A novel scalable process to the GSK3β inhibitor AZD8926 based on a heterocyclic ziegler coupling

Article abstract of DOI:10.1021/op300365e

Development of a new, safe, and scalable route to the GSK3β inhibitor, AZD8926, is presented. In brief, the process constitutes of (i) a synthesis of 1-(pyran-4-yl)-2-trifluoromethyl-imidazole, 14; (ii) a Ziegler-type coupling of lithiated 14 with commercially available 2-chloro-5-fluoropyrimidine via 1,2-addition over the 3,4-C-N bond; (iii) a copper-catalyzed dehydrogenative aromatization using oxygen as the stoichiometric oxidant; and (iv) an aromatic C-N bond formation using either a Buchwald-Hartwig coupling or an acid-catalyzed amination. This process circumvents the main issue in the early-phase route, in which serious process safety constraints were associated with the hazardous properties of the structure, formation, and reduction of 5-methyl-4- nitroisoxazole, 2 (4200 J/g). The new process has been demonstrated on a multigram, 2-L scale. The overall yield was improved from 4 to 14%, and the number of steps decreased from 12 to 10.

Full text of DOI:10.1021/op300365e

Article
pubs.acs.org/OPRD
A Novel Scalable Process to the GSK3β Inhibitor AZD8926 Based on a
Heterocyclic Ziegler Coupling
Anette Witt,*,∥ Peter Teodorovic, Mats Linderberg, Peter Johansson, and Anna Minidis*,⊥
Pharmaceutical Development R&D, Chemical Science, AstraZeneca, S-151 85 Sodertalje, Sweden
̈
̈
ABSTRACT: Development of a new, safe, and scalable route to the GSK3β inhibitor, AZD8926, is presented. In brief, the
process constitutes of (i) a synthesis of 1-(pyran-4-yl)-2-trifluoromethyl-imidazole, 14; (ii) a Ziegler-type coupling of lithiated 14
with commercially available 2-chloro-5-fluoropyrimidine via 1,2-addition over the 3,4-C−N bond; (iii) a copper-catalyzed
dehydrogenative aromatization using oxygen as the stoichiometric oxidant; and (iv) an aromatic C−N bond formation using
either a Buchwald−Hartwig coupling or an acid-catalyzed amination. This process circumvents the main issue in the early-phase
route, in which serious process safety constraints were associated with the hazardous properties of the structure, formation, and
reduction of 5-methyl-4-nitroisoxazole, 2 (4200 J/g). The new process has been demonstrated on a multigram, 2-L scale. The
overall yield was improved from 4 to 14%, and the number of steps decreased from 12 to 10.
to avoid expensive fluorination reagents. In the search for
suitable coupling partners, the 2-CF3-imidazole 14 (Scheme 2)
was identified, but up to this point compound 14 was unknown
in the literature. Herein we report the development of a
scalable route to the key intermediate 14 and thereafter the
development of a new scalable process to AZD8926 (9).
INTRODUCTION
■
AZD8926 (9) is a potent glycogen synthase kinase-3β
(GSK3β) inhibitor which has potential for treating several
CNS disorders, such as Alzheimer’s disease (AD), schizo-
phrenia, and chronic as well as acute neurodegenerative
diseases.1 When 9 entered into clinical development phase, a
closer evaluation of process safety and scale-up efficiency
factors was performed. Initial scale-up work for clinical material
utilized the original medicinal chemistry route2 (Scheme 1),
and 7.9 kg of the active pharmaceutical ingredient (API) was
produced in approximately 4−6% overall yield after optimiza-
tion. Several issues were identified as impediments to long-term
use of this route:
RESULTS AND DISCUSSION
■
Coupling Partner, 1-(Pyran-4-yl)-2-trifluoromethyl-
imidazole, 14. In order to utilize the proposed new route
using 2-CF3-imidazole, 14, as a building block (Scheme 2), a
new methodology for the preparation of such compounds had
to be developed because the existing methods were not
applicable to large-scale synthesis due to the use of either
photochemical or expensive methods or environmentally
unfriendly reagents.3 In the first step of the synthesis ethyl
trifluoroacetate and the amine 10 using triethylamine (TEA)
and toluene as solvent gave amide 11. The first approach to
prepare the chloroimine 12 adopted carbon tetrachloride
(CCl4) and THF as a solvent combination together with
triphenylphosphine (TPP) and the amide 11.4 Both TPP and
CCl4 are undesirable and not a long-term solution. Several
classical methods to form the chloroimine 12 from the amide
11 were then tested; however, often complex mixtures were
obtained, or no reaction occurred. The best result was achieved
using triphosgene and TEA. An undesired byproduct, ethyl
carbamoyl chloride, was detected which is a known possible
side reaction when using TEA with triphosgene.5
• Problems with the high energy of substances 2 (4200 J/
g) and 3 (1500 J/g) as well as problems with desired
reactions, including low onset temperatures and
accumulation issues.
• Low yield in the synthesis of the enaminone 5 (30−
36%).
• The fluorination of the enaminone 5 generated an
unstable intermediate, and before the reaction was
complete it was necessary to proceed with guanidine
cyclization to 6.
• The Selectfluor reagent is expensive, and despite
extensive optimization, the isolated yield after fluorina-
tion and cyclization to the pyrimidine 6 was only 50−
55%.
• The use of palladium (Pd) catalysis in the final step
required rework to reduce residual levels of Pd in the
API.
The byproduct formation was reduced when a solution of
triphosgene was charged over several hours at 50 °C to a
mixture of the amide 11 and TEA in a nonpolar solvent.
Toluene was the most nonpolar solvent used due to the
solubility profile of the amide 11. A variety of bases other than
These factors made the route unsuitable for further scale-up
and long-term supply of API. A large literature survey was
made, and our retrosynthetic analysis of 9 was directed towards
earlier introduction of cross-coupling chemistry and con-
struction of the imidazole moiety in a different way to avoid
intermediates with high energy content. Furthermore 5-fluoro-
substituted pyrimidines, commercially available in multikilo-
gram quantities, were investigated as potential starting materials
TEA were tested, such as Hunigs base, carbonates, and
̈
pyridine, without good results. The chloroimine 12 was treated
Received: December 20, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op300365e | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
telescoped synthetic steps as a sugar-like solid in an overall
yield of 34% from compound 10. The method was rapidly
scaled to approximately 500 g in a 5-L reactor, and no more
optimization was performed.
Scheme 1. Optimized medicinal chemistry route to
AZD8926 (9)
Cross-Coupling Investigations. With the key intermedi-
ate 14 in hand, a number of cross-coupling methods were
considered, involving either the 5H-imidazole using C−H
activation conditions7 or a 5-metal-imidazole as an intermediate
applying Suzuki or Negishi conditions.8 One of the synthetic
strategies was to use 2,4-dichloro-5-fluoropyrimidine (DCFP)
as starting material, which is commercially available in
multikilogram quantities. A screen of catalysts for the C−H
activation coupling between compound 14 and DCFP were
conducted in a few solvents (DMF, toluene, and 2-propanol)
and with a variety of bases (Cs2CO3, KOAc, t-BuONa, and
N,N-dicyclohexylmethylamine). All the work was performed in
a glovebox, but the desired compound 16 (Scheme 3) was
Scheme 3. New route to AZD8926 (9) from imidazole 14
Scheme 2. Preparation of imidazole 14 from
tetrahydropyran 10
never detected. The Negishi coupling was investigated in a
series of experiments using THF as solvent. A large variety of
catalysts were tested, and with one exception the reactions gave
poor or moderate conversion to the desired compound 16. The
only interesting result, with a conversion of 69%, was achieved
when the lithiated compound 14 was first reacted with I2 and
the formed 5-iodo-imidazole was treated with Rieke-zinc9 to
form imidazol-5-yl zinc iodide and thereafter coupled with
DCFP using Pd(dba)2 as the catalyst.
Fortunately, Bursavich and co-workers reported10a
a
convenient synthesis to diverse 2-amino-4-heteroarylpyrimi-
dines via a 2-chloropyrimidine intermediate, and our second
synthetic strategy generated compound 16 via this Ziegler-type
reaction.10 Nucleophilic attack of in situ lithiated 14 on
commercially available 2-chloro-5-fluoropyrimidine (CFP)
afforded compound 16, in reasonable yield, after subsequent
oxidative workup (Scheme 3). It was decided to focus our
efforts on the Ziegler reaction as a possible strategy for
multikilogram production.
with aminoacetaldehyde dimethyl acetal in toluene at room
temperature to form compound 13, and upon treatment with
sulfuric acid (H2SO4) in THF at room temperature the
imidazole 14 was synthesized. Use of aminoacetals and different
acids to form imidazoles from chloroimines is known in the
literature.6 The imidazole 14 (Scheme 2) was isolated after four
Lithiation and Addition. In order to achieve a process
fitted for a wider range of plants, it was desired to avoid the
B
dx.doi.org/10.1021/op300365e | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
need for cryogenic equipment during the lithiation of 14. This
could be achieved with a good temperature range between −20
to −5 °C which gave a stable conversion of >95%. Only THF
and 2-methyl-tetrahydrofuran (2-MeTHF) were investigated as
solvents. The use of 2-MeTHF gave a homogeneous solution
during lithiation compared to THF which gave a slurry;
however, regarding byproduct 18 formation,11 THF gave a
more selective addition of 5-Li-imidazole to CFP to form the
intermediate 15. Due to the inhomogeneous reaction when
using THF it was noted that the in-process control (IPC) by
LC−MS varied if care was not taken during sampling of the
reaction mixture. This approach to obtain a representative
sample was considered unrobust for the pilot plant. Therefore
an in-line mid-IR method was developed to follow the course of
the lithiation reaction by measuring the disappearance of 14.
Hexyl lithium was chosen as lithiating agent for safety and
economical reasons in the pilot plant, but n-butyl lithium gave
very similar results. A slight excess of the hexyl lithium (1.05−
1.1 equiv) was added, followed by a controlled addition of CFP
(1.05−1.1 equiv) dissolved in THF. The reaction temperature
was maintained below −15 °C during the addition to obtain a
high conversion. The reaction was very fast, and the conversion
was complete within 15 min, but a prolonged reaction time did
not lead to an increased level of byproduct 18.11
Oxidative Dehydrogenation. The formation of the
dihydropyrimidine 15 and the subsequent aromatization to
16 was initially telescoped using benzoquinone as oxidating
agent. The reaction worked well on a 10-g scale, but the
extractive workup turned out to be very difficult, since both
phases were almost black. Also, loss of fluoride was a competing
side reaction giving up to 30% of compound 18.11 Air-oxidation
of 15 in THF was also found to work reasonably well in the
laboratory but was at first considered unsuitable to scale up
from a process safety perspective. First, quenching the reaction
mixture containing compound 15 with acetic acid (HOAc)
followed by an extractive workup before the oxidation gave less
byproduct 18 (2−4%), and second, a good phase separation
was obtained. Alternative oxidating agents (NMO, H2O2,
cumene peroxide, and tert-butylperoxide) were investigated
with varied results. Yamamoto and co-workers have reported a
mild, practical procedure for oxidative dehydrogenation with a
catalytic amount of a Cu salt, K2CO3, and tert-butylhydroper-
oxide.12 Therefore, H2O2 in combination with Cu(OAc)2 as
catalyst was investigated, and promising results were obtained.
Furthermore, the dehydrogenation of 15 required addition of
TEA to proceed at a reasonable rate. However, the reaction was
not robust, and decomposition of H2O2 by the catalyst was
observed. It was decided to proceed with the work using air-
oxidation, with Cu(OAc)2 as the catalyst, but in another solvent
other than THF due to the intrinsic risk of peroxide formation.
The reaction worked best in acetonitrile (CH3CN), giving 93−
97% conversion of 15 to 16 and 60% isolated yield over three
complex steps starting from 14. The following procedure was
developed. The reaction mixture of 3-Li-15 was quenched with
HOAc in ethylacetate (EtOAc) at −10 to 10 °C, and then
water was added. After phase separation and solvent swap to
CH3CN, Cu(OAc)2 (0.05 equiv) and TEA (1 equiv) were
added. The solution was cooled to 10 °C, giving a slurry. A
mixture of 5% O2 and 95% N2, which unlike air was acceptable
in our pilot plant from a process safety perspective, was
bubbled through the slurry which gradually turned into a
solution. After 24 h (92% conversion) the solution was
concentrated under reduced pressure. To remove copper from
the product, a solution of 5% NH3 (aq) was added (effective
removal down to <50 ppm, without ammonia >2500 ppm).
The mixture was cooled to 0 °C, precipitating 16 with an
HPLC purity of 99% on a 120-g scale.
Palladium-Catalyzed Amination. To avoid the potential
problems with using palladium in the final step and risking
contamination of the API, one of the alternatives was to
perform a Pd-catalyzed coupling of 16 with a p-aminobenzoic
ester and thereafter transform the ester to an amide as the final
step. For solubility reasons, 4-aminobutyl- and ethyl-benzoate
were chosen for the screening experiments. THF was used as
solvent for the reactions. Various combinations of ligands, Pd-
sources, and bases were investigated. Out of the four 2-
(dialkyl)phosphinobiphenyl ligands tested, SPhos in combina-
tion with Pd(OAc)2 and K2CO3 gave 100% conversion of 16 to
17 within 3 h. The same conversion was achieved using (R)-
BINAP, but the reaction was considerably slower, gave
somewhat lower yield, and required a reaction time ≥7 h.
Replacing (R)-BINAP with ( )-BINAP prolonged the reaction
time considerably, as a consequence of the lower solubility of
the racemate. Also the carbene ligand Neolyst CX32 gave high
conversion to 17 (98%). For cost reasons (R)-BINAP was the
preferred ligand. Concerning the Pd-sources investigated,
Pd(OAc)2 gave a better selectivity than Pd(dba)2 when used
together with SPhos. In a reaction with Pd(dba)2, 17 reacted
further with a second molecule of 16 to give 11% of a tertiary
amine as a byproduct. Of the bases tested, both t-BuOK and t-
AmONa gave 2- and 5-alkoxy derivatives of 16 as byproducts in
various amounts (4−45%). Regarding the base, fortunately
K2CO3 worked better than Cs2CO3. Compound 17 was
obtained in 81−87% yield and 99−100% HPLC purity by
reacting 16 and 4-amino butyl benzoate (1.03 equiv) in THF at
60 °C under inert atmosphere, using (R or )-BINAP (0.03
equiv) and Pd(OAc)2 (0.015 equiv) as catalyst and K2CO3 (1.4
equiv) as base. The workup consisted of cooling to 30 °C,
addition of HOAc (1.4 equiv) in EtOAc and water, phase
separation, aqueous wash, solvent swap to ethanol (EtOH), and
crystallization. The coupling product 17 was isolated in 81%
yield and 99% HPLC purity on a 110-g scale. Analysis of the
product for Pd-content showed <10 ppm.
Acid-Mediated Amination. In parallel to palladium-
catalyzed amination, an alternative acid-mediated procedure
was investigated to avoid the use of a palladium catalyst. A few
examples of acid-mediated coupling between 2-chloropyrimi-
dines and anilines could be found in the literature. A first
approach was to use p-toluene sulfonic acid (pTsOH) in
dioxane at reflux,10a and second 1 equiv 32% hydrogen chloride
(HCl) in an alcohol at reflux was tried.13 A small screen of
reactions between compound 16 and p-aminobenzoic ester was
set up. The use of pTsOH in dioxane generated only about 10%
conversion, but the use of 37% HCl (2 equiv) in ethanol at
reflux looked interesting with >40% conversion. Therefore, the
higher boiling n-butanol (BuOH) was considered as solvent in
combination with butyl 4-aminobenzoate (butamben). When a
mixture of compound 16 and butamben in BuOH was heated
at reflux with 37% HCl (1.05 equiv), >90% conversion was
obtained. If the reaction was performed more concentrated
95% conversion was obtained after reacting overnight.
Butamben is a cheap starting material with a melting point of
88−90 °C and it is also soluble in dilute acidic water. Therefore
butamben (4 equiv) was considered as solvent by melting with
compound 16 at 120 °C with 37% HCl (1.05 equiv). A clear
solution was obtained, and within 5−6 h >97% conversion was
C
dx.doi.org/10.1021/op300365e | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
seen. Then 2 M HCl was added, and after filtration at 50 °C
compound 17 could be isolated in 85% yield. The main features
to investigate were the robustness prior to scale up and the
filtration of compound 17 due to solidification of butamben in
the receiver vessel during isolation. Butamben is fully soluble in
EtOH, and if ethanol was charged in the receiver vessel before
isolation no solidification was obtained. For the planned
manufacture in the pilot plant, 4 equiv of butamben would have
given a reaction volume under the minimum stirring volume of
the intended reactor. Therefore amounts of butamben between
6−10 equiv were investigated. The use of 6 equiv of butamben
proved to work satisfactorily, and in addition 10 equiv gave a
somewhat slower reaction, and more degradation could be seen
after reacting overnight. A reaction with 6 equiv of butamben
gave >98% conversion overnight. A degradation of approx-
imately 1−2% per day could be seen if the reaction mixture was
left for several days. A good operating range for the amount of
37% HCl was 1.1 equiv 0.2 equiv, and the temperature was
set to 110 °C. The reaction was scaled to 0.5 L, and compound
17 was isolated in 82% yield and 97% HPLC purity. Butamben
was the largest impurity (∼2%), which was easily removed
further downstream in the synthesis. Another advantage, in
addition to the reaction being palladium free, is that butamben
probably could be recycled in the longer term.
Ester Hydrolysis and Peptide Coupling. When perform-
ing the butyl ester hydrolysis in a mixture of EtOH and aqueous
KOH at 50 °C, about 4% of the 5-ethoxy analogue, from
displacement of fluoride by alkoxide, was formed as byproduct.
To suppress this side reaction, EtOH was replaced with the
more sterically hindered 2-propanol. Early in the hydrolysis,
some trans-esterification to i-Pr-ester occurred, but at the end
of the reaction >99.5% of the esters were converted to the
carboxylic acid. The best procedure found for the hydrolysis
was to heat 17 in a mixture of 2-propanol (8 relative volumes
(rel vol mL/g) to the weight of 17) and 4 M KOH (1.5 equiv)
at 55 °C for about 6 h. Quenching the reaction mixture with 2
M HCl (1.5 equiv) diluted with water (6 rel vol) resulted in
crystallization of the product which was isolated by filtration.
The formed carboxylic acid was then subjected to a peptide
coupling with N-methylpiperazine as nucleophile to form
compound 9. The method of choice, due to very good
knowledge in-house, was the two-phase system with 1-ethyl-
3(3-dimethylaminopropyl)carbodiimide (EDCI) and a water
solution of 1-hydroxybenzotriazole (HOBt, ∼20% w/w).14 The
reaction was performed in THF (6 rel vol) and water (2 rel vol)
together with N-methylmorpholine as base at 45 °C, and within
a few hours full conversion was obtained. After the reaction was
completed addition of a second solvent was needed to enable
an extractive work up. Several solvents were investigated, e.g., 2-
MeTHF, isopropylacetate (i-PrOAc), and EtOAc. Due to
precipitation of compound 9 during the extractions, only
EtOAc showed good solubility of 9 at higher temperature. The
extraction was performed at 55 °C followed by a brine wash.
The free-base product could be made to crystallize from a
solution of 9 at 55 °C by controlled addition of the antisolvent
heptane, and high-quality 9 was isolated in 87% yield from
compound 17.
addition reaction and an environmentally friendly oxidative
dehydrogenation of dihydropyrimidine 15 using copper
catalysis and oxygen (O2:N2 5:95) as the stoichiometric
oxidant. All of the major scale-up and quality issues with the
early phase route have been addressed. The new route design
and process development has resulted in removal of hazardous
reaction mixtures and high energy intermediates 2 and 3,
expensive and low yielding fluorination chemistry, as well as
avoiding Pd chemistry in the last synthetic step towards the
API. Imidazole 14 was synthesized in four telescoped synthetic
steps via the chloroimine 12 in 34% overall yield. Various
classical methods for the transformation to 12 were tested,
however triphosgene was found to give by far the best result
regarding yield and byproduct formation. Two well working
alternatives for aromatic amine coupling were developed for the
pyrimidine 16, one Pd catalyzed coupling using Buchwald-
Hartwig conditions and another acid-mediated reaction using
37% HCl. The overall yield for the ten-step process, starting
with 300 g of 10, was 14%, which could be compared to the
optimized medicinal chemistry process which gave 4% overall
yield. The close collaboration between synthetic and process
chemists, chemical engineers and analytical chemists greatly
facilitated this process research and development.
EXPERIMENTAL SECTION
■
1H NMR and 13C NMR spectra were recorded on a Bruker
Avance 400 NMR spectrometer. HPLC analyses were
performed on a Dionex P680 HPG, consisting of gradient
pump with a Gynkotek UVD 170S equipped with a Symmetry
Shield RP column (C18, 50 mm × 4.6 mm). The column
temperature was set to +25 °C and the flow rate to 1.0 mL/
min. UV detection was at 292 and 220 nm. A linear gradient
was applied, starting at 90% A (A: 95:5 water/MeCN with
0.05% HCOOH buffer) ending at 80% B (B: 5:95 water/
MeCN with 0.05% HCOOH buffer) over 12 min with 3 min
hold at 90% A before next sample. GC analyses were performed
on an Agilent HP6890 system equipped with an Ultra 2 (25 m
× 320 μm, 0.25 μm) column. HRMS analyses were performed
on a Waters Synapt G2 (Q-TOF), with mass resolution (fwhm)
around 20000.
N-(Tetrahydro-2H-pyran-4-yl)-2,2,2-trifluoroaceta-
mide (11). Tetrahydro-2H-pyran-4-amine acetate, 10 (300 g,
1.86 mol), was mixed in toluene (2.4 L) at 25 °C. TEA (389
mL, 2.79 mol) was charged in one portion, followed by
addition of ethyl trifluoroacetate (288 mL, 2.42 mol) over 30
min. The slurry reaction was left overnight. The reaction
solution was washed with saturated NaHCO3 (300 mL) and
followed by water (300 mL). The organic phase was distilled to
∼1.7 L (6 rel vol towards compound 10) of a solution of 11 in
toluene that was telescoped directly in the next step. The crude
product solution was sampled for GC analysis. GC analysis
indicated ∼100% purity.
N-(Tetrahydro-2H-pyran-4-yl)-2,2,2-trifluoroacetimi-
doyl Chloride (12). TEA (778 mL, 5.58 mol) was charged to
the solution containing compound 11 (1.86 mol theoretical) in
toluene at 50 °C. A solution of triphosgene (182 g, 0.61 mol) in
toluene (800 mL) was charged over 4 h so that 50 °C was
maintained and then left for an additional 4 h. The reaction
mixture was cooled to 25 °C over 1 h and left stirring
overnight. The crude product solution of 12 in toluene was
sampled for GC analysis and telescoped directly in the next
step. GC analysis indicated ∼85% purity, with approximately
∼4% of the unwanted carbamoyl chloride.
CONCLUSION
■
In summary, a new process for the manufacture of GSK3β
inhibitor 9 was developed in our laboratories. This process
relies on a new synthesis to an imidazole 14 with novel
substitution pattern, a true application and scale-up of a Ziegler
D
dx.doi.org/10.1021/op300365e | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
́
N-(2,2-Dimethoxyethyl)-N-(tetrahydro-2H-pyran-4-
concentrated under reduced pressure to approximately 900 mL
(7.5 rel vol towards 14) and adjusted to 20 °C. NH3 in water
(5%, 900 mL) was charged over 1.5 h to initiate precipitation.
The slurry was cooled to 0 °C over 2 h and left for 1 h. The
solids were collected, washed with MeCN/H2O (1:1, 300 mL),
and dried at 40 °C with vacuum to give the pyrimidine 16
(117.1 g, 60.5% yield, corrected for 98.7 wt % 1H NMR assay)
as a yellow-beige solid. MS (ESI) m/z 351 and 353 [M + H]+;
1H NMR (DMSO-d6) δ 1.93 (dd, 2H, J = 12 and 2 Hz), 2.21−
2.33 (m, 2H), 3.41 (dd, 2H, J = 11.6 and 10.4 Hz), 4.00 (dd,
2H, J = 11.6 and 4.4 Hz), 4.87−4.97 (m, 1H), 7.78 (d, 1H, J =
3.2 Hz), 9.09 (d, 1H, J = 2.4 Hz); 13C NMR (DMSO-d6) δ
31.89, 56.71, 66.67, 118.78 (CF3, JC,F = 267.9 Hz), 126.54 (C−
C, JC,F = 2.4 Hz), 134.55 (C−H, JC,F = 9.6 Hz), 137.79 (C−
CF3, JC,F = 38.8 Hz), 146.27 (C−C, JC,F =12.4 Hz), 149.77 (C−
H, JC,F = 25.5 Hz), 153.52 (C−Cl, JC,F = 3.8 Hz), 154.48 (C−F,
JC,F = 264.1 Hz); 99.6% HPLC purity. Reaction monitoring
(14, tR ∼3.77 min; 15, tR ∼5.22 min; 16, tR ∼6.96 min) was
conducted by HPLC analysis with UV detection at 220 nm.
HRMS m/z found 351.0635 [M + H]+, C13H12N4OF4Cl
requires 351.0636.
yl)-2,2,2-trifluoroacetamide (13). A solution of amino-
acetaldehyde dimethyl acetal (293 g, 2.79 mol) in toluene
(200 mL) was charged over 1 h to the solution containing
compound 12 (1.86 mol theoretical) in toluene at 25 °C. The
reaction mixture was left stirring at 25 °C overnight and then
was washed with saturated NaHCO3 (1 L). The toluene phase
was concentrated under reduced pressure to a brown oil, which
was stored at 5 °C. The crude product oil 13 was sampled for
GC analysis and telescoped directly in the next step. GC
analysis indicated ∼81% purity.
1-(Tetrahydro-2H-pyran-4-yl)-2-trifluoromethyl-1H-
imidazole (14). The oil containing 13 (1.86 mol theoretical)
was dissolved in THF (1.5 L) at 20 °C. A solution of H2SO4
(121 mL, 2.23 mol, 98%) in THF (500 mL) was charged over 2
h so the temperature was maintained below 25 °C. The
reaction solution was left for an additional 3 h at 25 °C and
then cooled to 5 °C. Charged i-PrOAc (1 L), followed by
saturated NaHCO3 (1 L). The pH was adjusted to
approximately pH 8 using 45% NaOH(aq). The water phase
was separated, and the organic phase was washed with water
(0.5 L). The organic phase was extracted with 4 M HCl (2 × 1
L). The acidic phases, containing the product, were combined
and concentrated to remove remaining THF. The pH was
carefully adjusted to approximately pH 8 using 45% NaOH(aq)
at such a rate that the temperature was below 10 °C. During
adjustment crystallization was obtained, but if the addition was
too fast, oiling out could be seen. The crystallization was left
stirring at 10 °C overnight. The solids were collected, washed
with cooled water (500 mL), and dried to give the 2-CF3-
imidazole 14 (143.6 g, corrected for 98.4 wt % 1H NMR assay,
0.641 mol) as an off-white sugar-like solid. The overall yield
from 10 in four steps was 34.5%. MS (ESI) m/z 221 [M + H]+;
1H NMR (DMSO-d6) δ 1.84−1.89 (m, 2H), 1.96−2.08 (m,
2H), 3.49 (dt, 2H, J = 12.0 and 2.0 Hz), 3.96 (dd, 2H, J = 11.6
and 4.8 Hz), 4.34−4.44 (m, 1H), 7.13 (d, 1H, J = 1.2 Hz), 7.81
(d, 1H, J = 1.2 Hz); 13C NMR (DMSO-d6) 33.57, 53.51, 65.96,
119.15 (CF3, JC,F = 267.2 Hz), 121.93, 128.61, 133.24 (C−CF3,
Butyl 4-({5-Fluoro-4-[1-(tetrahydro-2H-pyran-4-yl)-2-
(trifluoromethyl)-1H-imidazol-5-yl] pyrimidin-2-yl}-
amino)benzoate (17). Method A. Butamben (127.5 g, 0.66
mol) and compound 16 (40.0 g, 0.11 mol) were neat combined
and heated to 85 °C over 60 min. Concentrated HCl (10.1 mL,
0.12 mol, 37%) was added dropwise over 30 min to the clear-
brown solution. The reaction solution was heated at 110 °C
overnight and then cooled to 100 °C over 30 min, before
controlled addition of 2 M HCl (230 mL). The resulting
precipitate was stirred for about 30 min at 100 °C before
cooling to 50 °C over 1 h and was left an additional 2 h at 50
°C. The solids were collected at 50 °C, washed with 2 M HCl
(230 mL), and dried at 40 °C with vacuum to give the butyl-
1
ester 17 (51.0 g, 82% yield, corrected for 89.7 wt % H NMR
assay) as a yellow-brown solid. 96.6% HPLC purity. Caution!
Add EtOH (230 mL) in the receiver vessel during isolation of
17 to avoid solidification of butamben in the filtrate.
J
C,F = 37.8 Hz); 99.5% GC purity. HRMS m/z found 221.0903
Method B. Butamben (63.7 g, 0.32 mol) and compound 16
(111.4 g, 0.31 mol) were mixed with THF (500 mL) and the
slurry stirred at room temperature. K2CO3 (61.3 g, 0.44 mol)
was charged, followed by THF (580 mL). The mixture was
purged with N2. A solution of Pd(OAc)2 (0.86 g, 3.76 mmol)
and (R)-BINAP (4.83g, 7.52 mmol) in THF (150 mL) was
added to the slurry, and the vessel was inerted by evacuation
and refilling with N2. The mixture was heated to 60 °C with
rapid stirring overnight. The mixture was cooled to 30 °C.
EtOAc (1100 mL) and HOAc (25 mL, 0.44 mol) were added,
followed by water (330 mL), and a solution was obtained. The
lower water phase was separated off, and the organic phase was
washed with 5% NaCl in water (300 mL) and then
concentrated under reduced pressure until reaching half of
the original volume. EtOH (1.5 L) was added, and the
distillation was continued to half of the original volume (∼1 L).
The slurry was heated to 80 °C and stirred until a solution was
obtained. The solution was cooled to 65 °C over 2.5 h and left
for 1 h before cooling was continued to 5 °C with 10 °C/h.
The solids were collected, washed with cooled EtOH (200
mL), and dried at 50 °C with vacuum to give the butyl-ester 17
(131.0 g, 81% yield, corrected for 97.8 wt % 1H NMR assay) as
[M + H]+, C9H12N2OF3 requires 221.0902.
5-Fluoro-4-[1-(tetrahydro-2H-puran-4-yl)-2-(trifluoro-
methyl)-1H-imidazol-5-yl]pyrimidin-2-chloride (16). 2-
CF3-imidazole 14 (120.0 g, 0.55 mol) was dissolved in THF
(1080 mL) and cooled to −25 °C. The reactor was inerted by
evacuation and refilled with N2. Hexyllithium (2.47 M, 243 mL,
0.60 mol) was charged over 35 min so the temperature was
maintained below −15 °C. 2-Chloro-5-fluoro-pyrimidine (80.2
g, 0.60 mol) dissolved in THF (360 mL) was charged over 30
min so the temperature was maintained below −15 °C. The
reaction solution was left at −25 °C for 1 h. A solution of
HOAc (47 mL, 0.82 mol) in EtOAc (480 mL) was charged
over 30 min, and during the charging the temperature was
increased to 0 °C. Water (720 mL) was charged, and the
temperature was set to 20 °C. The water phase was separated
off, and the organic phase was washed with brine (1 L). A
solvent swap from THF to MeCN was performed to a final
volume of approximately 1.8 L of MeCN solution containing
compound 15 with precipitation at 10 °C. Cu(OAc)2 (5.44 g,
27.8 mmol) and TEA (76 mL, 0.55 mol) were added. A
mixture of 5% oxygen in nitrogen was bubbled through at a
flow rate of 300 mL/min overnight (approximately 50%
conversion after 5 h), and a clear-brown solution was obtained
when the reaction had finished. The organic solution was
1
a light-brown solid. MS (ESI) m/z 508 [M + H]+; H NMR
(DMSO-d6) δ 0.92 (dd, 3H, J = 7.6 and 7.2 Hz), 1.36−1.45 (m,
2H), 1.63−1.71 (m, 2H), 1.88−1.94 (m, 2H), 2.10−2.18 (m,
E
dx.doi.org/10.1021/op300365e | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
2H), 3.23 (t, 2H, J = 11.4 Hz), 3.80 (dd, 2H, J = 11.4 and 3.8
Hz); 4.23 (t, 2H, J = 6.6 Hz), 4.76−4.87 (m, 1H), 7.57 (d, 1H,
J = 2.0 Hz), 7.80 (d, 2H, J = 8.8 Hz), 7.90 (d, 2H, J = 8.8 Hz),
8.84 (d, 1H, J = 1.6 Hz), 10.25 (s, 1H); 13C NMR (DMSO-d6)
δ 13.57, 18.72, 30.29, 32.03, 55.97, 63.91, 66.43, 118.14, 118.89
(CF3, JC,F = 267.9 Hz), 122.72, 127.41, 130.06, 132.46 (C−H,
JC,F = 5.5 Hz), 136.47 (C−CF3, JC,F = 38.3 Hz), 144.23 (C−C,
JC,F = 12.5 Hz), 144.42, 147.94 (C−H, JC,F = 24.2 Hz), 150.27
(C−F, JC,F = 252.0 Hz), 155.70 (C−N, JC,F = 3.0 Hz), 165.43;
99.4% HPLC purity. Reaction monitoring (17, tR ∼10.15 min)
was conducted by HPLC analysis with UV detection at 292 nm.
HRMS m/z found 508.1972 [M + H]+, C24H26N5O3F4 requires
508.1972.
132.39 (C−H, JC,F = 5.6 Hz), 136.42 (C−CF3, JC,F = 38.3 Hz),
141.07, 144.06 (C−C, JC,F = 12.5 Hz), 147.96 (C−H, JC,F
24.7 Hz), 150.02 (C−F, JC,F = 251.2 Hz), 156.04 (C−N, JC,F
=
=
2.9 Hz), 168.89; 99.6% HPLC purity. Reaction monitoring (9,
tR ∼2.90 min) was conducted by HPLC analysis with UV
detection at 292 nm. HRMS m/z found 534.2250 [M + H]+,
C25H28N7O2F4 requires 534.2241.
AUTHOR INFORMATION
Corresponding Author
*E-mail: anna.minidis@sp.se; anna.minidis@gmail.com; anna.x.
minidis@astrazeneca.com.
■
Present Addresses
[4-[5-Fluoro-4-[3-tetrahydropyran-4-yl-2-
(trifluoromethyl)imidazol-4-yl]-pyrimidin-2-yl]-
aminophenyl]-(4-methylpiperazin-1-yl)-methanone (9).
Ester Hydrolysis. The butyl-ester 17 (130 g, 0.25 mol) was
mixed in 2-propanol (1040 mL) at 25 °C. KOH (4 M, 95 mL,
0.37 mol) was charged, and the reaction mixture was heated at
55 °C for 6 h. HCl (2 M, 189 mL, 0.38 mol) diluted with water
(780 mL) was charged to the reaction solution over 45 min,
while 55 °C was maintained. The slurry was cooled to 15 °C
over 2 h and left for 1 h. The solids were collected, washed with
water (380 mL), and dried at 55 °C with vacuum to give the
∥R&D Regular Product, Cambrex Karlskoga, SE-691 85
Karlskoga, Sweden. Tel.: +46 586 78 33 46. E-mail: anette.
witt.stranne@cambrex.com.
⊥Substance and Formulation, SP Process Development AB, SP
Technical Research Institute of Sweden, RISE, Forskargatan 18,
B341, Box 36, SE-151 21 Sodertalje, Sweden. Tel.: +46 10 516
̈
̈
65 11. E-mail: anna.minidis@sp.se.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
1
■
butyl acid (122.8 g, 98% yield, corrected for 98.5 wt % H
NMR assay) as a yellow solid. MS (ESI) m/z 452 [M + H]+;
1H NMR (DMSO-d6) δ 1.88−1.94 (m, 2H), 2.10−2.18 (m,
2H), 3.23 (t, 2H, J = 11.4 Hz), 3.81 (dd, 2H, J = 11.6 and 4.0
Hz), 4.77−4.84 (m, 1H), 7.57 (d, 1H, J = 2.0 Hz), 7.78 (d, 2H,
J = 8.8 Hz), 7.88 (d, 2H, J = 8.8 Hz), 8.83 (d, 1H, J = 1.6 Hz),
10.22 (s, 1H), 12.61 (br s, 1H); 13C NMR (DMSO-d6) δ 32.02,
55.96, 66.43, 118.08, 118.89 (CF3, JC,F = 267.9 Hz), 123.66,
127.43, 130.24, 132.45 (C−H, JC,F = 5.4 Hz), 136.48 (C−CF3,
JC,F = 38.2 Hz), 144.08, 144.21 (C−C, JC,F = 12.6 Hz), 147.94
(C−H, JC,F = 24.4 Hz), 150.22 (C−F, JC,F = 251.9 Hz), 155.76
(C−N, JC,F = 3.1 Hz), 167.01; 99.8% HPLC purity. Reaction
monitoring (acid, tR ∼7.25 min) was conducted by HPLC
analysis with UV detection at 292 nm. HRMS m/z found
452.1345 [M + H]+, C20H18N5O3F4 requires 452.1346.
We thank Erika Lindberg and Erica Tjerneld for engineering
support, David Aslan and Marie Berzelius for analytic support,
Hefeng Pan for the HRMS analyses, Mats Ridemark and
Tommi Ratilainen for the development of the Mid-IR analysis,
and Ann-Britt Fransson for chemistry production in 5-L scale.
We also thank Jim Brennan and Tobias Rein for helpful
suggestions and guidance during the course of this work.
REFERENCES
■
(1) (a) Bhat, R. V.; Budd, S. L. Neurosignals 2002, 11, 251. (b) Bhat,
R. V.; Budd Haeberlein, S. L.; Avila, J. J. Neurochem. 2004, 89, 1313.
(c) Johansson, S.; Jamsa, A.; Vasange, M.; Winblad, B.; Luthman;
̈
̈
̈
Cowburn, R. F. Neurochemistry 2006, 17, 907.
(2) Andersson, L., Arzel, E., Berg, S., Burrows, J., Hellberg, S., Huerta,
F., Pedersen, T., Rein, T., Rotticci, D., Staaf, K., Turek, D.
(Astrazeneca AB, Sweden). PCT Int. Appl. WO/2007/040440 A1,
2007.
(3) (a) Owen, D.; Plevey, R. G.; Tatlow, J. C. J. Fluorine Chem. 1981,
17, 179. (b) Kimoto, H.; Fujii, S.; Cohen, L. A. J. Org. Chem. 1982, 47,
2867. (c) Wiehn, M. S.; Vinogradova, E. V.; Togni, A. J. Fluorine Chem.
2010, 131, 951.
EDCI Coupling. The butyl acid (112.6 g, 0.24 mol) was
mixed in THF (675 mL) at 25 °C. N-Methyl-piperazine (29.3
mL, 0.26 mol), NMM (41 mL, 0.37 mol), and HOBt as a water
solution (17 mL, 24.4 mmol, 19.4%) were added to the
reaction solution. EDCI (72.7 g, 0.34 mol) dissolved in water
(220 mL) was added over 10 min. The reaction mixture was
heated at 45 °C for 1 h. EtOAc (675 mL) was added, and the
reaction solution was heated at 55 °C. The lower water phase
was separated off, and the organic phase was washed with 5%
NaCl in water (225 mL), and then the organic solution was left
stirring at 55 °C. Heptane (675 mL) was added over 1.5 h to
initiate crystallization. The slurry was cooled to 5 °C over 8 h
and left for at least 2 h. The solids were collected, washed with
a cooled mixture of EtOAc and heptane (1:1, 600 mL), and
dried at 60 °C with vacuum to give compound 9 (120.8 g, 89%
yield, corrected for 96.1 wt % 1H NMR assay) as a light-yellow
(4) Tamura, K.; Mizukami, H.; Maeda, K.; Watanabe, H.; Uneyama,
K. J. Org. Chem. 1993, 58, 32.
(5) Cotarca, L.; Delogu, P.; Nardelli, A.; Sunjic,
̌
́
V. Synthesis 1996,
553.
(6) (a) Frutos, R. P.; Rodríguez, S.; Patel, N.; Johnson, J.; Saha, A.;
Krishnamurthy, D.; Senanayake, C. H. Org. Process Res. Dev. 2007, 11,
1076. (b) Manley, P. J.; Balitza, A. E.; Bilodeau, M. T.; Coll, K. E.;
Hartman, G. D; McFall, R. C; Rickert, K. W.; Rodman, L. D.; Thomas,
K. A. Bioorg. Med. Chem. Lett. 2003, 13, 1673.
(7) (a) Pivsa-Art, S.; Satoh, T.; Kawamura, Y.; Miura, M.; Nomura,
M. Bull. Chem. Soc. Jpn. 1998, 71, 467. (b) Bellina, F.; Cauteruccio, S.;
Mannina, L.; Rossi, R.; Viel, S. Eur. J. Org. Chem. 2006, 693.
(c) Aoyagi, Y.; Inoue, A.; Koizumi, I.; Otha, A. Heterocycles 1992, 257.
1
solid. MS (ESI) m/z 534 [M + H]+; H NMR (DMSO-d6) δ
1.86−1.93 (m, 2H), 2.08−2.18 (m, 2H), 2.19 (s, 3H), 2.23−
2.38 (m, 4H), 3.22 (t, 2H, J = 11.4 Hz), 3.34−3.68 (m, 4H),
3.80 (dd, 2H, J = 11.4 and 3.8 Hz), 4.77−4.88 (m, 1H), 7.35
(d, 2H, J = 8.8 Hz), 7.57 (d, 1H, J = 2.4 Hz), 7.70 (d, 2H, J =
8.8 Hz), 8.05 (d, 1H, J = 1.6 Hz), 10.04 (s, 1H); 13C NMR
(DMSO-d6) δ 32.01, 45.57, 54.49 (piperazine), 55.96, 66.44,
118.65, 118.90 (CF3, JC,F = 267.8 Hz), 127.53, 127.89, 129.08,
(8) (a) Schnurch, M.; Flasik, R.; Khan, A. F.; Spina, M.; Mihovilovic,
̈
M. D.; Stanetty, P. Eur. J. Org. Chem. 2006, 3283. (b) Hueter, O. F.;
Brunner, H.-G.; Muehlebach, M.; O’Sullivan, A. (Syngenta Partic-
ipations AG, Switzerland). PCT Int. Appl. WO/2005/047281 A1,
2005. (c) Zhong, Y.-L.; Lee, J.; Reamer, R. A.; Askin, D. Org. Lett.
2004, 6, 929. (d) Dobler, M. R. Tetrahedron Lett. 2003, 44, 7115.
(9) Rieke, R. Aldrichimica Acta 2000, 33, 52.
F
dx.doi.org/10.1021/op300365e | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(10) (a) Bursavich, M. G.; Lombardi, S.; Gilbert, A. M. Org. Lett.
2005, 7, 4113. (b) Shin Cho, Y.; Borland, M.; Brain, C.; Chen, C. H.-
T.; Cheng, H.; Chopra, R.; Chung, K.; Groarke, J.; He, G.; Hou, Y.;
Kim, S.; Kovats, S.; Lu, Y.; O’Reilly, M.; Shen, J.; Smith, T.; Trakshel,
G.; Vogtle, M.; Xu, M.; Je Sung, M. J. Med. Chem. 2010, 53, 7938.
̈
(c) Harden, D. B.; Mokrosz, M. J.; Strekowski, L. J. Org. Chem. 1988,
53, 4137. (d) Ziegler, K.; Zeiser, H. Justus Liebigs. Ann. Chem. 1931,
174.
(11) The byproduct was not isolated and fully characterized. Mass
spectral analysis revealed a molecular weight (MS (ESI) m/z 333 and
335 [M + 1]) of the impurity for which structure 18 was tentatively
assigned.
(12) Yamamoto, K.; Chen, Y. G.; Buono, F. G. Org. Lett. 2005, 7,
4673.
(13) Gunzenhauser, S.; Balli, H. Helv. Chim. Acta 1988, 71, 33.
(14) (a) Nozaki, S. Org. Lett. 1997, 1. (b) Nozaki, S. Pept. Res. 1999,
54, 162.
G
dx.doi.org/10.1021/op300365e | Org. Process Res. Dev. XXXX, XXX, XXX−XXX