An improved synthesis of Pyran-3,5-dione: Application to the synthesis of ABT-598, a potassium channel opener, via Hantzsch reaction

Article abstract of DOI:10.1021/jo052226w

Ketoester 1 is cyclized to give pyran-3,5-dione 2 in 78% yield using a parallel addition of ketoester 1 and base NaO^tBu in refluxing THF. Compared to the previously reported procedures, these optimized conditions have significantly increased the yield of this transformation and the quality of pyran 2 and prove to be suitable for large-scale preparation. An application of 2 to the synthesis of ABT-598, a potassium channel opener, is demonstrated.

Full text of DOI:10.1021/jo052226w

An Improved Synthesis of Pyran-3,5-dione:
Application to the Synthesis of ABT-598, a
Potassium Channel Opener, via Hantzsch
Reaction
Wenke Li,* Gregory S. Wayne,
John E. Lallaman, Sou-Jen Chang, and
Steven J. Wittenberger
FIGURE 1. Structure of ABT-598.
Process Chemistry, Global Pharmaceutical Research and
DeVelopment, Abbott Laboratories,
SCHEME 1. Synthesis of Pyran-3,5-dionea
North Chicago, Illinois 60064-6285
wenke.li@abbott.com
ReceiVed October 26, 2005
a Reagents and conditions: (a) â-methallyl alcohol, KOtBu (2.2 equiv),
rt, 16 h; (b) 5 wt % Amberlyst-15 resin, MeOH, 60 °C, 8 h; (c) K2OsO4
(0.1 wt %), NaIO4 (2.3 equiv), THF/H2O, rt, 11 h; (d) NaOtBu (1.3 equiv),
THF, 65 °C.
Ketoester 1 is cyclized to give pyran-3,5-dione 2 in 78%
yield using a parallel addition of ketoester 1 and base NaOtBu
in refluxing THF. Compared to the previously reported
procedures, these optimized conditions have significantly
increased the yield of this transformation and the quality of
pyran 2 and prove to be suitable for large-scale preparation.
An application of 2 to the synthesis of ABT-598, a potassium
channel opener, is demonstrated.
developed a synthetic approach that was both efficient and
scalable (Scheme 1).
The synthesis began with coupling of chloroacetic acid and
methallyl alcohol. The use of chloroacetic acid as a substrate
allowed an acid-base workup to be utilized and enabled the
effective removal of excess methallyl alcohol. The major
impurity was 2-tert-butyloxyacetic acid, resulting from the
reaction between chloroacetic acid and potassium tert-butoxide.
This impurity was minimized by the use of an excess of
methallyl alcohol and by addition of the base to a solution of
the two substrates. The coupling product 3 was obtained as a
MTBE solution in 88% yield.
To avoid product loss during an aqueous workup due to the
relatively high water solubility of 4, the esterification reaction
was carried out under anhydrous conditions (MeOH, 5 wt % of
Amberlyst-15 resin). Once the reaction was complete (8 h, 60
°C), the resin was filtered off and the methanol removed by
distillation. Unfortunately, some product was lost due to co-
distillation with methanol, resulting in an assay yield of 80%.
A coupling of chloroacetic acid methyl ester and methallyl
alcohol could afford 4 directly. However the reaction was low
yielding and not clean. The two-step sequence we employed
provided methyl ester 4 in good yield and high purity without
requiring additional purification.
Pyran-1,3-dione 2 is an important building block which has
been incorporated into compounds of pharmaceutical interest
such as potassium channel opener ABT-598 for overactive
bladder1 (Figure 1) and quinolizinecarboxylic acid for antibacte-
rial agent.2 Compound 2 also found valuable utility in exploring
biological activities of steroid natural products as their 16-oxa-
homoanalogues.3 During the course of development of ABT-
598, we needed to prepare multi-kilograms of 2. Several
syntheses of diketone 2 were reported in the literature.4-6
However, these procedures utilized transformations and reagents
which were not suitable for our multi-kilogram scale, such as
hydromercuration and organocadmium reagents. Therefore, we
(1) (a) Carroll, W. A.; Agrios, K. A.; Altenbach, R. J.; Drizin, I.; Kort,
M. E. PCT Int. Appl. WO 2000024743A1, 2000, 145 pp. (b) Wayne, G.
S.; Li, W. U.S. Pat. Appl. Publ. US 2003153773A1, 2003, 4 pp.
(2) Ohya, S.; Masuda, N.; Kuroki, Y.; Inoue, T.; Okudo, M.; Iwata, T.;
Kokubo, K.; Mizuno, H.; Hagiwara, M. Jpn. Kokai Tokkyo Koho
JP2003261566A2, 2003, 146 pp.
The olefinic double bond of methyl ester 4 was converted to
the corresponding carbonyl functionality affording ketoester 1
by oxidative cleavage with catalytic K2OsO4 (0.1 wt %) and
NaIO4 (2.3 equiv) in THF/H2O. The active catalyst OsO4 was
generated by in situ oxidation of K2OsO4 with NaIO4. The use
of K2OsO4 as a precatalyst minimized the exposure to toxic
OsO4. Reducing the catalyst loading from 0.5 to 0.1 wt % did
(3) Terasawa, T.; Okada, T. J. Chem. Soc., Perkin Trans. 1 1978, 6,
576-584.
(4) Morgan, M. A.; Van Heyningen, E. J. Am. Chem. Soc. 1957, 79,
422-424.
(5) Altenbach, R. J.; Agrios, K.; Drizin, I.; Carroll, W. A. Synth. Commun.
2004, 34 (4), 557-565.
(6) Terasawa, T.; Okada, T. J. Org. Chem. 1977, 42, 1163-1169.
10.1021/jo052226w CCC: $33.50 © 2006 American Chemical Society
Published on Web 01/20/2006
J. Org. Chem. 2006, 71, 1725-1727
1725

TABLE 1. Effect of Base on Cyclizationa
TABLE 3. Solvent Effect on Cyclizationa
entry
base
yieldb (%)
entry
solvent
yield (%)
1
2
3
4
5
6
7
8
LiHMDS
KHMDS
NaHMDS
LiOMe
23
35
1
2
3
THF
1,4-dioxane
toluene
64
38
32
44
<1c
45
a Reaction was conducted at 65 °C using NaOtBu as base.
KOMe
NaOMe
68
50
64
KOtBu
NaOtBu
a The reaction was conducted in THF at 65 °C. b Yields were determined
by HPLC assay of the quenched reaction mixture versus a working standard.
c Some 1 remained.
TABLE 2. Effect of Temperature on Cyclizationa
entry
T (°C)
yield (%)
1
2
3
4
5
-20
-5
23
45
65
19
28
29
39
50
a The reaction was conducted in THF using KOtBu as base.
FIGURE 2. Effect of addition modes on % yield of cyclization.
higher yield than the base-to-ketoester protocol (64% versus
20%). Using this addition protocol (THF, NaOtBu, 65 °C) we
prepared 2 in 55% solution assay yield on kilogram scale. The
modest yield prompted us to identify further improved cycliza-
tion conditions. After analyzing the ketoester-to-base addition
mode we found that a critical drawback in this addition mode
was the varying concentration of base relative to ketoester 1.
At the beginning of the addition there is a large excess of base,
which decreases over time with the addition of ketoester 1. To
gain insight into how the reaction progressed, yields of the
reaction were determined over time during the addition of
ketoester 1.8 The results (Figure 2, Ketoester-to-base Addition)
indicated that the yield increased with the addition of the
ketoester (or with the decreased excess of base), suggesting a
large excess of base had an adverse effect on the reaction. To
circumvent this problem, a parallel addition method was
developed by which separate base and substrate solutions were
added simultaneously to refluxing THF. As shown in Figure 2
(Parallel Addition), constant high yields over the addition span
were achieved with this addition mode. Upon scaling up to
multi-kilogram scale, the new process averaged 75% assay yield
versus the 55% yield obtained with the ketoester-to-base addition
protocol. The quality of the product was also improved
significantly as a result of a cleaner reaction. The product can
be isolated using recrystallization from ethyl acetate without
resorting to column chromatography.
not affect reaction yield but slightly increased reaction time (12
h). After the reaction was complete, the precipitated salts were
removed by filtration and the aqueous filtrate was washed with
MTBE to remove nonpolar impurities and residual osmium. The
ketoester 1 was extracted three times with dichloromethane, and
after dichloromethane was removed in vacuo, ketoester 1 was
dissolved in THF. The yield of the oxidative cleavage was 89%.
The key step is the cyclization of 1 to form diketone 2. In
the literature, this was achieved by the addition of a ketoester
1 THF solution to NaH in THF at 0 °C with a 48% isolated
yield.6 For safety reasons, we replaced NaH with KOtBu. This
change resulted in a low yield (∼30%) and required chromato-
graphic purification. To identify a robust process for this
conversion, a detailed study of the cyclization conditions (base,
solvent, temperature, etc.) was initiated.
The majority of the screening experiments was conducted in
THF with ketoester 1 solution being added to a base solution.
The initial results showed that sodium methoxide provided the
highest yield (68%) among eight bases screened, and sodium
tert-butoxide gave the second highest (64%) (Table 1, entries
6 and 8). However, the sodium methoxide results were variable,
presumably due to its low solubility in THF.7 Meanwhile, the
NaOtBu reaction constantly produced high yields. A trend of
counterion (Li, Na, K) effect on yield was also observed with
Na+ superior to K+ and to Li+ (Table 1). As a result, NaOtBu
was chosen as the base for cyclization reaction.
A more efficient way to obtain the product is to isolate the
sodium salt of 2 (2.Na), which precipitates during the reaction.
Thus, after the reaction is complete, the reaction mixture is
simply quenched with 1 equiv of water to destroy any excess
NaOtBu and filtered. Later, this salt can be protonated and the
freed diketone 2 either isolated by recrystallization from EtOAc
or used without isolation.
The synthetic utility of pyran-3,5-dione 2 was demonstrated
in the synthesis of ABT-598, a potent potassium channel opener,
via Hantzsch reaction (Scheme 2). The required 4-fluoro-3-
iodobenzaldehyde (5) was synthesized in 55% isolated yield
employing a direct iodination approach using N-iodosuccinimide
Temperature has a big effect on the reaction. When the
reaction was conducted at different temperatures using KOtBu
in THF, the reaction yield increased steadily from 19% at -20
°C to 50% at 65 °C (Table 2). The effect of solvent on the
cyclization reaction was examined among THF,1,4-dioxane and
toluene. THF stood out as the solvent of choice achieving 64%
yield and was subsequently selected for scale-up (Table 3).
The addition mode was critical to achieving high yield. Earlier
experiments showed that ketoester-to-base protocol gave a much
(7) When conducting a 25 g run using NaOMe, the yield dropped abruptly
to 31%. We suspected that this might be due to the relatively low solubility
of NaOMe in THF. The use of stirbar in screening experiment could help
reduce the particle size of NaOMe and hence increase its surface area. This
grinding effect was not present when overhead stirring was used on 25 g
scale.
(8) The data were collected by analyzing aliquots of the reaction mixture
using pyridine as an internal reference.
1726 J. Org. Chem., Vol. 71, No. 4, 2006

SCHEME 2. Synthesis of ABT-598a
kg), the aqueous was extracted twice with MTBE (5.6 L each).
The combined organic layer was washed with brine solution (2.7
kg). Analysis of the product solution by HPLC indicated 955 g
(87.5%) of 3 present.
The MTBE of 3 solution (6.43 kg, 948 g 3, 7.29 mol) was
removed by distillation, and methanol was added. To the resulting
methanol solution of 3 (2 L) was added Amberlyst-15 ion-exchange
resin (50 g). The mixture was heated at 60 °C until <2% 3 remained
by GC analysis (8 h). The reaction was then cooled to room
temperature and filtered. The methanol of the filtrate was removed
by distillation and replaced with THF. Analysis of the product
solution by HPLC indicated 827 g (79%) of 4 present.
To the above THF solution of 4 (2.47 kg, 814 g 4, 5.64 mol)
were charged potassium osmate dihydrate (0.85 g, 2.30 mmol),
distilled water (930 mL), and THF (1.1 L). Then a solution of
sodium periodate (2.72 kg, 12.7 mol) in distilled water (23 L) was
added over 1 h. The reaction was stirred until <2% 4 remained by
HPLC analysis (∼11 h). The reaction mixture was filtered to remove
precipitated sodium iodate, and the filter cake was washed with
distilled water (3.3 L). The filtrate was washed with MTBE (4.4
L). The aqueous was then extracted four times with dichloromethane
(7 L each). Dichloromethane was removed by distillation and
replaced with THF. The water content was reduced to less than
0.2% by azeotropic distillation with THF several times. Analysis
of the product THF solution by HPLC indicated 738 g (89%) of 1
present. An analytical sample of 1 as a colorless oil was obtained
by stripping THF and purifying by column chromatography on silica
gel (hexane/ethyl acetate, 7:3): 1H NMR (400 MHz/CDCl3) δ 4.21
(s, 2H), 4.20 (s, 2H), 3.76 (s, 3H), 2.18 (s, 3H); 13C NMR (100
MHz/CDCl3) δ 205.1, 169.9, 76.5, 68.4, 52.3, 26.8; IR (neat) 1752,
1731, 1217, 1136 cm-1. Anal. Calcd for C6H10O4: C, 49.31; H,
6.90. Found: C, 49.30; H, 7.12.
Pyran-3,5-dione, Sodium Salt (2.Na). To a 50 L flask contain-
ing 10.5 L of THF at reflux were added simultaneously a solution
of sodium tert-butoxide (0.817 kg, 8.5 mol) in THF (12 L) and a
solution of ketoester 1 (1.03 kg, 7.05 mol) in THF (12 L) over 1
h. The addition was carried out using two metering pumps that
had been calibrated at 180 g/min. The base solution started first,
by about 100 mL. The resulting reaction mixture became hetero-
geneous with a yellow color. Following the completion of addition,
the reaction mixture was stirred at 65 °C for 5 min and then
quenched with water (0.13 L). Upon cooling to room temperature,
the slurry was filtered. The wet cake was washed with THF (2 L)
and was then dried under vacuum with nitrogen bleed at 60 °C to
provide 1.066 kg of product (70% potency based on salt; 746 g
product 2.Na; 77.5% yield). An analytical sample of 2 was obtained
as a pale yellow solid by converting 2.Na to 2 and recrystallizing
from ethyl acetate: mp 128-129 °C; 1H NMR (400 MHz/CDCl3)
δ 5.52 (s, 0.34 H, olefinic 4-H, enol form), 4.22 (s, 4 H, 2-H2 and
6-H2), 3.72 (s, 1.21 H, 4-H2, diketone form); 13C NMR (100 Mz/
CDCl3) δ 202.4 (3-C, 5-C, diketone form), 188.2 (3-C, 5-C, enol
form), 101.6 (4-C, enol form), 74.0 (2-C, 6-C, diketone form), 68.2
(2-C, 6-C, enol form), 55.9 (4-C, diketone form); IR (KBr) 1644,
1549, 1423, 1231 cm-1. Anal. Calcd for C5H6O3: C, 52.63; H,
5.30. Found: C, 52.63; H, 5.15.
a Reagents and conditions: (a) 6 M HCl; (b) NIS (1.2 equiv), AcOH/
H2SO4, 40 °C, 12 h; (c) NEt3, EtOAc/HOiPr, 50 °C, 1 h; (d) NH4OAc,
HOAc, 105 °C, 1 h.
(NIS) in AcOH/H2SO4 at 40 °C, a modification of Olah’s NIS/
triflic acid conditions for substituted benzonitriles.9 Because of
the competitive diiodination reaction, the reaction was allowed
to proceed to approximately 78% completion, thereby control-
ling the level of diiodo byproduct 5′ to < 2%. The sodium salt
of 2 was dissolved in a small amount of water and its pH
adjusted to 1.2 with 6 M HCl solution. The resulting 2 was
extracted into ethyl acetate and condensed with 0.5 equiv of
aldehyde 5 to give triethylamine salt 6 in 88% isolated yield
with >99% pa purity. This novel Hantszch reaction intermediate
is a new synthetic entity.1b Since the final product, ABT-598,
is very insoluble in most of common solvents for recrystalli-
zation, the isolation of triethylamine salt provides an important
opportunity for purification. The triethylamine salt then was
heated at 105 °C with ammonium acetate in acetic acid to form
ABT-598 as a yellow solid with ∼4% of pryan 7. Purification
using 1% KOH in EtOH removed pyran 7 and afforded pure
ABT-598 in 75% isolated yield.
In conclusion, we have developed an efficient synthesis for
pyran-3,5-dione 2. Optimized conditions for the key cyclization
were established resulting in substantial improvements in both
reaction yield and product purity. The synthetically important
pyran-3,5-dione was used in the preparation of ABT-598.
Experimental Section
2-Oxopropoxyacetic Acid Methyl Ester (1). To a solution of
chloroacetic acid (794 g, 8.4 mol) in THF (8 L) was added
â-methallyl alcohol (720 g, 9.98 mol). The mixture was cooled to
-2 °C, and a solution of potassium tert-butoxide (1.95 kg, 16.9
mol) in THF (10 L) was added slowly to the mixture. The reaction
was stirred at ambient temperature until <3% chloroacetic acid
remained by HPLC analysis (∼16 h). The reaction was then
quenched slowly with distilled water (5.6 L). The mixture was
washed twice with methyl tert-butyl ether (MTBE) (2.8 and 5.6
L). After the pH was adjusted to 1.6 with 6 M HCl solution (1.8
Acknowledgment. We thank Russell Hertzler for analytical
support and Ian Marsden for helpful discussions on NMR issues.
Note Added after ASAP Publication. Author Sou-Jen Chang
was added to the version published ASAP January 20, 2006;
the corrected version was published ASAP January 27, 2006.
Supporting Information Available: Experimental procedures
1
for compounds 5, 6, and ABT-598 and H NMR and 13C NMR
spectra for compounds 1, 2, 5, 6, and ABT-598. This material is
available free of charge via the Internet at http://pubs.acs.org.
(9) Olah, G. A.; Wang, Q.; Sandford, G.; Surya Prakashet, G. K. J. Org.
Chem. 1993, 58, 3194-3195.
JO052226W
J. Org. Chem, Vol. 71, No. 4, 2006 1727