Diastereomeric salt resolution based synthesis of LY503430, an AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) potentiator

Article abstract of DOI:10.1021/op0500741

This article describes the development and optimization of chemical reactions and subsequent preparation of the API LY503430 under cGMPs to fund first human dose (FHD) clinical evaluation as a potential therapeutic agent for Parkinson's disease. Reasons and rationale are presented for changes in solvents and reagents. One of the major developments presented here is the replacement of a chiral chromatography with a diastereomeric salt resolution. This article also discusses a preferred orientation issue with LY503430 which complicated the XRPD analysis.

Full text of DOI:10.1021/op0500741

Organic Process Research & Development 2005, 9, 621−628
Diastereomeric Salt Resolution Based Synthesis of LY503430, an AMPA
(r-Amino-3-hydroxy-5-methyl-4-isoxazolepropionic Acid) Potentiator
Nicholas A. Magnus,* James A. Aikins, Jason S. Cronin, William D. Diseroad, Amy D. Hargis, Michael E. LeTourneau,
Bruce E. Parker, Susan M. Reutzel-Edens, John P. Schafer, Michael A. Staszak, Gregory A. Stephenson,
Shella L. Tameze, and Lisa M. H. Zollars
Eli Lilly and Company, Chemical Product Research and DeVelopment DiVision, Indianapolis, Indiana 46285
Scheme 1
Abstract:
This article describes the development and optimization of
chemical reactions and subsequent preparation of the API
LY503430 under cGMPs to fund first human dose (FHD)
clinical evaluation as a potential therapeutic agent for Parkin-
son’s disease. Reasons and rationale are presented for changes
in solvents and reagents. One of the major developments
presented here is the replacement of a chiral chromatography
with a diastereomeric salt resolution. This article also discusses
a preferred orientation issue with LY503430 which complicated
the XRPD analysis.
Introduction
We were charged with preparing about a kilogram of
LY503430 to permit its evaluation in the clinic as a potential
2.2 This reaction had been performed in a binary solvent
t
system of toluene and THF, employing solid BuOK. The
reaction was refluxed for 24 h, worked up with water, dried
with MgSO4, concentrated to dryness, and slurried in hexanes
to facilitate filtration of the triphenylphosphine oxide byprod-
uct. The Wittig olefination had some development opportuni-
ties: (1) simplify the solvent system by going from a binary
therapeutic agent for Parkinson’s disease.1 LY503430 had
been prepared at the multigram scale in our process depart-
ment in laboratory glassware and had required a chiral
chromatography (Scheme 1). To prepare LY503430 at the
kilogram scale we needed to make the chemistry amenable
to “fixed” pilot plant equipment for the first few reactions
and, if possible, replace the chiral chromatography with a
diastereomeric salt resolution. The poor solubility of the
biphenyl acid 10 and ultimately LY503430 (and any biphe-
nyl-related impurities as a result of the synthesis of 10 and
LY503430) led us to develop an impurity control strategy
that relied on having high chemical and optical purity of the
Suzuki coupling partner 8 prior to assembling the biphenyl
linkage.
t
reaction solvent to a single solvent, (2) avoid solid BuOK
charges for safety reasons and better control of the reaction
temperature, (3) remove the drying-agent operation and
resulting waste, and (4) avoid concentrating to dryness to
proceed with the triphenylphosphine oxide precipitation. It
was found that the reaction ran cleanly to completion in THF
t
alone, and that the solid BuOK could be replaced with a 1
t
t
M BuOK solution in THF. After the 1 M BuOK in THF
was added to the reaction mixture of ketone 1 and methyl-
triphenylphosphonium bromide in THF over 0.5 h at 25 °C
and the reaction mixture held for 1.5 h, the conversion of 1
to 2 was complete. The workup was changed to adding the
reaction to a mixture of heptane and water. After phase
separation, the organic phase was concentrated to azeotro-
pically remove water and THF to give a dry slurry of the
phosphine oxide in heptane directly. The phosphine oxide
was removed very productively by filtration, and the alkene
2 was processed forward as a heptane stock solution. These
Results and Discussion
The first step in the synthesis is a Wittig olefination to
convert 4-iodoacetophenone 1 into the corresponding alkene
* To whom correspondence should be addressed. E-mail: magnus_nicholas@
lilly.com.
(1) Murray, T. K.; Whalley, K.; Robinson, C. S.; Ward, M. A.; Hicks, C. A.;
Caroline, A.; Lodge, D.; Vandergriff, J. L.; Baumbarger, P.; Siuda, E.; Gates,
M.; Ogden, A. M. J. Pharmacol. Exp. Ther. 2003, 306, 2, 752-762. Bender,
M. D.; Cantrell, B. U.; Fray, A. H.; Jones, W. D.; Miller, W. D.; Mitchell,
D.; Simon, R. L.; Zarrinmayeh, H.; Zimmreman, D. M. Monofluoroalkyl
Derivatives. (Eli Lilly and Co., Indiana, U.S.A.). PCT Int. Appl. WO 00/
66546, 2000.
(2) Bachman, B. G.; Carlson, C. L.; Robinson, M. J. Am. Chem. Soc. 1951,
73, 1964-1965. Perera, R. P.; Wimalasena, D. S.; Wimalasena, K. J. Med.
Chem. 2003, 13, 2599-2605.
10.1021/op0500741 CCC: $30.25 © 2005 American Chemical Society
Published on Web 07/27/2005
Vol. 9, No. 5, 2005 / Organic Process Research & Development
•
621

modifications allowed the process to run in standard pilot
plant equipment with yields of alkene 2 obtained in >95%.
Step 2 in Scheme 1 involves converting alkene 2 into the
fluorobromo adduct 3 by exposing 2 to the in situ generated
fluorine bromine mixed halogen.3 The original reaction
procedure was run by dissolving alkene 2 in dichloromethane
and adding triethylamine trihydrofluoride (TREAT) to that
mixture at 25 °C. At this point, the reaction mixture was
analyzed, and no reaction had taken place. The mixture was
cooled to 0-5 °C, and N-bromosuccinimide was introduced
to the reaction mixture to generate the mixed halogen and
effect fluorobromination of the alkene 2, presumably via
Markovnikov attack of a bromonium ion by fluoride. This
reaction mixture was quenched into aqueous ammonium
hydroxide, the organic phase was separated and dried with
MgSO4, and the adduct 3 was isolated as a crude oil after
concentration. Minor changes were made for pilot plant scale-
up. The heptane stock solution of 2 was diluted with
dichloromethane, and the reaction was conducted as de-
scribed above with similar results. The reaction was worked
up as before, but instead of concentrating the resulting
mixture to a crude oil, the heptane and dichloromethane
reaction solvents were replaced by DMF via azeotropic
distillation. The fluorobromo adduct 3 was isolated as a DMF
stock solution, due to 3 being an oil and handling more easily
as a stock solution and DMF being the preferred solvent for
the subsequent Gabriel reaction.4 On the basis of laboratory
use tests, the solution yields of 3 were estimated to be 90%.
Step 3 in Scheme 1 is a Gabriel reaction that was run by
combining the fluorobromo adduct 3 with potassium phthal-
imide 4 in DMSO and heating the resulting mixture to 130-
140 °C for about 18 h. The phthalimide adduct 5 was
precipitated from the reaction as a crude amorphous solid
by employing a slow water addition, and the solid was further
purified by crystallization from dichloromethane with hex-
anes as the anti-solvent. Evaluation of this reaction in our
hazard laboratory revealed a potentially dangerous exotherm
with an onset at 155 °C. Since DMF is a common solvent
for Gabriel reactions, we evaluated it as a substitute for
DMSO, found no exothermic issues and that the reaction
was virtually identical to the reaction employing DMSO as
the reaction solvent. These Gabriel reactions ran as slurries
with a large amount of potassium phthalimide 4 out of
solution. We speculated that the 18 h reaction time was due
to the poor solubility of the potassium phthalimide 4. To
remedy this problem, tetrabutylammonium bromide (TBAB)
was added to the reaction as a phase transfer catalyst, and
the reaction time was consequently reduced from 18 to 4 h.
As before, the phthalimide adduct 5 was precipitated from
the reaction by the slow addition of water. The crude
precipitate of 5 filtered well and was mainly contaminated
with potassium phthalimide 4. HPLC assays of the dried
crude precipitate indicated that it was about 70 wt % 5. The
70 wt % pure adduct 5 was found to perform acceptably in
the next phthalimide cleavage reaction so that no further
purification was required. The yields of 5 were estimated to
be about 65% over three steps starting from ketone 1 with a
total of 24.3 kg of crude 5 isolated (17.2 kg of actual 5 based
on weight percent purity).
Step 4, in Scheme 1 is the cleavage of the phthalimide 5
to give amine 6, which was originally achieved with
hydrazine.5 Hydrazine was not a chemical that we wanted
to handle in our pilot plant due to its toxicity. For the pilot
plant campaign we chose to use 40% aqueous methylamine
in ethanol to cleave the pthalimide group.6 While the
methylamine promoted phthalimide cleavage reaction worked
acceptably in the laboratory, it presented problems in the
pilot plant due to the reversibility of the reaction. The
N-methyl phthalimide that was generated as a byproduct in
the desired reaction was attacked by the primary amine 6 to
give the bis-secondary amide 11 (Scheme 2) during the
workup phase of the reaction, which led to yield fluctuation.
This issue was amplified in the pilot plant due to longer
residence times of the processing streams in the reactors.
However, the primary amine 6 could be purified by extracting
it into water as a hydrochloride salt, and the bis-secondary
amide 11 and N-methyl phthalimide were removed by
filtration and extraction with MTBE, respectively. The
hydrochloride salt of 6 was treated with aqueous sodium
hydroxide and the free base extracted into MTBE. The
MTBE extracts containing primary amine 6 were concen-
trated and delivered to our kilo labs as stock solutions. The
yields of amine 6 were variable (86, 61, and 56% over three
runs) due to a lack of understanding of how to work up the
reaction without causing reaction reversal. In total, 8.96 kg
of crude amine 6 were prepared as MTBE stock solutions.
The racemic sulfonamide 7 had been separated by chiral
chromatography (step 6 in Scheme 1). To avoid the costly
chromatography we developed a diastereomeric salt resolu-
tion of the precursor primary amine 6 with (S)-(-)-(2)-
pyrrolidone-5-carboxylic acid 12. The MTBE stock solutions
of 6 were concentrated to an oil and subsequently dissolved
in 5% aqueous 2-propanol (v/v). The resulting mixture was
heated to 65-70 °C, and a 5% aqueous 2-propanol solution
of the acid 12 was slowly introduced to the mixture, inducing
crystallization of the diastereomeric salt 13. The diastereo-
meric salt resolution proved to be a dynamic process with
the diastereomeric excess (de) of the crystals enriching over
time. The de of the isolated salt 13 ranged from 92.25 to
96.77%, and averaged at 94.5% over six runs with yields in
the 39-41% range.
The sulfonylation reaction (Scheme 1) was originally
conducted in dichloromethane by reacting racemic amine 6
with isopropylsulfonyl chloride in the presence of triethyl-
amine and catalytic DMAP.7 We wanted to avoid the use of
dichloromethane and DMAP in the preparation of sulfon-
amide 8; therefore, alternative solvents, reagents, and condi-
(3) Dolensky, B.; Kirk, K. L. J. Org. Chem. 2001, 66, 4687-4691. Alvernhe,
G.; Laurent, A.; Haufe, G. Synthesis 1987, 6, 562-564. McClinton, M. A.
Aldrichimica Acta 1995, 28, 2, 31-36.
(4) Gabriel, S. Ber. Dtsch. Chem. Ges. 1887, 20, 2224. Review: The Gabriel
Synthesis of Primary Amines. Gibson, M. S.; Bradshaw, R. W. Angew.
Chem., Int. Ed.. Engl. 1968, 7, 12, 919-930.
(5) Ing, H. R.; Manske, R. F. H. J. Chem. Soc. 1926, 2384.
(6) Motawia M. S.; Mengel, J.; Abdel-Megid, A. E. S.; Pedersen, E. B. Synthesis
1989, 384.
622
•
Vol. 9, No. 5, 2005 / Organic Process Research & Development

Scheme 2
tions were evaluated as substitutes. The use of carbonate and
phosphate bases proved encouraging, but large excesses of
isopropylsulfonyl chloride were necessary to achieve high
conversion. Toluene and THF emerged as the solvents of
choice for the sulfonylation, with a mixture of the two
solvents giving better impurity suppression than either solvent
by itself. The diastereomeric salt 13 was converted to its
free base with aqueous NaOH, extracted into MTBE, dried
with sodium sulfate, and concentrated to an oil. The oil was
dissolved in a mixture of toluene and THF, cooled to 0-5
°C, and treated with TEA, followed by a slow addition of
isopropylsulfonyl chloride to give about 95% conversion to
sulfonamide 8. The reaction was quenched with aqueous
HCl, and after the organic phase was separated and dried
with sodium sulfate, concentration gave a “technical grade”
solid of about 90% pure sulfonamide 8. Step 7 in the
synthesis of LY503430 was a Suzuki coupling to establish
the biphenyl linkage. The biphenyl motif in this chemistry
renders the molecule very insoluble and makes any biphenyl-
related impurities difficult to remove from the desired
products 10 and LY503430. Therefore, the crude sulfonamide
8 was recrystallized from acetonitrile and water, which
reduced the yield by 10 to 15%, but gave sulfonamide 8
with excellent purity. The isolated yields of purified sul-
fonamide 8 were consistently in the 76% range over four
batches with enantiomeric excesses of >99%.
filtered through Celite and acidified with aqueous HCl to
precipitate the acid 10 as a fine, slow-filtering solid with
isolated yields of about 70% with <95% purity. A Suzuki
reaction with similar coupling partners was being run in our
department with palladium black as the catalyst, which was
giving good results.9 We applied the palladium black Suzuki
reaction conditions to the coupling of sulfonamide 8 to
boronic acid 9. The sulfonamide 8 was combined with a
slight subcess of the boronic acid 9 (0.99 equiv relative to
the sulfonamide 8), dissolved in aqueous MeOH containing
K2CO3, catalytic palladium black was added, and the
resulting mixture was heated to 63-68 °C for 1 h. These
conditions gave >99% conversion of the boronic acid 9
(lower conversion levels and/or >0.99 equiv of 9 led to
difficulty removing the boronic acid from the product acid
10). The reaction mixture was filtered through Celite and
the Celite rinsed with warm water to displace product 10.
The resulting clarified mixture was heated to 60-65 °C and
acidified slowly with acetic acid to induce crystallization of
very high purity, rapid filtering acid 10 in about 88% yield
over five batches as the single desired enantiomer. The
palladium levels in the isolated acid 10 typically ranged from
3 to 8 ppm.
The last step in the synthesis of LY503430 was an amide
coupling that used diphenylphosphinic chloride with N-
methylmorpholine in THF to effect the coupling of methy-
lamine with the acid 10.10 The workup involved an aqueous
acidic wash, followed by an aqueous basic wash, drying the
organic phase with MgSO4, and removing the solvent to give
the amide LY503430 as a crude solid. The crude LY503430
The conditions we received for running the Suzuki
reaction to couple sulfonamide 8 to 4-carboxylphenylboronic
acid 9 were as follows: Pd(OAc)2, Ph3P, and Na2CO3 in
aqueous IPA at reflux for 3 h.8 This reaction mixture was
(7) Ornstein P. L.; Zimmerman, D. M.; Arnold, M. B.; Bleisch, T. J.; Cantrell,
B.; Simon, R.; Zarrinmayeh, H.; Baker, S. R.; Gates, M.; Tizzano, J. P.;
Bleakman, D.J. Med. Chem. 2000, 43, 4354-4358.
(8) Reviews: Suzuki, A. Acc. Chem. Res. 1982, 15, 178. Pure Appl. Chem.
1985, 57, 1749; 1991, 63, 419; 1994, 66, 213.
(9) Marck, G.; Villiger, A.; Buchecker, R. Tetrahedron Lett. 1994, 35, 3277-
3280.
(10) Jackson, A. G.; Kenner, G. W.; Moore, G. A.; Ramage, R.; Thorpe, W. D.
Tetrahedron Lett. 1976, 17, 3627-3630.
Vol. 9, No. 5, 2005 / Organic Process Research & Development
•
623

Figure 1. Lath morphology of LY503430 gave rise to preferred orientation issues when gathering the XRPD (X-ray powder
diffraction) data. Example of an XRPD pattern of LY503430 before and after grinding the sample to exemplify the preferred
orientation issue. The “ground” sample’s XRPD pattern in red has diffraction peaks that were not present prior to grinding.
was purified by recrystallization from acetone/water (1/ 4)
employing 130 volumes (mL/g) of solvent. We found that
the combination of 1,1′-carbonyldiimidazole (CDI) as the
coupling agent with EtOAc as the reaction solvent had some
operational advantages over the existing diphenylphosphinic
chloride procedure.11 The acid 10 was combined with CDI
in EtOAc at 23 °C to give a slurry, which became
homogeneous as the corresponding imidazolide was formed.
Eventually the imidazolide crystallized from solution, but
warming the mixture to 40 °C redissolved the imidazolide
to give a homogeneous solution. We took this opportunity
to polish filter the mixture prior to crystallizing the final
product. Slow subsurface introduction of a 2 M solution of
methylamine in THF to the imidazolide caused LY503430
to crystallize from solution (if the methylamine was intro-
duced into the headspace of the reactor, it would react with
the carbon dioxide produced from the CDI degradation, and
the carbonate of methylamine deposited as a white solid on
the reactor walls). After filtration, the resulting wet cake was
further purified by employing an EtOH re-slurry, affording
LY503430 with excellent chemical purity. This protocol
avoided a workup and unnecessary drying of the “technical”
grade API, which was being handled with a high level of
containment. The LY503430 was afforded in 85% yield, with
99.8 wt % potency by HPLC, and the undesired enantiomer
of LY503430 not detectable.
suggesting that preferred orientation effects complicated the
XRPD analysis. The crystals of LY503430 had a lath habit,
lacking in the third dimension as illustrated in Figure 1. This
crystal morphology can lead to preferred orientation issues
when conducting XRPD analyses.
To determine if preferred orientation complicated the
XRPD analysis, a lot of LY503430 was lightly ground using
an agate mortar and pestle, and an XRPD pattern was
collected. As shown in Figure 1, the diffraction peak
intensities, before (in black) and after (in red), were dramati-
cally affected by the grinding process. The black arrows in
Figure 1 point to new diffraction peaks in the XRPD pattern
that were not present prior to grinding. These results
confirmed the effects of preferred orientation; however, phase
transformations induced by even light grinding could not be
ruled out as a possible cause for the appearance of the new
diffraction peaks. Therefore, to unambiguously show that a
single crystalline form of LY503430 was present, thin
crystals of acceptable quality for X-ray structure analysis
were grown from EtOH-H2O, and the crystal structure was
solved. From the crystal structure, a theoretical powder
pattern was calculated. Aside from a slight shifting of a few
of the diffraction peaks due to the different temperatures at
which the simulated and experimental data were obtained
(148 K vs ambient temperature, respectively), there is
excellent agreement between the simulated pattern and the
experimental pattern of the ground material, Figure 2. This
result confirms that all of the peaks observed in the different
LY503430 powder patterns were due to a single crystalline
phase. The absolute stereochemistry of LY503430 was also
confirmed in the single crystal X-ray analysis using the sulfur
atom as the asymmetric reference.12
X-ray powder diffraction (XRPD), used to evaluate the
crystal phase composition of LY503430, showed consider-
able lot-to-lot variability in both the relative peak intensities
and the number of diffraction peaks, which raised a concern
that polymorphs might be present. Solid state 13C NMR
spectra (not shown) collected for lots of LY503430, which
gave different XRPD patterns, were virtually identical,
(11) Paul, R.; Anderson, G. W. J. Am. Chem. Soc. 1960, 82, 4596.
(12) X-ray data in the Supporting Information.
624
•
Vol. 9, No. 5, 2005 / Organic Process Research & Development

HPLC chiral and potency assay for acid 10: column:
ChiralPak AD, 250 nm × 4.6 mm; temperature: 30 °C;
detection: UV 270 nm; injection: 20 µL; flow rate: 1.0
mL/min; eluent: 85:15 (v:v) hexane (w/ 0.2% AcOH):IPA
(w/0.2% AcOH). Peaks: enantiomers of acid 10 at 20.0 (S-
enantiomer) and 26.6 (R-enantiomer) min.
HPLC chiral and potency assay for acid LY503430:
column: ChiralPak AD, 250 nm × 4.6 mm; temperature:
30 °C; detection: UV 270 nm; injection: 20 µL; flow rate:
1.0 mL/min.; eluent: 80:20 (v:v) hexane:IPA. Peaks: enan-
tiomers of acid 10 at 10.7 (S-enantiomer) and 15.1 (R-
enantiomer) min.
2-[2-Fluoro-2-(4-iodo-phenyl)-propyl]-isoindole-1,3-di-
one (5). Step 1. Under a nitrogen atmosphere, a reactor was
charged with THF (35 L) followed by 4-iodoacetophenone
1 (5.298 kg, 21.53 mol) at 20-25 °C. Methyltriphenylphos-
phonium bromide (8.492 kg, 23.77 mol) was added to the
reaction mixture, followed by a 20% tBuOK in THF solution
(14.6 kg, 25.85 mol) addition over about 0.5 h, between 20
and 30 °C. The reaction mixture was cooled to 22 °C and
stirred for about 1.5 h. Subsequent GC analysis of the
reaction mixture indicated complete consumption of the
4-iodoacetophenone 1. Heptane (30 L) was charged to a
quench vessel followed by DI water (45 L). The reaction
was charged to the quench vessel over about 0.5 h, between
20 and 30 °C (no exotherm observed), and the resulting
mixture was agitated for about 0.5 h. Agitation was stopped,
and the phases were allowed to separate over about 0.5 h.
The bottom aqueous phase was removed, and a 1% aqueous
NaCl solution (45 L) was charged to the organic mixture.
The resulting mixture was agitated for about 0.25 h and held
to separate for about 0.5 h (20-25 °C). The bottom aqueous
phase was removed, the reactor was set up for vacuum
distillation (40-55 °C under 250-375 mmHg vacuum), and
50 L of distillate was removed, producing a triphenylphos-
phine oxide slurry in heptane. The reactor contents were
cooled to 0-5 °C over about 0.5 h to further precipitate
triphenylphosphine oxide. The contents of the reactor were
passed through a filter into a second reactor to remove
precipitated triphenylphosphine oxide, and the oxide waste
cake was washed with cold heptane (20 L, 0-5 °C) to give
an end volume of 43 L.
Step 2. CH2Cl2 (20 L) and triethylamine trihydrofluoride
(7.998 kg, 49.61 mol) were charged to the alkene 2 heptane
mixture between 20 and 25 °C, and the resulting mixture
was cooled to 0-5 °C. N-Bromosuccinimide (4.184 kg, 23.51
mol) was charged to the reaction mixture in one portion,
producing an exotherm which was controlled with jacket
cooling (set to 0-5 °C). After the exotherm subsided, the
reaction mixture was heated to 20-25 °C over about 1.0 h.
After about 4.0 h of stirring the reaction mixture at 20-25
°C, GC indicated 7.9 area % of alkene 2 remaining (<2 area
% of alkene 2 was the criteria for reaction completion). An
additional portion of N-bromosuccinimide (0.80 kg, 4.50
mol) was added, and after about 1.0 h the GC area % of
alkene 2 was 1.8. (Note: GC area % criteria of alkene 2
can be met in <3.0 h with 1.2 equiv of NBS charged
initially). An aqueous ammonia solution was prepared by
Figure 2.
Conclusion
The diastereomeric salt resolution based synthesis of
LY503430 was developed and performed in a pilot plant and
kilo lab in eight isolated stages with an overall yield of
11.6%. A total of 2.19 kg of LY503430 was produced to
fund phase I clinical evaluation.
Experimental Section
Analytical Methods for the Scheme 2 Synthesis of
LY503430. GC conditions for the first stage of the synthesis:
column: J & W DB 1701 (30 m × 0.25 µm); carrier gas:
He, 1.1 mL @ 40 °C; injection: 1.0 µL @ 250 °C, split
50:1; temperature program: 40-280 °C at 15 °C/min, hold
at final temperature for 15 min; detection FID @ 300 °C;
sample prep: ∼5 mg/mL in CH2Cl2. Peaks: alkene 2 at 9.9
min, fluorobromo adduct 3 at 13.2 min, Phthalimide adduct
5 at 22.6 min, amine 6 at 12.2 min.
CE chiral assay for diastereomeric salt 13: column: 48
cm long × 50 µm i.d. uncoated capillary; temperature: 30
°C; detection: 232 nm; voltage: +25 kV; buffer: 25 mM
citrate/tris pH 2.5 + 2.5 mM 18C6H4 + 1.5 mM DM â CD;
injection: 50 mbar × 2 s; sample prep: ∼0.4 mg/mL in 1:1
(v:v) CH3CN:H2O. Peaks: enantiomers of amine 6 at 18.3
(S-enantiomer) and 19.4 (R-enantiomer) minutes.
HPLC potency assay for diastereomeric salt 13: col-
umn: Zorbax SB-phenyl 250 nm × 4.6 mm; temperature:
25 °C; detection: UV 229 nm; injection: 20 µL; eluent:
isocratic, 30% CH3CN (w/0.05% TFA): 70% Milli-Q H2O
(w/ 0.05% TFA); sample prep: ∼0.1 mg/mL in 1:1 (v:v)
CH3CN:H2O. Peaks: amine 6 at 6.2 min.
HPLC chiral assay for sulfonamide 8: column: Chiralpak
AD, 250 × 4.6 mm, 30 °C; detection: UV 230 nm; eluent:
95:5 v/v hexane/IPA; flow: 1.0 mL/min; injection: ∼20 µL;
sample prep: ∼0.3 mg/mL 1:1 v/v IPA/hexane. Peaks:
enantiomers of sulfonamide 8 at 16.6 (S-enantiomer) and 28.0
(R-enantiomer) min.
HPLC potency assay for sulfonamide 8: column: Zorbax
SB-phenyl 250 nm × 4.6 mm; temperature: 25 °C; detec-
tion: UV 232 nm; injection: 20 µL; eluent: isocratic, 55%
CH3CN (w/0.05% TFA): 45% Milli-Q H2O (w/ 0.05%
TFA); sample prep: ∼0.5 mg/mL in 1:1 (v:v) CH3CN:H2O.
Vol. 9, No. 5, 2005 / Organic Process Research & Development
•
625

combining DI water (15 L) and “strong (28-30%)” aqueous
ammonia solution (8 L). The reaction mixture was added to
the aqueous ammonia solution over about 0.5 h, in a
temperature range of 20-30 °C. The resulting mixture was
agitated for about 1.0 h, and the phases were allowed to
separate for about 0.5 h. The bottom phase (pH 10) was
removed, and DI water (21 L) was charged to the organic
layer. The resulting mixture was agitated for 0.25 h, and the
phases were allowed to separate for about 0.5 h (bottom
phase removed). MgSO4 (2.1 kg) was charged to the organic
phase, and the resulting mixture agitated for about 1.0 h
between 20 and 25 °C. The MgSO4 was filtered off and
washed with CH2Cl2 (20 L). The resulting organic mixture
was concentrated via distillation (30-60 °C under 150-275
mmHg vacuum) to an end volume of about 20 L. DMF (20-
25 L) was charged to the organic mixture, and a second
vacuum distillation performed (45-75 °C under 75-150
mmHg vacuum) to a final volume of about 34 L. The DMF
solution of fluorobromo adduct 3 (36.4 kg) was collected in
a drum, and a 90% yield (based on forward processing lab
data) was assumed for the first two steps.
Step 3. A reactor was charged with a DMF solution of
fluorobromo adduct 3 (61.6 kg, 28.87 mol), and the mixture
was held at 20-25 °C. To this mixture was added potassium
phthalimide 4 (8.1 kg, 43.73 mol) followed by tetrabuty-
lammonium bromide (0.94 kg, 2.92 mol). The reaction
mixture was heated to 150-155 °C and was sampled for
area % conversion after 3.0 h (1.7 area % fluorobromo adduct
3), and sampled again after 4.0 h (0.62 area % fluorobromo
adduct 3). The reaction mixture was cooled to 38-42 °C
over about 1.0 h, and DI water (95 L) was added to the
reaction mixture at 38-42 °C over 2.0 h. The resulting
mixture was cooled to 20-25 °C over 0.5 h and stirred for
1.0 h. The thick solids produced during the water addition
required a manual scrape with a clean spatula to remove them
from the walls of the reactor for filtration. The filter cake
was washed with DI water (25 L) and dried at 45-55 °C
under vacuum to constant weight; 12.47 kg of crude
phthalimide adduct 5 was isolated as a tan solid.
2-Fluoro-2-(4-iodo-phenyl)-propylamine (S)-(-)-(2)-
Pyrrolidone-5-carboxylic Acid Salt (13). Step 1. A reactor
was charged with ethanol (50 L) followed by the crude
phthalimide adduct 5 (8.0 kg, 70.8 wt %, 5.6 kg, 13.69
mol). At 20-25 °C, a 40% aqueous solution of methylamine
(28.8 kg or 32 L, 370.90 mol) was charged to the reaction
mixture and the reactor sealed. After 24.0 h, the reaction
was sampled, and 30 area % of the bis-secondary amide 11
was present by HPLC. After a further 2.0 h, a second sample
was taken, and the bis-secondary amide 11 was present at
the 29 area % level by HPLC. Due to the lack of forward
reaction, the reaction temperature was increased to 32 °C.
After an additional 24.0 h, the reaction was sampled and
0.6 area % of the bis-secondary amide 11 was present by
HPLC (criteria <1 area %). A reduced pressure distillation
was performed (40-60 °C under 100-150 mmHg) to
remove 55 kg of distillate. MTBE (56 L) was charged to
the reactor between 15 and 30 °C, followed by 1 M HCl
(48 L). The resulting mixture was mixed for 10 min, and
the phases were allowed to separate for 35 min. The lower
aqueous phase containing amine 6 was removed, the organic
phase was extracted with DI water (32 L), and the aqueous
phases were combined. MTBE (50 L) was charged to the
aqueous phase, the resulting mixture was agitated for 17 min,
and the phases were allowed to separate for 20 min. The
aqueous bottom phase containing amine 6 was drained, and
this MTBE extraction was repeated. The aqueous phase had
a pH of 5.32, causing the desired product amine 6 to be
extracted into the MTBE. Therefore, the combined organic
phases were extracted with 1 M HCl (26 L), and the resulting
aqueous phase was combined with the initial aqueous phase
giving a new pH of 1.2. This aqueous phase was extracted
with MTBE (2 × 40 L). Fresh MTBE (48 L) was charged
to the aqueous phase, followed by 5 M NaOH (6.4 kg) over
20 min (20 °C) to adjust the pH from 1.2 to 8.5. The organic
phase was removed, and the aqueous phase was back-
extracted with MTBE (48 L). The combined organic phases
containing amine 6 were concentrated via atmospheric
pressure distillation to afford 19.68 kg of amine 6 as an
MTBE stock solution (19.31 wt %; 3.8 kg of amine 6;
86.3% yield).
Step 2. The amine 6 that was produced in the pilot plant
as a stock solution in MTBE was taken into the kilo lab. A
portion of the amine 6 MTBE stock solution was concen-
trated via distillation on a rotovapor at 30-40 °C to give
1.61 kg (5.77 mol; this mass is used for “equivalent” and
“volume” calculations) of amine 6 as an oil (stored under
vacuum for 15.25 h). In a separate flask, 12 (S)-(-)-(2)-
pyrrolidone-5-carboxylic acid (0.336 kg, 2.60 mol) was
dissolved in 95/5 v/v n-propanol/DI water (4.03 L) between
40 and 50 °C. The amine 6 was dissolved in 95/5 v/v
n-propanol/DI water (8.06 L) at 20-30 °C, and the resulting
mixture was heated to 65-70 °C. The warm (40-50 °C)
aqueous n-propanol solution of 12 was added to the warm
(69 °C) amine 6 aqueous n-propanol mixture over 3.5 h.
Initial nucleation was observed after 2.7 L of the acid solution
was added, and a large amount of solids were present after
the addition was complete. After the acid addition was
1-Iodo-4-isopropenyl-benzene (2): 1H NMR (500 MHz,
DMSO-d6), δ 2.01 (s, 3H), 5.14 (s, 1H), 5.47 (s, 1H), 7.32
(d, 2H, J ) 8.5 Hz), 7.72 (d, 2H, J ) 8.5 Hz).
1-(2-Bromo-1-fluoro-1-methyl-ethyl)-4-iodo-benzene
1
(3): H NMR (500 MHz, CDCl3) δ 1.74-1.81 (d, 3H, J )
21.27 Hz), 3.54-3.65 (m, 2H), 7.10-7.13 (d, 2H, J ) 8.5
Hz), 7.71-7.74 (d, 2H, J ) 8.5 Hz).
2-[2-Fluoro-2-(4-iodo-phenyl)-propyl]-isoindole-1,3-di-
one (5): (sample was recrystallized from IPA) mp (DSC)
(5 °C/min) onset 133.17 °C, peak 134.83 °C. IR (neat) 3471,
1
3069, 2991, 2938, 1777, 1711, 1615, 1589 cm-1. H NMR
(300 MHz, DMSO-d6) δ 1.65 (d, 3H, J ) 23.2 Hz), 3.97-
4.02 (m, 2H), 7.21 (d, 2H, J ) 8.5 Hz), 7.71 (d, 2H, J ) 8.5
Hz), 7.85-7.90 (m, 4H). 13C NMR (75 MHz, DMSO-d6) δ
24.4, 46.4, 94.3, 96.5, 123.2, 126.7, 131.3, 134.5, 137.0,
141.4, 167.4. 19F NMR (282 MHz, DMSO-d6) δ -129.2.
Anal. Calcd for C17H13FINO2: C, 49.90; H, 3.20; N, 3.42.
Found: C, 49.95; H, 3.19; N, 3.27.
626
•
Vol. 9, No. 5, 2005 / Organic Process Research & Development

complete, the reaction mixture was held at 69 °C for 1.0 h
and was subsequently cooled to 60 °C over 1.0 h. The
mixture was held at 60 °C for 1.5 h, at which point the power
to the mantle was switched off, and the reaction mixture was
allowed to cool to ambient temperature (25 °C, 17.0 h) with
the mantle in place. A sample of the solids was taken and
washed with IPA for de determination (94.2% de; criterion
>90% de). After the de criterion was met, the batch was
filtered and washed with IPA (8 L) between 20 and 30 °C.
The crystalline salt 13 was dried under vacuum between 57
and 62 °C to constant weight (0.814 kg isolated; 39.5% yield
based on acid 12 with >94% de).
(5.6 L) (agitated 5 min, separated 5 min). The organic phase
was dried over sodium sulfate (1.0 kg), and the solvent was
removed on a rotovapor at 45 °C under vacuum to give crude
8 (949 g, 93.59 area % by HPLC) as a solid. The crude 8
was dissolved in CH3CN (3.3 L) between 77 and 82 °C, and
DI water (2.6 L) was added to the mixture over 1.0 h at 78
°C. The resulting mixture (cloudy) was held at 76-78 °C
for 19 min, and cooled to 43.5 °C over 2.5 h to promote
crystallization. The mixture was cooled further to 28 °C over
0.5 h, and then to 1.8 °C over 2.0 h. The batch was filtered,
washed with DI water (3.0 L), and dried under vacuum at
55-65 °C to constant weight. The 8 (790 g, 76.5% isolated
yield, ∼99 area % and 99.44% ee by HPLC) was isolated
as a crystalline solid.
2-Fluoro-2-(4-iodo-phenyl)-propylamine (6): 1H NMR
(500 MHz, DMSO-d6), δ 1.39 (bs, 2H, NH2), 1.55 (d, 3H,
J ) 23.0 Hz), 2.84 (2H, m), 7.15 (d, 2H, J ) 8.0), 7.71 (d,
2H, J ) 8.0).
(R)-Propane-2-sulfonic acid [2-fluoro-2-(4-iodo-phen-
yl)-propyl]-amide (8): mp (DSC) (5 °C/min) onset 142.5
2-Fluoro-2-(4-iodo-phenyl)-propylamine (S)-(-)-(2)-
pyrrolidone-5-carboxylic acid salt (13): mp (DSC) (5 °C/
min) onset 184.62 °C, peak 187.30 °C. [R]22D 9.3 (c ) 1.0,
°C, peak 143.2 °C. [R]22 25.1 (c ) 1.0, MeOH). IR (neat)
D
3323, 2997, 2940, 1586 cm-1. 1H NMR (300 MHz, DMSO-
d6, 60 °C) δ 1.12 (d, 3H, J ) 2.2 Hz), 1.14 (d, 3H, J ) 2.2
Hz), 1.58 (d, 3H, J ) 23.0 Hz), 3.02 (Sep, 1H, J ) 6.8 Hz),
3.28-3.45 (m, 2H), 7.18 (d, 2H, J ) 8.0 Hz), 7.31 (t, 1H,
J ) 6.5 Hz, NH), 7.73 (d, 2H, J ) 8.0 Hz). 13C NMR (75
MHz, DMSO-d6, 60 °C) δ 16.2, 24.1, 51.4, 52.0, 94.0, 96.8,
127.0, 137.0, 141.9. 19F NMR (282 MHz, DMSO-d6, 60 °C)
δ -131.3. Anal. Calcd for C12H17FINO2S: C, 37.41; H, 4.44;
N, 3.64. Found: C, 37.41; H, 4.44; N, 3.66.
1
MeOH). IR (neat) 3220, 2186, 1688, 1649, 1544 cm-1. H
NMR (500 MHz, DMSO-d6) δ 1.57 (d, 3H, J ) 23.0 Hz),
1.91-2.1 (m, 1H), 2.04-2.18 (m, 3H), 2.22-2.32 (m, 1H),
2.87-2.98 (m, 2H), 3.99-4.02 (m, 1H), 7.16 (d, 2H, J )
8.0 Hz), 7.63 (bs, 1H), 7.72 (d, 2H, J ) 8.0 Hz). 13C NMR
(75 MHz, DMSO-d6) δ 23.9, 24.5, 28.9, 50.4, 55.0, 93.1,
97.5, 126.6, 136.7, 142.6, 174.1, 175.5. 19F NMR (282 MHz,
DMSO-d6) δ -152.0. Anal. Calcd for C9H11FIN‚C5H7NO3:
C, 41.19; H, 4.44; N, 6.86. Found: C, 41.25; H, 4.48; N,
6.79.
(R)-4′-[1-Fluoro-1-methyl-2-(propane-2-sulfonylamino)-
ethyl]-biphenyl-4-carboxylic acid (10). Under an N2 atmo-
sphere, the sulfonamide 8 (627 g, 1.63 mol) was charged to
a reactor followed by the boronic acid 9 (267.4 g, 1.61 mol).
To the mixture was added potassium carbonate (449.9 g, 3.36
mol) followed by MeOH (9.40 L) and DI water (1.88 L),
between 15 and 35 °C. Palladium black (5.20 g, 48.8 mmol)
was added to the reaction mixture, and the temperature was
increased to 63-68 °C for 1.0 h (HPLC <1 area % 9). The
reaction mixture was cooled to 50-60 °C, filtered through
Hyflo Super-Cel (430 g), and the Hyflo Super-Cel was rinsed
with warm (50-60 °C) DI water (2.5 L) to displace desired
product. The resulting clarified mixture was heated to 60-
65 °C, and acetic acid (627 mL) was added dropwise over
1.0 h. The initial portion of the acid addition caused CO2
evolution (addition controlled) as the residual potassium
carbonate/bicarbonate was consumed. As the acid addition
proceeded, the product 10 began to crystallize. After the acid
addition was complete, the mixture was held for 10 min at
65 °C and was subsequently cooled to 35 °C over 1.25 h.
The resulting mixture was cooled to 4 °C over 1.0 h and
held for 0.5 h. The batch was filtered, washed with DI water
(2 × 3.1 L) and cold MeOH (2 × 1.57 L), and dried under
vacuum at 55-65 °C to constant weight. The 10 (543.47 g,
88.0% isolated yield, ∼99 area % and 100% ee by HPLC)
was isolated as a crystalline solid.
(R)-Propane-2-sulfonic Acid [2-Fluoro-2-(4-iodo-phen-
yl)-propyl]-amide (8). A reactor was charged with MTBE
(5.5 L) followed by salt 13 (1.1 kg, 2.69 mol) between 15
and 30 °C. A 1 N NaOH solution (5.06 L) was added over
12 min at 19-20 °C to the agitated reaction mixture (stirred
for 23 min, and the phases were allowed to separate for 4
min). The aqueous phase was removed and back-extracted
with a further 5.5 L of MTBE (agitated for 5 min, allowed
to separate for 3 min). The organic phases were combined
and dried over sodium sulfate (1.0 kg) for 18 min. The
solvent was removed via distillation on a rotovapor at 35
°C under vacuum. The free amine oil weighed 747 g (2.68
mol, 99.5% yield for the free base step), which was the mass
used to calculate the reagent equivalents for the sulfonamide
preparation. The free amine was dissolved in toluene (3.75
L) and THF (3.75 L) between 10 and 30 °C. The reaction
mixture was cooled to -5-5 °C, and treated with triethy-
lamine (1.353 kg, 13.37 mol) followed by the addition of
isopropylsulfonyl chloride (670 g, 4.71 mol) over 51 min.
The reaction mixture was stirred between 5 and 3 °C for
1.5 h and was subsequently sampled for HPLC analysis
(result: 95.3 area % 8; criteria >93 area % 8). The reaction
mixture was warmed to 10 °C, treated with cold 1 N HCl
(5.6 L), and agitated for 18 min. The phases separated in 15
min, and the bottom aqueous phase was removed. The
organic phase was washed with an additional 5.6 L of cold
1 N HCl (agitated 5 min, separated 4 min), followed by a 1
N NaOH (5.6 L) wash (agitated 5 min, separated 6 min).
The organic phase was finally washed with DI water
(R)-4′-[1-Fluoro-1-methyl-2-(propane-2-sulfonylamino)-
ethyl]-biphenyl-4-carboxylic acid (10): mp (DSC) (5 °C/
min) onset 183.51 °C, peak 185.00 °C. [R]22D 30.2 (c ) 1.0,
1
MeOH). IR (neat) 3202, 2989, 1700, 1607 cm-1. H NMR
(300 MHz, DMSO-d6) δ 1.13 (d, 3H, J ) 2.2 Hz), 1.16 (d,
3H, J ) 2.2 Hz), 1.65 (d, 3H, J ) 23.0 Hz), 3.03 (Sep, 1H,
Vol. 9, No. 5, 2005 / Organic Process Research & Development
•
627

J ) 6.8 Hz), 3.33-3.48 (2H, m), 7.36 (t, 1H, NH, J ) 6.4
Hz), 7.50 (d, 2H, J ) 8.2 Hz), 7.75 (d, 2H, J ) 8.2 Hz),
7.80 (d, 2H, J ) 8.2 Hz), 8.00 (d, 2H, J ) 8.2 Hz), 12.98
(s, 1H). 13C NMR (75 MHz, DMSO-d6) δ 16.2, 24.1, 51.5,
52.0, 96.8, 125.2, 126.8, 129.8, 130.0, 138.4, 142.1, 143.7,
167.1. 19F NMR (282 MHz, DMSO-d6) δ -130.8. Anal.
Calcd for C19H22FNO4S: C, 60.14; H, 5.84; N, 3.69; S, 8.45.
Found: C, 60.30; H, 5.86; N, 3.69; S, 8.45.
resulting product was transferred to a reactor for an EtOH
re-slurry. Filtered EtOH (13 L) was added to the LY503430
wet cake and the temperature was increased from 14 to 75
°C over 45 min, and held for 1.25 h. The heating mantle
was removed (75 °C), and the mixture was allowed to cool
to room temperature and stir overnight (24 °C). The product
LY503430 was filtered between 20 and 30 ° and washed
with filtered EtOH (3.35 L). The wet cake was dried under
vacuum at 60 °C, until constant weight. The isolated
LY503430 was ground to remove lumps with a mortar and
pestle, affording 1.135 kg (84.4% yield, 99.82 wt % purity
and 100% ee by HPLC) of colorless crystalline solid. Mp
(R)-4′-[1-Fluoro-1-methyl-2-(propane-2-sulfonylamino)-
ethyl]-biphenyl-4-carboxylic acid methylamide (LY503430).
Under an N2 atmosphere, the carboxylic acid 10 (1.30 kg,
3.43 mol) was charged to a reactor followed by CDI (1,1′-
carbonyldiimidazole) (611.1 g, 3.77 mol). To the mixture
was added EtOAc (13 L), and the resulting slurry was stirred
at about 22 °C for 1.5 h. After the mixture had stirred for
1.0 h, it had become homogeneous (imidazolide intermedi-
ate), and a sample was taken for conversion check by
preparing a derivative (2 drops of reaction quenched into 2
mL of 2 M methylamine in THF) (HPLC < 1 area % 10).
The reaction mixture was warmed to 40-50 °C to maintain
homogeneity and polish filtered through an inline stainless
steel Swagelock filter with a new 15 µm filter element into
a second reactor. Two molar methylamine in THF (2.06 L,
4.11 mol) was added subsurface (to avoid carbonate forma-
tion) through the Swagelock filter into the reaction mixture
between 40 and 50 °C over 19 min. The product LY503430
began to precipitate during the methylamine addition. After
the methylamine addition, the resulting slurry was stirred
for 1.16 h with the temperature reducing from 48 to 23 °C.
The mixture was cooled to 5 °C over 0.5 h and stirred for
1.5 h with an ending temperature of 0 °C. The product was
filtered and washed with 0-5 °C EtOAc (6.5 L). House
vacuum was pulled on the filter cake for 1.0 h, and the
(DSC) (5 °C/min) onset 186.42 °C, peak 187.06 °C. [R]22
D
31.0 (c ) 1.0, MeOH). IR (neat) 3321, 3286, 2994, 2944,
2882, 1634, 1545, 1451 cm-1. 1H NMR (300 MHz, DMSO-
d6) δ 1.13 (d, 3H, J ) 2.2 Hz), 1.15 (d, 3H, J ) 2.2 Hz),
1.66 (d, 3H, J ) 22.7 Hz), 2.80 (d, 3H, J ) 4.5 Hz), 3.04
(Sep, 1H, J ) 6.8 Hz), 3.38-3.54 (m, 2H), 7.37 (t, 1H, J )
6.4 Hz, NH), 7.49 (d, 2H, J ) 8.3 Hz), 7.74 (d, 2H, J ) 5.0
Hz), 7.82 (d, 2H, J ) 5.0 Hz), 7.92 (d, 2H, J ) 8.3 Hz),
8.48 (q, 1H, J ) 4.5 Hz). 13C NMR (75 MHz, DMSO-d6) δ
16.2, 24.1, 26.3, 51.7, 52.0, 97.2, 125.2, 126.5, 127.1, 127.6,
133.4, 138.5, 141.9, 166.2. 19F NMR (282 MHz, DMSO-d6)
δ -130.8. Anal. Calcd for C20H25FN2O3S: C, 61.20; H, 6.42;
N, 7.13; S, 8.16. Found: C, 61.11; H, 6.38; N, 7.07; S, 8.11.
Supporting Information Available
X-ray data. This material is available free of charge via
the Internet at http://pubs.acs.org.
Received for review May 16, 2005.
OP0500741
628
•
Vol. 9, No. 5, 2005 / Organic Process Research & Development