The manufacture of a homochiral 4-silyloxycyclopentenone intermediate for the synthesis of prostaglandin analogues

Article abstract of DOI:10.1021/op300188x

A process is described for the synthesis of kilogram quantities of homochiral 4-silyloxycyclopentenone (R)-1, a key intermediate useful for the synthesis of a plurality of prostaglandin analogue drugs. Cyclopentenone (R)-1 was synthesized in 14 isolated steps from furfural. Key steps in the synthesis include a Wittig reaction, Piancatelli rearrangement, and an enzymatic resolution featuring in situ recycling of the undesired enantiomer furnishing the desired homochiral alcohol in ≥99.5% ee. As a retort to the unsatisfactory coformation of about 8% at best of the trans-olefin in the Wittig reaction, a change to the order of several steps and the identification of a recrystallisable, amine salt derivative, 2, allowed the unwanted isomer to be controlled to as low as 0.2%.

Full text of DOI:10.1021/op300188x

Article
pubs.acs.org/OPRD
The Manufacture of a Homochiral 4‑Silyloxycyclopentenone
Intermediate for the Synthesis of Prostaglandin Analogues
Julian P. Henschke,*,† Yuanlian Liu,‡ Xiaohong Huang,‡ Yungfa Chen,† Dechao Meng,‡ Lizhen Xia,‡
Xiuqiong Wei,‡ Aiping Xie,‡ Danhong Li,‡ Qiang Huang,‡ Ting Sun,‡ Juan Wang,‡ Xuebin Gu,‡
Xinyan Huang,‡ Longhu Wang,† Jun Xiao,‡ and Shenhai Qiu‡
†ScinoPharm Taiwan, Ltd., 1 Nan-Ke Eighth Road, Tainan Science Industrial Park, Shan-Hua, Tainan, Taiwan 74144
‡ScinoPharm (Changshu) Pharmaceutical, Ltd., 16 Dong Zhou Road, Economic Development Zone, Changshu, Jiangsu Province,
China 215513
S
* Supporting Information
ABSTRACT: A process is described for the synthesis of kilogram quantities of homochiral 4-silyloxycyclopentenone (R)-1, a
key intermediate useful for the synthesis of a plurality of prostaglandin analogue drugs. Cyclopentenone (R)-1 was synthesized in
14 isolated steps from furfural. Key steps in the synthesis include a Wittig reaction, Piancatelli rearrangement, and an enzymatic
resolution featuring in situ recycling of the undesired enantiomer furnishing the desired homochiral alcohol in ≥99.5% ee. As a
retort to the unsatisfactory coformation of about 8% at best of the trans-olefin in the Wittig reaction, a change to the order of
several steps and the identification of a recrystallisable, amine salt derivative, 2, allowed the unwanted isomer to be controlled to
as low as 0.2%.
stage in the synthesis of APIs, such as bimatoprost and
lubiprostone. The advantages of the use of a single feedstock
for a plurality of products is obvious and includes significant
time and cost savings during the research and development
stage as well as a lowering of manufacturing costs.8
INTRODUCTION
■
Prostaglandins (PGs) are naturally occurring, bioactive,
polyunsaturated, fatty acid derivatives composed of 20 carbon
atoms. Despite being highly potent and playing important roles
in animal and human physiology, therapeutic applications of
prostaglandins and their synthetic analogues as active
pharmaceutical ingredients (APIs) have been surprisingly
limited, in spite of extensive research efforts in both academia
and industry beginning in the 1960s.1 In contrast to typical
small-molecule drugs, prostaglandins and their synthetic
analogues are stereochemically complex, their syntheses are
long, their intermediates are typically oils (rather than
crystallisable solids), and they can be chemically unstable.
Notwithstanding this, there are a variety of synthetic
prostaglandin analogues on the market available for the
treatment of a range of conditions including peptic ulcer,
hypertension, and fertility control in humans, and veterinary
uses.1 PGF2α analogues, such as latanoprost2 (Xalatan),
bimatoprost3a (Lumigan), and travoprost4 (Travatan) have
achieved success as first-line topical treatments of glaucoma.
More recently tafluprost5 (Zioptan) was also introduced into
the glaucoma market, whilst bimatoprost (Latisse) was
Although the production volumes of these prostaglandin
analogues would be small due to their extreme potency (a
single dose of each of these drugs is in the order of
micrograms), the upstream steps used to make the key
diverging (R)-1 intermediate needed to be robust. Because of
working on kilograms to tens of kilograms, it was preferable to
avoid particularly low or high temperatures, the use of air- and
moisture-sensitive reagents, or specialised equipment. Further,
the common intermediate had to be of high chemical and
stereochemical purity as to be able to provide the downstream
APIs with high stereochemical purity to meet the demands of
regulatory authorities. This was particularly important given
that the single chiral centre of 4-silyloxycyclopentenone (R)-1
was responsible, through chiral induction, for the stereo-
chemistry of up to three chiral centres formed in the
subsequent synthetic steps.
Of the three main approaches (Corey’s9 lactone approach,
the two-component approach,10 and the three-component11
approach) used for prostaglandin and analogue synthesis,12 we
favored the two-component approach, as this strategy would
provide the degree of divergence that we required. The two-
component approach comprises installation of the ω-side chain
via a 1,4-addition of a vinyl cuprate to a cyclopentenone system
already possessing the α-side chain (Figure 1). In this
contribution we report our first-generation process for this
additionally approved for cosmetic use.3b Lubiprostone,6
bicyclic PGE1 derivative, is the active drug substance in Amitiza,
a gastrointestinal drug used for the treatment of chronic
idiopathic constipation in adults.
It was envisioned that all of these APIs could be accessed
from the single, common intermediate, 4-silyloxycyclopente-
none (R)-17 (Figure 1). As part of the original platform
technology design, the isopropyl ester functionality of this
intermediate would act not only as part of the final molecular
structure of APIs, such as travoprost and tafluprost, but also as a
carboxyl protecting group that could be deprotected at a late
a
Received: July 15, 2012
Published: November 15, 2012
© 2012 American Chemical Society
1905
dx.doi.org/10.1021/op300188x | Org. Process Res. Dev. 2012, 16, 1905−1916

Organic Process Research & Development
Article
Figure 1. Use of cyclopentenone (R)-1 as a common intermediate in prostaglandin analogue synthesis.
Scheme 1. Retrosynthetic analysis of cyclopentenone (R)-1
a
key intermediate, with a focus on the Wittig reaction and how
the required levels of geometric and stereochemical purity were
achieved.
Scheme 2. Synthesis of aldehyde 4
RESULTS AND DISCUSSION
■
Attempted Synthesis of Cyclopentenone (R)-1, via
Aldehyde 3. Although a significant number of syntheses of 4-
hydroxy-2-cyclopenten-1-ones with α-side chains useful for
prostaglandin synthesis have been published,7,13 the first
approach to the requisite cyclopentenone intermediate (R)-1
that we investigated was based on our retrosynthetic analysis
shown in Scheme 1 (top pathway). This approach would rely
upon the Piancatelli rearrangement14 of 2-furylcarbinols to 2-
substituted 4-hydroxy-2-cyclopenten-1-ones, a cis-selective
Wittig reaction using the potentially labile aldehyde 3,15 and
the enzymatic resolution of 2-substituted 4-hydroxy-2-cyclo-
penten-1-ones.16
a
Reagents and conditions: (a) (i) Allylbromide, Zn, THF, 60−68 °C;
(ii) TBSCl, imidazole, 1:1 DMF/THF, r.t., 70−77%; (b) NMO,
K2OsO2(OH)4, (DHQ)2PHAL, acetone, H2O, 87%; (c) NalO4,
acetone, H2O, r.t., 92−95%.
(Scheme 2). Dihydroxylation of olefin 5 using NMO in the
presence of catalytic K2OsO2(OH)4 and (DHQ)2PHAL,18
affording diol 6, was followed by oxidative cleavage with NaIO4
in aqueous acetone at ambient temperature to give desired
aldehyde 4 in about 80% yield over two steps. Preliminary tests
suggest that more expedient routes to aldehyde 4 exist.19
On a 3-kg scale, furfural was converted to silyl ether 5 in one
reactor in 70−77% yield with 93−96% GC purity after
fractional distillation. Nineteen kilograms of olefin 5 was
dihydroxylated in a single batch, providing diol 6 (20.1 kg) in
87% yield that was then used without purification. Oxidative
cleavage of diol 6 proceeded smoothly on a 0.2−0.6 kg scale,
supplying oily aldehyde 4 in 92−95% yield (91−95% GC
purity), following rapid flushing through a plug of silica gel.
Aldehyde 4 was dissolved in THF under an inert atmosphere
and used directly to avoid oxidation.20 A 20.1-kg-scale run
Despite examining different solvents, temperatures, and
concentrations and testing normal and reverse addition
modes of the ylide and aldehyde, we were unable to convert
aldehyde 3 into the desired olefin using the Wittig reaction.
Multiple, unidentifiable products were formed due to the
instability of aldehyde 3 under basic conditions.17
Synthesis of Aldehyde 4. We envisioned that this
problem could be circumvented by exploiting the base-stable
nature of the furan ring of aldehyde 4, using it as a latent form
of the sensitive 4-silyloxycyclopent-2-enone ring system. This
would delay the Piancatelli rearrangement until after the α-side
chain had been installed by Wittig reaction (Scheme 1, bottom
pathway). To this end silyl ether 5 was prepared from furfural
in one pot by reaction with in situ formed allyl zinc bromide
followed by trapping of the resultant zinc alkoxide with TBSCl
1906
dx.doi.org/10.1021/op300188x | Org. Process Res. Dev. 2012, 16, 1905−1916

Organic Process Research & Development
Article
provided about 18 kg of crude aldehyde 4 with 93% purity by
GC.
Wittig Reaction of Aldehyde 4. Owing to the failure of
the Wittig reaction of aldehyde 4 and putative ester ylide 7
(Figure 2; instead yielding 3-(furan-2-yl)prop-2-enal and
absence of crystalline intermediates in our original process.
Although the trans-isomer eluted slightly faster than cis-isomer,
9, column chromatography did not present a viable method for
enrichment of the cis-isomer.24
With an immediate need to demonstrate the platform
technology depicted in Figure 1 and to allow development of
the GMP steps (i.e., (R)-1 through to the APIs) to begin, the
process was moved to scale-up, albeit with the intention to
return for a more thorough investigation of the Wittig reaction
to reduce the levels of trans-isomer (see below). The first lot of
ester 8 (10.1 kg; 14 batches of Wittig reaction and 3 batches of
esterification) was prepared by a contractor. The Wittig
reaction was conducted at −25 to −30 °C in THF using
NaHMDS as the base, followed by esterification (i-PrI, K2CO3
at 55 °C in acetone25) and column chromatography, providing
purified product 8 (91−95% GC purity). Desilylation (TBAF
in THF) of a sample of silyl ether 8 gave alcohol 14 that upon
HPLC analysis revealed that this lot of Wittig reaction product
contained 12.2 mol% (relative to the mixture of cis- and trans-
isomers) of the undesired trans-isomer.26
Figure 2. Miscellaneous compounds and Wittig reaction side products.
cyclopentenone) a two-step approach for the synthesis of
ester 8 involving the Wittig reaction followed by esterification
of the resulting carboxylic acid 9 was required (Scheme 3). A
study of the Wittig reaction was initiated with a base screen.
a
Scheme 3. Synthesis of ester 8
Investigations were resumed with a cosolvent study of the
Wittig reaction step in THF at −15 °C, generating ylide 10b
using NaHMDS as the base (Table 1, entry 1).27 The best
improvement in cis-/trans-selectivity was observed when
phosphoramide cosolvents HMPA and tripyrrolidinophosphor-
ic acid triamide (TPPA) were used (entries 4−9).28 Due to its
toxicity, HMPA was replaced with the pyrrolidine analogue,
TPPA.30 When the amount of TPPA in THF was increased
from 5 to 50% (entries 5 to 9) the trans-isomer steadily
decreased from 10.2 to 7.7 mol%, albeit with a large drop in
both the yield (from 75% down to 14%) and purity of olefin 8
along with a significant increase of diene side product 15. The
use of other polar additives in THF (entries 10−15) provided
no overall improvement and were inferior to the two
phosphoramide solvents tested. Using 10% TPPA in THF
and NaHMDS, a study of the influence of temperature (−20 to
−78 °C) on the yield and cis-/trans-selectivity (entries 16−19)
suggested that the best temperature was around −20 to −35
°C.
a
Reagents and conditions: (a) PPh3, PhMe, 87%; (b) (i) LiHMDS,
NaH, NaHMDS, t-BuOK, or KHMDS/THF, THF, 0−10 °C; (ii) 4,
THF, <0 °C; (c) (i) i-PrI, K2CO3, acetone, reflux; (ii) chromatog-
raphy; (d) TBAF, THF, 35−40 °C.
Demonstration of these conditions (0.4 kg scale of 4 based
on entry 17) in our kilo laboratory, followed by conversion of
analytical test samples to ester 14 (entries 20 and 21), showed
that about 8.0−8.3 mol% trans-isomer had formed (70−74%
yield of 8 over two steps from 4), consistent with laboratory-
scale runs.29
As a result of this study, the original conditions (LiHMDS in
THF) were improved such that the level of undesired trans-
isomer could be reduced from ∼20 mol% down to ∼8 mol%,
whilst maintaining the purity of the product, without
necessitating the use of dedicated cryogenic equipment.26
Although ostensibly small and falling short of our goal of ≤3%,
the improvement in selectivity had a significant impact on the
downstream API syntheses through reduced burden placed on
the API and API precursor purification operations. As will be
discussed below, this improvement was later operated together
with a process modification that ultimately allowed levels of less
than 1% of the trans-isomer to be achieved.
Ylides 10a−10c were generated in situ from the
phosphonium salt 11b (prepared from 5-bromopentanoic
acid (11a)) in THF using lithium, sodium, and potassium
bases, respectively, followed by the addition of a THF solution
of aldehyde 4. In a Wittig reaction of a hemiacetal and ylide 10c
(synthesized from 11b and t-BuOK) reported21 for the
synthesis of an advanced precursor of travoprost, good cis-/
trans-selectivity was observed with only 3% of the correspond-
ing trans-isomer being produced at 0 °C, and 5% at r.t. When
we reacted aldehyde 4 and ylide 10c, however, the main
product was a geometric mixture of dienes 12 (Figure 2)
accompanied by only a small amount of the desired olefin 9.
Surprisingly, treatment of ylide 10a, generated using n-BuLi in
THF at 0 °C, with aldehyde 4 provided ketone 13 (Figure 2)
instead of the desired olefin 9.22 Although ylides generated
using NaH in DMSO or KHMDS in THF did not provide any
olefin 9, the use of LiHMDS in THF led to the desired cis-
alkene 9, albeit contaminated with a relatively high 20 mol%
(w.r.t. cis-isomer) of the undesired trans-isomer trans-9.
Consistent with the “Lithium-Salt Effect”23 the undesired
trans-isomer was reduced to 13 mol% upon the use of sodium
ylide 10b, prepared with NaHMDS, in THF at −15 °C.
This isomer proved extremely troublesome; it could not be
purged by any means applicable to scale up due to the complete
Deprotection of Silyl Ether 8 and Piancatelli Re-
arrangement. Silyl ether 8 was treated with 1 equiv of TBAF
in THF at 35−40 °C to furnish alcohol 14 (Scheme 3) in a
quantitative yield on a 6.5 kg scale (77% GC purity). This
material was used as is in the next step without further
purification.
1907
dx.doi.org/10.1021/op300188x | Org. Process Res. Dev. 2012, 16, 1905−1916

Organic Process Research & Development
Article
a
Table 1. Influence of solvent and temperature on the Wittig
Scheme 4. Piancatelli rearrangement of 14
reaction
temp 8 purity 15 purity yield
14 cis-/
a
entry
solvent
THF
(°C) (area%) (area%) (%) 8 trans-ratio
1
−15
−15
−15
88
88
87
1.3
4.6
0.8
82
79
69
87.0/13.0
87.5/12.5
83.3/16.7
b
2
THF
3
50% PhMe
in THF
a
4
10% HMPA
in THF
−15
−15
−15
−15
−15
−15
−15
81
92
88
85
64
25
93
91
88
92
89
5.4
3.5
5.7
5.8
43
75
71
72
44
14
49
58
70
57
68
90.9/9.1
89.8/10.2
91.1/8.9
91.2/8.8
92.2/7.8
92.3/7.7
87.9/12.1
89.5/11.5
89.6/11.4
86.7/13.3
86.1/13.9
Reagents and conditions: (a) ZnCl2, hydroquinone (cat.), dioxane,
H2O, reflux; (b) chloral, Et3N, PhMe, r.t., 55% over two steps
5
5% TPPA in
THF
acid derivative (as the major product therefore requiring a
subsequent re-esterification step), the isopropyl ester function-
ality of 16 was left intact in the Piancatelli reaction.
6
10% TPPA
in THF
7
20% TPPA
in THF
Following aqueous work-up, a PhMe solution of the crude
Piancatelli rearrangement products rac-16 and rac-16b was
transformed in a second rearrangement reaction under known31
conditions (catalytic chloral in the presence of Et3N) into 4-
hydroxycyclopentenone rac-16, devoid of the thermodynami-
cally less stable isomer rac-16b, in about 55% yield calculated
over the two rearrangement steps from 2-furylcarbinol 14.
Attempts to improve the disappointing yield of the
furylcarbinol 14 to cyclopentenone rac-16 conversion, which
we attribute the first rearrangement step, will be addressed in
future process development.34
In a scale-up run the Piancatelli rearrangement was complete
within 6 h using 2-furylcarbinol 14 (manufactured from 15.8 kg
of purified 8), providing a 1.0:1.1 mixture of cyclopentenone
isomers rac-16 and rac-16b. This mixture converted smoothly
in the second rearrangement step into racemic 4-hydroxycy-
clopentenone rac-16 with 98−100% GC purity following
purification.
Enzymatic Resolution of 4-Hydroxcyclopentenone
rac-16. A critical part of the overall strategy35 for the synthesis
of homochiral cyclopentenone (R)-1 relied upon the enzymatic
resolution of the racemic alcohol precursor rac-16. Given the
only modest efficiency already endured in the Piancatelli
rearrangement, however, it was essential that both enantiomers
of the racemate could be fully utilized without unnecessary
wastage. Furthermore, given that the chiral centre at C4 of the
cyclopentenone was crucial for inducing the relative and
absolute configuration of other chiral centres in the
prostaglandin analogues, it was critical that we were able to
resolve the racemic 4-hydroxy-2-alkyl-cyclopentenone rac-16 to
a high degree of stereochemical purity.
8
30% TPPA
in THF
23
51
3.4
9
50% TPPA
in THF
10
11
12
13
14
10% DMPU
in THF
10% NMP in −15
4.1
5.8
5.5
4.0
THF
10% DMAC
in THF
−15
10% DME in −15
THF
10%
−15
TMEDA in
THF
15
16
17
18
19
20
21
10% DMSO
in THF
−15
−20
−35
−55
−78
−36
−36
61
23
43
79
81
61
11
74
70
91.3/8.7
91.9/8.1
91.8/8.2
91.4/8.6
10% TPPA
in THF
94.3
96.1
88
3.0
1.7
5.2
0.4
5.9
5.6
10% TPPA
in THF
10% TPPA
in THF
10% TPPA
in THF
96
not
analysed
10% TPPA
in THF
89
91.7/8.3
10% TPPA
in THF
92
92.0/8.0
a
Whereas the yield and purity (8 and 15) were determined following
esterification of 9, the cis-/trans-ratio (Wittig reaction selectivity) was
measured by analysis of 14. Double the normal volume of THF.
b
A key part of the overall strategy for the synthesis of
cyclopentenone (R)-1 relied upon the Piancatelli rearrange-
ment,14 which has been demonstrated in prostaglandin
analogue synthesis (enisoprost) by Dygos et al.31 When
Piancatelli’s ZnCl2 modification32 in aqueous dioxane31,33a,c
was applied to crude 2-furylcarbinol 14, a ∼1:1 to 2:1 mixture
of cyclopentenone rac-16 and its undesired isomer rac-16b was
obtained within 6−12 h (Scheme 4). Heating beyond reaction
completion encouraged further, secondary rearrangement of
rac-16b to rac-16; however, after 2.75 days about 10% rac-16b
still remained. With time and yield considerations in mind,
instead of converting 2-furylcarbinol 14 completely through to
cyclopentenone rac-16, it was preferred that upon complete
consumption of furylcarbinol 14 the Piancatelli rearrangement
was terminated, and the mixture of isomers (rac-16b and rac-
16) was then processed through the second rearrangement
step. Unlike that reported31 for the methyl ester analogue,
methyl 7-(2-hydroxy-5-oxo-3-cyclopenten-1-yl)-(4Z)-hepte-
noate, that underwent in situ hydrolysis back to its carboxylic
To address these requirements, a 4-hydroxycyclopentenone
enzymatic resolution protocol reported by Babiak and Wong,16
and subsequently modified by Spur et al.,33a was chosen (this
has not been reported previously for cyclopentenone rac-16).
The original protocol involved stereoselective acetylation of the
(R)-enantiomer of 4-hydroxy-2-alkyl-cyclopentenone substrates
with neat vinyl acetate in the presence of porcine pancreatic
lipase (PPL),36 chromatographic separation of ester/alcohol
enantiomorphic mixtures, and a Mitsunobu inversion. Spur et
al.’s subsequent modification, as demonstrated on methyl 7-(3-
hydroxy-5-oxo-1-cyclopenten-1-yl)-heptanoate, bypassed chro-
matographic separation of the first resolved pair (i.e., (S)-
alcohol and (R)-acetate) by instead directly employing in situ
Mitsunobu inversion of the (S)-alcohol of the (R)-acetate,
providing the product in 96% ee in 76% yield following
deacylation with guanidine.
Upon applying37 Babiak and Wong’s protocol16 to the
resolution of racemic alcohol rac-16 (Scheme 5) it was found
1908
dx.doi.org/10.1021/op300188x | Org. Process Res. Dev. 2012, 16, 1905−1916

Organic Process Research & Development
Article
a
of the silyl ether confirmed that it was composed of a 91:9 ratio
of the cis- and trans-isomers of (R)-1.42 This was consistent
with the mixture of geometric isomers observed in the Wittig
reaction feedstock that this material had originated from (see
above), showing that no enrichment of either isomer occurred
throughout the process. As will be described below, a recent
batch of (R)-1 comprising a 99.4:0.6 ratio of cis-/trans-isomers
was prepared using an improved process.
Scheme 5. Resolution of rac-16
Improved Process for (R)-1 with Reduced trans-
Isomer. As described above, the synthesis of diverging
intermediate (R)-1 proceeded on multikilogram scales;
however, significant improvements remained to be made to
the original process to improve overall efficiency. In addition to
resorting to a series of chromatographic purifications owing to
the nonsolid nature of all intermediates, the cis-/trans-selectivity
of the Wittig reaction was unsatisfactory. Apart from at least 8%
w/w of material (i.e., the trans-isomer) processed through this
route being useless, a costly burden was placed on the
downstream purification operations to remove the unwanted
isomers, which possessed very similar chromatographic
behaviours, from the APIs themselves. Having now supplied
sufficient homochiral (R)-1 for preparation of hundreds of
grams of prostaglandin analogue APIs, attention was focused
back to the earlier synthetic steps to resolve the trans-isomer
issue.
Attention was turned to the identification of a solid
intermediate derivative that would be amenable to crystal-
lization. Of the various intermediates that might form
crystalline derivatives, it was thought that carboxylic acid 9
from the Wittig reaction would provide the best choice
following its conversion to an amine salt. Unfortunately, of the
nitrogen bases tested43a only the ammonium salt of 9 appeared
to be a solid, and only so at low temperatures (< −20 °C), with
analysis revealing that no enrichment of the cis-isomer had
occurred.
a
Reagents and conditions: (a) Lipase PS (IM or SD), vinyl acetate,
neat, or n-heptane or MTBE, 30−50 °C; (b) HCO2H, Ph3P, DEAD,
THF, 0 °C−r.t.; (c) 0.5 M guanidine/MeOH, MeOH, 0−10 °C; (d)
TBSCl, imidazole, DMF, r.t.
that the resolution was slow, and we therefore examined other
lipases. Immobilized and free-form lipases derived from
Burkholderia cepacia, namely Lipase PS “Amano” SD and
Lipase PS “Amano” IM, provided much more rapid conversion
than when using PPL, however. Use of either neat vinyl acetate
or a mixture of vinyl acetate and MTBE as the solvent was most
preferred. In contrast to that reported16,33a for analogues of
racemic alcohol rac-16 that required 4−9 days for the
enzymatic resolution at r.t. to reach the desired end-point,
alcohol rac-16 was resolved in a matter of several hours
providing a high ee of the (R)-acetate (R)-17.38
The resolution was scaled-up to 4.4 kg of racemic alcohol
rac-16 (>98% GC purity) using neat vinyl acetate at 38−42 °C
and the free enzyme, Lipase PS “Amano” SD.39 The reaction
was terminated at 33 h when 2.7% of (R)-alcohol (R)-16
remained. Once filtration and concentration were complete, a
mixture of (R)-acetate (R)-17 with 96.7% ee and unreacted
(S)-alcohol (S)-16 with 89.6% ee were obtained. As per Spur et
al.’s adapted protocol,33a a Mitsunobu inversion of the (R)-
acetate (R)-17/(S)-alcohol (S)-16 mixture with formic acid was
conducted directly, without prior chromatographic separation,
and the resulting 1:1 mixture of the (R)-acetate and (R)-
formate (18) was purified and then treated with guanidine in
MeOH at 0−10 °C to provide 94% GC pure (R)-alcohol (R)-
16 with 91% ee in 92% yield (based on racemic alcohol rac-16).
An enantiopurity upgrade was then accomplished by enzymatic
reprocessing, but this time the resolution step was followed by
column chromatography to remove the undesired (S)-alcohol,
providing 4.34 kg of 98% GC pure (R)-acetate (R)-17 with
99.9% ee.40 Finally, guanidinolysis supplied 3.38 kg (87% yield)
of 94% GC pure (R)-alcohol (R)-16 in which its (S)-
enantiomer was not detected. Thus, over the double enzymatic
resolution, Mitsunobu inversion and guanidinolysis steps a 72%
yield of high stereochemical purity (R)-alcohol (R)-16 was
obtained calculated on the basis of both enantiomorphs of the
racemic alcohol input.
Various salts of the desilylated analogue, hydroxycarboxylic
acid 19, were examined next with the expectation that removal
of the lipophilic TBS group would impart better physical
properties and provide a robust crystalline solid (Scheme 6). A
screen of nitrogen bases in MTBE, EtOAc, or aqueous MeOH
(for the amino acids) was conducted. Of all bases screened43b
only the benzylamine series provided solids. Apart from the 4-
methoxybenzylamine salt 2,43c which possessed good physical
properties44 and being formed in good yield using 1 equiv of 4-
Scheme 6. Improved process for the synthesis of 14 and (R)-
a
1
Finally, protection41 of about 3.4 kg of (R)-16 with TBSCl in
DMF in the presence of imidazole at 20−25 °C furnished key
intermediate (R)-1 with 99.5% ee, 98% GC purity in 82% yield
following column chromatography (Scheme 5). HPLC analysis
a
Reagents and conditions: (a) (i) TBAF, MTBE, reflux; (ii) 0.5 M
HCl, 80−90%; (b) (i) MeOC6H4CH2NH2, MeCN, r.t., then 50 °C to
recrystallise, 92%; (ii) Optionally recrystallize, 93%; (c) i-PrI, Cs2CO3,
acetone, reflux, quantitative; (d) As per original process.
1909
dx.doi.org/10.1021/op300188x | Org. Process Res. Dev. 2012, 16, 1905−1916

Organic Process Research & Development
Article
methoxybenzylamine in EtOAc, the three others were initially
obtained as sticky solids. Due to the significant rate of
background aminolysis of EtOAc,45 further solvent screening
(MTBE, MIBK, MeCN) was required. MeCN was identified as
the most preferred solvent for recrystallisation of 4-methox-
ybenzylamine salt 2, supplying a high yield (94%) and high
enrichment of the cis-isomer.
With these results in hand, a scalable process to convert
crude, trans-isomer-contaminated Wittig reaction carboxylic
acid 9 into pure ester cis-14 was now in sight. On laboratory-to-
kilogram scales, desilylation of crude Wittig reaction product
946 with >70% potency assay and 80% HPLC purity with
TBAF in MTBE followed by a 0.5 M HCl quench provided
hydroxycarboxylic acid 19 with 62−65% potency assay in 80−
93% assay yield. Recrystallisation of the crude acid as its 4-
methoxybenzylamine salt from hot MeCN with seeding gave
83−92% assay yields of 4-methoxybenzylamine salt 2 with
98.0−98.5% HPLC purity. This salt was contaminated with less
than 1 mol% of the trans-isomer, demonstrating the utility of
the new process. Although this was low enough for our
requirements, if necessary the trans-isomer could be reduced
down to 0.2 mol% by a second recrystallisation of the salt from
MeCN, as was demonstrated on about half-kilogram scales
(92−97% yield).
Direct esterification of recrystallised salt 2, without a salt-
breaking step, was demonstrated on a 0.7-kg scale by treatment
with isopropyl iodide in the presence of Cs2CO347 under reflux
in acetone. Isopropyl ester 14, which was an intermediate in the
original process, was acquired as an oil in an almost quantitative
yield with ≥98% HPLC purity without purification. The
unalkylated amine was recovered from the aqueous phase of the
acid wash step for recycling.
Using the redesigned process described above, one
chromatography step was eliminated from the original process,
and with a single recrystallization of the salt intermediate 2 on a
1.0-kg scale, an overall yield of about 70% based on crude 9 was
achieved with the trans-isomer lowered to 0.7% by HPLC.
Conversion of this through the remaining steps of the synthetic
route provided the cyclopentenone (R)-1 with 99.8% ee and
only 0.62% of the trans-isomer, demonstrating good control of
chiral purity and adequate control of the geometric isomer.
chromatography steps might be eliminated by identification of
new crystalline derivatives of intermediates.
Finally, it is worth commenting on the choice of the
isopropyl ester of cyclopentenone (R)-1 that was integral to our
original strategy depicted in Figure 1. As we have disclosed
elsewhere,7b,48 this ester is not end-product or structure
limiting in that it serves as both a carboxyl protecting group
and as part of the final drug structure for some products (i.e.,
for travoprost7b) or as a protecting group that can be converted
downstream to an amide by aminolysis (i.e., for bimatoprost7b)
or to the parent carboxylic acid (i.e., for lubiprostone48) by
enzymatic hydrolysis. Further, this ester proved to be resistant
to hydrolysis in the Piancatelli rearrangement and trans-
esterification in the enzymatic resolution steps.
EXPERIMENTAL SECTION
■
General. HPLC analyses were conducted using Agilent
1100 and 1200 HPLC systems. GC analyses were conducted
using Agilent 6890 and 7890 GC systems. HPLC and GC
1
purity is reported by area%. H and 13C NMR spectra were
acquired using a Varian, Mercury Plus 300 MHz spectrometer.
EIMS data were acquired using an Agilent 7890 GC system
with a 5975C MSD. ESI data was acquired using an Agilent
1200 HPLC system coupled with a 6120 MSD. HRMS data
was acquired using a Bruker APEX III 7.0 FTMS system. IR
data was acquired using a Nicolet Avatar 360 FT-IR system.
The specific rotation was calculated from data obtained using a
Rudolph Autopol III polarimeter. Melting points were
measured using Shanghai Jingke Scientific Instrument Co.,
Ltd.’s WRR melting point apparatus. The enzymatic resolution
of rac-16 was monitored throughout (both enantiomorphs and
cis- and trans-geometric isomers of both the alcohol ((R)-16,
(S)-16, trans-(R)-16, trans-(S)-16 elute at about 49.0, 52.5,
55.3, 67.0 min, respectively) and the acetate esters ((R)-17,
(S)-17, trans-(R)-17, trans-(S)-17 elute at about 25.0, 29.6,
31.7, 35.0 min, respectively) are separated) using an Agilent
HPLC system (as above) monitoring at 216 nm equipped with
a Chiralcel OJ-H column (250 mm × 4.6 mm, 5 μm)
maintained at 25 °C using a 100:2 n-hexane/isopropanol
mobile phase (isocratic) with a flow rate of 1.2 mL/min. The ee
of (R)-1 ((R)-1 and (S)-1 elute at about 20.4 and 24.8 min,
respectively) was determined using an Agilent HPLC system
(as above) monitoring at 220 nm equipped with a Chiralcel
OD-3 column (150 mm × 4.6 mm, 3 μm) maintained at 20 °C
using a 99:1 n-hexane/isopropanol mobile phase (isocratic)
with a flow rate of 0.2 mL/min. The cis- and trans-geometric
isomer ratio of (R)-1 and trans-(R)-1 was determined using the
same HPLC system ((R)-1 and trans-(R)-1 elute at about 20.4
and 21.6 min, respectively). The cis- and trans-geometric isomer
ratio of 2 (cis-2 and trans-2 elute at 10.1 and 11.6 min,
respectively) was determined using an Agilent HPLC system
(as above) monitoring at 215 nm equipped with a Zorbax SB-
C18 column (150 mm × 4.6 mm, 5 μm) maintained at 30 °C
using a 90:10 10 mM K2HPO4 (pH = 6.0)/MeCN for 0−11
min (isocratic), then 90 to 50:10 to 50 10 mM K2HPO4 (pH =
6.0)/MeCN (linear gradient) from 11 to 15 min, then 50 to
20:50 to 80 10 mM K2HPO4 (pH = 6.0)/MeCN (linear
gradient) for 15−20 min, then 20:80 10 mM K2HPO4 (pH =
6.0)/MeCN (isocratic) for 20−28 min as the mobile phase
with a flow rate of 1.5 mL/min. The cis- and trans-geometric
isomer ratio of 14 (cis-14 and trans-14 elute at 14.6 and 15.3
min, respectively) was determined using an Agilent HPLC
system (as above) monitoring at 215 nm equipped with a
CONCLUSION
■
In summary, a robust and scalable process for the manufacture
of the homochiral key intermediate cyclopentenone (R)-1,
which is useful as an intermediate for the synthesis of
prostaglandin analogues, has been developed. Cyclopentenone
(R)-1 was produced on multikilogram scales with a high
chemical and chiral purity in 14 steps, starting with furfural.
Controlling the level of the undesired trans-isomer formed in
the Wittig reaction that tracked through to the APIs was later
addressed by development of improved Wittig reaction
conditions followed by recrystallisation of an amine salt
derivative of hydroxycarboxylic acid intermediate 19. The
high enantiopurity of cyclopentenone (R)-1 was achieved
through a double enzymatic resolution protocol adapted from
reported procedures, making use of both enantiomers of the
original racemate, in good yield.
Improvements in process efficiency are still required, and
these, with a particular focus on the Piancatelli rearrangement,
will be addressed in future process development. A more
expedient synthesis of aldehyde 4 is sought, and telescoping of
reaction steps would provide much benefit. Further, additional
1910
dx.doi.org/10.1021/op300188x | Org. Process Res. Dev. 2012, 16, 1905−1916

Organic Process Research & Development
Article
Zorbax SB-C18 column (150 mm × 4.6 mm, 5 μm) maintained
at 30 °C using a 80 to 60:20 to 40 10 mM KH2PO4 (pH =
3.0)/MeCN for 0−5 min (linear gradient), then 60:40 10 mM
KH2PO4 (pH = 3.0)/MeCN (isocratic) from 5 to 18 min, then
60 to 20:40 to 80 10 mM KH2PO4 (pH = 3.0)/MeCN (linear
gradient) for 18−20 min, then 20:80 10 mM KH2PO4 (pH =
3.0)/MeCN (isocratic) for 20−28 min as the mobile phase
with a flow rate of 1.5 mL/min.
washed with saturated aqueous NaCl (41.1 kg), and then
concentrated at <55 °C under reduced pressure, providing 20.1
kg (63.1 mol, 87.4%) of crude product 6 as a mixture of two
1
diastereomers* with 90% GC purity. H NMR (300 MHz,
CDCl3): δ −0.15 (s, 1.5H, H9), −0.04 (s, 1.5H, H9), 0.08 (s,
1.5H, H9), 0.10 (s, 1.5H, H9), 0.88 (s, 4.5H, H11), 0.90 (s,
4.5H, H11), 1.83 (ddd, J = 14.4 Hz, 4.5 Hz, 2.7 Hz, 0.5H, H3),
1.91−2.05 (m, 2H, H3 and OH), 2.13 (dt, J = 14.4 Hz, 9.0 Hz,
0.5H, H3), 3.23 (d, J = 2.7 Hz, 0.5H, OH), 3.34 (d, J = 2.4 Hz,
0.5H, OH), 3.48 (m, 1H, H2), 3.62 (m, 1H, H1), 3.92 (m, 1H,
H1), 4.98 (dd, J = 9.0 Hz, 4.5 Hz, 0.5H, H4), 5.09 (t, J = 5.1
Hz, 0.5H, H4), 6.22 (t, J = 3.0 Hz, 1H, H6), 6.32 (m, 1H, H7),
7.35 (m, 1H, H8). 13C NMR (75.45 MHz, CDCl3): δ 156.2,
156.0*, 141.9, 141.8*, 110.4, 110.3*, 106.8, 106.5*, 71.1, 69.3*,
68.5, 67.4*, 67.1, 67.0*, 39.8, 39.1*, 26.0, 18.3, 18.2*, −4.86,
−5.08*, −5.13. EIMS m/z 73 (24%, C3H9Si+ and/or
Synthesis of 4-(tert-Butyldimethylsilyloxy)-4-(furan-2-
yl)-butene (5). To a stirred mixture of zinc (2.475 kg, 38.08
mol) and anhydrous THF (5.874 kg) at 60−65 °C in a 20-L
reactor was added a 4% portion of allyl bromide (4.587 kg, 38
mol) and furfural (3.3 kg, 34.38 mol) in anhydrous THF (2.937
kg). The mixture was stirred at 60−65 °C until the reaction had
initiated (as indicated by reflux); then the remaining 96% of the
mixture was added dropwise without any heating. After the
addition was complete (3 h), the mixture was heated again until
the reaction was complete (GC analysis). The mixture was
cooled to <10 °C, transferred to a 50-L reactor, and a solution
of TBSCl (8.283 kg, 55.2 mol), imidazole (5.61 kg, 82.5 mol),
and THF (5.874 kg) in DMF (6.237 kg) was added over a 2.5 h
period at 10−20 °C. The mixture was then warmed to 20−35
°C and stirred until alcohol 3 was consumed (2 h; GC
analysis). The reaction solution was cooled to 10−15 °C; water
(3.3 kg) and n-heptane (2.244 kg) were added and then
separated, and the aqueous layer was extracted twice with n-
heptane (4.488 kg each). The combined organic layers were
washed twice with saturated aqueous NaCl (8.6 kg each),
washed with saturated aqueous NaHCO3 (3.3 kg), and then
concentrated at 55−60 °C under reduced pressure (<0.1 MPa)
to a volume of 4−5 L. The remaining residue was then
fractionally distilled under reduced pressure (7−8 mmHg) with
the fraction boiling at 95−101 °C being collected, giving 6.4 kg
C3H7O2 ), 75 (100, C2H7SiO+), 95 (23), 117 (77), 171 (18,
+
M.+ − C6H15Si), 211 (7, M.+ − C3H7O2). ESI (Negative) m/z
331 (100%, M[AcO]−), 345 (30, M[AcO]−). IR 3384, 2930,
2954, 2886, 2858, 1472, 1463, 1361, 1255, 1150, 1078, 1009,
837, 779, 736 cm−1.
Synthesis of 3-(tert-Butyldimethylsilyloxy)-3-(furan-2-
yl)propanal (4). To a stirred solution of NaIO4 (20.54 kg,
95.98 mol) in water (61.8 kg) in a 200-L reactor under N2 was
added a solution of 6 (20.1 kg, 79.13 mol) in acetone (57.64
kg) at 10−30 °C over a 2.5 h period. The resulting mixture was
stirred at 10−30 °C for 6 h; then over 24.5 h were added more
NaIO4 in four portions (1.3 kg, 6.08 mol, then 1.8 kg, 8.42 mol,
then 0.9 kg, 4.21 mol, 0.9 kg, 4.21 mol), water (1.5 kg), and
acetone (10 kg). Once complete (GC analysis), the mixture
was filtered to remove the solids and washed with MTBE (14.9
kg). The combined filtrate was separated, and the aqueous layer
was extracted with MTBE (14.9 kg); the combined organic
layer was washed with saturated aqueous NaCl (26.73 kg) and
then dried over anhydrous MgSO4 for 2 h under an atmosphere
of argon. The mixture was filtered through silica gel (6.63 kg),
and the filter cake was washed with MTBE (50 kg). The filtrate
was concentrated at <40 °C under reduced pressure to furnish
4 as a brown oil (17.8 kg; typical GC purity >90%) that was
1
(24.2 mol, 70%) of olefin 5 with 95.5% GC purity. H NMR
(300 MHz, CDCl3): δ −0.06 (s, 3H, H9), 0.05 (s, 3H, H9),
0.87 (s, 9H, H11), 2.56 (t, J = 6.6 Hz, 2H, H3), 4.71 (t, J = 6.6
Hz, 1H, H4), 5.04 (m, 1H, H1), 5.05 (dm, J = 24.0 Hz, 1H,
H1), 5.77 (m, 1H, H2), 6.17 (d, J = 3.3 Hz, 1H, H6), 6.30 (dd,
J = 3.3 Hz, 1.8 Hz, 1H, H7), 7.34 (dd, J = 1.8 Hz, 0.9 Hz, 1H,
H8). 13C NMR (75.45 MHz, CDCl3): δ 157.1, 141.5, 134.8,
117.4, 110.1, 105.1, 68.6, 41.8, 26.0, 18.4, −4.7, −4.8. EIMS m/
z 73 (63%, C3H9Si+), 75 (100, C2H7SiO+), 77 (21), 91 (19),
103 (16), 111 (8), 121 (7), 129 (8), 195 (49, M.+ − C4H9),
211, (32, M.+ − C3H5), 237 (1, M.+ − CH3). IR 2929, 2956,
2887, 2858, 1472, 1463, 1344, 1257, 1089, 1005, 915, 847, 777,
735, 598 cm−1.
Synthesis of 4-(tert-Butyldimethylsilyloxy)-4-(furan-2-
yl)-butane-1,2-diol (6). A mixture of 5 (19.09 kg, 96% purity,
72.2 mol), K2OsO2(OH)4 (0.1 kg, 0.271 mol), and
(DHQ)2PHAL (0.2 kg, 0.257 mol) in acetone (86.5 kg) in a
200-L reactor was cooled to 10−15 °C. A solution of NMO
(11.14 kg, 97% purity, 92.4 mol) in water (36.46 kg) was added
at such a rate that the temperature stayed within a range of 10−
25 °C. The reaction mixture was stirred at r.t. for 6.5 h; then
more K2OsO2(OH)4 (9 g, 0.024 mol), (DHQ)2PHAL (18 g,
0.023 mol), and NMO (0.984 kg, 8.4 mol) were added, and the
mixture was stirred for a further 2 h. A solution of Na2SO3
(15.68 kg, 125.4 mol) in water (47.34 kg) was added, and the
mixture was then heated to 40−43 °C for 1.5 h, filtered at 30−
40 °C, and washed with acetone (27.3 kg). The combined
filtrate was concentrated at <45 °C under reduced pressure.
The concentrated residue was extracted twice with EtOAc
(32.8 kg each), and the combined organic solutions were
1
used directly without purification in the next step. H NMR
(300 MHz, CDCl3): δ −0.06 (s, 3H, H8), 0.08 (s, 3H, H8),
0.85 (s, 9H, H10), 2.75 (ddd, J = 16.2 Hz, 4.8 Hz, 2.1 Hz, 1H,
H2), 2.96 (ddd, J = 16.2 Hz, 7.5 Hz, 2.7 Hz, 1H, H2), 5.24 (dd,
J = 7.5 Hz, 4.8 Hz, 1H, H3), 6.22 (d, J = 3.3 Hz, 1H, H5), 6.32
(dd, J = 3.3 Hz, 1.8 Hz, 1H, H6), 7.37 (dd, J = 1.8 Hz, 0.9 Hz,
1H, H7), 9.82 (dd, J = 2.7 Hz, 2.1 Hz, 1H, H1).13C NMR
(75.45 MHz, CDCl3): δ 201.0, 155.5, 142.2, 110.4, 106.8, 64.3,
50.2, 25.9, 18.3, −4.8, −5.0. IR 2956, 2930, 2888, 2858, 1728,
1472, 1464, 1362, 1343, 1256, 1149, 1097, 1007, 848, 779, 738
cm−1.
Synthesis of Isopropyl (Z)-8-(tert-Butyldimethylsily-
loxy)-8-(furan-2-yl)-oct-5-enoate (8) Using TPPA in
Wittig Reaction. A mixture of 5-bromopentanoic acid (11a;
3.50 kg, 19.33 mol), triphenylphosphine (7.60 kg, 28.98 mol),
and toluene (10.5 L) was heated under reflux for 6 h. The
resulting slurry was cooled to 25−30 °C, and the solids were
isolated by centrifuge and were washed with toluene (1.7 L)
and dried at 90−100 °C for 12 h providing (4-carboxybutyl)-
triphenylphosphonium bromide (11b; 7.42 kg, 16.74 mol). To
a solution of (4-carboxybutyl)triphenylphosphonium bromide
(0.90 kg, 2.03 mol) in a mixture of THF (3.36 kg) and TPPA
(0.96 kg) in a 20 L glass-lined reactor under an atmosphere of
nitrogen at 6−8 °C was added a solution of NaHMDS (2 M
1911
dx.doi.org/10.1021/op300188x | Org. Process Res. Dev. 2012, 16, 1905−1916

Organic Process Research & Development
Article
solution in THF, 1.85 kg, 4.01 mol) over a 0.92-h period to
maintain a temperature within 0−10 °C, furnishing a dark-
orange mixture that was further stirred for 0.5 h. The solution
was cooled to between −32 and −38 °C. A solution of
aldehyde 4 (0.405 kg, 1.43 mol based on GC purity of 90%) in
THF (1.80 kg) was then added over a 1-h period to maintain a
temperature of between −32 and −38 °C and was further
stirred at this temperature until the reaction was complete (GC
analysis; about 1 h). Acetone (0.115 kg) was added and stirred
for 30 min. EtOAc (3.79 kg) and saturated aqueous NH4Cl
(6.79 kg; adjusted to pH 1 with conc. HCl) were added at <5
°C, and the mixture was warmed to between ∼20 °C. Water
(2.43 kg) and solid NaCl (0.324 kg) were added, and after the
mixture stirred for 20 min, the aqueous phase was separated,
and water was added and stirred for 20 min. The aqueous layer
was separated and extracted with EtOAc (1.05 kg), and the
combined organic layer was concentrated at <55 °C under
reduced pressure (0.08−0.1 MPa) to provide crude 9 ((Z)-8-
(tert-butyldimethylsilyloxy)-8-(furan-2-yl)-oct-5-enoic acid; 1.9
1471, 1374, 1361, 1348, 1252, 1151, 1110, 1006, 968, 940, 836,
778, 735 cm−1.
Synthesis of Isopropyl (Z)-7-(3-Hydroxy-5-oxo-cyclo-
pent-1-en-1-yl)-hept-5-enoate (rac-16). A mixture of 8
(15.83 kg, 41.66 mol) and TBAF·3H2O (13.12 kg, 42.05 mol)
in THF (70.44 kg) in a 200-L reactor was stirred at 35−45 °C
until the starting material 8 was consumed (GC analysis).
Saturated aqueous NH4Cl (23.11 kg) and solid NaCl (0.158
kg) were added, and the mixture was vigorously stirred for 30
min. The aqueous phase was separated, extracted with EtOAc
(110.5 kg), and then combined with the organic phase and
concentrated at <55 °C under reduced pressure. The residue
was dissolved in EtOAc (110.5 kg) and was washed twice with
water (45.91 kg each); following evaporation under reduced
pressure isopropyl (Z)-8-(furan-2-yl)-8-hydroxy-oct-5-enoate
was furnished (14; analytical data for this compound can be
found below for its synthesis from 2) as a brown oil. To a
mixture of ZnCl2 (64.75 kg, 476 mol) and water (72.8 kg) in a
500-L reactor was added a solution of the crude 14 and
hydroquinone (4.6 g) in dioxane (86 kg). The mixture was
heated at reflux under an atmosphere of nitrogen until 14 was
consumed (6 h; GC analysis). The product mixture was cooled
to 50−60 °C, and the dioxane was evaporated under reduced
pressure until 72 kg of solvent had distilled. EtOAc (75.5 kg)
and saturated aqueous NH4Cl (44.16 kg) were added into the
vigorously stirred mixture. The aqueous phase was separated
and extracted with EtOAc (35.62 kg), and the combined
organic layer was washed with saturated aqueous NaCl (44.17
kg) and concentrated at <55 °C under reduced pressure (0.1
MPa) to give a black oil (a mixture of rac-16 and rac-16b). The
oil was dissolved in toluene (81.2 kg) in a 200-L reactor and
stirred in the presence of Et3N (5.52 kg, 54.65 mol) and chloral
(1.266 kg, 5.83 mol) at r.t. for 12 h. Extra portions of Et3N
(2.77 kg, 27.4 mol) and chloral (0.427 kg, 3.1 mol) were added,
and the mixture was stirred at r.t. for a further 3 h. The product
mixture was washed with a saturated aqueous solution of
NH4Cl (44.16 kg), and the aqueous phase was back-extracted
with PhMe (28.96 kg). The organic portions were combined
and washed with saturated aqueous NaCl (23.1 kg) and
concentrated under reduced pressure at <55 °C to give a brown
oil that was purified by chromatography (eluting with 1:2
EtOAc/n-heptane followed by an almost equal volume of 2:3
EtOAc/n-heptane) to furnish the title product rac-16 (4.43 kg,
16.6 mol) with 100% GC purity plus another fraction
consisting of 97% GC pure rac-16 (452 g, 1.6 mol); total
44% over three steps based on 8. 1H NMR (300 MHz, CDCl3):
δ 1.22 (d, J = 6.3 Hz, 6H, H2′), 1.68 (m, 2H, H3), 2.06 (m, 2H,
H4), 2.27 (t, J = 7.5 Hz, 2H, H2), 2.34 (dd, J = 18.6 Hz, 1.8 Hz,
1H, H10), 2.85 (dd, J = 18.6 Hz, 6.3 Hz, 1H, H10), 2.91 (m,
2H, H7), 4.98 (m, 2H, H11 and H1′), 5.53 (m, 2H, H5 and
H6), 7.16 (dd, J = 1.8 Hz, 0.9 Hz, 1H, H12). 13C NMR (75.45
MHz, CDCl3): δ 206.0, 173.5, 156.3, 146.9, 132.0, 125.4, 68.0,
68.0, 45.2, 34.2, 26.7, 24.8, 22.9, 22.0. EIMS m/z 41 (59%), 43
1
kg). H NMR (300 MHz, CDCl3): δ −0.05 (s, 3H, H1′), 0.05
(s, 3H, H1′), 0.87 (s, 9H, H2′), 1.66 (p, J = 7.5 Hz, 2H, H3),
2.07 (q, J = 6.9 Hz, 2H, H4), 2.32 (t, J = 7.5 Hz, 2H, H2), 2.54
(t, J = 6.3 Hz, 2H, H7), 4.69 (t, J = 6.3 Hz, 1H, H8), 5.41 (m,
2H, H5 and H6), 6.16 (d, J = 3.0 Hz, 1H, H10), 6.28 (dd, J =
3.0 Hz, 1.8 Hz, 1H, H11), 7.33 (dd, J = 1.8 Hz, 0.9 Hz, 1H,
H12). 13C NMR (75.45 MHz, CDCl3): δ 180.2, 157.1, 141.5,
130.7, 126.5, 110.2, 106.1, 68.7, 35.2, 33.7, 32.4, 26.8, 26.0,
24.7, 22.9, 18.4, −4.7, −4.8. EIMS m/z 73 (70%, C3H9Si+), 75
(76, C2H7SiO+), 81 (27), 211 (100, M.+ − C7H11O2), 212 (18),
263 (8, M.+ − C3H9Si), 281 (4, M.+ − C4H9). ESI (Negative)
m/z 337 ([M − H]−). IR 3016, 2955, 2930, 2857, 1710, 1472,
1463, 1412, 1255, 1151, 1087, 1006, 939, 837, 777, 735 cm−1.
In a 20-L glass-lined reactor, an acetone (3.0 kg) solution of
the crude product and K2CO3 (0.64 kg, 4.63 mol) and 2-
iodopropane (0.785 kg, 4.62 mol) was heated under reflux for 4
h. After the mixture was cooled to <30 °C, water (2.55 kg) and
MTBE (1.89 kg) were added, and the mixture was stirred for 20
min. Solid NaCl (0.219 kg, 0.5 P) was added, and the mixture
was stirred for 20 min; the aqueous layer was separated and
extracted with MTBE (0.6 kg), and the MTBE portions were
combined, washed once with saturated aqueous NaCl (1.07
kg), and concentrated under reduced pressure (0.09−0.1 MPa)
at <55 °C; the resulting oil was purified by column
chromatography (eluting with a 1:20 mixture of EtOAc and
n-heptane), and the fractions containing the title product were
combined and concentrated under reduced pressure to provide
8 (0.412 kg, 1.00 mol, 70% based on aldehyde 4) with 92.4%
GC purity (contaminated with 5.6% 12). A small sample was
converted to ester 14 and analysed by HPLC, showing a
8.0:92.0 trans-/cis- ratio. 1H NMR (300 MHz, CDCl3): δ −0.06
(s, 3H, H13), 0.05 (s, 3H, H13), 0.87 (s, 9H, H9), 1.22 (d, J =
6.3 Hz, 6H, H2′), 1.65 (m, 2H, H3), 2.04 (m, 2H, H4), 2.24 (t,
J = 7.5 Hz, 2H, H2), 2.54 (t, J = 6.0 Hz, 2H, H7), 4.68 (t, J =
6.6 Hz, 1H, H8), 5.00 (septet, J = 6.3 Hz, 1H, H1′), 5.40 (m,
2H, H5 and H6), 6.16 (d, J = 3.3 Hz, 1H, H10), 6.29 (dd, J =
3.3 Hz, 1.8 Hz, 1H, H11), 7.3 (dd, J = 1.8 Hz, 0.9 Hz, 1H,
H12). 13C NMR (75.45 MHz, CDCl3): δ 173.4, 157.1, 141.5,
131.0, 126.3, 110.2, 106.1, 68.7, 67.6, 35.2, 34.3, 32.1, 26.9,
26.0, 25.1, 22.9, 22.1, 18.4, −4.7, −4.8. EIMS m/z 73 (53%,
+
(100, C3H7 ), 55 (43), 67 (31), 77 (33), 79 (40), 91 (44), 94
(56), 107 (30), 119 (51, C8H7O+), 133 (54, C9H9O+), 146 (49,
C10H10O+), 160 (49, C11H12O+), 188 (37, M.+ − H2O −
C3H7O − H), 189 (46, M.+ − H2O − C3H7O), 206 (62, M.+ −
C3H7O − H), 207 (26, M.+ − C3H7O), 248 (11, M.+ − H2O).
ESI (Positive) m/z 267 (100%, MH+), 284 (57, MNH4 ). IR
+
C3H9Si+), 75 (27, C2H7SiO+), 189 (11), 211 (100, M.+
−
3446, 2980, 2936, 1713, 1456, 1419, 1375, 1316, 1260, 1199,
C10H17O2), 263 (6), 281 (2), 305 (1), 323 (3), 380 (0.2, M.+).
1147, 1109, 1037 cm−1.
+
ESI (Positive) m/z 403 (41%, MNa+), 398 (80, MNH4 ), 249
Synthesis of Isopropyl (3R,Z)-7-(3-Hydroxy-5-oxo-
cyclopent-1-en-1-yl)-hept-5-enoate ((R)-16). A mixture of
(100, MH+ − C6H15SiO). IR 2956, 2930, 2895, 2858, 1732,
1912
dx.doi.org/10.1021/op300188x | Org. Process Res. Dev. 2012, 16, 1905−1916

Organic Process Research & Development
Article
+
rac-16 (4.43 kg, 16.64 mol), vinyl acetate (33 kg), and Lipase
PS “Amano” SD (2.22 kg) was stirred in a 50-L reactor at 38−
42 °C until chiral HPLC analysis showed that (R)-16 was not
more than 3.0% (the reaction was terminated at 33 h when (R)-
16 was 2.7%). The reaction mixture was filtered through a plug
of Celite; the filter cake was washed three times with EtOAc (4
kg each) and then concentrated to give a yellow-brown oil
comprising a mixture of isopropyl (3R,Z)-7-(3-acetoxy-5-oxo-
cyclopent-1-en-1-yl)-hept-5-enoate ((R)-17; 96.7% ee) and
isopropyl (3S,Z)-7-(3-hydroxy-5-oxo-cyclopent-1-en-1-yl)-
hept-5-enoate ((S)-16; 89.6% ee). The mixture was dissolved
in THF (22.26 kg) in a 50-L reactor and cooled to 0−10 °C,
and triphenylphosphine (3.615 kg, 13.8 mol) and formic acid
(0.655 kg, 13.8 mol) were added, followed by addition of a
solution of diethyl azodicarboxylate (2.38 kg, 13.8 mol) in THF
(4.53 kg) over 5.5 h, maintaining a temperature of 0−10 °C.
The reaction mixture was allowed to warm to 20−25 °C and
stirred until (S)-16 was fully consumed (GC analysis; 1 h). The
product mixture was partially concentrated under reduced
pressure, and EtOAc (5.67 kg) was added to the concentrate
followed by n-heptane (12.85 kg) causing precipitation. The
mixture was filtered, and the filter cake was washed three times
with 1:5 EtOAc/n-heptane (3.1 kg each), concentrated under
reduced pressure, and purified by column chromatography
(eluting with 1:3 EtOAc/heptane) providing a mixture of (R)-
17 and isopropyl (3R,Z)-7-(3-formyloxy-5-oxo-cyclopent-1-en-
1-yl)-hept-5-enoate ((R)-18) with a combined GC purity of
96%. To a cold (−10 to 0 °C) MeOH (19.86 kg) solution of
the mixture in a 50-L reactor was added a solution of guanidine
in MeOH (0.5 M, 14.41 kg, 9.0 mol), and the mixture was
stirred at 0−10 °C until the esters were consumed (TLC
analysis; <30 min). AcOH (0.545 kg, 9.08 mol) was added, the
mixture was stirred for 5 min, and the mixture was then allowed
to warm to r.t. The product mixture was concentrated under
reduced pressure at <55 °C. The residue was diluted with
EtOAc (21.7 kg) and was washed with water (48.2 kg). The
aqueous layer was separated and was extracted with EtOAc
(21.7 kg); the combined organic layers were washed with water
(24.1 kg) and with saturated aqueous NaCl (24.1 kg) and then
were dried over MgSO4 for 2 h and filtered, and the filter cake
was washed three times with EtOAc (5 kg each) and
concentrated under reduced pressure to give (R)-16 (4.48 kg,
15.8 mol, 92% GC yield based on rac-16) with 94% GC purity
and 91% ee. A mixture of the above prepared enantioenriched
(R)-16, vinyl acetate (33.4 kg), and Lipase PS “Amano” SD
(2.24 kg) in a 50-L reactor was stirred at 38−42 °C until chiral
HPLC analysis showed that (R)-16 was not more than 2.0%
(observed: 1.7% at 34 h). The reaction mixture was filtered
through a plug of Celite; the filter cake was washed three times
with EtOAc (4.1 kg each) and then concentrated to give a
yellow-brown oil (5.0 kg) that was purified by column
chromatography (eluting with 1:3 EtOAc/heptane, followed
by EtOAc), providing the intermediate (R)-17 (4.34 kg, 13.8
mol, 83%) with 98.3% GC purity and 99.9% ee. 1H NMR (300
MHz, CDCl3): δ 1.23 (d, J = 6.3 Hz, 6H, H2′), 1.69 (p, J = 7.5
Hz, H3), 2.09 (s, 3H, H14), 2.11 (m, 2H, H4), 2.27 (t, J = 7.5
Hz, 2H, H12), 2.38 (dd, J = 18.9 Hz, 2.1 Hz, 1H, H10), 2.88
(dd, J = 18.9 Hz, 6.3 Hz, 1H, H10), 2.95 (d, J = 6.3 Hz, 2H,
H7), 5.00 (septet, J = 6.3 Hz, 1H, H1′), 5.51 (m, 2H, H5 and
H6), 5.75 (ddd, J = 6.3 Hz, 4.2 Hz, 2.1 Hz, 1H, H11), 7.15 (dd,
J = 4.2 Hz, 1.5 Hz,, 1H, H12). 13C NMR (75.45 MHz, CDCl3):
δ 204.5, 173.3, 170.8, 152.2, 148.8, 131.9, 125.1, 70.6, 67.7,
41.8, 34.3, 26.8, 24.9, 23.0, 22.1, 21.2. EIMS m/z 41 (27%), 43
(100, C3H7 ), 55 (29), 91 (33), 94 (50), 105 (18), 119 (53,
C8H7O+), 133 (29, C9H9O+), 146 (55, C10H10O+), 160 (32,
C11H12O+), 188 (35, M.+ − AcOH − C3H7O − H), 189 (53,
M.+ − AcOH − C3H7O), 206 (57, M.+ − C5H10O2), 248 (16,
M.+ − AcOH). ESI (Positive) m/z 309 (85%, MH+), 326 (100,
+
MNH4 ).
Some unreacted (R)-16 (0.46 kg, 34% ee, 65.9% GC purity)
was isolated for recycling.
To a cold (−10 to 0 °C) MeOH (16.85 kg) solution of the
enantiopure ester (R)-17 in a 50-L reactor was added a solution
of guanidine in MeOH (0.5 M, 11.09 kg, 6.9 mol), and the
mixture was stirred at 0−10 °C until the ester was consumed
(<30 min; TLC analysis). AcOH (0.426 kg, 7.0 mol) was
added, and the mixture was stirred for 5 min and was then
allowed to warm to r.t. The product mixture was concentrated
under reduced pressure at <55 °C, and the residue was diluted
with EtOAc (19.2 kg) and was washed with water (42.66 kg).
The aqueous layer was separated and extracted with EtOAc
(19.2 kg), and the combined organic layers were washed with
water (21.33 kg) and with saturated aqueous NaCl (21.33 kg)
and then were dried over MgSO4 for 2 h and then filtered. The
filter cake was washed three times with EtOAc (3.8 kg each)
and concentrated under reduced pressure to give (R)-16 (3.38
kg, 11.9 mol, 72% total yield based on rac-16) with 94% GC
purity and for which none of the enantiomer was detected by
chiral HPLC analysis. The NMR spectroscopic and mass
spectrometric data of this compound were the same as those of
the racemic precursor rac-16, which are reported above.
Synthesis of Isopropyl (3R,Z)-7-(3-(tert-Butyldimethyl-
silyloxy)-5-oxo-cyclopent-1-en-1-yl)-hept-5-enoate ((R)-
1). To a chilled (0−10 °C) solution of (R)-16 (3.38 kg, 11.9
mol) and imidazole (1.6 kg, 23.5 mol) in dry DMF (8.97 kg) in
a 30-L reactor was added a solution of TBSCl (2.7 kg, 18 mol)
in dry DMF (11.96 kg). The mixture was then stirred at 20−25
°C until the reaction was complete (TLC analysis; 2 h). The
product mixture was diluted with water (22.23 kg) and
extracted twice with MTBE (17 kg each). The combined
organic portions were washed with water (22.23 kg) and twice
with saturated aqueous NaCl (22.23 kg), concentrated under
reduced pressure, and then chromatographed (eluting first with
n-heptane (213 kg) and then 1:5 EtOAc/n-heptane (533 kg))
to furnish the title product (R)-1 (3.752 kg, 9.7 mol, 82% yield
based on (R)-16) with 99.5% ee and 98% GC purity, as a
90.8:9.2 ratio of cis-/trans-isomers as determined by HPLC
analysis. Another batch of (R)-1 that contained a 99.38:0.62
ratio of cis-/trans-isomers, as determined by HPLC analysis, was
prepared using purified salt 2. 1H NMR (300 MHz, CDCl3): δ
0.12 (s, 3H, H13), 0.13 (s, 3H, H13), 0.91 (s, 9H, H15), 1.23
(d, J = 6.3 Hz, 6H, H2′), 1.69 (p, J = 7.5 Hz, 2H, H3), 2.09 (q,
J = 6.9 Hz, 2H, H4), 2.27 (t, J = 7.5 Hz, 2H, H2), 2.29 (dd, J =
18.3 Hz, 2.1 Hz, 1H, H10), 2.77 (dd, J = 18.3 Hz, 6.0 Hz, 1H,
H10), 2.91 (m, 2H, H7), 4.89 (m, 1H, H11), 5.00 (septet, J =
6.3 Hz, 1H, H1′), 5.51 (m, 2H, H5 and H6), 7.04 (m, 1H,
H12). 13C NMR (75.45 MHz, CDCl3): δ 206.0, 173.3, 157.1,
146.0, 131.5, 125.6, 69.2, 67.7, 45.7, 34.3, 26.8, 26.0, 25.0, 22.8,
+
22.1, 18.4, −4.5. EIMS m/z 41 (24%), 43 (100, C3H7 ), 73 (59,
C3H9Si+), 75 (27, C2H7SiO+), 91 (28), 107 (20), 119 (31,
C8H7O+), 129 (21), 133 (13, C9H9O+), 143 (17), 161 (23,
C11H13O+), 171 (16), 189 (90, M.+- C6H16SiO − C3H7O), 206
(13), 221 (6), 248 (7, M+.-C6H16SiO), 263 (83), 281 (4), 321
(18, M.+ − C3H7O), 323 (6, M.+ − C4H9). ESI (Positive) m/z
381 (9%, MH+), 403 (100, MNa+), 419 (12, MK+). HRMS
(ESI, positive) found 381.2432 (calc. 380.2383 for
1913
dx.doi.org/10.1021/op300188x | Org. Process Res. Dev. 2012, 16, 1905−1916

Organic Process Research & Development
Article
C21H36O4Si), found 403.2260 (calc. 403.2281 for
C21H36NaO4Si), found 419.2040 (calc. 419.2020 for
C21H36KO4Si); [α]20D = +10.4 (c = 0.01, CHCl3). IR 2955,
2931, 2886, 2858, 1716, 1471, 1374, 1352, 1257, 1162, 1109,
1084, 967, 900, 836, 778, 669 cm−1.
off-white solid. If required, the salt could be recrystallised in
>92% yield from MeCN providing 0.22% by HPLC of the
trans-isomer. Mp 90.6−91.1 °C. 1H NMR (300 MHz, CDCl3):
δ 1.55 (m, 2H), 2.91−2.18 (m, 4H), 2.55 (m, 2H), 3.76 (s,
3H), 3.81 (s, 2H), 4.65 (t, J = 6.6 Hz, 1H), 5.27−5.52 (m, 2H),
5.82 (bs, 4H), 6.20 (d, J = 3.3 Hz, 1H), 6.29 (dd, J = 3.3 Hz, 1.8
Hz, 1H), 6.84 (d, J = 8.7 Hz, 2H), 67.24 (d, J = 8.7 Hz, 2H),
7.32 (dd, J = 1.8 Hz, 0.6 Hz, 1H). 13C NMR (75.45 MHz,
CDCl3): δ 180.8, 159.6, 156.9, 142.0, 132.6, 130.0, 125.7, 114.3,
110.4, 106.1, 67.4, 55.5, 43.6, 36.1, 34.0, 26.9, 25.8. LC−MS
analysis of 2 provides ESI mass spectrometry data the same as
for 19 reported above. IR 3357, 3009, 2956, 2915, 2839, 2644,
1616, 1586, 1519, 1460, 1422, 1406, 1302, 1253, 1225, 1180,
1028, 1010, 819, 749 cm−1.
Synthesis of (Z)-8-(furan-2-yl)-8-hydroxy-oct-5-enoic
acid (19). To a stirred solution of a partial concentrate of crude
9 (7.210 kg, 3.13 mol, 14.7% HPLC assay) and EtOAc (6.47
kg) in a 50 L reactor at r.t. was added n-heptane (19.6 kg). The
mixture was stirred at r.t. for 1 h, then separated and the upper
layer was extracted twice with 1:4 EtOAc/n-heptane (5.21 kg).
The combined organic layers was concentrated under reduced
pressure (<0.1 MPa) at 50 °C providing a thick, red-coloured
oil that was diluted with n-heptane (19.6 kg) causing
precipitation (Ph3PO). After stirring for 0.5 h, the mixture
was filtered through Celite (1.44 kg), and the filter cake was
washed twice with n-heptane (3.92 kg each). The combined
organic phases were washed three times with water (14.4 kg
each) and then concentrated at 50−55 °C under reduced
pressure (<0.1 MPa) to furnish semipurified 9 as a brown oil
(1.262 kg, 77.1% HPLC assay, yield 91.8%). To a MTBE (3.60
kg) solution in the oil in a 20-L reactor was added TBAF·3H2O
(1.63 kg, 5.15 mol), and the mixture was heated under reflux
for 7 h. After cooling to 30−35 °C, MTBE (4.32 kg) was
added, and the mixture was quenched with 0.5 M HCl (11.5
kg). The aqueous phase was extracted twice with MTBE (3.96
kg each). The combined organic phases were heated to 40 °C,
washed three times with water (5.35 kg each) at 40 °C, dried
over MgSO4 (0.66 kg) for 2 h, and filtered, and the filter cake
was washed three times with MTBE (1.0 kg each). The
solution was concentrated at 40−45 °C under reduced pressure
(<0.1 MPa) providing crude 19 (0.836 kg, 68.6% HPLC purity,
61.5% HPLC assay, 80% based on semipurified 9) as an oil. 1H
NMR (300 MHz, CDCl3): δ 1.69 (p, J = 7.5 Hz, 2H), 2.12 (q, J
= 7.2 Hz, 2H), 2.34 (t, J = 7.5 Hz, 2H), 2.61 (t, J = 6.6 Hz, 2H),
4.71 (t, J = 6.6 Hz, 1H), 5.46 (m, 2H), 6.24 (dd, J = 3.3 Hz, 0.6
Hz, 1H), 6.32 (dd, J = 3.3 Hz, 1.8 Hz, 1H), 7.37 (dd, J = 1.9
Hz, 0.6 Hz, 1H). 13C NMR (75.45 MHz, CDCl3): δ 179.6,
156.4, 142.2, 132.1, 125.6, 110.4, 106.4, 67.7, 33.8, 33.5, 26.8,
24.6. ESI (Negative) m/z 223 (100%, [M − H]−), 224 (12, [M
+ 1 − H]−), 447 (9, [2M − H]−). ESI (Positive) m/z 207
(100%, MH+ − H2O), 247 (20, MNa+), 413 (48, 2[M −
H2O]H+), 414 (15, 2[MH+ − H2O]).
Synthesis and Recrystallization of (4-Methoxyphen-
yl)-methanaminium (Z)-8-(Furan-2-yl)-8-hydroxy-oct-5-
enoate (2). To a MeCN solution (2.9 kg) of crude 19
(0.836 kg, 61.5% HPLC assay) under a nitrogen atmosphere in
a 10-L reactor was added a MeCN (1.24 kg) solution of 4-
methoxybenzylamine (0.32 kg, 2.33 mol) over a period of 20
min at 10−30 °C. An off-white solid precipitated from the
solution, and the mixture was stirred for 2 h at 16−24 °C. The
temperature was raised to 49 °C causing the solids to dissolve.
The temperature was slowly cooled to 45 °C, and seed crystals
of 2 (2.66 g, 7.4 mmol, HPLC ratio of 19/trans-19 was
99.3:0.03) were added to the solution at 45 °C and stirred
slowly for 2 h at 45 °C for crystallization. The mixture was
slowly cooled to 0−5 °C (at a rate of 10 °C/h) and stirred for 1
h at 0−5 °C. The cold mixture was filtered, and the filter cake
was washed three times with chilled (0−5 °C) MeCN (0.83 kg
each). The solids were dried at 50 °C under reduced pressure
(<0.1 MPa) for 4 h to furnish 2 (0.778 kg, 2.15 mol, 92.3%
yield based on HPLC assay) with 98.1% HPLC purity of the
cis-isomer and a 99.3%/0.7% HPLC ratio of 19/trans-19 of an
Synthesis of Isopropyl (Z)-8-(Furan-2-yl)-8-hydroxy-
oct-5-enoate (14). Isopropyl iodide (1.026 kg, 6.04 mol) and
Cs2CO3 (1.31 kg, 4.02 mol) were added to an acetone (5.75
kg) solution of 2 (0.728 kg, 2.01 mol, containing 0.74% of the
trans-isomer by HPLC) at 20−30 °C in a 20-L reactor. The
mixture was stirred under reflux for 6 h. The product mixture
was cooled to ≤35 °C, and MTBE (5.39 kg) and H2O (7.28
kg) were added; the mixture was stirred for 30 min. The
aqueous layer was separated and extracted with MTBE (2.70
kg). The combined organic layers were washed with 0.5 M HCl
(4.15 kg) and saturated aq NaCl (4.95 kg). The organic phase
was concentrated at 40 °C under reduced pressure (<0.1 MPa)
to furnish 14 (0.546 kg, quantitative yield based on 2) with
98.3% HPLC purity with 0.72% of the trans-isomer (this
material was used to provide (R)-1 with 0.62% trans-isomer) by
1
HPLC as an orange oil. H NMR (300 MHz, CDCl3): δ 1.23
(d, J = 6.3 Hz, 6H, H2′), 1.68 (p, J = 7.5 Hz, 2H, H3′), 2.10 (q,
J = 7.3 Hz, 2H, H4), 2.27 (t, J = 7.5 Hz, 2H, H2), 2.62 (t, J =
6.9 Hz, 2H, H7), 4.72 (t, J = 6.6 Hz, 1H, H8), 5.00 (septet, J =
6.3 Hz, 1H, H1′), 5.48 (m, 2H, H5 and H6), 6.25 (d, J = 3.3
Hz, 1H, H10), 6.33 (dd, J = 3.3 Hz, 1.8 Hz, 1H, H11), 7.38
(dd, J = 1.8 Hz, 0.6 Hz, 1H, H12). 13C NMR (75.45 MHz,
CDCl3): δ 173.5, 156.4, 141.1, 132.5, 125.5, 110.4, 106.3, 67.8,
67.6, 34.3, 33.9, 26.9, 25.9, 25.0, 22.1. ESI (Positive) m/z 208
(10%), 249 (100, MH+ − H2O), 266 (<5, MH+), 305.2 (<5,
MK+). EIMS m/z 41 (23%), 43 (14, C3H7 ), 68 (20), 82 (13),
+
+
+
97 (100, C5H5O2 ), 110 (34, C6H6O2 ), 128 (23), 133 (5,
C9H9O+), 145 (6), 146 (5), 170 (20), 189 (5, M.+ − C3H7O −
H2O), 207 (5, M.+ − C3H7O), 248 (4, M.+ − H2O), 266 (1,
M.+). IR 3446, 2980, 2936, 1729, 1375, 1224, 1181, 1146, 1109,
1010, 739, 597.
ASSOCIATED CONTENT
■
S
* Supporting Information
1H and 13C NMR spectra of 5, 6, 4, 8, rac-16, (R)-17, (R)-1,
19, 2, 14; HPLC chromatograms of (R)-17, (R)-16, (R)-1, 2; a
GC chromatogram of (R)-1 and a high-resolution mass
spectrum of (R)-1. Additional details on the synthesis of
aldehyde 3 from furfural, a low-temperature Wittig reaction,
and research of the first resolution step of rac-16. This material
is available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: julian.henschke@scinopharm.com.tw
Notes
The authors declare no competing financial interest.
1914
dx.doi.org/10.1021/op300188x | Org. Process Res. Dev. 2012, 16, 1905−1916

Organic Process Research & Development
Article
(15) For further details of the synthesis of aldehyde 3 see the
Supporting Information.
(16) (a) Babiak, K. A.; Ng, J. S.; Dygos, J. H.; Weyker, C. L.; Wang,
Y.-F.; Wong, C.-H. J. Org. Chem. 1990, 55, 3377. (b) Wong, C.-H.;
Wang, Y.-F.; Hennen, W. J.; Babiak, K. A.; Dygos, J. H.; Ng, J. S. U.S.
Patent 5,106,750, 1992.
ACKNOWLEDGMENTS
■
We thank Dr. Tsung-Yu Hsiao, Dr. Chi-Nung Hsiao, and Dr.
Penelope Henschke for proofing of the manuscript; Yafeng
Chao, Huadong Mei, Lei Yang, Shufeng Wang, Zhimi Hu for
routine IR, HPLC, and GC analysis; Yuxia Shao and Yu-Hui
Chen for acquiring/processing NMR spectra; Xuan Zhou for
collating analytical data; and Professor Lijun Xia for assistance
with chiral purity analysis and liaising with the Shanghai
Institute of Organic Chemistry for high-resolution mass
spectrometry.
(17) Facile, autocatalytic decomposition (contact with only 0.2 equiv
of ylide caused full decomposition) of the 4-silyloxy substituent via
deprotonation α- to the aldehyde was proposed to be responsible for
its base sensitivity. By contrast, when the 4-silyloxy substituent is
absent, successful Wittig reaction with the identical ylide is possible
(see Armstead, D. A.; Mann, J. Synth. Commun. 1985, 15, 1147).
(18) (a) See Ahrgren, L.; Sutin, L. Org. Process Res. Dev. 1997, 1, 425
for a description of a Sharpless asymmetric dihydroxylation on an
industrial scale. (b) To minimise risk to staff associated with the toxic,
reactive, and volatile nature of OsO4: (i) the non-volatile Os(VI) salt,
K2OsO2(OH)4 was used instead of OsO4, and (ii) this salt was used at
a level of 0.4 mol%. The Os(VI) precatalyst is oxidised in situ of the
enclosed reactor to OsO4 during the reaction. The reaction was
conducted at ambient temperature to minimise volatilisation. (c)
Upon work-up, the OsO4 was reduced to a more water-soluble form
using aqueous Na2SO3, and waste streams were handled as per local
environmental regulations. (d) Although the chiral centre generated at
C6 was destroyed in the next step, (DHQ)2PHAL provided a
REFERENCES
■
(1) Collins, P. W.; Djuric, S. W. Chem. Rev. 1993, 93, 1533.
(2) (a) Stjernschantz, J.; Resul, B. Drugs Future 1992, 17, 691.
(b) Resul, B.; Stjernschantz, J.; No, K.; Liljebris, C.; Selen, G.; Astin,
M.; Karlsson, M.; Bito, L. Z. J. Med. Chem. 1993, 36, 243.
(3) (a) Sorbera, L. A.; Leeson, P. A.; Rabasseda, X.; Castaner, J. Drugs
Future 2001, 26, 433. (b) Yolin, S.; Walt, J. G.; Earl, M. Dermatol. Surg.
2010, 36, 638. (c) Cohen, J. L. Dermatol. Surg. 2010, 36, 1361.
(4) Sorbera, L. A.; Castaner, J. Drugs Future 2000, 25, 41.
(5) Wang, Y.; Bolos, J.; Serradell, N. Drugs Future 2006, 31, 788.
(6) Sobera, L. A.; Castaner, J.; Mealy, N. E. Drugs Future 2004, 29,
336.
significant and useful rate acceleration; see Jacobsen, E. N.; Marko,
́
I.;
Mungall, W. S.; Schroder, G.; Sharpless, K. B. J. Am, Chem. Soc. 1988,
̈
́ ̌ ́ ́ ̌
(7) (a) Obadalova, I.; Pilarcík, T.; Slavíkova, M.; Hajícek, J. Chirality
2005, 17, S109. (b) Henschke, J. P.; Liu, Y. L.; Chen, Y. F.; Meng, D.
C.; Sun, T. U.S. Patent 7,897,795, 2011.
110, 1968. (e) Although we had ready access to this ligand and with it
only being used at 0.4 mol%, cheaper ligands and modified reaction
conditions may provide similar results (see Eames, J.; Mitchell, H. J.;
Nelson, A.; O’Brien, P.; Warren, S.; Wyatt, P. J. Chem. Soc., Perkin
Trans. 1 1999, 1095 and references therein). (f) One-pot
dihydroxylation and oxidative cleavage of olefin 5 using NaIO4 and
catalytic K2OsO2(OH)4 furnished aldehyde 4 in less than 20% yield.
(19) The addition of metalated derivatives of 1-bromo-2,2-
dimethoxyethane to furfural, in an analogous fashion to that report
Collins, P. W.; Kramer, S. W.; Gullikson, G. W. J. Med. Chem. 1987,
30, 1952 for a homologue of aldehyde 4 has shown promise..
(20) It is probable that the oily rather than solid nature of 4 meant
that it was susceptible to aerobic oxidation. Therefore, on
manufacturing scales where it was not appropriate to store 4 under
argon and/or at low temperatures we preferred to use it soon after its
preparation to avoid aerobic degradation.
(8) A reviewer questioned how the use of (R)-1 competes with the
use of the Corey lactone, a compound commonly used in
prostaglandin analogue synthesis on industrial scales. Through design,
both the two-component route (that we use (R)-1 for) and the Corey
lactone route allow divergence and can be used to provide a plurality
of products; both approaches have found use in industry for the
synthesis of prostaglandin analogues. From a cost-wise point of view, a
comparison between (R)-1 and the Corey lactone is not valid. Whilst
the Corey lactone comprises the cyclopentane moiety without either
side chain, (R)-1 already has the α-side chain installed and is a more
advanced prostaglandin intermediate. Moreover, whereas we have
shown,7b for example, that (R)-1 can be converted into travoprost (as
per the two-component approach) in three linear steps (1,4-conjugate
addition, reduction, and desilylation), the Corey lactone approach
requires another eight linear steps from the Corey lactone itself.4
Further, we reasoned that for the targets shown in Figure 1, the two-
component approach would allow divergence later in the synthesis.
Note that whilst travoprost and tafluprost possess identical α-side
chains and bimatoprost and lubiprostone require7b,48 only single
additional specific manipulations to the α-side chain late in the
synthesis, the greater structural variance of the ω-side chains between
targets means that it is best installed as late as possible.
(21) Boulton, L. T.; Brick, D.; Fox, M. E.; Jackson, M.; Lennon, I. C.;
McCague, R.; Parkin, N.; Rhodes, D.; Ruecroft, G. Org. Process Res.
Dev. 2002, 6, 138.
(22) This implies that the n-BuLi carbonyl 1,2-addition product in
solution was a gem-dioxy species rather than a ketone and therefore
resisted conversion to a tertiary alcohol or other products.
(23) Reitz, A. B.; Nortey, S. O.; Jordan, A. D., Jr.; Mutter, M. S.;
Maryanoff, B. E. J. Org. Chem. 1986, 51, 3302.
(24) During chromatography (5% EtOAc/n-heptane/AcOH) of 9
contaminated with 12 mol% trans-9, the first eight fractions were more
enriched with trans-9 (18 mol% to 14 mol%) than the crude input.
When the cis-isomer started to elute enriched (∼9 mol% trans-isomer
trans-9; factions 10−14) relative to the crude input, the HPLC purity
fell to ∼60% due to coelution of impurities.
(25) It was later found that it was better to replace K2CO3 with
Cs2CO3, resulting in fewer equivalents of this base and of isopropyl
iodide being required. DBU21 can also be used.
(9) Corey, E. J.; Weinshenker, N. M.; Schaaf, T. K.; Huber, W. J. Am.
Chem . Soc. 1969, 91, 5675.
(10) (a) Sih, C. J.; Price, P.; Sood, R.; Salomon, R. G.; Peruzzotti, G.;
Casey, M. J. Am. Chem. Soc. 1972, 94, 3643. (b) Kluge, A. F.; Untch, K.
G.; Fried, J. H. J. Am. Chem. Soc. 1972, 94, 7827. (c) Pappo, R.;
Collins, P. W. Tetrahedron Lett. 1972, 2627.
(11) (a) Noyori, R.; Suzuki, M. Angew. Chem., Int. Ed. Engl. 1984, 23,
847. (b) Suzuki, M.; Yanagisawa, A.; Noyori, R. J. Am. Chem. Soc.
1985, 107, 3348.
(26) Laboratory experiments showed improvements in selectivity at
lower temperatures. A scale-up run was conducted in-house (see
Supporting Information for experimental details) in a liquid nitrogen-
cooled reactor using 92% GC pure, precooled (−50 °C in a separate
reservoir) aldehyde 4. The Wittig reaction was conducted at −60 to
−68 °C. Following esterification and column chromatography, olefin 8
was desilylated and analysed by HPLC showing that the desired cis-
isomer had formed along with 9.5 mol% of the undesired trans-isomer.
Although this confirmed that lower temperatures improved selectivity,
(12) Das, S.; Chandrasekhar, S.; Yadav, J. S.; Gree
2007, 107, 3286.
(13) (a) Gruber, L.; Tomoskozi, I.; Major, E.; Kovac
Lett. 1974, 3729. (b) Collins, P. W; Dajani, E. Z.; Driskill, D. R.;
Bruhn, M. S.; Jung, C. J; Pappo, R. J. Med. Chem. 1977, 20, 1152.
́
, R. Chem. Rev.
́
s, G. Tetrahedron
̈
̈
̈
́
(c) Novak
Tetrahedron 1985, 41, 435.
́
, L.; Szan
́
tay, Cs.; Meisel, T.; Aszod
́
i, J.; Szabo,
́
E.; Fekete, J.
(14) (a) Piancatelli, G. Heterocycles 1982, 19, 1735. (b) Piancatelli,
G.; D’Auria, M.; D’Onofrio, F. Synthesis 1994, 867.
1915
dx.doi.org/10.1021/op300188x | Org. Process Res. Dev. 2012, 16, 1905−1916

Organic Process Research & Development
Article
the improvement was too small to justify the use of specialised
cryogenic equipment.
towards obtaining more enriched (R)-acetate ((R)-17; 96.7% ee) at
the expense of less enriched (S)-alcohol ((S)-16; 89.6% ee). The in-
process control method that we used specified that the ee of (R)-17
should not fall below 96.0%. Following the first resolution step, Spur’s
modification was adapted by utilisation of a purification step prior to
guanidinolysis followed by subjecting the enantioenriched (R)-16 to a
second enzymatic resolution (Spur did not report a direct second
resolution of their hydroxycyclopentenone without other manipu-
lations being used). By contrast Babiak and Wong’s second enzymatic
resolution was conducted on the chromatographically separated (R)-
hydroxycyclopentenone process stream.
(39) Although these were not the most expedient conditions
identified in laboratory tests, they allowed adequate control on
production scales where the several-hour turnaround time for
sampling, analysis, and response to the analytical result could
potentially lead to ee erosion due to over-reaction.
(27) cis-/trans-Selectivity was determined by HPLC analysis
following conversion of analytical samples of 9 to 14. No enrichment
of either isomer occurred during the analytical sample preparation.
(28) The use of 5-chloropentanoic acid in place of 5-bromopentanoic
acid (11a) in THF provided no improvement. When LiHMDS in
THF was retested with 8% of the phosphoramide cosolvent, a very
similar result in terms of cis-/trans-selectivity (9.5% trans-isomer),
purity (96%), and yield (74%) was seen as when NaHMDS was used
(entry 6). This indicated that the highly polar and Lewis basic solvent
countered the negative impact of lithium ions.
(29) The best cis-/trans-selectivities (6.6 and 7.7 mol% trans-isomer,
respectively) were obtained when KHMDS or t-BuOK was used in
10% HMPA in THF, but yields of less than 30% of olefin 8 were
observed with low purities (38−53%) and large amounts (22−36%) of
diene side-product 15.
(40) A quantity of 0.46 kg unreacted (R)-16 with 34% ee was
recovered during chromatography. This was recycled by addition to
feedstock rac-16 of a subsequent campaign.
(30) Berndt, M.; Holemann, A.; Niermann, A.; Bentz, C.; Zimmer,
̈
R.; Reissig, H. -U. Eur. J. Org. Chem. 2012, 1299.
(41) A reviewer noted that the syntheses7b,48 of prostaglandin
analogues using (R)-1 required protection of C11-O on two separate
occasions. This was a necessity given the oxidative and anionic reaction
conditions used in the total synthesis. It was also a necessity to free
C11-OH for the enzymatic resolution stage. Despite this, some return
in efficiency was achieved through (i) the first TBS protection step
being combined with the allylation reaction and (ii) the final
desilylations of C11-OTBS in travoprost and bimatoprost syntheses
being conducted concurrently with deprotection of the ω-side chain.7b
Corey reported9 the use of two separate C11-O protecting groups (Ac
and then THP) in his prototypical total synthesis of PGF2α.
(42) Although (R)-1 was produced in the original process as a
mixture of geometric isomers, this did not preclude its use in the
synthesis of subkilogram quantities of prostaglandin analogues meeting
the desired specification. The unwanted trans-isomer could be
removed during chromatography (which is viable21 given the small
volume of these products) and/or crystallisation of the prostaglandin
analogues.
(31) Dygos, J. H.; Adamek, J. P.; Babiak, K. A.; Behling, J. R.; Medich,
J. R.; Ng, J. S.; Wieczorek, J. J. J. Org. Chem. 1991, 56, 2549.
(32) Piancatelli, G.; Scettri, A.; David, G.; D’Auria, M. Tetrahedron
1978, 34, 2775.
(33) (a) Rodríguez, A.; Nomen, M.; Spur, B. W.; Godfroid, J. -J. Eur.
J. Org. Chem. 1999, 2655. (b) Rodríguez, A. R.; Spur, B. W.
Tetrahedron Lett. 2003, 44, 7411. (c) From a toxicological and
environmental standpoint, 1,4-dioxane is an undesirable solvent; see
Laird, T. Org. Process Res. Dev. 2012, 16, 1. (d) Its impact was reduced
by its use at only moderate volume/weight ratios (∼5−7 v/w) with an
approximately equal portion of water. No 1,4-dioxane was detected in
the target compounds by headspace GC analysis. Although the
Piancatelli rearrangement has been reported to proceed in aq acetone,
it was reported32 to provide “poor yields” and “extremely low rate”
when the substituent α- to the hydroxyl group was non-aromatic.
Unfortunately solubility proved to be a constraint when using a
completely aqueous system; see D’Auria, M. Heterocycles 2000, 52, 185
for the rearrangement..
(43) (a) MeNH2, BnNH2, DBU, piperidine, α-methyl-benzylamine,
pyridine, and ammonia. (b) DMAP, DABCO, DBU, N-methylbenzyl-
amine, N-methylmorpholine, Ph2CHNH2, Bn2NH, benzylamine, 4-
chlorobenzylamine, 4-bromobenzylamine, 4-methoxybenzylamine, L-
arginine, L-lysine, and L-histidine. (c) Although the benzylamine salt
was the first useful salt identified and reported by us,7b its 4-methoxy
derivative was later found to possess superior properties. The
benzylamine salt was obtained in disappointing yield with only a
moderate enrichment of the cis-isomer (96.4/3.6 cis-/trans- ratio)
being achieved after recrystallisation.
(44) Good solubility, stability (stable at 50 °C in vacuo and no
exothermic events observed during DSC analysis) and melting point
(mp 90.6−91.1 °C).
(45) At 55 °C HPLC purity of N-acetyl-4-methoxybenzylamine was
detected at 6% at 1 h and 31% after 4 h.
(46) To obtain 9 of sufficient quality for desilylation and subsequent,
direct crystallisation as its salt 2, it was mixed with EtOAc/n-heptane,
the insoluble oil layer was separated, and triphenylphosphine oxide
was precipitated and filtered off upon dilution with n-heptane.
Following extraction with water, 80−90% HPLC purity/70−80%
HPLC potency assay 9 in almost quantitative recovery on a 0.5 kg
scale was obtained.
(34) In related prostaglandin analogue syntheses using similar
conditions, yields of 40−72% for the sequential rearrangement and
isomerisation have been reported; see Clissold, D. W.; Craig, S. W.;
Gadikota, R. R.; He, M.; Jurayj, J. F.; Kazerani, S.; Rannala, E.; Sharma,
P. K. U.S. Patent 7,109,371, 2006 (and refs 31 and 33 herein).
Addition of 0.1−110 mol% hydroquinone, conducting the reaction in
the dark or under an inert atmosphere have so far proven unhelpful. A
future solvent screen to eliminate the use of dioxane altogether, or to
minimise its volume further, should focus on water-miscible ethers or
alcohols (transesterification needs to be taken into account), and a
reinvestigation of the use of acetone is required.
(35) The strategic combination of the Piancatelli rearrangement and
an enzymatic resolution in the synthesis of nonracemic prostaglandin
E1- and E-type phytoprostanes has been reported previously.33a,b
Those syntheses were demonstrated on scales of ≤30 g, unlike the
multikilogram scale reported herein, and different furylcarbinols, 4-
hydroxycyclopentenones, enzymes, and resolution conditions were
used.
(36) A shortcoming of this protocol was that very long reaction times
(4−9 days) were required.
(37) For further details see the Supporting Information.
(38) (a) Almost complete conversion of (R)-16 occurred in 6 h
providing a 97.4% ee of (R)-17 leaving unreacted (S)-16 with 87.6%
ee when using 1.4 vol vinyl acetate and 5 vol MTBE and 5% w/w
Lipase PS “Amano” IM at 50 °C. (R)-16 was consumed within 3 h,
providing 94.6% ee of (R)-17 leaving (S)-16 with 100% ee when using
25 w/w% of (w.r.t. rac-16) Lipase PS “Amano” IM in 8 vol vinyl
acetate at 40 °C. (b) The experimental procedure reported herein is
an adaptation of both Babiak and Wong’s16 method and Spur’s33a
modification. Whereas Babiak and Wong focused on producing highly
enantioenriched (S)-alcohols, the approach reported in this con-
tribution (demonstrated on a multikilogram scale) was directed
(47) Cs2CO3 was preferred to K2CO3 (used in the synthesis of 8)
and required less isopropyl iodide and less base, and the esterification
was more rapid.
(48) Henschke, J. P.; Liu, Y. L.; Xia, L. Z.; Chen, Y. F. PCT Int. Appl.
Pub. WO/2012/048447 A1, 2012.
1916
dx.doi.org/10.1021/op300188x | Org. Process Res. Dev. 2012, 16, 1905−1916