A practical enantioselective synthesis of a novel peptide deformylase inhibitor

Article abstract of DOI:10.1021/op050165y

A practical synthesis of the peptide deformylase inhibitor LBM415,(2S)-N-(5-fluoro-1-oxido-2-pyridinyl)-1-[(2R)-2-[(formylhydroxyamino) methyl]-1-oxohexyl]-2-pyrrolidinecarboxamide, magnesium salt 11, is described. The key chiral intermediate, (2S)-N-(5-f

Full text of DOI:10.1021/op050165y

Organic Process Research & Development 2006, 10, 78−93
A Practical Enantioselective Synthesis of a Novel Peptide Deformylase Inhibitor
Joel Slade,*,† David Parker,† Michael Girgis,† Martin Mueller,† James Vivelo,† Hui Liu,† Joginder Bajwa,†
Guang-Pei Chen,† Joseph Carosi,‡ Paul Lee,‡ Apurva Chaudhary,‡ Dana Wambser,| Kapa Prasad,† Kathryn Bracken,§
Karl Dean,§ Helmut Boehnke,‡ Oljan Repicˇ,† and Thomas J. Blacklock†
Process R & D, Chemical and Analytical DeVelopment and Pilot Plant Operations, Chemical and Analytical
DeVelopment, NoVartis Institutes for Biomedical Research, One Health Plaza, East HanoVer, New Jersey 07936, U.S.A.,
and Department of Infectious Diseases, NoVartis Institutes for Biomedical Research, 100 Technology Square,
Cambridge, Massachusetts 02139 U.S.A.
Abstract:
A practical synthesis of the peptide deformylase inhibitor
LBM415,(2S)-N-(5-fluoro-1-oxido-2-pyridinyl)-1-[(2R)-2-[(formyl-
hydroxyamino)methyl]-1-oxohexyl]-2-pyrrolidinecarboxamide,
magnesium salt 11, is described. The key chiral intermediate,
(2S)-N-(5-fluoro-2-pyridinyl)-1-[(2R)-2-[[formyl(phenylmeth-
oxy)amino]methyl]-1-oxohexyl]-2-pyrrolidinecarboxamide 8, was
Figure 1. LBM415.
made by coupling the corresponding amino acid 7 with the
carboxamide 25 prepared from -proline and 2-amino-5-
L
Due to the straightforward nature of this route, it was
decided to retain the overall strategy for the construction of
the molecule for the first Pilot Plant campaign, while
employing the necessary modifications to allow for an
acceptable fit in the pilot plant. Mainly, we were concerned
with the stability of the drug substance along with the large
number of chromatographic purifications required and the
use of expensive, environmentally unfriendly or undesirable
solvents and reagents [e.g., O-(7-azabenzotriazol-1-yl)-
N,N,N′,N′-tetramethyluronium hexafluorophosphate (HATU),
m-chloroperoxybenzoic acid (MCPBA), and methylene chlo-
ride]. The observed instability of the free acid form 10 was
particularly important since no solid form of the compound
had as yet been prepared. Thus, in addition to the develop-
ment of a viable process for the production of this material,
we also sought to identify a solid, stable, crystalline
derivative. In this paper we provide a full description of the
synthesis of LBM415, obtained in 16% overall yield in 13
steps.
fluoropyridine. Following oxidation of the pyridine nitrogen,
selective hydrogenolysis of the benzyl group afforded the free
acid of the drug substance, which was converted to the
magnesium salt in situ with magnesium chloride. The product
was obtained in an overall yield of 16% with an ee > 99%.
Introduction
LBM415-NX (Figure 1) has been shown to be an
extremely potent inhibitor of peptide deformylase (PDF)1 in
vitro while at the same time showing good antibacterial
activity in vivo. Importantly, LBM415 was found to be active
against pathogenic bacterial strains that have acquired
resistance to other antibacterial agents. This PDF inhibitor
has the potential to be developed into an antibacterial agent
with the ability to eradicate many bacterial infections,
including those that are refractory to treatment with existing
agents. We have developed a practical synthesis to produce
multikilogram quantities of this potential medicinal agent as
a single enantiomer.
The Discovery synthesis of the free acid form 10 2 was
designed in a modular fashion and was obtained in 9 linear
(12 total) steps from mostly readily available starting
materials (see Scheme 1).
Results and Discussion
Following the route elucidated by the Drug Discovery
group, our synthesis of LBM415 is illustrated in Scheme 2.
n-Butylacrylic acid 2 was prepared from n-butylmalonic acid
1 and aqueous formaldehyde via a Knoevenagel condensa-
tion3,4 using piperidine as the base. The oily product 2 was
obtained in sufficient quality so that no further purification
was necessary. In the original procedure developed in Drug
Discovery, oxazolidinone chemistry5 was used to prepare 3
which was purified by preparative chromatography followed
by recrystallization from ether/hexane.
* To whom correspondence should be addressed. E-mail: joel.slade@
pharma.novartis.com.
† Process R & D, Chemical and Analytical Development.
‡ Pilot Plant Operations, Chemical and Analytical Development.
§ Department of Infectious Diseases.
| Current address: Colgate-Palmolive, Morris Plains, New Jersey 07950.
(1) (a) For a discussion of PDF inhibitors, see: Huntington, K. M.; Yi, T.;
Wei, Y.; Pei, D. Biochemistry 2000, 39, 4543 and references therein. (b)
Pratt, L. M.; Beckett, R. P.; Davies, S. J.; Launchbury, S. B.; Miller, A.;
Spavold, Z. M.; Todd, R. S.; Whittaker, M. Bioorg. Med. Chem. 2001, 11,
2585.
In addition to the obvious problems associated with
preparative chromatography, there were several other aspects
(2) (a) Patel, D. V.; Yuan, Z.; Jain, R. K.; Garcia Alvarez, S.; Jacobs, J. PCT
Int. Appl. WO 102790, 2002, 69 pp. (b) Patel, D. V.; Yuan, Z.; Jain, R. K.;
Garcia Alvarez, S.; Jacobs, J. PCT Int. Appl. WO 102791, 2002, 52 pp. (c)
Jain, R.; Sundram, A.; Lopez, S.; Neckerman, G.; Wu, C.; Hackbarth, C.;
Chen, D.; Wang, W.; Ryder, N. S.; Weidmann, B.; Patel, D.; Trias, J.; White,
R.; Yuan, Z. Bioorg. Med. Chem. Lett. 2002, 12, 3595.
(3) Jones, G. Org. React. 1967, 15, 204.
(4) Wilk, B. K. Tetrahedron 1997, 53, 7097.
(5) (a) Gage, J. R.; Evans, D. A. Org. Synth. 1989, 68, 77. (b) Evans, D. A.
Aldrichchimica Acta 1982, 15, 23.
78
•
Vol. 10, No. 1, 2006 / Organic Process Research & Development
10.1021/op050165y CCC: $33.50 © 2006 American Chemical Society
Published on Web 12/16/2005

Scheme 1. Discovery synthesis of LBM415
of this procedure which needed to be addressed. These were
as follows:
(i) the use of low temperature (this would result in lower
throughput due to the size of the low-temperature reactor in
the pilot plant);
(ii) the need to recrystallize from diethyl ether/hexane;
and
(iii) the use of n-butyllithium which would generate large
quantities of butane gas in the pilot plant.
As the unreacted (S)-4-benzyl-2-oxazolidinone could not
be removed by recrystallization, we reasoned that if all of
the oxazolidinone could be consumed in the reaction, then
purification would be simplified. Based on this assumption,
a series of experiments was carried out in which the
2/oxazolidinone ratio was increased until all of the latter
reagent was consumed. By using a 10% excess of acid 2,
complete consumption of the oxazolidinone occurred. The
excess acid present in the product could be easily removed
using a basic workup procedure.
To make the isolation of 3 safe, the original crystallization
solvents (ether/hexane) were replaced by methyl tert-butyl
ether/heptane (1/9). This mixture afforded a white crystalline
Vol. 10, No. 1, 2006 / Organic Process Research & Development
•
79

Scheme 2. Process R & D synthesis of LBM415
solid in 75% overall yield with high purity (>98.5% by
HPLC area normalization). To prevent the buildup of static
charge during the centrifugation of the batch, 5% ethanol in
heptane was used as the final cake wash.
80
•
Vol. 10, No. 1, 2006 / Organic Process Research & Development

Scheme 3. Preparation of compound 3
Another important issue to be addressed was that of
throughput. Since n-butyllithium was used as the base to
deprotonate the oxazolidinone, a reaction temperature of -78
°C was required to prevent side reactions, the main one being
nucleophilic attack at the carbonyl of the oxazolidinone ring.
This necessitated the use of the low temperature reactor in
the pilot plant which only had limited capacity. Another
drawback to n-BuLi was the fact that butane was produced
as a byproduct. This was not desirable in the pilot plant due
to the flammability of this gas. To address this problem
lithium diisopropylamide (LDA) was investigated. The use
of LDA had two advantages: no flammable gas was released
on deprotonation, and the selectivity was preserved even at
-20 °C with the formation of little or no side products. In
our initial experiments, the LDA was used as a 2.0 M
solution in THF/heptane/ethylbenzene. However, at the
outset, we encountered some difficulties in reproducing both
the yield and purity of this reaction. Sometimes the oxazo-
lidinone was consumed, and in some cases, as much as 4%
still remained. Even this amount was found to cause problems
with the isolation (vide supra).
In conjunction with this, we observed that during the
addition of the LDA to the oxazolidinone, the color of the
reaction at the end point was variable. Sometimes it was a
clear yellow solution, and sometimes it was a dark red-brown
solution. Further experimentation revealed that when the end
point was yellow, not all of the oxazolidinone was consumed,
but when the color changed to red-brown, no oxazolidinone
could be detected. Furthermore we knew from previous
experience that if less than 1 equiv of LDA was used, the
residual oxazolidinone could not be removed easily by
recrystallization and that if excess LDA was employed, the
yield and purity of 3 would decrease.
min. If the color became yellow again, additional quantities
of LDA were added until the red-brown color persisted. The
anion mixture was then transferred to the solution containing
the mixed anhydride (Scheme 3). This procedure was found
to work well in the pilot plant as the color change was easily
observed through the sight glass.
For the preparation of intermediate 5, the original Drug
Discovery protocol was modified. Thus, 2 equiv of O-
benzylhydroxylamine were reacted at 50 °C with 3 in the
absence of solvent. Ethyl acetate and 4 equiv of p-toluene-
sulfonic were added to precipitate the toluenesulfonic salt
of O-benzylhydroxylamine. The solvent was then replaced
with MTBE, and the desired diastereomer crystallized in 69%
yield. A sample of the undesired diastereoisomer was isolated
by chromatography from the mother liquors (NMR as well
as HPLC indicated that the desired diastereoisomer 5 was
obtained in greater than 99% diastereomeric purity).
One other problem which we addressed was the exother-
micity of the reaction. Calorimetric studies indicated that an
exotherm was occurring at 33 °C (∆H > 29 J/g) as the
reaction was being heated. Since the reaction mixture was a
thick oil, it was feared that heat might accumulate in the
reactor due to the poor heat transfer to the cooling medium.
To eliminate this possibility, the reaction was heated in stages
over a 3-h period until the desired temperature (50 °C) was
reached.
The diastereoselective conjugate addition of an achiral
amine nucleophile to a chiral R,â-unsaturated carbonyl
component is well-known in the literature.7 Specifically for
the case of the conjugate addition of O-benzylhydroxylamine
or its analogues, several examples have been reported.8-13
In the present case, we feel that the ratio of the
diastereomers is set based on the protonation of the si face
which is favored for steric reasons. Fortuitously, the desired
diastereoisomer is a solid and crystallizes from the reaction
mixture. The selective enolate protonation argument has been
Although the concentration of each lot of LDA could be
ascertained, we felt that use of the color change as a guide
to the completeness of the deprotonation would be more
reliable. Thus, the following protocol was used for the
development of the process. The dark LDA solution was
added slowly to the oxazolidinone in THF at -20 °C. The
addition was continued until the color of the solution changed
from yellow to dark red-brown.6 At this point, the addition
was stopped and the mixture was held at -20 °C for 15
(7) (a) For a recent review, see: Sibi, M. P.; Liu, M. Tetrahedron 2002, 58,
7991. (b) Juaristi, E.; Garcia-Barradas, O. In EnantioselectiVe Synthesis of
â-Amino Acids; Juaristi, E., Ed.; Wiley-VCH: New York, 1997; Chapter
7.
(8) Ishikaiwa, T.; Nagai, K.; Kudoh, T.; Saito, S. Synlett 1995, 1171.
(9) Volonterio, A.; Bravo, P.; Zanda, M. Org. Lett. 2000, 2, 1827.
(10) Ishikawa, T.; Nagai, K.; Kudoh, T.; Saito, S. Synlett 1998, 1291.
(11) Fallborg, L.; Jorgenson, K. A. J. Chem. Soc., Perkin Trans. 1 1996, 2823.
(12) Sibi, M. P.; Shay, J. J.; Liu, M.; Jasperse, C. P. J. Am. Chem. Soc. 1998,
120, 6615.
(6) A similar color change has been noted for the N-acylation of an oxazoli-
dinone using n-BuLi; see: Hoekstra, M. S.; Sobieray, D. M.; Schwindt, M.
A.; Mulhern, T. A.; Todd, M. G.; Huckabee, B. K.; Hendrickson, V. S.;
Franklin, L. C.; Granger, E. J.; Karrick, G. L. Org. Process Res. DeV. 1997,
1, 26.
(13) Cardillo, G.; Gentilucci, L.; Gianotti, M.; Kim, H.; Perciaccante, R.;
Tolomelli, A. Tetrahedron: Asymmetry 2001, 12, 2395.
Vol. 10, No. 1, 2006 / Organic Process Research & Development
•
81

Scheme 4. Discovery route to amino acid 7 free base: formylation of compound 5
Scheme 5. Removal of the chiral auxiliary from 12
Scheme 6. Development route to intermediate 7
put forth to explain the stereochemistry of the nucleophilic
conjugate addition of carbon nucleophiles.14
In the original Discovery synthesis, the formylation step
was executed prior to the removal of the chiral auxiliary
(Scheme 4). Thus, a formic acid solution of 5 was treated
with a mixture of formic acid/acetic anhydride (15 equiv/2
equiv) at 0 °C. After an aqueous workup and preparative
chromatography, product 12 was isolated in 40% overall
yield. The major impurity in the crude product (10%) was
identified as the N-acetyl derivative 13.
This procedure was modified by carrying out the formyl-
ation at -15 °C. When the reaction was conducted in THF
using 8 equiv of formic acid and 2 equiv of acetic anhy-
dride,15 12 was obtained in 75% yield with only 1.5% of
the N-acetyl impurity 13 present.
However, problems were encountered in the following
step, i.e., removal of the chiral auxiliary (Scheme 5). Under
the standard reaction conditions (2.3 equiv of 30% hydrogen
peroxide and 1.2 equiv of lithium hydroxide)16 14 was
obtained in low yield and purity. Along with the chiral
auxiliary, the product was contaminated with intractable polar
materials and as much as 10% of the deformylated side
product 15. The latter was identified by comparison with an
authentic sample and may have resulted from the reaction
between the product 14 and the lithium hydroperoxide.
In addition to the difficulties mentioned above, we found
that 14 was a rather unstable oil which had to be kept from
light and stored cold. Because of these problems, it was
decided to reverse the order of the steps and remove the chiral
auxiliary prior to formylation (see Scheme 6).
It was expected that this approach would produce a less
deformylated side product and also allow us to separate the
oxazolidinone from 6 by acid/base extraction. In addition, it
was thought that the properties of 14 could be improved by
preparing a stable salt, thus simplifying the isolation of this
important intermediate.
This approach was found to be much more suitable for
scale-up; however, several problems still existed. In addition
to those associated with the use of hydrogen peroxide in the
pilot plant (e.g., liberation of oxygen from the reaction
mixture due to decomposition of the peroxide17), the stability
of 6 was also a concern. In one instance in the laboratory, a
solution of 6 was concentrated at 50-55 °C on the rotary
(14) Mohrig, J. R.; Fu, S. S.; King, R. W.; Warnet, R.; Gustafson, G. J. Am.
Chem. Soc. 1990, 112, 3665.
(15) Wilkinson, H. S.; Hett, R.; Tanoury, G. J.; Senanayake, C. H.; Wald, S. A.
Org. Process. Res. DeV. 2000, 4, 567.
(16) Evans, D. A.; Britton, T. C.; Ellman, J. A. Tetrahedron Lett. 1987, 28,
6141.
(17) Astbury, G. R. Org. Process Res. DeV. 2002, 6, 893.
82
•
Vol. 10, No. 1, 2006 / Organic Process Research & Development

Scheme 7. Reaction of 25 with morpholinium salt 17
evaporator. Upon removal, the neat oil was observed to begin
offgassing and the internal temperature rose to 150 °C. The
sample changed from a colorless oil to a yellow solid.
Analysis by HPLC, NMR, and MS indicated that 6 had
decomposed to a variety of products. The main constituent
was identified as the â-amino acid 16. Although a definitive
explanation is not yet available, we feel that the decomposi-
tion may have been brought about by residual amounts of
sodium sulfite (used in the work up to decompose the
hydrogen peroxide) which could, with heat, cleave the N-O
bond of the hydroxylamine.
For the preparation of 25, the protected proline 23 was
coupled with 2-amino-5-fluoro-pyridine using CDMT, and
the resulting amide 24, deprotected with HBr/HOAc. The
latter conditions were chosen because it was found that the
free base obtained by hydrogenolysis of 24 was low melting
and difficult to isolate in pure form, whereas the HBr salt
was a crystalline solid.
Initially, we envisioned using 2-chloro-4,6-dimethoxy-
1,3,5-triazine (CDMT)18 as the coupling agent for the
synthesis of 8, since we had employed this reagent success-
fully on several previous projects. Thus, 7 and 25 were
allowed to react in acetonitrile in the presence of a slight
excess of CDMT and 4 equiv of N-methylmorpholine.
Analysis of the crude product indicated that it contained 54%
of 8 and 33% of an impurity. The impurity was identified
as 18, the product resulting from the reaction of 25 with the
morpholinium salt 17 (Scheme 7).19 Apparently, the proline
amide 25 can compete effectively with the carboxylic acid
7 for the morpholinium salt. Allowing the acid 7 to react
with the salt at room temperature prior to the introduction
of 25 reduced the level of this impurity to 15%.
Due to the fact that 18 could only be removed from the
product by chromatography, alternative conditions for the
coupling reaction were investigated. Other coupling agents
tried were 1-propanephosphonic acid cyclic anhydride (T3P),20
isobutyl chloroformate,21 and 1-[3-(dimethylamino)propyl]-
3-ethylcarbodiimide hydrochloride (EDCI).22 Of these, the
latter reagent was the only one which afforded appreciable
amounts of product.
During the early part of our development work for this
step, we observed the formation of a new impurity which
was rather difficult to remove. Using NMR and LC/MS, the
compound was identified as the acylated urea 20. This is a
rather common type of side product which is observed in
carbodiimide-mediated peptide bond formation reactions and
comes about by rearrangement of the adduct 19 (Scheme
8).23 The formation of this compound was suppressed by
adding HOBt to the reaction mixture.24 Thus, the rearrange-
ment of 19 was precluded by the reaction of this intermediate
With regard to the use of aqueous hydrogen peroxide in
the pilot plant, it was felt that the best way to preclude the
accumulation of oxygen in the headspace above the reaction
mixture was to use a rapid nitrogen sweep across the batch,
thus maintaining the level below 8-10% (v/v) (the minimum
oxygen level necessary for combustion).17 To monitor the
oxygen level, the reactor to be used was fitted with an oxygen
analyzer.
Because of our laboratory experiences with 6, a significant
number of safety tests were carried out to be certain that
this intermediate could be handled safely on a large scale. It
was found that the neat oil was very unstable with an onset
exotherm at 48 °C (both Radex and DSC) with a 612-823
J/g energy release. In addition, it was observed that samples
of 6 became less stable over time (lower exotherm detection
temperature) with decomposition to the â-amino acid being
observed. Based on the stability data, it was decided to store
intermediate 6 as a cold ethyl acetate solution and to use it
as soon as possible.
Intermediate 6 was N-formylated using the conditions
mentioned above. However, to improve the yield and reduce
the amount of the N-acetyl impurity present, the reaction
was carried out at lower temperature (-15 °C) and the
amount of formic acid was increased from 4 to 8 equiv.
As with 6, the free acid was unstable to heat and light.
Therefore, the compound was isolated as the dicyclohexyl-
amine salt 7. To further streamline the process, the formyl-
ation step was carried out in ethyl acetate so that no solvent
exchange or removal was necessary. The overall yield for
the two steps was 74%, and the purity was greater than 99%
(by HPLC area normalization).
(18) (a) Kaminski, Z. J. Synthesis 1987 917. (b) Garrett, C. E.; Jiang, X.; Prasad,
P.; Repic, O. Tetrahedron Lett. 2002, 43, 4161.
(19) Kaminski, Z. J.; Panetti, P.; Rudzinski, J. J. Org. Chem. 1998, 63, 4248.
(20) Wissman, H.; Kleiner, H. J. Angew. Chem. 1980, 92, 129.
(21) For a review, see: Ray, T. In Encyclopedia of Reagents for Organic
Synthesis; Paquette, L., Ed.; John Wiley and Sons: New York; Vol. 4, p
2752. In the DSC, an exotherm was noted beginning at 150 °C [∆H ≈
1129 J/g].
(22) Sheehan, J. C.; Cruickshank, P. A.; Boshart, G. L. J. Org. Chem. 1961, 26,
2525.
(23) Lloyd-Williams, P.; Albericio, F.; Giralt, E. Chemical Approaches to the
Synthesis of Peptides and Proteins; CRC Press: Boca Raton, FL, 1997; p
50.
(24) Ko¨nig, W.; Geiger, R. Chem. Ber. 1970, 103, 788.
Vol. 10, No. 1, 2006 / Organic Process Research & Development
•
83

Scheme 8. Rearrangement of EDCI adduct 19
comparable or even higher rates than the amine to afford
mixtures of starting material and product. It was concluded
that the amine/N-oxide ratio could be dependent on the
nucleophilicity of the N-oxide. Addition of an excess of
dioxirane would result in the decomposition of the N-oxide
to release 1O2. The authors proposed the mechanism shown
in Scheme 11. Using this hypothesis, the reaction of Oxone
in acetone/water (dimethyldioxirane) with 8 could be rep-
resented in the way shown in Scheme 12.
As a final verification that this was indeed occurring, a
sample of 9, free of 8, was subjected to the oxidation
conditions described above. After an overnight hold at room
temperature, the reaction mixture was found to contain 1.4%
of 8. Thus, the reversibility of this reaction was proven, and
we concluded that it would not be possible under these
conditions to effect complete conversion of 8 to 9 due to
the apparent reversibility of the reaction brought about by
the propensity of the product to react with the excess oxidant.
As an alternative to Oxone, we decided to investigate the
use of the urea/hydrogen peroxide/phthalic anhydride (UHP)28
system for this oxidation since the oxidation of 4-tert-
butylpyridine to the N-oxide had been reported using these
conditions.29
The advantages to using this reagent mixture were as
follows: the mild conditions, the stability of the solid oxidant,
the ease of handling, and the fact that the byproducts were
water and base-soluble. Since UHP has only limited solubility
in nonpolar organic solvents, we chose to investigate
acetonitrile and ethyl acetate since the polarity of these two
solvents would be sufficient to solubilize the reagent so that
the reaction would proceed at a reasonable rate.
In both solvents the reactions were complete after 7 h at
room temperature using 2 equiv of UHP and 3 equiv of
phthalic anhydride. Since the results were similar, ethyl
acetate was chosen as the reaction solvent, since the workup
using this solvent was simpler (aqueous sodium carbonate
wash).
Due to the fact that small amounts of water can increase
the solubility of UHP in ethyl acetate and that a higher
concentration of UHP in solution should accelerate the
reaction rate, the reaction conditions were modified to include
the addition of 10% water by volume to the mixture. This
reduced the reaction time to less than 5 h. As mentioned
previously, since a method for the crystallization of 9 was
not yet available, an assay method was developed, and the
material was carried forward to the next step as an ethyl
acetate solution.
with HOBt to form the HOBt ester which then reacted with
25 to give 8.
As anhydrous HOBt dust is explosive in nature,21 we
conducted the reaction in a two-phase mixture of ethyl acetate
and water (2.5/1 v/v). The water served to remove polar
impurities and allowed for a simpler workup and purification.
The product 8 was then recrystallized to high purity with
only minimal losses.
The original procedure used for the oxidation of 8
(MCPBA/methylene chloride) was not considered for reasons
of safety and ecology. One major requirement with regard
to any new reaction conditions was that complete conversion
of 8 to 9 was essential. This was because during the following
debenzylation step, any compound 8 remaining would also
debenzylate to give the same impurity one would obtain from
cleavage of the N-O bond of the pyridine N-oxide (see
Scheme 9). We had previously observed the latter side
reaction during our laboratory work and had found that 21
could only be removed by chromatography (at this point,
we had yet to isolate 9 as a solid).
Initially, we investigated the use of Oxone (potassium
hydrogen persulfate) for carrying out this oxidation.25 Based
on these early studies we found that the best conditions
involved the use of an acetone/water mixture with potassium
bicarbonate as a base to reduce the acidity of the reagent.
Otherwise, protonation of the pyridine nitrogen would occur
rendering it inactive to oxidation. The actual oxidant in this
case is dimethyldioxirane 22 (see Scheme 10).26 Although
the reaction worked reasonably well, it could not be driven
to completion even when a large excess of oxidant was
employed. The best we could accomplish using 4-5 equiv
of Oxone and long reaction times (2-4 days) was about 98%
conversion with 1-2% 8 still remaining.
Recently, a report appeared27 in which the authors
described the release of singlet oxygen from the reaction of
dimethyldioxirane with amine N-oxides. In this case it was
observed that when dimethyl dioxirane was used to prepare
amine N-oxides, in some instances it was not possible to
achieve complete conversion of the amine to the N-oxide
despite the use of a large excess of dimethyl dioxirane.
According to the experimental data, the authors concluded
that the oxidation of tertiary amines and nitrogen heteroarenes
by dioxiranes to N-oxides may be a reversible process. The
in situ generated N-oxide may react with the dioxirane at
Conversion of 9 to LBM415. In the Discovery synthesis,
the debenzylation of 9 was accomplished by catalytic
hydrogenation under 1 atm of hydrogen using 10% Pd/C in
a 1:1 mixture of ethyl acetate/ethanol. Under these conditions,
approximately 10% of 21 was formed resulting from cleavage
of the pyridine N-O bond. Since 10 was not a solid, the
only method of purification was preparative chromatography.
Upon removal of the mobile phase, the drug substance was
isolated as a solidified foam in moderate yield.
(25) Bell, T. W.; Young-Moon, C.; Firestone, A.; Healy, K.; Liu, J.; Ludwig,
R.; Rothenberger, S. D. Org. Synth. 1990, 69, 226.
(26) Murray, R. W.; Jeyaraman, R. J. Org. Chem. 1985, 50, 2847.
(27) Ferrer, M.; Sa´nchez-Baeza, F.; Messeguer, A.; Adam, W.; Golsch, D.; Go¨rth,
F.; Kiefer, W.; Nagel, V. Eur. J. Org. Chem. 1998, 2527.
(28) For a review, see: Heaney, H. Aldrichchimica Acta 1993, 26, 35.
(29) Kaczmarek, L.; Balicki, R.; Nantka-Namirski, P. Chem. Ber. 1992, 125,
1965.
84
•
Vol. 10, No. 1, 2006 / Organic Process Research & Development

Scheme 9. Debenzylation of compound 9
Scheme 10. Oxidation with dimethyldioxirane in acetone
(300 g) was not feasible due to the instability of 10 to
prolonged exposure to silica gel. To circumvent this problem,
alternative hydrogenolysis conditions were sought which, it
was hoped, would minimize the undesired side reaction
(reduction of the N-oxide) and yet still promote the deben-
zylation reaction.
At this point, we did not have a scaleable procedure for
the final step, nor did we have a method of isolating the
drug substance in a solid form. Thus, a parallel investigation
was initiated wherein various hydrogenolysis conditions were
tried while, at the same time, salt screening was carried out.
Switching the catalyst from Pd/C to 5% Pd/BaSO4 (40
wt % catalyst loading) in ethyl acetate afforded between 2%
and 3% of the deoxy impurity 21. Both commercial and in
situ reduced Pd/BaSO4 were tried. With the Lindlar catalyst
(Pd/CaCO3 with Pb), the reaction was too slow to be
practical. Using a more polar solvent such as ethanol served
Scheme 11. Reaction of N-oxides with excess
dimethyldioxirane
This was acceptable for small-scale preparations; however,
we learned that chromatographic purification on a large scale
Scheme 12. Reaction of 9 with dimethyldioxirane
Vol. 10, No. 1, 2006 / Organic Process Research & Development
•
85

Figure 2. Kinetics of the debenzylation of 9 and the formation of 10 in various solvents.
to increase the amount of the deoxy impurity up to as much
as 16%.
9, 10, and the deoxy impurity 21 using external standards.
The reactivity and selectivity in more polar (EtOH, THF)
and less polar solvents (MTBE) were also investigated
(keeping reactant and catalyst loadings as well as pressure
constant). The debenzylation was significantly faster in EtOH
and THF relative to EtOAc (see Figure 2).
Thus, the best conditions for this reaction involved the
use of unreduced 5% Pd/BaSO4 (40 wt %) in ethyl acetate
at 20 °C and 22 psi of hydrogen for 20 h. These conditions
afforded complete conversion of 9 while at the same time
giving 2.5 wt % of the deoxy impurity 21.
Based upon the conditions described above, a series of
experiments was conducted to quantify the impact of
temperature and solvent on the kinetics and selectivity in
order to further optimize this reaction. As mentioned
previously, minimizing the amount of the deoxy impurity
21 was especially important because of the purification
issues.
Using ethyl acetate as the solvent, a Pd/BaSO4 loading
of 3.8 wt %, and 25 psig pressure, the reaction was more
rapid at 20 °C but also resulted in good conversion of 9 to
10; however a significantly greater amount of 21 was also
formed (4.3%). The conversion and yield were calculated
from component concentrations, the latter based on experi-
mental determination of absolute HPLC response factors for
With MTBE, the reaction was very slow, with only a 73%
conversion of 9 after 58 h. The conversion 9 and the
formation of 10 followed zero-order kinetics, based on the
almost linear variation of yield with reaction time, consistent
with high catalyst surface coverage of reactant and products.
The best selectivity for 21 (viz., yield/conversion) of 0.019
was obtained with EtOAc at 10 °C (Table 1). However,
subsequent experiments showed that better selectivities could
be obtained using transfer hydrogenation. Consequently,
catalytic hydrogenation using molecular hydrogen was not
considered further.
Hydrogenolysis of 9 by catalytic transfer hydrogenation30
in ethanol with 5% Pd/C (10 wt %), ammonium formate
(30) For a review, see: Johnstone, R. A. W.; Wilby, A. H. Chem. ReV. 1985,
85, 129.
86
•
Vol. 10, No. 1, 2006 / Organic Process Research & Development

Table 1. Conversion of 9 and product yields at final reaction
times
change resulted in an increase in the amount of 21 produced.
We found that different lots of catalyst had different activities
and that the reaction concentration was related to the activity
of the catalyst. For Pd/C (old stock from the pilot plant),
the reaction was carried out with ∼14% concentration of 9
in ethanol (w/v) and the amount of 21 produced was 0.7%.
However, a fresh batch of catalyst which was ordered for
the Pilot Plant campaign gave 1.3% of 21 under the same
conditions. This was attributed to the higher activity of the
catalyst. By decreasing the concentration to 10%, the amount
of 21 produced decreased to below 0.6%. Thus, the concen-
tration for all of the pilot plant batches was lowered
accordingly.
With regard to what is reasonable for the pilot plant, a
heating and cooling rate of about 1 °C/min is practical. Thus,
the reaction was carried out in such a way that the
temperature was increased from 20 to 65 °C over ∼40 min
and from 65 to 72 °C over 15 min. At this point, there was
usually about 5% of 9 remaining. Then, an additional heating
period of 10 min was carried out before cooling was started.
Subsequently, it was decided to do without this final heating,
since the cooling temperature ramp (from 72 to 50 °C over
20 min) was sufficient to allow the reaction to continue, and
thus the amount of unreacted 9 was reduced to 0-1%.
In addition, the amount of ammonium formate was
increased from 1.25 to 1.3 equiv to help increase the rate.
However, we found that a large excess of ammonium formate
would result in more 21 being formed. We found that
hydrogenolysis did not occur at 20 °C. Analysis by HPLC
indicated that no 10 was formed at 20 °C for 20 min. At
about 40 °C, gas evolution from the reaction mixture was
obvious, and white solids, presumably ammonium carbonate,
were observed on the wall of the condenser.
When the catalyst (10% Pd/C, 50% wet) and solid
ammonium formate were combined, an exotherm accompa-
nied by gas evolution was observed. Thus it was decided
that the Pd/C should be added first and that the ammonium
formate would be added after the ethanol had been intro-
duced.
Since 10 was prepared as an ethyl acetate solution, a
solvent switch to ethanol was necessary for the salt formation
step. This was carried out in the following manner. After
concentrating 10 from ∼25% to 45% at ∼200 mbar, ethanol
was added and the batch was distilled at reduced pressure
(less than 80 mbar). This procedure effectively reduced the
residual ethyl acetate to below 1%.
It was found that filtration through Celite was effective
in removing the catalyst. The residual Pd content was below
0.5 ppm in the crystallized product using this procedure.31
The filtrate was found to be stable at low temperature. For
example, at -14 °C for 7 days, the decomposition of 10
was about 1% by HPLC. Thus, this solution could be used
as a hold point in the pilot plant.
reaction conversion
conditions
time (h)
of 9 (%)
21 (%) selectivity
EtOH, 10 °C
MTBE, 10 °C
THF, 5 °C
22
58
21
46
98.7
73.0
96.2
96.2
8.2
1.65
6.1
0.083
0.023
0.061
0.019
EtOAc, 10 °C
1.85
(1.25 equiv), and sodium bicarbonate (3 equiv) was inves-
tigated, and the results looked promising. Our experiments
indicated that the reaction could be carried out without the
sodium bicarbonate, which had been added to neutralize any
acid that was present. Under these conditions, hydrogenolysis
was complete in 30-60 min for small scale reactions and
10 was obtained in 90-95% purity (HPLC area %) with
0.7 ( 0.2% of 21. Prolonged reaction times resulted in lower
purity with the formation of more side products. Subse-
quently it was found that 21, as well as most of the other
impurities, could be removed or significantly reduced through
salt formation and recrystallization.
The hydroxylamine 10 is relatively stable in ethanol. After
heating at 75 °C for 2 h, the purity of a 10% solution of 10
in ethanol decreased by only 1-2%. However, the decom-
position of 10 in the transfer-hydrogenolysis reaction mixture
was found to be much faster; about 1-2% for every 15 min
at 75 °C. It was assumed that the acceleration of decomposi-
tion was due to the presence of ammonia, which is a
byproduct of the reaction and which could react with 10 by
a number of pathways, including deformylation, and cleavage
of the proline amide bond.
Supporting evidence for this assumption came from a
calorimetry experiment (see below) in which a slightly
positive nitrogen pressure (1.03-1.1 bar) was maintained
throughout the reaction. This prevented the ammonia gener-
ated in the reaction from escaping. After an extra 30 min of
heating at 73 °C (for calibration purposes), the amount of
10 present in the reaction mixture had decreased to 79.2%
by HPLC with a concomitant increase in impurity levels.
This is about a 10% lower concentration than expected.
However, the magnesium salt (vide infra) obtained from this
reaction mixture was still of excellent purity (98.2%),
although the product was isolated in lower yield (62% vs
75%).
Substituting formic acid or calcium formate for am-
monium formate resulted in no product formation. Addition
of acetic acid or sodium dihydrogen phosphate did not help
reduce the amount of decomposition observed. A nitrogen
sweep over the reaction mixture was found to be helpful.
For the pilot plant batches, the nitrogen flow was about 1.7
times that of the lab. The purity of 10 in the reaction mixture
was 92.6% under these conditions.
Catalyst, concentration, number of equivalents of am-
monium formate, and heating profile were some other factors
that were found to influence the course of the reaction. Since
5% Pd/C did not always afford complete conversion of 9,
even with extended time and additional ammonium formate,
we decided to use 10% Pd/C (50% water wet) instead. This
The reaction was carried out in the pilot plant in the
following manner. After increasing the temperature from 25
°C to 71 °C over 60 min, the mixture was cooled to 15 °C
(31) For leading methods of removing residual amounts of palladium from
organic compounds, see for example: Thayer, A. Chem. Eng. News 2005,
83 (36), 55. Garrett, C. E.; Prasad, K. AdV. Synth. Catal. 2004, 346, 889.
Vol. 10, No. 1, 2006 / Organic Process Research & Development
•
87

ammonium formate decomposition remained in the reaction
mixture.
As white solids were found on the reactor fittings after
the reaction, it is hypothesized that at least some NH3 reacted
to form ammonium carbonate,
2NH3 + CO2 + H2O f (NH4)2CO3
with water originating from the 50 wt % wet Pd/C catalyst.
Ammonium bicarbonate might also form. The high solubility
of ammonia in the ethanol solvent and the possibility of salt
formation by its reaction with 10 are also consistent with
the observed low volume of evolved ammonia gas.
Salt Prescreening and Manufacture. As was mentioned
previously, a solid derivative of 10 was needed to facilitate
the isolation, to provide needed stability, and to assist with
the formulation studies. Specifically, we sought to take
advantage of the moderate acidity (pKa ) 8) of the proton
on the hydroxyl group attached to the N-formylated nitrogen.
Since we were unsuccessful using organic bases for salt
formation, we reasoned that we might have a better chance
of success with a metal salt. Initially, formation of the sodium
salt was investigated; however, this compound was found
to be an amorphous solid. Our salt screening indicated that
magnesium and calcium could form salts. This was later
extended to zinc. A search of the literature provided little
information with regard to the formation of salts of hydrox-
ylamines using any of these metals.33
It was found possible to prepare the three salts by
treatment of 10 with NaOH in water, followed by the addition
of CaCl2, MgCl2, or ZnSO4. The purity of the salts was
slightly better than that of the starting material, and two of
the salts (Mg and Zn) were crystalline after drying. Crystal-
lization of the Ca and Zn salts was problematic because both
gave amorphous precipitates before crystallization. During
the production of the Ca salt, gel formation occurred,
resulting in poor processability (i.e., a viscous mixture that
filtered slowly). The Mg salt was the only one of the three
that initially precipitated as a crystalline material. Starting
from 10, about 6 g of each salt were prepared for the salt
program and preformulation studies.
Figure 3. Schematic diagram of the apparatus used for the
determination of the offgas volume, flow rate, and composition.
over ∼55 min. At this point, the in-process analysis indicated
virtually complete conversion [less than 1% of 9 remaining].
To ensure a safe scale-up in the pilot plant, an experiment
was performed to quantify heat and gas evolution rates. The
apparatus comprised a Mettler-Toledo RC1 reaction calo-
rimeter with a wet test meter (Ritter Model TG05) placed at
the reactor exit to monitor offgas volume. The offgas
composition was also determined on-line using a GC
equipped with a thermal conductivity detector and a sampling
loop placed upstream of the wet test meter (Figure 3). A
flow of nitrogen gas was maintained throughout the experi-
ment. Formation of offgas containing CO2 and NH3 was
expected based on the in situ decomposition of ammonium
formate:32
NH4+HCOO- f NH3 + CO2 + H2
The debenzylation was performed by first adding 8.9 g of
catalyst (10% Pd/C, 50% wet), 444 g of 9 dissolved in EtOH
(containing 69 g of 9), and 11.6 g of ammonium formate to
the reactor at room temperature. The mixture was subse-
quently heated under agitation to 65 °C over 40 min, heated
to 72 °C over 16 min, held at that temperature for 5 min,
and finally cooled to below room temperature (ca. 15 °C)
over 1 h. The feasibility of these heating and cooling times
in the pilot plant was confirmed with the project engineer.
The reaction was slightly exothermic initially (Figure 4) but
became endothermic as gas was evolved. The maximum gas
evolution rate (69 g of 9 basis] of 100 cm3/min and the
maximum heat evolution rate (ca. 10 W/kg) were both
acceptable for safe scale-up.
A basic procedure for the preparation of these salts was
developed: an aqueous solution of 10 in water was filtered,
and 1 equiv of NaOH was added. Then, 0.5 equiv of an
inorganic salt (ZnSO4, MgCl2, CaCl2) was introduced. After
stirring for the designated time period, the salt was isolated
by filtration, washed, and dried.
An improvement which was made involved adding the
0.5 equiv of NaOH before the first filtration. This served to
increase the solubility of 10 in water, and thus the total
amount of water could be reduced resulting in an increased
yield for the Zn and Mg salts.
The yields were 77%, 70%, and 36% for the zinc,
magnesium, and calcium salts, respectively. The amount of
the deoxy impurity (now present as the salt) was nearly the
same as that in 10 for the Zn and Mg salts (2.0 to 2.3%), or
even higher for the calcium salt (from 2.3 to 3.5%).
GC analysis indicated that the offgas contained carbon
dioxide but no ammonia. However, the presence of some
ammonia in the offgas was not ruled out based on the pH
paper color change (indicating the presence of a base) when
the paper was exposed to the offgas. The capability of the
GC method to detect ammonia was checked in a separate
experiment in which aqueous NH4OH was placed in the
calorimeter and heated to 45 °C to desorb NH3 gas; NH3
was easily detected by GC (Figure 5). We therefore infer
that the volume of ammonia evolved in the debenzylation
of 9 was very small and that most of the NH3 produced by
(33) Metal complexes of simple hydroxylamines are known; see for example:
Sarukhabnov, M. A.; Val’dman, S. S.; Parpiev, N. A. Zhurnal Neorgan-
icheskoi Khimii 1973, 18, 838; Chem. Abs. 1973, 131508.
(32) Rajagopal, S.; Spatola, A. F. J. Org. Chem. 1995, 60, 1347.
88
•
Vol. 10, No. 1, 2006 / Organic Process Research & Development

Figure 4. Heat and offgas evolution rates.
Figure 5. Offgas rate and composition in which NH4OH was charged to the RC-1 calorimeter to test the GC method.
A considerable effort was made to remove this impurity
from 11. In a solubility test, the magnesium salt 11 (made
from material which contained 2.2% of 23) was found to
dissolve readily in ethanol (1 g/5 mL) below 40 °C and
partially to dissolve in ethanol/water, THF/water, and aque-
ous MgCl2. However, no significant reduction in the deoxy
impurity content was observed after digestion of the Mg salt
in any of these solvents at elevated temperatures (40-45
°C) for 10 h. In addition, recrystallization from ethanol/water,
ethyl acetate, acetone/water, methanol/water, THF/water,
2-propanol, or acetonitrile proved to be ineffective since the
magnesium salt of the deoxy impurity was less soluble than
the magnesium salt 11.
However, when the recrystallization method was applied
to the magnesium salt 11 which contained a smaller amount
of the deoxymagnesium salt impurity (0.5-0.8%), the level
of the deoxy compound present was reduced to 0.3-0.5%.
Other impurities, although present at higher levels, could also
be removed. Thus, a solution of 10 in acetone (or ethanol or
methanol) was treated with aqueous MgCl2 and aqueous
NaOH at 40-45 °C followed by the slow addition of water
and digestion at this temperature before cooling. The
magnesium salt 11 was obtained in 65-70% yield and
g98.5% purity. Thus we felt that salt formation was a viable
method of effecting purification as long as the level of the
deoxy impurity 21 was kept below 2% in 10.
Due to concerns regarding residual acetone in the drug
substance, ethanol was chosen as the solvent for the
precipitation of the salt. Thus, 10 in ethanol was first
concentrated under a vacuum to 1 g/3.3 mL of EtOH. To
this solution was added an aqueous MgCl2 solution (0.5
equiv) followed by an aqueous NaOH (∼1 equiv) to pH 9.1.
At this stage, the concentration was 1 g of 10 in 3.3 mL of
EtOH and 3.3 mL of H2O, and precipitation of the salt began.
After digestion at 45 °C for 1 h, another 3.3 mL of water/g
was added over ∼1 h, and the digestion was allowed to
proceed for another hour before cooling the suspension to
20 °C and isolation of the solids.
Additionally, the question of the possible decomposition
of the Mg salt at 45 °C was addressed. This was based on
Vol. 10, No. 1, 2006 / Organic Process Research & Development
•
89

Figure 6. Percent loss on drying vs time.
some preliminary data which indicated that ∼93% decom-
position of a 0.1% aqueous solution of 10 occurred at 50 °C
in pH 9 solution after 7 days. Based on our experiences with
the Mg salt, we felt that this compound was much more
stable than the free acid under the conditions we were using.
This is because we had been able to recrystallize the material
to high purity using the ethanol/water solvent mixture under
conditions similar to the isolation conditions.
To support this assumption, an additional experiment was
carried out prior to the Pilot Plant campaign. A solution of
11 (1 wt % in 1:2 ethanol/water, pH 9.6, almost saturated at
45 °C) was heated at 45 °C for 20 h. HPLC analysis indicated
91% purity (i.e., <10% decomposition). Since we have
calculated that only about 10% of the salt was actually in
solution, this would translate as 1% decomposition of the
total amount of Mg salt. It is thus reasonable to believe the
decomposition loss during salt formation is merely 1-2%.
We also observed that when the impurity level in 11 was
higher, the loss of material in the mother liquor was also
higher. The dissolution loss in the mother liquor ranged from
6 to 12 mg/mL for crude 11.
Since the crystalline form of the magnesium salt was
determined to be a tetrahydrate, and since it was known that
the anhydrous material was not crystalline, there was some
concern regarding overdrying of the product in the pilot plant.
Therefore, a drying study was conducted using vacuum
thermogravimetry. When the Mg salt was dried by vacuum
thermogravimetry at 40-45 °C and 20 mbar, a sharp increase
in loss on drying occurred, approaching a nearly constant
(plateau) value. However, with longer times, the sample
continued to lose mass at a much slower rate (see Figure 6).
Thus, overdrying of the product in the pilot plant could occur
if a longer drying time is used. To avoid significant
overdrying, recommendations were made to closely monitor
the loss on drying at periodic intervals. In addition, the
residual ethanol values after drying times of 4 and 24 h
indicated that small amounts of ethanol were still present,
meaning that a small portion of ethanol remained bound in
the solid (thermodynamic activity < 1).
In summary, we have developed a practical enantiose-
lective route to LBM415. The conjugate addition of O-
benzylhydroxylamine to oxazolidinone 3 furnished tosylate
5 (after salt formation) with good selectivity. Thus, chemical
resolution was not required. Removal of the chiral auxiliary
followed by N-formylation gave optically pure 7 in good
overall yield. Coupling of acid 7 with proline amide 25 was
carried out using the readily available peptide coupling agent
EDCI/HOBt and provided key intermediate 8 in good yield
and high optical and chemical purity. Oxidation to 9 using
relatively stable and nonhazardous urea/hydrogen peroxide
was followed by debenzylation and salt formation without
the need for isolation of either of the previous intermediates
and gave LBM415 in 16% overall yield (ee > 99%) in 12
steps.
Experimental Section
2-Methylenehexanoic Acid (2). A suspension of n-
butylmalonic acid 1 (110 kg, 680 mol) in absolute ethanol
(633 L) was treated at room temperature with 37% aqueous
formaldehyde (258 kg, 3180 mol). The suspension was stirred
at 20 °C until all of the solids had dissolved (15 min).
Piperidine (70 kg, 810 mol) was added at such a rate that
the internal temperature was kept below 45 °C. The time of
the addition was 20 min. The reaction mixture was stirred
at 45 °C for 1 h, heated to 72 °C, and held at this temperature
for 30 min. The batch was then heated to the reflux
temperature and held for 2 h. At this point, the reaction
mixture was concentrated under vacuum at 60 °C to a
predetermined volume (396 L). After the reaction mixture
cooled to 20 °C, 2 N HCl (880 kg, 1760 mol) and ethyl
acetate (440 L) were charged to the batch while the
temperature was maintained below 27 °C. The aqueous layer
was separated and back extracted with ethyl acetate (220
L). The organic portions were combined and washed with 2
N HCl (220 kg, 440 mol), saturated sodium chloride (220
kg), and water (220 kg). Most of the ethyl acetate was
removed by distillation at atmospheric pressure (final volume
385 L), and after the solution cooled to ambient temperature,
90
•
Vol. 10, No. 1, 2006 / Organic Process Research & Development

heptane (376 kg) was added. Distillation at atmospheric
pressure to a predetermined volume (495 L) was followed
by filtration at 55 °C to afford 2 as a solution in heptane
(72.1 kg, 82%): 1H NMR (CDCl3) δ 6.30 (s, 1H), 5.66 (s,
1H), 2.32 (t, J ) 7.0 Hz, 2H), 1.44 (m, 4H), 0.94 (t, J ) 7.0
Hz, 4H). Note: Due to concerns regarding the stability of
this compound, it was not purified further but was used “as
is” in the following step.
tert-butyl methyl ether (191 and 47 L). The combined organic
portions were washed with water (47 L), and the tert-butyl
methyl ether was removed by distillation under reduced
pressure to afford O-benzylhydroxylamine free base as a
clear oil.34 The oil was added to 3 (33.3 kg, 116 mol), and
the mixture was heated to 50 °C for 22 h. At this point, the
reaction mixture was cooled to room temperature and diluted
with ethyl acetate (139 L). This mixture of the free base 4
was added to a solution of p-toluenesulfonic acid monohy-
drate (88 kg, 463 mol) in ethyl acetate (325 L) followed by
an ethyl acetate rinse (118 L). The suspension was stirred
for 1.5 h and filtered to remove the O-benzylhydroxylamine
p-toluenesulfonate salt. The salt was washed with ethyl
acetate (115 L), and the wash was combined with the filtrate.
The filtrate was concentrated under reduced pressure to a
final volume (170 L) and cooled to 45 °C at which point
tert-butyl methyl ether (764 L) was added. The mixture was
cooled to 22 °C and stirred for 2 h. The resulting solids were
filtered, washed with water-saturated tert-butyl methyl ether
(231 L), and dried in vacuo at 50 °C for 12 h to give 5 [45.9
(4S)-3-(2-Methylene-1-oxohexyl)-4-(phenylmethyl)-2-
oxazolidinone (3). A solution of 2-methylenehexanoic acid
(21.6 kg, 170 mol) in THF (233 L) was cooled to -20 °C.
To this solution was added N,N-diisopropylethylamine (27.8
kg, 215 mol) followed by a THF rinse (22 L). After stirring
at -20 °C for 30 min, pivaloyl chloride (19.7 kg, 163 mol)
was added followed by a THF rinse (22 L). The mixture
was stirred for 30 min at -20 °C, allowed to warm to 10-
15 °C for 30 min, and then cooled back down to -20 °C. In
a separate vessel, a solution consisting of (S)-benzyl-2-
oxazolidinone (26.4 kg, 149 mol) in THF (136 L) was
prepared and cooled to -20 °C. A 2 M solution of lithium
diisopropylamide in heptane (68 kg, 166 mol) was added at
such a rate that the temperature was maintained below -20
°C. Upon completion of the addition, the solution of the anion
was added to the solution of the mixed anhydride while the
temperature was maintained below -20 °C. The reaction
mixture was held for 30 min at -20 °C, allowed to warm
slowly to 15 °C, and held at this temperature for 30 min. At
this point, a suspension consisting of 1 M potassium
bicarbonate solution (150 kg, 690 mol) and Celite (5.5 kg)
was introduced, and the mixture was stirred for 20 min. The
suspension was filtered, and the filter cake was washed with
ethyl acetate (49 kg, 54 L). The mixture was filtered, and
the aqueous portion separated. The organic portion was
washed twice with 1 M potassium bicarbonate (150 kg) and
concentrated under reduced pressure to a final volume (50
L). To the concentrate was added ethyl acetate (136 L). The
ethyl acetate solution was washed twice with water (55 L)
and concentrated to a final volume (50 L) under reduced
pressure. The concentrate was diluted with ethyl acetate (95
L), and the distillation was repeated until no more distillate
was observed. The concentrate was warmed to 50 °C, and
heptane (136 L) was added to form a solution. The mixture
was concentrated under reduced pressure to a final volume
(65 L). The concentrate was dissolved in tert-butyl methyl
ether (40.8 L) and warmed to 45 °C. The solution was diluted
with heptane (368 L), cooled to 0 °C, and held at this
temperature for 2 h. The solids were filtered, washed with a
cold (0 °C) 95/5 (w/w) heptane/ethanol solution, and dried
in vacuo at 30 °C overnight to give 3 [29.4 kg, 63%
(corrected)] as a white crystalline solid: mp 46 °C; 1H NMR
(CDCl3) δ 7.29 (m, 5H), 5.40 (d, J ) 7 Hz, 2H), 4.44 (m,
1H), 4.22 (m, 2H), 3.37 (m, 1H), 2.82 (m, 1H), 2.39 (m,
1H), 1.43 (m, 4H), 0.93 (m, 3H).
1
kg, 68% (corrected)]: mp 132 °C; H NMR (DMSO-d6) δ
10.7 (s br, 1H), 7.52 (d, J ) 6 Hz, 2H), 7.38 (m, 6H), 7.26
(m, 3H), 7.13 (m, 4H), 4.98 (s, 2H), 4.56 (m, 1H), 4.29 (m,
1H), 4.19 (m, 1H), 4.02 (d, J ) 8 Hz, 1H), 3.70 (m, 1H),
3.41 (d, J ) 12 Hz, 1H), 2.84 (d, J ) 12 Hz, 1H), 2.50 (s,
3H), 2.24 (m, 1H), 1.63 (m, 1H), 1.48 (m, 1H), 1.26 (m,
4H), 0.86 (m, 3H).
(2R)-2-[[(Phenylmethoxy)amino]methyl]hexanoic Acid
(6). A suspension of 5 (27.5 kg, 0.47 mol) in ethyl acetate
(95 L) and water (24 L) was treated with a 1 M aqueous
solution of sodium carbonate (96 kg). After the mixture was
stirred for 15 min, the aqueous layer was separated and
discarded. The organic layer was washed with water (2 ×
24 L) and concentrated under reduced pressure until the ethyl
acetate content was less than 8 wt %. The concentrate was
dissolved in THF (129 L) and water (41 L) and cooled to 0
°C. To this was added 30% hydrogen peroxide (12.3 kg,
108 mol), followed by a 4.5% lithium hydroxide solution
(50.6 kg, 54 mol), while the temperature was maintained at
0 °C. The reaction mixture was stirred at this temperature
for 45 min. The reaction was quenched by the addition of a
cold (0 °C) aqueous solution of sodium sulfite (20.5 kg, 163
mol in 403 L of water). The reaction mixture was allowed
to warm to 20 °C and stirred at this temperature for 30 min.
The solution was concentrated under reduced pressure
(maximum jacket temperature ) 33 °C) to a final volume
of 560 L. The concentrate was extracted with ethyl acetate
(6 × 58 L) and then acidified with 3 N HCl (42.1 kg) to a
final pH of 4-5. The product was extracted into ethyl acetate
(2 × 113 L), and the extracts were combined and washed
with water (52 L). Due to the very limited stability of 6, the
compound was held as a solution for use in the next step.
The solution was assayed for content of 6 (7.9 kg, 66%
yield). A small sample of the solution was removed and
(4S)-3-[(2R)-1-Oxo-2-[[(phenylmethoxy)amino]methyl]-
hexyl]-4-(phenylmethyl)-2-oxazolidinone, p-Toluenesulfon-
ic Acid Salt (5). O-Benzylhydroxylamine hydrochloride
(43.7 kg, 274 mol) was dissolved in 2 N potassium hydroxide
(300 L) with stirring. The solution was extracted twice with
1
carefully concentrated for characterization: H NMR (CDCl3)
δ 7.37 (s, 5H), 6.86 (s br, 2H), 4.75 (t, J ) 12 Hz, 2H), 3.19
(34) O-Benzylhydroxylamine free base (oil) was found to be stable at 160 °C
for 10 h (RADEX dynamic autoclave) and at 120 °C for 16 h (stainless
steel pan) (RADEX isoperibolic autoclave).
Vol. 10, No. 1, 2006 / Organic Process Research & Development
•
91

(m, 2H), 2.77 (m, 1H), 1.70 (m, 1H), 1.55 (m, 1H), 1.35
(m, 4H), 0.90 (m, 3H).
for 12 h to give 25 [48.7 kg, 84% (corrected for the purity
of the limiting reagent)]: mp 180 °C (dec); 1H NMR
(DMSO-d6) δ 11.2 (s, 2H), 9.35 (broad s, 1H), 8.73 (broad
s, 1H), 8.40 (s, 1H), 8.09 (m, 1H), 7.84 (m, 1H), 4.43 (m,
1H), 3.28 (s, 1H), 2.42 (m, 1H), 1.96 (m, 3H); 13C NMR
(75.5 MHz, CDCl3) δ 167.33, 157.52, 154.21, 149.83,
147.40, 114.99, 59.36, 45.72, 29.60, 23.46; MS (ES+) 210
(MH+, free base).
(2R)-2-[Formyl(phenylmethoxy)amino]methylhexa-
noic Acid Dicyclohexylamine Salt (7). An ethyl acetate
solution of 6 (16.2 kg, 64 mol, total weight of solution )
384 kg) was cooled to -10 °C and treated with a cold (5
°C) solution of acetic anhydride (13.2 kg, 129 mol) and 96%
formic acid (23.7 kg, 514 mol), which was added over 20
min. The reaction mixture was stirred an additional 20 min
at -10 °C. Water (5 L) was added, and the reaction mixture
was allowed to slowly warm to 15 °C. The ethyl acetate
was removed by distillation in vacuo while the internal
temperature was maintained below 40 °C to a final volume
(60 L). Toluene (210 L) was added, and the distillation was
continued to reach a final volume (60 L). The toluene
addition and distillation were carried out a second time. At
this point, the mixture was cooled to room temperature, and
a sample was taken to determine the acetic acid content of
the batch (<2%). A solution of dicyclohexylamine (14 kg,
77 mol) in heptane (206 L) was added over 15 min. Seed
crystals of 7 were added, and the solution was stirred for 2
h during which time the precipitation occurred. The suspen-
sion was diluted with heptane (149 L) and held for 8 h at
room temperature with stirring. The solids were isolated by
filtration, washed with a solution of dicyclohexylamine (2.9
kg, 16 mol) in heptane (86 L), and dried at a maximum
temperature of 27 °C in vacuo for 8 h to give 7 [20.6 kg,
(2S)-N-(5-Fluoro-2-pyridinyl)-1-[(2R)-2-[[formyl(phen-
ylmethoxy)amino]methyl]-1-oxohexyl-2-pyrrolidinecar-
boxamide (8). A solution of 7 (41.7 kg, 90.1 mol) in ethyl
acetate (360 L) was treated at room temperature with a 10%
solution of citric acid (360 kg), and the mixture was stirred
until all of the solids had dissolved. The aqueous layer was
separated, and the organic layer was washed with water
(2 × 270 L). Compound 25 (40.1 kg, 108.1 mol) was added,
and the mixture was cooled to 5 °C. Water (72 L) was
introduced followed by 1-hydroxybenzotriazole hydrate
(HOBt) (15.4 kg, 99.1 mol) and 1-(3-dimethylaminopropyl)-
3-ethyl carbodiimide hydrochloride (98%) (48.4 kg, 252.2
mol). Another water (72 L) addition was followed by the
slow introduction of N-methylmorpholine (57.5 kg, 567.5
mol) while the temperature was maintained below 20 °C.
The reaction mixture was stirred at 21 °C for 10 h or until
the amount of 7 remaining was less than or equal to 2% by
HPLC area normalization. The aqueous layer was separated
and discarded. The organic layer was washed with water
(4 × 270 L) to remove the last traces of HOBt. The ethyl
acetate solution containing 8 was split into two equal portions
(85.7 kg each), and each portion was filtered through a bed
of silica gel (50 kg) using ethyl acetate as the eluent.
Fractions (approximately 20 kg each) were collected and
analyzed by HPLC. The appropriate fractions were com-
bined, and this solution was concentrated by distillation under
vacuum (maximum jacket temperature ) 45 °C) to a final
volume (260 L). The batch was heated to 55 °C, and heptane
(815 L) was added. The solution was cooled to 45 °C, seeded,
cooled further to 35 °C, held at this temperature for 2 h,
and then cooled to 20 °C, and held at this point for a
minimum of 8 h. Finally, the suspension was cooled to -10
°C and stirred for another 2 h. The product was collected
by centrifugation and washed with a mixture of heptane (94
L), ethyl acetate (48 L), ethanol (2.7 L), and water (0.14 L).
The filter cake was dried at 42 °C in vacuo for 8 h, and the
crude product was recrystallized from ethyl acetate (173 L)/
heptane (695 L) to afford 8 [28.9 kg, 68% (corrected)] as
an off-white crystalline solid: mp 98 °C; 1H NMR (DMSO-
d6) δ 10.89, 10.63 (s, 1H), 8.29 (s, 1H), 8.01 (m, 2H), 7.70
(m, 1H), 7.35 (s, 5H), 4.81 (s, 2H), 4.54 (broad s, 1H), 3.52
(m, 4H), 2.92 (m, 1H), 1.85 (m, 4H), 1.20 (m, 6H), 0.81 (s,
3H); Assay by HPLC 98.9%; Purity by DSC 98.2%.
(2S)-N-(5-Fluoro-1-oxido-2-pyridinyl)-1-[(2R)-2-[[formyl-
(phenylmethoxy)-amino]methyl]-1-oxohexyl]-2-pyrrolidine-
carboxamide (9). A solution of 8 (12.8 kg, 27 mol) and
urea-hydrogen peroxide (7.7 kg, 81.9 mol) in ethyl acetate
(117 L) was cooled to 10 °C, and solid phthalic anhydride
(12.0 kg, 81 mol) was added in portions. The reaction
mixture was stirred at 20 °C for 6 h at which point there
1
69% (corrected)]: mp 83-86 °C; H NMR (DMSO-d6) δ
8.04 (d, 1H), 7.40 (m, 5H), 4.91 (s, 2H), 3.50 (m, 2H), 2.77
(m, 2H), 2.51 (m, 2H), 1.49 (m, 27H), 0.84 (s, 3H).
(2S)-N-(5-Fluoro-2-pyridinyl)-2-pyrrolidinecarboxa-
mide Dihydrobromide (25). A mixture of (2S)-1,2-pyrro-
lidinecarboxylic acid 1-(phenylmethyl)ester 23 (50.6 kg, 203
mol) and 2-amino-5-fluoropyridine (17.5 kg, 156 mol) in
ethyl acetate (67 L) was treated with a solution of 2-chloro-
4,6-dimethoxy-1,3,5-triazine (32.9 kg, 188 mol) in ethyl
acetate (67 L) followed by a rinse with ethyl acetate (33 L).
This was followed by the addition of N-methylmorpholine
(23.7 kg, 243 mol) at such a rate that the temperature was
maintained below 25 °C. The reaction mixture was stirred
at 22 °C for 5 h. At this point, water (156 L) was added and
the suspension was filtered. The filter cake was washed with
ethyl acetate (46 L), and the wash was combined with the
filtrate. The aqueous layer was separated and extracted with
ethyl acetate (2 × 46 L). The ethyl acetate layers were
combined and washed with 1 N sodium hydroxide (46.8 kg)
and saturated sodium chloride solution (2 × 62 kg). The
solution was concentrated at atmospheric pressure (95 °C
maximum jacket temperature) to a final volume (100 L). At
this point, the water content by Karl Fischer titration was
<0.5%. The solution of crude (2S)-2-[[(5-fluoro-2-pyridinyl)-
amino]carbonyl]phenylmethyl ester 24 was diluted with ethyl
acetate (37 L) and cooled to 5 °C. Hydrogen bromide solution
(30 wt % in acetic acid, 166.1 kg, 616 mol) was added while
the temperature was maintained below 10 °C. After stirring
for 1 h, the batch was warmed to 23 °C and stirred for an
additional 2 h. The solids were collected by centrifugation,
washed with ethyl acetate (69 L), and dried at 45 °C in vacuo
92
•
Vol. 10, No. 1, 2006 / Organic Process Research & Development

was less than 0.4% 8 remaining by HPLC analysis (area
normalization). After the reaction mixture was cooled to 5
°C, an 8.6% sodium sulfite solution (159.7 kg) was intro-
duced while the temperature was maintained below 25 °C.
At the completion of the addition, the batch was held at 22
°C for 55 min. The top layer was tested for the absence of
peroxides, and the aqueous layer was separated and dis-
carded. The organic layer was washed with 0.9 M sodium
carbonate (160.4 kg) and a 12.7% sodium chloride solution
(109.5 kg). The organic portion was concentrated in vacuo
(maximum batch temperature ) 50 °C) to a final volume
(60 L) (water content by Karl Fischer titration e1.0%) which
was assayed by HPLC for the content of 9 [12.8 kg, 96%
(corrected)]. A sample was removed and concentrated to
dryness for characterization purposes: 1H NMR (CDCl3) δ
10.39 (s, 1H), 8.42 (m, 1H), 8.17 (m, 2H), 7.40 (s, 5H), 7.08
(m, 1H), 4.83 (m, 2H), 4.41 (broad d, 1H), 3.81 (m, 2H),
3.58 (m, 2H), 3.02 (m, 1H), 1.68 (m, 10H), 0.85 (s, 3H);
MS (ES+) 487 (MH+).
(2S)-N-(5-Fluoro-1-oxido-2-pyridinyl)-1-[(2R)-2-[(formyl-
hydroxyamino)methyl]-1-oxohexyl]-2-pyrrolidinecarbox-
amide Magnesium Salt (2:1) (11). The ethyl acetate solution
of 9 from the previous step (53.1 kg, 26 mol) was
concentrated under reduced pressure (maximum batch tem-
perature ) 50 °C) to a final volume (24 L). The batch was
cooled to room temperature, and ethanol (71 L) was added.
The distillation was resumed under the same conditions to a
final volume (24 L). Ethanol (72 L) was added, and the
distillation was carried out once more to the same final
volume (ethyl acetate content e 4%). The ethanol solution
of 9 was added to 10% Pd/C (1.3 kg, 50% water wet), and
to this suspension was added ammonium formate (2.1 kg,
32.3 mol). The batch was heated to an internal temperature
of 65 °C over 40 min. The temperature was then raised to
72 °C over 14 min and held at this temperature an additional
6 min. The mixture was cooled to 10 °C rapidly (ap-
proximately 1 °C/min cooling rate), and a sample was taken
to check for completeness (e6% of 9 remaining). The
reaction was filtered through Celite (5 kg), and the filter cake
was washed with ethanol (38 L) which was combined with
the batch. The solution containing 10 was concentrated under
reduced pressure (maximum jacket temperature ) 50 °C)
to a final volume (41 L). After cooling to 25 °C, water (14.7
L) was added followed by a solution of magnesium chloride
(13.5 kg) which was prepared from magnesium chloride
hexahydrate (5.5 kg, 27 mol) and water (23 L). The reaction
mixture was heated to 43 °C (internal temperature), and 3
N sodium hydroxide solution (10.1 kg) was added over 15
min to afford a final pH of 8.7. After stirring for 70 min at
43 °C, water (33.1 L) was added over 1 h and the suspension
was stirred for an additional 1 h at 43 °C. The batch was
cooled to 20 °C and stirred for 1 h. The solids were collected
by centrifugation, washed with a cold (5 °C) solution of water
(13 L) and ethanol (6.5 L), and dried in vacuo at 42 °C for
4.5 h to afford 11 [9.1 kg, 79% (corrected)] as an off-white
crystalline solid. A second batch was prepared, and the
combined products (17.8 kg) were recrystallized from
ethanol/water using the following procedure. The salt was
dissolved in absolute ethanol (86 L) and heated to 40 °C to
effect dissolution. Upon cooling to room temperature, the
solution was filtered and the filter and lines were rinsed with
ethanol (19 L), after which the rinse was combined with the
batch. The solution was concentrated under reduced pressure
(maximum jacket temperature ) 50 °C) to a final volume
(65 L). The concentrate was warmed to 44 °C, and water
(51.5 L) was added. Stirring was continued at this temper-
ature for 1 h after precipitation started. Water (51.5 L) was
added, and stirring was continued for an additional 55 min.
The suspension was cooled to 3 °C and stirred for a total of
3 h. The solids were collected by centrifugation, washed with
cold (3 °C) ethanol/water (7L/13.8 L), and dried in vacuo
for 7.5 h to afford LBM415 (11) [16.0 kg, 70% overall
1
(corrected)]: H NMR (CD3OD) δ 8.55 (s, 1H), 8.50 (t, 1H),
7.55 (s, 1H), 7.45 (t, 1H), 4.76 (broad s, 1H), 4.03 (broad s,
1H), 3.70 (broad s, 2H), 3.44 (broad s, 1H), 3.41 (broad s,
1H), 2.30 (broad s, 1H), 2.07 (broad s, 3H), 1.60 (broad s,
1H), 1.46 (broad s, 2H), 1.35 (broad s, 3H), 0.91 (t, 3H);
13C NMR (75.5 MHz, CD3OD) δ 175.66, 172.49, 156.38,
155.74, 143.03, 128.86, 118.41, 116.45, 62.59, 54.70, 49.28,
42.45, 31.50, 30.43, 30.18, 25.87, 24.00, 14.35. Anal. Calcd
for C18H24FN4O5‚1/2Mg‚2H2O: C, 48.72; H, 6.36; N, 12.62.
Found: C, 48.64; H, 6.23; N, 12.53.
Acknowledgment
We thank the Analytical Research and Development group
for their support and assistance throughout this work. We
extend our appreciation to Dr. Hanumantha Marepalli for
his help in analyzing the NMR spectra of compound 11 and
Robert Duval for the safety tests. We also thank the reviewers
for many valuable suggestions.
Note Added after ASAP Publication: This article was
published on the web December 16, 2005 with a minor error
in Table 1. The version posted December 21, 2005 and the
print version are correct.
Received for review September 8, 2005.
OP050165Y
Vol. 10, No. 1, 2006 / Organic Process Research & Development
•
93