Development of a Large-Scale Route to an MCH1 Receptor Antagonist: Investigation of a Staudinger Ketene-Imine Cycloaddition in Batch and Flow Mode

Article abstract of DOI:10.1021/acs.oprd.5b00319

A practical large-scale route to an MCH₁ receptor antagonist is described. A Staudinger β-lactam synthesis of an imine and an in situ generated ketene was utilized as a key step for the preparation of a spiro-azetidine building block. The reaction was demonstrated in both batch and flow mode and a comparison of these techniques is described.

Full text of DOI:10.1021/acs.oprd.5b00319

Article
pubs.acs.org/OPRD
Development of a Large-Scale Route to an MCH1 Receptor
Antagonist: Investigation of a Staudinger Ketene−Imine
Cycloaddition in Batch and Flow Mode
Staffan Karlsson,*,† Rolf Bergman,† Christian Lofberg,† Peter R. Moore,‡ Fritiof Ponten,†
́
̈
Joakim Tholander,† and Henrik Sorensen*,†
̈
†Medicinal Chemistry, Cardiovascular & Metabolic Diseases iMed, AstraZeneca R & D, Molndal SE-431 83, Sweden
̈
‡Global Chemical Development, AstraZeneca R & D, Macclesfield SK10 4NX, United Kingdom
S
* Supporting Information
ABSTRACT: A practical large-scale route to an MCH1 receptor antagonist is described. A Staudinger β-lactam synthesis of an
imine and an in situ generated ketene was utilized as a key step for the preparation of a spiro-azetidine building block. The
reaction was demonstrated in both batch and flow mode and a comparison of these techniques is described.
Staudinger lactam synthesis5 of an N-protected methyl imine
16 and the ketene generated from tetrahydrofuran-3-carbonyl
INTRODUCTION
■
Melanin-concentrating hormone (MCH) receptor antagonists
chloride 14 followed by reduction of the resulting azetidinone
have been subject to extensive studies for the treatment of
17 (Scheme 4).6 Thus, we planned to generate the acid
chloride 14 from the easily accessible commercially available
carboxylic acid 13 and use this as the crude solution in the next
reaction with N-methylene-1-phenylmethanamine 16.
obesity1,2 and have resulted in many patents.3,4 During our
screening program, compound 1 (Scheme 1) was identified for
extended studies as an MCH1 receptor antagonist, and more
than 300 g were required.
A prerequisite was to find a clean and selective way to
transform the carboxylic acid to the acid chloride as it was
envisaged that use of excess reagent in the acid chloride
formation may have a negative impact on the subsequent
Staudinger synthesis. We found that. in the presence of a
catalytic amount of pyridine, an almost stoichiometric amount
of thionyl chloride as chlorinating agent was enough to give
complete the conversion to desired acid chloride 14 within 90
min. The N-methylene-1-phenylmethanamine 16 required for
the Staudinger synthesis was easily obtained from precursor 15
through treatment with BF3-OEt2.7 The imine was used as a
freshly prepared solution in DCM, as slow decomposition was
observed over time. Our initial efforts with the reaction of acid
chloride 14 and imine 16 failed, and no trace of product 17
could be detected. Several attempts were made to run the
reaction at different temperatures and with different orders of
addition of reagents. In the event, successful reaction was
achieved when all reagents were mixed at −78 to −37 °C
followed by slow warming of the reaction mixture. At
approximately −10 °C, a sudden and quite exothermic reaction
was observed resulting in clean conversion to desired lactam 17
in good yield. Using this procedure, the reaction could be
reliably performed on a 1.3 mol scale. Through three repeated
syntheses on this scale, we were able to deliver >700 g of lactam
17 in 87% yield, which supported our initial needs. Despite the
successful result, we were concerned about further scale-up of
this reaction. In addition to the cryogenic conditions required,
which are difficult to apply on a larger scale, we believed it
First Generation Synthesis. The approach shown in
Scheme 1 for the first generation synthesis of 1 was a logical
choice in the sense that the attachment of amine A to fragment
BCD was made as the last step through reductive amination,
making the synthesis of an array of screening compounds with
different terminal amines a seemingly simple task. The detailed
first generation synthesis was performed as shown in Scheme 2.
For scale-up purposes, the route was deemed impractical for
several reasons: (a) toxic sodium cyanide was used as a catalyst
for the formation of amide 3; (b) aldehyde 5 was practically
insoluble in all solvents except for DCM, and its isolation by
filtration took several hours, even on less than a 10 g scale; (c)
HPLC purification of the poorly soluble API 1 required very
large amounts of solvent; and (d) the sequence for preparation
of building block 6, depicted in Scheme 3, encompassed an
average yielding anion formation/alkylation of 8 at −78 °C, a
hazardous OsO4-mediated oxidation of 9, an impractical DIAD
(diisopropyl azodicarboxylate)-mediated Mitsunobu reaction of
diol 11, and two time-consuming chromatographic purifica-
tions.
RESULTS AND DISCUSSION
■
Development of a Staudinger Ketene−Imine Cyclo-
addition for the Synthesis of Building Block 6. With a
limited time frame for optimization of the first generation
synthesis, we decided to focus on finding a scale-up friendly
route to spiro-azetidine 6 while retaining most of our previously
developed route but in a modified sequence to avoid problems
associated with aldehyde 5. Our goal was to find a scalable,
robust, safe, and chromatography-free synthesis. We envisioned
that spiro-azetidine building block 6 could be obtained via a
Received: October 6, 2015
© XXXX American Chemical Society
A
DOI: 10.1021/acs.oprd.5b00319
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 1. Retrosynthetic Strategies for the MCH1 Receptor Antagonist 1
Scheme 2. First Generation Synthesis of the MCH1 Receptor Antagonist 1
Scheme 3. First Generation Synthesis of Spiro-Azetidine 6
Scheme 4. Large Scale Synthesis of Spiro-Azetidine 6
would be difficult to control the exotherm resulting from an “all
in” procedure. Also, the addition of Et3N to the acidic mixture
of acid chloride 14 was highly exothermic and would be very
time-consuming on a larger scale. We realized that for long-
term supply of lactam 17 an alternative strategy was required. A
Staudinger synthesis in flow was considered as a safe
alternative.5f,8 Reaction in flow would benefit from efficient
heat transfer of the highly exothermic reaction, and the sudden
B
DOI: 10.1021/acs.oprd.5b00319
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
exotherm associated with this reaction would also be easier to
control in a flow mode. Therefore, an experiment was set up in
which preformed acid chloride 14 was mixed in a 0.6 mL glass
chip cooled at 0 °C with preformed imine 16 (1 equiv)
followed by reaction with Et3N (neat, 3 equiv) in a second 0.6
mL glass chip cooled at 0 °C. Although the lactam was obtained
in good quality, the flow conditions resulted in heavy
precipitation of salts that quickly blocked the glass chip and
made the operation impossible. Running the reaction using
more dilute conditions solved the problem with precipitation
but delivered product 17 in low yield and with poor quality. In
the search for a more suitable base for this reaction, six different
amines that did not give rise to precipitation were identified. N-
Methylpiperidine was selected on the basis that it provided the
In batch mode, although N-methylpiperidine gave a lower
yield (68%) of product 17 compared with Et3N (87%), the
quality after extractions was comparable, and therefore, further
time-consuming purification procedures of the crude mixture
were unnecessary. Using N-methylpiperidine as the base,
optimization of the Staudinger synthesis under continuous
processing conditions using an in-line IR detector was then
undertaken (Scheme 5).
We found that lactam 17 had a specific CO IR-band (1750
cm−1) different from that of starting acid chloride 14 (1795
cm−1). Consequently, these wave numbers were used for
monitoring the relative amounts of each of these components
in the eluent (Figure 1).
As shown in Figure 1, the Staudinger synthesis of acid
chloride 14 and imine 16 was best performed at temperatures
of 20 °C (at points c, g, and h). Significantly lower yield was
indicated when the reaction was performed at 0 °C (at b).
During the evaluated conditions, the residence time was of
minor importance, although marginally higher yield of the
product was indicated at lower flow rates. Furthermore, it was
found that the best way to perform the reaction was to mix all
of the reagents at the same time to minimize the risk of
undesired side reactions. For example, premixing of acid
chloride 14 with the amine base followed by reaction with
imine 16 gave no product, presumably due to ketene formation
followed by oligomerization. Premixing of 16 with 14 followed
by mixing with the amine base was also found to be inferior for
the outcome of the reaction, and premixing of the amine base
with 16 resulted in regeneration of triazinane imine-precursor
15. Using the optimized conditions10 in Figure 1 at 20 °C and
using a flow rate of 1 mL/min of acid chloride (1.05 M) 14,
1.31 mL/min of imine 16 (0.8 M, 1 equiv), and 0.38 mL/min
of N-methylpiperidine (3 equiv), we were able to run the
reaction continuously for 5 h with no interruptions. By using
in-line FT-IR monitoring of the reaction, we could conclude
that the yield of 17 was stable over time. However, we observed
that the starting pregenerated imine 16 was not stable over time
at room temperature, and therefore, this was used as an ice-
cooled solution during the course of the reaction.11 After
workup of the continuous 5 h reaction, we isolated the desired
lactam in 56% effective yield. The yield is comparable to what
was obtained in batch mode with N-methylpiperidine but
inferior to the batch mode using Et3N (87% effective yield) as
the amine base. We then searched for a method to reduce
lactam 17 to the corresponding azetidine 18.12 The reducing
agents BH3-DMS and LiAlH4 both resulted in a significant
amount of the amino alcohol byproduct resulting from ring-
opening and reduction of the lactam. Instead, we turned our
attention to a milder reduction of the lactam using a
combination of LiAlH4 and AlCl3.13 To our delight, it was
found that this method selectively gave desired azetidine 18 in
good yield. Much time was then spent on finding optimal
conditions for the workup and removal of Al salts. It was found
that quenching with EtOAc followed by the addition of 2-
aminoethanol furnished a suspension of Al salts that could be
removed through filtration. The reaction was run on a 500 g
scale, and desired azetidine 18 could be obtained in 81%
effective yield. As the crude product was obtained in high
quality (90% w/w), no further purification was necessary. In
the subsequent step, our plan was to remove the benzyl
protecting group through conventional hydrogenation using
Pd/C as catalyst. Because azetidine 6 is very hydrophilic,
extractions should be avoided. Thus, after a filtration of the Pd
1
required product at the highest quality as determined by H
NMR (Table 1).
Table 1. Screen of Amine Bases for the Staudinger Synthesis
a
in Flow
a
Conditions: acid chloride 14 and preformed imine 16 were each
pumped at a rate of 1 mL/min as 0.86 M solutions in DCM. These
were allowed to mix in a 0.6 mL glass reactor chip at 0 °C and then
reacted in a second ice-cooled chip (0.6 mL) with the amine bases
(neat, 3 equiv) followed by collection into small vials. 1H NMR
(CDCl3) was recorded on the crude mixtures.
C
DOI: 10.1021/acs.oprd.5b00319
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 5. Photo and Schematic View of the Experimental Setup
catalyst, we hoped to be able to use resulting product 6 in the
next step without further purification. However, after screening
several catalysts, such as PtO2, Pd/C, and Pd(OH)2, for this
transformation under different hydrogen pressures, we found
that either incomplete reactions or mixtures of products were
obtained in combination with inconsistent results. Fortunately,
transfer hydrogenation conditions using formic acid or salts
thereof in the presence of large amounts of Pd/C catalyst
selectively furnished desired product 6. The most consistent
results were obtained using highly diluted ammonium formate
in EtOH as a hydrogen source. Because of time constraints, this
method was selected for further scale-up despite the obvious
drawbacks in terms of high Pd/C loading and large reaction
volume. After the reaction was complete, azetidine 6 was
protected as the BOC derivative 12 to allow for aqueous
washing and removal of salts. During this operation, some
reaction of ammonia with BOC2O was also observed. However,
this byproduct did not interfere in subsequent steps. Desired
BOC-protected azetidine building block 12 was obtained in
87% effective yield (2 steps) on a 400 g scale. We found that no
further purification of building block 12 was necessary (80% w/
w), and the material obtained was used as such in the following
step. By treatment of 12 with 6.5 equiv of TFA, we could, after
concentration, isolate compound 6 as a trifluoroacetate salt.
Large-Scale Synthesis of 1. During the initial large-scale
development work, it was observed that aminolysis of
carboxylic ester 2 could take place without cyanide catalysis.
For scaling up, it was reasoned that this facile transformation
may be utilized in aminolysis of oxadiazole ester 2 as a final
step, thus avoiding the virtually insoluble14 and difficult to
isolate aldehyde 5. Thus, intermediate aldehyde 21 containing
fragments B and C (Scheme 1 and 6) was targeted as our first
intermediate for large-scale synthesis. Two routes to aldehyde
21 were explored. The first route, shown in Scheme 6, gave 21
in excellent yield by alkylation of 4-hydroxybenzaldehyde with
mesylate 20. Mesylate 20 was readily prepared from BOC-
protected azetidin-3-ol 19 under standard conditions.
However, we found an alternative route to aldehyde 21 by
the aromatic nucleophilic substitution of 4-fluorobenzaldehyde
with 19 in the presence of potassium hydroxide (Scheme 7).
Despite a lower yield (67%) for the latter route, it was chosen
for our large-scale synthesis based on facile isolation of
aldehyde 21 and avoidance of the additional step of mesylation
of 19. Spiroazetidine 6 was subsequently reductively alkylated
with aldehyde 21 to give BOC-protected amine 22. This
fragment was deprotected using TFA, and after neutralization,
free amine 23 was isolated as an oil after extractive workup. The
final step was accomplished by facile reaction of amine 23 with
known ester 215 in MeOH. Crystalline API 1 was isolated by
simple filtration of the reaction mixture. Ester 2 was easily
prepared using a chromatography-free one pot procedure as
shown in Scheme 8 starting from 4-methoxybenzohydrazide via
in situ generated 24. Compound 24 was ring closed with
thionyl chloride to give ester 2.
D
DOI: 10.1021/acs.oprd.5b00319
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
reaction by FT-IR spectroscopy in real time facilitated the
optimization process.
EXPERIMENTAL SECTION
■
All materials were purchased from commercial suppliers and
used as such without further purification. All reactions were
performed under an atmosphere of nitrogen. Continuous flow
reactions were performed in PFA coiled tube reactors using a
Vaportec instrument (E-series) coupled to a Mettler Toledo
React IR 15 instrument integrated with a DiComp (diamond)
in-line flow cell. IPCs (In Process Control) were recorded
1
either by HPLC or H NMR analysis of the crude reaction
1
mixtures. Assays were determined by H NMR integration
Figure 1. IR recording of eluent from the flow reaction as pictured in
Scheme 5. Compounds 14 (1.05 M) and 16 (0.8 M) were prepared as
solutions in DCM, and N-methylpiperidine was used as the neat liquid.
The three solutions were mixed in a 4-way junction and then allowed
to react in a 2 mL standard coil tube reactor followed by the IR
detector, 75 cm tubing, a BPR (back pressure regulator), and another
32 cm tubing.9 Arrows indicate (a) 14, 1.05 M, 1 mL/min, 0 °C; (b)
14, 1 mL/min, 16, 1.31 mL/min, and N-methylpiperidine, 0.38 mL/
min, 0 °C; (c) 14, 1 mL/min, 16, 1.31 mL/min, and N-
methylpiperidine, 0.38 mL/min, 20 °C; (d) 14, 1 mL/min, 16, 1.31
mL/min, and N-methylpiperidine, 0.38 mL/min, 40 °C; (e) 14, 4 mL/
min, 16, 5.24 mL/min, and N-methylpiperidine, 1.52 mL/min, 40 °C;
(f) 14, 2 mL/min, 16, 2.62 mL/min, and N-methylpiperidine, 0.76
mL/min, 40 °C; (g) 14, 1 mL/min, 16, 1.31 mL/min, and N-
methylpiperidine, 0.38 mL/min, 20 °C; and (h) 14, 0.5 mL/min, 16,
0.66 mL/min, and N-methylpiperidine, 0.19 mL/min, 20 °C.
using benzyl benzoate or maleic acid as internal standards. High
resolution mass spectrometry was performed on a QTOF 6530
instrument (Agilent), mass precision 5 ppm. NMR measure-
ments were performed using a Bruker Avance III spectrometer.
LC/MS analyses were recorded on a Waters ZMD, LC column
x Terra MS C8 (Waters) detection with an HP 1100 MS-
detector diode array.
2-Benzyl-6-oxa-2-azaspiro[3.4]octan-1-one (17),
Batch Mode. A 2 L 3-necked round bottomed flask was
charged with carboxylic acid 13 (152 g, 1.31 mol), DCM (1.1
L), and pyridine (2.2 mL, 26 mmol). The mixture was heated
to 30 °C in a water bath. Sulfurous dichloride (100 mL, 1.37
mol) was added over 10 min. The reaction temperature was
1
maintained in the interval of 25 to 30 °C. After 90 min, H
NMR analysis of the crude mixture indicated full conversion to
the corresponding acid chloride 14. The mixture was cooled in
a dry ice bath. At a reaction temperature of −72 °C, Et3N (397
g, 3.93 mol) was slowly added while keeping the reaction
temperature below −50 °C. In a separate flask at 20 °C, to a
solution of 1,3,5-tribenzyl-1,3,5-triazinane 15 (159 g, 0.44 mol)
in DCM (250 mL) was added BF3-OEt2 (186 g, 1.31 mol)
within 5 min. After 20 min of stirring, 1H NMR analysis of the
crude mixture indicated full conversion to desired intermediate
N-methylene-1-phenylmethanamine 16. This mixture was now
slowly (15 min) added to the crude dark brown solution of acid
chloride 14 above. The reaction temperature was kept below
−37 °C during this addition. The mixture was slowly allowed to
attain room temperature in the ice bath. When the reaction
temperature reached approximately −10 °C, a rapid increase in
Scheme 6. Alternative Route to Aldehyde 21
CONCLUSIONS
■
A convergent large-scale route to MCH1 receptor antagonist 1
was developed in 37% overall yield. Compared with the first
generation synthesis, key achievements were the exclusion of
chromatography in the sequence and the development of a new
route to spiroazetidine 6 in which the handling of toxic and
hazardous reagents were avoided. The sensitive Staudinger
synthesis used for the formation of spiroazetidinone 17 was
optimized in both batch and flow mode, and monitoring of the
1
reaction temperature to 12 °C was detected. H NMR analysis
of the crude mixture indicated full conversion of imine 16 to
desired product 17. Water (500 mL) was added, and the
1
biphasic mixture was stirred vigorously overnight. H NMR
analysis of the crude mixture indicated complete decomposition
of an unwanted Et3N-BF3 complex. The organic layer was
washed with 2 M KHSO4 solution (aq, 500 mL). Some
Scheme 7. Large-Scale Synthesis of 1
E
DOI: 10.1021/acs.oprd.5b00319
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 8. Synthesis of Oxadiazole Ethyl Ester 2
precipitate formed was removed by filtration. The organic layer
was washed with NaHCO3 solution (aq, sat., 500 mL) and
water (500 mL) followed by concentration to give the title
compound as a pale brown nonviscous oil (257.6 g, 96% w/w,
mol) dissolved in THF (0.5 L) was slowly added over 1.5 h.
The reaction temperature was kept below 10 °C during this
1
addition. After an additional 1 h of stirring at 5 °C, H NMR
analysis indicated full conversion. With the mantle temperature
set at −5 °C, the reaction was quenched by careful addition of
EtOAc (311 mL) over 1.5 h. The reaction temperature was
kept below 10 °C during this addition. The mixture was stirred
at 5 °C for 24 h followed by cooling to 0 °C. 2-Aminoethanol
(1039 g) was added over 1 h. At 10 °C, the resulting white
thick suspension was stirred for 22 h followed by filtration and
rinsing with THF (2 L). The filtrate was concentrated. The
residue was dissolved in DCM (3 L) followed by washing with
water (1.7 L). The organic layer was filtered through a Celite
filter (6 μm) followed by concentration. Water present was
azeotropically removed using DCM (0.5 L). This furnished the
title compound as a yellow oil (414 g, 90% w/w, 81% effective
1
87% effective yield). H NMR (400 MHz, CDCl3): δ 7.20−
7.38 (m, 5H), 4.42 (d, J = 15.0 Hz, 1H), 4.36 (d, J = 15.0 Hz,
1H), 4.01 (d, J = 9.3 Hz, 1H), 3.97 (d, J = 9.3 Hz, 1H), 3.83−
3.93 (m, 2H), 3.24 (d, J = 5.5 Hz, 1H), 3.20 (d, J = 5.5 Hz,
1H), 2.42 (dt, J = 7.3, 12.8 Hz, 1H), 2.11 (dt, J = 6.6, 12.8 Hz,
1H). 13C NMR (101 MHz, CDCl3): δ 170.1, 135.3, 128.8 (2
Cs), 127.9 (2 Cs), 127.7, 72.0, 68.2, 59.9, 53.6, 45.8, 32.8.
HRMS (ESI): [M + H]+ m/z calcd for C13H16NO2, 218.1181;
found, 218.1171.
2-Benzyl-6-oxa-2-azaspiro[3.4]octan-1-one (17), Flow
Mode. Reaction performed in a continuous flow tube reactor
(2 mL) coupled to an FT-IR flow cell. Under ice-cooling, N-
methylene-1-phenylmethanamine 16 was generated from 1,3,5-
tribenzyl-1,3,5-triazinane 15 and BF3-OEt2 (3 equiv) using the
same procedure as above. The resulting imine was diluted with
DCM to give a final concentration of the imine of 0.8 M. This
was used freshly as an ice-cooled solution. Acid chloride 14 was
prepared using the same procedure as described above and was
used as a 1.05 M solution in DCM. N-Methylpiperidine was
used as the amine base for the Staudinger synthesis and
pumped as a neat solution. The imine (0.8 M, 1.31 mL/min),
acid chloride (1.05 M, 1.0 mL/min), and N-methylpiperidine
(neat, 0.38 mL/min) were mixed in a 4-way junction and then
directly reacted in a 2 mL tube reactor followed by in-line
recording with an FT-IR flow cell. This was followed by
collection of the stream into a 2 L receiver flask equipped with a
magnetic stir bar containing 600 mL of water. The back
pressure regulator was set at 1.5 bar, and the temperature of the
reactor was set at 20 °C. The continuous reaction was run for
295 min corresponding to 41.7 g (310 mmol) of starting acid
chloride 14 consumed. The biphasic product mixture was
stirred at 20 °C over the weekend followed by separation of the
layers. The organic layer was extracted with KHSO4 solution
(2M, 500 mL) followed by NaHCO3 (aq, sat., 300 mL) and
water (200 mL). The organic layer was concentrated to give the
title compound as a pale yellow nonviscous oil (44 g, 85% w/w,
56% effective yield). NMR data were in agreement with those
given above.
1
yield). H NMR (400 MHz, MeOD): δ 7.24−7.34 (m, 5H),
3.79 (s, 2H), 3.74 (t, J = 7.0 Hz, 2H), 3.63 (s, 2H), 3.26−3.30
(m, 4H), 2.09 (t, J = 7.0 Hz, 2H). 13C NMR (MeOD, 101
MHz): δ 138.4, 129.9 (2 Cs), 129.5 (2 Cs), 128.4, 77.9, 68.3,
64.7 (2 Cs), 64.2, 42.4, 38.6. HRMS (ESI): [M + H]+ m/z
calcd for C13H18NO, 204.1388; found, 204.1387.
tert-Butyl 6-Oxa-2-azaspiro[3.4]octane-2-carboxylate
(12). A 25 L reactor was charged with Pd/C (10% on charcoal,
390 g) followed by the addition of 2-benzyl-6-oxa-2-
azaspiro[3.4]octane 18 (414 g, 90% w/w, 1.83 mol) dissolved
in EtOH (16.5 L). Ammonium formate (462 g, 7.33 mol)
dissolved in water (4.1 L) was added, and the mixture was
heated at 50 °C for 20 h. 1H NMR analysis of the crude mixture
indicated >98% conversion to amine 12. To the crude mixture
at 10 °C were added Et3N (555 g, 5.5 mol) and di-tert-butyl
1
dicarbonate (583 g, 2.67 mol) in small portions over 2 h. H
NMR analysis of the crude mixture indicated >98% conversion.
The reaction mixture was filtered through a Celite filter (Seitz,
K200) and rinsed with EtOH (2 × 2.5 L). The filtrate was
concentrated to dryness. To the residue were added MTBE (4
L) and water (2 L). The pH of the aqueous layer was adjusted
to ∼7 using 3.8 M NaOH solution. The organic layer was
filtered through a Celite filter followed by concentration to give
the title compound as a pale yellow oil (423 g, 80% w/w, 87%
1
effective yield). H NMR (600 MHz, MeOD): δ 4.85 (s, 2H),
3.86−3.92 (m, 3H), 3.82 (s, 2H), 3.81 (t, J = 7.0 Hz, 1H), 2.15
(t, J = 7.0 Hz, 2H), 1.44 (s, 9H). 13C NMR (MeOD, 101
MHz): δ 158.0, 81.0, 79.7, 77.7, 68.4, 41.5, 38.6, 28.6. HRMS
(ESI): [M + H]+ m/z calcd for C11H20NO3, 214.1443; found,
214.1448.
tert-Butyl 3-Hydroxyazetidine-1-carboxylate (19). In a
5 L reactor, azetidine-3-ol hydrochloride (301 g, 2.75 mol) was
dissolved in MeOH (2.25 L) and Et3N (508 mL, 3.67 mol) at
20 °C. The mixture was cooled to −17 °C, and di-tert-butyl
dicarbonate (500 g, 2.29 mol) was added portion wise at a rate
such that the reaction temperature was kept below −5 °C. The
2-Benzyl-6-oxa-2-azaspiro[3.4]octane (18). In a 10 L
reactor, to a −20 °C solution of THF (2.8 L) was added in
small portions AlCl3 (368.8 g, 2.76 mol). With a mantle
temperature set at −5 °C, to the white suspension was slowly
added a solution of LiAlH4 (1.0 M in THF, 2.72 L, 2.72 mol)
over 1.5 h. The reaction temperature was kept below 3 °C
during this addition. The resulting homogeneous solution was
stirred at 20 °C for 1 h followed by recooling to −2 °C. With
the mantle temperature set at −10 °C, a solution of 2-benzyl-6-
oxa-2-azaspiro[3.4]octan-1-one 17 (507.8 g, 97% w/w, 2.27
F
DOI: 10.1021/acs.oprd.5b00319
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
mixture was stirred at 0 °C for 17 h and then concentrated at
40 °C to give a residue that was dissolved in EtOAc (1.6 L).
The solution was washed with water (0.6 L), 10% aqueous
citric acid (2 × 0.6 L), 4% aqueous NaHCO3 solution (1 L),
and 8% aqueous NaHCO3. The EtOAc was removed under
reduced pressure with the aid of heptane (3 × 320 mL) to give
the title compound as a white solid (325.9 g, 100% w/w, 82%
solution, 3.05 mol) was added followed by a mixture of toluene
(750 mL) and heptane (750 mL). After thorough mixing, the
aqueous phase had a pH of 4.81. The organic phase was
removed, and the aqueous one was washed with another
portion of 150 mL of toluene and 150 mL of heptane. After
separation, MTBE (1000 mL) was added to the aqueous phase
followed by 50% sodium hydroxide (403 g, 5 equiv) solution to
give a pH of 9.5. To the aqueous phase were added another
portion of MTBE (1000 mL) and sodium hydroxide (120 g
50% solution, 1.5 equiv) to give a pH of 9.74. The organic
phases were pooled, dried (potassium carbonate), and then
filtered through Celite. The solution was concentrated to 1.5 L
and diluted to 4.5 L with heptane. The mixture was heated to
45 °C, then cooled to 5 °C over 3 h and stirred at 5 °C for 16
h. The crystals formed were isolated by filtration, washed with
heptane (400 mL), and dried under reduced pressure at 25 °C
to give the title compound as colorless crystals (211 g, 97% w/
w, 56% effective yield). By reworking the mother liquor and the
lining of the reactor, another crop (60 g, 97% w/w, 16%
1
effective yield). H NMR (400 MHz, CDCl3): δ 1.46 (s, 9H),
3.83 (dd, J = 10.0, 4.4 Hz, 2H), 4.15 (dd, J = 10.0, 6.7 Hz, 2H),
4.58 (m, 1H). 13C NMR (101 MHz, CDCl3): δ 156.62, 79.89,
61.23, 59.02, 28.45.
tert-Butyl 3-(4-Formylphenoxy)azetidine-1-carboxy-
late (21). In a 10 L reactor at a temperature of 20−25 °C,
solid potassium hydroxide (86%, 313 g, 4.8 mol) was added
over a 20 min period (slightly exothermic reaction) to a
solution of tert-butyl 3-hydroxyazetidine-1-carboxylate 19 (343
g, 1.98 mol) and 4-fluorobenzaldehyde (248 g, 2.00 mol) in
DMF (1.8 L). The mixture was stirred for 1.5 h at 20 °C
whereupon MTBE (7 L) and water (1 L) were added. The
organic phase was washed successively with water (3 × 1 L),
5% aqueous citric acid (200 mL), 8% aqueous NaHCO3 (200
mL), and brine (500 mL). The mixture was dried (magnesium
sulfate) and concentrated to approximately 3 L followed by the
addition of heptane (1 L). The mixture was stirred for 1 h at 20
°C, cooled to 0 °C, and diluted with more heptane (2 L). The
suspension formed was filtered, and the crystals isolated were
1
effective yield) was isolated. H NMR (500 MHz, MeOD): δ
1.45 (s, 9H), 2.10 (t, J = 7.0, 7.0 Hz, 2H), 3.27 (dd, 4H), 3.57
(s, 2H), 3.75 (t, J = 7.0, 7.0 Hz, 2H), 3.79 (s, 2H), 3.84−3.92
(m, 2H), 4.28−4.37 (m, 2H), 4.91−4.99 (m, 1H), 6.77 (d, J =
8.6 Hz, 2H), 7.23 (d, J = 8.6 Hz, 2H). 13C NMR (126 MHz,
CDCl3): δ 155.91, 155.50, 130.95, 129.66, 114.22, 79.52, 77.10,
67.23, 65.45, 64.03, 62.73, 56.18, 41.29, 37.66, 28.19. HRMS
(ESI): [M + H]+ m/z calcd for C21H31N2O4, 375.2284; found,
375.2296.
1
dried to give the title compound (369 g, 67%). H NMR (400
MHz, CDCl3): δ 1.45 (s, 9H), 4.03 (dd, J = 10.5, 4.1 Hz, 2H),
4.34 (m, 2H), 4.96 (tt, J = 6.4, 6.4, 4.1, 4.1 Hz, 1H), 6.85 (d, J =
8.7 Hz, 2H), 7.84 (d, J = 8.8 Hz, 2H), 9.90 (s, 1H). 13C NMR
(101 MHz, CDCl3): δ 190.65, 161.52, 156.04, 132.14, 130.69,
114.97, 80.05, 66.18, 56.22, 28.37.
2-(4-(Azetidin-3-yloxy)benzyl)-6-oxa-2-azaspiro[3.4]-
octane (23). In a 5 L reactor, trifluoroacetic acid (860 g, 7.54
mol) was added at 30 °C over a 50 min period to a solution of
tert-butyl 3-(4-(6-oxa-2-azaspiro[3.4]octan-2-ylmethyl)-
phenoxy)azetidine-1-carboxylate 22 (321 g, 0.84 mol) in
DCM (400 mL). After 3 h of reaction time from the final
addition of trifluoroacetic acid, the reaction mixture was cooled
to 5 °C, and sodium hydroxide solution (1.48 L of a 6.25 M
solution in water, 9.25 mol) was added portion wise
(exothermic). The mixture was stirred for 15 min at 20 °C
followed by extractions with DCM (250 mL + 2 × 500 mL),
eventually adding MeOH (250 mL) to facilitate separation of
the phases. The combined organic extracts were dried
(potassium carbonate), and the solvent was removed under
reduced pressure, eventually with the aid of MeOH to give the
title compound as a pale oil (251.6 g, 80% w/w, 87% effective
yield). 1H NMR (400 MHz, MeOD): δ 2.09 (t, J = 7.0, 7.0 Hz,
2H), 3.2−3.29 (m, 4H), 3.55 (s, 2H), 3.62−3.68 (m, 2H), 3.74
(t, J = 7.0, 7.0 Hz, 2H), 3.78 (s, 2H), 3.86−3.95 (m, 2H), 5.00
(p, J = 6.2, 6.2, 6.1, 6.1 Hz, 1H), 6.74 (d, J = 8.5 Hz, 2H), 7.21
(d, J = 8.5 Hz, 2H). 13C NMR (151 MHz, MeOD): δ 157.66,
131.32, 131.10, 115.60, 77.88, 71.57, 68.26, 64.54, 63.48, 54.75,
42.33, 38.58.
6-Oxa-2-azaspiro[3.4]octane (6) x 2 TFA. In a 3 L
reactor, tert-butyl 6-oxa-2-azaspiro[3.4]octane-2-carboxylate 12
(260 g, 80% w/w, 975 mmol) was added at a temperature of
18−20 °C to trifluoroacetic acid (600 mL) over a 30 min
period. The reaction was slightly exothermic and effervescent.
The temperature was increased to 25 °C, and the mixture was
allowed to stir for an additional 135 min. The trifluoroacetic
acid was removed by coevaporation with 3 × 1000 mL MeOH
at a temperature of 35 °C to give the title compound as a
bis(trifluoroacetate) pale oil (351 g, 95% w/w, 100% effective
yield). 1H NMR (400 MHz, MeOD): δ 2.25 (t, J = 7.1, 7.1 Hz,
2H), 3.80 (t, J = 7.1, 7.1 Hz, 2H), 3.89 (s, 2H), 4.09 (s, 4H).
13C NMR (101 MHz, MeOD): δ 162.41, 117.66, 77.02, 68.39,
56.39, 44.80, 38.31.
tert-Butyl 3-(4-(6-Oxa-2-azaspiro[3.4]octan-2-
ylmethyl)phenoxy)azetidine-1-carboxylate (22). A 10 L
reactor was charged with 6-oxa-2-azaspiro[3.4]octane bis-
(trifluoroacetate) 6 x 2 TFA (351 g, 975 mmol), EtOAc (2.5
L), and DIPEA (455 g, 3.52 mol). tert-Butyl 3-(4-
formylphenoxy)azetidine-1-carboxylate 21 (275 g, 991 mmol)
was added portion wise over a 15 min period to give a clear
solution. The mixture was cooled to 17 °C, and sodium
triacetoxyborohydride (255 g, 1.2 mol) was added over a 10
min period. The reaction was slightly exothermic, and the
temperature reached 27 °C during the addition. The mixture
was stirred at 20 °C for 20 h. Another portion of sodium
triacetoxyborohydride (54 g, 255 mmol) was added, which
completed the reaction as determined by 1H NMR after
another 24 h reaction time. Water (1.350 L) was added, and
most of the organic solvent was removed under reduced
pressure at 35 °C. Dilute acetic acid (3.66 L of a 5% aqueous
(3-(4-(6-Oxa-2-azaspiro[3.4]octan-2-ylmethyl)-
phenoxy)azetidin-1-yl)(5-(4-methoxyphenyl)-1,3,4-oxa-
diazol-2-yl)methanone (1). In a 10 L reactor, ethyl 5-(4-
methoxyphenyl)-1,3,4-oxadiazole-2-carboxylate 2 (182 g, 732
mmol) was added at 20 °C over 1 min to a solution of 2-(4-
(azetidin-3-yloxy)benzyl)-6-oxa-2-azaspiro[3.4]octane 23 (251
g 80% w/w, 732 mmol) in MeOH (2.5 L). Precipitation was
evident after 5 min. After 13 h, the crystals formed were
isolated by filtration and washed with MeOH (0.5 L). Drying
under reduced pressure at 35 °C gave the title compound as
1
off-tan crystals (334.6 g, 99.6% w/w, 96% effective yield). H
NMR (400 MHz, CDCl3): δ 2.10 (t, J = 6.9, 6.9 Hz, 2H),
G
DOI: 10.1021/acs.oprd.5b00319
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
3.16−3.29 (m, 4H), 3.56 (s, 2H), 3.77 (t, J = 6.9, 6.9 Hz, 2H),
3.84 (s, 2H), 3.90 (s, 3H), 4.26−4.37 (m, 1H), 4.64 (ddd, J =
11.3, 6.2, 1.6 Hz, 1H), 4.74 (ddd, J = 10.9, 3.5, 1.5 Hz, 1H),
5.01−5.16 (m, 2H), 6.68−6.78 (m, 2H), 6.98−7.07 (m, 2H),
7.23 (d, J = 8.6 Hz, 2H), 8.06−8.14 (m, 2H). 13C NMR (101
MHz, CDCl3): δ 165.54, 163.03, 157.28, 155.33, 153.26,
131.76, 129.90, 129.44, 115.19, 114.63, 67.48, 66.17, 64.34,
62.94, 60.92, 56.09, 55.53, 41.51, 37.89. HRMS (ESI): [M +
H]+ m/z calcd for C26H29N4O5, 477.2138; found, 477.2148.
Ethyl 2-(2-(4-Methoxybenzoyl)hydrazinyl)-2-oxoace-
tate (24). In a 5 L reactor, Et3N (304 g, 3.00 mol) was
added to 4-methoxybenzohydrazide (200 g, 1.2 mol) in anisole
(1.8 L). Ethyl 2-chloro-2-oxoacetate (197 g, 1.44 mol) was
added over a period of 1 h. The addition was slightly
exothermic. After addition, the mixture was heated and kept at
50 °C for 1 h. The mixture was used in the subsequent step
without further processing or characterization.
Ethyl 5-(4-Methoxyphenyl)-1,3,4-oxadiazole-2-car-
boxylate (2).14 At a temperature of 20 °C, thionyl chloride
(174 g, 1.44 mol) was added to the reaction mixture from the
preparation of intermediate ethyl 2-(2-(4-methoxybenzoyl)-
hydrazinyl)-2-oxoacetate 24. The temperature was increased to
95 °C. After a period of 3 h, minor amounts of intermediate 24
were still present as indicated by HPLC-MS (pH 3). Addition
of more thionyl chloride (29 g, 0.24 mol) followed by
continued heating at 95 °C for another 2 h completed the
reaction. The mixture was cooled to 10 °C, and water (1.6 L)
and magnesium sulfate (160 g) were added followed by an
adjustment of the temperature to 40 °C. The phases were
separated and the organic phase was washed twice with
aqueous sodium carbonate (190 g dissolved in 1.6 L of water in
each wash). The mixture was concentrated to 1 L, and heptane
(1.6 L) was added followed by heating to 55 °C for 1 h. The
mixture was cooled to 10 °C over a period of 4 h and then
stirred at 10 °C for 16 h. Filtration and washing with heptane
(2 × 640 mL) followed by drying under reduced pressure at 40
°C for 16 h gave the title compound as an off white solid (202
g, 99% w/w, 67% effective yield over two steps). 1H NMR (400
MHz, CDCl3): δ 1.48 (t, J = 7.1, 7.1 Hz, 3H), 3.89 (s, 3H),
4.54 (q, J = 7.1, 7.1, 7.1 Hz, 2H), 6.97−7.1 (m, 2H), 8.04−8.16
(m, 2H). 13C NMR (126 MHz, CDCl3, 30 °C): δ 166.56,
163.31, 156.22, 154.64, 129.61, 115.26, 114.79, 63.48, 55.64,
14.21.
Process Development, Forskargatan 20J, SE-151 36 Sodertalje,
̈ ̈
Sweden) and co-workers for supply and optimized synthesis
protocols of ester 2.
REFERENCES
■
(1) Johansson, A.; Lofberg, C.; von Unge, S.; Antonsson, M.;
Hogner, A.; Hayes, M.; Judkins, R.; Ploj, K.; Benthem, B.; Linden
Fredenwall, M.; Li, L.; Persson, J.; Bergman, R. Manuscript in
preparation.
̈
́
, D.;
(2) (a) MacNeil, D. J. Front. Endocrinol. 2013, 4, 1−14. (b) Hogberg,
̈
T.; Frimurer, T. M.; Sasmal, P. K. Bioorg. Med. Chem. Lett. 2012, 22,
6039−6047. (c) Luthin, D. R. Life Sci. 2007, 81, 423−440.
(3) Johansson, A. Expert Opin. Ther. Pat. 2011, 21, 905−925.
(4) Johansson, A.; Lofberg, C. Expert Opin. Ther. Pat. 2015, 25, 193−
̈
207.
(5) (a) Kamath, A.; Ojima, I. Tetrahedron 2012, 68, 10640−10664.
and references cited therein. (b) Singh, G. S.; Sudheesh, S. ARKIVOC
(Gainesville, FL, U. S.) 2014, 337−385. (c) Brandi, A.; Cicchi, S.;
Cordero, F. M. Chem. Rev. 2008, 108, 3988−4035. (d) Fu, N.; Tidwell,
T. T. Tetrahedron 2008, 64, 10465−10496. (e) Pitts, C. R.; Lectka, T.
Chem. Rev. 2014, 114, 7930−7953. (f) Hafner, A.; Ley, S. V. Synlett
2015, 26, 1470−1474. (g) Allen, A. D.; Tidwell, T. T. Chem. Rev.
2013, 113, 7287−7342.
(6) (a) Macías, A.; Ramallal, A. M.; Alonso, E.; del Pozo, C.;
Gonzal
Pozo, C.; Gonzal
́
ez, J. J. Org. Chem. 2006, 71, 7721−7730. (b) Alonso, E.; del
́
ez, J. Synlett 2002, 2002, 0069−0072. (c) Alonso, E.;
Pozo, C. d.; Gonzalez, J. J. Chem. Soc., Perkin Trans. 1 2002, 571−576.
(7) Cainelli, G.; Galletti, P.; Giacomini, D. Synlett 1998, 1998, 611−
612.
(8) Generation of ketenes in flow: Henry, C.; Bolien, D.; Ibanescu,
B.; Bloodworth, S.; Harrowven, D. C.; Zhang, X.; Craven, A.;
Sneddon, H. F.; Whitby, R. J. Eur. J. Org. Chem. 2015, 2015, 1491−
1499.
(9) Diameter pf 1 mm.
(10) The reaction volume in the setup was calculated to 3.2 mL
(including the IR flow cell and subsequent tubing), resulting in a
residence time of 71 s.
(11) Later, we have shown that in-line premixing of triazinane 15
dissolved in CH2Cl2 with BF3-OEt2 (3 equiv) at 20 °C requires only
seconds of residence time to transform to the corresponding imine.
Consequently, the preferred setup would include the generation of the
imine in flow.
(12) AstraZeneca AB; AstraZeneca UK Ltd; Preparation of azetidine
derivatives for treatment of melanin related disorders.
WO2013011285A1, Jan 24, 2013; p 57−59.
(13) Ojima, I.; Zhao, M.; Yamato, T.; Nakahashi, K.; Yamashita, M.;
Abe, R. J. Org. Chem. 1991, 56, 5263−5277.
(14) Aldehyde 5 has medium solubility in DCM.
ASSOCIATED CONTENT
* Supporting Information
(15) (a) Dost, J.; Heschel, M.; Stein, J. J. Prakt. Chem. (Leipzig) 1985,
327, 109−116. (b) Dost, J.; Stein, J.; Heschel, M. Z. Chem. (Stuttgart,
Ger.) 1986, 26, 203−204. (c) Kumar, N. N. B.; Kuznetsov, D. M.;
Kutateladze, A. G. Org. Lett. 2015, 17, 438−441.
■
S
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.oprd.5b00319.
1H NMR, 13C NMR, HRMS, and HPLC analyses (PDF)
AUTHOR INFORMATION
Corresponding Authors
*E-mail: staffan.po.karlsson@astrazeneca.com.
*E-mail: henrik.sorensen@astrazeneca.com.
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank our lead optimization group for valuable discussions,
our NMR specialist group for help with structure elucidation,
and our Structure Analysis group for HRMS and HPLC purity
analyses. Thanks are also due to Esmail Yousefi (currently at SP
H
DOI: 10.1021/acs.oprd.5b00319
Org. Process Res. Dev. XXXX, XXX, XXX−XXX