Complementary syntheses of N,O-protected-(S)-2-methylserine on a multikilogram scale

Article abstract of DOI:10.1021/op100299d

Two complementary and scalable approaches have been used to manufacture multikilogram quantities of N,O-protected-(S)-2-methylserine. The first approach uses a diastereomeric salt resolution of 2-methylserine methyl ester as the (1S)-(+)-camphorsulfonate salt, and was used to rapidly access 15 kg of (S)-3-tert-butoxycarbonyl-2,2,4-trimethyl-1,3-oxazolidine-4-carboxylic acid with >99% ee. The second approach involves a stereoselective enolate methylation of a chiral cyclic l-serine derivative under cryogenic conditions. The four-step telescoped process, starting from l-serine methyl ester, was used to manufacture 20 kg of (2R,4S)-2-tert-butyl-3-tert-butoxycarbonyl-4-methyl-1,3- oxazolidine-4-carboxylic acid in 52% overall yield and 98% ee. The advantages and disadvantages for scale-up of both approaches are discussed.

Full text of DOI:10.1021/op100299d

ARTICLE
pubs.acs.org/OPRD
Complementary Syntheses of N,O-Protected-(S)-2-methylserine on a
Multikilogram Scale
Michael S. Anson,† Hugh F. Clark,† Paul Evans,† Martin E. Fox,‡ Jonathan P. Graham,*,†
Natalie N. Griffiths,† Graham Meek,‡ James A. Ramsden,‡ Alastair J. Roberts,† Shaun Simmonds,‡
Matthew D. Walker,† and Matthew Willets§
†Chemical Development Division, GlaxoSmithKline Research and Development Ltd., Gunnels Wood Road, Stevenage,
Hertfordshire SG1 2NY, U.K.
‡Chirotech Technology Ltd., A subsidiary of Dr. Reddy’s Laboratories (EU) Ltd., Unit 162 Cambridge Science Park, Milton Road,
Cambridge CB4 0GH, U.K.
§Chirotech Technology Ltd., A subsidiary of Dr. Reddy’s Laboratories (EU) Ltd., PO Box 6 Steanard Lane, Mirfield,
West Yorkshire WF14 8QB, U.K.
S Supporting Information
b
ABSTRACT: Two complementary and scalable approaches have been used to manufacture multikilogram quantities of
N,O-protected-(S)-2-methylserine. The first approach uses a diastereomeric salt resolution of 2-methylserine methyl ester as the
(1S)-(þ)-camphorsulfonate salt, and was used to rapidly access 15 kg of (S)-3-tert-butoxycarbonyl-2,2,4-trimethyl-1,3-oxazolidine-
4-carboxylic acid with >99% ee. The second approach involves a stereoselective enolate methylation of a chiral cyclic L-serine derivative
under cryogenic conditions. The four-step telescoped process, starting from L-serine methyl ester, was used to manufacture 20 kg of
(2R,4S)-2-tert-butyl-3-tert-butoxycarbonyl-4-methyl-1,3-oxazolidine-4-carboxylic acid in 52% overall yield and 98% ee. The advantages
and disadvantages for scale-up of both approaches are discussed.
’ INTRODUCTION
to resolve racemic 2-methylserine methyl ester using an inexpensive
resolving agent was available,6 and the chemistry would be amenable
to processing in standard plant equipment.
GSK1842799 (Scheme 1, 1) is a potent and selective sphin-
gosine-1-phosphate receptor subtype 1 (S1P1) agonist, and as such
has potential utility for the treatment of diseases or conditions associ-
ated with inappropriate immune responses, including transplant rejec-
tion and autoimmune diseases, such as multiple sclerosis and pso-
riasis.1 A key structural feature of 1is the chiral quaternary centre. This
centre can be derived from (S)-2-methylserine, with the dimethyl
oxazolidine 2 and the tert-butyl oxazolidine 3 both having been used
for this purpose in gram-scale preparations of 1 (Scheme 1).2
To support the development activities for 1, we required rapid
access to a suitable (S)-2-methylserine derivative. There are many
methods available for accessing quaternary R-amino acids,3 including
specific methods for 2-methylserine,4 but their preparation on a multi-
kilogram scale can remain a challenge. Indeed, we initially contracted a
CRO partner to synthesise 4 kg of 2, but they were unable to develop
a suitable process that could deliver beyond a gram scale. Given the
difficulties our initial CRO partner had in identifying a suitable pro-
cess, and in order to reduce supply risk, we chose to explore two
complementary methodologies. These approaches were chosen for
investigation on the basis of the potential for rapid scale-up and
probability of delivering sufficient material of appropriate quality.
The first approach, targeting the manufacture of 15 kg of
dimethyl oxazolidine 2, was conducted by Chirotech.5 This approach
involved the diastereomeric salt resolution of racemic 2-methyl serine
methyl ester, with subsequent conversion to the dimethyl oxazolidine
2 (Scheme 2, Route A). Despite the inherent inefficiency of a
resolution step, this approach was attractive since a literature method
The second approach, undertaken at GSK, involved the dias-
tereoselective alkylation of an L-serine derivative using self-regenera-
tion of stereocentres (SRS) methodology7 to access tert-butyl oxazo-
lidine 3 (Scheme 2, Route B). This approach would allow rapid
access to 3, and the desired R-methylation of L-serine using SRS
chemistry is known.8 However, we were concerned about the suita-
bility of this chemistry to scale-up due to the propensity of the oxazo-
lidine enolates to undergo β-elimination, even at low temperature.9
Furthermore, to allow rapid manufacture on a multikilogram scale we
wished to avoid purification by chromatography; therefore, the only
opportunity to provide an upgrade to the chemical and enantiomeric
purity would be by crystallisation of the acid 3.10
The development of both approaches to enable the multi-
kilogram-scale manufacture of 2 and 3 is discussed below.
’ RESOLUTION-BASED ROUTE
We were attracted to the preparation of oxazolidine 2 via the
resolution of 2-methylserine, since the chemistry would all be
amenable to processing in standard plant equipment and requires
inexpensive and readily available starting materials, including the
resolving agent (1S)-(þ)-camphorsulfonic acid (Scheme 3).6
Received: November 9, 2010
Published: January 18, 2011
r
2011 American Chemical Society
389
dx.doi.org/10.1021/op100299d Org. Process Res. Dev. 2011, 15, 389–397
|

Organic Process Research & Development
ARTICLE
Scheme 1. Use of dimethyl oxazolidine 2 or tert-butyl oxazolidine 3 for the preparation of GSK1842799 (1)
scale-up of this reaction to a 1-kg scale were sublimation of ammo-
nium carbonate during the initial hydantoin formation, causing
blockages in the headspace above the reaction, and outgassing of
ammonia during addition of potassium hydroxide. Additional carry-
over of ammonium carbonate occurred during this outgassing.
However, the problems presented by these issues were relatively
minor on a 1-kg scale and did not prevent the further scale-up of this
chemistry. Potassium sulfate is insoluble in water-alcohol mixtures,
and 2-methylserine is soluble in water-methanol, so after the potas-
sium hydroxide hydrolysis and sulfuric acid neutralisation, methanol
was added to the aqueous reaction solution to precipitate this salt.
After removal of the solid potassium sulfate by filtration, addition of
acetone as an antisolvent to the solution of 2-methylserine 7 in
water-methanol was used to achieve crystallisation. The amino acid
was isolated as the hemihydrate in around 50% yield from hydro-
xyacetone. The ash content of 1.3-4%, resulting from incomplete
removal of inorganics, and water content of 7%, arising from the
compound being a hemihydrate, were acceptable for use in the
remaining synthetic steps. After transfer of the process to an external
contractor, it was found that the anhydrous compound could be
prepared from the hemihydrate by heating to 65 °C under vacuum.
The literature method for the esterification employed metha-
nolic hydrochloric acid, generated in situ using thionyl chloride,
to give the methyl ester hydrochloride 8.13 This was followed by
acid exchange with (1S)-(þ)-camphor-10-sulfonic acid to give
the camphorsulfonate salt 9, where the desired (S)-enantiomer of
the product crystallised as the less-soluble salt.6 For the 1-kg
campaign, the esterification reaction was run at a higher con-
centration than the literature procedure, but otherwise with little
modification. As with the literature procedure, thionyl chloride
was used for in situ generation of a solution of hydrogen chloride
in methanol for convenience. The esterification reaction required
heating to reflux for 16-24 h to reach a high level of conversion.
Acid exchange was achieved by addition of camphorsulfonic acid
to the solution of ester hydrochloride 8, after which the majority
of the methanol and hydrogen chloride were removed by
distillation. In addition to acid exchange, solvent exchange from
methanol to a less polar solvent was required to crystallise the
camphorsulfonate salt 9. Dichloromethane, used in the literature
procedure, was too low boiling for adequate removal of metha-
nol, and its use presented difficulties on the pilot plant. However,
a facile solvent-swap from methanol into DME was achieved,
which allowed the crystallisation of the salt 9 in high yield and in
75-80% diastereomeric excess. This was upgraded to 90-95%
diastereomeric excess by heating with acetone to give the salt 9 in
35-40% overall yield from 2-methylserine hydrate. Further
upgrade to higher enantiomeric excess was achieved at a later
stage in the synthesis. The use of camphorsulfonic acid in the
esterification, which would remove the need for acid exchange,
was tried, but the reaction was found to be too sluggish in
refluxing methanol.
Scheme 2. Approaches to (S)-(2)-methylserine derivatives 2
and 3: (A) via diastereoselective salt resolution of racemic
2-methylserine; (B) from L-serine methyl ester via diastereo-
selective alkylation of 4
The initial objective was for a 1 kg laboratory-scale make
followed by a 15 kg pilot-scale campaign.
Racemic 2-methylserine 7 was not commercially available on
sufficient scale, so an inexpensive synthesis readily scaled to
multikg quantities was required. An approach via a Bucherer-Bergs
reaction of hydroxyacetone with subsequent hydrolysis was attrac-
tive from the points of view of inexpensive raw materials, brevity, and
simplicity.11,12 The amino acid synthesis comprises two steps
(Scheme 4), the first being the reaction of hydroxyacetone with
ammonium carbonate and cyanide in an aqueous medium to give a
hydantoin 12, and the second, basic hydrolysis to the amino acid 7.
This approach presents the issue of isolation of the highly polar,
water-soluble amino acid 7 from an aqueous reaction medium
containing inorganic salts. In the literature procedure, this issue
was addressed first by isolation of the intermediate hydantoin 12 and
second by the use of barium hydroxide for the basic hydrolysis.
Neutralisation with sulfuric acid gave insoluble barium sulfate and an
aqueous solution of the amino acid 7 free of inorganic salts from
which the compound could be isolated. We found that the literature
conditions gave the rearranged oxazolidinone 13 rather than the
expected hydantoin 12. Like the amino acid 7, the intermediate 13 is
also highly polar and water soluble, hence, difficult to isolate, and
therefore was formed in aqueous solution and hydrolysed in situ
without isolation. On the basis of previous experience on another
project, we expected the literature barium sulfate procedure to give
rise to a finely divided precipitate that is very slow to remove by
filtration on large scale. Therefore, this was replaced with a potassium
sulfate precipitation, expected to give rise to a more granular solid
that is easier to remove by filtration. Thus, potassium cyanide was
used as the cyanide source, potassium hydroxide for the hydantoin
hydrolysis, and sulfuric acid for the neutralisation. A slight excess of
hydroxyacetone was used in the reaction to avoid the presence
of cyanide during the neutralisation. Difficulties identified during
390
dx.doi.org/10.1021/op100299d |Org. Process Res. Dev. 2011, 15, 389–397

Organic Process Research & Development
ARTICLE
Scheme 3. Resolution-based route
Scheme 4. Bucherer-Bergs reaction of hydroxyacetone
The literature method14,15 for the formation of acetonide 11
from the N-Boc methyl ester 10 used 2,2-dimethoxypropane as a
reagent and boron trifluoride etherate as a catalyst. Sulfuric acid
was found to be equally effective as a catalyst, while being cheaper
and more easily handled on scale. With orthophosphoric acid as the
catalyst, the reaction stopped at the 2-methoxypropyl ether inter-
mediate 15 (Figure 1), showing the C-O bond-forming step to
occur first, the slower C-N bond-forming step not occurring with
the weaker acid catalyst. The reaction required substoichiometric
rather than truly catalytic quantities of sulfuric acid. Conditions used
for scale-up were slow addition of 0.25 equivalents of sulfuric acid
to a solution of the N-Boc methyl ester 10 in 5 equivalents of 2,
2-dimethoxypropane, which gave a 95% conversion to the acetonide
11 with the remaining 5% being N-Boc methyl ester 10 and inter-
mediate 15. The reaction was quenched under anhydrous conditions
by addition of triethylamine, after which the excess reagent was
distilled away from the acetonide 11, a mobile oil, which was not
purified prior to the final deprotection. Cleavage of the methyl ester
with sodium hydroxide in methanol was relatively slow but clean,
with complete conversion being achieved at 20 °C for 64 h or 2-3 h
at reflux. The product was partitioned into water as the sodium salt
and base-insoluble material was extracted into toluene. After acid-
ification with orthophosphoric acid, the acid 2 was extracted into
toluene. The product was crystallised by addition of heptane anti-
solvent to a solution in hot toluene followed by cooling. In this crys-
tallisation, an efficient upgrade to >99.9% ee was achieved, with prod-
uct in the liquors of around 50% ee being obtained. The chemical
Figure 1. Side product 14 and intermediate 15.
Boc protection of the salt 96 was readily carried out using Boc-
anhydride (1.2 equiv) and triethylamine. THF was used as the
reaction solvent instead of the literature chloroform. A side
reaction, formation of the O-Boc ester 14 (Figure 1), was
detected in this reaction, requiring only a modest excess of
Boc-anhydride to be used. Aqueous workup with sodium carbon-
ate was used to remove camphorsulfonate salts. Toluene was
used as an extraction solvent so that the product 10 could be
dried azeotropically before the next stage of chemistry, acetonide
formation, which requires anhydrous conditions. We were
unable to crystallise the single enantiomer Boc-carbamate 10,
which we were only able to obtain as a liquid. However, the
racemate is a crystalline solid, mp 77 °C. These properties
allowed an upgrade of the enantiomeric excess by crystallisation
of racemic 10, leaving liquors of high enantiomeric excess, with
up to 98% being achieved. However, an efficient, combined
upgrade and chemical purification was possible later in the
synthesis, and the Boc carbamate 10 was carried through crude
to the acetonide formation.
391
dx.doi.org/10.1021/op100299d |Org. Process Res. Dev. 2011, 15, 389–397

Organic Process Research & Development
ARTICLE
purity of the product was also well within the specification of 98%, no
single impurity >0.2% by HPLC. On a 500 g to 1 kg scale, the acid 2
was obtained in about 18% overall yield from the racemic 2-methyl-
serine hemihydrate 7. The only purification steps required were
crystallisation of 2-methylserine hemihydrate 7, the camphorsulfo-
nate salt 9 in the resolution and final product 2.
For the 15-kg pilot campaign using the resolution process,
the racemic 2-methylserine 7 process was transferred to an
external contractor, where 80 kg were manufactured. In the
esterification step, gaseous hydrogen chloride was used instead
of thionyl chloride,16 but otherwise, the process was run with no
significant modification compared to the 1-kg scale. Starting
with a 45.4-kg batch of 2-methylserine hemihydrate, a 16.7-kg
batch of acid 2 with >99.9% ee was obtained, in an overall yield
of 18.2%.
step that had been stripped to dryness under high vacuum, the
reaction proceeded smoothly to completion. However, using the
concentrated solution following partial distillation, our reaction
consistently appeared to stall at about 60% conversion, and would
not proceed further even with an additional charge of Boc-anhydride,
1
or by addition of DMAP or triethylamine. H NMR analysis of
reaction samples in d4-methanol showed the same ratio of open-
chain and oxazolidine ring diastereomers was present when the
reaction had stalled as in the initial mixture. To gain insight into why
the reaction was not reaching completion, the reaction was investi-
gated in d8-THF, to allow direct analysis of the reaction mixture by
NMR. The equilibrium ratio for the oxazolidine diastereomers is
known to be solvent dependent,17 and in d8-THF the initial mixture
shows a similar ratio of the cis- and trans-diastereomers, but with
considerably less of the open-chain imine (53:44:3 ratio of cis-, trans-,
and open chain). However, at the point when the reaction stalls, only
a single component in addition to the desired product 4 is observed
by 1H NMR, which was identified as the trans-oxazolidine 16trans. On
adding a drop of D2O to the NMR sample, the reaction again started
to proceed to give the desired Boc-protected cis-oxazolidine 4. It
appears that a component in the reaction that is not removed during
the partial distillation of 16 inhibits the equilibrium between the
oxazolidine diastereomers but addition of a protic source to the
reaction reinstates the equilibrium (hence, why the equilibrium
mixture was seen when sampling the reaction into d4-methanol).
On spiking individually 2-methyl pentane, pivalaldehyde, and
triethylamine into the reaction with a sample of 16 that had been
stripped to dryness, only the reaction with added triethylamine
stalled, indicating that it is carryover of triethylamine that is prevent-
ing the oxazolidine equilibrium. To avoid having to ensure that all
triethylamine is removed from the reaction, we investigated the
possibility of adding a protic source to allow the equilibration to
continue. Since the literature workup involves partitioning of the
completed reaction between aqueous sodium bicarbonate and
diethyl ether, we charged the aqueous base to a stalled reaction,
which then proceeded cleanly to completion on stirring at ambient
temperature overnight. Boc-protected oxazolidine 4 could then be
directly extracted into a suitable solvent, and we chose to use
2-methylpentane, since it is safer to use on scale than diethyl ether
and also minimizes the total number of solvents used in the process.
We found that replacing the sodium bicarbonate solution with 1 M
potassium carbonate gave a more rapid conversion to 4, presumably
due to the more homogeneous reaction mixture achieved with the
lower ionic strength aqueous system. Using this process, we were
able to reduce the excess of Boc-anhydride required to 1.1 equiv.
Even so, we did not want to carryover any Boc-anhydride into the
critical stereoselective alkylation and so sought a method to remove
the excess. The literature method removed the excess Boc-anhydride
by distillation,8c but this was not considered a feasible option on scale.
Instead, we treated the 2-methylpentane solution with N-methylpi-
perazine, which rapidly reacted with the remaining Boc-anhydride.
The resulting N-Boc-N0-methylpiperazine, together with unreacted
N-methylpiperazine, was readily removed with a dilute HCl wash.
After further washes with aqueous potassium carbonate and then
water, the organic solution was concentrated by distillation to
provide 4 as an oil. Due to the thermal properties of 4, we were
restricted in plant to a maximum safe operating temperature of
50 °C, and we ran the distillation at a reduced pressure of ∼50 mbar.
The concentrate, which contained some residual THF (typically
∼20% w/w) and tert-butanol (typically ∼3% w/w; 15 mol %), a
byproduct from the Boc-anhydride, was diluted further with THF for
direct use in the next step. An estimation of the yield from the first
’ L-SERINE ALKYLATION ROUTE
The route to oxazolidine 3 via the stereoselective alkylation of
the L-serine derivative 4 was very attractive, due to the short
synthetic sequence and inexpensive and readily available starting
materials (Scheme 5). Our main concern prior to commencing
work was around the stability of the oxazolidine enolate during
the alkylation step,9 given the inevitable longer processing times
at larger scale. We also required the identification of a suitable
isolation of 3 that would enable us to control the output purity.
We targeted the production of 15 kg of 3 with an ee of >97%.
We adapted Seebach’s conditions for the condensation of
L-serine methyl ester hydrochloride with pivalaldehyde to provide
a mixture of the cis-, trans-, and open-chain oxazolidine diaste-
reomers 16 (Scheme 6).8b Analysis of the mixture by 1H NMR in
d4-methanol showed a 47:39:14 mixture of the cis-, trans-, and
open-chain oxazolidine diastereomers 16. Due to safety concerns
and to reduce the risk of emissions of volatile hydrocarbons, we
changed solvent from pentane to 2-methylpentane, and we were
also able to reduce the excess of pivalaldehyde used from 2 to 1.2
equiv. The reaction was heated under Dean-Stark conditions
until no further water was collected (∼4 h), with reaction
1
completion confirmed by H NMR analysis in d4-methanol.
Upon cooling, the triethylamine hydrochloride was removed by
filtration and the filtrate concentrated by distillation. Due to the
inefficiency of removing all of the volatiles when at low volume in
the reaction vessel, we distilled to a level of ∼3 volumes to provide
a concentrated solution of 16 for direct use in the next step. This
approach saved significant processing time, compared to transfer-
ring the reaction mixture to alternative equipment for stripping to
dryness, but meant that there was carryover of about two volumes
of 2-methyl pentane, as well as the excess pivalaldehyde and
triethylamine from the reaction.
Following the literature conditions for the Boc-protection,
which involve the reaction of the oxazolidine mixture with 2 equiv of
Boc-anhydride in THF,8c exclusive formation of the single diaste-
reomer 4was achieved, since only in the cis-oxazolidine is the nitrogen
sterically available to react. Using material from the condensation
Scheme 5. Diasteoselective alkylation route to 3
392
dx.doi.org/10.1021/op100299d |Org. Process Res. Dev. 2011, 15, 389–397

Organic Process Research & Development
ARTICLE
Scheme 6. Formation of N-Boc cis-oxazolidine 4
Scheme 7. Diastereoselective alkylation route to 3
two steps was achieved either by an NMR assay or by loss on drying,
with the yield typically in the range 65-75%. Given the timelines we
were working to, efforts were not made to improve upon this
moderate yield, but it appeared that the majority of the mass balance
loss occurred during the filtration of solids in the first step, with
unreacted L-serine methyl ester trapped with the triethylamine
hydrochloride.
Our main concern with the diastereoselective alkylation was
the sensitivity of the chemistry to temperature; thus, a series of
experiments were performed using the literature conditions
at different temperatures.8c When the reaction was run above
-50 °C, very little desired product was observed, presumably
due to decomposition via a β-elimination pathway. In contrast,
our results indicated that the reaction could be performed at up
to -50 °C, which we had confidence would be achieved in our
cryogenic pilot-plant equipment. Indeed, we rapidly scaled the
chemistry into a 10-L jacketed vessel, which performed in a
fashion analogous to that of the small-scale reactions. With
increased confidence that we would be able to run the chemistry
on a pilot-plant scale, we sought to optimise the process for
transfer to the pilot-plant and identify conditions to telescope
with the hydrolysis step.
To avoid unnecessary cooling, the LDA solution was prepared
at 0 ( 5 °C, with reduced quantities of diisopropyl amine (1.5
equiv), nBuLi (1.6 M in hexanes, 1.3 equiv), and THF (9 vol)
(Scheme 7). The amount of base employed could be further
reduced but was maintained at 1.3 equiv of nBuLi to ensure that
complete conversion to the enolate would still occur in the
presence of adventitious water or with higher levels of tert-
butanol from the Boc-protection step contaminating the input 4.
Robustness studies indicated that the enolate formation could be
performed in the temperature range -65 ( 10 °C, which could
be maintained by controlled addition of the concentrated THF
solution of 4. Attempts were made to follow the reaction by in situ
ReactIR analysis, but the absorbance at cold temperature was too
large to decipher any trends. Instead, process monitoring was per-
formed by 1H NMR, with reaction samples immediately quenched
into cold d4-methanol prior to analysis. Complete enolate formation,
demonstrated by full deuterium incorporation at the 2-position, was
routinely achieved within 30 min of addition of the substrate. Efforts
to substitute the volatile methyl iodide for dimethyl sulfate were
unsuccessful, but the excess of methyl iodide required could be
reduced from 2.7 to 1.7 equiv. Again, a slow addition rate was utilized
to maintain the temperature range of -65 ( 10 °C, with 1H NMR
393
dx.doi.org/10.1021/op100299d |Org. Process Res. Dev. 2011, 15, 389–397

Organic Process Research & Development
ARTICLE
Figure 2. Rate of gas flow during LDA preparation and subsequent
Figure 3. Rate of gas flow and MeI content upon heating reaction after
warming.
NaOH addition.
hour after addition of the sodium hydroxide, and the rate of
evolution of methyl iodide fell during the heating period to
50 °C. However, the rate of evolution of butane increased during
the reaction heat-up (Figure 3). To minimize the flow of butane,
a stepwise temperature ramp was incorporated into the process.
Although the rate of evolution of butane still increased during the
heat-up, the maximum rate of gas evolution was reduced and
meant that we could stay within our consented limits for
emission of butane.
In the pilot plant, with 22.5 kg input of L-serine methyl ester
hydrochloride, the condensation and Boc-protection chemistry
worked as planned, although the calculated solution yield of 4
was at the lower end of what we expected, at 64%. This solution
was taken directly into the alkylation and hydrolysis steps, which
again performed without issue in 81% yield. A total of 21.5 kg of
oxazolidine 3 with 98% ee and >99% chemical purity by HPLC
was obtained, with an overall yield of 52% from L-serine methyl
ester hydrochloride.
analysis showing complete conversion to 17 after ∼3h, withonlythe
single diastereomer observed. We found that a simplified workup
was adequate to provide a solution of 17 suitable for direct use in the
hydrolysis step. Upon reaction completion, a methanol quench
(0.2 vol) was performed at -65 ( 10 °C prior to warming the reac-
tion mixture to 20 °C. A single wash with 22% w/w aqueous
ammonium chloride solution removed the inorganics and diisopropyl-
amine, to provide a clear solution of 17 in THF and hexanes.
Hydrolysis of the methyl ester proceeded smoothly upon
addition of aqueous sodium hydroxide and methanol, with the
reaction typically complete within 90 min at 50 ( 5 °C. Removal of
the volatiles (THF, hexanes, and methanol) by distillation and
dilution with water provided an aqueous solution of the sodium
carboxylate salt, with the acid 3 then crystallised by slow addition of
acetic acid. Upon cooling to 0 ( 5 °C, the solids were collected by
filtration, washed with water, and dried under vacuum to provide a
white solid of 3, in typically 70-80% yield from 4, giving an overall
yield of ∼50% from L-serine methyl ester hydrochloride 5. The chiral
purity of 3 was initially demonstrated to be acceptable by processing
a lab-scale sample of 3 through to drug substance 1, which had
an enantiomeric excess of >99%. A chiral GC method was then
developed, to allow direct analysis of 3.
Prior to commencing the pilot-plant campaign, we needed to
demonstrate that the maximum rate of emission of the volatile
organic compounds was within consented limits. Our main concern
was the rate of evolution of butane, and we also needed to
demonstrate the fate of excess methyl iodide. Whilst we showed
that we could exchange nBuLi with nHexLi for the enolate forma-
tion, thereby replacing the butane byproduct with the less volatile
hexane, this change would have led to a delay in commencing the
pilot-plant campaign. Investigations using an RC1 and Hiden Gas
analyser were performed to understand the rate of butane evolution
and to identify a suitable control strategy. Although butane has a
boiling point of -1 °C, it has a high solubility in organic solvents and
so can be considered under our reaction conditions more like a
highly volatile organic solvent than a gas. During the addition of
nBuLi to diisopropylamine and subsequent stir period at 0 °C very
little butane was evolved (see Figure 2). The rate of evolution of
butane increased but remained low on warming to 25 °C, meaning
that even if the cooling system failed there would not be a sudden
increase in the rate of butane emissions.
’ CONCLUSIONS
Two routes to N,O-protected-(S)-2-methylserine have been
developed, with the manufacture of >15 kg each of (S)-3-tert-
butoxycarbonyl-2,2,4-trimethyl-1,3-oxazolidine-4-carboxylic acid 2
and (2R,4S)-2-tert-butyl-3-tert-butoxycarbonyl-4-methyl-1,3-oxazoli-
dine-4-carboxylic acid 3achieved. A six-step, resolution-based synthe-
sis was used to prepare 2, in about 9% overall yield from hydroxy-
acetone. The chemistry was simple, robust, and inexpensive and
hence suitable for rapid scale-up. A four-step telescoped synthesis,
involving a diastereoselective alkylation, was used to prepare 3 in
about 50% overall yield from L-serine methyl ester. The chemistry
required cryogenic conditions for the alkylation, but the route was
shorter and higher-yielding than the resolution route. The use of 2
and 3 in the preparation of the S1P1 agonist 1 will be the subject of
a future communication.
’ EXPERIMENTAL SECTION
4-Methyloxazolidine-2-one-4-carboxamide13.11,12. Potassium
cyanide (17.6 g, 135 mmol), and ammonium chloride (14.4 g,
270 mmol) were placed in a flask. A solution of hydroxyacetone
(10.0 g, 135 mmol) in ethanol (100 mL) and water (100 mL) was
added. The suspension was stirred at room temperature for 3 h, then
ammonium carbonate (65 g, 675 mmol) was added. The reaction was
heated to 60 °C for 7 h, then allowed to cool to room tempera-
ture. The solvent was evaporated, ethanol (100 mL) was added, the
No gas evolution was detected during the low-temperature
chemistry, with low levels of butane and trace levels of methyl
iodide detected on warming to 20 °C. 1H NMR analysis showed
that the methyl iodide was consumed over a period of about one
394
dx.doi.org/10.1021/op100299d |Org. Process Res. Dev. 2011, 15, 389–397

Organic Process Research & Development
ARTICLE
suspension was filtered, and the solvent was evaporated. The residue
was evaporated with methanol (3 ꢀ 100 mL) to give the oxazolidi-
none 13 as a brown foam which solidified slowly on standing (18.8 g,
87%). 1H NMR (400 MHz, CDCl3): δ 8.09 (br s, 1H), 7.48 (br s,
1H), 7.37 (br s, 1H), 4.37 (d, J= 8.8 Hz, 1H), 4.03 (d, J=8.4Hz,1H),
1.39 (s, 3H). 13C NMR (100 MHz, CDCl3) δ 175.1 (s), 157.9 (s),
73.4 (t), 60.9 (s), and 24.2 (q).
(S)-2-Methylserine Methyl Ester (1S)-(þ)-Camphor-10-sul-
fonate 9.6. (1S)-(þ)-Camphor-10-sulfonic acid (1.95 kg, 8.40
mol) was charged to the solution of 2-methylserine methyl ester
hydrochloride 8. An endotherm of 2 °C occurred. The solvent was
removed using a rotary evaporator (water bath at 50 °C) to a mass of
4.01 kg. Methanol (5 L) was added, and then the solvent was
removed using a rotary evaporator to a mass of 3.49 kg. DME (2.5 L)
was added. The solvent was evaporated using a rotary evaporator
such that 2 L of distillate was collected. DME (2.5 L) was added to
obtain a mobile slurry, then this was transferred to the 20-L vessel,
using further DME (2.5 L). The white suspension was stirred at
ambient temperature for 65 h, after which the temperature was
measured at 10 °C. The suspension was filtered, and the solid was
washed with fresh DME (3 ꢀ 500 mL) and then dried to give the
crude salt 9 as a white, granular solid (1.28 kg). The diastereomeric
excess was 78% (S), liquors 73% (R). The salt, acetone (3.03 L), and
methanol (0.54 L) were charged to a 20-L jacketed vessel. The
suspension was heated to reflux (55-58 °C) for 2 h, cooled to room
temperature over 18 h, and then filtered. The solid was washed with
acetone (2 ꢀ 500 mL) and dried to give the salt 9as a white, granular
solid (1.12 kg, 36.5%). The diastereomeric excess was 89% (S),
liquors 23% (R). 1H NMR (400 MHz, DMSO): δ 8.4 (br s, 3 H),
5.81 (t, J = 4.8 Hz, 1 H), 3.77 (dd, J = 10.8, 4.8 Hz, 1H), 3.77 (s,
3H), 3.77 (s, 3 H), 3.53 (dd, J = 11.0, 2.6 Hz, 1H), 2.87 (d, J =
14.4 Hz, 1H), 2.70 (quintet, J = 10.2, 1H), 2.37 (d, J = 14.8 Hz, 1H),
2.24 (d, J= 18.3, 3.6 Hz, 1H), 1.94 (t, J= 4.4 Hz, 1H), 1.88-1.82 (m,
1H), 1.80 (d, J = 18.4 Hz, 1H), 1.36 (s, 3 H), 1.27 (q, J = 10.0 Hz,
2H), 1.05(s, 3H), and0.74(s, 1H). 13CNMR(100MHz, DMSO):
δ 216.6 (s), 171.0 (s), 65.0 (t), 61.2 (s), 58.6 (s), 53.5 (q), 47.4
(s), 47.0 (t), 42.6 (t), 42.5 (d), 26.8 (t), 24.5 (t), 20.5 q), 19.9 (q),
and 18.6 (t).
(S)-N-tert-Butoxycarbonyl-2-methylserine Methyl Ester
10.6. THF (2.0 L) and the camphorsulfonate salt 9 (∼90% de,
1.93 kg, 5.29 mol) were charged to a nitrogen-purged 10-L vessel at
20 °C. Triethylamine (884 mL, 6.34 mol) was added. An endotherm
of 4 °C occurred. A solution of Boc-anhydride (1384 g, 6.34 mol)
and THF (940 mL) was added over 2.5 h. An exotherm to 24 °Cand
gas evolution were observed. The suspension was stirred at 20 °C for
18 h. A solution of sodium carbonate (585 g, 5.52 mol) in water
(6.60 L) was prepared in a 20-L jacketed vessel. The reaction mixture
was added into the solution with stirring. Toluene (3.50 L) was
added. The mixture was stirred at 17 °C at 160 rpm for 30 min. The
phases were separated, and the aqueous phase was extracted with
toluene (2 ꢀ 2.00 L). The combined organic phases were dried with
sodium sulfate (220 g) and filtered. The filtrate was dried in two
batches (rotary evaporator, 30 mbar 50 °C bath temp) to give crude
(S)-N-tert-butoxycarbonyl-2-methylserine methyl ester 10 (∼90%
ee) as a viscous, pale-yellow liquid (1.41 kg). The aqueous phase was
extracted with toluene and concentrated separately to give ∼8 g of
product. 1H NMR (400 MHz, CDCl3): δ 5.29 (br s, 1 H), 3.98 (dd,
J = 11.2, 5.6 Hz, 1H), 3.80-3.75 (m, 1H), 3.78 (s, 3H), 3.3 (br s,
1H) 1.47 (s, 3H), and 1.46 (s, 9H). 13C NMR (100 MHz, CDCl3):
δ 174.3 (s), 155.8 (s), 80.7 (s), 67.3 (t), 61.4 (s), 53.1 (q), 28.6 (q),
and 21.2 (q).
2-Methylserine 7.11,12. A 20-L jacketed vessel was charged
with ammonium carbonate (2279 g, 23.7 mol) and water (3.0 L).
Ammonia solution (S.G. 0.88, 1180 mL, ∼17 mol) and potassium
cyanide (1000 g, 15.4 mol) were added. The solids were rinsed into
the reactor with water (400 mL). Hydroxyacetone (1173 g, 15.8
mol) was added via dropping funnel over 3 h. The temperature of
the reaction mixture was kept at 22-24 °C during the addition and
then stirred at 20 °C for 30 min. The mixture was heated to 65 °C
over 2 h. Above ∼60 °C gas evolution (ammonia) became
apparent. The mixture was slowly heated to 85 °C over 4 h. The
cyanide concentration of the mixture at this point was measured at
∼0.25 g/L using a cyanide ion-selective electrode; this represented
>99.5% of the cyanide consumed. The reaction mixture was cooled,
and potassium hydroxide (4177 g, 85%, 63.3 mol) was added in
100-140 g portions over 90 min. Ammonia was discharged from
the reaction during the addition. The mixture was heated to 92 °C
overnight. The mixture was cooled to 20 °C and was again checked
for free cyanide and contained trace amounts. Methanol (6 L) was
added, leading to the crystallization of potassium carbonate. Sulfuric
acid was added (50%, 700 mL to pH 13; then 98% 2000 mL to pH
5.8). The head space was monitored for HCN at pH 10.7 (no
HCN) and pH 9.3 (3 ppm); below pH 8 carbon dioxide
effervescence was significant. The potassium sulfate was removed
by filtration (can be very slow), and the cake was washed with
aqueous methanol (1:1, 2 ꢀ 1 L). The supernatant (10 L) was
returned to the 20-L vessel, acetone (5 L) was added, the mixture
was seeded and stirred for 1 h, and then acetone (5 L) was added
slowly. The product was collected and washed with acetone/water/
methanol (2:1:1, 1 L) and acetone (1 L), and was dried in the
vacuum oven at 30 °C overnight to give 2-methylserine 7 as the
hemihydrate, a slightly off-white solid (1019 g, 52% yield). Karl
Fisher Analysis 7.0% water. 1H NMR (400 MHz, DMSO): δ 7 (br
s, 4H), 3.43 (d, J = 11.0 Hz), 3.30 (d, J = 11.0 Hz, 1H), and 1.09 (s,
3H). 1H NMR (400 MHz, D2O): δ 3.91 (d, J = 12.2 Hz, 1H), 3.65
(d, J = 12.2 Hz), and 1.41 (s, 3H). 13C NMR (100 MHz, D2O): δ 7
175.4 (s), 64.8 (t), 62.6 (s), and 18.5 (q).
2-Methylserine Methyl Ester Hydrochloride 8.13. A nitro-
gen-purged 20-L jacketed vessel with the jacket held at 20 °C was
charged with methanol (10 L) and 2-methylserine hemihydrate 7
(1.00 kg, 8.40 mol). Thionyl chloride (1210 mL, 16.8 mol) was
added over 3 h and 20 min, during which the maximum tem-
perature reached was 32 °C. The solution was heated to reflux
(55-58 °C) for 16 h, then allowed to cool to room temperature
to give a solution of 2-methylserine hydrochloride 8 in metha-
nol/HCl. A small aliquot was removed and concentrated to give
2-methylserine methyl ester hydrochloride 8 as a viscous, pale-
1
yellow oil. H NMR analysis showed this to be a roughly 20:1
mixture of ester and amino acid hydrochlorides. 1H NMR (400
MHz, DMSO) δ 8.6 (br s, 3H), 3.74 (s, 3H), 3.73 (d, J = 10.8 Hz,
1H), 3.63 (d, J = 10.8 Hz, 1H), and 1.39 (s, 3H). 13C NMR (100
MHz, DMSO): δ 171.0 (s), 64.9 (t), 61.4 (s), 53.3 (q), 18.4 (q).
2-Methylserine hydrochloride: 1H NMR (400 MHz, DMSO) δ
8.4 (br s, 3H), 3.72 (d, J = 11.2 Hz, 1H), 3.59 (d, J = 11.2 Hz, 1H),
and 1.35 (s, 3H). 13C NMR (100 MHz, DMSO): δ 172.2 (s),
64.8 (t), 61.0 (s), 18.6 (q).
(S)-Methyl 3-tert-Butoxycarbonyl-2,2,4-trimethyloxazoli-
dine-4-carboxylate 1113. (S)-Boc-2-methylserine methyl ester
10 (1.41 kg, ∼5.2 mmol) and 2,2-dimethoxypropane (3.24 L, 26.4
mol) were charged to a nitrogen-purged 10 L vessel at 20 °C.
Concentrated sulfuric acid (70 mL, 1.3 mol) was added over 3 h,
then the dark brown solution was stirred at 20 °C for 30 min. GC
analysis showed a 96.5:2.5 ratio of starting material and product. The
reaction was quenched with triethylamine (368 mL, 2.64 mmol).
395
dx.doi.org/10.1021/op100299d |Org. Process Res. Dev. 2011, 15, 389–397

Organic Process Research & Development
ARTICLE
An exotherm of 2 °C occurred. The solvent was evaporated using a
rotary evaporator to give crude (S)-methyl 3-tert-butoxycarbo-
nyl-2,2,4-trimethyloxazolidine-4-carboxylate 11 as a biphasic liquid
(1649 g). Spectroscopic data for this compound were obtained using
a sample purified by flash chromatography on silica eluting with
heptane/ethyl acetate (9:1). 1H NMR (400 MHz, CDCl3): δ 4.10
and 4.08 (d, J = 8.4 Hz and d, J= 8.4 Hz, 1H), 3.81 and 3.79 (d, J = 8.4
Hz and d, J = 8.4 Hz, 1H), 1.62, 1.60, 1.58, and 1.55 (4 s, 12H), and
1.48 and 1.41 (2 s, 9H). 13C NMR (100 MHz, CDCl3): δ 173.3 (s),
173.1 (s), 151.8 (s), 151.2 (s), 96.6 (s), 95.7 (s), 80.9 (s), 80.7 (s),
74.1 (t), 73.7 (t), 66.2 (s), 65.7 (s), 52.9 (q), 52.8 (q), 28.7 (q), 28.6
(q), 27.4 (q), 26.7 (q), 25.0 (q), 23.7 (q), 22.3 (q), and 21.2 (q).
(S)-3-tert-Butoxycarbonyl-2,2,4-trimethyl-1,3-oxazolidine-
4-carboxylic Acid 2. Crude (S)-methyl 3-tert-butoxycarbonyl-
2,2,4-trimethyloxazolidine-4-carboxylate 11 (1649 g) and metha-
nol (2.5 L) were placed in a nitrogen-purged 10-L vessel at 20 °C.
Sodium hydroxide (423 g, 10.6 mol) was added. An exotherm to
60 °C occurred. The solution was allowed to cool to 20 °C and
was stirred at 20 °C for 6 h, then stirred at 30 °C for 16 h. A 20-L
jacketed vessel was charged with water (5.8 L), and the crude
reaction mixture was added. Toluene (3 L) was added, and the
mixture was stirred vigorously for 30 min. The organic phase was
removed, and 85% orthophosphoric acid (∼920 mL) was added
to the aqueous phase over 1.5 h to pH 3.0. The product was
extracted into toluene (3 L) with vigorous stirring for 1 h. The
solvent was removed on a rotary evaporator to give 925 g of crude
product as a pale-yellow solid. The aqueous layer was extracted
with a further portion of toluene (1.5 L each), giving a further 40 g
of crude product. The crude product and toluene (920 mL) were
placed in a nitrogen-purged 10-L jacketed vessel. The suspension
was heated to 70 °C at which point a clear, yellow solution
resulted. Heptane (2.76 L) was added over 30 min, during which
the internal temperature was maintained at 60-70 °C. The clear,
yellow solution was seeded with a little of the crude product, then
cooled to 50 °C over 1.5 h, stirred at 50 °C for 2 h, cooled to 45 °C
over 1.5 h, then 20 °C over 16 h; the suspension was then filtered
(the solid tended to stick to the sides of the vessel), and the solid
was washed with heptane/toluene (3:1, 2 ꢀ 500 mL) and dried
under vacuum to give 2 as a white, granular solid (724 g, 53% from
the salt). The enantiomeric excess was >99.9%, liquors 48% ee.
The liquors were returned to the vessel and cooled to þ4 °C for 6
h, then were filtered to give a further 19 g of product of lower
purity with 99.7% ee;liquors 42% ee. Mp 124 °C. 20[R]D -27.9 (c
= 1.00, CHCl3). 1H NMR(400 MHz, CDCl3): δ 4.37 and 4.16(d,
J = 8.8 Hz and d, J = 8.8 Hz, 1H), 3.85 and 3.79 (d, J = 8.8 Hz and
d, J = 8.8 Hz, 1 H), 1.62, 1.58, and 1.55 (3s, 12 H), and 1.50 and
1.43 (2s, 9 H). 13C NMR (100 MHz, CDCl3): δ 179.3 (s), 177.7
(s), 152.7 (s), 151.2 (s), 96.9 (s), 96.1 (s), 81.9 (s), 81.3 (s), 74.2
(t), 73.5 (t), 66.2 (s), 65.6 (s), 28.8 (q), 28.7 (q), 27.0 (q), 26.6
(q), 25.3 (q), 23.7 (q), 22.0 (q), and 20.9 (q). HRMS (ES þve;
MHþ) calcd for C12H22NO5 260.1498, found 260.1500.
the reaction was stirred at 20 ( 5 °C for 6 h. A 20% w/w aqueous
solution of potassium carbonate (135 kg) was charged, and the
resultant biphasic mixture stirred at 20 ( 5 °C for a further 14 h,
with reaction completion confirmed by 1H NMR. 2-Methylpentane
(113 L) was added, and the phases were separated. The aqueous
layer was discarded, and N-methyl piperazine (7.2 kg) was charged
to the organic solution and stirred for ∼1 h. The organic phase was
washed with 1 M HCl (2 ꢀ 68 kg), 20% w/w aqueous K2CO3 (82
kg), and finally with water (67 kg). The volatiles were removed
under vacuum by distillation (-0.65 bar, 30-35 °C), and the
concentrate was diluted with THF (4.5 L) to afford a 69% w/w
solution18 of 4 (38.6 kg; equivalent to 26.5 kg of 4, 64%). 1H NMR
(400 MHz, d4-MeOH) 4.99 (s, 1H), 4.72 (dd, J = 8.1, 5.6 Hz, 1H),
4.24 (dd, J = 8.8, 5.6 Hz, 1H), 4.13 (t, J = 8.3 Hz, 1H), 3.74 (s, 3H),
1.48 (s, 9H), 0.93 (s, 9H).
(2R,4S)-2-tert-Butyl-3-tert-butoxycarbonyl-4-methyl-1,
3-oxazolidine-4-carboxylic Acid 3. A solution of diisopropylamine
(14.0 kg, 1.5 equiv) in anhydrous tetrahydrofuran (211 kg) was
cooled to 0 ( 5 °C, and 1.6 M n-butyl lithium in hexanes (51.0 kg,
1.3 equiv) was charged, maintaining the temperature in the range 0
( 5 °C. The reaction mixture was cooled to -65 ( 5 °C, and the
69% w/w solution of 4 in THF (38.4 kg; equivalent to 26.4 kg of 4)
was charged and rinsed in with THF (12 kg), maintaining the
temperature in the range -65 ( 5 °C. After 45 min, the enolate
1
formation was confirmed to be complete by H NMR analysis.
Methyl iodide (21.9 kg, 1.7 equiv) was charged and rinsed in with
THF (4 kg), maintaining the temperature in the range -65 ( 5 °C.
The reaction was stirred at -65 ( 5 °C for 3 h and sampled for
reaction completion by 1H NMR. MeOH (4.6 kg) was charged and
the reaction warmed to 20 ( 5 °C before washing with 22% w/w
aqueous ammonium chloride solution (198 kg). MeOH (127 kg),
water (53 kg), and 32% w/w sodium hydroxide (46 kg) were
charged, and the reaction was warmed to 50 ( 5 °C over 2.5 h and
stirred at this temperature for a further 1.5 h. The reaction was shown
to be complete by HPLC analysis. The reaction was concentrated to
a level of ∼132 L by distillation (∼450 L distillate removed), diluted
with water (255 kg), and adjusted to 50 ( 5 °C. Glacial acetic acid
(41.8 kg) was added over 1 h, rinsing in with water (10 kg), and the
resulting slurry was cooled to 0 ( 5 °C over 2 h and aged at this
temperature for a further hour. The solids were collected by filtration,
washed with 10.5% w/w aqueous glacial acetic acid solution (132 kg)
and then water (2 ꢀ 132 kg), and dried under reduced pressure at 50
( 5 °C for 6 h to provide 3 as a white powder (21.5 kg, 81%) with
1
97.8% ee and >99% area purity by HPLC. H NMR (400 MHz,
CDCl3): δ 5.05 (s, 1H), 4.28 (d, J= 8.3 Hz, 1 H), 3.85 (d, J =8.3 Hz,
1 H), 1.60 (s, 3 H), 1.45 (s, 9 H), and 0.99 (s, 9 H). 13C NMR (100
MHz, CDCl3): δ 175.1 (s), 155.6 (s), 98.4 (s), 96.1 (d), 82.3 (s),
78.3 (t), 67.6 (s), 40.3 (s), 28.5 (q), 27.1 (q), and 21.8 (q). HRMS
(ES þve; MHþ) calcd for C14H26NO5 288.1811, found 288.1817.
’ ASSOCIATED CONTENT
(2R,4S)-2-tert-Butyl-3-tert-butoxycarbonyl-1,3-oxazolidine-
4-carboxylic Acid Methyl Ester 4. L-Serine methyl ester hydro-
chloride (22.5 kg, 1 equiv), triethylamine (16.0 kg, 1.1 equiv), and
pivalaldehyde (15.1 kg, 1.2 equiv) were slurried in 2-methylpentane
(225 L). The reaction was heated to reflux under Dean-Stark
conditions for 4 h, and reaction completion was confirmed by 1H
NMR. The slurry was cooled to 25 °C, and the solids were removed
by filtration, washing through with 2-methylpentane (3 ꢀ 67 L).
The filtrate was concentrated to ∼56 L by atmospheric distillation
to provide a concentrated solution of 16 in 2-methylpentane. THF
(100 L) and Boc anhydride (33.8 kg, 1.1 equiv) were charged, and
S
Supporting Information. Methods for the preparation of
b
and spectroscopic data for side product 14 and reaction intermediate
15; methods of purity analysis for 2-methylserine 7, (S)-3-tert-
butoxycarbonyl-2,2,4-trimethyl-1,3-oxazolidine-4-carboxylic acid 2
and (2R,4S)-2-tert-butyl-3-tert-butoxycarbonyl-4-methyl-1,3-oxazoli-
dine-4-carboxylic acid 3; methods of chiral analysis for N-tert-butoxy-
carbonyl-2-methylserine methyl ester 10, camphorsulfonate salt 9,
methyl ester 11, (S)-3-tert-butoxycarbonyl-2,2,4-trimethyl-1,3-oxa-
zolidine-4-carboxylic acid 2 and (2R,4S)-2-tert-butyl-3-tert-butoxy-
carbonyl-4-methyl-1,3-oxazolidine-4-carboxylic acid 3; in-process
396
dx.doi.org/10.1021/op100299d |Org. Process Res. Dev. 2011, 15, 389–397

Organic Process Research & Development
ARTICLE
methods of analysis for preparation of N-tert-butoxycarbonyl-
2-methylserine methyl ester 10, methyl ester 11 and (2R,4S)-2-tert-
butyl-3-tert-butoxycarbonyl-4-methyl-1,3-oxazolidine-4-carboxylic
acid 3. This material is available free of charge via the Internet at
http://pubs.acs.org.
(17) F€ul€op, F.; Pihlaja, K. Tetrahedron 1993, 6701.
(18) Solution concentration determination: a portion of the THF
solution (30.0 g) was concentrated to dryness to provide a clear oil of 4
(20.6 g).
’ AUTHOR INFORMATION
Corresponding Author
jonathan.p.graham@gsk.com.
’ ACKNOWLEDGMENT
We are grateful to Robert Williams at Chirotech and Jamie
Russell and Charles Dexter at GSK for assistance with pilot-plant
operations, and we thank Kelly Goulston and Fiona King at GSK
for key analytical support.
’ REFERENCES
(1) (a) Evindar, G.; Deng, H.; Bernier, S.; Yao, G.; Coffin, A.; Yang,
H. WO 2008016692; (b) Anson, M. S.; Day, C. J.; Graham, J. P.; Kindon,
L. J.; Woolham, G. R. WO 2009098193.
(2) Initial discovery and preparation of GSK1842799 was performed
by Praecis Pharmaceuticals Inc., and multigram preparations of (2S)-2-
amino-2-{5-[4-(octyloxy)-3-(trifluoromethyl)phenyl]-1,3,4-thiadiazol-
2-yl}-1-propanol 1, using both 2 and 3, were performed under contract by
Albany Molecular Research, Inc.
(3) For reviews on the synthesis of R,R-disubstituted R-amino acids,
see (a) Vogt, H.; Br€ase, S. Org. Biomol. Chem. 2007, 5, 406. (b) Ohfune,
Y.; Shinada, T. Eur. J. Org. Chem. 2005, 5127. (c) With, T. In Organic
Synthesis Highlights IV; Schmalz, H.-G., Ed.; Wiley: Weinheim, 2000;
pp 26-33.
(4) For a review on the synthesis of 2-methylserine, see Nakano, K.;
Kotsuki, H.; Ichikawa, Y. Org. Prep. Proced. Int. 2008, 40, 67.
(5) Now Dr. Reddy’s Laboratories, Chirotech Technology Ltd.
(6) Jones, M. A.; Hislop, A. D.; Snaith, J. S. Org. Biomol. Chem. 2006,
4, 3769.
(7) For a review on the concept of self-regeneration of stereocentres,
see Seebach, D.; Sting, A. R.; Hoffmann, M. Angew. Chem., Int. Ed. 1996,
35, 2708.
(8) (a) Seebach, D.; Aebi, J. D. Tetrahedron Lett. 1984, 25, 2545. (b)
Seebach, D.; Aebi, J. D.; Gander-Coquoz, M.; Naef, R. Helv. Chim. Acta
1987, 70, 1194. (c) Brunner, M.; Saarenketo, P.; Straub, T.; Rissanen, K.;
Koskinen, A. M. P. Eur. J. Org. Chem. 2004, 3879–3883. (d) Sankyo
Company, Limited. Patent Application EP1806344 A1, 2007.
(9) Seebach notes the decomposition of the lithium enolate of
methyl-(2R,4S)-2-(1,1-dimethylethyl)-3-formyl-1,3-oxazolidine-4-car-
boxylate even at -78 °C, see ref 8a. In the Sankyo Company patent (see
ref 8d) higher temperatures, typically -25 to 10 °C, are used for the
alkylation, although only crude yields and no details of stereoselectivity
are disclosed.
(10) With the N-formyl group employed by Seebach (see ref 8a), it is
possible to crystallise the desired oxazolidine diastereomer prior to
alkylation. However, the N-formyl group would not be compatible with
our planned downstream chemistry. When the formyl group is replaced
with a Boc group, as used by Koskinen (ref 8c), none of the inter-
mediates to 3 are crystalline.
(11) Abshire, C. J. Experientia 1968, 24, 778–779.
(12) Abshire, C. J.; Berlinguet, L. Can. J. Chem. 1965, 5, 1232–1234.
(13) Easton, C. J.; Ivory, A. J.; Smith, C. A. J. Chem. Soc., Perkin Trans.
2 1997, 503.
(14) Clemens, J. J.; Davis, M.-D.; Lynch, K. R.; MacDonald, T. L.
Bioorg. Med. Chem. Lett. 2005, 15, 3568–3572.
(15) Lynch, K. R.; MacDonald, T. L. WO 2004/010949, 2004.
(16) Hansen, D.; Webber, R. K.; Pitzele, B. S.; Awasthi, A. K.;
Sikorski, J.; Trivedi, M.; Moorman, A. WO 2002/022561, 2002.
397
dx.doi.org/10.1021/op100299d |Org. Process Res. Dev. 2011, 15, 389–397