Practical asymmetric synthesis of a chiral piperazinone derivative

Article abstract of DOI:10.1021/op400150w

A practical asymmetric route to a chiral piperazinone derivative, a fragment of MK-3207, is reported. The amine-bearing benzylic stereocenter is introduced via an asymmetric Pd-catalyzed hydrogenation of a cyclic sulfimidate in the presence of a chiral phosphine ligand. An efficient synthesis of the hydrogenation substrate is described, together with process development of the hydrogenation step and elaboration of the resulting cyclic sulfamate product to the desired piperazinone.

Full text of DOI:10.1021/op400150w

Article
pubs.acs.org/OPRD
Practical Asymmetric Synthesis of a Chiral Piperazinone Derivative
Mark McLaughlin,* Kevin Belyk, Cheng-yi Chen, Xin Linghu, Jun Pan, Gang Qian, Robert A. Reamer,
and Yingju Xu
Department of Process Research, Merck Research Laboratories, Merck & Co., Inc., Rahway, New Jersey 07065, United States
S
* Supporting Information
ABSTRACT: A practical asymmetric route to a chiral piperazinone derivative, a fragment of MK-3207, is reported. The amine-
bearing benzylic stereocenter is introduced via an asymmetric Pd-catalyzed hydrogenation of a cyclic sulfimidate in the presence
of a chiral phosphine ligand. An efficient synthesis of the hydrogenation substrate is described, together with process
development of the hydrogenation step and elaboration of the resulting cyclic sulfamate product to the desired piperazinone.
synthetic efficiency, better stereocontrol, and greater overall
economy.
INTRODUCTION
■
The calcitonin gene-related peptide (CGRP) has been
implicated in the pathogenesis of migraine headaches.1 The
development of small molecule CGRP-receptor antagonists
offers potential for new treatments for this debilitating
condition, which affects an estimated 13% of the general
population.2 As part of a drug development program at Merck,
practical access to multikilogram quantities of clinical candidate
MK-3207 was required.3 This paper describes a second
generation catalytic asymmetric approach to the key piper-
azinone intermediate 1.
Second Generation Synthesis. Retrosynthetic analysis for
the proposed new route to the piperazinone acid 1 intermediate
is shown in Scheme 2. The first disconnection reveals the core
piperazinone heterocycle 13, and it was anticipated that
selective alkylation of the amide NH would be achievable
under appropriate conditions. A benefit of this disconnection is
that the source of the acid sidechain is now a readily available α-
haloacetate reagent. This obviates a process robustness issue in
the first generation synthesis, in which the unstable ethyl
glycine free base is used as the source of this molecular
fragment. To access the core piperazinone 13, it was recognized
that a cyclic sulfamate intermediate, 14, with regiochemistry
“reversed” from the first generation synthesis could significantly
shorten the synthetic sequence and confer several additional
advantages. In addition, cyclic sulfamates participate in ring-
opening/ring-closing reaction sequences with bifunctional
reagents (e.g., amino acids) to afford piperazinones in a single
step.4
Because of steric encumbrance, cycloleucine appeared to be a
challenging reaction partner for this process, but in light of the
overall synthetic advantage conferred by this disconnection, it
was worthy of investigation. Having identified the key chiral
intermediate, a solution to the problem of the benzylic
stereocenter was required. Asymmetric hydrogenation of
unsaturated compounds represents one of the most attractive
and practical options for control of functionalized stereo-
centers.5 Consequently, we targeted cyclic sulfimidate 15 as a
likely precursor of the chiral sulfamate and sought to derive this
cyclic imine from a hydroxyacetophenone starting material
(16).
Clinical candidate MK-3207 comprises two main molecular
fragments; piperazinone-acid 1 and aniline 2 (Figure 1). These
two fragments are united via a late-stage amide formation. The
focus of this report is synthesis design and process development
of a commercially viable approach to the piperazinone acid 1.
First Generation Synthesis. Early GMP deliveries of MK-
3207 utilized the synthesis of the piperazinone acid 1 shown in
Scheme 1. Although this route ably supported initial animal
toxicology and clinical studies, it was recognized that an
alternative approach would be required in the long term. The
first generation route to piperazinone acid 1 comprises 10 steps
in the longest linear sequence. There are several key issues with
respect to process robustness/efficiency and economics.
Despite extensive study, the asymmetric hydrogenation to set
the benzylic stereocenter is only moderately effective and
typically delivers material with enantiomeric excess in the range
60−70%. This lack of stereocontrol necessitates an upgrade via
the dibenzoyltartaric acid salt in the downstream chemistry,
compromising the overall process efficiency. Additionally, both
the aminoketone 6 and ethyl glycine intermediates have limited
chemical stability and create issues with process robustness.
Furthermore, cost analysis of this route to piperazinone acid 1
identifies the synthetic amino acid “cycloleucine” as a major
contributor. Since this reagent constitutes an integral part of the
molecular structure of MK-3207, its use is essentially
mandatory in any synthesis. Consequently, introduction of
this component at such an early stage in a 10-step sequence is
not the ideal strategy from an economic standpoint. Taking all
of these factors into account, we set out to design a new
approach to the piperazinone acid 1 that would provide higher
Hydroxyacetophenone Synthesis. α-Hydroxyacetophe-
nones are useful synthetic intermediates amenable to various
asymmetric transformations capable of generating valuable
chiral compounds.6 Although α-hydroxyacetophenones have
been featured many times in the chemical literature, there are
only a limited number of reports directly focused on general
procedures for their preparation, which is perhaps indicative of
Received: June 6, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op400150w | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 1. Amide bond disconnection of MK-3207 revealing the two key molecular fragments, piperazinone acid 1 and aniline 2.
Scheme 1. First generation synthesis of piperazinone acid 1
Cyclic Sulfimidate Synthesis. Cyclic sulfimidates similar
to 15 have been described previously in the literature. Initial
laboratory investigations quickly revealed the known methods
of preparation to be entirely unsuitable for large-scale
operation.9 The literature method (Scheme 3) relies on an in
situ preparation of sulfamoyl chloride, an unstable compound
that is not readily commercially available on-scale. Reported
procedures involve a neat reaction between N-chlorosulfony-
lisocyanate (CSI) and formic acid (95% aqueous), which is
relatively hazardous. This reaction is highly exothermic and
releases 2 mol of gas (carbon monoxide and carbon dioxide),
and the mixture solidifies as conversion proceeds, affecting
agitation and making control of the exotherm even more
problematic. In the next stage, sulfamoyl chloride is combined
with the α-hydroxyacetophenone 16 in the presence of pyridine
to yield the O-sulfamoyl intermediate 17, which is cyclized/
dehydrated to the cyclic sulfimidate via thermal treatment
during workup. A significant side-reaction here is nucleophilic
attack by chloride ion on the uncyclized O-sulfamoyl
intermediate to yield the α-chloroacetophenone 18. Pyridine
is a poor choice of base in this regard because pyridinium
hydrochloride has reasonable solubility in the reaction medium
and facilitates side-product formation (vide infra). In addition
to having a detrimental effect on yield, the α-chloroacetophe-
none 18 is a severe lachrymator, creating handling/industrial
hygiene issues during workup.
Scheme 2. Retrosynthesis of piperazinone acid 1
underlying stability issues associated with these intermediates.7
Indeed, we encountered significant stability problems during
our early attempts to work with our particular intermediate
(16). These difficulties were eventually resolved when an
understanding of the compound’s instability to both oxygen
and neutral-to-basic pH was obtained. The details of this
chemistry have been described in a separate report.8 In
summary, following process development, we were ultimately
able to access the desired hydroxyacetophenone in practical
fashion, and this enabled study of the subsequent steps in our
proposed synthesis.
B
dx.doi.org/10.1021/op400150w | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 3. Literature conditions applied to synthesis of cyclic sulfimidate 15
Scheme 4. Improved synthesis of cyclic sulfimidate 15
This procedure was deemed unsuitable for large-scale
synthesis, and we developed a new process that is both safer
and higher yielding (Scheme 4). The reaction of tert-butyl
alcohol with CSI is essentially quantitative and cleanly
generates N-Boc-sulfamoyl chloride. Since this process is a
simple addition reaction, there are no gaseous byproducts. In
addition, the reaction is conveniently run using an appropriate
solvent (e.g., THF, 2-Me-THF), allowing for good control of
the exotherm via rate of addition of reagent. Range-finding
experiments indicate that over-reaction of excess tert-butyl
alcohol with the initially formed N-Boc-sulfamoyl chloride is
not an issue under the reaction conditions.10 The resulting
solution of N-Boc-sulfamoyl chloride has good stability,
affording an acceptable operating window for large-scale
processing.11 Combination with the hydroxyacetophenone 16
generates negligible exotherm because there is no reaction until
the subsequent addition of triethylamine; the O-sulfamoylation
exotherm is controlled via addition rate of the base. O-
Sulfamoylation generates intermediates (17 and 19) that are
activated toward nucleophilic displacement by the chloride ion.
The reaction temperature and age time also has a significant
impact on the amount of the α-chloroacetophenone impurity
formed, with >90% conversion after 24 h age at room
temperature using pyridine as the base.
substrate solubility and rate of reaction significantly. Above 0
°C, the solubility of triethylamine hydrochloride becomes
greater, and the α-chloroacetophenone can accumulate over
time. Under the optimal conditions, the O-sulfamoylation
reaction is typically complete within 30 min, and the reaction is
then quenched by the addition of 0.5 M NaHSO4, maintaining
the batch temperature around 0 °C. At this stage, the use of 2-
Me-THF as solvent allows for a direct phase separation and
rejection of triethylamine hydrochloride into the lower aqueous
phase. A second polishing wash with additional 0.5 M NaHSO4
ensures a negligible concentration of chloride ion in the organic
phase and renders the intermediate O-sulfamoyl compound
stable for continued processing at elevated temperatures.12
Although it is possible to isolate the N-Boc-protected O-
sulfamoyl intermediate 19 via crystallization, the opportunity to
through-process to the desired cyclic sulfimidate 15 was
attractive. Accordingly, the wet 2-Me-THF solution of
uncyclized intermediate 19 is treated with a catalytic quantity
of TsOH·H2O (1 mol %) and heated to reflux to effect
sequential N-Boc deprotection and cyclization/dehydration to
the cyclic sulfimidate 15. The N-Boc group remains largely
intact until the batch temperature reaches >60 °C and then
undergoes smooth cleavage under the action of catalytic acid.
Over the course of several hours, the conversion to cyclic
sulfimidate 15 reaches >95%, and then close to complete
conversion can be attained via azeotropic distillation of the 2-
Me-THF.13 The final isolation involves cooling and washing
with water to remove residual inorganics (NaHSO4) from the
organic phase, followed by crystallization of 15 from a mixture
To mitigate these issues, we selected triethylamine as the
base and 2-Me-THF as the solvent and defined the operating
temperature range as −5 to 0 °C. Triethylamine hydrochloride
precipitates from solution, which suppresses α-chloroacetophe-
none formation because of a reduced concentration of
dissolved chloride ion. Cooling to <−10 °C decreases the
C
dx.doi.org/10.1021/op400150w | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
of 2-Me-THF/heptanes. The overall assay yield from
hydroxyketone is 94% and the isolated yield is 86%.
Table 1. Preliminary screening of catalyst/ligand
combinations for asymmetric hydrogenation of cyclic
Cyclic Imine Asymmetric Hydrogenation. Concurrent
with our process development work, a literature report
appeared on the asymmetric hydrogenation of cyclic
sulfimidates.9 In this report, the levels of conversion and
enantiocontrol were excellent across a variety of substrates, so
we were highly optimistic regarding our particular compound
15. However, the optimal conditions described by Zhou have
several potential drawbacks with respect to large-scale
operation. From an economic perspective, the use of
trifluoroethanol as solvent, Pd(TFA)2 as a catalyst precursor,
and (S,S)-Binaphane as the chiral ligand would all contribute to
a high cost for the key asymmetric transformation in the new
route. Furthermore, in practical terms, the requirement for
relatively high pressures of hydrogen gas (500 psi) could limit
options for implementing this chemistry at vendors lacking
appropriate plant capabilities. To this end, we embarked upon a
systematic study of the reaction conditions for this key
transformation in the hope of finding an improved process.
For our initial screening of reaction conditions (Scheme 5,
Figure 2), we elected to retain certain aspects of the published
sulfimidate 15
entry
ligand
metal
% ee
1
2
3
4
5
A
B
B
B
C
Pd
Pd
Rh
Ir
96
98
99
99
97
Pd
Table 2. Optimization of catalyst precursor and loading,
solvent, hydrogen pressure, and reaction temperature
entry metal precursor
solvent
H2 (psi)
S/C
% conv % ee
1
2
3
4
5
6
7
8
9
Pd(TFA)2
Pd(TFA)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
TFE
500
80
80
40
20
40
20
40
20
200
200
>99.5
>99.5
>99.5
>99.5
>99.5
>99.5
84
96
96
96
96
96
96
96
96
96
MeOH
MeOH
MeOH
MeOH
MeOH
MeOH
MeOH
MeOH
200
Scheme 5. Conditions for initial evaluation of metal/ligand
for asymmetric hydrogenation
330
330
500
500
1000
1000
52
14
several process advantages, such as cost reduction, alleviation of
industrial hygiene concerns around TFE, and the opportunity
for a simple product isolation (vide infra). Also notable is the
option to conduct the hydrogenation using significantly lower
hydrogen pressures and still obtain good reaction performance.
In the final process, the cyclic sulfimidate is dissolved in 5 vol
of MeOH and subject to 0.3 mol % catalyst loading (0.33 mol
% ligand) under 40 psi hydrogen at 40 °C (Scheme 6). After
complete conversion, the batch is treated with carbon and
filtered, and the desired product is isolated via crystallization
following addition of water. The isolated yield is 94%, and the
white solid is typically >99 wt % pure. The measured
enantiomeric excess of the isolated material matches the in-
process-control assay, indicating no upgrade is available via this
crystallization. Subsequent crystallization of a downstream
intermediate affords the necessary upgrade in stereochemical
purity.
procedure (such as the solvent (TFE) and hydrogen pressure
of 500 psi) and vary the parameters of catalyst precursor and
ligand. We also ran the control experiment in which the exact
literature conditions were used. A summary of results is shown
in Table 1.
The control experiment using Pd(TFA)2 and the binaphane
ligand gave a result similar to that published. Complete
conversion was reached in all cases, and excellent enantiomeric
excess was also observed using Rh, Ir, and Pd catalyst
precursors in conjunction with several alternative commercially
available phosphine ligands. We selected the Josiphos ligand for
further study because this was already available to us on
production scale for support of another Merck marketed drug
(Januvia).14
Examining the other reaction parameters (Table 2), we were
pleased to quickly establish that Pd(OAc)2 can replace
Pd(TFA)2 with similar performance. More significantly, in
contrast to the literature report, replacement of TFE with
MeOH as the reaction solvent was equally effective and confers
Piperazinone Formation. Cyclic sulfamates are known to
undergo reactions with amino esters that ultimately cascade to
piperazinones.4 For the target piperazinone 13, the amino ester
required is the unnatural but commercially available “cyclo-
Figure 2. Ligands used for initial evaluation of the asymmetric hydrogenation.
D
dx.doi.org/10.1021/op400150w | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
sulfamate 14 at 70 °C led to ring-opening, and then 5 M
aqueous HCl was used to effect cleavage of the N-sulfate;
however, the two-phase nature of the toluene/water system
made the hydrolysis very slow. Addition of an acid-stable
cosolvent (in this case, 1,2-DME) was necessary to make the
phases partially miscible and increase the rate of hydrolysis to a
practical level. After hydrolysis, the majority of substrate was
present in the acidic aqueous phase (as determined by HPLC
assay), whereas the organic phase contained some dark
polymerized material, which was separated via a phase cut at
this stage. Adjustment of the pH via treatment with 10 M
aqueous NaOH caused lactam formation and partitioning into
the freshly replaced organic phase. After phase cut and solvent
switch, the desired lactam was isolated via crystallization.
Although the overall yield was reasonable, a close
examination of the various operations revealed several
opportunities for process improvements. The overall piper-
azinone process involves four distinct stages: free-base of
cycloleucine, ring-opening of cyclic sulfamate 14, N-sulfate
cleavage, and pH adjustment/ring closure to the piperazinone
13.
Scheme 6. Optimized conditions for asymmetric
hydrogenation of 15
leucine”. Because of the steric hindrance around the quaternary
center, cycloleucine is an extremely poor nucleophile. In fact, in
contrast to more typical amino esters, cycloleucine free-base is
relatively stable for extended periods and does not polymerize
to any significant extent.15 This lack of reactivity necessitated
significant process development to achieve an acceptable rate of
reaction with cyclic sulfamate 14 and eventual yield of the
desired piperazinone 13. The reaction conditions used to gain
proof-of-concept for this particular piperazinone formation are
shown in Scheme 7.
In developing this overall process, the following parameters
received close attention (Scheme 8):
(a) volatility of cycloleucine during free-base process
(b) dipolar aprotic solvent to enhance nucleophilic dis-
placement
Scheme 7. Initial conditions used to prepare piperazinone 13
(c) absence of extraneous nucleophiles (including water and
certain solvents)
(d) water miscibility/pH stability across a wide range for N-
sulfate cleavage and lactamization
To address the sluggish nature of the ring-opening step, an
evaluation of solvents was made, with particular focus on those
likely to facilitate formation of the initially formed highly polar
reaction intermediate (N-sulfate 20). Protic solvents, such as
MeOH and EtOH, were excluded after experiments confirmed
partial solvolysis of the cyclic sulfamate substrate 14. More
surprising and discouraging, however, were observations that
typical dipolar aprotic solvents, such as THF, MeCN, DMF,
DMAc, and NMP, did not provide improved results. In fact, in
the case of DMF and DMAc, these solvents are sufficiently
nucleophilic to be competitive with cycloleucine and
Cycloleucine is available commercially as a hydrochloride salt
that needs to be converted to free-base before reaction with
cyclic sulfamate 14. In our initial studies, cycloleucine free-base
was extracted into toluene from aqueous potassium phosphate
tribasic. Concentration of the organic phase allowed for
azeotropic drying prior to reaction with the cyclic sulfamate.16
Heating the dry toluene solution of cycloleucine and cyclic
Scheme 8. Reaction sequence for piperazinone formation
E
dx.doi.org/10.1021/op400150w | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
consequently lead to unwanted side-reactions and poor reaction
profiles.
adjustment to pH 1 with aqueous HCl leads to precipitation
of the crude intermediate HCl salt. Reslurry of this HCl salt in
water followed by treatment with 1 equiv of NaOAc and
heating to 80 °C allows for neutralization of the HCl and
crystallization of the product 1 as the neutral form. This
crystallization also affords the necessary upgrade in enantio-
purity to deliver material that matches the previously
established purity specifications for this regulated API starting
material (>99 LCAP, 99.9 LC wt % purity, 99.7% ee). In
addition, using piperazinone acid 1 (from the new chemistry)
for the final amide bond formation to generate MK-3207 API
was also shown to provide material matching the established
specification.
Following these observations, we turned our attention to
sulfolane, a solvent that has been gaining in popularity for a
variety of reactions and appeared to be a good candidate for our
particular process.17 Sulfolane has one of the highest dielectric
constants of any solvent and was expected to promote the
initial desired ring-opening to yield N-sulfate 20. In addition
and in contrast to other common dipolar aprotic solvents,
sulfolane is nonnucleophilic, which virtually eliminates non-
productive solvolytic process, such as those encountered with
DMF. Last, sulfolane is aqueous miscible and stable at low pH,
which helps streamline the overall piperazinone-forming
process.
The finalized process for piperazinone formation is as
follows: Cycloleucine free-base is generated via partition of the
hydrochloride salt between aqueous K3PO4 and MTBE
followed by azeotropic drying of the MTBE layer.18 The
relative volatility of MTBE ensures minimal loss of the
cycloleucine free-base during this distillation process. Upon
reaching the desired water content specification (<100 ppm),
the MTBE solution is concentrated and transferred into a
solution of the cyclic sulfamate in sulfolane. Further distillation
under reduced pressure removes MTBE from the system, and
the resulting sulfolane solution of reactants is heated to 70 °C
to promote the desired ring-opening.19 Typical conversion is
>95% after heating for 10 h. Addition of 5 M aqueous HCl and
continued heating leads to hydrolysis of the N-sulfate; the
aqueous miscibility of sulfolane renders the reaction system
single phase and facilitates this hydrolysis.
Next, the system is adjusted to pH 10 using aqueous NaOH
and heated to ensure complete lactamization. At the end of
reaction, the mixture is extracted with MTBE, and after washing
with brine, the majority of the sulfolane is rejected to the
aqueous phase. The MTBE layer is decolorized with carbon,
and the product piperazinone 13 can ultimately be isolated
(vide infra) via crystallization following solvent switch into
heptane. The typical corrected isolated yield is 70% (96% ee
unchanged from input stream).20 Given the lack of stereo-
chemical upgrade achieved via the isolation of 13, it appeared
more attractive to through-process this intermediate into the
final N-alkylation step. To our satisfaction, the crystallization
after the N-alkylation step proved highly robust and
consistently afforded good overall purity and sufficient
stereochemical upgrade (vide infra).
CONCLUSION
■
In summary, we have developed a practical asymmetric
synthesis of 1, the chiral piperazinone acid fragment of
CGRP clinical development candidate MK-3207 (Scheme
10). A scalable process for preparation of the prochiral cyclic
sulfimidate 15 was defined, and an efficient asymmetric
hydrogenation of this intermediate was used to set the key
benzylic stereocenter. Significant improvements to the hydro-
genation procedure were realized. The final process is
conducted at mild temperature (40 °C) under relatively low
pressure of hydrogen (40 psi) with low catalyst loading (0.3
mol %) and affords the desired product 14 in 96−97% ee.
Further, the use of MeOH solvent allows for simple product
isolation via addition of water and crystallization at the end of
the reaction. Elaboration of the chiral cyclic sulfamate 14 into
piperazinone 13 was achieved via a cascade reaction sequence
with cycloleucine. The low reactivity of cycloleucine neces-
sitated careful control of reaction conditions, and sulfolane was
identified as a uniquely effective solvent for this process. Last,
selective N-alkylation of the piperazinone amide with ethyl
bromoacetate followed by in situ ester hydrolysis generated the
target piperazinone acid 1 in high yield. Final crystallization of
the piperazinone acid 1 results in consistent upgrade of
stereochemical purity to afford material that matches the
established specification for material prepared via the first-
generation chemistry. In addition, comparison of both routes
indicates the second-generation approach comprises fewer
steps; has greater overall yield; and, as a result, is significantly
more economical. Consequently, the new piperazinone acid
chemistry based on cyclic sulfimidate asymmetric hydro-
genation was chosen as the long-term synthesis for this
fragment of clinical development candidate MK-3207.23
Piperazinone Amide N-Alkylation. Alkylation of the
piperazinone amide NH is accomplished via deprotonation
using NaHMDS in THF followed by treatment with ethyl
bromoacetate (Scheme 9).21,22 The resulting ester is hydro-
lyzed in situ during workup with aqueous LiOH, and
EXPERIMENTAL SECTION
■
All materials were purchased from commercial suppliers. All
reagents and solvents were used as supplied by manufacturers.
1H and 13C NMR spectra were obtained using a Bruker 500
MHz spectrometer in the solvents indicated. HPLC was
performed using an Agilent 1100 series instrument. HPLC
method details for both reaction monitoring and % ee
determination can be found in the Supporting Information,
together with all relevant NMR spectra.
Scheme 9. Piperazinone amide N-alkylation
Preparation of 4-(3,5-Difluorophenyl)-5H-1,2,3-oxa-
thiazole 2,2-Dioxide (15). A solution of 2-Me-THF (600
mL, 20 ppm water content by KF titration) and t-BuOH (73
mL, 762 mmol, 1.3 equiv, 570 ppm water content by KF
titration) was cooled to between −10 and 0 °C. Neat N-
chlorosulfonylisocyanate (CSI) (61 mL, 702 mmol, 1.2 equiv)
F
dx.doi.org/10.1021/op400150w | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 10. The second generation synthesis of piperazinone acid 1
was charged at −5 to 0 C°. This solution of N-Boc-sulfamoyl
chloride was aged for 30 min. A solution of hydroxyketone 16
(102 g, 580 mmol, 1 equiv) in 2-Me-THF (600 mL) was
prepared (water content of solution was 322 ppm by KF
titration). The solution of hydroxyketone 16 was transferred
into the cooled (<0 °C) solution of N-Boc-sulfamoyl chloride
with negligible exotherm. Neat Et3N (114 mL, 818 mmol, 1.4
equiv) was charged at −10 to 0 C°. A precipitate of Et3N·HCl
formed. Temperature control at this stage and throughout
workup of this step is critical to minimize formation of
undesired side-product (α-chloroacetophenone derivative).
After 1 h age between −10 and 0 C°, the reaction was
quenched with 0.5 M NaHSO4 (552 mL, 0.5 equiv) at 0−5 °C.
Additional water (300 mL) was added. The lower aqueous
phase was cut, and the pale yellow organic phase was washed
with 0.5 M NaHSO4 (552 mL, 0.5 equiv). After cutting the
lower aqueous phase, catalytic TsOH·H2O (1.1 g, 5.8 mmol,
0.01 equiv) was added, and the mixture was heated to 75 °C
(reflux condition). After 14 h under standard reflux, the
reaction was driven to completion via azeotropic water removal.
The solution was cooled to 25 °C, and water (1 L) was added.
The lower aqueous phase was cut, and the orange organic phase
was dried via azeotropic distillation (KF < 500 ppm) and then
concentrated to ∼0.2 L, during which time the product
crystallized. Heptanes (0.4 L) were slowly added (at 20 °C) as
antisolvent until a final ratio of 1:2 2-Me-THF/heptane was
achieved. The resultant slurry of product 15 was cooled to 0 °C
and aged for 1 h before filtration. The cake was washed with 2
bed volumes of 1:2 mixture of 2-Me-THF/heptanes mixture
then, finally, with 2 bed volumes of heptanes. After drying, 116
g of 15 was obtained in >99 LCAP and >99 wt % purity by 1H
NMR vs 1,3,5-trimethoxybenzene as standard. The corrected
isolated yield was 86% from 16.
by mixing Pd(OAc)2 (337 mg, 1.5 mmol, 0.3 mol %) and
phosphine ligand (S)-(−)-1-[(R)-2-(diphenylphosphino)-
ferrocenyl]ethyldi-t-butylphosphine (788 mg, 1.65 mmol, 0.33
mol %) in CH2Cl2 (40 mL) at room temperature for 15 min.
The catalyst solution was then transferred to the substrate in
MeOH under vacuum, and the system was placed under an
atmosphere of H2 (40 psi). The autoclave was heated to 40 °C
for 16 h. After this time, sampling for HPLC analysis indicated
>99% conversion and 97% assay yield. The reaction mixture
was treated with carbon (Ecosorb C-941) and then filtered over
Celite to afford a light brown solution. Addition of water to this
MeOH solution induced crystallization of the sulfamate
product 14, which was collected by filtration and rinsed with
further portions of MeOH/water. After drying at 40 °C under a
nitrogen sweep, the product 14, 110 g, was obtained as a white
solid. The material was >99 LCAP, 96.4% ee and >99 wt %
1
purity by H NMR vs 1,3,5-trimethoxybenzene as standard.
The corrected isolated yield was 94% from 15.
1H NMR (500 MHz, DMSO): δ 4.42−4.44 (m, 1 H), 4.97−
4.99 (m, 1 H), 5.17−5.18 (m, 1 H), 7.20−7.22 (m, 3 H), 8.63
(d, J = 6.1 Hz, 1 H). 13C NMR (126 MHz, DMSO): δ 57.8 (t,
JCF = 2 Hz), 75.0, 104.2 (t, JCF = 26 Hz), 110.4 (m), 143.1 (t,
JCF = 9 Hz), 162.9 (dd, JCF = 247 and 13 Hz). HRMS calcd. for
C8H7SO3NF2 (M + Na), 258.0012; found, 258.0005.
Preparation of 8-(3,5-Difluorophenyl)-6,9-
diazaspiro[4.5]decan-10-one (13). Cycloleucine hydro-
chloride salt (40.8 g, 0.23 mol, 1.7 equiv) was free-based via
partition between MTBE (3 × 100 mL) and aqueous potassium
phosphate (78.4 g, 0.37 mol, 2.7 equiv in 200 mL of water).
The combined MTBE extract was distilled at constant volume
under atmospheric pressure (with continuous addition of fresh
400 mL MTBE) to effect azeotropic water removal until the
water content measured by KF titration was <100 ppm.
Sulfolane (40 mL) was added, and the majority of MTBE was
removed via continued distillation. Cyclic sulfamate 3 (32.4 g,
0.14 mol, 1 equiv) was charged, and the resultant solution was
heated to 50 °C for 7 h. After this time, aqueous HCl (100 mL
of 5 M, 0.50 mol, 3.7 equiv) was charged, and the mixture was
heated to 60 °C for 3 h. The mixture was then pH-adjusted via
addition of 10 M NaOH and heated to 60 °C for a further 2 h
before cooling to room temperature. MTBE (300 mL) and 15%
aq NaCl (450 mL) were charged, and the resulting two-phase
1H NMR (500 MHz, DMSO): δ 6.09 (s, 2 H), 7.78−7.76
(m, 3 H). 13C NMR (126 MHz, DMSO): δ 77.6, 111.3 (t, JCF
=
26 Hz), 113.2 (m), 130.8 (t, JCF = 10 Hz), 163.9 (dd, JCF = 250
and 13 Hz), 177.6 (t, JCF = 3 Hz). HRMS calcd. for
C8H5SO3NF2 (M + Na), 255.9856; found, 255.9857.
Preparation of 4-(3,5-Difluorophenyl)-1,2,3-oxathia-
zolidine 2,2-Dioxide (14). Cyclic imine 15 (116 g, 0.50
mol, 1 equiv) was charged to an autoclave vessel, followed by
MeOH (1.2 L). In a separate vessel, the catalyst was prepared
G
dx.doi.org/10.1021/op400150w | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Notes
mixture was agitated for 10 min. The lower aqueous phase was
cut, and the organic phase was washed with more 15% aq NaCl
(2 × 300 mL) to remove sulfolane. The organic phase was then
treated with Ecosorb C941 and filtered, and the solvent was
switched to heptane to effect crystallization of the piperazinone
13. After filtration, the product 13 was rinsed with heptane
(100 mL) and dried to give a yellow solid (26.50 g, 97.5%
NMR wt % for 70% corrected isolated yield).
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors wish to thank T. J. Novak for the rapid acquisition
of high resolution mass spectrometry data.
REFERENCES
1H NMR (500 MHz, DMSO): δ 1.61−1.66 (m, 6 H), 1.93−
1.96 (m, 1 H), 2.03−2.06 (m, 1 H), 2.67 (dd, J = 13.4, 6.9 Hz,
1 H), 3.12 (dd, J = 13.4, 4.6 Hz, 1 H), 4.56−4.59 (m, 1 H),
6.99−7.02 (m, 2 H), 7.12−7.13 (m, 1 H), 7.90 (s, 1 H). 13C
NMR (126 MHz, DMSO): δ 25.3, 25.4, 38.4, 38.5, 47.0, 56.8,
66.2, 103.0 (t, JCF = 26 Hz), 110.1 (m) 147.3 (t, JCF = 9 Hz),
162.8 (dd, JCF = 247 and 13 Hz), 175.7. HRMS calcd. for
C14H16N2OF2 (M + H), 267.1309; found, 267.1304.
■
(1) (a) Ho, T. W.; Edvinsson, L.; Goadsby, P. J. Nat. Rev. Neurol.
2010, 6, 573. (b) Durham, P. L. Headache 2008, 48, 1269.
(2) Stovner, L. J.; Hagen, K.; Jensen, R.; Katsarava, Z.; Lipton, R. B.;
Scher, A. I.; Steiner, T. J.; Zwart, J. A. Cephalalgia 2007, 27, 193.
(3) For the discovery of MK-3207, see: Bell, I. M.; Gallicchio, S. N.;
Wood, M. R.; Quigley, A. G.; Stump, C. A.; Zartman, C. B.; Fay, J. F.;
Li, C.-C.; Lynch, J. J.; Moore, E. L.; Mosser, S. M.; Prueksaritanont, T.;
Regan, C. P.; Roller, S.; Salvatore, C. A.; Kane, S. A.; Vacca, J. P.;
Selnick, H. G. Med. Chem. Lett. 2010, 1, 24−29.
(4) (a) Williams, A. J.; Chakthong, S.; Gray, D.; Lawrence, R. M.;
Gallagher, T. Org. Lett. 2003, 5, 811. (b) Bower, J. F.; Rujirawanicha,
J.; Gallagher, T. Org. Biomol. Chem. 2010, 8, 1505.
Preparation of (R)-2-(8-(3,5-Difluorophenyl)-10-oxo-
6,9-diazaspiro[4.5]decan-9-yl)acetic Acid (1). A solution
of piperazinone 13 (4 g, 14.6 mmol, 1 equiv) in THF (16 mL)
was cooled to −20 °C, and a THF solution of NaHMDS (1 M,
18.9 mL, 18.9 mmol, 1.3 equiv) was charged dropwise,
maintaining the internal temperature <−10 °C. A homoge-
neous light brown solution was obtained. A solution of ethyl
bromoacetate (2.3 mL, 20.4 mmol, 1.4 equiv) in THF (8 mL)
was charged, again maintaining the internal temperature <−10
°C. A precipitate of NaBr was observed. The mixture was
allowed to reach 10 °C, and then aqueous LiOH (5 wt %, 19.0
mL, 43.7 mmol, 3 equiv) was charged to effect in situ hydrolysis
of the ethyl ester. After 2 h, the hydrolysis was complete, and
the mixture was partitioned between water (10 mL) and
heptane (35 mL). The lower, basic aqueous phase (containing
the product acid 1) was separated and cooled to 0−5 °C, then
carefully acidified to a final pH of 1 using concentrated HCl.
The product acid 1 crystallized (as the HCl salt) during pH
adjustment and was collected via filtration. The cake was
washed with 2 M aqueous HCl (20 mL) and briefly dried
before being reslurried in water (40 mL). The slurry was heated
to 80 °C, which dissolved the solid. Treatment of this solution
with NaOAc (1.2 g, 14.6 mmol, 1 equiv) neutralized the HCl
salt and caused recrystallization of the desired acid 1. After
filtration and drying, the acid 1 was obtained as a white solid
(4.1 g, > 99 LCAP, > 97 wt % purity by NMR vs 1,3,5-
trimethoxybenzene as standard, 99.9 LC wt % purity, 99.7% ee
(5) (a) Chen, Q.-A.; Ye, Z.-S.; Duan, Y.; Zhou, Y.-G. Chem. Soc. Rev.
2013, 42, 497. (b) Etayo, P.; Vidal-Ferran, A. Chem. Soc. Rev. 2013, 42,
728.
(6) (a) For example, asymmetric hydrogenation affording 1,2-diols,
see: Ohkuma, T.; Utsumi, N.; Watanabe, M.; Arai, N.; Murata, K. Org.
Lett. 2007, 9, 2565. (b) Asymmetric Mannich-type reactions affording
β-amino-alcohols, see: Matsunaga, S.; Kumagai, N.; Harada, S.;
Shibasaki, M. J. Am. Chem. Soc. 2003, 125, 4712. (c) Trost, B. M.;
Jaratjaroonphong, J.; Reutrakul, V. J. Am. Chem. Soc. 2006, 128, 2778.
(7) For acetophenone oxidation, see: (a) Moriarty, R. M.; Berglund,
B. A.; Penmasta, R. Tetrahedron Lett. 1992, 33, 6065. (b) Chen, C.;
Feng, X.; Zhang, G.; Zhao, Q.; Huang, G. Synthesis 2008, 3205. Via
the dimethyl acetal, see: (c) Moriarty, R. M.; Hou, K.-C.; Prakash, I.;
Arora, S. K. Org. Synth. 1986, 64, 138. Oxidation of silyl enol ethers,
see: (d) Moriarty, R. M.; Prakash, O.; Duncan, M. P. Synthesis 1985,
943. (e) Le, J. C.-D.; Pagenkopf, B. L. J. Org. Chem. 2004, 69, 4177.
From α-haloacetophenones, see: (f) Zhu, G.; Casalnuovo, A. L.;
Zhang, X. J. Org. Chem. 1998, 63, 8100. (g) Wong, F. F.; Chang, P.-W.;
Lin, H.-C.; You, B.-J.; Huang, J.-J.; Lin, S.-K. J. Organomet. Chem. 2009,
694, 3452. For enzymatic cleavage of α-acetoxyacetophenones, see:
́ ́
(h) Paizs, C.; Tosa, M.; Majdik, C.; Bodai, V.; Novak, L.; Irimie, F. D.;
Poppe, L. J. Chem. Soc., Perkins Trans. 1 2002, 2400. For
hydroxymethylation of benzaldehydes, see: (i) Demir, A. S.; Ahyan,
P.; Igdir, A. C.; Duygu, A. N. Tetrahedron 2004, 60, 6509. (j) Kuhl, N.;
Glorious, F. Chem. Commun. 2011, 47, 573.
(8) McLaughlin, M.; Belyk, K. M.; Qian, G.; Reamer, R. A.; Chen, C.
J. Org. Chem. 2012, 77, 5144.
1
and corrected isolated yield of 85%). H NMR (500 MHz,
DMSO): δ 1.67−1.81 (m, 6 H), 1.93−2.01 (m, 1 H), 2.07−
2.14 (m, 1 H), 2.83 (dd, J = 13.5, 3.8 Hz, 1 H), 3.21 (d, J = 17.1
Hz, 1H), 3.25−3.32 (m, 1H), 4.22 (d, J = 17.1 Hz, 1 H), 4.67
(s, 1 H), 7.02−7.06 (m, 2 H), 7.13−7.17 (m, 1 H). 13C NMR
(126 MHz, DMSO): δ 25.2, 25.5, 38.6, 39.3, 46.8, 47.9, 62.2,
67.1, 103.4 (t, JCF = 26 Hz), 111.0 (m), 145.6 (t, JCF = 9 Hz),
162.8 (dd, JCF = 247 and 13 Hz), 170.5, 175.2. HRMS calcd. for
C16H18N2O3F2 (M + H), 325.1364; found, 325.1361.
(9) Wang, Y.-Q.; Yu, C.-B.; Wang, D.-W.; Wang, X.-B.; Zhou, Y.-G.
Org. Lett. 2008, 10, 2071.
(10) In contrast, when the system is deficient in tert-butyl alcohol, the
excess CSI can react directly with the hydroxyacetophenone to
produce a carbamate side-product that is difficult to reject.
(11) Stability of prepared N-Boc-sulfamoyl chloride was demon-
strated via 24 h hold prior to use in subsequent reaction, and
equivalent results were observed.
(12) It is important to avoid a final brine wash of the organic layer
because this will leave residual chloride in the organic phase and
ultimately generate the α-chloroacetophenone side-product.
(13) The cyclic imine is stable to water at the elevated temperatures
required for solvent distillation.
ASSOCIATED CONTENT
■
S
* Supporting Information
Copies of relevant NMR spectra and chromatograms are
provided. This material is available free of charge via the
Internet at http://pubs.acs.org.
(14) Hansen, K. B.; Hsiao, Y.; Xu, F.; Rivera, N.; Clausen, A.;
Kubryk, M.; Krska, S.; Rosner, T.; Simmons, B.; Balsells, J.; Ikemoto,
N.; Sun, Y.; Spindler, F.; Malan, C.; Grabowski, E. J. J.; Armstrong, J.
D. J. Am. Chem. Soc. 2009, 131, 8798.
(15) We observed <5% degradation, even after several weeks’ storage
of the free-base at room temperature.
(16) Because of the volatility of cycloleucine free-base, some material
AUTHOR INFORMATION
■
Corresponding Author
E-mail: mark_mclaughlin@merck.com.
is lost via codistillation during azeotropic drying with toluene.
H
dx.doi.org/10.1021/op400150w | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(17) Tilstam, U. Org. Process Res. Dev. 2012, 16, 1273.
(18) A final brine is not conducted since this was shown to leave
residual chloride in the MTBE layer, which reduced the percent AY for
the sulfamate ring opening via competitive nucleophilic attack by
chloride.
(19) Range-finding studies demonstrated up to 10 vol % MTBE can
be tolerated in the reaction system.
(20) Separate studies using isolated N-sulfate intermediate
demonstrated that hydrolysis and lactamization occur in very high
yield (95%), indicating the overall yield of 70% is associated with the
ring-opening step.
(21) Ethyl bromoacetate is a potential genotoxic impurity; however,
the strongly basic hydrolysis step followed by two different
crystallizations (HCl salt and then final neutral form) is likely to
consume/reject this reagent to below default threshold of toxicological
concern. Prior to implementation of this new chemistry for GMP
production, the development of MK-3207 was discontinued.
(22) Other bases studied include LiHMDS, KHMDS, and KOt-Bu,
but inferior results were observed.
(23) During initial demonstration of the new process for synthesis,
development of MK-3207 was discontinued. As a result, experimental
details are provided for the 100 g scale early demonstration runs.
I
dx.doi.org/10.1021/op400150w | Org. Process Res. Dev. XXXX, XXX, XXX−XXX