Process development for a key synthetic intermediate of LY2140023, a clinical candidate for the treatment of schizophrenia

Article abstract of DOI:10.1021/op100325h

To fuel clinical development of the experimental CNS medicine LY2140023, we developed a scalable route for the multistep synthesis of a pivotal synthetic intermediate. The core of the conformationally restricted glutamic acid-based amino acid analogue was built via a Rh-catalyzed cyclopropanation of thiophene. Regioselective functionalization of the remaining double bond was achieved by a hydroboration/oxidation sequence followed by a Bucherer-Bergs reaction to give a hydantoin with the targeted l-glutamic acid configuration. Subsequent resolution, oxidation state, and protecting group manipulations gave the key intermediate in an overall nine-step scalable streamlined route starting from thiophene.

Full text of DOI:10.1021/op100325h

ARTICLE
pubs.acs.org/OPRD
Process Development for a Key Synthetic Intermediate of LY2140023,
a Clinical Candidate for the Treatment of Schizophrenia
,
†,||
,‡
‡
‡
‡
Mario Waser,* Eric D. Moher,* Sandy S. K. Borders, Marvin M. Hansen, David W. Hoard,
‡
‡
‡
‡
‡
Michael E. Laurila, Michael E. LeTourneau, Richard D. Miller, Michael L. Phillips, Kevin A. Sullivan,
‡
‡
§
†
†
Jeffrey A. Ward, Chaoyu Xie, Cheryl A. Bye, Tanja Leitner, Brigitte Herzog-Krimbacher,
†
†
Marcus Kordian, and Martin M €u llner
†
DSM Fine Chemicals Austria Nfg GmbH & Co KG, St.-Peter-Straße 25, 4021 Linz, Austria
‡
Chemical Product Research and Development, Eli Lilly and Company, Indianapolis, Indiana 46285, United States
§
Analytical Sciences Research and Development, Eli Lilly and Company, Indianapolis, Indiana 46285, United States
ABSTRACT: To fuel clinical development of the experimental CNS medicine LY2140023, we developed a scalable route for the
multistep synthesis of a pivotal synthetic intermediate. The core of the conformationally restricted glutamic acid-based amino acid
analogue was built via a Rh-catalyzed cyclopropanation of thiophene. Regioselective functionalization of the remaining double bond
was achieved by a hydroboration/oxidation sequence followed by a BuchererꢀBergs reaction to give a hydantoin with the targeted
L-glutamic acid configuration. Subsequent resolution, oxidation state, and protecting group manipulations gave the key intermediate
in an overall nine-step scalable streamlined route starting from thiophene.
’
INTRODUCTION
Interest in this novel antipsychotic agent prompted the develop-
ment of a reliable, safe, and scalable process for the synthesis of
intermediate dimethyl ester 10. This work describes the develop-
ment of processes that have now been successfully demonstrated
at pilot plant scale, allowing practical multikilogram production
of the API.
L-Glutamic acid (1) is the primary excitatory amino acid
neurotransmitter in the mammalian central nervous system
CNS), taking part in several physiological and pathophysiolo-
(
gical processes (e.g., anxiety and schizophrenia). It influences
neuronal transmission in part by activation of G-protein-coupled
metabotropic glutamate (mGlu) receptors.
1
ꢀ3
mGlu receptors
are highly heterogeneous with respect to their structure, func-
tion, and localization within the central nervous system and
represent promising targets for therapeutic intervention in a
’
RESULTS AND DISCUSSION
Overview of the First-Generation Route. The initially de-
scribed first-generation multistep laboratory scale synthesis of the
free base of diester 10 is shown in Scheme 1. The synthetic
strategy was straightforward: thiophene and ethyl diazoacetate
4
ꢀ6
multiplicity of CNS disorders.
11
Although altered glutamate neurotransmission has been
linked to schizophrenia for decades, all commonly prescribed
7
antipsychotics act on dopamine receptors. The dense loca-
(
11) were brought together in a rhodium(II)-catalyzed cyclopro-
14
tion of mGlu2/3 (group II) receptors in limbic and forebrain
areas (associated with drug abuse, anxiety, and schizophrenia)
suggests that these receptors may be alternative pharmaco-
logical targets for the development of new medications for
panation to fashion the bicyclo[3.1.0]hexane structural frame-
work in one step while leaving a suitable handle for further
functionalization. Hydroboration of the remaining olefin in 12
was followed by Swern oxidation, affording ketone 14 as a
substrate for an amino acid synthesis targeting the L-glutamate
15
8
these disorders.
As part of a program aimed at identifying highly potent and
selective agonists for mGlu2/3 receptors, a series of novel, con-
formationally restricted bicyclic glutamic acid-based amino acids
configuration. In the event, 14 was subjected to BuchererꢀBergs
16
conditions giving rise to a masked amino acid in the form of
racemic hydantoin 15, which was crystallized from a 4:1 mixture of
diastereomers. Separation of enantiomers by diastereomeric salt
formation and crystallization followed by salt break gave single
isomer (ꢀ)-15. The sequence of hydantoin hydrolysis, exhaustive
protection, and chromatographic purification provided sulfide 18
as a suitably soluble substrate for conversion to sulfone 19 upon
treatment with m-CPBA in methylene chloride. Removal of the
tert-butyl carbamate with trifluoroacetic acid followed by treat-
ment with aqueous bicarbonate, extractive workup, and solvent
9ꢀ12
have been investigated over the past two decades (Figure 1).
Among these novel non-natural amino acids, bicyclic sulfone 8
LY404039) was found to be a promising selective agonist for
(
mGlu2/3 receptors showing antipsychotic potential both in
12
animal studies and in humans. However, 8 suffers from low
human oral bioavailability. After a significant effort, methionine
amide 9 (LY2140023) emerged as a suitable replacement. The
amide linkage is hydrolyzed in vivo, affording the active mGlu2/3
receptor agonist 8 (Figure 2) with suitable bioavailability. Amide
9
, which is obtained from 8 by way of key intermediate amino
diester 10, has shown efficacy for the oral treatment for schizo-
phrenia in a phase II clinical study.
Received: December 8, 2010
Published: September 27, 2011
12,13
r 2011 American Chemical Society
1266
dx.doi.org/10.1021/op100325h
|Org. Process Res. Dev. 2011, 15, 1266–1274

Organic Process Research & Development
ARTICLE
Figure 1. Structures of representative L-glutamic acid (1)-derived mGlu2/3 receptor agonists 2ꢀ8.
9
5% (w/w) purity as an off-white solid. It is noteworthy that
the cyclopropanation reaction is stereoselective for the isomer
1
4
having an exo-oriented carboethoxy substituent as shown (12).
The epimeric endo isomer has not been detected in either the
reaction mixture or the isolated material. The process described
herein has been transferred to multiple contract manufac-
turers whereby several metric tons of 12 have been produced
1
8
to date.
Hydroboration. Regioselective hydroxylation of the double
1
9
bond of 12 could be achieved by a hydroboration using
Figure 2. Relationship among LY2140023 (9), diester intermediate 10,
and agonist 8.
BH THF in place of thexyl borane, followed by a methanol
3
3
quench and subsequent oxidation to the corresponding second-
ary alcohol 13. Moving away from the original procedure of using
thexyl borane resulted in a simpler reaction sequence, and the
workup process did not have to contend with the 2,3-dimethyl-
-butanol (thexyl alcohol) byproduct. As in the bulky thexyl borane
process, the predominant byproduct is the β-epimeric alcohol
∼98:2 β:α). The hydroboration required 1.3 equiv of BH
removal in vacuo gave the free base of amine 10 in approximately
1
2 chemical steps and 0.5% overall yield from 11.
Transformation of the First-Generation Route into a Scal-
2
able Second-Generation Route. The first-generation synthesis,
while strategically straightforward and logical, involved wasteful
protectionꢀdeprotection steps together with several chromato-
graphic purifications and low yields, making it unsuitable for
multikilogram scale. Therefore, this route was developed and
streamlined, resulting in a nine-step, chromatography free, high-
er-yielding second-generation route (Scheme 2).
(
3
3
THF to achieve full conversion in e6 h at 0 °C. Experiments
with less borane resulted in longer reaction times and inferior
yields, perhaps indicating that a portion of the borane is
sequestered by process impurities or rendered inactive through
complexation with the sulfur atom of sulfide 12. Proton NMR
analysis of the methanol-quenched hydroboration reaction mix-
ture revealed a mixure of intermediates, making it difficult to
discern if the major product is a mono- or dialkyl borane. After
the methanol quench, oxidation of the methyl borinate/boronate
mixture to alcohol 13 could be achieved by using a refluxing
aqueous solution (50%) of N-methylmorpholine N-oxide
Cyclopropanation. The original rhodium(II) acetate (0.3
mol %)-catalyzed cyclopropanation of thiophene by ethyl diazo-
acetate suffered from a low yield after chromatographic purifica-
tion. A catalyst screen was conducted in search of a more efficient
transformation. We investigated less expensive copper catalysts
such as CuPF (CH CN) only to find that dimerization of ethyl
6
3
4
17
diazoacetate was the predominant reaction pathway. Dilution
of the reaction mixture with large volumes of thiophene par-
tially suppressed the dimerization but was not practical espe-
cially because the yield remained depressed compared to that for
the rhodium(II) process. Additionally, asymmetric copper and
rhodium catalyst systems were tested but provided lowered
(
NMO). Aqueous NMO bears advantages over alkaline hydro-
gen peroxide and sodium perborate in terms of safety, byproduct
profile (e.g., products of overoxidation to sulfoxide, or hydroxide-
induced events such as ester hydrolysis and ring fragmentation to
give 21), and convenience (liquid charge vs solid perborate
charge). The use of NMO is a direct corollary to the use of
trimethylamine N-oxide as demonstrated by Kabalka. We
found trimethylamine N-oxide to work as well as NMO; how-
ever, the limited availability at multikilogram scale led us to select
NMO for the oxidation. The oxidation did not occur with
pyridine N-oxide at room temperature in contrast to amine N-
oxides. Under these new conditions, alcohol 13 was obtained
after extractive workup in 75ꢀ80% yield as a solution in EtOAc
of sufficient purity to be used directly in the next step. Further
purification of 12 was not necessary, and isolation would have
been a challenge because of the low melting (mp 52 °C). It is
17
2
0
yields of racemic 12. We eventually turned to rhodium(II)
octanoate as a catalyst because it was more readily available and
less expensive. While its use in the cyclopropanation resulted in a
yield similar to that with rhodium(II) acetate, a lower loading of
0
.05 mol % could be realized. Following complete consump-
tion of 11, the thiophene solvent was removed by wiped film
distillation; a second-pass distillation then provided cyclopro-
pane 12 as a crude oil. Dissolution of the oil in methanol
followed by cooling to low temperatures produced a slurry
from which 12 was isolated in 30ꢀ35% yield and greater than
1
267
dx.doi.org/10.1021/op100325h |Org. Process Res. Dev. 2011, 15, 1266–1274

Organic Process Research & Development
ARTICLE
Scheme 1. First-Generation Synthesis of 10
Scheme 2. Second-Generation Synthesis of 10
worth noting that in experiments in which 4ꢀ5 equiv of
methanol was added to quench the hydroboration mixture,
NMO oxidation was usually complete within 4 h, whereas in
the case of addition of a larger excess of methanol (>15 equiv),
the subsequent oxidation required at least 10 h. This fact might
be explained by competing coordination of methanol to the
boron resulting in an increased reaction time. Lending some
additional support to this hypothesis is the observation that a
1
268
dx.doi.org/10.1021/op100325h |Org. Process Res. Dev. 2011, 15, 1266–1274

Organic Process Research & Development
ARTICLE
Table 1. Comparison of the Hydroboration of 12 Using 1.3
equiv of BH THF or BH DMS
3
3
3
3
T (°C)
t (h)
corrected yield (%)
BH
BH
BH
BH
3
3
3
3
THF
DMS
DMS
DMS
0
2
19
6
70
50
68
35
3
3
3
3
0
5ꢀ10
25
Figure 3. Some hydroboration and ParikhꢀDoering byproducts.
2
additional reagent in the event of a stall. After some brief
experimentation, we settled on conducting the reaction in an
EtOAc/DMSO mixture with H €u nig’s base and 2.5 equiv of
much faster rate of oxidation can be achieved (1.5 h at room
temperature) by using crystalline, anhydrous NMO (Aldrich,
9
7%) where water is not present to compete with NMO for
SO pyridine at 0 °C. The reaction was usually finished within
3
3
coordination to boron. Perhaps another explanation for the
enhanced rate under anhydrous conditions results from a differ-
ence in rate of oxidation between boronate esters and boronic
acids, the latter likely being the oxidation substrate in the aqueous
environment at 60 °C.
1ꢀ2 h, and the resulting 14 could be obtained by crystallization
from aqueous or neat i-PrOH in ∼75% yield with a purity of
>90% (w/w), which was sufficient for the subsequent steps.
However, in the case of prolonged reaction times at elevated
temperatures (room temperature or greater) with larger excesses
of oxidant, the product began to overoxidize to dimers such as 22
(Figure 3). Additionally, the reaction was consistently plagued by
the methyl thiomethyl ether byproduct 23 (∼5%) typical of
Although the use of BH THF as the hydroboration agent for
3
3
the conversion of 12 to 13 was found to be reliable on the lab and
on the pilot plant scale in many settings, we sought to render the
process suitable for execution in a manufacturing environment
where the reagent’s low autoignition temperature (90 °C) would
be prohibitive because of safety policies. We therefore investi-
15b
alcohol oxidations using activated DMSO. With this knowl-
edge, the hydroboration/ParikhꢀDoering telescope sequence
was reliably conducted on a 30ꢀ120 kg scale.
gated the borane dimethyl sulfide complex (BH DMS) as a
Hydantoin Formation. The BuchererꢀBergs reaction to
install the masked amino acid functionality was examined with
the primary goal of increasing the yield without compromising
safety for the inherently hazardous cyanide using step. The
hydantoin formation reaction usually finished within 16 h at
30ꢀ35 °C, with care taken to maintain the headspace contain-
ing the gaseous reagents necessary for the transformation (e.g.,
hydrogen cyanide, ammonia, and carbon dioxide). Upon com-
pletion of the formation of hydantoin, the product mixture was
comprised of ∼75% ethyl and ∼25% methyl esters because of
exchange with the methanol solvent. Instead of isolation of the
mixture, the esters were saponified in the same pot and
subsequently crystallized when the pH was lowered to give
desired isomer (()-15 together with 15ꢀ20% racemic diaster-
eomer 24. We have found that desired diastereomer (()-15
kinetically crystallizes out in ∼40ꢀ45% yield upon acidification
of the saponification mixture, after which time (()-15 crystal-
lizes along with (()-24 (i.e., 24 + 25) as a 1:1 mixture. When
the system is left to completely relax over 12ꢀ24 h, a yield of
65ꢀ70% for (()-15 was realized at a 300 gal scale. Unwanted
isomers 24 and 25 are effectively removed in the next step,
resolution (see below). An important safety measure was that
the hydantoin forming reaction could be efficiently and repro-
ducibly conducted with only 1 equiv of potassium cyanide,
which led to reduced exposure through raw material and waste
stream handling. With respect to the quality of ammonium
3
3
possible substitution. Although known to be less hazardous (e.g.,
autoignition temperature of 206 °C) than the THF complex, it is
therefore less reactive. Not surprisingly, significant differences
were encountered between the two reagents as evidenced by the
experimental data listed in Table 1.
As expected, when the reaction was conducted at the same
temperature (0 °C), a significantly longer reaction time (19 h)
was required to achieve full conversion of the starting material,
resulting in a significant decrease in yield. In addition, at 25 °C
full conversion was achieved for the DMS complex within 2 h,
accompanied by a poor yield. The lowered yield is attributed to
competitive reduction of the ethyl ester in the event of a
prolonged reaction time or a higher reaction temperature.
Fortunately, when the reaction was conducted at the midrange
temperature of 5ꢀ10 °C, full conversion was usually achieved
after 6ꢀ8 h, resulting in a comparable yield and purity as in the
BH THF process. The quality of 13 had an important impact
3
3
on the downstream steps. Using 13 with a low purity [<75%
w/w)] in the next step, oxidation (13 f 14), resulted in a low
(
yield of a dark and tacky product that resulted in further
reduced yields in the next steps due to poor crystallization
efficiencies. Using 13 with a purity of >85% (w/w) resulted in
1
4, a powdery yellow solid of high quality [>90% (w/w) in the
assay] that performed well in the subsequent steps. Accord-
ingly, the BH DMS process was performed successfully on the
3
3
2
2
6
0 kg scale, delivering 13 with a yield and quality similar to
carbonate, we observed a significant influence on the yield.
Using a batch of reagent with a low ammonia content (<30%,
those of BH THF but in a safer manner.
3
3
Alcohol Oxidation. The oxidation to ketone 14 was initially
achieved by a Swern oxidation (Scheme 1). However, the
cryogenic temperatures and use of methylene chloride made this
procedure undesireable for large scale production. Therefore, the
indicative of a high NH
HCO fraction), at least 3 equiv of
4 3
“(NH CO ” was necessary to ensure >90% conversion of 14
)
4
2
3
to hydantoins. Using less would result in the full consumption
of 14, but only 75ꢀ80% of hydantoins were formed as a result
of competing decomposition of starting material, leading to
21
process was changed to a ParikhꢀDoering oxidation using
SO pyridine that could be conducted out at more practical
a decreased isolated yield of ∼55%. If (NH
high content of ammonia (30ꢀ34%, indicative of a high
NH CO NH fraction) was used, only 2 equiv was necessary
4 2 3
) CO with a
3
3
temperatures potentially using less toxic solvents. Furthermore,
unlike a Swern-type process using oxalyl chloride or trifluoroa-
cetic anhydride as an activator, the ParikhꢀDoering method
is amenable to conversion control through introduction of
4
2
2
to ensure >90% conversion to hydantoins. As we found the
ammonia content of (NH ) CO to vary from batch to batch,
4
2
3
1
269
dx.doi.org/10.1021/op100325h |Org. Process Res. Dev. 2011, 15, 1266–1274

Organic Process Research & Development
ARTICLE
scale (40 kg) proceeded smoothly, achieving an average yield
of 24% from 14 to (ꢀ)-15 with improved cycle time and
throughput.
Sulfide Oxidation. At the point of sulfone formation, the
second-generation synthesis diverges significantly from the first-
generation synthesis. The poor organic solubility of hydantoin 15
necessitated the deprotectionꢀreprotection sequence upon
going from 15 to sulfone 19 by way of amino diacid 17 and
carbamate diester 18. For instance, 15 is completely insoluble in
typical organic solvents at reasonable temperatures and concen-
trations, whereas 18 is readily soluble in methylene chloride, for
example, in which it can be efficiently oxidized with m-CPBA.
Attempting to avoid the seemingly extra steps as well as
methylene chloride and m-CPBA, we explored options for the
direct conversion of sulfide 15 to sulfone 20. Fortunately, the
sodium salt of 15 is sufficiently soluble in an aqueous environ-
ment to allow the use of water-soluble oxidizing agents such as
Figure 4. Diastereomeric hydantoins purged in the resolution of (()-15.
approximately 3.5 equiv was often used to ensure consistent
conversion to (()-15.
Resolution and Salt Break. With a mixture of (()-15 and
(
()-24 in hand, we were gratified to find that resolution with 1.2
equiv of (R)-phenylglycinol in a 4:1 EtOH/H O mixture gave
2
ammonium salt 16 with a high enantiomeric purity (>97:3 er for
1
5 vs its enantiomer) in up to 38% yield after isolation by
crystallization. Furthermore, unwanted diastereomers 24 and 25
Figure 4) were effectively removed. Subsequent acidification of
24
Oxone or hydrogen peroxide in conjunction with metal
(
25
catalysts (e.g., MeReO or Na WO ). Ultimately, the conver-
3
2
4
an aqueous solution of 16 effected salt break with concomitant
crystallization of free acid 15 in 90ꢀ95% yield on a 300 gal scale
with a further upgrade in stereochemical purity to greater than
sion, yield, economy, and throughput for the hydrogen peroxide/
tungstate system were found to be superior to those of other
methods. Conducting the oxidation at 70 °C in the presence of
9
9.95:0.05 er and dr. With a high stereoisomeric purity estab-
0
.5 mol % sodium tungstate dihydrate ensured a satisfactory rate
lished for hydantoin 15, possessing four nonepimerizable
stereocenters, high stereoisomeric purity is ensured for down-
stream products.
2
3
for complete conversion to sulfone with less than 0.5% inter-
mediate sulfoxide remaining at the end of the reaction, typically
4
h after controlled addition of 35% hydrogen peroxide. An
Over the course of conducting several pilot plant campaigns
utilizing the procedures described above for the sequence span-
ning the conversion of ketone 14 to hydantoin 15, we generally
found that the isolation of (()-15 via crystallization was a
bottleneck on a plant scale with respect to cycle time and
throughput. Slow crystallization rates were occasionally experi-
enced as a function of decreased starting material purity. Under
such circumstances, there could be substantial filtrate losses
resulting in 5ꢀ10% lower than expected yields. Furthermore,
the filtration and drying of (()-15 were often excessively time-
consuming as the material frequently crystallized as a thick
pastelike slurry of fine particles. To avoid the filtrate losses and
eliminate long filtration and drying times, an extractive workup
process was designed. After ester saponification with sodium
hydroxide, the pH was lowered to 7, whereby an extraction with
ethyl acetate was employed to remove organic-soluble impurities.
Further lowering the aqueous layer pH to 1 and washing with
tetrahydrofuran effectively extracted (()-15 with a <1% loss to
the aqueous phase (72ꢀ74% corrected solution yield, total dr = 4:1).
On the basis of solubility data, tetrahydrofuran was selected as
the solvent for extraction of the racemic hydantoin acids 15
and 24. Because of the high salt content of the reaction mixture,
phase separation was not a problem. The tetrahydrofuran solu-
tion of hydantoins was then solvent exchanged into ethanol
important aspect of the oxidation is pH control, which ensures
that sulfur oxidation is competitive with base-induced peroxide
decomposition. Using 3 equiv of 35% hydrogen peroxide with an
initial reaction pH of 5ꢀ6 allowed for optimal performance
without stalling. Isolated yields of 91ꢀ94% with purities of >99%
(
w/w) were obtained after the pH had been lowered to induce
crystallization of acid 20 followed by filtration and drying. The
process could be reproduced across multiple plant batches on a
4
5ꢀ75 kg scale. The typical level of tungsten content in the
product was <300 ppm on a pilot plant scale. At such levels,
further processing of 20 through the hydrolysis and crystal-
lization of the next step ensured that the tungsten specification of
6
0 ppm for API 9 would be met (see below).
Hydantoin Hydrolysis. Previous experience with analogous
0
molecules bearing the 5,5 -disubstituted hydantoin moiety has
shown the hydrolysis to be a difficult transformation requiring
harsh conditions whether in an alkaline or acidic environment
9
(
see below). As a result, we were not optimistic that a direct
conversion of hydantoin 20 to dimethyl ester 10 would be
feasible; therefore, a two-step approach of hydantoin hydrolysis
followed by esterification was pursued. The initial hydantoin
cleavage process that we utilized for early material deliveries
employed 50% sodium hydroxide at reflux. The conversion was
slow, taking more than 40ꢀ50 h for acceptable starting material
consumption, and the yield was only moderate at 80ꢀ85%
because of the competitive production of multiple byproducts.
The product could be crystallized upon lowering the pH to the
range of its isoelectric point (pH 1ꢀ2). The desire for a faster
reaction rate and better yield led to the exploration of alternative
conditions for the hydrolysis such as strong mineral acids at
reflux. The use of 48% HBr (8.9 M) as the reaction solvent gave a
significantly improved yield versus that of the sodium hydroxide
method. Of the acids screened, hydrobromic acid was superior in
terms of rate and ease of workup. For example, there was a 2-fold
rate increase over that of 9 M hydrochloric acid (18 h vs 36 h to
reach completion), and the use of 9 M sulfuric acid, while
followed by dilution with H O, and treatment with (R)-phenyl-
2
glycinol and triethylamine. As part of this telescope process
development effort, we found that the amount of phenylglycinol
could be reduced by substitution in part with an inexpensive
achiral base such as triethylamine while the yield and enantio-
meric purity were maintained. It was important that the total
amount of base equaled the total amount of carboxylic acids
present in the form of (()-15 and isomers 24 and 25. Using 0.85
equiv of phenylglycinol combined with 0.35 equiv of Et N
3
resulted in an isolated yield of 36ꢀ38% of 16 in the lab and on
a pilot plant scale with an enantiomeric purity of >95:5 er for 15
versus its enantiomer. Execution of the overall process on a plant
1
270
dx.doi.org/10.1021/op100325h |Org. Process Res. Dev. 2011, 15, 1266–1274

Organic Process Research & Development
ARTICLE
effecting a hydrolysis rate comparable to that of HBr, gave a
product that was significantly contaminated with sodium bisulfate.
The marked rate difference between HBr and HCl is presumably
due to the higher acid concentration and temperature at reflux
steps and 0.4% yield, including multiple chromatographic pur-
ifications, was streamlined significantly by removing unnecessary
steps and chromatographies. The process hazards were reduced
and the space time yield was increased. The second-generation
process, by which more than 1 metric ton of 10 has been afforded
to date, consists of nine steps starting from thiophene and ethyl
diazoacetate with a 4% overall yield.
(8.9 and 5.5 M and 126 and 108 °C, respectively). Increasing the
pH to the isoelectric point range upon completion of the hydrolysis,
followed by filtration and drying, gave 8 (as its zwitterionic form)
in 94ꢀ96% yields and >99% (w/w) purity at a 9ꢀ70 kg scale.
The tungsten content at this stage was consistently less than
’
EXPERIMENTAL SECTION
1
0 ppm. Interestingly, we found that the addition of 85% phos-
phoric acid to the hydrolysis reaction medium afforded, after pH
adjustment, crystals of 8 with a rectangular shape that were more
readily filtered. This is in contrast to the small, needlelike crystals
of 8 that were obtained previously and which filtered poorly.
Esterification. Bis-esterification of diacid 8 was best conducted
utilizing Fischer-type conditions to minimize waste and expense.
The key to many such esterifications is the ability to adequately
remove the water byproduct to drive the equilibrating mixture
toward product formation, thereby maximizing conversion and
yield. We found that typical mineral acids such as concentrated
hydrochloric or sulfuric acids did not provide acceptable conver-
sion. The use of gaseous hydrogen chloride would presumably give
a higher level of conversion, but our preference was to avoid using
gaseous reagents if possible because of the relative handling
inconvenience. Wefoundacceptable conversioncould be achieved
byrefluxing8 in methanolinthe presence of trimethylsilylchloride
General. Monitoring of reaction progress and assay of the
final compounds were conducted by HPLC and comparison to
reference substances. The individual HPLC conditions can be
found in the experimental procedures where applicable. Full
characterization data, including HRMS data, are provided for
new compounds 10 and 20.
(()-(1R,5R,6S)-2-Thiabicyclo[3.1.0]hex-3-ene-6-carboxylic
Acid, Ethyl Ester (12). Caution! Ethyl diazoacetate is a toxic
and explosive substance. Consult the MSDS and other safety-
related information prior to handling. Follow appropriate
safety precautions. A solution of ethyl diazoacetate (11) (300 g,
2.63 mol) in thiophene (1.6 L) was added over 8 h to a 45 °C
solution of Rh (oct) (0.968 g, 1.24 mmol) in thiophene (2 L).
2 4
After being stirred for an additional 2ꢀ3 h, the reaction mixture
was cooled to room temperature. Of the total mass (3985 g),
3240 g (81%) was concentrated by rotary evaporation to an
orange-colored oil that was further distilled using a wiped film
distillation apparatus at 1.5 Torr and 23 °C to remove residual
thiophene prior to product distillation. The product was distilled
at 120 °C and 1.3 Torr to produce 187 g (44% purity-corrected
yield) of crude 12 as a yellow oil. After dissolution of a 23.5 g
portion of the oil in MeOH (2 mL per 1 g of distillate) and after it
had cooled to ꢀ10 °C, seed crystals were added and the resulting
slurry was further cooled to ꢀ45 °C for 2ꢀ3 h. Filtration and
washing with ꢀ45 °C MeOH (1 ꢁ 1 mL per gram of distillate)
2
6
(
3.5 equiv). Typical conversions under these conditions were
greater than 97% at scales from 22 to 80 kg. Over the course of the
reaction, product 10 would crystallize out of solution. To achieve a
higher crystallization yield, tert-butanol was added as an antisol-
vent. Apart from solubility considerations, tert-butanol was se-
lected as the antisolvent because of its miscibility in the reaction
system and inability to undergo ester exchange with the product.
Once the resultant slurry had been cooled, filtered, and dried,
dimethyl ester amine hydrochloride salt 10 was produced in
2
8
8
9ꢀ91% yield. The product at this stage was an unstable methanol
gave 12.6 g (31% overall yield) of 12 as an off-white solid: mp
+ 1
solvate that would slowly convert to a stable hydrate form. The
36 °C (DSC); FDMS M = 170; H NMR (CDCl , 300 MHz) δ
3
27
hydrate form performed much better downstream; thus, we
added a form conversion step to deliberately generate the hydrated
species in a controlled manner. The form conversion consisted of
reslurrying the methanol solvate in heptane followed by the addi-
tion of wet tert-butanol at room temperature to give the hydrated
form of 10 in quantitative yield with purities of >97% (w/w).
Heptane was chosen as the reslurry solvent to eliminate filtrate
losses. However, the water was poorly dispersed when added
directly to the heptane slurry, producing a clumpy heterogeneous
product. When co-administered with tert-butanol, the water was
well mixed throughout the slurry, giving the product as a uniform
powder. Under these conditions, the conversion was facile, being
limited by mass transfer according to the rate of water intro-
duction. Because of the inability of X-ray powder diffraction to
distinguish between the forms, we found that either Raman
1.09 (dd, 1H, J = 3 and 3 Hz), 1.26 (t, 3H, J = 7 Hz), 3.04 (ddd,
1H, J = 2.6, 3.3, and 7.5 Hz), 3.49 (ddd, 1H, J = 1.8, 3.0, and 7.5
Hz), 4.15 (q, 2H, J = 7 Hz), 5.88 (dd, 1H, J = 5.6 and 2.6 Hz),
13
6.15 (dd, 1H, J = 1.8 and 5.6 Hz); C NMR (CDCl , 75 MHz) δ
3
14.3, 23.9, 34.1, 39.0, 60.9, 122.9, 127.7, 174.1; FTIR (ATR)
3074 (w), 2985 (w), 1701 (s), 1553 (w), 1372 (s), 1349 (s), 1274 (s),
ꢀ
1
1219 (s), 1157 (s), 1009 (s), 796 (s) cm . Reaction monitoring
and product (12, retention time of ∼16 min) assay determina-
tion were conducted by HPLC analysis with UV detection at
215 nm. The instrument was equipped with an Atlantis d-C18,
3 μm, 150 mm ꢁ 4.6 mm column. The solvent system consisted
of acetonitrile and water (5:95 ramped to 80:20 over 25 min)
running at a flow rate of 1.5 mL/min.
(()-(1R,5S,6S)-2-Thia-4-oxobicyclo[3.1.0]hexane-6-carboxylic
Acid, Ethyl Ester (14) (boraneꢀdimethylsulfide method).
1
29
spectroscopy or H NMR (monitoring the disappearance of
Neat BH DMS (71 mL, 0.77 mol) was added dropwise over
3
3
methanol) could be employed to monitor the conversion. How-
ever, as the rate of conversion coincides with the rate of addition of
water, monitoring the conversion is typically unnecessary.
20 min to a cold (0 °C) solution of 12 (97 g, 0.57 mol) in THF
(640 mL), keeping the temperature below 5 °C. The mixture was
stirred for 5ꢀ6 h at 5ꢀ10 °C. After the mixture had cooled
to ꢀ10 °C, MeOH (90 mL) was added dropwise over 30 min,
keeping the temperature below 0 °C. After the mixture had been
stirred at 0 °C for 30 min, an aqueous solution of N-methylmor-
pholine N-oxide [50% (w/w), 138 mL] was added over 30 min. The
resulting mixture was refluxed for 4 h, cooled to room tempera-
ture, and diluted with EtOAc (1 L) and extracted with aqueous
’
CONCLUSION
A scalable and robust process for production of synthetic
intermediate 10 was developed and successfully implemented at a
pilot plant scale. The first generation process, which gave 10 in 12
1
271
dx.doi.org/10.1021/op100325h |Org. Process Res. Dev. 2011, 15, 1266–1274

Organic Process Research & Development
ARTICLE
sulfuric acid [5% (w/w), 580 mL]. The organic layer was washed
with a 1:1 mixture of brine and NaOH (1 N) (2 ꢁ 1 L) and once
with brine (500 mL). The organic layer was azeotropically
distilled under vacuum to ensure a solution water content of
and water with 0.05% trifluoroacetic acid (5:95 ramped to 80:20
over 25 min) running at a flow rate of 1.5 mL/min.
0
0
0
0
(ꢀ)-(1 R,4S,5 S,6 S)-2,5-Dioxospiro[imidazolidine-4,4 -
0
[2]thiabicyclo[3.1.0]hexane]-6 -carboxylic Acid [(ꢀ)-15]
<
0.3% (w/w) by KF to ultimately provide ∼180 mL of a solution
[telescope from 14 to (ꢀ)-15 with extractive isolation of
(()-15]. Caution! Potassium cyanide and hydrogen cyanide
are extremely toxic substances. Consult the MSDS and other
safety-related information prior to handling. Follow appro-
of 13 (∼70 g, ∼0.37 mol) in EtOAc that was directly used for the
subsequent oxidation.
To the solution described above were added additional EtOAc
400 mL) and N,N-diisopropylethylamine (508 mL, 2.95 mol),
(
priate safety precautions. A mixture of (NH ) CO (30.7 g,
4 2 3
and the mixture was cooled to 0 °C, followed by the addition of a
0.32 mol) and KCN (6.1 g, 0.093 mol) in 119 mL of methanol
was stirred at room temperature for 75 min, at which time a
solution of 14 (17.4 g, 0.093 mol) in 119 mL of methanol was
introduced. The reaction vessel was closed, and the mixture was
solution of SO pyridine [95% (w/w), 159 g, 0.95 mol] in
3
3
DMSO (635 mL) over 10 min, keeping the temperature below
2
5 °C. After the mixture had been stirred at 0 °C for 1ꢀ2 h,
aqueous HCl (3 N) was added over 10ꢀ15 min. After phase
stirred at 30ꢀ35 °C for 16 h. After addition of H
2
O (50 mL), the
separation, the aqueous layer was re-extracted with EtOAc
mixture was concentrated by vacuum distillation to 50 mL
followed by the addition of aqueous NaOH [20% (w/w),
53 mL] and stirred for 2 h (25 °C). Concentrated HCl
(16 mL) was added to adjust the pH to 7ꢀ8, followed by an
extraction with EtOAc (55 mL). The aqueous phase was further
acidified to pH 1 upon addition of concentrated HCl (16 mL).
The resulting slurry was twice extracted with THF (100 mL +
50 mL) to give a solution of (()-15 (∼15 g, ∼0.066 mol,
70ꢀ74% yield) in THF that was directly used for the resolution.
This solution was concentrated by vacuum distillation to
∼30 mL; EtOH (150 mL) was added and the mixture again
vacuum distilled to ∼30 mL. After addition of EtOH (120 mL)
(
430 mL), and the combined organic layers were washed sequentially
with a saturated solution of NaHCO (870 mL) and brine (575 mL).
3
The EtOAc was replaced with i-PrOH by vacuum distillation to
provide a solution of 14 in i-PrOH (180 mL). After cooling to
0
°C, the resulting yellow suspension was stirred for 10 h. After
the mixture had been filtered, washed with cold i-PrOH (80 mL),
and dried in vacuo, 14 was obtained as a yellow solid in 50% yield
+
(
1
59 g, 90% potency, 0.285 mol): mp 56ꢀ57 °C; FDMS M =
1
86; H NMR (CDCl , 300 MHz) δ 1.27 (t, 3H, J = 7 Hz), 2.34
3
(
1
dd, 1H, J = 3 and 3 Hz), 2.58ꢀ2.60 (m, 1H), 3.19ꢀ3.23 (m,
1
3
H), 3.30 (d, 2H, J = 2 Hz), 4.16 (q, 2H, J = 7 Hz); C NMR
(
CDCl , 75 MHz) δ 14.2, 29.3, 30.8, 34.1, 34.9, 61.7, 168.8,
and enough H O (10 mL in this case) to bring the total water
2
3
2
1
06.4; FTIR (ATR) 2991 (w), 1713 (s), 1402 (m), 1379 (m),
268 (m), 1182 (s), 1005 (m), 830 (m) cm . Reaction monitor-
content to 15 wt %, (R)-phenylglycinol (7.7 g, 0.056 mol) and
triethylamine (3.2 mL, 0.023 mol) were added. The mixture was
refluxed for 1 h and cooled to 0 °C over 9 h and stirred at 0 °C for
an additional 5 h. After the sample had been filtered, washed with
ꢀ1
ing and product (14, retention time of ∼2.9 min) assay deter-
mination were conducted by HPLC analysis with UV detection
at 215 nm. The instrument was equipped with an Atlantis d-C18,
a cold EtOH/H O mixture (4:1), and dried in vacuo, 16 was
2
3
μm, 50 mm ꢁ 4.6 mm column. The solvent system consisted of
isolated as a gray to brown solid (∼9 g, ∼0.025 mol) in ∼38%
acetonitrile and water (20:80, isocratic) running at a flow rate
of 2 mL/min.
yield (for the resolution step).
This solid was then suspended in H O (54 mL), heated to
2
(
()-(1R,5S,6S)-2-Thia-4-hydroxybicyclo[3.1.0]hexane-6-
70ꢀ75 °C, and acidified with concentrated HCl (9 mL, target pH
of <1). The resulting suspension was cooled to 0 °C and stirred
for 3 h. After the mixture had been filtered, washed with cold
carboxylic Acid, Ethyl Ester (13) (boraneꢀtetrahydrofuran
method). To a ꢀ10 °C solution of 12 (35 kg, 206 mol) in 125 L
of THF was added BH THF (1 M in THF, 267 L, 267 mol)
H O, and dried in vacuo, (ꢀ)-15 was obtained as an off-white
3
3
2
29
over 75 min at <0 °C. After being stirred at ꢀ5 °C for 5ꢀ6 h, the
solid (5.1 g, 0.022 mol) in 24% yield (based on 14) and >99% ee
2
5
mixture was cooled to ꢀ10 °C and treated with methanol (23 kg,
and >99% de: [α]
ꢀ202 (c = 1.0, MeOH); mp 330 °C
D
+
1
7
20 mol) over 1 h at <0 °C. After the mixture had been stirred
(DSC); FDMS M = 228; H NMR (DMSO-d , 300 MHz) δ
6
at ꢀ5 °C for 30 min, 50% aqueous N-methylmorpholine N-oxide
2.01 (t, 1H, J = 2 Hz), 2.29 (dd, 1H, J = 3 and 8 Hz), 2.69 (d, 1H,
(
58 kg, 247 mol) was added over 10ꢀ15 min, heated to 65 °C,
J = 12 Hz), 2.89 (dd, 1H, J = 3 and 8 Hz), 3.15 (d, 1H, J = 12 Hz),
1
3
and stirred for 5ꢀ6 h. The mixture was cooled to 10ꢀ15 °C,
8.10 (br s, 1H), 10.75 (br s, 1H), 12.52 (br s, 1H); C NMR
diluted with EtOAc (320 L), and extracted with 0.5 N HCl (1 ꢁ
(DMSO-d , 75 MHz) δ 24.6, 29.7, 34.2, 34.9, 70.7, 156.0, 171.6,
6
3
50 L). The organic layer was washed with a 1:1 brine/1 N
175.0; FTIR (ATR) 3211 (m), 3079 (w), 1785 (w), 1727 (s),
NaOH mixture (2 ꢁ 350 L) followed by brine (1 ꢁ 175 L). The
product-containing organic layer was azeotropically distilled
under vacuum to a target KF of 0.3%. The resulting ethyl acetate
solution was forward processed as outlined for the bora-
neꢀdimethylsulfide procedure described above. Reaction mon-
itoring (product 13 retention time of ∼1.3 min) were performed
by HPLC with UV detection at 215 nm. The instrument was
equipped with an Atlantis d-C18, 3 μm, 50 mm ꢁ 4.6 mm
column. The solvent system consisted of acetonitrile, trifluor-
oacetic acid, and water (20:80:0.05, isocratic) running at a flow
rate of 2 mL/min. Product (13, retention time of ∼8.2 min)
solution assay determination was conducted by HPLC analysis
with UV detection at 215 nm. The instrument was equipped with
an Atlantis d-C18, 3 μm, 150 mm ꢁ 4.6 mm column. The solvent
system consisted of acetonitrile with 0.05% trifluoroacetic acid
1713 (s), 1702 (s), 1430 (m), 1397 (m), 1244 (m), 1179 (m),
ꢀ1
1124 (m), 1022 (m), 882 (m), 731 (s), 715 (s), 631 (s) cm
.
See the HPLC method information in the nontelescoped,
individual procedures below.
0
0
0
0
(()-(1 R,4S,5 S,6 S)-2,5-Dioxospiro[imidazolidine-4,4 -[2]-
0
thiabicyclo[3.1.0]hexane]-6 -carboxylic Acid [(()-15] [isolation
of (()-15 by crystallization]. Caution! Potassium cyanide and
hydrogen cyanide are extremely toxic substances. Consult the
MSDS and other safety-related information prior to handling.
Follow appropriatesafetyprecautions.A mixture of (NH ) CO
4 2 3
(10.34 g, 107.6 mmol) and KCN (3.51 g, 53.9 mmol) in
methanol (150 mL) was stirred at room temperature for
60ꢀ90 min in a closed vessel. Ketone 14 (10.0 g, 53.7 mmol)
was charged in one portion as a solid and the flask resealed. The
reaction mixture was allowed to stir at 30ꢀ32 °C for 20ꢀ24 h.
1
272
dx.doi.org/10.1021/op100325h |Org. Process Res. Dev. 2011, 15, 1266–1274

Organic Process Research & Development
ARTICLE
1
3
The methanol volume was reduced to approximately 20 mL by
vacuum distillation, at which time 2.75 N NaOH (55 mL) was
charged followed by stirring at 20ꢀ30 °C for 1ꢀ2 h. Treatment
with water (55 mL) followed by adjustment of the pH from ∼13
to 2.0 using dropwise addition of concentrated (12 N) HCl
1H), 13.15 (s, 1H); C NMR (DMSO-d , 125 MHz) δ 21.9,
6
31.7, 44.2, 52.6, 62.7, 156.4, 169.9, 174.4; FTIR (KBr) 3317 (s),
3250 (s), 3211 (s), 3086 (w), 1791 (s), 1742 (s), 1713 (s), 1327 (s),
ꢀ
1
+
1192 (s), 1140 (s) cm ; HRMS m/z found 261.0172 (M + H) ,
C H N O S requires 261.0176. Reaction monitoring and pro-
8
9 2 6
(
∼20 mL) and seeding with (()-15 resulted in product crystal-
lization. The slurry was allowed to stir at room temperature for
ꢀ2 h and then at 0ꢀ3 °C for 12ꢀ36 h. The product was
isolated by suction filtration, rinsed with cold (0ꢀ3 °C) water
2 ꢁ 10 mL), and dried in a vacuum oven at 40ꢀ45 °C to afford
0.69 g of a solid consisting of a 4.6:1 ratio of (()-15 to (()-24.
duct (20, retention time of ∼6.1 min) assay determination were
conducted by HPLC analysis with UV detection at 215 nm.
The instrument was equipped with a Zorbax SB-C8, 3.5 μm,
150 mm ꢁ 4.6 mm column. The solvent system consisted of
5% (v) acetonitrile and 95% (v) 20 mM aqueous phosphate
buffer containing 16.5 mM tetrabutylammonium chloride
1
(
1
The purity-corrected yield was 69% for (()-15. Reaction
monitoring and product (12, retention time of ∼8.6 min; 24,
retention time of ∼11 min) assay determination were conducted
by HPLC analysis with UV detection at 210 nm. The instrument
was equipped with an Atlantis d-C18, 3 μm, 150 mm ꢁ 4.6 mm
column. The solvent system consisted of acetonitrile and water
with 0.1% phosphoric acid (5:95 ramped to 45:55 over 30 min)
running at a flow rate of 1.0 mL/min.
(
pH 3) running at a flow rate of 1.5 mL/min (isocratic).
(ꢀ)-(1R,4S,5S,6S)-4-Amino-2-thiabicyclo[3.1.0]hexane-4,
6-dicarboxylic Acid, 2,2-Dioxide (8). To aqueous 48% HBr
(192.4 kg) and aqueous 85% H PO (34.0 kg) was added 43 kg
(165 mol) of 20. The mixture was heated to 130 °C and allowed to
stir for 25 h. After the mixture had cooled to 80 °C, H O (137 L)
3
4
2
was added followed slowly by 90ꢀ95 L of 10 N NaOH. After the
mixture had cooled further to 25 °C, the pH (1.7) was within the
target range of 1.4ꢀ1.8 and did not require further adjustment.
The slurry was cooled to 2 °C and allowed to stir for 4 h. Filtration,
washingwith 0 °C water(2 ꢁ 135 L), washing with 0 °C methanol
0
0
0
0
(
ꢀ)-(1 R,4S,5 S,6 S)-2,5-Dioxospiro[imidazolidine-4,4 -[2]-
0
thiabicyclo[3.1.0]hexane]-6 -carboxylic Acid [(ꢀ)-15] [resolution
of (()-15 that had been isolated by crystallization]. To a
4
(
.3:1 diastereomeric mixture of hydantoins (()-15 and (()-24
(
1 ꢁ 270 L), and vacuum drying at 65 °C gave 37 kg (94%) of 8 as
D D
20
20
39.3 kg, 172 mol) along with 24.4 kg (178 mol) of (R)-
an off-white solid: [α] ꢀ18 (c = 1.02, 1 N HCl); [α] ꢀ76
+
phenylglycinol were added 540 L of ethanol and 135 L of water.
The slurry was heated to 78ꢀ82 °C with stirring until complete
dissolution was achieved. The mixture was allowed to cool to
(
c = 1.52, 1 N NaOH); mp 308 °C (DSC); FDMS M + 1 = 236;
1
H NMR (D O, KOD) δ 2.22 (m, 1H), 2.63ꢀ2.66 (m, 1H) 3.08
2
(
d, 1H, J = 14.9 Hz), 3.32ꢀ3.40 (m, 1H), 3.66 (d, 1H, J = 14.9
13
0
°C over 7 h and held for 3 h. The slurry was filtered, washed
Hz); C NMR (D O, KOD) δ 24.7, 32.0, 43.0, 54.6, 60.49, 174.2,
2
with a 4:1 ethanol/water mixture (1 ꢁ 137 L), and blown with
1
1
8
74.3; FTIR (KBr) 3439 (m), 3068 (m), 2995 (m), 2944 (m),
709 (s), 1653 (m), 1542 (m), 1324 (s), 1248 (s), 1192 (s),
nitrogen for 4 h. Analysis of the wet cake revealed the presence of
ꢀ
1
1
3 wt % ethanol, 6 wt % water, 1.2 area % (+)-15, and 1.9 area %
43 (m) cm . Reaction monitoring and product (8, retention
diastereomers 24 and 25 combined. The filter cake was dissolved
with 137 L of water with stirring at 65ꢀ70 °C and then
transferred and rinsed (20 L) into a glass-lined tank and cooled
to 50ꢀ55 °C. Treatment with 32% HCl to pH ∼0 produced a
slurry that was cooled to 0 °C over 2 h, and then stirred for 2 h,
filtered, washed with 0 °C water (2 ꢁ 25 L), and dried in vacuo at
time of ∼2.7 min) assay determination were conducted by HPLC
analysis with UV detection at 215 nm. The instrument was
equipped with a Zorbax SB-C8, 3.5 μm, 150 mm ꢁ 4.6 mm
column. The solvent system consisted of 5% (v) acetonitrile and
9
5% (v) 20 mM aqueous phosphate buffer containing 16.5 mM
tetrabutylammonium chloride (pH 3) running at a flow rate of
6
0 °C to give 13.2 kg of (ꢀ)-15 (38% corrected yield from 14) as
1
.5 mL/min (isocratic).
ꢀ)-(1R,4S,5S,6S)-4-Amino-2-thiabicyclo[3.1.0]hexane-4,
-dicarboxylic Acid, Dimethyl Ester, 2,2-Dioxide, Hydro-
a brown solid with high isomeric purity (>99.9:0.1 er and dr by
(
HPLC area %). Product [(ꢀ)-15, retention time of ∼8.2 min;
6
(
+)-15, retention time of ∼5.0 min] enantiomeric purities were
chloride (10). A slurry of 8 (18.2 kg, 98.8 wt % via HPLC assay,
76.4 mol) in MeOH (63 L) was heated to 48ꢀ53 °C followed by
the addition of TMSCl (30.0 kg, 276 mol). The biphasic mixture
was warmed to 55ꢀ60 °C and allowed to stir for 1ꢀ2 h, at which
time the mixture was seeded with hydrate 10 (∼0.4 wt %) to induce
product crystallization if spontaneous nucleation had not already
occurred. After the mixture had been stirred for a total of 21 h,
t-BuOH (63 L, warmed to 45ꢀ55 °C) was added over 30 min.
The resultant slurry was cooled to 28 °C over 1 h and allowed to
stir for an additional 3.5 h. The product was isolated by pressure
filtration and washed with t-BuOH (1 ꢁ 90ꢀ95 L) and then
dried at 50 °C to give the methanol solvate of 10, which was
directly converted to the hydrate form of 10 within the agitated
filter dryer. Thusly, the dried filter cake was suspended in heptane
(126 L) by being stirred at 25 °C and then was treated over 110 min
with a solution of aqueous t-BuOH prepared from 23.8 kg of
t-BuOH and 1.6 kg of water. After the mixture had been stirred at
25 °C for 75 min, pressure filtered, washed with heptane (1 ꢁ
determined by HPLC analysis with UV detection at 215 nm. The
instrument was equipped with a Chiralpak IA, 5 μm, 250 mm ꢁ
4
.6 mm column. The solvent system was a hexane/ethanol/
trifluoroacetic acid mixture [60:40:0.1 (v/v), isocratic] running
at a flow rate of 1.0 mL/min.
0
0
0
0
(
ꢀ)-(1 R,4S,5 S,6 S)-2,5-Dioxospiro[imidazolidine-4,4 -[2]-
0 0 0
thiabicyclo[3.1.0]hexane]-6 -carboxylic Acid, 2 ,2 -Dioxide
20). To a mixture of water (64.2 L) and 50% NaOH (13.4 kg)
at 20ꢀ30 °C was added 0.32 kg of Na WO 2H O (0.97 mol)
(
2
4
3
2
followed by 43 kg (189 mol) of (ꢀ)-15. The mixture (pH 5.1)
was heated to 65ꢀ75 °C, at which time 35% H O (55.7 kg, 573
2
2
mol) was added over 4 h to maintain the temperature range target.
After being stirred at 70 °C for 4 h, the mixture was slowly
acidified via addition of 32% HCl to a target pH of 0ꢀ1. The
resultant slurry was cooled to 0 °C over 1 h and then stirred for
1
h, filtered, and washed with 20 °C i-PrOH (2 ꢁ 110 L). The wet
cake was dried in vacuo at 50 °C to give 44.9 kg (91%) of 20 as an
25
off-white solid: [α] ꢀ49 (c = 1.19, 1 N NaOH); mp 353 °C
30ꢀ35 L), and dried at 50 °C, 10 was isolated as a white solid in
D
1
25
(
DSC); H NMR (DMSO-d , 500 MHz) δ 2.39 (t, 1H, J = 4 Hz),
91% yield (22.2 kg, 93.9 wt % HPLC assay, 69.6 mol): [α]
6
D
25
2
1
.80 (dd, 1H, J = 7 and 4 Hz), 3.03 (d, 1H, J = 16 Hz), 3.74 (dd,
H, J = 7 and 4 Hz), 3.85 (d, 1H, J = 15 Hz), 8.13 (s, 1H), 10.99 (s,
ꢀ22 (c = 1.00, MeOH); [α] ꢀ21 (c = 1.01, H O); mp 194 °C
D
2
1
(DSC); H NMR (DMSO-d , 400 MHz) δ 2.96 (dd, 1H, J = 7.2
6
1
273
dx.doi.org/10.1021/op100325h |Org. Process Res. Dev. 2011, 15, 1266–1274

Organic Process Research & Development
ARTICLE
and 4.2 Hz), 3.35 (t, 1H, J = 4.0 Hz), 3.70 (s, 3H), 3.79 (d, 1H, J =
B. G.; Andis, S. L.; Kingston, A.; Schoepp, D. D. J. Med. Chem. 2007,
50, 233.
13
1
5.2 Hz), 3.83 (s, 3H), 3.95ꢀ4.02 (m, 2H), 9.52 (br s, 3H); C
(12) Patil, S. T.; Zhang, L.; Martenyi, F.; Lowe, S. W.; Jackson, K. A.;
NMR (DMSO-d , 100 MHz) δ 167.8, 167.1, 58.2, 54.0, 52.7,
6
Andreev, B. V.; Avedisova, A. S.; Bardenstein, L. M.; Gurovich, I. Y.;
Morozova, M. A.; Mosolov, S. N.; Neznanov, N. G.; Reznik, A. M.;
Smulevich, A. B.; Tochilov, V. A.; Johnson, B. G.; Monn, J. A.; Schoepp,
D. D. Nat. Med. 2007, 13, 1102.
5
1.1, 42.7, 29.2, 21.0; FTIR (ATR) 3559 (w), 3182 (w), 3012
(w), 1753 (m), 1730 (s), 1649 (w), 1581 (w), 1324 (s), 1301 (s),
ꢀ
1
1
210 (s), 1174 (s), 1119 (s), 767 (s) cm ; FIA-APCI-MS
(
positive ion mode) m/z (M + H) 264 (free base); water content
(
(
13) Moher, E. D.; Monn, J. A. U.S. Patent 7,671,082, 2010.
14) (a) Gillespie, R. J.; Porter, A. E. A. J. Chem. Soc., Perkin Trans. 1
3
.2% by KF; chloride content 11.4% by potentiometric titration
+
3 9 14 6
(
(
(
AgNO ); HRMS m/z found 264.0533 (M + H) , C H NO S
1979, 2624. (b) Tranmer, G. K.; Capretta, A. Tetrahedron 1998,
54, 15499.
(15) (a) Mancuso, A. J.; Swern, D. Synthesis 1981, 165. (b) Omura,
K.; Swern, D. Tetrahedron 1978, 34, 1651.
free base) requires 264.0536. Reaction monitoring and product
10, retention time of ∼6.6 min) assay determination were con-
ducted by HPLC analysis with UV detection at 210 nm. The
(16) (a) Ware, E. Chem. Rev. 1950, 46, 403. (b) Bucherer, H. T.;
instrument was equipped with a Gemini C-18, 3 μm, 150 mm ꢁ
Fischbeck, H. T. J. Prakt. Chem. 1934, 140, 69. (c) Bucherer, H. T.;
Steiner, W. J. Prakt. Chem. 1934, 140, 291.(d) Bergs, H. German Patent
4
.6 mm column. The solvent system was a 5:95 (v/v) acetoni-
trile/aqueous buffer [22 mM NaH PO containing 10 mM
2
4
5
66094, 1929.
17) For example, see: Doyle, M. P.; Peterson, C. S.; Parker, D. L.
Angew. Chem., Int. Ed. 1996, 35, 1334.
18) Because of the hazards associated with ethyl diazoacetate
Clark, J. D.; Shah, A. S.; Peterson, J. C. Thermochim. Acta 2002,
92, 177) and the potential for other hazards associated with heating
tetrabutylammonium chloride (pH to 7.6 with NaOH)] mixture
running at a flow rate of 1.0 mL/min (isocratic).
(
(
’
AUTHOR INFORMATION
(
3
Corresponding Author
*
and distilling crude 12, process safety analysis has been conducted prior
to scale-up by the individual manufacturers responsible for the supply of
E-mail: mario.waser@jku.at (M.W.) or e.moher@lilly.com
E.D.M.).
(
1
2.
(
(
19) Brown, H. C.; Rao, B. C. S. J. Am. Chem. Soc. 1956, 78, 5694.
20) Kabalka, G. W.; Hedgecock, H. C., Jr. J. Org. Chem. 1975,
Present Addresses
Institute of Organic Chemistry, Johannes Kepler University
Linz, Altenberger Straße 69, 4040 Linz, Austria.
4
0, 1776.
(
(
21) Parikh, J. R.; Doering, W. v. E. J. Am. Chem. Soc. 1967, 89, 5505.
22) Commercially available (NH ) CO is a mixture of NH HCO
3
4
2
3
4
4 2 2
and NH CO NH in varying ratios. The quality in terms of ammonia
’
ACKNOWLEDGMENT
content was determined according to the USP NF monograph for
ammonium carbonate.
We thank David Anderson, Brian Axe, Mindy Forst, Steven
(23) In theory, C-6 would appear to be an epimerizable stereocenter
Bandy, Robert Montgomery, and Zachary Hart for their important
analytical contributions. Gary Rhodes, Tricia Aust, and Erica Buxton
are recognized for oversight of pilot plant operations. Additionally,
we are grateful for the numerous experiments involving the
attempted epimerization of 12 by collaborators at University
College Cork (Cork, Ireland) in the laboratory of Professor Anita
Maguire, namely, Drs. Marie Kissane and Orla McNamara.
because of the attachment of an “acidic” proton. However, several
attempts to incorporate deuterium at C-6 of 12, 13, and 15 via base-
induced enolization followed by quenching with D O were uniformly
2
met with failure; the proteo starting material was recovered unchanged
or suffered decomposition. From these results, we consider C-6 to be
nonepimerizable.
’
REFERENCES
(1) Hollmann, M.; Heinemann, S. Annu. Rev. Neurosci. 1994, 17, 31.
(
2) Nakanishi, S.; Masu, M. Annu. Rev. Biophys. Biomol. Struct. 1994,
(
24) (a) Webb, K. S. Tetrahedron Lett. 1994, 35, 3457. (b) Yang, D.;
Yip, Y.-C.; Jiao, G.-S.; Wong, M.-K. J. Org. Chem. 1998, 63, 8952.
25) (a) Schultz, H. S.; Freyermuth, H. B.; Buc, S. R. J. Org. Chem.
963, 28, 1140. (b) Yamazaki, S. Bull. Chem. Soc. Jpn. 1996, 69, 2955.
2
3, 319.
(
3) Schoepp, D. D.; Jane, D. E.; Monn, J. A. Neuropharmacology
(
1
999, 38, 1431.
1
(
(
4) Marino, M. J.; Conn, J. P. Curr. Drug Targets 2002, 1, 239.
5) Pellicciari, R.; Constantino, G. Curr. Opin. Chem. Biol. 1999,
(
(
26) Brook, M. A.; Chan, T. H. Synthesis 1983, 201.
27) For instance, the methanol solvate immediately became a
3
, 433.
semisolid, sticky mass upon exposure to the THF solvent used in the
next step. The hydrate gave a free-flowing slurry by contrast.
(
moisture and oxygen is recommended because of the low melting point
and degradation potential.
(
(
6) Salt, T. E. Drug Dev. Res. 2001, 54, 129.
(
7) Tamminga, C. A.; Lahti, A. C. Int. Clin. Psychopharmacol. 1996,
28) Refrigerated storage of 12 in protective packaging to exclude
1
1 (Suppl. 2), 73.
8) Swanson, C. J.; Bures, M.; Johnson, M. P.; Linden, A.-M.; Monn,
J. A.; Schoepp, D. D. Nat. Rev. Drug Discovery 2005, 4, 131.
9) Monn, J. A.; Valli, M. J.; Massey, S. M.; Wright, R. A.; Salhoff,
(
29) For an illustrative overview of the handling of boranes at
(
scale, see: Atkins, W. J.; Burkhardt, E. R.; Matos, K. Org. Process Res.
Dev. 2006, 10, 1292.
C. R.; Johnson, B. G.; Howe, T.; Alt, C. A.; Rhodes, G. A.; Robey, R. L.;
Griffey, K. R.; Tizzano, J. P.; Kallman, M. J.; Helton, D. R.; Schoepp,
D. D. J. Med. Chem. 1997, 40, 528.
(10) Monn, J. A.; Valli, M. J.; Massey, S. M.; Hansen, M. M.; Kress,
T. J.; Wepsiec, J. P.; Harkness, A. R.; Grutsch, J. L.; Wright, R. A.;
Johnson, B. G.; Andis, S. L.; Kingston, A.; Tomlinson, R.; Lewis, R.;
Griffey, K. R.; Tizzano, J. P.; Schoepp, D. D. J. Med. Chem. 1999,
4
2, 1027.
11) Monn, J. A.; Massey, S. M.; Valli, M. J.; Henry, S. S.; Stephenson,
G. A.; Bures, M.; Herin, M.; Catlow, J.; Giera, D.; Wright, R. A.; Johnson,
(
1
274
dx.doi.org/10.1021/op100325h |Org. Process Res. Dev. 2011, 15, 1266–1274