Synthesis of filibuvir. Part II. Second-generation synthesis of a 6,6-disubstituted 2H-pyranone via dieckmann cyclization of a β-acetoxy ester

Article abstract of DOI:10.1021/op400236r

This paper describes an improved sequence for the conversion of an oxazolidinone (3) to a β-keto lactone (5). The primary drivers behind this change were the modest and variable yields observed in the intramolecular cyclization to generate the β-keto lactone. Changing the cyclization substrate from oxazolidinone to alkyl ester offered a significantly improved cyclization, as well as improvements in the alkyne hydrogenation. Selection of the optimal substrates for methanolysis and intermediate salt formation are also described.

Full text of DOI:10.1021/op400236r

Article
pubs.acs.org/OPRD
Synthesis of Filibuvir. Part II. Second-Generation Synthesis of a 6,6-
Disubstituted 2H‑Pyranone via Dieckmann Cyclization of a β‑Acetoxy
Ester
Zhihui Peng,*,† John A. Ragan,*,† Roberto Colon-Cruz,† Brian G. Conway,† Eric M. Cordi,†
Kyle Leeman,‡ Leo J. Letendre,† Li-Jen Ping,† Janice E. Sieser,† Robert A. Singer,† Gregory W. Sluggett,‡
Holly Strohmeyer,‡ and Brian C. Vanderplas†
†Chemical Research & Development and ‡Analytical Research & Development, Pfizer Worldwide Research & Development, Eastern
Point Road, Groton, Connecticut
Jon Blunt, Nicola Mawby, Kevin Meldrum, and Stuart Taylor§
Chemical Research & Development, Process Development Facility Ramsgate Road, Sandwich, Kent, United Kingdom
S
* Supporting Information
ABSTRACT: This paper describes an improved sequence for the conversion of an oxazolidinone (3) to a β-keto lactone (5).
The primary drivers behind this change were the modest and variable yields observed in the intramolecular cyclization to
generate the β-keto lactone. Changing the cyclization substrate from oxazolidinone to alkyl ester offered a significantly improved
cyclization, as well as improvements in the alkyne hydrogenation. Selection of the optimal substrates for methanolysis and
intermediate salt formation are also described.
ethyl ester 8, a reasonable 76% in situ yield of 1 was observed.
However, the lengthy synthesis of substrate 8 was not
competitive with the route ultimately developed (vide infra).
INTRODUCTION
■
In the previous paper (Synthesis of Filibuvir. Part I) we
described an approach to filibuvir (1) in which control of the
C−O chirality was achieved through Evans aldol methodology.
This led to a diastereoselective synthesis of propargylic alcohol
3 (Scheme 1). Through a sequence of Sonogashira coupling,
acylation, and hydrogenation, this intermediate was converted
to acetate 4, which in turn served as a substrate for a
Dieckmann-type cyclization to form β-keto lactone 5. There
were several problems encountered with this route, the most
significant being low and variable yields in the cyclization of 4
to form 5, particularly upon scale-up. Our analysis of the
competing elimination pathways in this reaction suggested that
changing the acyl oxazolidinone to a less acidic carbonyl group
might improve the efficiency of the cyclization. This paper
describes the results of these studies, leading to a more efficient
conversion of 3 to β-keto lactone 5 via Dieckmann cyclization
of a β-acetoxy methyl ester.
OPTIMIZATION OF THE CYCLIZATION SUBSTRATE
■
As discussed in the preceding paper, elimination via C2
deprotonation to form α,β-unsaturated acyloxazolidinones
was problematic during the base-mediated cyclization (Scheme
3). We reasoned that changing the oxazolidinone to other
carbonyl groups might alter the relative acidities of C2 and C3′
and reduce the level of elimination byproducts, thereby
improving the efficiency of the desired cyclization.
We prepared several carboxylic acid derivatives and studied
their cyclizations, as summarized in Table 1 (see Supporting
Information for details on preparation of substrates 10−13).
The substrates are arranged in order of decreasing elimination.
Entry 2 represents the original cyclization substrate (4).
Thioester 10 exhibited more elimination than the oxazolidi-
none. Weinreb amide 11 gave comparable results to the
oxazolidinone. Ethyl ester 12 gave the best results with 90%
cyclization and 7% elimination. Pyrrolidine amide 13 suffered
no elimination, but also failed to cyclize (likely reflecting the
poor leaving group ability of the pyrrolidinyl anion). This trend
tracks reasonably well with the estimated acidity of C2 (these
estimates reflect the consensus evaluation of several chemists
on the team, with guidance from the literature where
available).1
ALTERNATIVE ROUTE INVESTIGATION
■
In addition to the successful ester Dieckmann route (vide
infra), we also considered a convergent Dieckmann approach in
which the triazolopyrimidine fragment was present in the
cyclization substrate (in the route ultimately developed, this
fragment is installed via a reductive coupling after formation of
the β-keto lactone; see the subsequent paper in this series,
Synthesis of Filibuvir. Part III). Two substrates were prepared
to study this cyclization, as shown in Scheme 1 (see Supporting
Information for details on their preparation as well as
evaluation of some related substrates). With oxazolidinone 7,
only trace amounts of 1 were formed by HPLC analysis. With
Received: August 27, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op400236r | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 1. Evans aldol approach to filibuvir (1)
Scheme 2. Convergent Dieckmann approach with oxazolidinone (7) and ester (8) substrates
Scheme 3. Deprotonation regioselectivity impact on cyclization vs elimination pathways
of alkynyl acid 15, and precludes approaches wherein an alkynyl
pyridine is formed in the presence of the free acid. This
cyclization could also be metal-catalyzed, as there are numerous
literature examples of such transformations, including with
palladium.3
On the other hand, we could hydrolyze oxazolidinone 3 to
form acid 17, which was isolated as the crystalline dicyclohexyl-
amine salt in 75% yield (Scheme 5). Although the Sonogashira
reaction of acid 17 with bromopyridine did not provide any
METHANOLYSIS OF THE OXAZOLIDINONE
■
These studies demonstrated that changing the oxazolidinone to
an alkyl ester would offer a significant yield improvement. We
first considered hydrolysis of the oxazolidinone substrates
followed by esterification of the resultant acids to prepare the
ester substrate. Interestingly, the saponification (aq LiOH,
H2O2)2 of oxazolidinone 14, an intermediate in our first-
generation process, led to isolation of enol lactone 16 (Scheme
4). This product apparently arises from spontaneous cyclization
B
dx.doi.org/10.1021/op400236r | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Table 1. Base-mediated cyclization of several carboxylic acid derivatives
a
Cyclization yields in toluene followed the same trend as above, but were 6−14% lower than in THF due to increased formation of elimination
b
products. The only observed product was deacetylation to form the tertiary alcohol.
Scheme 4. Hydrolysis of the acyloxazolidinone 14 and enol lactone formation
Scheme 5. Hydrolysis of oxazolidinone 3, esterification, and Sonogashira reaction
Scheme 6. Alcoholysis of acyloxazolidinone 4
C
dx.doi.org/10.1021/op400236r | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Table 2. Substrate and reagent optimization for the alcoholysis
entry
substrate
conditions
exo/endo (%)
1
2
3
4
5
6
7
8
20
14
MeMgBr, MeOH, THF
45/49
50/25
NaOMe, MeOH
MgBr2 or MgCl2, MeOH/THF
MeMgBr, MeOH/THF
60−80/8−20
85/8
Mg(OMe)2, MeOH/THF
Me(OMe)2, AcCl, MeOH/THF
i-PrMgCl, MeOH/THF
Mg(OMe)2, AcCl, MeOH/THF
80/13
86/7
3
93−95/2−5
93−95/2−5
Scheme 7. Telescoped methanolysis, Sonogashira, and hydrogenation sequence to 22
coupling product, the Fischer esterification of 17 followed by
the Sonogashira reaction of the resultant ester 18 afforded
coupling product 19 cleanly. The overall in situ yield from 3 to
19 was a modest 65%, which was not competitive with the
direct methanolysis approach (vide infra).
significant byproduct with substrate 20 (entry 1). With the
alkynyl pyridine 14, NaOMe/MeOH gave significant endocy-
clic product (entry 2), but showed reasonable selectivities with
magnesium salts (entries 3−6). The best results were obtained
with terminal alkyne 3, which provided >20:1 exo selectivity
with MeOMgCl (entries 7−8).8−10 This reagent was
conveniently prepared in situ either by addition of methanol
to commercial i-PrMgCl in THF or treatment of a methanol
solution of Mg(OMe)2 with HCl (generated in situ with AcCl).
We chose the Grignard method due to its consistency and
lower cost. We also found that 0.5 equiv of MeOMgCl were
sufficient to achieve the same selectivity and reaction rate as a
full equivalent. Further lowering the amount of MeOMgCl not
only slowed the reaction but also gave more endocyclic
byproduct.
We then studied alcoholysis, hoping to form the desired ester
directly. Our initial studies focused on oxazolidinone 4, based
on our desire to make use of the large quantities of this
intermediate that had previously been prepared in our pilot
plant (Scheme 6). Unfortunately, 4 was not a good substrate
for alcoholysis. Most conditions either provided minimal
reaction or gave significant levels of endocyclic product.4
Gadolinium triflate (Gd(OTf)3)5−7 was the only effective Lewis
acid that gave a reasonable 91:9 exo/endo selectivity, but at
least 10% loading of the catalyst was required. While this result
was mechanistically interesting, it was unlikely to be viable on a
multikilogram scale. We soon found improved results with
magnesium alkoxide reagents and other substrates (Table 2).
Several other substrates were next evaluated in magnesium-
mediated methanolyses (Table 2). Endocyclic alcoholysis was a
We preferred to avoid isolation of methyl ester 18 due to its
noncrystalline nature and lack of UV chromophore. Fortu-
nately, we found that the methanolysis product solution could
be taken directly into the Sonogashira coupling to provide
alkynyl pyridine 19 with excellent conversions (Scheme 7). The
D
dx.doi.org/10.1021/op400236r | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 8. Optimized process for conversion of oxazolidinone 3 to β-keto lactone 5
magnesium salts from the methanolysis reaction did not
interfere with the copper and palladium reagents used in the
Sonogashira coupling. As described in the previous paper
(Synthesis of Filibuvir. Part I), the alkyne hydrogenation in the
oxazolidinone series (i.e., 19 to 21, where OMe in the ester is
replaced with the benzyloxazolidinone) had required high
loadings of Pd/C catalyst (25 wt %), a result we attributed to
the presence of residual phosphine and copper species from the
Sonogashira coupling. With the change from oxazolidinone to
methyl ester, we now found that the solubility properties of
alkynyl alcohol 19 were such that it could be extracted from the
organic phase with aqueous HCl (this was not possible with the
less polar oxazolidinone). This allowed for a more robust
cleanup of the Sonogashira product via the following sequence:
(i) aqueous citric acid wash (pH 4−5) to remove inorganic
salts; (ii) aqueous HCl extraction of pyridine 19; (iii)
neutralization with aq base and extraction of 19 into a fresh
organic phase (toluene or i-PrOAc). This sequence provided an
effective purge of residual phosphine and copper impurities
from the Sonogashira coupling, and allowed for significantly
lower catalyst loadings in the alkyne hydrogenation (5 wt %).
The hydrogenation of 19 occurred smoothly in a variety of
solvents (MeOH, toluene, EtOAc, i-PrOAc, i-PrOH, THF),
although the rate was faster in alcohol solvents. A variety of
catalysts were screened with several commercially available
catalysts giving good conversions at 30 °C and 50 psi H2 (e.g.,
Johnson-Mathey type A402028−10). On scale the hydro-
genation was run in 15 volumes of i-PrOAc containing 3%
MeOH, with 92% isolated yield (15 kg scale) following
isolation of the oxalate salt of 21 (vide infra) (pilot plant yields
for all reactions are presented in Scheme 8). Acylation of the
alcohol was achieved under similar conditions to those used in
the earlier sequence (Synthesis of Filibuvir. Part I), with acetic
anhydride and methanesulfonic acid in toluene providing clean
conversion to the desired product 22. These conditions were
identified after initial screening showed Cu(OTf)2 to be an
effective catalyst for acylation of this hindered, tertiary
alcohol.11 Subsequent experiments demonstrated that the
copper triflate could be replaced by a strong acid such as
methanesulfonic acid, suggesting that the role of the copper
triflate was to provide catalytic quantities of trifluoromethane-
sulfonic acid.
SELECTION OF AN INTERMEDIATE FOR
ISOLATION
■
With the four step sequence to convert 3 to 22 defined, we next
turned to selection of the optimal isolation strategy (Scheme
7). There were three potential isolation points, as crystalline
salts of pyridines 19, 21, and 22 had been isolated. Ideally we
sought a single isolation point that would minimize filtrations
and isolations by telescoping two of the three steps. A variety of
salts were screened, from which several promising leads
emerged, including the D-DBTA (dibenzoyl tartaric acid) and
oxalic acid salts of 19, the D-DBTA, L-DBTA and oxalic acid
salts of 21, and the D- and L-DBTA salts of 22. From these
studies, the preferred salt emerged as the oxalate of 21. This
was based on its ease of isolation (reasonably fast filtrations), its
chemical stability, the low cost of oxalic acid, and the high
potency of the resulting salt (i.e., the low molecular weight of
oxalic acid relative to the pyridine).
ACYLATION AND DIECKMANN CYCLIZATION
■
Acylation of alcohol 21 could be performed directly on the
oxalate salt, avoiding the need for a separate salt break.
Acylation conditions were similar to those with the analogous
oxazolidinone substrate; attempts to replace methanesulfonic
acid with other acid catalysts (HCl, H2SO4) or to directly use
acetyl chloride with no added acid were unsuccessful. The
equivalents of MsOH and Ac2O in the acylation were studied,
E
dx.doi.org/10.1021/op400236r | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
and optimal conditions were found to be >1.5 equiv of MsOH
and >2 equiv Ac2O.
Table 4. Time variations for three addition protocols
isolated yield
(%)
Off-gassing was observed during the addition of MsOH and
Ac2O to the oxalate salt, consistent with literature reports.12−14
We quantified the volume of gas generated and found it to be
∼20% of the theoretical volume of CO2 and CO that would
form from complete decomposition of oxalic acid (Figure 1).
Although full decomposition was not observed, risk/hazard
analyses for scale-up were predicated on the complete
conversion of oxalic acid to CO2 and CO.
mode of addition
addition time
normal addition (LHMDS to substrate) 5−10 min
87−88
85−87
81
30 or 90 min
5 h
inverse addition (Substrate to LHMDS) 5−10 min
5 h
86
65
dual stream addition
1 or 5 h
88−90
observed. However, for longer addition times (5 h), as would
be required for larger scale campaigns, the dual stream addition
developed for the oxazolidinone substrate was clearly preferred,
as it exhibited no variation between 1 and 5 h addition times.
The mass balances for all the experiments in Table 4 were
excellent, with the reduced yield of β-keto lactone correspond-
ing to an increase in the level of elimination byproducts.
We also observed a source-dependence for the LHMDS,
similar to that observed with the oxazolidinone substrate. With
LHMDS prepared from n-BuLi and HN(TMS)2 the yield was
consistently 88−90%. However, in the case of one supplier
using a lithium metal process (Li metal, 2-methyl-1,3-butadiene,
HN(TMS)2), a consistently lower yield of 81% was observed.
Similar observations with LDA prepared from n-BuLi vs Li
metal/styrene have also been reported.15,16 In the course of
investigating the cause of this variation, we identified two
additional suppliers who used the lithium metal process, but
with those sources the higher yield was consistently obtained
(88−90%). We also found that if the “low olefin content” grade
of LHMDS was used from the initial supplier, the yield
improved to 85−87%. We studied several potential culprits,
including the presence of residual 2-methyl-2-butene and small
quantities of other metal contaminants (e.g., sodium amide
bases, LiCl, and Li-alkoxides), but none of these accounted for
the observed discrepancy. In all cases where the yield decreased,
an increased level of elimination impurities was observed (i.e.,
the total mass balance was consistent). While these results
continue to intrigue us, our successful identification of at least
three viable commercial suppliers (Optima, BASF, and
Chemetall “Low Olefin Content” LHMDS) for the reagent
and the relatively modest yield variations have attenuated our
concern with this unexplained LHMDS source variation.
Gratifyingly, the outcome of these studies upon scale-up to
15−30 kg batches in our pilot plant were dramatic and positive,
as shown in Table 5. Although fewer batches have been
processed with this new substrate, the average yield difference is
significant (58.1% over 14 batches for the oxazolidinone
substrate, vs 79.7% over 3 batches for ester 23).
Figure 1. Decomposition of oxalic acid.
The Dieckmann cyclization was more efficient in THF than
in toluene (90% vs 76% yield), but we found that up to 1−2
volumes (mL/g substrate) of toluene, the optimal solvent for
acylation of 21, were well tolerated. Thus, following addition of
aqueous sodium carbonate, a water wash, and phase separation,
the toluene phase was distilled to 1−2 volumes, and THF was
added prior to execution of the next reaction. This operation
also served as an effective method for removal of any trace
water following the workup.
Optimization of the base used in the Dieckmann cyclization
is summarized in Table 3. Although optimized on the ethyl
Table 3. Evaluation of base in the Dieckmann cyclization
in situ yield
of 5 (%)
cyclization conditions
other products
LiHMDS, THF, −20 °C
LiHMDS, Et3N, THF, −20 °C
NaHMDS, THF, −20 °C
KHMDS, THF, −20 °C
90
89
43
31
5−6% elimination
7−8% elimination
50% elimination
35% elimination, other
byproducts
KOt-Bu, THF, −20 °C
8
15% SM, 60%
elimination, other
byproducts
LiOt-Bu, THF, −20 to 0 °C
LDA, THF, −20 to −10 °C
8
20% SM, 70% elimination
30
15% elimination, other
byproducts
FINAL PROCESS DESCRIPTION
LiTMP, THF, −20 to −10 °C
40
15% elimination, other
byproducts
■
The optimized sequence to convert alkyne 3 to β-keto lactone
5 is shown in Scheme 8, including yields on pilot plant scale
(16 kg of oxazolidinone 3).
ester 23, comparable results were observed with the methyl
ester. These experiments indicated that LHMDS was the
preferred reagent, consistent with observations made with the
analogous oxazolidinone substrate.
Although the yields with the ester substrate were consistently
higher than those with oxazolidinone, we did observe some
yield dependence on mode of addition and source of LHMDS,
similar to those observed with the oxazolidinone substrate
(Synthesis of Filibuvir. Part I). These are summarized in Table
4. For rapid additions (<10 min), little if any difference was
The key features of the optimized synthesis relative to the
previous route are as follows: (1) An improved yield for the
Dieckmann cyclization (22 to 5·H2O) from an average of 58%
to 80% on multi-kilogram-scale; (2) reduced catalyst load in the
alkyne hydrogenation (the enhanced solubility of alkynyl
alcohol 19 in aqueous acid relative to the analogous
oxazolidinone acetate allowed for a more robust purge of
copper and phosphine contaminants from the Sonogashira
coupling, and thus a reduced catalyst requirement from 25 to 5
F
dx.doi.org/10.1021/op400236r | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Table 5. Campaign yields for cyclization of ester vs
oxazoidinone substrates
Table 7. Sample solution preparation
volume
sample
quantity
0.5 g
(mL)
dissolving solvent
3 to 18 reaction
50
acetonitrile/water
(50/50, v/v)
mixture
18 to 19 reaction
0.5 g
100
100
25
acetonitrile/water
(80/20, v/v)
mixture
19 product solution 0.5 g
acetonitrile/water
(80/20, v/v)
19 to 21 reaction
mixture
0.5 g
acetonitrile/water
(80/20, v/v)
21 oxalic acid salt
50 mg
0.5 mL
25
acetonitrile/water
(80/20, v/v)
21 to 22 reaction
mixture
25
acetonitrile
22 product solution 0.5 g
100
25
acetonitrile
22 to 5 reaction
0.5 mL each
layer
acetonitrile/water
(80/20, v/v)
mixture
5·H2O
20 mg
25
acetonitrile/water
(80/20, v/v)
Table 8. Approximate retention times
wt %; (3) an efficient conversion of oxazolidinone 3 to methyl
ester 18 with MeOMgCl, conveniently prepared from i-PrMgCl
and MeOH.
Development of a reductive coupling for the conversion of β-
keto lactone 5·H2O to filibuvir (1) will be described in the next
paper in this journal (Synthesis of Filibuvir. Part III).
cmpd
retention time (min)
3
6.6
4.8
3.4
6.0
6.8
6.5
7.4
3.9
18
4
4-bromo-2,6-diethylpyridine
19
21
22
5
EXPERIMENTAL SECTION
■
Analytical Methods. The UPLC method used for reaction
monitoring and analysis of isolated intermediates employed a
Waters Acquity series instrument (Waters Corp., Milford, MA),
an Acquity BEH C8 column (100 × 2.1 mm, 1.7 μm particle
size), and a 10 mM ammonium acetate/acetonitrile gradient as
shown in the table below with a re-equilibration time of
approximately 2 min. The flow rate was 0.4 mL/min and the
column temperature was 30 °C. The injection volume was 2 μL
and UV detection was performed at 220 nm.
15 °C over 2 h. A sample was analyzed by UPLC for reaction
completion.
Upon complete methanolysis (<0.1% acyloxazolidinone 3),
the vessel was charged with 4-bromo-2,6-diethylpyridine p-
toluenesulfonic acid (17.3 kg, 99.6 wt %, 44.5 mol), toluene
(41.5 L), and triethylamine (13.5 L, 134 mol). The reactor was
then purged with nitrogen. Bis(triphenylphosphine) palladium-
(II) chloride (312 g, 0.445 mol) was charged, and the reactor
purged with nitrogen. The resulting slurry was stirred at 25 to
30 °C for 30 min, and copper(I) iodide (127 g, 0.666 mol) was
added. The system was purged with nitrogen and warmed to 60
to 65 °C for 10 h.
Gradient Program for UPLC Analysis. See Table 6
Table 6. Gradient program for UPLC analysis
time (min)
% mobile phase A
% mobile phase B
0
95
5
5
95
95
10
10.5
5
Upon reaction completion (UPLC analysis, <3% bromopyr-
idine), the reaction mixture was cooled to ambient temperature
and quenched into a solution of citric acid monohydrate (5.20
kg, 24.5 mol) and water (80 L). The reactor was rinsed with
toluene (22 L). The two-phase quench solution was stirred for
15 min, and the layers were separated. The dark brown organic
layer was cooled to 0 to 5 °C and extracted with two portions
of prechilled aqueous HCl (1 M, 66.8 L for each extraction).
The first aq HCl addition was exothermic, and was added at a
rate such that the temperature remained <10 °C. To the
combined product-containing aqueous extractions was added
isopropyl acetate (128 L), and the two-phase mixture was
cooled to 0 to 5 °C. Aqueous sodium carbonate (20%, 77 L)
was added at such a rate that off-gassing was controlled. The
layers were separated, and the lower aqueous phase was
checked for pH (target pH 8−9). The product-containing
organic phase was washed with water (64 L), and the organic
was assayed by HPLC (12.7 kg of 19, yield 86.9%) and used
directly in the next reaction.
Sample Solution Preparation. See Table 7.
The approximate retention times are listed in Table 8.
(R)-Methyl 3-Cyclopentyl-5-(2,6-diethylpyridin-4-yl)-3-hy-
droxypent-4-ynoate (19). A 250 L reactor was charged with
dry THF (32 L) and cooled to 0 to 5 °C. Isopropyl magnesium
chloride (20.6 wt % in THF, 11.1 kg, 22.3 mol) was added, and
the resulting solution was cooled to −5 to 0 °C. A solution of
methanol (1.35 L, 33.4 mol) in THF (16 L) was added to the
Grignard solution over 20 min (CAUTION: this addition was
highly exothermic and evolved propane gas; the addition was
done at a rate such that the internal temperature remained <25
°C). An additional 32 L portion of methanol was then added,
and the solution of MeOMgCl was stirred until the temperature
had returned to −5 to 0 °C (1−2 h required). Solid acyl
oxazolidinone 3 (16.0 kg, 95 wt %, 44.5 mol) was added, and
the solution was stirred for 3 h at −5 to 0 °C, then warmed to
G
dx.doi.org/10.1021/op400236r | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
An aliquot of 19 was isolated for characterization by flash
chromatography (20% to 40% EtOAc in hexanes) as a colorless
oil: 1H NMR (400 MHz, CDCl3): δ 6.93 (s, 2H), 4.45 (s, 1H),
3.74 (s, 3H), 2.83 (d, 1H, J = 15.8), 2.74 (q, 4H, J = 7.6 Hz),
2.68 (d, 1H, J = 15.8), 2.19 (m, 1H), 1.90−1.60 (m, 5H), 1.60−
1.50 (m, 3H), 1.25 (t, 6H, J = 7.6). 13C NMR (100 MHz,
CDCl3): δ 172.4, 162.9, 131.0, 121.0, 93.1, 82.6, 71.6, 51.9,
49.5, 44.7, 31.5, 31.3, 28.2, 27.4, 25.8, 13.9. IR (KBr pellet, thin
film, cm−1): 3493 (br), 2954, 2868, 1723, 1597, 1548, 1406,
1349, 1202, 1169. HRMS (ESI+): m/z calcd. for C20H28NO3
330.20637; found 330.20642.
(R)-Methyl 3-Cyclopentyl-5-(2,6-diethylpyridin-4-yl)-3-hy-
droxypentanoate, oxalic acid salt (21). A 250 L hastelloy
pressure reactor was charged with a solution consisting of
alkyne 19 (12.3 kg, 37.2 mol) and isopropyl acetate (total
solution weigh 135.2 kg, ∼155 L), rinsing with an additional
portion of isopropyl acetate (37 L). 10% Pd/C (0.613 kg, 5 wt
%, 50% water wet, Johnson−Mathey Type A402028−10) was
charged, followed by methanol (4.9 L). The vessel was purged
with nitrogen (3×), and then hydrogen (3×) at 25 °C,
pressurized to 50 psig (4.5 barg) of hydrogen and stirred at 30
°C for 4 h, at which time hydrogen uptake was complete. The
vessel was depressurized and purged with nitrogen, and an
aliquot was analyzed for reaction completion (<1% cis-olefin).
The reaction mixture was filtered through a glass fiber filter,
rinsing with 49 L of isopropyl acetate.
The solution of free base was washed sequentially, first with
25 L of aqueous NaHCO3 (1.09 kg NaHCO3 (13.0 mol)
dissolved in 25 L water), and then with 25 L of water. The
organic layer was concentrated under vacuum (0.6 bar) to a
volume of ∼98 L (8 L/kg substrate). An additional 86 L of
isopropyl acetate was added and distillation resumed to a
volume of ∼12.3 L (10 L/kg substrate). Water content of an
aliquot was assayed by Karl−Fischer titration (target <0.1%).
Isopropanol (22 L) was added, and the solution was heated to
80 °C. A solution of oxalic acid (4.02 kg, 44.6 mol, 1.2 equiv) in
122.5 L isopropyl acetate was added over 5−10 min, such that
the internal temperature remained >75 °C. The reaction was
cooled to 74 °C at 0.25 °C/min, and seed crystals (61.3 g) were
added. The resulting slurry was stirred at 74 °C for 60 min,
then cooled to 66 °C at 0.1 °C/min. The slurry was reheated to
70 °C at 0.25 °C/min, and granulated for at least 60 min. The
slurry was cooled to 55 °C at 0.1 °C/min, then to 20 °C at 0.25
°C/min. Solids were collected by filtration, rinsing with two
portions of isopropyl acetate (49 L each). The solids were dried
at 0.1 bar and 60 °C for 24 h to provide the product oxalate salt
of 21 (21·(CO2H)2) as white, fluffy solids (14.2 kg, 92% yield).
1H NMR (400 MHz, CDCl3): δ 11.0−10.0 (b, 2H), 7.24 (s,
2H), 3.69 (s, 3H), 3.02 (q, 4H, J = 7.6), 2.86 (m, 2H), 2.61 (d,
1H, J = 15.2), 2.52 (d, 1H, J = 15.2), 2.05 (m, 1H), 1.85 (m, 2
H), 1.70−1.45 (m, 5H), 1.45−1.35 (m, 3H), 1.30 (t, 6H, J =
7.6). 13C (100 MHz, CDCl3): δ 173.1, 163.1, 162.5, 158.4,
122.6, 73.9, 51.9, 47.9, 40.9, 38.3, 30.6, 26.6, 26.5, 26.4, 25.64,
25.58, 13.2. IR (KBr pellet, thin film, cm−1): 3558 (br), 3074,
2948, 2870, 2661, 1714, 1632, 1323, 1193, 1164. HRMS (ESI
+): m/z calcd. for C20H32NO3 334.23767; found 334.23779.
HPLC purity (achiral): 98.9%; oxalic acid content 21.5%
(theory 21.3%); water <0.1%; residual solvents <0.1%.
toluene (1078 L), and cooled to 15 °C. Methanesulfonic acid
(36.5 L, 562 mol, 1.70 equiv) was added at a rate such that the
internal temperature remained below 25 °C. Acetic anhydride
(93.8 L, 992 mol, 3.0 equiv) was charged rapidly (this addition
was not exothermic). The resulting two-phase mixture was
stirred at high agitation for at least 2 h. When complete by
UPLC analysis (<2% SM), the reaction mixture was cooled to
10 °C and quenched by addition of aqueous KOH (296.8 kg of
50 wt % aq KOH, 2645 mol, diluted with 840 L water), added
at a rate such that the internal temperature remained <30 °C.
The resulting solution was stirred at 25 °C for 2 h, then allowed
to settle for 15 min. The aqueous pH was confirmed to be in
target (pH >6), and the phases were separated. The organic
phase was washed with water (700 L), then vacuum
concentrated (0.1 bar) to a volume of ∼350 L. The resulting
toluene solution of product (22) was assayed for water content
by Karl−Fischer titration (target <0.05%). When within target,
it was transferred to a tared vessel, rinsing with THF (280 L),
assayed for potency (∼117.9 kg, 95%), and held for use in the
cyclization (it is stable to storage at this point up to 7 days).
Characterization data of (R)-Methyl 3-acetoxy-3-cyclo-
pentyl-5-(2,6-diethylpyridin-4-yl)pentanoate (22) is from a
sample isolated by flash chromatography (20% to 40% EtOAc
1
in hexanes) as a colorless oil: H NMR (400 MHz, CDCl3): δ
6.82 (s, 2H), 3.69 (s, 3H), 3.08 (d, 1H, J = 14.7), 3.02 (d, 1H, J
= 14.7), 2.78 (q, 4H, J = 7.5), 2.70−2.60 (m, 3H), 2.36 (m,
1H), 2.26 (m, 1 H), 2.05 (s, 3H), 1.80−1.40 (m, 8H), 1.30 (t,
6H, J = 7.5). 13C NMR (100 MHz, CDCl3): δ 170.6, 170.4,
162.7, 151.3, 119.0, 85.0, 51.4, 47.2, 39.5, 37.0, 31.2, 29.9, 26.9,
26.7, 25.3, 25.1, 22.0, 14.1. IR (KBr pellet, thin film, cm−1):
2962, 2871, 1731, 1605, 1565, 1435, 1366, 1242, 1205, 1161.
HRMS (ESI+): m/z calcd. for C22H34NO4 376.24824; found
376.24827.
Cyclization. The following procedure involved the simulta-
neous addition of substrate and LHMDS solutions to a reactor
vessel. Vessel #1 was the reactor vessel, vessel #2 was the
substrate charge vessel, and vessel #3 was the LHMDS charge
vessel. Vessel #1 was charged with anhydrous THF (193 L),
and cooled to −15 to −20 °C. Vessel #2 was charged with a
THF−toluene solution of 22, prepared as described above,
containing 117.9 kg (314 mol) of substrate in ∼554 L total
solution volume, rinsing with 31 L anhydrous THF. This
solution was cooled to −15 °C. Vessel #3 was charged with
LHMDS (550 L of a 23.1 wt % THF solution, 691 mol, 2.2
equiv), rinsing with 35.4 L THF as a line rinse, and cooled to
−10 °C. The contents of charge vessels #2 and #3 were
simultaneously transferred to reactor vessel #1 at such a rate
that the internal temperature remained <5 °C (on this scale the
addition required 2 h). Upon complete addition, stirring was
maintained at −15 °C for 30 min, and the reaction was sampled
for completion (target <2% remaining SM). The reaction was
warmed to −5 °C and quenched with cold (<5 °C) water (707
L), maintaining an internal temperature below 20 °C. Toluene
(236 L) was then added, and the mixture stirred for 30 min at 0
to 5 °C. The mixture was warmed to 20 °C and allowed to
settle. The aqueous phase (containing product at this pH) was
removed, and the organic phase was rinsed with an additional
236 L portion of water. The aqueous phases were combined. A
solution of citric acid monohydrate (66 kg, 314 mol, 1.0 equiv)
in 236 L water was prepared in a separate vessel, and added to
the aqueous product solution. The pH was monitored
throughout the addition until a pH of 6.5−7.5 was reached
(the first 40−50% of the citric acid solution was added rapidly,
(R)-6-Cyclopentyl-6-(2-(2,6-diethylpyridin-4-yl)ethyl)-
dihyro-2H-pyran-2,4(3H)-dione hydrate (5). Acylation. A
2500 L glass-lined reactor was charged with (R)-methyl 3-
cyclopentyl-5-(2,6-diethylpyridin-4-yl)-3-hydroxypentanoate,
oxalic acid salt (21·(CO2H)2) (140 kg, 330.6 mol) and dry
H
dx.doi.org/10.1021/op400236r | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(9) Herbert, B.; Kim, I. H.; Kirk, K. L. J. Org. Chem. 2001, 66, 4892.
(10) Schneider, C. Eur. J. Org. Chem. 1998, 1661.
(11) Chandra, K. L.; Saravanan, P.; Singh, R. K.; Singh, V. K.
Tetrahedron 2002, 58, 1369.
(12) Whitford, E. L. J. Am. Chem. Soc. 1925, 47, 2934.
(13) Danforth, J. D.; Dix, J. J. Am. Chem. Soc. 1971, 93, 6843.
(14) Ervasti, H. K.; Lee, R.; Burgers, P. C.; Ruttink, P. J. A.; Terlouw,
J. K. Int. J. Mass Spectrom. 2006, 249/250, 240.
but then slowed to allow the pH to stabilize as the end point
was approached). Product crystallization began, and the slurry
was held with stirring for 60 min. Additional citric acid was
added slowly until a pH of 5.5−6.5 was reached (target pH
6.0). 75−85% of the prepared citric acid solution was required
to reach the final target pH. The resulting slurry was stirred for
at least 30 min, and the solids were collected by filtration,
rinsing the vessel and filter cake with water (2 × 236 L). The
solids were dried at 35 °C under vacuum to provide the
product (5·H2O) as a white solid (94.6 kg, 87.5% yield).
(15) Hoepker, A. C.; Gupta, L.; Ma, Y.; Faggin, M. F.; Collum, D. B.
J. Am. Chem. Soc. 2011, 133, 7135.
(16) Schreiber, S. L.; Ragan, J. A.; Standaert, R. F. Strategies Tactics
1
Org. Synth. 1991, 3, 417.
Data obtained from a laboratory pilot: H NMR (400 MHz,
CD3OD): (note that the protons on C3 of the dihydropyrone
are not visible due to deuterium exchange with solvent): δ 7.08
(s, 2H), 2.77 (q, 4H, J = 7.6), 2.70−2.65 (m, 3H), 2.43−2.42
(m, 2H), 2.09−2.06 (m, 2H), 1.75−1.4 (m, 8H), 1.27 (t, 6H, J
= 7.6). 13C NMR (100 MHz, CD3OD): (note that the C3
dihydropyrone resonance is broadened due to deuterium
exchange): δ 175.9, 170.3, 161.9, 155.7, 120.5, 89 (br), 83.8,
47.0, 37.8, 33.6, 29.6, 29.5, 26.8, 26.7, 25.5, 13.4. IR (thin film,
KBr disc, cm−1): 3294 (br), 2963, 2870, 1656, 1605. MS (CI):
344.2 (M − H)+. HPLC purity (achiral): 99.5%; 5.1% water
(theory for monohydrate 5.0%); residual solvents 0.27%
(MeTHF and MeOH).
ASSOCIATED CONTENT
* Supporting Information
■
S
Preparation of alternative cyclization substrates 7 and 8
(Scheme 2) and other alternative cyclization studies.
Preparation of cyclization substrates 10−13 (Table 1). This
material is available free of charge via the Internet at http://
pubs.acs.org.
AUTHOR INFORMATION
Corresponding Authors
*E-mail: zhihui.peng@pfizer.com.
*E-mail: john.a.ragan@pfizer.com.
■
Present Address
§Stuart Taylor, Canterbury Christ Church University, North
Holmes Road, Canterbury, Kent, U.K.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors thank Dave Damon and Jerry Weisenburger of
Pfizer for the salt screening studies on 19, 21, 22, and Terry
Rathman (Optima Chemicals) for several helpful conversations
regarding the manufacture and performance of lithium
hexamethyldisilazide.
REFERENCES
■
(1) Ripin, D. H. B. In Practical Synthetic Organic Chemistry; Caron, S.,
Ed.; John Wiley & Sons, 2011; Chapter 18.
(2) Evans, D. A.; Britton, T. C.; Ellman, J. A.; Dorow, R. L. J. Am.
Chem. Soc. 1990, 112, 4011.
(3) Sashida, H.; Kawamukai, A. Synthesis 1999, 1145.
(4) Orita, A.; Nagano, Y.; Hirano, J.; Otera, J. Synlett 2001, 637.
(5) Ishihara, Y.; Mendoza, A.; Baran, P. S. Tetrahedron 2013, 69,
5685.
(6) Yang, Z.; Wang, Z.; Bai, S.; Liu, X.; Lin, L.; Feng, X. Org. Lett.
2011, 13, 596.
(7) Murarka, S.; Zhang, C.; Konieczynska, M. D.; Seidel, D. Org. Lett.
2009, 11, 129.
(8) Evans, D. A.; Weber, A. E. J. Am. Chem. Soc. 1987, 109, 7151.
I
dx.doi.org/10.1021/op400236r | Org. Process Res. Dev. XXXX, XXX, XXX−XXX