An Efficient and Cost-Effective Synthesis of Pagoclone

Article abstract of DOI:10.1021/op034060b

The compound (+)-2-(7-chloro-1,8-naphthyridin-2-yl)-3S-(5-methyl-2- oxohexyl)-l-isoindolinone (pagoclone) shows anxiolytic activity due to partial agonism of the benzodiazepine site of the GABAA receptor. We describe the development of an economical and practical process for a 100+ kg pilot plant production used to supply development needs. For the key reaction, a β-keto phosphonium salt was prepared by selectively reacting a primary α-bromo ketone with triphenylphosphine in the presence of a secondary α-bromo ketone. A novel Wittig reaction with a 1-isoindolinone was used to produce racemic pagoclone. The enantiomerically pure drug substance was prepared by hydrolyzing a γ-lactam and resolving the resulting enantiomeric carboxylic acids with (+)-ephedrine hemihydrate. An alternate resolution, involving chiral multicolumn chromatography (MCC) was also developed. The synthesis was completed by a racemization-free lactam formation to afford pagoclone.

Full text of DOI:10.1021/op034060b

Organic Process Research & Development 2003, 7, 851−855
An Efficient and Cost-Effective Synthesis of Pagoclone
Timothy L. Stuk,*,† Bryce K. Assink,† Ronald C. Bates, Jr.,‡ David T. Erdman,† Victor Fedij,† Sandra M. Jennings,†
Jennifer A. Lassig,† Randy J. Smith,‡ and Traci L. Smith†
Pfizer Global Research and DeVelopment, Holland Laboratories, 188 Howard AVenue, Holland, Michigan 49424, and
Bioprocess R&D, MS 4053 Eastern Point Road, Groton, Connecticut 06340
Scheme 1
Abstract:
The compound (+)-2-(7-chloro-1,8-naphthyridin-2-yl)-3S-(5-
methyl-2-oxohexyl)-1-isoindolinone (pagoclone) shows anxiolytic
activity due to partial agonism of the benzodiazepine site of
the GABAA receptor. We describe the development of an
economical and practical process for a 100+ kg pilot plant
production used to supply development needs. For the key
reaction, a â-keto phosphonium salt was prepared by selectively
reacting a primary r-bromo ketone with triphenylphosphine
in the presence of a secondary r-bromo ketone. A novel Wittig
reaction with a 1-isoindolinone was used to produce racemic
pagoclone. The enantiomerically pure drug substance was
prepared by hydrolyzing a γ-lactam and resolving the resulting
enantiomeric carboxylic acids with (+)-ephedrine hemihydrate.
An alternate resolution, involving chiral multicolumn chroma-
tography (MCC) was also developed. The synthesis was com-
pleted by a racemization-free lactam formation to afford
pagoclone.
Our approach was based on the assumption that, under
certain conditions, the hydroxy-isoindolinone 4 would be in
equilibrium with the aldehyde tautomer 5 (Scheme 2). If this
were the case, a Wittig olefination of that aldehyde followed
by a Michael-type ring closure would afford the desired
racemic product, 3.
Results and Discussion
Our process began with the synthesis of naphthyridine 6
from 2,6-diaminopyridine and malic acid using a modifica-
tion of a literature procedure3 (Scheme 3). Hazard evaluation
of our reaction revealed two safety concerns that had to be
addressed prior to scale-up. The first was the significant
adiabatic heat rise that occurred when the diamine was added
to the concd sulfuric acid (∆Tad ) 192 °C). The material
could not be added to a cold solvent because, somewhat
surprisingly, the dissolution would require days to complete.
The solution was to add the material in six portions at 40
°C and allow the exotherm (ca. 30 °C) to subside (ca. 30
min) prior to each addition. The second concern was the
generation of an equivalent of carbon monoxide after the
addition of the malic acid. Fortunately, evaluation of the
headspace gases showed that the CO was not liberated until
the reaction had reached 65 °C. Therefore, charging the malic
acid to a 25 °C reaction through an open man-way was not
considered to be a hazard to the operators. We found that
the free base of the naphthyridine (6) was a very fine powder
and formed a clay during filtration. The sulfuric acid salt,
on the other hand, could be isolated as a fast-filtering, yellow
powder simply by addition of brine to the reaction mixture.4
The sulfuric acid salt of the product was used crude (93%
pure) in the subsequent step.
Introduction
Pagoclone (1) is a partial agonist at the benzodiazepine
site of the GABAA receptor and is active in preclinical
models predictive of anxiolytic activity. This licensed
compound1 was under development for the treatment of
general anxiety disorder (GAD) and panic disorder.
It was found that the known route2 to racemic pagoclone
(3), which involved the alkylation of dichloride 2 and
subsequent decarboxylation (Scheme 1), provided low through-
put and was also prohibitively expensive. Because of the
large volumes of drug substance required for the development
program, it was necessary to invent a more practical
approach.
The synthesis of the requisite ylide precursor, phospho-
nium salt (10), was not as straightforward as we had initially
* To whom correspondence should be addressed. E-mail: timothy.stuk@
pfizer.com. Telephone: (616) 392-2375 ext. 2401. Fax: (616) 392-8916.
† Pfizer Global Research and Development.
(3) The free base of compound 6 has been reported: Newkome, G.; Garbis,
S.; Majestic, V.; Fronczek, F.; Chiari, G. J. Org. Chem. 1981, 46(5), 833.
This protocol, in our hands, produced a low yield of product contaminated
with large amounts of Na2SO4.
(4) It was also observed that, because of the low basicity of the compound, a
water wash of the filter cake would remove the acid and form the free-base
clay (with substantial product loss). To avoid the clay formation, the sulfuric
acid salt was chosen as the form for isolation.
‡ Bioprocess R&D.
(1) Pagoclone was originally licensed by Warner-Lambert (now Pfizer, Inc.)
from Interneuron Pharmaceuticals (now Indevus Pharmaceuticals) who in
turn licensed it from Rhone-Poulenc Rorer (now Aventis).
(2) Bourzat, J.-D., et al. (Rhone-Poulenc Sante). U.S. Patent 4,960,779, October,
1990.
10.1021/op034060b CCC: $25.00 © 2003 American Chemical Society
Published on Web 09/17/2003
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
851

Scheme 2
Scheme 3
Scheme 5
in a nonpolar solvent, such as tert-butyl methyl ether, we
have found that the undesired pathway is extremely slow
and the desired product (10) can be isolated almost exclu-
sively. A small amount (1-3%) of triphenylphosphine oxide
is evident in the product and is the result of either traces of
oxygen in the reaction or, more likely, the formation of small
amounts of 11 which is hydrolyzed during analysis. Because
11 is simply converted back to the original ketone during
ylide formation, it did not serve as a source of concern. The
conversion to the phosphonium salt is very efficient, giving
a two-step yield of 57%.
Scheme 4
The ylide formed rapidly in a biphasic system of aqueous
carbonate and xylenes; the phosphonium salt is initially
insoluble in both phases, but after 15 min the ylide is formed
and two clear phases are achieved. After removing the
aqueous layer and refluxing the ylide with hydroxy-isoin-
dolinone 4 in xylenes (ca. 136 °C) for 24 h (Scheme 5), an
85% yield of the desired adduct 3 was realized. A screen of
alternate reaction solvents showed that only nonpolar aprotic
solvents gave acceptable yields and that addition of polar
cosolvents (e.g., DMSO, DMF, DMI, DMP, and the like)
greatly decreased the conversion. The reaction proceeds at
temperatures as low as 110 °C, but this is impractical because
the rate of ylide decomposition is competitive with the
desired reaction at this temperature. After the reaction is
complete, a solvent switch to 2-propanol allows for the
isolation of racemic pagoclone (3) as a white crystalline solid
with >99% chemical purity.
A key difficulty in this synthesis was the resolution of
the enantiomers of 3; the molecule is not susceptible to salt
formation, and we could find no practical enzymatic resolu-
tion. (Lactam hydrolysis was investigated, for example.) We
also looked briefly at making diastereomeric derivatives of
the ketone functionality, but found that the ketone was very
sterically crowded, making derivatization inefficient at best.
A complicating factor was the ease with which the resolved
ketone will racemize under basic, acidic, and thermal
conditions. Deuterium labeling studies showed that the
mechanism for racemization was a retro-Michael/Michael
reaction, as shown in Scheme 6. The low activation energy
of this racemization pathway made deprotection of the ketone
derivatives almost impossible.
hoped. With common brominating agents the preferred
position of bromination is known to be the more substituted
carbon.5 An alternative is to brominate the silyl enol ether
of the kinetic enolate,6 but we chose not to pursue this at
large scale because of the cost it would impart to this
potentially inexpensive component. Instead, we experimented
with brominating methyl iso-amyl ketone in methanol, a
solvent which is known7 to give decent levels of bromination
at the primary carbon. The best result obtained for our
substrate was a 63:37 ratio of C-1 (8) to C-3 (9) bromination
(Scheme 4), with a small amount (<5%) of 1,3-dibromina-
tion.7 We initially attempted to purify the resulting R-bro-
moketones by fractional distillation, but their similar boiling
points (207 and 200 °C for 8 and 9, respectively) and their
flagrantly lachrymose nature encouraged us to rapidly seek
other avenues. The solution was to take advantage of the
reactivity differences of the C-1 and C-3 bromides with
triphenylphosphine. It is known8 that secondary and tertiary
R-bromoketones tend, instead of a normal Sn2 displacement
of the halide, to form an enolphosphonium compound by an
SN2′ mechanism (Scheme 4). By running the displacement
(5) March, J. AdVanced Organic Chemistry, 3rd ed.; John Wiley & Sons: New
York, 1985; pp 529-530.
(6) Blanco, L.; Amice, P.; Conia, J. Synthesis 1976, 194.
(7) Ha, H.-J.; Lee, S.-K.; Ha, Y.-J.; Park, J.-W. Synth. Commun. 1994, 24(18),
2557.
(8) Further dilution of the reaction will give an increasing ratio, but the cost of
decreased throughput negates this benefit. Bromine was used as the limiting
reagent in order to minimize the amount of 1,3-dibromo product.
The method used for the initial deliveries of drug
substance was to open the lactam under hydrolytic conditions
to afford carboxylic acid 13 which can be resolved via
852
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development

Scheme 6
ethanol (15 vol) at 20 °C for several hours, a small amount
of racemic material could be removed by filtration, and the
material in solution was >98% ee. After concentration and
recrystallization from ethanol, ca. 90% of the material was
salvaged.
The above-described resolution process had an undesirable
number of steps, was the bottleneck in the plant, and was
responsible for 80% of the cost of goods. Therefore, an
alternate means of resolution was pursued. It was found that
the enantiomers of pagoclone separate well on a number of
chiral HPLC stationary phases, and this led us to investigate
chiral multicolumn chromatography11 as a means to a cost-
effective resolution. The process (see Experimental Section
for details) was tested on a 26 kg batch of racemic feed and
produced a 95% yield of >99% ee (+)-pagoclone in the
raffinate. The (-)-enatiomer could be racemized by heating
the extract stream with a solution of aqueous base (KOH,
for example). This recycling of the extract would allow for
an almost quantitative yield on the resolution, and the
estimated throughput would be 25 kg of (+)-pagoclone per
day using 30-cm columns. This translates to about 2.0 kg of
pure enantiomer per kg of stationary phase per day. We could
have expected even greater productivity, except that in our
case the throughput was limited by the low solubility of the
racemic material in the mobile phase (17 g/L).
Scheme 7
Scheme 8
diastereomeric salt formation. As shown in Scheme 7, the
challenge in this ring-opening reaction was to avoid over-
hydrolysis to the enone 14 which, under most reaction
conditions tried, was a competitive or even dominant
reaction. The conditions employed for pilot runs involved
hydrolysis with aqueous hydroxide in a two-phase mixture
of 1,2-dimethoxyethane and THF at 34 °C for at least 24 h.
Even with these highly optimized conditions, the reaction
could only be run to about 98% completion and the amount
of byproduct 14 was in the 5-8% range.
After removal of the residual racemic pagoclone (3) by
precipitation, the carboxylic acid 13 could be isolated as a
stable white powder in 85% yield by precipitation from
aqueous methanol. The acid was resolved using the known9
ephedrine salt formation (Scheme 8) to provide the acid 15
in >98% ee and 80% theoretical yield. The salt can then be
neutralized and the lactam reformed with carbonyldiimida-
zole (CDI) or acetic anhydride/imidazole to produce (+)-
pagoclone (1) in 95% yield. The (-)-enantiomer of the acid
can be recovered from the resolution step and recycled. We
were unsuccessful in scrambling the stereocenter in the open
form (13) because â-elimination of the amine to 14 was rapid
and irreversible, but once the lactam was reformed with CDI
the material racemized rapidly in the ring-opening step.
(Scheme 7).
Until the final batch of pagoclone was produced, we
operated under the assumption that it was critical to obtain
high ee in the resolution step because crystallization of the
final drug substance did not improve enantiometric purity.
However, the final batch was partially racemized10 in the
final isolation step to 88% ee, and an alternate purification
had to be devised out of necessity. It was known that the
racemic crystal of pagoclone is less soluble and higher-
melting than the pure enantiomer (174 vs 165 °C), and this
fact was used to our advantage. We found that after stirring
the batch in a mixture of acetone (25 vol) and absolute
Although there would be capital expenses incurred in the
commission of a chromatography unit, the cost savings
gained by using this chromatography as a continuous process
are substantial. The cost of the final drug substance would
be decreased by >60%.
Conclusion
Starting from inexpensive raw materials, we have devel-
oped a practical, scalable, and cost-effective commercial
synthesis of pagoclone. Use of asymmetric multicolumn
chromatography for the resolution of the enantiomers was
demonstrated as an alternative to classical resolution.
Experimental Section
2-Amino-7-hydroxy-1,8-naphthyridine Sulfuric Acid
Salt (6). 2,6-Diaminopyridine (150 kg, 1375 mol) was added
in six portions to concd sulfuric acid (1030 kg) at 40-50
°C. The solution was allowed to exotherm, and the next
portion was not added until the temperature was within range.
After the solution became homogeneous, the reaction was
cooled to 25 °C and D,L-malic acid (185 kg, 1380 mol) was
added in one portion. The reaction was heated to 110-120
°C over 1 h. CAUTION: Carbon monoxide is evolved during
this time. After 1 h, the reaction was cooled to 25 °C and
transferred into a cold aqueous sodium chloride solution (10
°C, 150 kg in 1125 L of water), maintaining the temperature
below 50 °C. The slurry was stirred for 2 h at 20 °C and
then filtered. The filter cake was washed with hexanes (270
L) to drive out residual water. The solid was dried under
vacuum at 60 °C to afford 303 kg (85% yield) of a pale
1
(9) David-Comte et al. (Rhone-Poulenc Rorer). U.S Patent 5,498,716, March,
yellow powder. MS (DCI) M + 1 at 162, 100%; H NMR
1996.
(200 MHz, DMSO-d6): δ 4.32 (br s, 5H, -OH, -NH2,
(10) The racemization was most likely due to contamination of the crystallization
solution with traces of base, although this could not be confirmed in
retrospect.
(11) This process is similar to simulated moving bed chromatography.
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
853

H2SO4), 6.41 (d, J ) 9.0 Hz, 1H), 6.56 (d, J ) 9.0 Hz, 1H),
7.85 (d, J ) 6.6 Hz, 1H), 7.91 (d, J ) 6.6 Hz, 1H); mp
>350 °C.
distillation conditions to remove residual water, then under
reflux). The reaction mixture was cooled to 80 °C and then
vacuum distilled to remove the majority of the xylenes. To
the resulting slurry was charged 2-propanol (2284 L). The
slurry was heated to reflux and then cooled to 20 °C. The
solids were collected by filtration, washed with 2-propanol
(600 L) and methanol (300 L), and dried under vacuum at
60 °C to afford 170 kg (82%) of the title compound as an
off-white solid. APCI/MS: M + H+ at 408, 100%; 1H NMR
(200 MHz, DMSO-d6): 8.87 (d, J ) 8.8, 1H), 8.61 (m, 2H),
7.93 (d, J ) 7.0, 1H), 7.74 (m, 4H), 6.05 (m, 1H), 3.62 (m,
1H), 3.28 (dd, J ) 7.0, 17.2, 1H), 2.42 (m, 2H), 1.35 (m,
3H), 0.79 (d, J ) 6.2, 6H); mp 173-174 °C.
[(5-Methyl-2-oxo)-hexyl]-triphenylphosphonium Bro-
mide (8). To a solution of methanol (850 L) and 5-methyl-
2-hexanone (152 kg, 1330 mol) at 0 °C was added bromine
(185 kg, 1156 mol) such that the temperature remained below
15 °C. The solution was stirred at 10 °C for about 2 h. An
exotherm (ca. 10 °C) occurs when the reaction is about 80%
complete; after this subsides, the reaction is complete. The
reaction was quenched with water (148 L). After the mixture
was stirred for 30 min, tert-butyl methyl ether (1100 kg)
was added, followed by a sodium chloride solution (133 kg
in 740 L of water). After the mixture was stirred for 15 min,14
the aqueous layer was discarded. The organics were further
washed with a sodium bicarbonate solution (36 kg in 744 L
of water) and a sodium chloride solution (as above). The
solvent was vacuum distilled, replaced with tert-butyl methyl
ether (550 kg), and redistilled. To the cooled bromoketone
product in tert-butyl methyl ether (281 kg, 10 °C) was added
a solution of triphenylphosphine (303 kg, 1156 mol) in tert-
butyl methyl ether (281 kg). After 12 h at 20 °C, the solids
were filtered and washed with tert-butyl methyl ether (115
kg). The material was dried under vacuum for at least 12 h
at 40 °C to afford 300 kg (57%) of white crystals. 1H NMR:
7.87 (m, 15H), 5.66 (dd, J ) 2.9, 12.8, 2H), 2.71 (m, 2H),
1.35 (m, 3H), 0.80 (d, J ) 6.2, 6H); CI (MS) M at 455,
100%.
(()-2-[1-(7-Chloro-1,8-naphthyridin-2-ylamino)-6-meth-
yl-3-oxo-heptyl]-benzoic Acid (13). A reactor was charged
with racemic pagoclone (3) (111 kg, 272 mol), 1,2-
dimethoxyethane (404 kg), and tetrahydrofuran (633 L),
followed by addition of a potassium hydroxide solution (85
kg in 1100 L of water). The solution was stirred at 34 °C
for at least 30 h. The reaction was cooled to 20 °C, and the
lower aqueous layer was discarded. Water (610 L) was
added, and the pH was adjusted to 9 with aqueous hydro-
chloric acid (4N). The tetrahydrofuran was removed by
vacuum distillation. Water (350 L) was added, and the pH
was adjusted to 11.5 with aqueous potassium hydroxide
(1.4N). The precipitate (residual racemic pagoclone) was
removed by filtration. After adding dichloromethane (1027
kg), the aqueous layer was acidified to less than pH 1.4.
The organic layer was washed with water (450 L) and
concentrated under vacuum. The residue was precipitated
from methanol (300 L) and water (320 L), filtered, and dried
2-Hydroxy-7-N-phthalimidyl-1,8-naphthyridine (7). A
reactor was charged with naphthyridine salt (6) (165 kg, 636
mol), phthalic anhydride (246 kg, 1660 mol), and glacial
acetic acid (850 kg) and cooled to 20 °C. Triethylamine was
added (257 kg, 2540 mol) while maintaining the temperature
below 30 °C. The reaction was heated to 115 °C and held
for 5 h. After cooling to below 30 °C, methanol (900 L)
was added, and the slurry was stirred for 30 min. The solids
were filtered and rinsed with methanol (440 L). The product
was dried under vacuum at 60 °C to afford 176 kg (95%) of
a tan powder. 1H NMR: 12.45 (s, 1H), 8.43 (d, J ) 8.1 Hz,
1H), 8.09 (m, 5H), 7.49 (d, J ) 8.1, 1H), 6.74 (d, J ) 9.5,
1H); CI (MS) M + 1 at 292, 100%.
(()-2-(7-Chloro-1,8-naphthyridin-2-yl)-3-hydroxy-1-
isoindolinone (4). A reactor was charged with hydroxy-
phthalimide (7) (120 kg, 412 mol), sodium chloride12 (1.2
kg), acetonitrile (755 kg), and dimethylformamide (1.5 kg).
The mixture was heated to reflux (ca. 83 °C), and a solution
of phosphorus oxychloride (69.1 kg, 450 mol) in acetonitrile
(5 kg) was added. After 4 h at reflux, the mixture was cooled
to 5 °C, and acetonitrile (240 kg) and potassium hydroxide
(201 kg of a 45% aqueous solution) were added. After the
mixture was stirred for 15 min, the pH was adjusted to 8
with additional potassium hydroxide if necessary. A solution
of potassium borohydride (83.3 kg, 1544 mol) in water (900
L) was added to the reaction at 0-15 °C. After the mixture
was stirred for 1 h at 20-30 °C, the reaction was quenched
by addition of glacial acetic acid (630 kg) and water (5 L).
After the mixture was stirred for 15 min, the solids were
collected by filtration and washed with water (2 × 300 L)
and methanol (2 × 300 L). The solids were dried under
vacuum at 75 °C to afford 109 kg (85%) of a tan solid. MS
1
(DCI) M + 1 at 312, 100%; H NMR (400 MHz, DMSO-
d6): 8.58 (d, J ) 10.5 Hz, 1H), 8.52 (d, J ) 10.5 Hz), 8.45
(d, J ) 8.7 Hz, 1H), 7.83 (ddd, J ) 7.7, 1.0, 1.0 Hz, 1H),
7.77 (ddd, J ) 7.7, 7.7, 1.4 Hz, 1H), 7.72 (ddd, J ) 7.7,
1.4, 1.4 Hz, 1H), 7.63 (dd, J ) 7.7, 1.4 Hz), 7.60 (d, J )
8.7 Hz, 1H), 7.05 (s, 1H), 6.95 (br s, 1H); 13C NMR (100
MHz, DMSO-d6) 166, 154, 153, 153, 144, 141, 139, 134,
130, 130, 124, 123, 122, 119, 116.
(()-2-(7-Chloro-1,8-naphthyridin-2-yl)-3-(5-methyl-2-
oxo-hexyl)-1-isoindolinone (3). While being stirred, the
following were charged to a reactor in this order: water
(1185 L), sodium carbonate (120 kg, 1700 mol), phospho-
nium salt (10) (370 kg, 813 mol), and xylenes (1350 kg).
After the mixture was stirred for 30 min, the reaction became
clear, and the lower aqueous layer was removed. The or-
ganics were then washed13 with a sodium carbonate solution
(60 kg in 1185 L of water). To the organics was added the
hydroxy compound (4) (158 kg, 507 mol). The reaction
mixture was heated to 136 °C for 24 h (initially under
(12) The product was very fine and difficult to isolate by centrifuge unless sodium
chloride was added to the reaction mixture.
(13) The purpose of this additional wash is to remove residual bromide. If this
is not done, up to 0.5% of the resulting product will have bromo rather
than chloro-naphthyridine.
(14) The reaction was initially monitored for completion by GC and was
complicated by some methyl ketal formation during the reaction. This stir
allows adequate time for the ketals to be converted to ketone.
854
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development

at 50 °C under vacuum to afford 99.7 kg (86%) of a white
2H), 7.93 (d, J ) 7.0, 1H), 7.74 (m, 4H), 6.05 (m, 1H), 3.62
(m, 1H), 3.28 (dd, J ) 7.0, 17.2, 1H), 2.42 (m, 2H), 1.35
(m, 3H), 0.79 (d, J ) 6.2, 6H); 13C NMR (100 MHz, DMSO-
d6): δ 80.74 (C2), 115.98 (C19), 119.068 (C17), 121.91
(C15), 123.39 (C9), 124.10 (C6), 129.90 (C8), 130.23 (C4),
133.90 (C7), 139.40 (C18), 140.53 (C16), 144.45 (C3),
152.69 (C12), 153.06 (C10), 153.77 (C14), 166.43 (C5); CI
1
powder. DCI/MS: M + H+ at 426, 100%; H NMR (200
MHz, DMSO-d6): δ 13.5 (br s, 1H), 8.25 (d, J ) 8 Hz,
1H), 8.04 (d, J ) 9 Hz, 1H), 7.85 (d, J ) 9 Hz, 1H), 7.80
(d, J ) 9 Hz, 1H), 7.60 (dd, J ) 8, 8 Hz, 1H), 7.45 (dd, J
) 8, 8 Hz, 1H), 7.30 (dd, J ) 8, 8 Hz, 1H), 7.15 (d, J ) 9
Hz), 6.90 (d, J ) 9 Hz, 1H), 2.9 (m, 2H), 2.5 (m, 2H), 1.3
(m, 3H), 0.8 (d, 6H); mp 173-174 °C.
(MS) M + 1 at 408, 100%; [R]20 ) +135° (c ) 1,
D
(+)-2-(7-Chloro-1,8-naphthyridin-2-yl)-3S-(5-methyl-
2-oxohexyl)-1-isoindolinone (1). A reactor was charged with
100 kg of carboxylic acid (13) (234 mol), ethanol (385 kg),
water (23 L), and (1S,2R)-ephedrine hemihydrate (42.8 kg,
246 mol). The reaction was heated at 40 °C until homoge-
neous, filtered, and cooled to 20 °C until onset of crystal-
lization and then to 0 °C for 2 h. The precipitate was filtered
and washed with a solution of ethanol (221 kg) and water
(11 L). This solid was added to a solution of concd
hydrochloric acid (12.8 kg), water (133 L), and dichlo-
romethane (535 kg). After 15 min, the aqueous layer was
removed, and the organics were washed with water (135 L).
The dichloromethane was distilled at atmospheric pressure
to a volume of 250 L, and then a solution of N,N-
carbonyldiimidazole (30.3 kg) in dichloromethane (234 kg)
was added. After 15 min, water (256 L) was added. The
aqueous layer was removed, and the organics were washed
with additional water (256 L). The organics were concen-
trated by distillation and replaced with ethanol (530 kg). The
slurry was cooled to 5 °C for 3 h and filtered. The solids
were dried under vacuum at 60 °C to afford 35 kg (63% of
theory, 32% yield) of a white crystalline material. 1H NMR
(200 MHz, DMSO-d6): δ 8.87 (d, J ) 8.8, 1H), 8.61 (m,
dichloromethane); mp 169 °C.
Multicolumn Chromatography Conditions. A 26 kg lot
of racemic material (3) was purified using a mobile phase
of 90% toluene and 10% 2-propanol and a feed of 17 gm/L
in the mobile phase. The chiral stationary phase was 720
gm of (S,S)-Whelk-O 1 available from Regis Technologies
which was packed into 6 × 5 cmID columns. The feed flow
rate was 59 mL/min, and the raffinate produced a stream of
product with >99.5% optical purity. The product was isolated
by concentration and crystallization from ethanol. The extract
stream was racemized with 0.1% (v/v) of 0.5N potassium
hydroxide at 50 °C for 2 h. Following racemization, the
solution was washed with one volume 0.1N hydrochloric acid
and 1 vol of water.
Acknowledgment
The authors thank Dr. Sechoing Lin, Craig Rossman,
Cecilia Sandee, and Dr. Tom Zennie for analyses of the
compounds and isolation of impurities.
Received for review May 19, 2003.
OP034060B
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
855