Exploratory Process Development of a Novel Diacylglycerol Acyltransferase-1 (DGAT-1) Inhibitor

Article abstract of DOI:10.1021/op400215h

A practical large-scale synthesis was developed for 1, a DGAT-1 inhibitor, involving an aza-Michael reaction, amidation, Dieckman cyclization, and conjugate addition of cyanamide followed by cyclization, to form the fused 4-amino-7,8-dihydropyrido[4,3-d]p

Full text of DOI:10.1021/op400215h

Article
pubs.acs.org/OPRD
Exploratory Process Development of a Novel Diacylglycerol
Acyltransferase‑1 (DGAT-1) Inhibitor
Andrei Shavnya,* Yong Tao,* Susan C. Lilley, Kevin B. Bahnck, Michael J. Munchhof, Asaad Nematalla,
Michael Waldo, and David R. Bill
Groton Laboratories, Pfizer Worldwide Research and Development, Eastern Point Road, Groton, Connecticut 06340, United States
S
* Supporting Information
ABSTRACT: A practical large-scale synthesis was developed for 1, a DGAT-1 inhibitor, involving an aza-Michael reaction,
amidation, Dieckman cyclization, and conjugate addition of cyanamide followed by cyclization, to form the fused 4-amino-7,8-
dihydropyrido[4,3-d]pyrimidin-5-one scaffold. The enabled process presented here substantially improved safety (in particular,
due to eliminating a nitration step and optimizing a high-energy intermediate step), reproducibility, and scalability, resulting in
delivery of a multikilogram quantity of the API with high purity. The controls of API quality and particle size were also discussed.
after thorough investigation of several modes of retrosynthetic
bond disconnection, proved to be reliable and versatile in
expansive analogue production. This synthesis started with aza-
Michael reaction of aniline 2 to ethyl acrylate, followed by N-
acylation with cyanoacetic acid, leading to intermediate 3,
which was converted to the key vinylogous carbamic acid 4
through a Dieckmann cyclization.3 Upon conversion of 4 to the
corresponding chloride 5 and further to vinyl methylcarbamate
6, the latter was reacted with cyanamide under basic conditions
to afford the penultimate intermediate 7. The 2-methoxy-4-
aminopyrimidine system formation, leading to target 1, was
effected by acid-catalyzed addition of methanol.4 While many
well thought-out solutions were involved in the overall
discovery route design, especially the efficient building of the
4-amino-7,8-dihydropyrido[4,3-d]pyrimidin-5-one scaffold,
multikilogram delivery of API required further investigation
to ensure high purity and yield as well as to enable suitable
process safety, reproducibility, and operational efficiency on a
kilogram scale. The API quality control, including impurity
profile, residual solvent content, polymorph forms, and particle
size distribution, was also addressed.
Preparation of Aniline 2. 1-Aminophenyl-cyclobutanecar-
bonitrile 2 was synthesized by reduction of the corresponding
nitro compound 8, which itself was prepared via two different
pathways as depicted in Scheme 2. In the original medicinal
chemistry approach, nitration of 1-phenyl-cyclobutanecarboni-
trile 9 with potassium nitrate in concentrated sulfuric acid at 0
°C cleanly afforded 8 in 90% yield. However, these reaction
conditions were expected to have a potential run-away hazard
during scale-up, as a delayed sharp exothermic event occurred
on a 10 g scale. In an effort to find homogeneous conditions
suitable for a safer flow process, we unsuccessfully examined a
milder variant of batch nitration, in particular using a solution
of 70% aqueous nitric acid in acetic acid at 70 °C. We also
envisioned that 8 could be prepared by cycloalkylation5 of
readily available nitrile 10, which already carried the nitro
INTRODUCTION
■
Enzyme acyl-CoA:diacylglycerol acyltransferase-1 (DGAT-1)
catalyzes the rate-limiting step in triglyceride synthesis. It has
recently emerged as an attractive target for therapeutic
intervention in the treatment of Type II diabetes and obesity.1
As part of the program aimed at finding an orally active API, a
series of highly potent and selective DGAT-1 inhibitors was
recently discovered carrying a novel 4-amino-7,8-
dihydropyrido[4,3-d]pyrimidin-5-one scaffold.2 Compound A
(Figure 1) was first identified as a promising lead to inhibit the
Figure 1. Structures of compounds A and 1.
DGAT-1-mediated pathway. Afterwards, analogue 1 demon-
strated a better biological profile and a wider therapeutic index
(TI) and was therefore selected for clinical development. Due
to the structural similarity of these two compounds, useful
synthetic solutions initially found for compound A were
adapted to expedite scale-up development of 1 with the aim
of quick delivery for regulatory toxicity and early clinical
studies. Overall results of this work evidenced the efficiency of
the general synthetic strategy originally employed by our
medicinal chemists. Herein we describe the early process
development work that allowed for successful transition from
gram-scale laboratory preparation to multikilogram production
of 1 in a kilo lab.
RESULTS AND DISCUSSION
■
Synthetic Route Selection. The synthesis outlined in
Scheme 1 was the original discovery route to compound 1 that
was prepared as part of 4-amino-7,8-dihydropyrido[4,3-d]-
pyrimidin-5-one series of DGAT-1 inhibitors. This strategy,
Received: August 10, 2013
Published: November 12, 2013
© 2013 American Chemical Society
1510
dx.doi.org/10.1021/op400215h | Org. Process Res. Dev. 2013, 17, 1510−1516

Organic Process Research & Development
Article
Scheme 1. Medicinal Chemistry synthetic route to target 1
Scheme 2. Synthesis of the starting aniline 2
group. After screening a number of alkylating reactants, bases,
solvents, and phase transfer catalysts at various temperatures,
optimal conditions were found that employed 1.2 equiv of 1,3-
dibromopropane, 2.6 equiv of potassium carbonate, and 0.1
equiv of tetrabutylammonium bromide (TBAB) in acetone at
55 °C. The reaction profile showed ∼80% of 8 with low levels
of three side products (Figure 2). Purification via silica gel plug
In the original discovery synthesis, transfer hydrogenation
with ammonium formate and dry 10% palladium on carbon
cleanly converted 8 to 2. Our safety protocol for scale-up,
however, required using wet palladium on carbon (containing
50% of water), which unexpectedly led to significantly more
byproducts. Additionally, formation of ammonium formate in
the reflux condenser under the reaction conditions (70 °C)
became a serious safety issue. Other transfer hydrogenation
reagents, e.g., formic acid and cyclohexadiene, did not improve
the reaction profile. Instead, the conventional hydrogenation
with hydrogen gas (30−50 psi) in the presence of wet 5%
palladium on carbon (10 wt % loading) in methanol cleanly
converted 8 to aniline 2 at a temperature above 45 °C. The
exothermic process was controlled between 45 to 60 °C by the
adjusting hydrogen feeding rate, agitation speed, and cooling
capacity, and aniline 2 was obtained in 98% yield.
Figure 2. Byproducts of cyclobutanation.
Preparation of Vinylogous Carbamic Acid 4. A
telescoped three-step conversion was initially developed for a
multigram scale to make the key intermediate 4 as shown in
Scheme 3 (Method 1). Aniline 2 was reacted with excess ethyl
acrylate (2 equiv) through an aza-Michael reaction in
filtration, followed by recrystallization afforded 8 in 49−56%
yields with >95% purity (HPLC). Even though the cyclo-
butanation approach gave lower throughput, it was still selected
for scale-up due to safety considerations, and 8.5 kg of 8 was
successfully produced.
Scheme 3. Telescoped Route to Key Intermediate 4
1511
dx.doi.org/10.1021/op400215h | Org. Process Res. Dev. 2013, 17, 1510−1516

Organic Process Research & Development
Article
Scheme 4. Synthesis of 1
Scheme 5. Formation of 1 and impurities
triethylamine and ethanol at 80 °C to give a mixture of the
desired monoalkylated 11 and dialkylated side product 12 in
∼4:1 ratio. Without purification, this mixture was subjected to
two sequential chemical conversions: N-acylation with cyano-
acetic acid in the presence of N,N ′-diisopropylcarbodiimide
(DIC) and DMAP in DMF to give intermediate 3, and then an
intramolecular Dieckmann-type cyclization with DBU in THF
to afford key intermediate 4. An efficient isolation protocol that
utilized the acidic nature of vinylogous carbamic acid 4 was
developed as follows. Partition of the reaction mixture between
water and ethyl acetate kept most lipophilic impurities,
including 12, in the organic layer, while extracting 4 into the
aqueous phase as the DBU salt 13. Acidification of the aqueous
phase with hydrochloric acid resulted in precipitation of 4 in
96.2% purity (HPLC). Using this method resulted in ∼10 g of
intermediate 4 prepared in 50% yield over three steps.
This strategy was adopted for scale-up; enabling work was
focused on improving process safety and efficiency and led to
conditions outlined as Method 2 (Scheme 3). Ethyl acrylate is
known to undergo polymerization at elevated temperature,6
and therefore excess reagent should be minimized. The use of
acetic acid as the reaction solvent successfully reduced the
amount of ethyl acrylate required for the reaction to 1.1 equiv
and lowered the reaction temperature to 60 °C. The new
conditions resulted in a mixture of 11, 12, and 2 with the ratio
of 15:1:1. The original N-acylation using N,N′-diisopropylcar-
bodiimide (DIC) as the amide coupling reagent generated
diisopropyl urea, which was difficult to purge and complicated
intermediate isolation in the downstream process. T3P
(propanephosphonic acid anhydride, cyclic trimer, 50%
solution in ethyl acetate)7,8 was found to be an excellent
substitute. Reaction of crude 11 with cyanoacetic acid in the
presence of T3P and TEA in EtOAc and MTBE afforded crude
3 with >90% HPLC purity. Dieckmann-type condensation of
crude 3 was effected by DBU in refluxing THF. Solvent switch
from THF to EtOAc, followed by water extraction, effectively
removed 12 and other lipophilic impurities, and product 4 was
precipitated by adjusting the pH of the aqueous layer with
hydrochloric acid to pH < 2. Trituration with methanol
enhanced the purity of 4 to >98%. The yields were in the range
of 50−54% over three steps on a kilogram scale.
Building the Pyrimidine Ring To Form Target 1. Ten
grams of 1 was quickly prepared for early biological tests
(Scheme 4). O-Methylation of 4 was carried out as two
transformations in one pot: chlorination by Vilsmeier method9
with oxalyl chloride in the presence of catalytic DMF (10 mol
%) in dichloromethane at −10 to 20 °C led to vinylogous acid
chloride 5; methanolysis of the latter at 60 °C provided
vinylogous methyl carbamate 6 in ∼70% yield. Reaction of 6
with cyanamide in the presence of sodium methoxide in
methanol afforded the penultimate compound 7, which
precipitated during the reaction and was isolated by filtration
in 90% yield. While direct conversion of 5 to 7 was possible,
this reaction was lower yielding, and hydrolysis of 5 to the
1512
dx.doi.org/10.1021/op400215h | Org. Process Res. Dev. 2013, 17, 1510−1516

Organic Process Research & Development
Article
Scheme 6. Enabled process for 1
starting material 4 was observed. Treatment of 7 with hot
methanol in the presence of 1 equiv of sulfuric acid generated
O-methylisourea 17, which cyclized in situ to form the desired
2-methoxy-4-aminopyrimidine system (Scheme 5). Crude 1
was precipitated out by neutralizing the reaction mixture with 1
N KOH. The isolated crude 1 was typically contaminated with
10−20% of each of the side products 14 and 15.
good solubility in aqueous methanol and remained soluble after
neutralization with aqueous KOH. When sulfuric acid was
replaced with methanesulfonic acid (Scheme 6), the filtration
was dramatically improved and 5.1 kg of crude 1 was obtained
by this method in 86% yield with 81% HPLC purity mostly
associated with two major impurities 14 and 15 at ∼7% each. It
should be noted that weaker acids, such as trifluoroacetic,
phosphoric, p-toluenesulfonic, acetic, and citric acids, resulted
in low conversion; hydrochloric acid led predominantly to
chlorinated product 16 (Scheme 5)
The strategy described above was adopted with the following
modifications to improve process safety and scale-up feasibility.
The rate of off-gassing during the chlorination stage was
controlled by increasing the initial temperature from −10 to 10
°C, reducing the amount of DMF to 5% mol equiv and adding
oxalyl chloride (1.2 equiv) at a rate that allowed maintaining
the reaction temperature below 15 °C. A scrubber of aqueous
NaOH was also introduced to trap hydrogen chloride. Dilution
of the reaction mixture with methanol at 35 °C for 20 h
smoothly converted 5 to 6. Removal of dichloromethane by
evaporation resulted in full precipitation of 6 and 5.2 kg
(largest-scale run) of this material was obtained as a white
powder in 88% yield with 96.2% HPLC purity (Scheme 6).
Analysis of the DSC showed that the penultimate
intermediate 7 had an energy potential of −510.4 J/g with an
onset temperature of 91 °C. Our safety policy required at least
100 °C difference between the process temperature and the
DSC onset temperature for scale-up of high-energy com-
pounds, so it was decided not to isolate intermediate 7. Thus,
base-promoted cyanamidation and acid-promoted cyclization
were carried out in one pot (Scheme 6). In addition, excess
cyanamide was minimized as the reagent had a very high energy
potential (−1461.7 J/g) with low onset temperature of 108.5
°C. Compound 6 was cleanly converted to 7 in 2 h at 20 °C
with 1.1 equiv of cyanamide and 1.2 equiv of sodium methoxide
in methanol. The reaction mixture was then subjected to strong
acid at elevated temperature to effect the cyclization. Similar to
the original synthesis, sulfuric acid (2.5 mol equiv) afforded a
relatively clean cyclized target product 1 (with ∼75% HPLC
purity). However, the original isolation protocol did not work
well in the new process because neutralization with aqueous
NaOH or KOH resulted in a large amount of sodium or
potassium sulfates that coprecipitated with 1 as a very fine solid
and made the product filtration extremely slow. Methanesul-
fonic acid (2.8 mol equiv) was found to be more effective for
the cyclization and generated a slightly better reaction profile. It
was critical that sodium and potassium methanesulfonates had
Purification of 1 To Meet the Purity Requirement for
Clinical Use. Target 1 exhibited a unique solubility profile.
The drug substance was very soluble in acetic acid, partially
soluble in hot DMF, DMAc or DMSO, but poorly soluble in
other organic solvents and water. The original two-step
purification protocol afforded 1 in 20−40% yields on gram
scale: trituration with hot methanol mainly purged off 14 to
enhance the purity to ∼85%, which was followed by
recrystallization in hot aqueous acetic acid to further upgrade
purity to >98% with <1% of 15. The latter operation purged
most impurities but also hydrolyzed the labile methoxy group
to convert a small amount of 1 back to side product 15 at
elevated temperature. Consequently, both yield and purity
varied dramatically from batch to batch, especially during scale-
up syntheses. All attempts to recrystallize crude 1 in acetic acid
and water at ambient temperature resulted in unacceptable
purity enhancement. The need for an effective purification
protocol made it necessary to thoroughly investigate new
conditions. Compound 1 had a good solubility response to
temperature in mixtures of 5−20% acetic acid with other polar
solvents, such as MeOH, EtOH, acetone, MeTHF, DMF, and
DMAc. In these mixtures, it dissolved at 60−70 °C and
crystallized out when cooled. However, only a combination of
acetic acid with DMF or DMAc demonstrated efficiency to
purge impurities. The final process involved dissolving crude 1
in DMF and acetic acid at 65 °C, followed by cooling to −20
°C at rate of 1 °C per minute and cold filtration. The purified 1
was obtained in 56% yield as a white powder with 99.3% HPLC
purity and no single impurity over 0.2%. Raising the final
temperature to 0 °C resulted in an additional ∼10% loss in
yield. The main drawback of this purification was that the
isolated 1 contained high level of residual DMF (>0.5 wt %),
which was impossible to lower by drying, probably due to
formation of a solvate.
1513
dx.doi.org/10.1021/op400215h | Org. Process Res. Dev. 2013, 17, 1510−1516

Organic Process Research & Development
Article
In order to purge DMF to acceptable ICH levels for clinical
use,10 recrystallization in an acetic acid/water mixture at
ambient temperature was applied. It should also be pointed out
that only one polymorph, Form I, was obtained in all
conditions of recrystallization. However, the particle size was
typically too large to get adequate dissolution and content
uniformity for regulatory toxicity studies and clinical trials due
to its low water solubility (0.01 mg/mL at pH 6.5). Therefore,
high-shear, wet milling technology was adopted in the final
recrystallization to achieve the recommended particle size
D[4,3] < 30 μm and D[v, 0.95] < 60 μm. The above purified 1
was dissolved in acetic acid (3v) at 35 °C, filtered through 0.5
μm in-line filter and cooled to 20 °C. Addition of water (12v)
over 1.5 h resulted in crystallization of 1. The slurry was further
cooled to 10 °C and wet-milled. As a result, 2.6 kg of target 1
was obtained in 93% recovery with 99.63% HPLC purity. The
particle size distribution met the recommended standard with
D[4,3] = 25 μm and D[v, 0.95] = 58 μm.
was cooled to 0 °C and stirred for 2 h. Solid was filtered off and
dried in vacuo to obtain the title compound as a straw-colored
solid (4.25 kg, 49%). HPLC purity was 95%. 1H NMR
(CDCl3) δ 2.11−2.19 (m, 1H), 2.46−2.56 (m, 1H), 2.63−2.70
(m, 2H), 2.87−2.95 (m, 2H), 7.62 (d, J = 8.8 Hz, 2H), 8.27 (d,
J = 8.8 Hz, 2H); 13C NMR (CDCl3) δ 17.2, 34.9, 40.2, 123.3,
124.4, 126.9, 146.8, 147.6; GC/MS: m/z = 202 (M).
1-(4-Aminophenyl)cyclobutanecarbonitrile (2). A 10 L
autoclave reactor purged with nitrogen was charged with 8 (505
g, 2.5 mol), wet 5% palladium on carbon (8 g), and methanol
(8 L). The reactor was pressurized with 50 psi of hydrogen.
The mixture was agitated for 3 h under 45−55 psi of hydrogen
gas, maintaining reaction temperature below 60 °C with
external cooling. Upon cessation of hydrogen uptake, the
reaction mixture was cooled to 20 °C. Hydrogen was vented
and replaced with nitrogen. The catalyst was filtered off through
Celite, and the filter cake was washed with methanol while
being kept under flow of nitrogen. The combined filtrates were
concentrated in vacuo to give the crude title compound as a
colorless oily residue, which solidified upon standing (422 g,
CONCLUSIONS
■
1
98%). HPLC purity was over 95.8%. H NMR spectrum and
A novel DGAT-1 inhibitor 1 was synthesized from aniline 2
through an efficient Dieckmann-type cyclization to form
vinylogous carbamic acid 4, followed by cyanamide incorpo-
ration and acid-promoted cyclization. The process was
developed for scale-up by telescoping many conversions and
designing safety into the process. Compared to the original
Medicinal Chemistry synthesis, the overall safety, robustness,
and reproducibility were significantly improved; HPLC purity
of the final product was increased from 99.08% to 99.63%. The
enabled process, summarized in Scheme 6 provided 2.6 kg of
API in overall 24% yield, whereas the best yield achieved in the
original synthesis was 14%. The particle size distribution was
also controlled through wet milling.
MS were consistent with the literature data.2b,11 1H NMR
(CDCl3) δ 2.00−2.09 (m, 1H), 2.34−2.42 (m, 1H), 2.52−2.60
(m, 2H), 2.74−2.82 (m, 2H), 3.79 (br s, 2H), 6.70 (d, J = 8.80
Hz, 2H), 7.20 (d, J = 8.80 Hz, 2H); 13C NMR (CDCl3) δ 16.9,
34.8, 39.6, 115.3, 124.8, 126.6, 129.7, 145.8; LC/MS (API+) m/
z: 173.1 [M + 1]+; mp 48−49 °C.
Ethyl 3-(4-(1-Cyanocyclobutyl)phenylamino)-
propanoate (11). A mixture of 2 (7.16 kg, 41.58 mol),
ethyl acrylate (4.9 L, 44.94 mol), and acetic acid (7.3 L, 127.7
mol) was stirred at 60 °C for 2 d. The mixture was cooled to 0
°C, and MTBE (50 L) was added. The obtained mixture was
successively washed with 0.5 M hydrochloric acid (15 L), water
(10 L), and 10% wt aqueous potassium bicarbonate (10 L).
Each aqueous solution was sequentially extracted with an
additional portion of MTBE (4 L). The combined organic
layers were concentrated in vacuo to ∼9 L of the residual
volume.12 This obtained solution of crude 11 was directly used
in the next step without further purification (assumed 100%
yield of 11 for calculations).
Ethyl 3-(2-Cyano-N-(4-(1-cyanocyclobutyl)phenyl)-
acetamido)propanoate (3). The above solution of crude
11 in MTBE (assumed 41.5 mol) was combined with ethyl
acetate (40 L), cyanoacetic acid (3.2 kg, 37.6 mol), and
triethylamine (12 L, 86.7 mol). The mixture was cooled to −5
°C and 1-propanephosphonic acid cyclic anhydride (T3P, 50%
in ethyl acetate, 20 L, 34.0 mol) was added over 2 h while
maintaining the temperature below 0 °C. After addition was
complete, the reaction mixture was stirred at −5 °C for 1 h,
then warmed to 20 °C in 2 h and stirred at this temperature for
18 h. The reaction mixture was diluted with ethyl acetate (40
L) and successively washed with 10% aqueous potassium
bicarbonate (40 L), 0.5 M hydrochloric acid (20 L), and 25%
wt sodium chloride (20 L). The organic layer was concentrated
in vacuo to ∼12 kg of the residual oil. This crude 3 was directly
used in next step without further purification (assumed 100%
yield of 3 for calculations).
EXPERIMENTAL SECTION
■
General. Liquid chromatography mass spectrometry
(LCMS) was performed on an Agilent 1100 Series (Waters
Atlantis C18 column, 4.6 mm × 50 mm, 5 μm; 95% water/
acetonitrile linear gradient to 5% water/acetonitrile over 4 min,
hold at 5% water/acetonitrile to 5 min, trifluoroacetic acid
modifier (0.05%); flow rate = 2.0 mL/min). Reaction
monitoring and purity of intermediates and the final compound
were checked by HPLC in the following conditions: Column:
Zorbax SB-CN, 5 μm, 4.6 mm × 150 mm; Column
Temperature: 30 °C; Flow Rate: 2 mL/min; Detection: UV
@ 210 nm; Mobile phase: A: 0.2% phosphoric acid in water, B:
Acetonitrile; Linear Gradient: from 95% of A to 5% of A within
15 min. HPLC purity was reported at 210 nm wavelength.
1-(4-Nitrophenyl)cyclobutanecarbonitrile (8). A mix-
ture of acetone (80 L), powdered potassium carbonate (15.35
kg, 111 mol), 2-(4-nitrophenyl)acetonitrile (10, 7.0 kg, 43.2
mol), 1,3-dibromopropane (5.13 L, 50.6 mol), and tetrabuty-
lammonium bromide (1.0 kg, 3.1 mol) was stirred at 55 °C for
20 h. The reaction was cooled to 20 °C, and solids were filtered
off with an acetone wash. The filtrates were concentrated in
vacuo at 50 °C to ∼20 L. The residual slurry was extracted with
MTBE (3 × 12 L) at about 50 °C. The combined MTBE
extracts were concentrated in vacuo at 50 °C, and the residue
was loaded on a short 20 kg silica gel column. The column was
eluted with a 1:2 mixture of ethyl acetate and heptane. The
obtained solution was concentrated in vacuo until a precipitate
began to form (residual volume was about 40 L). The slurry
1-(4-(1-Cyanocyclobutyl)phenyl)-4-hydroxy-2-oxo-
1,2,5,6-tetrahydropyridine-3-carbonitrile (4). A mixture of
the above crude 3 (assumed 41.5 mol), 1,8-diazabicyclo[5.4.0]-
undec-7-ene (DBU, 5.6 L, 37.4 mol), and THF (80 L) was
stirred at 60 °C for 1 h. The reaction mixture was cooled to 20
°C, then water (80 L) and ethyl acetate (60 L) were added
1514
dx.doi.org/10.1021/op400215h | Org. Process Res. Dev. 2013, 17, 1510−1516

Organic Process Research & Development
Article
successively. The biphasic mixture was stirred for 30 min at 20
°C and the aqueous phase was separated and additionally
washed with ethyl acetate (40 L). The two combined organic
phases were back-extracted with water (40 L). The combined
aqueous solution was heated to 30 °C, and 6 M hydrochloric
acid (∼8 L) was added over 2 h to reach pH = 1−2. The
resulting slurry was cooled to 20 °C, stirred for 18 h, and then
filtered. The filter cake was sequentially washed with water (6
L) and ethyl acetate (6 L) and dried under nitrogen blow for 18
h. The obtained solid (6.38 kg) was stirred in methanol (22 L)
at 20 °C for 18 h and then filtered off, washed with methanol (2
L), and dried in vacuo at 50 °C for 18 h to obtain the title
compound as a beige solid (5.80 kg, 49% over three steps from
was slowly added, followed by 34% wt aqueous KOH (7.9 kg,
47.48 mol) to reach pH > 10 while maintaining the reaction
temperature below 25 °C. The resulting slurry was agitated at
20 °C for 2 h and filtered. The filter cake was washed with 3:1
(v/v) of methanol and water (20 L) and water (14 L), and then
dried in a tray dryer under vacuum at 55 °C for 18 h to obtain
the crude title product 1 as an off-white powder (5.07 kg, 86%).
HPLC purity was 86.7%, associated with 4.2% of 14 and 5.3%
of 15.
Purification of Crude 1-(4-(4-Amino-2-methoxy-5-
oxo-7,8-dihydropyrido[4,3-d]pyrimidin-6(5H)-yl)-
phenyl)cyclobutanecarbonitrile (1). A mixture of DMF (32
L), acetic acid (1.6 L), and crude 1 (5.04 kg) was heated to 70
°C over 30 min and agitated at this temperature for 15 min.
The resulting solution was cooled to −20 °C at a rate of one
degree per minute. The slurry was agitated at this temperature
for 1 h and then filtered through a Nutsche filter. The cake was
washed with methanol (10 L) and blown with nitrogen flow for
3 h. [Note: By adding water (90 L) to the mother liquor, an
additional 1.88 kg of a second crop of 1 with 90% HPLC purity
was obtained as an off-white powder in 37% yield.] In-process
control indicated HPLC purity 99.3%, associated with 0.28% of
14 and 0.14% of 15. The entire solid was charged back to the
100 L reactor, and acetic acid (13.4 L) was added. The mixture
was heated to 35 °C until all solids dissolved. The solution was
transferred through 0.5 μm in-line filter into a reactor, which
had been set up under speck-free conditions, connected with
Silverson 200HLS high shear wet mill (HSWM) with
Lasentech FBRM probe and preheated to 35 °C. Acetic acid
(2 L) was used to rinse the first reactor. The combined solution
was cooled to 20 °C. Water (61.5 L) was added slowly through
0.5 μm in-line filter over 90 min. The resulting slurry was
agitated at 20 °C for 1 h and cooled to 10 °C. The slurry was
circulated to HSWM with square-holed stator at RPM 5400 for
2 h at 10 °C until meeting particle size target. The milled slurry
was filtered through a Nutsche filter. The cake was washed with
water (10 L) and dried in a tray dryer under vacuum at 60 °C
for 18 h to obtain the title compound 1 as a white powder (2.61
kg, 51.8%). HPLC purity was 99.63%, associated with 0.16% of
14 and 0.13% of 15. Particle Size: D[4, 3] = 25 μm, D[v, 0.95]
= 58 μm. Residual Solvents: acetic acid 0.4 wt %, water 0.1 wt
% and DMF <0.1 wt %. 1H NMR (DMSO-d6) δ 1.93−2.05 (m,
1H), 2.18−2.32 (m, 1H), 2.55−2.65 (m, 2H), 2.68−2.77 (m,
2H), 2.93 (t, J = 6.7 Hz, 2H), 3.83 (s, 3H), 3.88 (t, J = 6.7 Hz,
2H), 7.39 (d, J = 8.6 Hz, 2H), 7.46 (d, J = 8.6 Hz, 2H), 7.78 (d,
J = 3.9 Hz, 1H), 8.32 (d, J = 3.9 Hz, 1H). 13C NMR (DMSO-
d6) δ 17.5, 31.4, 34.6, 47.5, 54.9, 98.8, 125.0, 126.6, 126.7,
137.7, 142.8, 164.9, 165.3, 165.9, 171.0; HRMS (m/z):
calculated for C19H19N5O2, [M + H]+ 350.1612; found
350.1620. Elemental analysis: calculated for C19H19N5O2: C
65.32, H 5.48, N 20.04; found: C 65.40, H 5.45, N 20.16.
1
2 to 4). HPLC purity was 97.9%. H NMR (DMSO-d6) δ
1.92−2.04 (m, 1H), 2.17−2.31 (m, 1H), 2.54−2.64 (m, 2H),
2.66−2.75 (m, 2H), 2.80 (t, J = 6.7 Hz, 2H), 3.78 (t, J = 6.7 Hz,
2H), 7.31 (d, J = 8.6 Hz, 2H), 7.43 (d, J = 8.6 Hz, 2H). 13C
NMR (DMSO-d6) δ 17.5, 29.8, 34.6, 46.2, 84.3, 115.6, 125.0,
126.2, 126.6, 137.3, 142.6, 163.1, 182.3. HRMS (m/z):
calculated for C17H15N3O2, [M + H]+ 294.1237; found
294.1242.
1-(4-(1-Cyanocyclobutyl)phenyl)-4-methoxy-2-oxo-
1,2,5,6-tetrahydropyridine-3-carbonitrile (6). To a stirred,
cooled to 10 °C mixture of dichloromethane (40 L) 4 (5.7 kg,
19.43 mol) and DMF (75 mL, 0.97 mol) was gradually added
neat oxalyl chloride (2.96 kg, 23.32 mol), while maintaining the
reaction temperature below 15 °C. After the addition was
complete, the mixture was warmed to 20 °C and agitated at this
temperature for 2 h. The initial insoluble solid dissolved, and
in-process control indicated that all 4 was converted to 5.
Methanol (34 L) was added in 1 h, while maintaining the
reaction temperature below 30 °C. The mixture was heated to
35 °C in 30 min and agitated at this temperature for 21 h. In-
process control indicated that over 97% of 5 was converted to
6. The solution was concentrated in vacuo to residual volume
∼40 L. Methanol (40 L) was added, and the mixture was
concentrated under vacuum again to residual volume ∼40 L.
Additional methanol (17 L) was added. The resulting slurry
was agitated at 20 °C for 2 h and filtered. The filter cake was
washed with methanol (11 L) and dried in a tray dryer under
vacuum at 40 °C for 18 h to obtain the title compound 6 as a
1
white powder (5.24 kg, 88%). HPLC purity was 96.2%. H
NMR (DMSO-d6) δ 1.96−2.06 (m, 1 H), 2.22−2.33 (m, 1H),
2.57−2.66 (m, 2H), 2.70−2.79 (m, 2H), 3.07 (t, J = 6.6 Hz,
2H), 3.88 (t, J = 6.6 Hz, 2H), 4.05 (s, 3H), 7.38 (d, J = 8.3 Hz,
2H), 7.47 (d, J = 8.3 Hz, 2H). 13C NMR (DMSO-d6) δ 17.5,
25.9, 34.6, 46.0, 58.5, 86.6, 114.9, 125.0, 126.2, 126.6, 137.6,
142.2, 162.2, 183.4; HRMS (m/z): calculated for C18H17N3O2,
[M + H]+ 308.1394; found 308.1402.
1-(4-(4-Amino-2-methoxy-5-oxo-7,8-dihydropyrido-
[ 4 , 3 - d ] p y r i m i d i n - 6 ( 5 H ) - y l ) p h e n y l ) -
cyclobutanecarbonitrile (crude 1). To a stirred mixture of
methanol (61.5 L), 6 (5.21 kg, 16.96 mol), and cyanamide (790
g, 18.82 mol) was added over 30 min a 25% solution of sodium
methoxide in methanol (4.4 kg, 20.35 mol) while maintaining
the reaction temperature below 25 °C. The mixture was
agitated at 20 °C for 2 h. In-process control indicated that over
98% of 6 was converted to 7. Methanesulfonic acid (4.56 kg,
47.48 mol) was added over 30 min while maintaining the
reaction temperature below 30 °C. The temperature was raised
to 50 °C in 2 h, and the mixture agitated at this temperature for
18 h. In-process control indicated that over 98% of 7 was
consumed. The thin slurry was cooled to 20 °C. Water (20 L)
ASSOCIATED CONTENT
■
S
* Supporting Information
NMR spectra of 1 and 4. This material is available free of
charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Authors
*E-mail: andrei.shavnya@pfizer.com
*E-mail: yong.tao@pfizer.com
1515
dx.doi.org/10.1021/op400215h | Org. Process Res. Dev. 2013, 17, 1510−1516

Organic Process Research & Development
Article
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Dr. Yun Huang and Mr. Jack King for analytical
support, Pfizer Groton kilo lab team for successful execution of
the clinical batch preparation, and also Dr. Bryan Li and Dr.
Aaron C. Smith for reviewing of this manuscript.
REFERENCES
■
(1) (a) Birch, A. M.; Buckett, L. K.; Turnbull, A. V. Opin. Drug
Discovery Dev. 2010, 13, 489. (b) Zammit, V. A.; Buckett, L. K.;
Turnbull, A. V.; Wure, H. Pharmacol. Ther. 2008, 118, 295.
(2) (a) Dow, R. L.; Munchhof, M. J. U.S. Patent Appl.2010/0197590.
(b) Aspnes, G. E.; Dow, R. L.; Munchhof, M. J. U.S. Patent Appl.
2010/0197591. (c) Bahnck, K. B.; Shavnya, A.; Tao, Y.; Lilley, S. C.;
Andrews, M. P.; Aspnes, G. E.; Bernhardson, D. J.; Bill, D. R.;
Bundesmann, M. W.; Dow, R. L.; Karki, K.; Le, T.; Li, Q.; Munchhof,
M. J.; Nematalla, A.; Nihlawi, M.; Patel, L.; Perreault, C.; Waldo, M.
Synthesis 2012, 44, 3152.
(3) (a) Yendapally, R.; Hurdle, J. G.; Carson, E. I.; Lee, R. B.; Lee, R.
E. J. Med. Chem. 2008, 51, 1487. (b) Kulkarni, B. A.; Ganesan, A.
Angew. Chem., Int. Ed. 1997, 109, 2565.
(4) (a) Schmidt, H.-W.; Koitz, G.; Junek, H. J. Heterocycl. Chem.
1987, 24, 1305. (b) Perez, M. A.; Soto, J. L.; Guzman, F.; Alcala, H. J.
Chem. Soc., Perkin Trans. 1 1985, 87.
(5) Barbasiewicz, M.; Marciniak, K.; Fedorynski, M. Tetrahedron Lett.
2006, 47, 3871.
(6) (a) Ethyl Acrylate: Material Safety Data Sheet; The Dow
Chemical Company, Jan. 18, 2007, pp 5−6. (b) Acrylate Esters: A
Summary of Safety and Handling, 3rd ed.; Intercompany Committee for
the Safety and Handling of Acrylic Monomers (ICSHAM), 2002, pp
9−10.
(7) T3P is a registered trademark of Archimica GmbH. See www.
archimica.com
(8) (a) Dunetz, J. R.; Xiang, Y.; Baldwin, A.; Ringling, J. Org. Lett.
2011, 13, 5048. (b) Campeau, L.-C.; Dolman, S. J.; Gauvreau, D.;
Corley, E.; Liu, J.; Guidry, E. N.; Ouellet, S. G.; Steinhuebel, D.;
Weisel, M.; O’Shea, P. D. Org. Process Res. Dev. 2011, 15, 1138.
(c) Patterson, D. E.; Powers, J. D.; LeBlanc, M.; Sharkey, T.; Boehler,
E.; Irdam, E.; Osterhout, M. H. Org. Process Res. Dev. 2009, 13, 900.
For a recent review on T3P chemistry: (d) Basavaprabhu;
Vishwanatha, T. M.; Panguluri, N. R.; Sureshbabu, V. V. Synthesis
2013, 45, 1569.
(9) (a) Li, J. J. Vilsmeier mechanism for acid chloride formation.
Name Reactions: A Collection of Detailed Reaction Mechanisms, 3rd
expanded ed.; Springer: New York, 2006; p 605. (b) Kaabak, L. V.;
Baranov, Y. I.; Kalitina, M. I.; Orlov, O. E. Russ. J. Gen. Chem.
(Translation of Zh. Obshch. Khim.) 1998, 68, 117.
(10) ICH Harmonised Tripartite Guideline Impurities: Guideline for
Residual Solvents Q3C(R5); www.ich.org/fileadmin/Public_Web_Site/
ICH_Products/Guidelines/Quality/Q3C/Step4/Q3C_R5_Step4.pdf
(11) Chen, Guoqing P. U.S. Patent Appl. 2004/0259916.
(12) All vacuum exhausts were passed through a scrubber filled with
MeOH/NH4OH/H2O (3:1:1) to destroy remaining ethyl acrylate.
1516
dx.doi.org/10.1021/op400215h | Org. Process Res. Dev. 2013, 17, 1510−1516