including 1,1′-carbonyl diimidazole (CDI), 1-ethyl-3-(3-dim-
ethylaminopropyl)carbodiimide (EDC), benzotriazole-1-yl-oxy-
tris-pyrrolidino-phosphonium hexafluorophosphate (PYBOP),
HBTU, and cyanuric chloride. In all cases, these coupling agents
gave either poor conversion, lower yields, and/or an increased
level of impurities in final API 1. A second screen of less
common coupling agents identified two possible replacements;
chlorodimethoxytriazine (CDMT)6 and n-propanephosphonic
acid cyclic anhydride (T3P)7,8 5 were effective for the coupling
reaction, going to completion in <4 h. The CDMT reaction,
* Corresponding author. E-mail: daniel.e.patterson@gsk.com. Telephone:
(919) 483-1266. Fax: (919) 483-3706.
(1) (a) Haffner, C. D.; McDougald, D. L.; Randhawa, A. S.; Reister, S. M.;
Lenhard, J. M. PCT Int. Appl. WO 2003/002531 A, 2003; CAN 138:
900712003. (b) Thomas, J. M.; von Geldern, T. W. Drug DeV. Res.
2006, 67, 627.
(2) (a) For recent reports, see: Hanson, R. L.; Goldberg, S. L.; Brzozowski,
D. B.; Tully, T. P.; Cazzulino, D.; Parker, W. L.; Lyngberg, O. K.;
Vu, T. C.; Wong, M. K.; Patel, R. N. AdV. Synth. Catal. 2007, 349,
1369. (b) Feng, J.; Zhang, Z.; Wallace, M. B.; Stafford, J. A.; Kaldor,
S. W.; Kassel, D. B.; Navre, M.; Shi, L.; Skene, R. J.; Asakawa, T.;
Takeuchi, K.; Xu, R.; Webb, D. R.; Gwaltney, S. L. J. Med. Chem.
2007, 50, 2297. (c) Pei, Z.; Li, X.; von Geldern, T. W.; Longnecker,
K.; Pireh, D.; Stewart, K. D.; Bakes, B. J.; Lai, C.; Lubben, T. H.;
Ballaron, S. J.; Beno, D. W. A.; Kempf-Grote, A. J.; Sham, H. L.;
Trevillyan, J. M. J. Med. Chem. 2007, 50, 1983.
(4) Dunn, P. J.; Hoffman, W.; Kang, Y.; Mitchell, J. C.; Snowden, M. J.
Org. Process Res. DeV. 2005, 9, 956.
(3) (a) Patterson, D. E.; Xie, S.; Jones, L. A.; Osterhout, M. H.; Henry,
C. G.; Roper, T. D. Org. Process Res. DeV. 2007, 11, 624. For another
approach to 4, see: Deng, G.; Ye, D.; Li, Y.; He, L.; Zhou, Y.; Wang,
J.; Li, J.; Jiang, H.; Liu, H. Tetrahedron 2008, 64, 10512. (c) For a
synthesis of 2 see: Doi, M.; Nishi, Y.; Kiritoshi, N.; Iwata, T.; Nago,
M.; Nakano, H.; Uchiyama, S.; Nakazawa, T.; Wakamiya, T.;
Kobayashi, Y. Tetrahedron 2002, 58, 8453.
(5) Data from GSK Process Safety group. Energy of decomposition
determined by DSC testing at 2 K/min. Maximum operating temper-
ature determined by ARC testing.
(6) (a) Garrett, C. E.; Jiang, X.; Prasad, K.; Repic, O. Tetrahedron Lett.
2002, 43, 4161. (b) Kaminski, Z. Synthesis 1987, 10, 917.
(7) T3P is a registered trademark of Clariant.
900
•
Vol. 13, No. 5, 2009 / Organic Process Research & Development
10.1021/op900178d CCC: $40.75  2009 American Chemical Society
Published on Web 08/20/2009

Scheme 2. Coupling using T3P
treated with ammonium bicarbonate as the ammonia source.9
The use of dichloromethane as a solvent was key to keeping
the product in solution while allowing the excess ammonium
bicarbonate to be filtered off. This procedure worked well,
giving a 90% yield, but the filtration to remove the ammonium
bicarbonate was undesirable on large scale. The reaction time
was also long, and the process was not especially robust; one
run in the pilot plant resulted in a low yield (<80%), presumably
due to crystallization of the product during the filtration of the
excess ammonium bicarbonate.
In developing a new process for the amidation, Boc-
anhydride was still used to activate the acid, but alternative
ammonia sources such as ammonia in alcohol solvents were
screened. These reactions worked well, but these reagents
introduced a large amount of excess alcohol as solvent. This
caused problems in the workup, since the product has relatively
high solubility in alcoholic solvents and therefore required
removal of a large amount of solvent to achieve a good yield
in the crystallization. Aqueous ammonium hydroxide was
examined as an ammonia source since it is available in high
concentration and is inexpensive and readily available. Initial
concerns regarding water addition to the system were not
realized, and the reaction worked exceedingly well. The acid
was treated with Boc-anhydride and pyridine in ethyl acetate
for 2-3 h, and then 28% aqueous ammonium hydroxide was
added slowly (Scheme 3). Upon addition, the product crashed
out of solution, and the reaction was complete within 1 h. The
product was isolated by partial solvent exchange to heptane
followed by filtration to afford the amide 6 in 97-99% yield.
The new procedure proceeded in higher yield and shorter
reaction time, avoided the use of DCM, and was more robust
since it avoided the need to filter off excess reagent.
Careful analysis of the amidation gave some interesting
mechanistic insight into the reaction. The initial reaction of 2
with Boc-anhydride gave the activated acid 7 as expected, but
it also formed an almost equal amount of the dimeric anhydride
8 (Scheme 4).10 Upon addition of aqueous ammonium hydrox-
ide, the two intermediates reacted very quickly to give amide
6. The reaction of 8 with ammonium hydroxide formed one
equivalent of the desired amide but also regenerated one
equivalent of the starting acid 2. In order for the reaction to
proceed to completion, the reaction of the regenerated 2 with
Boc-anhydride and subsequent conversion to 6 must be faster
than the reaction of ammonia with Boc-anhydride. The fact that
the reaction did proceed to >99% conversion under these
conditions suggests that the rates are significantly different.
Further evidence for this rate difference comes from an
Scheme 3. Stage 1 using aqueous ammonium hydroxide
however, was not as clean as that with T3P, giving higher levels
of impurities and dark color in both the reaction solution and
the product. The T3P-mediated reaction proceeded cleanly in
4 h at room temperature in ethyl acetate to give, after
deprotection with p-toluenesulfonic acid, 88% yield of dena-
gliptin tosylate in >98% purity (Scheme 2). No epimerization
of the R-chiral center was observed in the coupling reaction.
Diisopropylethylamine was used as base, and a slight excess
(1.05 equiv) of amine 3 and T3P (1.2-1.4 equiv) were
necessary for a robust and a reproducible process. The order
of addition was very important for the coupling reaction, with
the best results obtained when T3P was added last. The reaction
did not work well if the acid was first activated with T3P and
the amine subsequently added. This was due to activation of
the acid followed by reaction with a second equivalent of the
acid to form an anhydride (identified by LC-MS) which was
less reactive. In addition, the reaction worked better if the
tosylate salt was stirred for 30 min after the addition of base to
ensure complete freebasing before the T3P was added. T3P had
several advantages over HATU in the reaction. The cost of T3P
was significantly lower (>10 times), and there were lower safety
hazards associated with T3P (compared to those with HATU).
In addition, the workup of the T3P reaction was very simple,
requiring just two water washes to remove the byproducts and
give a clean solution of the desired product. This was in contrast
with many common coupling agents that require multiple
washes and/or the use of base, etc. to remove the byproducts
of the coupling agent.
A second major problem with the initial route involved the
two-step synthesis of the nitrile 3. The yield for this simple
transformation was typically only 60-65%, which severely
impacted the cost of goods and throughput of the synthesis.
When we initially developed the conversion of 2 to 6 (Scheme
3) we, recognized that an aqueous workup had to be avoided
because of the aqueous solubility of the product. This precluded
most of the typical conditions for amidation that go through
the acid chloride. Initially, the acid was activated with Boc-
anhydride and pyridine in dichloromethane (DCM) and then
(8) (a) Wissmann, H.; Kleiner, H. J. Angew. Chem., Int. Ed. Engl. 1980,
19, 133. (b) Boggs, S. D.; Cobb, J. D.; Gudmundsson, K. S.; Jones,
L. J.; Matsuoka, R. T.; Millar, A.; Patterson, D. E.; Samano, V.; Trone,
M. D.; Xie, S.; Zhou, X.-M. Org. Process Res. DeV. 2007, 11, 539.
(c) Helal, C. J.; Kang, Z.; Lucas, J. C.; Bohall, B. R. Org. Lett. 2004,
6, 1853.
(9) Pozdnev, V. F. Tetrahedron Lett. 1995, 36, 7115.
(10) The identity of the intermediates 7 and 8 are based on LC-MS
analysis.
Vol. 13, No. 5, 2009 / Organic Process Research & Development
•
901

Scheme 4. Mechanism of stage 1
Scheme 5. Synthesis of 9
Scheme 6. Synthesis of 11 by one-pot coupling/dehydration
experiment where the aqueous ammonium hydroxide was
present prior to the Boc-anhydride addition. Under these
conditions, the reaction went to >80% completion, indicating
that the formation of the reactive intermediate 7 was faster than
decomposition of Boc-anhydride by reaction with ammonia.
The conversion of 6 to 3 (stage 2) by treatment with
trifluoroacetic anhydride to effect dehydration followed by
deprotection with p- toluenesulfonic acid was problematic due
to its low yield. The low yield was due to the difficulty in
crystallizing the final product which led to a large yield loss in
the mother liquors. Some acid-catalyzed decomposition to
unknown byproduct was also observed during the deprotection
step. A large effort was made to optimize this reaction by
looking at alternate reagents, solvents, and crystallization
conditions, but in our hands, it was not possible to improve the
yield above 70%.
The improvements to the peptide coupling and amidation
steps detailed above represented major improvements from the
initial synthesis. The overall yield was improved from 42% to
55%, but more importantly, the cost of goods and robustness
of the process were greatly improved. This route was used to
supply phase II and to make the initial phase III clinical supplies
(>1 t of API). It was still a goal of the project team to improve
the overall yield of the synthesis, which led us to investigate a
new route. Because the project was at a late stage of develop-
ment, the new synthesis would have to provide material that
matched or had an improved impurity profile compared to that
of the current route. It was preferred that the route use the same
starting materials, since reliable suppliers of both starting
materials had already been established. Therefore, it was decided
to change the order of steps to perform the coupling before the
dehydration in hopes of avoiding the issues related to dehydra-
tion at an early stage. It was hoped that the amino-amide
intermediate 9 (Scheme 5) would be easier to isolate and could
be synthesized in higher yield than the amino-nitrile 3. The risk
of this approach, based on previous experience, was that the
dehydration to the nitrile would be done late in the synthesis.
Since the conditions for dehydration are often harsh, there was
a risk of decomposition or formation of new impurities during
this late stage dehydration.
amino-amide 9 which precipitated as its tosylate salt and could
be filtered directly in high yield (Scheme 5).
The conditions used for the coupling in the initial route
translated nicely to the coupling of amino amide 9 with Boc-
protected amino acid 4. Treatment with T3P and diisopropyl-
ethylamine in ethyl acetate led to complete conversion to the
desired coupled product 10 with no epimerization. The develop-
ment of the dehydration proved to be more challenging. We
screened a number of common dehydrating agents including
phosphorous oxychloride, trifluoroacetic anhydride, tosyl chlo-
ride, oxalyl chloride, thionyl chloride, and cyanuric chloride,
but all formed impurities or did not go to completion. T3P and
diisopropylethylamine were examined for the dehydration, but
even at 50 °C with a large excess of reagent, the reaction did
not go to completion.11 p-Toluenesulfonic anhydride appeared
to be a viable way forward, but our inability to find a large-
scale supplier of this reagent precluded its use. We finally had
success with methanesulfonic anhydride as the dehydrating
agent. The transformation proceeded very cleanly with pyridine
as the base at 50 °C, giving 11 in 90% yield (see Scheme 6).
This material was taken on to give API with an improved
impurity profile. A process was developed for the coupling
/dehydration whereby after the T3P coupling, the reaction was
washed with water twice, and dried by atmospheric distillation
to <0.1 wt %/vol water. Pyridine and methanesulfonic anhydride
were added, and the reaction was heated to 55 °C for 1 h. After
workup and three washes, the product was crystallized from
isopropanol/water to give 11 in 90% yield. This process was
successfully demonstrated on 150 kg scale, giving material of
acceptable purity.
During development, it was observed that if the T3P
coupling was run at high temperature, a significant amount of
dehydration of the product to form 11 was also observed. This
led us to reexamine T3P for the dehydration step. To our
surprise, by raising the temperature and lowering the amount
Stage 1 of the new route was identical to stage 1 of the initial
route and was not changed. Stage 2 was a simple deprotection
of Boc-amide 6 to give fluoroprolinamide 9. Multiple acids and
solvents were screened, but the best conditions were p-
toluenesulfonic acid in ethyl acetate. These conditions gave
(11) (a) Meudt, A.; Wisdom, R. Chim. Oggi. 2008, 26, 28. (b) Garcia,
A. L. L. Synlett 2007, 8, 1328. (c) Meudt, A.; Scherer, S.; Nerdinger,
S. PCT Int. Appl. WO 2005/070879 A1; CAN 143:172649, 2005.
902
•
Vol. 13, No. 5, 2009 / Organic Process Research & Development

Scheme 7. Coupling/dehydration mediated by T3P
Table 1. Degradation of 1 under drying conditions (16 h 55
°C)
conditions
12A (%) 12B (%) 12C (%)
1 washed wet cake
1 unwashed wet cake
-
-
-
-
-
3.7
1 + IPA (20 wt %)
0.02
0.05
3.4
-
-
1 + H2O (10 wt %)
-
-
1 + TsOH (5 wt %)
-
-
Scheme 8. Synthesis of denagliptin tosylate (intermediate
grade)
1 + TsOH (5 wt %), IPA (20 wt %)
1.2
1.2
0.3
The final step of the synthesis was a recrystallization from
isopropanol and water. A recrystallization was performed for
two main reasons. First, the use of a separate, final-stage
crystallization led to a more reproducible impurity profile and
consistent physical properties of the final API. In addition, a
separate crystallization was needed to control the level of
isopropyl tosylate, a potential genotoxic impurity, in the API.
Isopropyl tosylate was formed in the Boc-deprotection step
when p-toluenesulfonic acid was heated in isopropanol. The
levels in the reaction solution could be as high as 5000 ppm,
and the isolated crude API typically contained 20-60 ppm. This
was not an issue, as any isopropyl tosylate was readily
decomposed in isopropanol/water mixtures in the final crystal-
lization. The final crystallization was a straightforward cooling
crystallization. One issue with this process was that at high
temperature, in isopropanol/water mixtures, the API decomposes
by cyclization of the amine onto the nitrile to give amidine 13
(and/or) diketopiperazine 14 impurities (Scheme 9).
The initial recrystallization process was run by dissolving
the API in 10 volumes of a 2:3 IPA/water mixture and heating
to 80 °C to dissolve. However, in stability studies of the reaction
mixture, it was observed that at 80 °C the decomposition to
the amidine impurities was rapid. As the process was scaled
up, it was recognized that the batch would be held at high
temperature for extended periods of time as the reaction was
heated, filtered while hot, and then cooled back down. The
stability of 1 at various temperatures is shown in Figure 1. The
data indicate that decomposition was significantly slower at 70
°C and negligible at temperatures under 60 °C. On the basis of
these data, a lower temperature recrystallization was desired.
Solubility studies of denagliptin tosylate in varying ratios
of isopropanol and water showed that by lowering the initial
concentration of water in the solvent system it was possible to
dissolve the material at 70 °C without increasing the total
volumes above 10 (Figure 2). In practice, the crystallization
was run in 10 volumes of 1:1 isopropanol/water. The slurry
was heated to 70-72 °C to dissolve, filtered while hot, and
then cooled to 5 °C. At 5 °C an additional 3 volumes of water
was added to push more material out of solution, giving an
isolated yield of 92%. By changing to a lower temperature
process, the risk of degradation was mitigated.
of base, it was possible to do both the coupling and the
dehydration in one pot with T3P as the reagent (Scheme 7). In
practice, the coupling was run as previously, and after the
reaction was complete, a second portion of T3P was added,
and the reaction was heated to 78 °C. This led to complete
conversion to the desired product in 2 h, and crystallization gave
the desired product in 89% yield (>99% purity). This process
had an advantage in cycle time over the methanesulfonic
anhydride process in that it did not require aqueous washes or
an azeotropic drying distillation before the dehydration step.
In addition, the removal of methanesulfonic anhydride from
the process was advantageous from a genotoxic impurity
perspective since, when using methanesulfonic anhydride, there
is the possibility of forming alkyl methanesulfonate esters,
especially since alcoholic solvents are used in subsequent steps.
Switching to the T3P process removed the potential to form
these genotoxic impurities.
The final two steps of the synthesis were similar to the initial
route. Deprotection of 11 by treatment with p-toluenesulfonic
acid in isopropanol proceeded smoothly to give 1 in 97% yield
(Scheme 8). In order to control the gas evolution, a solution of
p-toluenesulfonic acid was added to a solution of 11 in
isopropanol at 70 °C over 1 h. The reaction was stirred for an
additional 6 h at 70 °C until completion. In some plant-scale
batches, three impurities (12A-C, Scheme 8) were observed
in the isolated product, that were not present in typical lab-
scale batches. In addition, some of these batches had different
levels of these three impurities in different spins of the same
batch (batches where the reaction was run in a single batch,
but were filtered in a centrifuge and dried in portions).
Studies of the Boc-deprotection showed that none of these
impurities were formed during the reaction, but were formed
in the dryer. Spiking studies showed that these impurities only
formed if there was excess p-toluenesulfonic acid in the filter
cake during drying (Table 1). Interestingly, these impurities only
form in the solid phase, and heating 1 in solution with
p-toluenesulfonic acid failed to give these impurities under a
variety of conditions.
The final manufacturing route to denagliptin tosylate is
shown in Scheme 10. This route is notable for its high yield
(>70% over five steps), and excellent scalability and operability.
The new route changed the order of steps to effect dehydration
at a late stage, which solved the major problems associated with
the original route. This route overcame the low yield and
stability issues related to the synthesis of nitrile 3. In addition,
T3P was identified as a cost-effective, safe alternative to HATU,
greatly improving the overall cost of goods of the process. By
The instability of 1 to drying in the presence of excess
p-toluenesulfonic acid highlighted the need for efficient washing
of the filter cake prior to drying. In practice, washing the cake
twice in a filter dryer with two volumes of isopropanol was
sufficient to remove the free toluenesulfonic acid from the cake
and avoid impurity formation.
Vol. 13, No. 5, 2009 / Organic Process Research & Development
•
903

Scheme 9. High temperature decomposition of denagliptin tosylate
using T3P for both the coupling and the dehydration, it was
possible to perform these two operations very cleanly in a single
stage. This transformation underscores the utility of T3P for
use on large scale. The development of the new route also
removed dichloromethane from the synthesis and removed two
salt filtrations from the process, improving the safety and
operability of the process. This route proved to be very robust
on scale and was used to produce more than a metric ton of
API.
(4S)-1-(tert-Butoxycarbonyl)-4-fluoro-L-prolinamide (6).
A mixture of 2 (100 kg, 429 mol), ethyl acetate (1000 L, 10
vol), and di-tert-butyl dicarbonate (117 L, 511 mol) was treated
with pyridine (42 L, 523 mol) and was stirred at 25 °C for
2-3 h. A solution of 28-30% w/w ammonium hydroxide (35
L, 528 mol) was charged over at least 20 min. (NOTE: Strong
off-gassing observed!) The reaction was stirred at 25 °C for
3 h. The solvent was removed by vacuum distillation (50-100
mbar) to 400 L (4 vol). Heptane (900 L, 9 vol) was charged,
and solvent was removed by vacuum distillation (50-100 mbar)
to 800 L (8 vol). The reaction was cooled to 5 °C and held for
30 min. The batch was filtered, washed with heptane (200 L, 2
vol), and dried at 75 °C to give 95.6 kg (96%) of 6 as a white
solid. Mp: 161 - 164 °C. 1H NMR (400 MHz, DMSO-d6) δ
7.20 (m, 2H), 5.22 (d, J ) 54 Hz, 1H), 4.13 (m, 1H), 3.60 (m,
2H), 2.39 (m, 1H), 2.16 (m, 1H), 1.39(s, 9H). 13C NMR (100
MHz, DMSO-d6) δ 173.8, 154.1, 92.0 (d, J ) 173 Hz), 79.4,
59.1, 53.6, 37.6, 28.5. Anal. Calcd For C10H17FN2O4: C, 51.72;
H, 7.38; N, 12.06. Found: C, 51.58; H, 7.42; N, 12.08.
(4S)-Fluoro-(2S)-cyanopyrrolidine p-Toluenesulfonate (3).
A mixture of isopropyl acetate (250 L, 5 vol), 6 (50 kg, 215
mol) and pyridine (38.3 L, 474 mol) was treated with trifluo-
roacetic anhydride (33.5 L, 237 mol). The reaction was stirred
at 20 °C for 1 h and filtered, and the filtercake was washed
with isopropyl acetate (100 L, 2 vol). To the filtrate was added
water (100 L, 2 vol), and the mixture was stirred for 15 min
and allowed to settle. The layers were separated, and the
aqueous layer was back-extracted with isopropyl acetate (100
L, 2 vol). The combined organic layer was concentrated under
vacuum (50-100 mbar) to 150 L (3 vol), and was treated with
p-toluenesulfonic acid monohydrate (82 kg, 430 mol). The
reaction was stirred at 20 °C for 6 h, and isopropanol (150 L,
3 vol) was added. The reaction was cooled to 5 °C over 30
min, filtered, washed with isopropyl acetate (50 L, 1 vol), and
dried at 50 °C to give 38.2 kg (62%) of 3 as a white solid. Mp:
156-157 °C. 1H NMR (400 MHz, DMSO-d6) δ 10.2 (m, 2H),
7.53 (d, J ) 8.2 Hz, 2H), 7.15 (d, J ) 8.2 Hz, 2H), 5.53 (d, J
) 52 Hz, 1H), 5.02 (m, 1H), 3.59 (m, 2H), 2.62 (m, 2H), 2.30
(s, 3H). 13C NMR (100 MHz, DMSO-d6) δ 145.4, 138.6, 128.7,
126.0, 116.8, 92.3 (d, J ) 175 Hz), 52.5, 45.7, 37.5, 21.3. Anal.
Calcd For C12H15FN2O3S: C, 50.34; H, 5.28; N, 9.78. Found:
C, 50.33; H, 5.29; N, 9.72.
Experimental Section
General Methods. HPLC purity was determined on a
Hewlett-Packard series 1100 system using Phenomenex Luna
C18 (2) columns (50 mm × 2 mm, 3 µm), and a mixture of
water and acetonitrile with trifluoroacetic acid as mobile phase
(gradient at a flow rate of 1.0 mL/min and UV detector at 220
nm).
Figure 1. Stability of denagliptin tosylate in isopropanol/water.
(4S)-Fluoro-L-prolinamide p-Toluenesulfonate (9). A mix-
ture of p-toluenesulfonic acid monohydrate (110.7 kg, 581 mol)
and ethyl acetate (270 L, 3 vol), was heated to 35 °C until all
of the solids dissolved. This solution was added over 60 min
to a solution of 6 (90 kg, 387 mol) in ethyl acetate (270 L, 3
vol) at 50 °C. The reaction was stirred at 50 °C for 2-3 h. The
batch was cooled to 5 °C and stirred for 30 min. The reaction
Figure 2. Solubility of denagliptin tosylate in IPA/water.
904
•
Vol. 13, No. 5, 2009 / Organic Process Research & Development

Scheme 10. Final manufacturing route to denagliptin tosylate
was filtered, washed with ethyl acetate (90 L, 1 vol), and dried
with 7 (94.3 kg, 309 mol). The reaction was stirred for 30 min
at room temperature, and 50% wt T3P in EtOAc (270.2 kg,
422 mol) was added while keeping the batch temperature less
than 50 °C. The reaction was stirred at 40-50 °C for 30 min,
and then a second portion of 50 w/w % T3P in EtOAc (270.2
kg, 422 mol) was added. The reaction was heated to 75-80
°C and held for 6-12 h. The reaction was cooled to 25 °C.
Water (530 L, 5 vol) was added, and the reaction was stirred
15 min. The stirring was stopped, the layers were allowed to
separate, and the bottom layer was removed. Water (318 L, 3
vol) was added, and the reaction was stirred 15 min. The stirring
was stopped, the layers were allowed to separate, and the bottom
layer was removed. Solvent was removed by atmospheric
distillation to approximately 3 volumes. Isopropanol (424 L, 4
vol) was added, and solvent was removed by atmospheric
distillation to approximately 3 volumes. Isopropanol (424 L, 4
vol) was added, and solvent was removed by atmospheric
distillation to 5 volumes. The batch was cooled to 65-75 °C,
and water (265 L, 2.5 vol) was charged into the batch at a
constant rate over approximately 60 min, keeping the temper-
ature at 65-75 °C. The batch was seeded at 70 °C with 0.005
wt of 1. The batch was held at 70 °C for at least 30 min. Water
(424 L, 4 vol) was charged into the batch at a constant rate
over at least 60 min, keeping the temperature at 65-75 °C.
The batch was cooled to 20 at 0.5 °C/min and held for 15
min. The batch was filtered and washed with 35 v/v%
isopropanol in water (318 L, 3 vol). The solids were dried
at 75 °C to give 123 kg (97%) of 11 as a white solid.
at 50 °C to give 112 kg (95%) of 9 as a white solid. Mp:
1
190-193 °C. H NMR (500 MHz, DMSO-d16) δ 9.10 (bs,
2H), 8.00 (s, 1H), 7.72 (s, 1H), 7.50 (d, J ) 8.3 Hz, 2H), 7.13
(d, J ) 8.3 Hz, 2H), 5.38 (dt, J ) 52.3, 3.7 Hz, 1H), 4.35 (dd,
J ) 10.7, 3.4 Hz, 1H), 3.58 (ddd, J ) 19.5, 13.7, 2.0 Hz, 1H),
3.44 (t, J ) 37.1, 13.2, 3.7 Hz, 1H), 2.62 (m, 1H), 2.31 (m,
1H), 2.28 (s, 3H). 13C NMR (100 MHz, DMSO-d6) δ 169.9,
145.7, 138.4, 129.3, 128.6, 126.0, 93.2, 92.4 (d, J ) 174 Hz),
58.0 (d, J ) 80 Hz), 52.1 (d, J ) 23 Hz), 37.2, 21.3. Anal.
Calcd For C12H17FN2O4: C, 47.36; H, 5.63; N, 9.20. Found: C,
47.39; H, 5.63; N, 9.16.
Initial Synthesis of (2S)-Cyano-(4S)-fluoro-1-pyrrolidi-
nyl]-2-oxoethylcarbamate p-Toluenesulfonate (Denagliptin
Tosylate) (1). A mixture of 4 (100 kg, 265 mol) and 3 (78 kg,
272 mol) in ethyl acetate (700 L, 7 vol) at 0 °C was treated
with Hunig’s base (153 L, 875 mol), and the reaction was stirred
for 30 min at 0 °C. T3P in EtOAc (50 w/w%, 221 L, 371 mol)
was added, keeping the reaction temperature below 5 °C. The
reaction was stirred at 0 °C for 20 min, then heated to 20 °C
and stirred for 4 h. The reactor was charged with water (600 L,
6 vol), and the mixture was stirred for 10 min. The layers were
separated, and the organic layer was washed with water (600
L, 6 vol). The ethyl acetate solution of 8 was solvent exchanged
under vacuum (50-100 mbar) to isopropanol (600 L final
volume, 6 vol). To this slurry was added a solution of
p-toluenesulfonic acid monohydrate (101 kg, 530 mol) in
isopropanol (400 L, 4 vol). The solution was heated to 60 °C
for 6 h. The reaction was cooled to 5 °C and stirred at this
temperature for 30 min. The reaction was filtered, washed with
isopropanol (200 L, 2 vol), and dried at 50 °C to give 127 kg
(88%) of 1 as a white solid. Mp: 241-245 °C dec. 1H NMR
(500 MHz, DMSO-d6) δ 8.36 (bs, 3H), 7.73 (m, 2H), 7.51 (d,
J ) 8.1 Hz, 2H), 7.36 (m, 2H), 7.24 (m, 2H), 7.14 (d, J ) 8.1,
2H), 7.05 (m, 2H), 5.34 (d, J ) 53.0 Hz, 1H), 4.91 (m, 2H),
4.45 (d, J ) 11.4 Hz, 1H), 3.84 (ddd, J ) 39.2, 12.5, 3.4, 1H),
3.39 (dd, J ) 24.2, 12.5, 1H), 2.62 (m, 1H), 2.34 (m, 1H),
2.30 (s, 3H), 2.24 (m, 1H). 13C NMR (100 MHz, DMSO-d6) δ
167.8, 162.2 (d, J ) 244.5 Hz), 162.0 (d, J ) 244.5 Hz), 145.7,
138.4, 134.7, 134.3, 131.6, 131.5, 130.6, 130.5, 128.6, 126.0,
117.8, 116.3, 116.1, 115.9, 92.0 (d, J ) 175.1) 53.8, 53.7, 53.6,
52.5, 45.1, 35.8, 21.2. Anal. Calcd For C27H26F3N3O4S: C,
59.44; H, 4.80; N, 7.70. Found: C, 59.52; H, 4.85; N, 7.70.
tert-Butyl{(1S)-1-[bis(4-fluorophenyl)methyl]-2-[(2S)-cy-
ano-(4S)-fluoro-1-pyrrolidinyl]-2-oxoethylcarbamate (11). A
mixture of diisopropylethylamine (145.1 kg, 1124 mol) and 4
(106 kg, 281 mol) in ethyl acetate (424 L, 4 vol), was treated
1
Mp: 205-208 °C dec. H NMR (500 MHz, DMSO-d6) δ
7.47 (m, 2H), 7.30 (dd, J ) 8.7, 5.5 Hz, 2H), 7.13 (t, J )
8.9 Hz, 2H), 7.00 (t, J ) 8.9 Hz, 2H), 5.37 (d, J ) 52
Hz, 1H), 5.02 (dd, J ) 11.7, 8.7 Hz, 1H), 4.87 (d, J )
9.0 Hz, 1H), 4.43 (d, J ) 11.7 Hz, 1H), 3.86 (ddd, J )
39.3, 12.6, 3.4 Hz, 1H), 3.7 (dd, J ) 24.2, 12.6 Hz, 1H),
2.41 (m, 1H), 2.31 (m, 1H), 1.28 (s, 9H). 13C NMR (100
MHz, DMSO-d6) δ 170.1, 162.0, 160.0, 155.3, 136.9,
136.5, 130.6, 130.0, 117.8, 115.2, 114.8, 92.6 (d, J ) 140
Hz), 78.4, 54.6, 52.9, 51.2, 44.2, 35.4, 28.8. Anal. Calcd
For C25H26F3N3O3: C, 63.42; H, 5.53; N, 8.87. Found: C,
63.45;H, 5.62; N, 8.89.
Denagliptin Tosylate (1). To a mixture of 11 (110 kg,
232 mmol) in isopropanol (550 L, 5 vol) at 70 °C was
added a solution of p-toluenesulfonic acid monohydrate
(88.4 kg, 464 mol) in isopropanol (550 L, 5 vol) over
one hour while maintaining the temperature at 70 °C. After
the addition, the reaction was stirred at 70 °C for 6 h.
The batch was cooled to 20 °C, held for 30 min, filtered,
Vol. 13, No. 5, 2009 / Organic Process Research & Development
•
905

and washed with isopropanol (2 × 220 L, 2 vol). The
solids were dried at 55 °C to give 118 kg (89%) of 1 as
a white solid.
pulled dry. The solids were dried at 55 °C to give 91.9
kg (92%) of 1 as a white solid.
Recrystallization of Denagliptin Tosylate (1). A mixture
of denagliptin tosylate (100 kg, 183 mol) and isopropanol
(500 L, 5 vol) and water (500 L, 5 vol), was heated until
all the solids dissolved (approximately 72 °C). The hot
solution was filtered into another vessel. The solution was
cooled to approximately 5 °C, and water (300 L, 3 vol)
was added. The reaction was stirred at this temperature
for 30 min and was filtered. The filtercake was washed
with filtered isopropanol (2 × 200 L, 2 × 2 vol), and
Supporting Information Available
Copies of the NMR spectra of all intermediates and the final
product. This material is available free of charge via the Internet
at http://pubs.acs.org.
Received for review July 13, 2009.
OP900178D
906
•
Vol. 13, No. 5, 2009 / Organic Process Research & Development