Modified chelation-controlled reduction of an N -acryloyloxazolidin-2-one

Article abstract of DOI:10.1021/op100074m

A key step in the synthesis of an optically active aminoalcohol-containing active pharmaceutical ingredient (API) involved the diasteroselective reduction of a chiral 3-acryloyl-4-benzyloxazolidin-2-one. Preliminary work identified that excellent facial s

Full text of DOI:10.1021/op100074m

Organic Process Research & Development 2010, 14, 1448–1452
Modified Chelation-Controlled Reduction of an N-Acryloyloxazolidin-2-one†
Terrence J. Connolly,*,‡ Zhixian Ding,‡ Yumin Gong,§ Michael F. MacEwan,‡ Jan Szeliga,§ and Asaf Alimardanov⊥
Wyeth Research, 401 North Middletown Road, Pearl RiVer, New York 10965, United States
Abstract:
of diastereomers, shown as 3 and dist-3 in Figure 1, with the
latter being favored by a ratio of 75:25 (entry 1, Table 1).
Addition of 1.2 equiv of magnesium bromide led to a reversed
sense of stereoinduction, and 3 was formed over dist-3 in a
ratio of 95:5.
A key step in the synthesis of an optically active aminoalcohol-
containing active pharmaceutical ingredient (API) involved the
diasteroselective reduction of a chiral 3-acryloyl-4-benzyloxazoli-
din-2-one. Preliminary work identified that excellent facial selectiv-
ity could be achieved by performing the hydrogenation in
tetrahydrofuran in the presence of magnesium bromide. During
an intermediate scale-up to support a 500-g batch of API, a side
reaction between the product and magnesium bromide was
observed that led to a significant deterioration in the isolated yield
of product. An examination of alternative chelators and processing
solvents identified that magnesium chloride in 2-methyltetrahy-
drofuran offered comparable facial selectivity with significantly
diminished liabilities for scale-up. This revised process was
incorporated into campaigns to produce larger lots of API and
afforded the product in 85% yield, averaged over 18 batches.
These results were explained on the basis of the preferred
conformation of 2. In the absence of Lewis acid, compound 2
existed predominantly as the anti-2 conformer. Literature
precedent based on other oxazolidin-2-one derivatives supports
the major conformation as anti-2 based on dipole-dipole
interactions between the two carbonyl groups.6,7 Addition of
magnesium bromide appeared to change the preferred confor-
mation to the syn-2 form, and the major product from
hydrogenation became 3. Others have reported similar magne-
sium salt-mediated reversals in the facial selectivity of R,ꢀ-
unsaturated imides during conjugate additions and oxidations
and have invoked similar mechanistic explanations.8,9
The effects of reaction temperature and applied pressure of
hydrogen were evaluated briefly (entries 2-6, Table 1). This
limited data set indicated that the working pressure of hydrogen
could be reduced to 40 psig if the temperature was maintained
at 50 °C. At lower temperatures, conversion slowed signifi-
cantly, and the reduction was not complete after 4 h (entries
4-6, Table 1).10
Introduction
Recently, we disclosed details on the route selection and
early scale-up efforts that resulted in the successful synthesis
of 5 from 11 (Scheme 1). Conversion of 2 to 3 used a newly
discovered diastereoselective hydrogenation protocol, followed
by an enolization-azidation sequence at -40 °C2,3 based on
Evans methodology.4 The details surrounding the development
of conditions for conversion of 2 to 3 and the robust process
that resulted are the subjects of this communication.
To support the preparation of a 500-g batch of active
pharmaceutical ingredient (API) 5, the reduction was scaled up
15-fold and performed in a kilo-lab reactor at 50 °C and 55
psig hydrogen. Laboratory experiments had shown that the
solubility of magnesium bromide was sufficient enough in a
mixture of IPA and water that it would not interfere with the
isolation of 3. Therefore, following removal of the catalyst, the
solvent was exchanged to IPA and water added to effect
crystallization of the product. Although laboratory yields had
been 80%, the first scale-up to the kilolab resulted in 33% yield.
An impurity was found in the filtrate that had m/z of 369, and
a stress test done after the kilolab run showed that 3 reacted
with MgBr2 at elevated temperature in a mixture of THF and
IPA to generate a peak with m/z 369, presumably formed via
Results and Discussion
As reported previously, direct hydrogenation of 2 over dry
palladium on carbon5 at 450 psig hydrogen produced a mixture
† Wyeth was acquired by Pfizer on October 15, 2009.
* Address correspondence to this author at the above mailing address, attention
B222/2125. E-mail: terrence.connolly@pfizer.com.
‡ Chemical Development.
§ Analytical and Quality Science.
⊥ Discovery Synthetic Chemistry.
(1) Alimardanov, A.; Nikitenko, A.; Connolly, T. J.; Feigelson, G.; Chan,
A. W.; Ding, Z.; Ghosh, M.; Shi, X.; Ren, J.; Hansen, E.; Farr, R.;
MacEwan, M.; Tadayon, S.; Springer, D.; Kreft, A. F.; Potoski, J. R.
Org. Process Res. DeV. 2009, 13, 1161–1168.
6
11,12 (Figure 2). Preparative LC was eventually used to obtain
(2) Safety screening performed in-house generated results in line with
those reported previously by Merck workers for triisopropylbenzene
sulfonyl azide. See: Tuma, L. D. Thermochim. Acta 1994, 243, 161–
167.
(6) Prashad, M.; Liu, Y.; Kim, H.-Y.; Repic, O.; Blacklock, T. J.
Tetrahedron: Asymmetry 1999, 10, 3479–3482.
(7) Davies has published single-crystal data for valine-derived products
related to 3 that also support the anti-relationship between the carbonyl
groups of the oxazolidinone and N-acyl groups. See: Bull, S. D.;
Davies, S. G.; Garner, A. C.; Kruchinin, D.; Key, M.-S.; Roberts,
P. M.; Savory, E. D.; Smith, A. D.; Thomson, J. E. Org. Biomol. Chem.
2006, 294, 5–2964.
(3) Details on the work that enabled the conversion of 3 to 4 to be
performed at -40 to -50 °C have been communicated recently. See:
Connolly, T. J.; Hansen, E. C.; MacEwan, M. F. Org. Process Res.
DeV. 2010, 14, 466–469.
(4) Evans, D. A.; Britton, T. C.; Ellman, J. A.; Dorow, R. L. J. Am. Chem.
Soc. 1990, 112, 4011–4030.
(8) Adam, W.; Zhang, A. Eur. J. Org. Chem. 2004, 147, 152.
(9) Kanai, M.; Muraoka, A.; Tanaka, T.; Sawada, M.; Ikota, N.; Tomioka,
K. Tetrahedron Lett. 1995, 36, 9349–9352.
(5) The process does require that dry catalyst be used since water has a
significant detrimental impact on the process. Palladium on alumina
can be used in cases where handling pyrophoric, dry palladium on
carbon is not acceptable.
(10) This screening was performed using an Endeavor Catalyst Screening
System from Biotage/Argonaut. Longer reaction times at lower
temperatures were not evaluated.
1448
•
Vol. 14, No. 6, 2010 / Organic Process Research & Development
10.1021/op100074m  2010 American Chemical Society
Published on Web 10/21/2010

Scheme 1. Condensed synthetic scheme for the conversion of 1 to 5
Figure 1. Conformer equilibrium and facial selectivity observed during hydrogenation.
Table 1. Screening results for reduction of 2 with MgBr2
entry
additive
temperature (°C)
hydrogen pressure (psig)
conversion (%)
ratioa3:dist-3
1
2
3
4
5
6
none
50
50
50
25
25
25
450
450-80
40
100
100
99
25:75
95:5
95:5
94:6
96:4
97:3
MgBr2
MgBr2
MgBr2
MgBr2
MgBr2
450
89
80
40
56
57
a Determined by HPLC.
Figure 2. Degradation products and impurities generated during MgBr2-mediated reduction.
a fraction that contained both m/z 449 and m/z 369 impurities,
and the structures were assigned as 6 and 7, respectively, on
the basis of 1D- and 2D-NMR techniques.13 A re-examination
of the MgBr2-mediated reduction showed that 6 and 7 were
also formed during the reaction and not just during the workup
and isolation process. Additionally, another degradation product
having m/z 371 was identified and assumed to be one of two
structures and shown as 8 since 6 and 7 could be reduced under
the hydrogenation conditions.14
The process was stabilized to provide the required amount
of 3 to enable timely delivery of 5 by further dilution of the
reduction conditions, incorporation of an aqueous work up to
remove MgBr2,15 and the use of vacuum during the solvent
exchange (Tj < 50 °C). Incorporation of these process changes
afforded a 76% yield of 3 when the process was scaled up to
(11) A mass of 369 corresponds to loss of HBr from 6 (m/z 449).
(12) Compound 6 could be detected in stress tests performed using 3,
MgBr2, and THF only. After 18 h at 65 °C in THF, the relative amount
of 3:7:6 formed was 2:1:2. Performing the same stress test in 1:1 THF/
IPA provided relative amounts for 3:7:6 of 1:2:0.
(13) Details are provided in the Supporting Information. It should be noted
that both 6 and 7 were hydrolyzed in the NMR sample tube when
DMSO was used as the solvent. The fraction isolated by prep-LC
initially contained three components; 6, 7, and a product consistent
with hydrolysis of 7. After three days, the sample contained 4
components: the three components mentioned above and the product
that was the result of hydrolysis of 6. We assume that the hydrolysis
product originally identified was an artifact of the prep-LC conditions
since the reduction is performed under anhydrous conditions.
(14) This byproduct was not isolated to determine the structure since it
was a secondary byproduct resulting from hydrogenation of the primary
byproducts. Limiting the amount of 6 and 7 formed during the reaction
would reduce the amount of 8, regardless of structure.
(15) Following removal of the catalyst and addition of MTBE, the reaction
mixture was washed three times with an aqueous brine solution to
remove MgBr2.
Vol. 14, No. 6, 2010 / Organic Process Research & Development
•
1449

Table 2. Expanded chelator screen in THF
provide increased selectivity with MgCl2 or attenuate the
decomposition observed with MgBr2. Results from the screening
are shown in Table 3 and are sorted in terms of conversion.
Generally, ethers performed better than other solvents in terms
of conversion. The notable exception was 1,2-dimethoxyethane,
which performed quite poorly, potentially due to a poisoning
effect caused by its ability to act as a chelator for metals.
Although the highest selectivity was achieved with MgBr2 and
MgBr2-OEt2, significant decomposition was also observed
(entries 8 and 9). The additive that offered the best balance of
selectivity with minimal decomposition and therefore lowest
liability for scale-up was MgCl2 in 2-methyltetrahydrofuran
(entry 1). It is worth noting that running the reaction at elevated
temperatures in 2-methyltetrahydrofuran with magnesium chlo-
ride offered better facial selectivity than the original MgBr2/
THF conditions.
The magnesium chloride-2-methyltetrahydrofuran conditions
were selected for further evaluation. Experiments demonstrated
that the solvent volume could be reduced from 20 volumes to
15 volumes without adversely impacting the diastereoselectivity
of the reduction. Decreasing solvent use further to 10 volumes
caused a slight drop in selectivity, but in all cases the isolated
yield was >80%, and decomposition of the product was minimal
(<1%), even when the reaction mixture was left under hydrogen
at 70-75 °C for 18 h.
entry
additive
conversion (%)
ratio 3:dist-3
1
2
MgBr2
100
100
100
100
100
100
0
100
25
42
98:2
MgCl2
80:20
36:64
41:59
49:51
24:76
-
3
MgSO4
Mg(OTf)2
Ca(OTf)2
Cu(OTf)2
CuCl2
4
5
6
7
8
FeCl2
43:57
22:88
28:72
9
ZnCl2
ZnBr2
a
10
a A significant unidentified impurity was observed with ZnBr2.
1 kg of 2; however, the process was viewed as having significant
liabilities for further scale-up. Even with the measures taken,
approximately 5% byproducts were formed during the 4 h
reduction.16 The total amount of the two primary impurities (6
and 7) increased at a rate of approximately 2%/h once the
reaction had progressed to ∼97% completion. Further, MgBr2
still remained in the organic layer even after three brine washes17
and continued to cause formation of 6 and 7 during the solvent
exchange required for isolation.
Re-examination of Alternative Chelators
In light of the identified issues and the knowledge that
processing time would be extended significantly upon further
scale-up,18 the reduction was screened with other potential
chelators, and the results are shown in Table 2.19
Of the potential chelators examined, only the previously
identified MgBr2 and MgCl2 showed any promise as additives
for the hydrogenation that would maintain the desired facial
selectivity. Another round of screening was performed that
focused on these two additives and MgBr2-OEt220 in various
solvents with the goal of identifying a solvent that would either
One final improvement was made when the reaction mixture
was adjusted to the reaction temperature prior to charging
hydrogen which led to consistently better ratios of 3:dist-3. The
isolated yield using these conditions was 87-90% in the lab
and averaged 85% over 18 hydrogenations in the kilogram
laboratory at a 2.1-kg scale.21 The improved selectivity observed
when the reaction mixture was heated prior to charging
hydrogen was presumably related to the solubility of MgCl2.
If hydrogen was applied to the system before all of the MgCl2
Table 3. Expanded solvent screen focused on MgBr2, MgCl2, and MgBr2 ·OEt2 as chelators
entry
additive
solvent
temp (°C)
conv (%)
ratio 3:dist-3
decomp (%)
1
MgCl2
2-MeTHFa
75
75
60
60
60
80
75
75
75
80
80
80
80
80
80
80
80
80
75
80
80
80
80
80
100
100
100
100
100
100
100
100
100
100
100
100
100
99
98:2
95:5
1.4
0.5
4.8
0
11.1
0
0.5
26.1
26.5
0.5
0.6
2.1
0.5
0
2
MgBr2 ·OEt2
MgBr2
2,2-DMPb
3
tetrahydrofuran
tetrahydrofuran
tetrahydrofuran
2-methoxyethanol
2,2-DMPb
96:4
4
MgCl2
90:10
97:3
5
MgBr2 ·OEt2
MgBr2 ·OEt2
MgCl2
6
84:16
77:33
100:0
100:0
46:54
43:57
27:73
25:75
83:17
59:41
18:82
58:42
27:73
98:2
7
8
MgBr2 ·OEt2
2-MeTHFa
9
MgBr2
2-MeTHFa
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
MgCl2
diethoxymethane
2-methoxyethanol
NMPc
MgCl2
MgCl2
MgCl2
toluene
MgBr2
2-methoxyethanol
MgBr2
NMPc
99
1.8
5.7
2.2
10.1
0
MgBr2
diethoxymethane
NMPc
97
MgBr2 ·OEt2
MgBr2 ·OEt2
MgBr2
96
diethoxymethane
96
2,2-DMPb
79
MgBr2
toluene
toluene
1,2-dimethoxyethane
1,2-dimethoxyethane
1,2-dimethoxyethane
62
91:9
70:30
nd
nd
nd
21
MgBr2 ·OEt2
MgBr2 ·OEt2
MgCl2
53
7
<5
nd
<5
nd
MgBr2
<5
nd
a 2-Methyltetrahydrofuran. b 2,2-Dimethoxypropane. c N-Methylpyrrolidinone.
1450
•
Vol. 14, No. 6, 2010 / Organic Process Research & Development

was in solution, reduction of the nonchelated conformer 2a
would dominate, leading to the formation of dist-3 and a lower
ratio of 3:dist-3. Residual water also adversely impacted the
overall yield and observed selectivity. The isolated yield of 3
dropped from 90% to 86% when the water content of the
palladium on carbon increased from 1.3% to 3.6%,22 and use
of 50% water-wet catalyst caused the ratio of 3:dist-3 to
decrease to approximately 70:30. The decreased yield and
selectivity were presumably an impact of hydrolysis of the
Lewis acid that resulted in there being less available for
chelation.
reported robust reaction conditions and expect that this meth-
odology could be useful for related substrates.
Experimental Section
General. Reaction monitoring was carried out on a Waters
Alliance HPLC with PDA monitoring (215 nm) equipped with
a Waters Symmetry C18 column (4.6 mm × 150 mm) at 25
°C. Elution was performed with a flow rate of 1.0 mL/min and
an isocratic method using 30% mobile phase A (water/
acetronitrile, 95:5) and 70% mobile phase B (water/acetonitrile,
5:95). The mobile phase was modified with 0.07% NH4Cl as
buffer. Observed retention times were as follows: MgBr2 (1.4
min), 6 (5.9 min), 5 (6.2 min), dist-3 (6.4 min), 3 (6.8 min), 2
(7.2 min).
Conclusion
A Lewis acid-mediated hydrogenation of an N-acryloylox-
azolidin-2-one was developed. Addition of specific Lewis acids
to the reaction mixture caused the facial selectivity of the
hydrogenation to be reversed. Although magnesium bromide
was used successfully on a small laboratory scale, stress tests
showed that the product was degraded by the magnesium
bromide when processing times were extended. Magnesium
chloride was identified as a suitable replacement; however,
2-methyltetrahydrofuran was required as the solvent, and the
reaction temperature had to be increased to 75 °C. The process
was scaled up to a kilo-lab, and over the 18 batches run to date,
the isolated yield has averaged 85%.
The change in facial selectivity may be best explained on
the basis of a shift in the preferred conformer of the starting
material in solution caused by Lewis acid coordination of the
two carbonyl groups. Attempts to further elucidate the reaction
mechanism using either NMR or in situ FTIR were not
successful. The in situ FTIR profile and heat-flow pattern did
reveal that the reduction was slower when conducted in the
presence of magnesium chloride. These data are consistent with
Lewis acid coordination of the carbonyl groups that would lead
to lower electron density of the carbon-carbon bond.23 Al-
though we have not been successful in obtaining direct evidence
that supports the proposed reaction mechanism, we have
Reaction screening was performed in an Endeavor Catalyst
Screening System from Biotage/Argonaut using 10-mL cells
with a fill volume of 3 mL. Each experiment was performed
using 0.2 g of 2 and 1.2 molar equiv of chelator. The vessels
were adjusted to the specified operating temperature at 250 rpm
and held at that temperature before hydrogen was introduced.
Upon introduction of hydrogen (60 psig), stirring was increased
to 500 rpm, and these conditions were maintained through the
course of the experiment.
Lab-Scale Demonstration. To a 2-L Parr pressure reactor,
compound 2 (82 g, 200 mmol), dry palladium on carbon (10%
Pd, 8.3 g, 7.8 mmol), MgCl2 (23 g, 240 mmol, 1.2 equiv), and
2-methyltetrahydrofuran (1230 mL, 15 parts v/w) were added.
The mixture was heated to 70 °C and held for 20 min, and the
atmosphere inside the pressure vessel was converted to hydro-
gen (55 psig). The reaction mixture was stirred at 70 °C,
maintaining hydrogen pressure at 55 psig until the conversion
to 2 to 3 was greater than 95% (AN HPLC, typically 1 h). The
reactor was purged of hydrogen, inerted with nitrogen, and
cooled to 40 °C. The reaction mixture was filtered through Celite
(82 g, 1 part w/w) at 40 °C under nitrogen (Caution: Spent
catalyst is pyrophoric!), and the reactor and cake were rinsed
forward with MTBE (820 mL, 10 parts v/w). The combined
filtrate was washed with brine (410 mL, 5 parts v/w) to remove
MgCl2 and was concentrated to a volume (330 mL, 4 parts v/w)
at 40 °C under vacuum. To the concentrate was charged IPA
(820 mL, 10 parts v/w). The mixture was concentrated to 330
mL (4 parts v/w) at 40 °C under vacuum. The concentrate was
heated to 70-80 °C to give a cloudy solution, and water was
added water (160 mL, 2 parts v/w) to afford a suspension. The
resulting suspension was cooled to ambient temperature and
filtered. The filter cake was dried at 60 °C under vacuum to
give 71.8 g of 3 as a white solid in 87% yield with purity of
98% (w/w). The material was found to be identical to that
prepared previously.1
(16) In process analysis of a scale-up run performed on 1 kg of 2 showed
1.6% 2, 4.8% 7, and 5.0% 6 after 4 h of reaction.
(17) Magnesium bromide had sufficient absorption in the UV region that
it could be monitored by HPLC. After three brine washes, the organic
layer still contained 2 area % MgBr2.
(18) McConville, F. X. The Pilot Plant Real Book; FXM Engineering and
Design: Worcester, MA, 2002.
(19) For screening purposes, all reductions were performed in THF at 50
°C with 50 psig hydrogen using dry palladium on carbon (10 wt %
Pd) and 1.2 equiv of additive.
(20) Magnesium bromide diethyletherate was examined when the Process
Safety Group raised concerns regarding the high heat of dissolution
of MgBr2 in THF. Use of pre-solvated magnesium bromide was
expected to lower the exotherm noted when MgBr2 and THF were
mixed.
(21) A total of 32 kg of 3 was prepared in six batches. Each batch of 3
required three hydrogenations of 2 that were then combined, providing
isolated batch sizes of 5.0-5.5 kg of 3.
(22) Palladium on alumina was also evaluated for this process. It offers
the advantages of being dry and safer to handle than dry palladium
on carbon. In our evaluation, the reaction took slightly longer, but the
isolated yield was approximately 88%. The decision to scale up the
dry palladium on carbon was based largely on the fact that the dry
palladium on carbon had been ordered to support the MgBr2 campaign
and was in stock.
(23) Partial NMR spectra for starting material in the presence and absence
of magnesium salts, and in-situ FTIR and heat-flow data for hydro-
genations performed in the presence and absence of magnesium
chloride are available as Supporting Information.
Kilolab-Scale Experiment. To a 40-L hydrogenator were
charged compound 2 (2.1 kg, 5.1 mol), dry palladium on carbon
(10% Pd, 0.21 kg) (Caution: Dry catalyst is pyrophoric!),
anhydrous MgCl2 (0.59 kg, 6.2 mol, 1.2 equiv), and 2-meth-
yltetrahydrofuran (27.1 kg). The mixture was heated to 70-75
°C and held for 15 min, and then the atmosphere inside the
pressure vessel was converted to hydrogen (60-65 psig) while
the contents were stirred at 900-1100 rpm. The reaction
mixture was held under these conditions until the pressure drop
on the supply line slowed to <10 psig/15 min. The reactor
Vol. 14, No. 6, 2010 / Organic Process Research & Development
•
1451

contents were cooled to 40-45 °C, and the atmosphere inside
the reactor was converted from hydrogen to nitrogen. The
reaction was sampled for HPLC, and 0% 2 was detected after
2.5 h. The reactor contents were filtered through a Sparkler filter
(Caution: Spent catalyst is pyrophoric!), and then the reactor
and filter were rinsed forward with MTBE (15 kg). The filtrate
was collected in a common receiving vessel and held for later
processing. The hydrogenations were repeated twice more on
the same scale. When the third reduction was complete, the
combined filtrates were washed with a solution of water (32
kg) and sodium chloride (3.1 kg). The organic layer was
separated and concentrated to 20-25 L final volume under
vacuum with a maximum jacket temperature of 35 °C (max.
pot temperature was 21 °C). 2-Propanol (50 kg) was charged,
and the reactor contents were reduced to a final volume of
20-25 L.24 The reactor contents were adjusted to 70-80 °C,
and water (12 kg) was added over 10 min while the batch
temperature was maintained at 65-80 °C. The reactor contents
were cooled to 20-25 °C over 1 h and then filtered and rinsed
with a mixture of 2-propanol (5.0 kg) and water (6.3 kg). The
product was dried on the filter for 24 h at ambient temperature
under a stream of nitrogen and was discharged.25 The filter was
discharged to provide 3 as a white solid, 5.64 kg (89.5% yield)
with a purity >99% (w/w).
Acknowledgment
We thank the Pearl River kilogram laboratory team, Process
Safety Group and Analytical and Quality Science team members
for their support of this work.
Supporting Information Available
Partial NMR spectra for starting material 2 in the presence
and absence of magnesium salts, in situ FTIR and heat-flow
data for hydrogenations of 2 performed in the presence and
absence of magnesium chloride, details of the preparative LC
separation conditions used to isolate 5 and 6 and NMR data
used to identify 5 and 6 are available as Supporting Information.
This material is available free of charge via the Internet at
http://pubs.acs.org.
Received for review January 29, 2010.
OP100074M
(24) In-process GC analysis showed there to be 1.9% 2-methyltetrahydro-
furan present at this point.
(25) After 24 h, the product had a water content of 0.03% (KF method)
and residual IPA of 0.11%.
1452
•
Vol. 14, No. 6, 2010 / Organic Process Research & Development