The synthesis of the high-potency sweetener, NC-00637. Part 1: The synthesis of (S)-2-methylhexanoic acid

Article abstract of DOI:10.1021/op025616i

The synthesis of the high potency sweetener candidate NC-00637 (1) required large quantities of (S)-2-methylhexanoic acid (2). This acid was first prepared in small quantities by the use of chiral auxiliaries. For large quantities, resolution by classical means and an enzymatic method were investigated. Asymmetric hydrogenation provided a workable solution.

Full text of DOI:10.1021/op025616i

Organic Process Research & Development 2003, 7, 369−378
The Synthesis of the High-Potency Sweetener, NC-00637. Part 1: The
Synthesis of (S)-2-Methylhexanoic Acid
David J. Ager,*,† Scott Babler, Diane E. Froen, Scott A. Laneman, David P. Pantaleone, Indra Prakash,‡ and Ben Zhi
NutraSweet R&D, 601 East Kensington Road, Mount Prospect Illinois 60056, U.S.A.
Scheme 1
Abstract:
The synthesis of the high potency sweetener candidate NC-00637
(1) required large quantities of (S)-2-methylhexanoic acid (2).
This acid was first prepared in small quantities by the use of
chiral auxiliaries. For large quantities, resolution by classical
means and an enzymatic method were investigated. Asymmetric
hydrogenation provided a workable solution.
Introduction
Over the past decades, the human desire for sweet, while
tempered by the trend for reduced calories, has seen
increasing sales of high-potency, low-calorie sweeteners such
as saccharin, aspartame, and acesulfam-K.1 As part of a
continuing effort to identify high-potency sweeteners with
an excellent taste profile and increased stability compared
to aspartame, the molecule NC-00637 (1) was identified as
a potential candidate.2,3 Before use as a sweetening agent,
any candidate compound must undergo clinical studies to
show that it is safe for human use and that there are no
adverse biological effects; as a consequence, large quantities
of compound are needed for these tests. As food additive
studies require extensive animal studiessthe only desired
effect on humans in the case of NC-00637 (1) being sweet
tastesa large-scale synthesis has to be “locked in” at a very
early stage in development. In this way impurity profiles
can be controlled. However, the short time frame for route
development means that many potential routes and solutions
cannot be investigated.
Inspection of the structure of NC-00637 reveals two
obvious disconnections, the two amide bonds (Scheme 1).4
The molecule can then be assembled in a convergent manner
from the three components, (S)-2-methylhexanoic acid (2),
L-glutamic acid (3), and the pyridinylamine 4. Each of these
components, or their subsequent coupling reactions, provided
a synthetic challenge. This contribution discusses the methods
used to prepare the deceptively simple acid 2.
synthesis. In addition to an asymmetric hydrogenation, a
resolution approach using salt formation and enzymatic
resolutions were investigated. However, there was an im-
mediate need for small quantities of the target molecule, and
we first investigated the use of chiral auxiliaries for the
preparation of the acid 2. This material was needed to
determine which isomer of 2 was in the sweetener candidate
as there was some ambiguity.4
Preparation by Alkylation Methods
Although (S)-2-methylhexanoic acid (2) has a relatively
simple constitution, the presence of a single asymmetric
centre has made this the key contributor to the synthetic
challenge. A number of methods do exist in the literature
for the preparation of carboxylic acids with an asymmetric
centre at the R-carbon atom (vide infra) including alkylation
methods. It was not envisioned that these methods would
provide a large-scale economical process, but they would
solve the short-term need for a few grams of material. In
addition, at this early stage, the enantiomeric excess (ee)
required for the 2-methylhexanoic acid (2) that would lead
to NC-00637 of acceptable diastereoisomeric purity was not
known. We, therefore, strived in our initial studies to obtain
as high an ee as possible. Subsequent studies showed that
an ee of >90% was acceptable as in-process purifications
removed the diastereoisomer derived from (R)-2-methyl-
hexanoic acid.4
At the time this work commenced, there were no large-
scale methods to access the hexanoic acid 2 by asymmetric
† Current address: RCCorp, 805 Darfield Drive, Raleigh, NC 27615.
E-mail: ScribusTed@aol.com.
Oxazolidinone Approach. The first approach was based
on the use of an oxazolidinone as an auxiliary.5 In this study
only the oxazolidinone 5 derived from valine was tried, and
the methodology followed that reported by Evans for
analogous compounds.6 Thus, the oxazolidinone 5 was
acylated with hexanoyl chloride after deprotonation with
‡ Current address: NutraSweet, 1801 Maple Ave., Evanston, IL 60201.
(1) Ager, D. J.; Pantaleone, D. P.; Henderson, S. A.;. Katritzky, A. R.; Prakash,
I.; Walters, D. E. Angew. Chem. 1998, 110, 1901; Angew Chem., Int. Ed.
1998, 37, 1803.
(2) DuBois, G. E. Sweeteners, Nonnutritive. In Encyclopedia of Food Science
and Technology; Hui, Y. H., Ed.; John Wiley: New York, 1991; Vol. 4,
pp 2470-2487.
(3) Nofre, C.; Tinti, J.-M. U.S. Patent 5,196,540, 1993.
(4) The chemistry for other components of NC-00637 and the complete
synthesis will be published soon.
(5) Ager, D. J.; Prakash, I.; Schaad, D. R. Chem. ReV. 1996, 96, 835.
(6) Evans, D. A.; Bartroli, J.; Shih, T. L. J. Am. Chem. Soc. 1981, 103, 2127.
10.1021/op025616i CCC: $25.00 © 2003 American Chemical Society
Published on Web 04/18/2003
Vol. 7, No. 3, 2003 / Organic Process Research & Development
•
369

Scheme 2
Scheme 3
Scheme 4
n-butyllithium to give 7 (Scheme 2). The required acid 2
was produced in 80% ee. In this sequence, none of the
intermediates was purified extensively, as the goal was to
prepare the acid 2.
Oxazoline Method. As a comparison, the oxazoline
auxiliary developed by Meyers was used in a similar
alkylation approach (Scheme 3).7 The oxazoline 9 was
methylated to provide the ethyl analogue 10. A second
alkylation then gave the dialkylated derivative 11 that was
then hydrolyzed to the product 2. By this sequence the
desired acid was formed with an ee of 62%. A variant
employing a camphor derivative has been reported to give
higher ee’s, but this was not attempted.8
Oppolzer’s Sultam Auxiliary. The reactions used were
standard for alkylations with Oppolzer’s sultam as the chiral
auxiliary (Scheme 4).5 The sultam 12 was acylated by
reaction with sodium hydride followed by hexanoyl chloride
to give the N-acyl derivative 13 in 88% yield.9 Deprotonation
with n-butyllithium and subsequent reaction of the enolate
with methyl iodide in the presence of HMPA gave the
alkylated product 14 in about 75% yield. Hydrolysis of 14
gave (R)-2-methylhexanoic acid (ent-2) with an ee of 90%.10
The ee of ent-2 was lower if the acylation was performed
with propionyl chloride and the alkylation with n-butyl
iodide; the yield of the alkylation step was also significantly
lower. It has been reported that sodium hexamethyldisilazide
can be used as the base in the alkylation step and this avoids
the use of HMPA,9 but this was not attempted.
Although this method gave material of acceptable enan-
tiomeric purity that was used to prepare small quantities of
1, it was not going to be a viable for a large-scale synthesis.
At the time this work was performed, we could only find 5
kg of the required sultam (ent-12), and the price made its
use prohibitive without multiple recycles. Rather than embark
on a programme to prepare the sultam, our efforts were
concentrated on the resolution and catalytic methods.
Other Approaches. A number of other approaches were
considered for the preparation of (S)-2-methylhexanoic acid
(2). In some cases, as the target molecule 2 is relatively
simple, there are many variations on a general theme. The
use of other auxiliaries in an alkylation approach such as
the use of pseudoephedrine11 was not tried, as the short-
comings outlined above would not be overcome. Other
approaches with chiral auxiliaries such as the use of
menthone in an SN2′ cuprate displacement on a chiral
carbonate12 were not pursued due to time and resource
constraints. The methodology used had to be “tried-and-
tested” and familiar to us.
Reduction of 2-methylhexenoic acid (15), or an ester of
this acid, by hydroboration was investigated. A conjugate
(11) Myers, A. G.; Yang, B. H.; Chen, H.; McKinstry, L.; Kopecky, D. J.;
Gleason, J. L. J. Am. Chem. Soc. 1997, 119, 6496; Myers, A. G.; Yang, B.
H.; Chen. H.; Gleason, J. L. J. Am. Chem. Soc. 1994, 116, 9361.
(12) Spino, C.; Beaulieu, C.; Lafreniere, J. J. Org. Chem. 2000, 65, 7091;
compare: Tseng, C. C.; Yen, S. J.; Goering, H. L. J. Org. Chem. 1986, 51,
2884; Tseng, C. C.; Goering, H. L. J. Org. Chem. 1983, 48, 3986.
(7) Meyers, A. I. Acc. Chem. Res. 1978, 11, 3756 and references therein.
(8) Chandrasekhar, S.; Kauser, A. Tetrahedron: Asymmetry 2000, 11, 2249.
(9) Oppolzer, W.; Moretti, R.; Thomi, S. Tetrahedron Lett. 1989, 30, 5603.
(10) Oppolzer, W.; Marco-Contelles, J. HelV. Chim. Acta 1986, 69, 1699.
370
•
Vol. 7, No. 3, 2003 / Organic Process Research & Development

Scheme 5
In our hands, the salt only had to be recrystallised twice
from acetone to provide the acid 2 with an ee of 92%. In
our case, GC analysis of the acid 2 by derivatisation was
used to monitor the reaction rather than optical rotation that
may account for the experimental differences, as our analyses
were not susceptible to contamination by small amounts of
quinine. The subsequent reactions to NC-00637 involve the
formation of diastereoisomers, and the small amount of
R-isomer of 2 was easily removed.4
This resolution method was the first to be successful at
scale, and as it was simple, the method was used to make
over 100 kg of the chiral acid 2. It was reproducible at the
larger scale and consistently gave material of the same purity.
Scale-up was just a multiplication of amounts from the
laboratory procedure. The yield of material after all manipu-
lations was about 60% of theoretical. However, the time
involved and volumes of solvents coupled with the need to
move the process between plants indicated that recycles
would be difficult. In addition, some salts were carried over
to the distillation step and this made the pot residues very
viscous and difficult to clean up. As a result, we continued
to look for alternative approaches. However, our findings
from this study did allow us to use the method to clean up
the optical purity of material obtained by other methods, such
as asymmetric hydrogenation.
Enzymatic Approaches. Considerable time and effort
was spent looking at enzymatic approaches to 2. As well as
methods to obtain the acid directly, approaches that looked
at the use of just one isomer of racemic 2-methylhexanoic
acid in the coupling reactions were also investigated.4 The
desired acid 2 has been prepared by enzymatic hydrolysis
of ethyl 2-methylhexanoate with PS-30.15 The conversion
was found to be good (45%), but the ee was only 74%. The
ee could be increased by lower conversions. Obviously, the
undesired isomer is hydrolyzed at a rate that has an affect
on the product ee with longer reaction times. This was not
considered to be a workable process due to the need for large
recycle loops and the problems associated with isolation of
the product from dilute aqueous solution.
A number of enzymes were screened with various esters
of racemic 2-methylhexanoic acid. PPL, PFL, PLE, and other
common lipases and esterases resulted in hydrolysis to
provide racemic or up to 30% ee of 2-methylhexanoic acid
when small esters were used, such as ethyl or methyl.16 The
iso-amyl esters generally gave better selectivity. The Amano
PS30 lipase gave an S:R ratio of 91:9 when the iso-amyl
ester was hydrolyzed at pH 9. The reaction was slow. A study
that varied the reaction conditions resulted in an increase to
85% ee for the formation of 2 from the methyl ester. The
enzyme could be recycled through a number of runs without
loss of activity. However, the initial cost of the enzyme still
did not provide a cost-effective method when compared to
that for the quinine resolution or asymmetric hydrogenation.
It was found that PS-800 performed much better than PS-
30 to give a 94% ee at 55% conversion (of the S-isomer)
reduction was required and it was thought that the boron
adduct could be reduced to the desired compound. Use of a
chiral ligand on boron, as in IpcBH2 could provide the
asymmetric induction. All our attempts at hydroboration of
15, even in the achiral series, were unsuccessful.
To obtain samples of the acid 2 and its enantiomer, and
then prepare the intermediates, isomers, and other compounds
needed to develop NC-00637 through the clinical phases, a
chromatographic separation of racemic 2-methylhexanoic
acid was undertaken as an external collaboration. The method
was a direct scale-up of the analytical procedure described
in the Experimental Section. The racemic 2-methylhexanoic
acid was converted to the acid chloride and then condensed
with the amino alcohol derived from D-phenylglycine.13
Chromatographic separation of the diastereoisomers followed
by hydrolysis gave material with >98% ee, but the route
was tedious as the peak separation was not good. After this,
the catalytic hydrogenation method outlined below was used.
Resolution
The resolution of 2-methylhexanoic acid has been reported
in the literature by formation of the quinine salt (Scheme
5).14 The process involved eight recrystallisations of the
2-methylhexanoic acid-quinine salt, and the yield was not
reported. The racemate was commercially available (from
Lancaster Synthesis) and could be produced at scale. A
classical resolution of this type was investigated as it could
provide kilogram quantities of material with a relatively
cheap resolving agent being employed. It was envisioned
that the quinine could be isolated by extraction and recycled
if necessary. In addition, the R-isomer of the acid could be
racemised by a variety of methods and then put back through
the process. If the amount of quinine was reduced, the
amount of salt that crystallized from the mixture was also
reduced. As quinine gave us a cheap, workable method, no
in-depth study was performed to see if other bases gave better
performance.
In our initial, laboratory studies the first crop of crystals
(55% based on the total amount of acid and base used) had
an R:S content of 26:74. Crystallisation of this mixture
resulted in a 75% recovery and an increase in the isomer
ratio to 14:86. A further recrystallisation gave 86% recovery
and an R:S ratio of 9:91. Further crystallizations did increase
the S-content by 1-2%, but the losses were about 10 wt %.
We, therefore, went with two recrystallisations and adjusted
the amount of solvent to minimize losses. This also resulted
in a slight increase in ee.
(13) Compare: Karl, V.; Kaunzinger, A.; Gutser, J.; Steuer, P.; Angles-Angel,
J.; Mosandl, A. Chirality 1994, 6, 420.
(14) Levene, P. A.; Bass, L. J. Biol. Chem. 1926, 70, 211.
(15) Engel, K.-H. Tetrahedron: Asymmetry 1991, 2, 165.
(16) Compare: Ozaki, E.; Uragaki, T. Jpn. Patent 09065891; Chem. Abs. 1997,
127, 276440.
Vol. 7, No. 3, 2003 / Organic Process Research & Development
•
371

Scheme 6
with an 8-fold decrease in the amount of enzyme needed.
The reaction still took over 3 days.
As an alternative to hydrolysis of an ester, ester formation
was also considered; again this is known in the literature
for the esterification of racemic 2-methylhexanoic acid using
Candida cylindracea lipase (CCL)11 and by the transesteri-
fication of racemic octyl 2-methylhexanoate with oleic acid
and CCL.17 In esterification or transesterification reactions,
the conversions were poor (8-15%), but better R:S ratios
(>10:90) are obtained. For the same reasons as outlined for
the hydrolysis reactions, these approaches were not consid-
ered further.
In both the hydrolysis and ester formation approaches, a
cosolvent such as DMSO had the advantage to ensure contact
of reagents with the enzymes. Use of this cosolvent
complicated the isolation procedure. In the hydrolysis
reactions, it was found that filtration of the enzyme solution
prior to the first run removed much of the material that gave
rise to emulsions during the isolation steps. Emulsions were
still a major problem and with better approaches available,
the enzymatic methods were dropped.
Hydrogenation Approach. At the commencement of this
work in the early 1990s, there were relatively few asymmetric
hydrogenation catalysts although a number of reports had
appeared in the literature describing the reductions of R,â-
unsaturated carboxylic acids to the corresponding saturated
acids with high enantiomeric excesses.18 In particular,
ruthenium-BINAP complexes have proven expeditious for
this transformation.19
The two available to us, through Monsanto, were Knowles’
catalyst20 and a ruthenium-BINAP system that was under
development for the synthesis of naproxen.21-23 With our
eyes on a robust process, 2-methyl-2-hexenoic acid (15) was
chosen as the only option for substrate. The other possibility
was to use an exo-methylene analogue, but this was
potentially problematic due to isomerisation of the unsat-
uration into the more substituted position. In addition, the
acid 15 is available as the E-isomer as it is used in the flavour
industry.
genation was carried out in a mixture of methylene chloride,
methanol, and water at varying amounts. The catalyst and
substrate concentrations were also varied. In all cases, the
reaction was asymmetric with ee’s of g80%.
Although the ee was not high, as noted above, the
selectivity was workable as downstream processing allowed
for purification as diastereoisomers are formed.
The reaction was scaled up to >50-kg runs, and at scale,
the ability to exclude oxygen was increased. This allowed
the amount of catalyst to be decreased significantly. Com-
pared to catalyst usage in the laboratory runs, usage was
reduced by about 100-fold at scale. In addition, the amount
of solvent could also be reduced so that almost a third of
the reactor content was substrate. Degassing of the substrate
in solution was relatively straightforward. However, the
sequence involved multiple purges, evacuations, and gas-
filling steps. As oxygen was present especially at the start
of the sequence, the catalyst did not survive if added to the
substrate prior to oxygen removal. A method was therefore
required to introduce the catalyst after the solution had been
purged. There are engineering solutions, but these were not
considered to be general so that the methodology can be run
in a number of plants and the method can be readily
transferred. This eliminated such solutions to the problem
as the use of a basket in the reactor. Our solution was to use
small cans that are similar to those used for dispensing liquids
for high-pressure chromatography or carbonated soda (Figure
1). This approach relies upon the observation that the solid
ruthenium catalyst is air-stable in the solid state.
The Ru(BINAP) catalyst (16) was placed as a solid in a
can linked to the reactor. The headspace could be deoxy-
genated through the reactor or in a separate sequence. In
another can was placed methylene chloride that was also
deoxygenated. Once the reactor, lines, catalyst container, and
methylene chloride were deoxygenated, the methylene
chloride was pushed under nitrogen pressure into the
container with the catalyst. The mixture was then agitateds
this was usually done by a magnetic stirrer bar that worked
through the can even on gallon scale. Laboratory experiments
had shown that dissolution was rapid in methylene chloride.
After about 0.25 h, the catalyst solution was again pushed
by nitrogen pressure into the reactor, which was then put
under hydrogen. Because the reaction does not start until
being heated, the agitation during the heat-up period was
sufficient to allow dispersion and dissolution.
Knowles’ catalyst, [Rh(COD)DIPAMP]BF4, did reduce
the R,â-unsaturated acid 15 under a variety of conditions,
but in all cases racemic 2-methylhexanoic acid resulted.
The ruthenium catalyst of the type [Ru(S-BINAP)XY]n
(where X and Y are either organic or halogen ligands [Our
catalyst has X ) Y ) Cl.]) does provide for useful
asymmetric induction to provide 2 (Scheme 6).22,23 In our
case the catalyst was [Ru(S-BINAP)Cl2]n (16). The hydro-
(17) Engel, K.-H. J. Am. Oil Chem. Soc. 1992, 69, 146.
(18) Asymmetric Synthesis; Morrison, J. D., Ed.; Academic Press: Orlando, 1985;
Vol. 5.
(19) Mashima, K.; Kusano, K.; Ohta, T.; Noyori, R.; Takaya, H. J. Chem. Soc.,
Chem. Commun. 1989, 1208; Noyori, R.; Takaya, H. Acc. Chem. Res. 1990,
23, 345; Noyori, R. Science 1990, 248, 1194; Takaya, H.; Ohta, T.;
Mashima, K.; Noyori, R. Pure Appl. 1990, 62, 1135; Noyori, R. Chem.
Soc. ReV. 1989, 187; Uemura, T.; Zhang, X.; Matsumura, K.; Sayo, N.;
Kumobayashi, H.; Ohta, T.; Nozaki, K.; Takaya, H. J. Org. Chem. 1996,
61, 5510; Zhang, X.; Uemura, T.; Matsumura, K.; Sayo, N.; Kumobayashi,
H.; Tayaya, H. Synlett 1994, 501.
Some Observations on the Reaction and Mechanism
The 2-methylhexenoic acid (15) used for these reductions
was exclusively the E-isomer (>99%). This material is used
in the flavour industry and is prepared by a Wittig reaction.
We spent a little time looking at the preparation of 15 and
found that a Claisen condensation gave variable isomeric
ratios. A Knoevenagel approach gave low yields (<50%)
due to self-condensation of the aldehyde. A Perkin reaction
(20) Knowles, W. S. Acc. Chem. Res. 1983, 16, 106.
(21) Chan, A. S. C.; Laneman, S. A. U.S. Patent 5,198,561, 1993.
(22) Chan, A. S. C.; Laneman, S. A. U.S. Patent 5,202,473, 1993.
(23) Chan, A. S. C. CHEMTECH 1993, 23, 46.
372
•
Vol. 7, No. 3, 2003 / Organic Process Research & Development

Figure 1. Reactor configuration for hydrogenations.
Figure 2. Mechanistic scheme for the [Ru(BINAP)(O2CR)2]-catalyzed hydrogenation of tiglic acid.
between butyraldehyde and propionic anhydride in the
presence of pyridine as base gave almost exclusively the
desired E-isomer of 15, but the yield (∼60%) was lower than
that from a Wittig-type approach, and purification was
tedious and required multiple distillations. The Wittig
approach was the low-cost route. Our supplier, Bedoukian
Research, used this approach.
Studies on the reduction of tiglic acid with Ru(BINAP)-
(OAc)2 have shown that the hydrogen addition to the
carbon-carbon double bond is not solely derived from a
hydrogen molecule; the â-hydrogen comes from gaseous
hydrogen while the proton is acquired from the solvent.26
This was confirmed to be correct in our case by deuterium-
labeling experiments with deuterium gas and deuterium
oxide.
the basis of the kinetics of these catalytic hydrogenations.27
Displacement of the ligand on the catalyst by the substrate,
in this case, tiglic acid, is followed by reduction of the double
bond and the dissociation of the product from the catalyst
(Figure 2). A number of assumptions were used to derive
the rate law from this catalytic cycle: The carboxylate
substitution equilibrium step is rapid, subsequent addition
of H2 is turnover-limiting, and concentration of the catalyst
is low compared to that of the substrate.
A study was made of the effects of a number of variables
including temperature, concentration, addition of base, and
solvents for the hexenoic acid derivative at laboratory scale.
There was no observable effect from the addition of base,
unlike the case with Noyori-type catalysts or in the synthesis
of naproxen.23 Pressures lower than those needed to obtain
high ee with naproxen could be used with the same catalyst
system and the enoic acid. The catalyst concentration was
not completely optimized but was derived from experiments
where the amount was halvedsall other conditions were kept
constantsuntil the reaction time became intolerable. These
data were used as the starting point for the larger-scale
The rate was reported to be first-order with respect to
hydrogen uptake.27 A mechanism for Ru(BINAP)(O2CR)2-
catalyzed reduction of an enoic acid has been postulated on
(24) Chan, A.; Laneman, S. U.S. Patent 5,144,050, 1992.
(25) Chan, A. U.S. Patent 4,994,607, 1991.
(26) Ohta, T.; Takaya, H.; Noyori, R. Tetrahedron Lett. 1990, 31, 7189.
(27) Ashby, M. T.; Halpern, J. J. Am. Chem. Soc. 1991, 113, 589.
Vol. 7, No. 3, 2003 / Organic Process Research & Development
•
373

Figure 4. Computer printout of % theoretical hydrogenation
uptake.
Figure 3. % Theoretical hydrogenation uptake vs time for
the hydrogenation of 2-methyl-2-hexenoic acid with [Ru(S-
BINAP)Cl2]n.
reductions, and as stated above, the amount of catalyst could
be significantly decreased. Again, this was achieved by
halving the amount of catalyst used in a run until hydrogen
uptake did not occur after a few minutes at 80 °C. If this
failure occurred, addition of more catalyst then allowed the
reduction to proceed.
The theoretical percentage of hydrogen uptake was
monitored over numerous runs and the curve is quite unusual.
It has an inflection point at ∼50% theoretical hydrogen
(Figure 3). All of the asymmetric hydrogenations show a
similar phenomenon, complete with inflection point. The
variability was often due to deoxygenation or other variables
that could not be quantified, as under identical conditions,
significant variations were observed in laboratory runs.
Figure 3 shows three runs under identical experimental
conditions including the same lots of catalyst, substrate and
solvents. At larger scale, this variation was considerably less
noticeable. The inflection point does not differ significantly
from 50% even when the catalyst loading is varied. The
hydrogen uptake begins immediately upon addition of
catalyst in Halpern’s tiglic acid case.27 In our system, there
is a slight lag time even after the reaction solution reaches
the 80 °C reaction temperature. Figure 4 shows a computer
printout of a monitored hydrogenation.
Our results do not fit Halpern’s proposed rate equation,
which is first-order in alkene and hydrogen. Figure 5 shows
representative plots of experimental hydrogen uptake derived
from Figure 3 with an additional run, again, under identical
conditions. From this a plot of hydrogen concentration, as
ln[(H2 uptake)∞ - (H2 uptake)t] vs time should give a straight
line if Halpern’s hypothesis holds for our system.27
Reactions were stopped at various stages, and the mixtures
were analyzed by GC. The hydrogen uptake was found to
Figure 5. Plot of H2 uptake vs time for the Ru(BINAP)Cl2-
catalyzed hydrogenation of 2-methyl-2-hexenoic acid (15).
correspond to the amount the reaction had proceeded when
compared to the amounts of 15, 2, and ent-2 present. In
addition, the amount of asymmetric induction was found to
be constant throughout the reaction.
The hydrogenation was also monitored by FT infrared
spectroscopy in a modified Fischer-Porter apparatus. Thus,
the hydrogen uptake could also be monitored during the
reaction. The IR spectra of the substrate and the product are
sufficiently different in the carbonyl region and at 1200-
1100 cm-1 to allow the reaction to be followed. The plot of
percent theoretical hydrogen uptake showed the usual inflec-
tion point at about 50% of theory. The superimposed spectra
(Figure 6) in the carbonyl region show that the substrate
disappears during the reaction without any significant
changes in rate, while the product builds from the zero time
point throughout the reaction.
The data from the IR study and hydrogen uptake showed
the same kinetics. This gave us the confidence that either
374
•
Vol. 7, No. 3, 2003 / Organic Process Research & Development

Figure 6. Superimposed infrared spectra.
method could be used to monitor the reaction. Why our
system that uses the catalyst 16 differs from the findings of
Halpern with tiglic acid could be due to a number of reasons
and factors: In the tiglic acid system, reduction was observed
as soon as hydrogen pressure was applied, whereas our
system required heating to 80 °C. The catalyst precursor 16
is polymeric in nature, unlike that used in Halpern’s studies,
and some dissociation to an active species may have to occur.
Thus, as the reaction progresses, more and more catalyst is
entering the catalytic cycle. For the reduction of the hexenoic
acid 15, the concentration of substrate is much higher than
for the tiglic acid study, and thus, some factors that were
not important for tiglic acid may become noticeable.
Certainly oxygen inhibition seems to play an important role
as the amount of catalyst needed at larger scale was 2 orders
of magnitude less than that for laboratory-scale runs. In
addition, it was found that 25 purges were more than
sufficient to ensure removal of oxygen.
Obviously, more studies could be performed with the
[Ru(BINAP)Cl2]n catalyst system to determine why there is
run-to-run variability as well as to unravel the mechanism,
but time did not allow this.
As studies were continuing towards the target molecule,
NC-00637, while these reactions were being performed, it
was found that a 90% ee for 2 was acceptable because the
diastereoisomers resulting from reaction of ent-2 were simple
to remove in subsequent steps.4
quinoline and forms a salt with the R-isomer ent-2. However,
this method was not used at scale due to the problems with
the residue as seen during the quinine resolution approach
(vide supra).
One interesting observation was that in one 22-kg run,
after distillation of 2, it turned blue on exposure to air.
Although metal ions could not be detected in the material,
redistillation gave a colourless, air-stable product.
Summary
A large-scale method was required to obtain (S)-2-
methylhexanoic acid (2), a component of the sweetener
candidate, NC-00637 (1). The use of chiral auxiliaries
provided the small quantities necessary for preliminary
biological testing. For larger amounts of the acid 2, a classical
resolution approach using quinine was found to be workable.
Although a large number of methods were explored for
enzymatic resolution, none were found to be cost-effective
compared to alternatives. Asymmetric hydrogenation of the
enoic acid 15 did provide a direct asymmetric approach to 2
although only 90% ee could be attained. This was not a
problem as subsequent reactions to 1 removed the undesired
isomer.
Experimental Section
The melting points are uncorrected. IR spectra (Nujol)
1
were recorded on a Nicolet FT-IR spectrometer. H NMR
It was found that the addition of quinine to the acid 2
prior to the final distillation did not result in an increase in
ee of the final product, but if quinidine (∼10 mol %) were
added, then ee’s of 95-98% could be achieved. Presumably
this is because the quinidine is a pseudoenantiomer of
spectra were recorded on a GE 300 spectrometer in CDCl3
using TMS as internal standard. Optical rotation was
measured on a Perkin-Elmer 241 Polarimeter.
Laboratory hydrogenations were performed in mechani-
cally stirred Parr reactors with external heating. In some
Vol. 7, No. 3, 2003 / Organic Process Research & Development
•
375

cases, Fischer-Porter bottles fitted with appropriate inlet and
outlet tubes and valves were used for screenings; external
heat was applied through an oil batch, and stirring was
magnetic. The systems included pressure sensors that were
computer-monitored to allow for the reactions to be followed
by hydrogen uptake.
1 M NaN(SiMe3)2 in THF (21.1 mL, 21.1 mmol) was added
slowly. After 30 min, MeI (13.6 g, 96 mmol) was added,
and the solution was warmed to -20 °C. At this time,
saturated aqueous NH4Cl was added, and the mixture was
extracted with ethyl acetate (3 × 50 mL). The extracts were
washed with brine and dried (MgSO4). Evaporation of the
solvents under vacuum gave 4.05 g (91%) of crude product
8.
The compound 8 from the previous step was dissolved
in MeOH (34.5 mL) and stirred at 0 °C. Then 2 N KOH
(34.5 mL)was added and the mixture stirred for 45 min. The
mixture was then extracted with ethyl acetate (2 × 25 mL).
The aqueous phase was acidified to pH 2 and extracted with
ethyl acetate (3 × 30 mL). The combined extracts were dried
(Na2SO4) and evaporated to give the product 2 liquid with
an ee of 80%.
Oxazoline Route. To a solution of diisopropylamine (2.9
mL, 21 mmol) in dry THF (54 mL) at -78 °C was added
1.6 M n-butyllithium in hexane (13.1 mL, 21 mmol). After
20 min, a solution of the oxazoline 9 (4.1 g, 20 mmol) in
THF (19 mL) was added portionwise over 15 min. After
stirring for 45 min, MeI (1.5 mL, 24 mmol) in THF (10 mL)
was added slowly. After an additional 2 h, the mixture was
warmed to -50 °C and poured into ice-water (250 mL).
The mixture was extracted with ether (3 × 40 mL). The
combined extracts were washed with brine and dried
(MgSO4). After removal of the solvents under reduced
pressure 4.04 g of product 10 was obtained (92%).
To a solution of diisopropylamine (2.7 mL, 19.2 mmol)
in THF (20 mL) at 0 °C was added 1.6 M n-butyllithium
(12 mL, 19.2 mmol). After 10 min, this solution was added
dropwise to a solution of 10 in THF (50 mL) at -78 °C.
After 30 min, n-butyl iodide (2.5 mL, 21.9 mmol) was added
dropwise and stirred for 2 h. After warming to room
temperature, the mixture was poured into ice-water (100
mL), which was extracted with ether (2 × 120 mL). The
combined extracts were dried (MgSO4). The crude product
was purified by chromatography (SiO2 eluting with 20%
EtOAc/hexanes). About 2.7 g of product 11 was obtained
(67%) along with 0.78 g of starting material 10.
To 4.5 N HCl (36 mL) was added 1.8 g of 11, and the
resultant mixture was heated under reflux for 3.5 h. After
cooling to room temperature, the mixture was extracted with
ether (2 × 80 mL) and dried (MgSO4). After removal of the
solvents in vacuo, the acid 2 was purified by distillation
(bulb-to-bulb). The S/R isomer ratio was 81/19 and gave 1.0
g.
Sultam Route. Sodium hydride (0.48 g of a 50%
suspension in oil, 10 mmol) was washed with dry THF (2
× 5 mL) under nitrogen. The washings were removed by
decantation. Dry THF (25 mL) was then added to the solid
and the mixture cooled in an ice bath. A solution of the
sultam 12 (2.15 g, 10 mmol) in THF (50 mL) was added
dropwise over 30 min and then stirred for an addition 1 h.
Saturated aqueous NH4Cl solution (100 mL) was added and
the mixture extracted with ethyl acetate (2 × 50 mL). The
combined extracts were washed with 1 M NaOH (2 × 25
mL) and brine (25 mL), dried (MgSO4), filtered, and
Racemic 2-methylhexanoic acid was purchased from
Lancaster Synthesis. The 2-methyl-2-hexenoic acid was
purchased from Bedoukian Research, Danbury, CT.
Analytical Methods. Unless otherwise noted, the method
used to determine ee’s was an automated, precolumn
derivatisation procedure using 1-3(dimethylaminopropyl)-
3-ethylcarbodiimide hydrochloride (EDC), 4-(dimethyl-
amino)pyridine (DMAP), and R-phenylglycinol and a Hewlett-
Packard 1090 HPLC system. A Supelco LC-8, 3 µ, 25 mm
× 46 mm i.d. column was used at 40 °C. The mobile phase
was Millipore water (solvent A) and acetonitrile (solvent B)
with detection at 210 nm. The column was equilibrated with
77% A and 23% B. After sample injection (10 µL), the
system was held at these conditions for 15 min followed by
18 min with a gradient change to 0% A and 100% B. These
conditions were held for another 4 min followed by
equilibration with 77% A and 23% B. Retention times were
13.4 and 15.6 min for the R- and S-isomers, respectively.
An achiral method was achieved by diluting a sample with
water or mobile phase. This was then injected (15 mL) onto
a Supelcosil LC-18 5 mm, 25 cm × 4.6 mm column eluting
at 1 mL/min with a mobile phase comprising buffer (800
mL of 0.01 M potassium phosphate, monobasic and 0.001
M tetrabutylammonium phosphate) and acetonitrile (200
mL). Detection was at 200 nm. 2-Methylhexanoic acid had
a retention time of 5.2 min, and the 2-methyl-2-hexenoic
acid, 6.8 min.
A GC method was used for final and in process analyses.
Column was a Supelco permethylated cyclodextrin poly-
siloxane fused silica capillary (60 m × 0.25 mm i.d., 0.25
micrometer film) at 130° (isothermal). Detection was by
flame ionization. Samples were injected at ∼1 mg/mL in
acetonitrile. Septum purge was 2 mL/min, column flow 0.6
mL/min, split vent 25 mL/min, aux gas 30 mL/min, air 330
mL/min, H2 30 mL/min. Injector temperature was 250 °C
and the column head pressure ∼21 psi. Retention times were
S-2-methylhexanoic acid 17.99 min, R-2-methlyhexanoic acid
18.65 min, Z-2-methyl-2-hexenoic acid 21.94, min and E-2-
methyl-2-hexenoic acid, 23.57 min.
Oxazolidinone Route. To a solution of the oxazolidinone
5 (2.5 g, 19.4 mmol) in dry THF (65 mL) was added 1.6 M
n-butyllithium in hexane (13.3 mL, 21.3 mmol) at -78 °C.
After stirring for 30 min, hexanoyl chloride (6) (2.87 g, 21.3
mmol) was added neat. The mixture was warmed to room
temperature and stirred overnight. The reaction mixture was
quenched with saturated aqueous NH4Cl and extracted with
ethyl acetate. The combined organic phase was washed with
brine and dried (MgSO4). After removing solvents, 4.36 g
of crude product 7 was obtained (99%).
The N-acyl compound 7 from the previous step was
dissolved in dry THF (75 mL) and cooled to -78 °C when
376
•
Vol. 7, No. 3, 2003 / Organic Process Research & Development

evaporated to give the crude acyl derivative 13 (2.76 g, 88%)
which was used without further purification.
glass beads to aid mixing. Racemic 2-methylhexanoic acid
ester (0.5 mmol) was added to 0.1 M potassium phosphate
buffer (pH 7.0) in order to make a final volume of 1.0 mL.
Enzymes to be screened were first prepared by dissolving
50 mg in 10 mM phosphate buffer (5 mL) at pH 7.0. After
dissolution, the reaction was initiated by the addition of 50
µL of enzyme solution. The vials were sealed with Teflon
caps and placed in a rotary shaker (300 rpm) overnight at
ambient temperature. At the end of the incubation period,
20 µL of 6 N HCl was added followed by ether (1.0 mL).
After mixing, the ether was removed and the extraction
repeated. The combined organic extracts were evaporated
to dryness and dissolved in acetonitrile (0.1 mL) for HPLC
analysis.
To a solution of the sultam 13 (2.50 g, 7.98 mmol) in
THF (35 mL) that had been cooled to -78° was added 1.6
M n-butyllithium in hexane (5 mL, 8 mmol). The mixture
was stirred for 0.5 h when a solution of methyl iodide (0.5
mL, 5 mmol) in HMPA (3 mL) was added. The reaction
was allowed to come to ambient temperature over 2 h.
Saturated aqueous NH4Cl solution (100 mL) was added and
the mixture extracted with ethyl acetate (2 × 50 mL). The
combined extracts were washed with water (2 × 25 mL)
and brine (25 mL), dried (MgSO4), filtered, and evaporated
to give the crude alkylated acyl derivative 14 (1.96 g, 75%)
which was used without further purification.
The alkylated product 14 (1.96 g, 5.99 mmol) was
dissolved in a solution of lithium hydroxide (0.143 g, 12
mmol) in 1:1 aqueous THF (50 mL). The reaction was
monitored by TLC (SiO2, 20% EtOAc/hexanes). After
stirring at ambient temperature for 3h, the solution was
acidified with 2 M hydrochloric acid to pH 2. The mixture
was extracted with ethyl acetate (3 × 30 mL), dried
(Na2SO4), filtered, evaporated under reduced pressure, and
then distilled (bulb-to-bulb) to give the product (0.43 g, 60%)
which had an R:S isomer ratio of 95:5.
Resolution with Quinine.14 To a solution of racemic
2-methylhexanoic acid (78 g, 0.6 mol) in acetone (600 mL)
was added quinine (194.4 g, 0.6 mol) and the mixture heated
to reflux. The clear solution was cooled to room temperature,
and the crystallized solid was collected by filtration, washed
with acetone (60 mL), and dried in air. This solid was
recrystallized two more times with acetone (2 × 360 mL).
The resulting solid was added to hydrochloric acid (1 N,
400 mL). The clear solution, thus obtained, was extracted
with ethyl acetate (2 × 90 mL), washed with water (100
mL), and concentrated on a rotary evaporator to get crude
S-2-methylhexanoic acid (31.4 g, 80%). Distillation of the
crude product at 48-49 °C/0.12 mm gave pure S-2-
methylhexanoic acid as a colorless oil (23.48 g, 60%, R:S
ratio 4:96).
Large Scale. In a 30-gal reactor, racemic 2-methyl-
hexanoic acid (7.8 kg) and quinine (19.4 kg) in acetone (60
L) were heated under reflux for 1 h. The mixture was cooled
to ambient temperature overnight. The solid was collected
on a Nutsche filter, rinsed with acetone (6 L), and dried in
a nitrogen stream. This solid was then added to acetone (36
L) and heated under reflux for 1 h after which time visible
inspection showed that all solids had dissolved. The mixture
was cooled to ambient temperature over 4 h. The solid was
collected on a Nutsche filter, rinsed with acetone (5 L) and
dried under a nitrogen stream. This last recrystallisation was
then repeated. The resultant solid was added to 1 N HCl (40
L) and extracted with EtOAc (2 × 90 L). The combined
extracts were washed with water (10 L). The EtOAc was
removed by vacuum distillation to give the crude acid 2
(∼3.0 kg). Redistillation was performed as described for
asymmetric hydrogenations in a thermosyphon.
pH Stat Reaction. Racemic ethyl 2-methylhexanoate
(1.37 g, 8.7 mmol) was mixed with potassium phosphate
buffer (9.0 mL of 5.0 mM solution, pH 8.0). The mixture
was vigorously stirred at room temperature and the reaction
initiated by the addition of solid PS-30 lipase (Amano). The
reaction pH was maintained by the addition of 1 N NaOH
using a Radiometer pH stat system. After the rate of base
addition leveled off, the reaction was terminated by acidi-
fication to pH 2.0 with 6 N HCl. Extraction in a manner
similar to that described in the screening experiments
provided material for analysis.
To investigate isolation procedures this reaction was also
performed with 70 g (0.44 mol) of the ester in 5.0 mM buffer
(0.93 L, pH 7.0). The enzyme was added as a solid (3.5 g).
Asymmetric Hydrogenation. Ruthenium-(S-BINAP)
catalyst, [Ru(S-BINAP)Cl2]n (0.135 g) was placed in a 450-
mL pressure vessel. Methylene chloride (27 mL) was placed
in a second 450-mL pressure vessel. Then methanol (225
mL), water (27 mL), and 2-methyl-2-hexenoic acid (40.5 g)
were placed in a third 450-mL pressure vessel. Each vessel
was purged 100 times with nitrogen. The methylene chloride
was then pushed over to the catalyst by a positive nitrogen
pressure. The resultant mixture was stirred until the catalyst
dissolved. The water, methanol, and substrate mixture was
then added to the catalyst solution by use of a positive
nitrogen pressure. The reaction mixture was purged four
times with hydrogen. The reaction vessel was charged to 50
psig with hydrogen and heated to 80 °C. The reaction was
run until the theoretical amount of hydrogen had been taken
up. The reaction vessel was purged with nitrogen. The
reaction mixture was concentrated by distillation under
vacuum (up to 40 °C/25 mmHg). The crude product was
then purified by vacuum distillation (bp 65-70 °C/0.2
mmHg). The isolated yield of the product 2 was typically
85%. The product was analyzed by derivatisation; the acid
chloride was prepared by reaction with thionyl chloride,
followed by reaction with (S)-R-phenylethylamine. The
diastereoisomeric ratio was determined by HPLC or GC. The
ratio of S:R varied from 86:14 to 92:8, a typical run giving
around 90:10. [R]D ) +17.4° (neat): NMR (CDCl3) δ 0.9
(3H, t) 1.18 (3H, d), 1.3-1.8 (6H, m), 2.47 (1H, sex), and
11 (1H, br).
Enzymatic Methods. Enzymatic screening studies were
carried out in 2-mL HPLC vials containing several small
Large Scale. At larger scale, the catalyst was dissolved
in CH2Cl2 and this solution was then added to the substrate
Vol. 7, No. 3, 2003 / Organic Process Research & Development
•
377

in aqueous methanol (Figure 1). In one 2-gal canister was
placed [Ru(S-BINAP)Cl2]n (16) (22.68 g) and a magnetic
stirrer bar. Methylene chloride (5.2 L) was placed in the
second canister. These tanks were purged 25 times with N2
up to 30 psig, releasing the pressure at the end of the
sequence. In a 20-gal autoclave fitted with a dip tube for
gas introduction, was charged the enoic acid 15 (6.8 kg),
methanol (38 L), and DI water (4.5 L). The reactor was
purged three times with N2 to 10 psig through the dip tube.
When all purges were complete, the CH2Cl2 was pushed into
the catalyst canister by N2. The mixture was stirred for 15
min. Again by use of N2 pressure, the catalyst solution was
added to the reactor. The reactor was purged by pressurizing
to 100 psig for 2 min through three cycles to ensure that
there were no gas leaks. The system was then purged with
H2 through four cycles. Hydrogen pressure was adjusted to
50 psig. The reaction was heated to 80 °C. The hydrogen
pressure was monitored by a pressure transducer and the
reactor filled up from a reservoir as needed to keep the
pressure at 50 psig. When hydrogen uptake ceased, the
hydrogen was vented, and the system was purged three times
with N2 (20 psig). The contents of the reactor were
transferred to 50-gal drums, rinsing with methanol (10 L).
For one reaction a 200-gal autoclave was used. In this
case, the amount of catalyst was kept the same so that the
2-gal canisters could be used. Thus, the catalyst 16 (22.68
g) was dissolved in CH2Cl2 (5.2 L) as above. This was added
to a purged solution of the enoic acid 15 (51 kg) in methanol
(180 L) and water (30 L). All other operations were
performed as described for the 20-gal series.
most of the methanol. When ∼70 kg remained, the contents
of the reactor were transferred to a 30-gal reactor system.
The volume was reduced again under vacuum, heating to
30 °C. When the contents of the reactor were about 45 kg,
the vacuum was released and the mixture allowed to stand
for 0.5 h. The lower layer was removed to waste. The upper
layer was placed in a 50-L thermosyphon, rinsing in with
methanol (5 kg). Silicone oil (15 kg) was added as chaser.
If the volume was too large, the crude 2 was added as
methanol was removed. The last amounts of methanol were
removed under vacuum with gentle heating. The product was
collected >100 °C at 28-29 in. Hg. Overall yields were
cumulative and ∼75%.
Kinetic Experiments. The hydrogenations were run in
both Fischer-Porter bottles and Parr reactors. In both cases
pressure transducers were used to monitor the hydrogen
pressure. All solvents and substrates were degassed with
nitrogen prior to use. In a typical run, 2-methyl-2-hexenoic
acid (36.7 g, 316 mmol), catalyst (135 mg), water (27 mL),
and methanol (225 mL) were combined in the reactor in a
glovebox. The vessel was then sealed and transferred from
the glovebox. After 10 cycles of hydrogen, the vessel was
charged to 50 psig and heated to 80 °C. The hydrogen uptake
was monitored via a computerized hydrogenation system.
Acknowledgment
We thank Mike Scaros, Mike Prunier, and Pete Yonan
for the use of their automated hydrogenation apparatus at
Searle R & D as well as John Ryan, Applied Systems,
Incorporated, for use of the infrared reaction monitoring
system.
Workup of these reactions was performed at a different
site. The contents of two drums were charged to a 100-gal
reactor system, rinsing with MeOH (2 L). This was then put
under vacuum (∼25 in. Hg) and warmed to 30 °C to remove
Received for review December 20, 2002.
OP025616I
378
•
Vol. 7, No. 3, 2003 / Organic Process Research & Development