Development of an alternative process for the manufacture of a key starting material for cefovecin sodium

Article abstract of DOI:10.1021/op700091d

A process has been developed for the use of trimethylphosphite for the formation of the six-membered 3,6-dihydro-2H-[1,3]-thiazine ring in the cephem architecture by an intramolecular Horner-Emmons-Wadsworth condensation. The process is a suitable alternative to the traditional Wittig process, which uses trimethylphosphine. The process developed is a highly telescoped reaction pathway consisting of at least six known reaction intermediates that was scaled for production use to produce 2.

Full text of DOI:10.1021/op700091d

Organic Process Research & Development 2007, 11, 742−746
Development of an Alternative Process for the Manufacture of a Key Starting
Material for Cefovecin Sodium
Timothy Norris,*,† Isao Nagakura,‡ Hiromasa Morita,‡ Grant McLachlan,§ and Joe Desneves§
Chemical Research and DeVelopment, Pfizer Inc Global Research and DeVelopment, Eastern Point Road,
Groton, Connecticut 06340, U.S.A., Pfizer Nagoya, 5-2, Taketoyo-cho, Chita-gun, Aichi, 470-2393 Japan, and
Institute of Drug Technology, 45 Wadhurst DriVe, Boronia, VIC 3155, Australia
Abstract:
original synthetic procedure using Wittig chemistry derives
from the patent literature3 and is currently used on a large-
scale basis for manufacture of 2 from 3 involving five
telescoped steps, in which none of the intermediates are
isolated, as shown in Scheme 1.
A process has been developed for the use of trimethylphosphite
for the formation of the six-membered 3,6-dihydro-2H-[1,3]-
thiazine ring in the cephem architecture by an intramolecular
Horner-Emmons-Wadsworth condensation. The process is a
suitable alternative to the traditional Wittig process, which uses
trimethylphosphine. The process developed is a highly tele-
scoped reaction pathway consisting of at least six known
reaction intermediates that was scaled for production use to
produce 2.
The â-lactam 3 is converted to a chloro compound using
thionyl chloride. After aqueous workup the product 4 is
reacted with trimethylphosphine (Me3P) to form the phos-
phonium salt 5. This is unusual because the standard
phosphine reagents are tributylphosphine or triphenylphos-
phines.4 Triphenylphosphine is unreactive with 4, and
tributlyphosphine and triethylphosphite gave poor yields.
Me3P solves this problem and was used successfully on a
large scale. A sodium bicarbonate wash gives the ylide 6,
which reacts intramolecularly with the ketone moiety to form
the six-membered ring, over 24 h. The cyclised product 7 is
treated with phosphorous pentachloride and isobutanol to
achieve removal of the phenacyl protecting group. Compound
2 precipitates from the reaction solution in high purity.
Introduction
Cefovecin 1 is a potent stable antibiotic targeted for
companion animals. Cefovecin features a chiral tetrahydro-
furan ring substituent at C-3, which is responsible for the
unique activity and stability profile. The synthesis of
Cefovecin is described in the patent literature.1 Starting from
penicillin G this synthesis consists of 13 transformations,
many of which are telescoped steps. The intermediate
products are often variable mixtures of diastereoisomers. It
is not until the cephem intermediate 2 is reached that a single
crystalline diastereoisomer is obtained. As a result, this
material was targeted as a key control gate in the synthesis,
and therefore the synthesis is critical to the establishment of
a commercial process.
Results and Discusion
Horner-Emmons-Wadsworth Approach. There are
a number of disadvantages to the use of Me3P in this process.
First, Me3P is expensive, contributing approximately 25%
to the total cost of goods for the process. Me3P is also
extremely hazardous, as it is highly flammable and toxic with
environmental issues. The yields from the reaction were
highly variable, and the intermediates are relatively unstable.
An alternative to Me3P involves the use of trimethylphos-
phite [P(OMe)3] in the synthesis. The synthetic approach with
P(OMe)3 is presented in Scheme 2. Steps one and five are
the same as those in the Me3P route. Replacement of Me3P
by P(OMe)3 leads to the formation of a phosphonate ester
8, which then undergoes cyclisation to 7.
Chlorination. Chlorination of 3 using thionyl chloride
in dichloromethane with 2-picoline gives 4 in near quantita-
tive yield. The optimum conditions for the SOCl2 charge
was found to require 1.1 equiv, based on the initial charge
of 3. Lower charges gave incomplete conversion to 4.
In order to avoid side-product formation the reaction must
be cooled. Cooling a solution of 3 and 2-picoline in
dichloromethane from ambient temperature to -20 °C
Parallel conversions from penicillin G Via penicillin G
sulfoxide and rearrangement with trialkyl phosphites lead
to the starting material for the synthesis of 3 and related
derivatives; this chemistry is reported in the literature.2 The
* Author for correspondence. E-mail: timothy.norris@pfizer.com.
† Pfizer Inc. Global Research and Development.
‡ Pfizer Nagoya.
§ Institute of Drug Technology.
(1) (a) Bateson, J. H.; Burton, G.; Fell, S. C. M. U.S. Patent 6001997, Dec 14,
1999. (b) Bateson, J. H.; Burton, G.; Fell, S. C. M. U.S. Patent 6020329,
Feb 1, 2000. (c) Bateson, J. H.; Burton, G.; Fell, S. C. M. U.S. Patent
6077952, Jun 20, 2000.
(3) (a) Colberg, J. C.; Tucker, J. L.; Zennoni, M.; Giovianni, F.; Donadelli, A.
WO 2002046199. (b) Colberg, J. C.; Zennoni, M.; Giovianni, F.; Donadelli,
A. WO 2002046198.
(4) Bateson, J. H.; Burton, G.; Fell, S. C. M.; Smulders, H. C. J. Antibiot. 1994,
47, 253-256.
(2) Cooper R. D. G.; Jose, F. L. J. Am. Chem. Soc. 1970, 92, 2575.
742
•
Vol. 11, No. 4, 2007 / Organic Process Research & Development
10.1021/op700091d CCC: $37.00 © 2007 American Chemical Society
Published on Web 06/08/2007

Scheme 1. Synthesis of 2 via the trimethyl phosphine (Me3P) route
resulted in some precipitation. Addition of thionyl chloride
to this suspension at -20 °C gave a higher amount of
unreacted starting material, which could not be chlorinated
by addition of excess thionyl chloride. Therefore, a portion
of the total thionyl chloride charge (10%) was added before
precipitation commenced, at -15 °C. The solution was then
cooled to -20 °C, and the remaining thionyl chloride was
added slowly at or below this temperature. The product is
significantly more soluble in dichloromethane, and therefore
no precipitation was observed using this procedure.
3 and 4 are diastereomeric mixtures of the hydroxy and
chloro epimers, respectively. TLC of the chlorination reaction
mixture showed clean conversion of 3 to 4, with a small
amount of unreacted 3 and baseline material. None of the
diastereomers were resolved. The four possible diastereomers
were resolved using reversed-phase HPLC. However, the RP-
HPLC result was not consistent with the TLC. It indicated
that the reaction mixture contained approximately 50% of
product 4 (mainly one epimer) and 50% of starting material
3, also mainly one epimer. Normal-phase HPLC was
consistent with the TLC and showed the reaction conditions
used gave greater than 90% conversion to 4, with 3-10%
of 3 remaining. These observations suggested that one epimer
of the product hydrolyses rapidly in the RP-HPLC, whereas
the other is relatively stable.
reaction conditions5 and was developed for the preparation
of phosphonate 8.
Trimethylphosphite, triethylphosphite, and tributylphos-
phite do not react with the chloro compound 4, and therefore
the chloride was converted into iodide by reaction with
sodium iodide (Finkelstein reaction). This was initially
performed by adding sodium iodide to the 4 reaction solution,
after the aqueous workup and drying. This procedure gave
inconsistent yields and purities of the iodo compound 9, due
to the low solubility of sodium iodide in dichloromethane.
Dosing known amounts of water into the reaction mixture
and analysis by KF titration showed that when the dichlo-
romethane contained sufficient water to dissolve enough
sodium iodide to allow the reaction to proceed, significant
hydrolysis occurred.
Alternative solvents for the Finkelstein reaction were
evaluated. The use of acetone (and other ketone containing
solvents, e.g., methyl ethyl ketone) was avoided due to its
potential to compete with the internal ketone during the
cyclization of 8. Acetonitrile was found to be a good solvent
for the halide exchange in terms of both product yield and
quality. Some degradation occurred if the 4 reaction solution
was evaporated to dryness and the residue dissolved in
acetonitrile. However, the halide exchange reaction could
be performed by drying and then concentration of the 4
reaction solution after workup and drying and then diluting
with acetonitrile, followed by addition of the sodium iodide.
The charge of sodium iodide is critical to the yield of 8
(and 9). Insufficient sodium iodide (e.g., due to lack of
availability due to solubility issues) results in a yield
reduction through incomplete reaction of 4. Excess sodium
iodide (greater than the optimum charge noted below) causes
decomposition of 8, after its formation by reaction of the
Fortunately, although the reaction was quenched into
saturated brine and dried over magnesium sulfate before
proceeding with the phosphonate formation, the workup
procedure did not cause any significant hydrolysis of 4.
Halide Exchange and Formation of Phosphonate 8.
Phosphonates are commonly prepared by reaction of an alkyl
halide with the trialkylphosphite (Arbuzov reaction) or by
reaction of an alkali metal derivative of the dialkyl phosphate
(Michaelis reaction). The Arbuzov reaction offers simpler
(5) Boutagy J.; Thomas, R. Chem. ReV. 1974, 1, 87-99.
Vol. 11, No. 4, 2007 / Organic Process Research & Development
•
743

Scheme 2. Synthesis of 2 via the trimethyl phosphite [P(OMe)3] route
excess iodide anions with the phosphonate methyl groups,
as shown in Figure 1. The optimum charge was determined
as 1.05 mol equiv, based on the initial charge of 3.6 Halide
exchange is complete within minutes of the addition of
sodium iodide.
Our experience with the Wittig synthesis suggested that
the use of the least sterically hindered trialkylphosphite for
the Arbuzov reaction with 9 would be advantageous. In our
hands, only trimethylphosphite [P(OMe)3] gave good conver-
sion of 9 to the corresponding phosphonate derivative.
The phosphonate 8 was prepared by addition of P(OMe)3
to the solution of 9. The reaction of P(OMe)3 with 9 is
exothermic, and this required careful control since the rate
of production of the phosphate impurity was increased with
temperature (as shown in Figure 1). The exotherm was
controlled by cooling the solution of 9 to below 5 °C before
the addition and by slow addition of the P(OMe)3 as a
solution in dichloromethane.
The optimum P(OMe)3 charge was found to be 1.45 mol
equiv, based on the initial charge of 3. Lower charges of
P(OMe)3 gave incomplete conversion of 9 to 8, and higher
charges gave rise to problems later in the synthesis (in the
PCl5 deprotection) due to the telescopic design of the process.
The phosphonate is fully formed after reaction for 1.5 h
at room temperature. An HPLC solution assay for 8 was
developed, which showed a yield of 75% from 3. It was
important to determine the content of 8 in the reaction
mixture so that subsequent reagent charges could be based
on the result.
Cyclization. The cephem six-membered ring cyclization
is performed by addition of lithium chloride7 and an organic
(6) The amount of the phosphate impurity in the reactions with minimal or no
excess sodium iodide was initially very low. It was found to increase in the
reaction solutions on storage for 24 h at ambient temperature if excess
sodium iodide was present. However, 8 solutions may be held at lower
temperatures (4 °C and -20 °C) for up to 4 days.
(7) Blanchette, M. A.; Choy, W.; Davis, J. T.; Essenfeld, A. P.; Masamune, S.;
Roush, W. R.; Sakai, T. Tetradedron Lett. 1984, 25, 2183-2186 and
references cited within.
744
•
Vol. 11, No. 4, 2007 / Organic Process Research & Development

formation of the double-bond isomer. The amount of 8 in
the reaction solution was determined by HPLC assay, and
the DIPEA and lithium chloride charges were based on the
result.
After addition of DIPEA and lithium chloride the solution
was stirred at ambient temperature to effect the cyclization,
which required more than 16 h to go to completion. The use
of a higher reaction temperature and/or significantly longer
reaction times led to an increase in side products and lower
yield.
It was found that residual water in the cyclization reaction
mixture resulted in the formation of impurities and lower
yields. Therefore, the solution of 8 was dried over magnesium
sulfate before addition of the lithium chloride and DIPEA.10
The yield of 7 from 3 in the above reaction is comparable
to that in the original Me3P route.11
Deprotection. The deprotection uses the standard condi-
tions in cephalosporin chemistry, phosphorous pentachloride,
picoline, and then isobutanol.12 Compounds 9 and 8 require
the presence of acetonitrile in the reaction solution. However,
it was necessary to remove the acetonitrile prior to proceed-
ing with the final deprotection of compound 7 reaction
because acetonitrile reacts with the phosphorous pentachlo-
ride, and it increases the solubility of the product (2) in the
reaction mixture and results in a lower yield.13
There were two opportunities to remove the acetontrile:
after formation of 8 or after formation of 7. There are two
methods for removal of the acetontrile: distillation and phase
extraction. Studies found that removal of the acetonitrile after
formation of 8 by distillation affected the product impurity
profile and yield of 7. Likewise, removal of acetonitrile by
extraction of the reaction mixture containing 8, prior to
cyclisation, leads to emulsions, low yield, and recovery and
issues with the reaction water content when proceeding to
the subsequent cyclisation step. Thus the only step available
to remove the acetonitrile was immediately prior to the
deprotection of 7. The reaction mixture was extracted with
acid solution to remove the DIPEA salts, followed by brine,
and this removed some of the acetonitrile. The reaction
mixture was then distilled twice, to ensure complete removal
of acetonitrile.
Two major issues with the conversion of 7 to 2 are the
residual water content and control of reaction temperature/
exotherms. These are common to both the phosphite and
phosphine routes. The water content needs to be low, and
this is achieved through the distillation procedure to remove
the acetonitrile.
In addition, it has been found that the deprotection
reaction works consistently well on 7 that has been isolated
and purified14 but is variable when using 7 produced via this
telescoped series of reactions from 3. This suggests that some
Figure 1. Proposed mechanism for the formation of the
phosphate impurity.
soluble base to the 8 reaction solution. The reaction proceeds
via formation of a (stabilized) phosphonate anion, which
cyclizes internally to give the product, 7, which contains the
fully formed bicyclic cephalosporin nucleus.
At least 2 mol equiv of lithium chloride were required
for successful cyclization. Excess lithium chloride had no
deleterious effect.
A number of bases were investigated, and diisopropyl-
ethylamine, DIPEA, was found to be the most successful.8
Weaker bases than DIPEA were unsuccessful, probably
because they are not able to deprotonate the phosphonate.
A major difference between the phosphite and phosphine
routes is the potential for ∆2-3 double bond isomerisation
during the cyclisation step in the phosphite method. Isomer-
ization of the double bond in the cephalosporin ring is
promoted by a base.9 In the Wittig synthesis the ylide is
formed by treatment of the phosphonium salt in dichlo-
romethane with aqueous sodium bicarbonate. The organic
phase is separated and allowed to cyclize at ambient
temperature, which takes up to 16 h. Since DIPEA is a
stronger base than bicarbonate and it is difficult to remove
from the reaction until after the cyclization, the DIPEA
charge is critical. The amount of isomerization is directly
related to the DIPEA charge. The optimal amount of DIPEA
is in the range 1.20 to 1.50 equiv, based on the mole amount
of 8. This ensures complete reaction and minimizes the
(10) Compound 8 interferes with the KF reagents, and the water content of the
reaction solution was unknown. The magnesium sulfate was sufficient to
provide a clean reaction.
(11) See ref 6.
(12) Standard deprotection conditions in cephalosporin chemistry.
(13) If the product does not precipitate from the reaction mixture, it decomposes
reasonably rapidly.
(14) A reference standard of 7 was prepared by coupling of 2 with phenylacetic
acid.
(8) DBU and other soluble bases can also be used successfully. The use of
hetergeneous bases was not successful.
(9) Base causes double bond isomerization.
Vol. 11, No. 4, 2007 / Organic Process Research & Development
•
745

other component(s) in the reaction solution have a detrimental
effect on the deprotection reaction.
through a bed of celite, and the celite was washed with
dichloromethane (150 mL). The phosphonate content was
determined by HPLC assay (36.5 g, 56.1mmol). Lithium
chloride (5.11 g) (120.5 mmol, 2.15 equiv relative to the
phosphonate) and DIPEA (12.6 mL) (72.3 mmol, 1.29 equiv
relative to the phosphonate) were added. The solution was
stirred at ambient temperature for 16 h. The reaction solution
was successively washed with 1% aqueous hydrochloric acid
(400 mL) and 20% brine solution (2 × 400 mL). The organic
phase was dried with powered 4 Å molecular sieves (22.3
g) and celite (20.3 g). The desiccant was decanted off through
a plug of silica G (43 g) and washed with dichloromethane
(200 mL). The solution was concentrated to a thick oil on a
rotary evaporator at less than 35 °C, and dichloromethane
(350 mL) was added. This solution was then reconcentrated
to a thick oil on a rotary evaporator at less than 35 °C, and
dichloromethane (350 mL) was added. The water content
was determined to be 140 ppm. The 7 content was deter-
mined by HPLC assay as 25.76 g (49.2 mmol, 67.0% yield
from 3, 87.6% for the cyclization). The solution was cooled
to -55 °C, and phosphorous pentachloride (30.4 g)
(147.4mmol, 3.0 equiv of 7) was charged. After 5 min,
2-picoline (29 mL) (293.6mmol, 6.0 equiv of 7) was added,
maintaining the temperature below -40 °C. An exotherm
was observed. The solution was stirred for 1 h below -20
°C. At this stage the reaction was a thick slurry. It was cooled
to below -50 °C, and isobutanol (205 mL, 2.02mol) was
charged. This caused the reaction to warm to -30 °C. The
solution was allowed to warm to ambient temperature, and
after stirring for 1 h, a seed crystal of 2 was added. The
solution was stirred for 16 h in a closed system to avoid
evaporation of the dichloromethane. The solid was collected
by filtration. The solid was washed with dichloromethane
(2 × 100 mL). The solid was dried to constant weight at 40
°C under high vacuum to give 2 (18.4 g) (41.64mmol, 56.7%
yield from 3, 84.6% yield from 7).
Dimethylphosphate (DMP) is a byproduct of the cycliza-
tion reaction. DMP and the excess P(OMe)3 were shown to
negatively effect the deprotection and are not removed by
the aqueous workup of 7. Based on this observation the
excess P(OMe)3 charge used in the preparation of 8 from 9
was kept at a minimum, which was found to be 1.45. There
is some data to suggest that 10 Å molecular sieves remove
the phosphorous compounds from the reaction mixture.
Conclusion
The phosphite route has been shown to be robust. On a
50 g scale three lots of 2 were prepared in very similar yields
(50-55%). The overall yield is comparable with the best
achievable with the phosphine method. Batches of 2 prepared
by the Horner-Emmons-Wadsworth method were found
to have a similar impurity profile to that produced by the
original phosphine (Wittig) method. They have been used
to prepare Cefovecin 1 that met all of the current test
specifications for drug substance release, and the process was
subsequently scaled for production operations.
Experimental Section
(6R,7R)-4-Nitrobenzyl 7-Amino-8-oxo-3-((S)-tetrahy-
drofuran-2-yl)-5-thia-1-aza-bicyclo[4.2.0]oct-2-ene-2-car-
boxylate Hydrochloride Salt, 2. Compound 3 (51.19 g)
(80% potency, 73.4 mmol) was dissolved in dichloromethane
(750 mL). 2-Picoline (11.8 mL) (119.5 mmol, 1.63 equiv)
was added, and the solution was cooled to -15 °C. Thionyl
chloride (7.6 mL) (104.19 mmol, 1.42 equiv) was added in
one portion (over approximately 3 min). The reaction was
stirred for 1 h below -20 °C. It was washed with 20% brine
solution (2 × 250 mL) and dried over 40 g of magnesium
sulfate, for 10 min at ambient temperature. The desiccant
was filtered off and washed with 100 mL of dichloromethane.
The filtrate was concentrated to 150 mL on a rotary
evaporator at less than 35 °C. Acetonitrile (150 mL) was
added, and the solution was further concentrated to 200 mL
at less than 35 °C. The solution was cooled to less than 5
°C. Sodium iodide (11.59 g) (119.5 mmol, 1.05 equiv to 3)
was charged followed by trimethylphosphite (12.6 mL)
(106.8 mmol, 1.45 equiv to 3) dissolved in dichloromethane
(10 mL), added dropwise over 10 min. The temperature was
maintained at or below 5 °C during the addition. No
exotherm was observed on this scale. The solution was
allowed to warm to room temperature over 1.5 h. The
phosphonate content was determined by HPLC assay (36.49
g, 56.2 mmol). This corresponds to a yield of 76.5% for the
two steps. Dichloromethane (500 mL) was added (total
volume approximately 700 mL). Activated carbon (17 g) and
magnesium sulfate (20.1 g) were added, and the mixture was
stirred for 10 min. The mixture was clarified by filtration
Mp 202 °C decomp. 1H NMR (400 MHz, D2O/CD3CN):
δ 1.62 (m, 1H), 1.85 (m, 2H), 2.15 (m, 1H), 3.37 (d, 1H, J
17.6 Hz), 3.57 (d, 1H, J 17.6 Hz), 3.7(m, 1H), 3.85 (m, 1H),
4.91 (dd, 1H), 4.95 (d, 1H, J 4.9 Hz), 5.13 (d, 1H, J 4.9
Hz), 5.31 (dd, 2H), 7.58 (d, 2H, J 8.9 Hz), 8.18 (d, 2H, J
8.9 Hz). 13C NMR (100 MHz, D2O/CD3CN): δ 24.90, 26.92,
32.71, 67.59, 55.96, 59.34, 70.58, 78.21, 123.02, 143.80,
124.81, 130.08, 142.13, 148.84, 161.86, 162.59. MS (m/z)
406.3 amu (M + H), C18H20ClN3O6S requires: C, 48.92; H,
4.70; N, 9.36; S, 7.26. Found: C, 49.00; H, 4.74; N, 9.44;
S, 6.84%
Characterization and isolation of normally nonisolated
intermediates found between 3 and 2 in the synthetic pathway
are available from the patent literature.15
Received for review April 27, 2007.
OP700091D
(15) Morita, H.; Nagakura, I.; Norris, T. WO 2005092900.
746
•
Vol. 11, No. 4, 2007 / Organic Process Research & Development