Efficient and practical synthesis of a potent anti-MRSA β- methylcarbapenem containing a releasable side chain

Article abstract of DOI:10.1021/ja992486t

We describe a convergent synthesis of the MRSA β-methyl carbapenem 1, wherein the molecule is assembled from the naphthosultam side chain 2 and the allylic carbonate of the β-Me carbapenem piece 6. The β-Me stereochemistry of 6 is set up in a novel titani

Full text of DOI:10.1021/ja992486t

VOLUME 121, NUMBER 49
DECEMBER 15, 1999
© Copyright 1999 by the
American Chemical Society
Efficient and Practical Synthesis of a Potent Anti-MRSA
â-Methylcarbapenem Containing a Releasable Side Chain
Guy R. Humphrey,* Ross A. Miller,* Philip J. Pye,* Kai Rossen,*
Robert A. Reamer, Ashok Maliakal, Scott S. Ceglia,
Edward J. J. Grabowski, R. P. Volante, and Paul J. Reider
Contribution from the Department of Process Research, Merck Research Laboratories, P.O. Box 2000,
Rahway, New Jersey 07065
ReceiVed July 14, 1999
Abstract: We describe a convergent synthesis of the MRSA â-methyl carbapenem 1, wherein the molecule is
assembled from the naphthosultam side chain 2 and the allylic carbonate of the â-Me carbapenem piece 6.
The â-Me stereochemistry of 6 is set up in a novel titanium enolate addition into the TBDMS acetoxy azetidinone
5. The benzenesulfonate salt of 1 is endowed with exceptional stability.
The ability of pathogenic bacteria to rapidly develop resis-
tance to the existing repertoire of antibacterial agents poses a
constant challenge for medicinal chemistry and the pharmaceuti-
cal industry. Consequently, an urgent need exists to expand the
quality of the arsenal of compounds at the disposal of the
medical community in order to prevent an upsurge of often-
fatal bacterial infections. Prominent among the many antibacte-
rial agents are the â-lactams, and especially the carbapenems,
such as imipenem.1 Carbapenems are endowed with broad-
spectrum antibacterial properties combined with good tolerability
and are consequently one of the most potent weapons in the
fight against bacterial infections. Work from numerous labo-
ratories has revealed some important structure-activity relation-
ships: addition of a â-Me group in the 1-position increases the
metabolic and chemical stability, while attachment of aryl groups
either directly or through a heteroatom linker to the 2-position
of the carbapenem results in exquisitely potent antibacterial
agents.2 A recent, very promising carbapenem development
candidate with anti-MRSA (Methicillin-resistant Staphylococcus
aureus) activity is 1, which attaches a naphthosultam side chain
2 through a methylene linker to the carbapenem nucleus3
(Scheme 1). This arrangement results in expulsion of the side
chain when the â-lactam has acylated the surface of red blood
cells. Consequently, the red blood cells are not labeled with
the potentially immunogenic naphthosultam side chain and it
could be hoped that hemolytic anemia, i.e., the lysis of the
marked red blood cells by the immune system, would not pose
a problem for this compound.4 Retrosynthetically, we planned
to assemble 1 by a Pd-catalyzed coupling5 between the allylic
carbonate of the â-methyl carbapenem 6 and the fully elaborated
naphthosultam side chain 26 (Scheme 2). Two different strategies
were pursued for the preparation of the carbapenem coupling
partner 6. In the first approach, a Pd-catalyzed Stille coupling
between the enoltriflate 3 and n-Bu3SnCH2OH is used to install
the methylene linker on the carbapenem nucleus.7 While this
approach has been used to successfully prepare material, the
route is long and requires significant efforts to ensure that the
(3) (a) Ratcliffe, R. W.; Wilkening, R. R.; Wildonger, K. J.; Waddell,
S. T.; Santorelli, G. M.; Parker, D. L., Jr.; Morgan, J. D.; Blizzard, T. A.;
Hammond, M. L.; Heck, J. V.; Huber, J.; Kohler, J.; Dorso, K. L.; St. Rose,
E.; Sundelof, J. G.; May, W. J.; Hammond, G. G. Bioorg. Med. Chem. Lett
1999, 9, 679. (b) Wilkening, R. R.; Ratcliffe, R. W.; Wildonger, K. J.;
Cama, L. D.; Dykstra, K. D.; DiNinno, F. P.; Blizzard, T. A.; Hammond,
M. L.; Heck, J. V.; Dorso, K. L.; Rose, E. St.; Kohler, J.; Hammond, G. G.
Bioorg. Med. Chem. Lett 1999, 9, 673.
* E-mail: kai_rossen@merck.com.
(1) Backes, J. In Methoden der Organischen Chemie (Houben Weyl);
Klamann, D., Ed.; Georg Thieme Verlag: Stuttgart, 1991; Band E16b.
(2) Kawamoto, I. Drugs Future 1998, 23, 181.
10.1021/ja992486t CCC: $18.00 © 1999 American Chemical Society
Published on Web 11/24/1999

11262 J. Am. Chem. Soc., Vol. 121, No. 49, 1999
Humphrey et al.
Scheme 1
Scheme 2
final drug is free from the organotin byproducts. Our goal was
develop a short, practical, and high yielding, as well as
environmentally benign, route to 1 that prepares the carbapenem
nucleus containing the activated methylene linker (6) directly
from the commercially available TBDMS acetoxy azetidinone
5. To this end, the â-methyl carbapenem coupling partner 6
would be made using a reductive cyclization of the oxalimide
7,8 which should in turn be easily attainable from the ketone
8a. The key problem is thus the ability to prepare the â-methyl-
containing hydroxymethyl ketone 8a in an efficient manner from
5.
Scheme 3
The highly desirable properties of â-methyl-containing car-
bapenems and the difficult synthetic accessibility of this class
of compounds have resulted in a large body of publications,
mostly from industrial laboratories.9 Unfortunately, the published
approaches target the â-Me carboxylic acid 9 and require
complex and cumbersomely prepared synthons of the propionic
acid enolate to achieve the desired diastereoselectivity in the
addition. Additionally, while the required extension from the
carboxylic acid to the hydroxymethyl ketone can be ac-
complished, a direct approach from the TBDMS acetoxy
azetidinone 5 would be preferable.10
of the enolate prepared from 11 by the addition into LHMDS
resulted in a 33% yield of 10, albeit as a 1:1 mixture of R- and
â-Me diastereomers. A systematic variation of reaction condi-
tions was undertaken at this point.
To this end, we decided to examine enolates and enolate
equivalents of the commercially available 1-hydroxy-2-butanone
(11) (Scheme 3). Initial results appeared promising: addition
Surprisingly, no reports on the use of the 1-hydroxy-2-
butanone enolate could be found in the literature and it was
clear that both the regiochemistry and enolate geometry would
have to be controlled. Addition of 11 to an excess of LHMDS
followed by a quench with TMSCl resulted in poor regioselec-
tivity of enolate formation, but with good selectivity for the
Z(O) enol ether 12. All attempts to improve on this result by
use of different bases or the addition of salts resulted in lower
selectivities. Fortunately, the combination of TMSOTf and Et3N
at -60 °C in CH2Cl2 cleanly led to the desired Z(O) enol ether
12. Reaction of 12 with acetoxy azetidinone 5 using ZnCl2
catalysis gave a small selectivity for the undesired R-Me
stereochemistry (66:34). Changing the NH group of the lactam
for the NTBDMS group (13) and use of TMSOTf as catalyst
improved the diastereoselectivity of the reaction to give a 10:1
ratio of the undesired R-diastereomer (Scheme 4).While the high
diastereoselectivity to the undesired R-diastereomer was disap-
pointing, it was encouraging to obtain the desired enolate
regiochemistry using a combination of Lewis acid and base.
To expand on this lead 11 was protected as the carbonate using
standard conditions (isobutylchloroformate, pyridine, toluene)
and the resulting 14 was treated with TiCl4 in toluene (Scheme
(4) (a) Rosen, H.; Hajdu, R.; Silver, L.; Kropp, H.; Dorso, K.; Kohler,
J.; Sundelof, J.G.; Huber, J.; Hammond, G. G.; Jackson, J. J.; Gill; C. J.;
Thompson, R.; Pelak, B. A.; Epstein, T.; Jeffrey H.; Lankas, G.; Wilkening,
R. R.; Wildonger, K. J.; Blizzard, T. A.; DiNinno, F. P.; Ratcliffe, R. W.;
Heck, J. V.; Kozarich, J. W.; Hammond, M. L. Science 1999, 283, 703. (b)
Lankas, G. R.; Coleman, J. B.; Klein, H. J.; Bailly, Y. Toxicology 1996,
108, 207.
(5) For a review of Pd-catalyzed π-allyl coupling, see: Godleski, S. A.
In ComprehensiVe Organic Synthesis; Trost, B. M., Fleming, I., Ed.;
Pergamon Press: Oxford, 1991; Vol. 4, p 585.
(6) The synthesis of the side chain 2 is disclosed in a separate
publication: Miller, R. A.; Humphrey, G. R.; Lieberman, D. R.; Ceglia, S.
S.; Grabowski, E. J. J. J. Org. Chem., manuscript submitted.
(7) Yasuda, N.; Yang, C.; Wells, K. M.; Jensen, M. S.; Hughes, D. L.
Tetrahedron Lett. 1999, 40, 427.
(8) For some recent applications of this reductive cyclization, see: (a)
King, S. A.; Pipik, B.; Thompson, A. S.; DeCamp, A.; Verhoeven, T. R.
Tetrahedron Lett. 1995, 36, 4563. (b) Hanesian, S.; Rozema, M. J. J. Am.
Chem. Soc. 1996, 118, 9884.
(9) For a comprehensive review of the chemistry of 9, see: Berks, A.
H. Tetrahedron 1996, 52, 331.
(10) (a) Yang, C. Yasuda, N. Bioorg. Med. Chem. Lett. 1998, 8, 255.
(b) Hu, X. E.; Demuth, T. P. J. Org. Chem. 1998, 63, 1719. (c) We have
realized an alternative, much simpler approach from the acid 9 to the
hydroxymethyl ketone â-10: activation of 9 using carbonyldiimidazole is
followed by reaction with the glycolic acid enolate. Acid workup with
concomitant decarboxylation gives â-10 in 72% isolated yield. (d) Choi,
W. B. Chem. Commun. (Cambridge) 1998, 17, 1817.

Synthesis of a Anti-MRSA â-Methylcarbapenem
J. Am. Chem. Soc., Vol. 121, No. 49, 1999 11263
Scheme 6a
Scheme 4
Scheme 5a
a Conditions: (a) 0.5 equiv of 2,6-lutidine, 3 mol % Pd(OAc)2, 9
mol % (BuO)3P, NMP, 35 °C; (b) TfOH, H2O, IPA pH ) 2.6; pH )
6.85 MOPS buffer, 5% Pd/C, 40 psi H2; PhSO3Na, H2O, IPA.
from TBDMS to the more labile TES group is performed.
Removal of the TBDMS ether in 8a was best accomplished
using aqueous 2 N HCl in a homogeneous CH3CN/toluene
mixture, followed by neutralization of the acid with solid
NaHCO3 and isolation of 8b after a filtration to remove salts
and a concentration from toluene. This simple procedure gave
8b in 64% overall yield from 5 and additionally served to
improve the diastereomeric purity in the crystallization of 8b
(>99% â-Me in isolated product). Reprotection of the hydroxy-
ethyl side chain was accomplished using standard conditions
(TESCl, imidazole, toluene/CH3CN), and the resulting 8c was
not isolated but used directly in the next step. The elaboration
of 8c to the oxalimide 7 was accomplished by addition of 8c to
a slurry of the pyridinium salt of p-nitrobenzyl oxalyl chloride
in toluene. Aqueous work up and crystallization gave the
oxalimide 7 in 97% overall yield. The reductive cyclization of
7 was easily accomplished using 5 equiv of P(OEt)3 in refluxing
heptane for 3 days.13,8 Oxidation of the excess phosphite to
triethyl phosphate (H2O2 in pH 6 phosphate buffer) was followed
by aqueous washes to remove the phosphate. A subsequent
solvent switch from heptane to 2-butoxyethanol and addition
of water precipitated the desired 6 in 85% yield (Scheme 5).
a Conditions: (a) 14, TiCl4, Bu3N, PhMe, -40 °C to -5 °C; (b) 2
N HCl (aq), MeCN, PhMe, 64% from 5; (c) TESCl, imid, PhMe,
MeCN; (d) pyr, ClOCCO2pNB, PhMe, 97% from 8b; (e) POEt3,
heptane, 92 °C, 85%.
5). A yellow precipitate formed,11 which was treated at -40
°C with 4 equiv of n-Bu3N.12 To the resulting deep red solution
of the Ti enolate was added a solution of 5 in toluene, and the
mixture was allowed to stir for 5 h at 5 °C. An inverse quench
into 2 N HCl resulted in a 82% yield of the desired product 8a
with excellent 95:5 â:R stereoselectiVity (Scheme 5).
To gain some understanding of the mechanism, we attempted
to elucidate the nature of the intermediate Ti enolate. Direct
quench of the Ti enolate with electrophiles other than DCl/D2O
failed, but low-temperature transmetalation with MeLi followed
by addition of TMSCl gave predominantly the Z(O) silyl enolate
12 (>10:1 Z:E). This apparent regio- and stereoselective
formation of the Z(O) enolate was confirmed by NMR experi-
ments, where the enolate was generated in toluene-d8 using
Et3N-d15. Again, the resulting enolate was the Z(O) enolate (10:1
Z:E). While conclusive statements on the mechanism of the
reaction between the Ti enolate with acetoxy azetidinone are
clearly not possible without extensive experimental work, the
observed enolate geometry is consistent with a closed cyclic
transition state in the addition of the titanium Z(O) enolate to
the putative acyl iminium intermediate.
The next synthetic challenge for the elaboration of 1 was the
coupling of the naphthosultam side chain 2 to 6 (Scheme 6).
The desirable convergent route would bring in the fully
elaborated side chain to convert 6 in one step to the protected
precursor of 1 (16), thus, minimizing manipulations with the
sensitive carbapenem. The choice of carbonate protection of
the hydroxymethyl side chain of 6 offers the opportunity to
exploit this protecting group as an activating group by utilizing
the chemistry of cationic Pd-π-allyl systems.14 Significantly,
both the generation of the Pd-π-allyl species and its coupling
with the complex dicationic 1,8-naphthosultam DABCO aceta-
With an efficient access to 8a at hand, we were now in a
position to complete the synthesis of the key penem nucleus 6
(Scheme 5). The lability of carbapenem 1 made it necessary to
use protecting groups throughout the sequence that could be
removed using a mild set of conditions for the final elaboration
of the drug substance. Consequently, a protecting group change
(11) Poll, T.; Metter; J. O.; Helmchen G. Angew. Chem., Int. Ed. Engl
1985, 24, 112.
(12) (a) Evans, D. A.; Rieger, D. L.; Bilodeau, M. T.; Urpi, F. J. Am.
Chem. Soc. 1991, 113, 1047. (b) Yoshida, Y.; Hayashi, R.; Sumbara, H.;
Tanabe, Y. Tetrahedron Lett. 1997, 38, 8727.
(13) Heptane was chosen for its commercial availability; running the
reaction in nonane or in heptane under pressure at 120 °C allows completion
of the reaction in 3 h.

11264 J. Am. Chem. Soc., Vol. 121, No. 49, 1999
Humphrey et al.
mide nucleophile 2 would clearly rank among the most complex
and challenging examples for this reaction recorded in the
literature.7
salt is endowed with exceptional thermal stability in the
crystalline state: essentially no decomposition is observed after
heating the solid salt for 2 h at 100 °C, in marked contrast to
observations for other salts of 1, or other carbapenems in
general.
Conventionally, the activation of allylic carbonates to the Pd-
π-allyl species is done using Ph3P both to reduce the Pd(OAc)2
catalyst precursor to the catalytically active Pd(0) species and
to prevent the precipitation of metallic Pd black. Indeed,
coupling of the partly elaborated side chain 15 with the isobutyl
carbonate 6 occurs in essentially quantitative yield with 3 mol
% Pd(OAc)2 and 9 mol % of Ph3P in a biphasic toluene/aqueous
Rochelle’s salt solution system at 80 °C in 2 h. Unfortunately,
it was not possible to use these conditions with the fully
elaborated side chain 2, as Hofmann elimination of the
quaternary DABCO piece became dominant. Thus, reaction
conditions were varied with the goal of finding a more highly
reactive catalyst system that would avoid this problem. Contrary
to literature reports, phosphite ligands resulted in catalysts of
exquisite reactivity: use of (BuO)3P instead of Ph3P with 15
gives complete conversion at 35 °C in less than 30 min! Indeed,
application of this highly active catalyst system makes it possible
to couple the carbapenem piece 6 with the fully elaborated side
chain 2 using NMP as solvent and 2,6-lutidine as base in
quantitative assay yield.15 Isolation of the protected 1 (16) is
accomplished in a straightforward manner from the reaction
mixture by the addition of water and isopropyl alcohol, which
furnishes crystalline product in excellent purity in 97% yield.
A key issue for the use of metal-catalyzed reactions for the
preparation of pharmaceuticals is the removal of the metal
catalysts down to trace levels in the bulk drug. Fortunately, the
crystallization after the Pd-π allyl coupling leaves the Pd metal
largely in the mother liquors, the Pd level in the isolated bulk
is very low at <100 ppm Pd.
In summary, we have described a simple and convergent
process for the preparation of the anti-MRSA â-Me carbapenem
1, which allows the preparation of this important antibiotic in
38% overall yield from the readily available acetoxy azetidinone
5. Preparation of the â-Me carbapenem coupling partner 6 is
accomplished in five steps from the readily available acetoxy
azetidinone in 52% yield, whereby the required â-Me stereo-
chemistry is set up in a highly diastereoselective reaction using
a novel titanium enolate. Subsequent attachment of the complete
side chain 2 uses a Pd-π-allyl cation intermediate and occurs
in near quantitative yield. Subsequent deprotections and isolation
of the desired 1 as the exceptionally stable benzenesulfonate
salt occurs in a remarkable 73% yield from 6. The route
described uses environmentally acceptable chemistry, is ame-
nable to large-scale preparations, and has been used to prepare
1 on a multikilogram scale.
Experimental Section
General Procedure. All reactions were carried out in glassware that
was dried in a N2 stream overnight. All reagents were commercial grade
and were used as received. BASF is a commercial supplier of
1-hydroxy-2-butanone. tert-Butyldimethylsilyl acetoxy azetidinone was
purchased from Takasago. HPLC purities refer to area % measured at
210 nm.
p-Nitrobenzyloxalyl Chloride.16 A 100-L flask equipped with an
overhead stirrer, thermocouple, nitrogen inlet, and an outlet leading to
a 50% aqueous sodium hydroxide scrubber was charged with dichlo-
romethane (44 L) and the temperature was adjusted to 6 °C. Oxalyl
chloride (5.81 kg) was added followed by p-nitrobenzyl alcohol (4.49
kg) in ∼500 g portions over 30 min. The reaction mixture was aged at
6 °C for 2 h. Dichloromethane (40 L) was removed by distillation over
4 h. Toluene (9.4 L) was added and heating was resumed. When the
distillation was complete the temperature was allowed to rise to 73 °C
for 10 min. The heating was stopped and cyclohexane (1 L) was added
to precipitate the undesired bis p-nitrobenzyl oxalate. Bis-p-nitrobenzyl
oxalate was removed by filtration under a blanket of nitrogen.
Cyclohexane (33 L) was added and the slurry was aged for 2 h at 11
°C. The precipitated p-nitrobenzyloxalyl chloride was filtered off and
allowed to dry under a nitrogen sweep to afford p-nitrobenzyloxalyl
chloride (5.31 kg, 74% yield).
The completion of the synthesis requires the removal of the
TES and p-PNB protecting groups from 16, as well as the
development of a procedure that allows for the isolation of the
unprotected 1. To this end, the TES group was removed using
3.5 mol % of triflic acid in IPA at 20 °C for 18 h. Adjustment
of the pH to 6.9 by the addition of aqueous MOPS buffer,
addition of 5% Pd/C catalyst and hydrogenolysis at 3 bar H2
for 1 h gave the desired free 1. After the filtration of the Pd/C
catalyst, the nonpolar byproducts of the deprotection steps could
be removed by washing the buffered aqueous phase with toluene
to give 1 in 90% yield.
The isolation of pure, crystalline, stable carbapenem products
from deprotection solutions is a formidable problem as these
products have limited stability in aqueous buffered solutions
due, principally, to â-lactam solvolysis. Initial purification is
usually achieved via resin chromatography followed by reverse
osmosis concentration of the rich cuts. Crystallization, if possible
at all, is typically difficult even for the chromatographed
material. Initial isolations of 1 were done in this standard
manner. While typical salts of 1 are not suitable for direct
isolation (i.e., chloride or sulfate), we were pleased to note that
addition of sodium benzenesulfonate to a worked-up hydrogena-
tion reaction mixture led to the ready crystallization of 1 as its
benzenesulfonate salt trihydrate (X ) PhSO3-) in 75% yield.
The isolated product has excellent purity and is essentially free
from palladium (<2 ppm). In addition to providing a simple
and elegant method for the isolation of 1, the benzenesulfonate
2-Oxobutyl Isobutyl Carbonate (14). A 100-L flask equipped with
an overhead stirrer, thermocouple, and a nitrogen inlet was charged
with 1-hydroxy-2-butanone (5.14 kg, 58.4 mol), hexanes (12 L),
cyclohexane (18 L), and pyridine (6.16 L). The temperature was
adjusted to 10 °C. iso-Butyl chloroformate (7.90 L) was added over
2.5 h. At the end of the reaction, water (20 L) was added, and the
mixture was aged with stirring for 30 min. The lower aqueous layer
was cut away, and the organic layer was further washed with aqueous
HCl (0.5 N; 20.7 L) and water (2 × 20 L). The organic phase was
concentrated to afford 9.68 kg of the 2-oxobutyl isobutyl carbonate
(14) as a pale yellow oil: 1H NMR (400.25 MHz, CDCl3) δ 4.60 (s, 2
H), 3.89 (d, J ) 6.8 Hz, 2 H), 2.39 (q, J ) 7.2 Hz, 2 H), 1.93 (m, 1
H), 1.02 (t, J ) 7.2 Hz, 3 H), 0.89 (d, J ) 6.8 Hz, 6 H); 13C NMR
(100.64 MHz, CDCl3) δ 204.0, 154.7, 74.4, 70.0, 31.6, 27.6, 18.6, 6.8;
IR (neat) 2966, 2878, 1755.7 cm-1. Anal. Calcd for C9H16O4: C, 57.41;
H, 8.57. Found: C, 57.36; H, 8.59.
Hydroxymethyl Ketone Isobutyl Carbonate (8a). A 100-L flask
equipped with an overhead stirrer, thermocouple, and a nitrogen inlet
was charged with 2-oxobutyl isobutyl carbonate (14) (5.47 kg, 58.4
mol) and toluene (28 L) and the temperature was adjusted to -40 °C.
Titanium(IV) chloride (3.05 L) was added over 45 min and the slurry
(14) Mitsunobu reactions of a 2-(hydroxymethyl)carbapenem with N-
nucleophiles have been reported: Arnould, J. C.; Landier, R.; Pasquet, M.
J. Tetrahedron Lett. 1992, 33, 7133.
(15) A strong ligand dependence is noted for the coupling between 6
and 2 in NMP/2,6-lutidine with 3 mol % Pd(OAc)2. The conversion after
5 h at 30 °C is 10% for Ph3P, 93% for (BuO)3P and <2% for Ph3As.
(16) Togo, H.; Fujii, M.; Yokoyama, M. Bull. Chem. Soc. Jpn. 1991,
64, 57.

Synthesis of a Anti-MRSA â-Methylcarbapenem
J. Am. Chem. Soc., Vol. 121, No. 49, 1999 11265
was aged for a further 30 min at -45 °C. Tributylamine (13.5 L) was
added over 1.5 h. The reaction mixture was warmed to -20 °C over 1
h. A solution of tert-butyldimethylsilyl acetoxy azetidinone 5 (4.01 kg,
14.8 mol) in toluene (22 L) was added to the 100-L flask over 1 h.
After 5 h the reaction was quenched by an inverse addition into water
at 5 °C. The organic layer was washed further with aqueous HCl (1.2
N; 2 × 72 L). The organic layer was collected (44.8 kg; â:R ) 96:4;)
nitrogen sweep to afford 9.37 kg of 7 as a pale yellow solid (97%
yield): mp 60-62 °C; [R]25 -81.59° (c 2.50, CHCl3); 1H NMR
D
(400.25 MHz, CDCl3) δ 8.24 (d, J ) 8.0 Hz, 2H), 7.59 (d, J ) 8.0 Hz,
2H), 5.41 (s, 2H), 4.76 (d, J ) 17.3 Hz, 1H), 4.63 (d, J ) 17.3 Hz,
1H), 4.46 (m, 1H), 4.32 (m, 1H), 3.93 (d, J ) 6.8 Hz, 2H), 3.60-3.51
(m, 2H), 1.98 (m, 1H), 1.27 (d, J ) 6.8 Hz, 3H), 1.22 (d, J ) 6.4 Hz,
3H), 0.95 (d, J ) 6.8 Hz, 6H), 0.89 (t, J ) 8.0 Hz, 9H), 0.54 (q, J )
8.0 Hz, 6H); 13C NMR (100.64 MHz, CDCl3) δ 204.5, 164.8, 159.4,
156.0, 154.7, 148.1, 141.1, 129.0, 123.8, 74.8, 69.9, 66.7, 64.7, 61.4,
54.5, 40.9, 27.8, 22.5, 18.8, 14.0, 6.7, 4.8; IR (KBr) 1810, 1754, 1729,
1685, 1607 cm-1. Anal. Calcd for C29H42N2O11Si: C, 55.71; H, 6.73;
N, 4.81; Si, 4.49. Found: C, 55.91; H, 6.47; N, 4.24; Si, 4.24.
and used in the subsequent step: mp 64.0-66.0 °C; [R]25 -5.56° (c
D
2.50, CHCl3); 1H NMR (400.25 MHz, CDCl3) δ 6.14 (s, 1H), 4.74 (d,
J ) 17.0 Hz, 1H), 4.68 (d, J ) 17.0 Hz, 1H), 4.16 (m, 1H), 3.95 (d,
J ) 6.8 Hz, 2H), 3.89 (dd, J ) 2.0, 4.8 Hz, 1H), 2.93-2.86 (m, 2H),
1.99 (m, 1H), 1.20 (d, J ) 7.2 Hz, 3H), 1.18 (d, J ) 6.4 Hz, 3H), 0.95
(d, J ) 6.8 Hz, 6H), 0.86 (s, 9H), 0.06 (s, 3H), 0.05 (s, 3H); 13C NMR
(100.64 MHz, CDCl3) δ 205.6, 168.2, 154.8, 74.8, 69.7, 65.5, 61.8,
51.0, 44.6, 27.8, 25.8, 22.5, 18.8, 17.9, 11.6, -4.3, -5.0; IR (KBr)
3468, 2955, 1763, 1748, 1734, 1722 cm-1. Anal. Calcd for C20H37-
NO6Si: C, 57.69; H, 8.89; N, 3.61 Si, 7.70. Found: C, 57.69; H, 8.70;
N, 3.31; Si, 7.20.
[4S-[4r,5â,6â(S*)]]-4-Methyl-3-[[[(2-methylpropoxy)carbonyl]-
oxy]methyl]-7-oxo-6-[1-[(triethylsilyl)oxy]ethyl]-1-azabicyclo[3.2.0]-
hepten-2-ene-2-carboxylic Acid (4-Nitrophenyl)methyl Ester (6). A
72-L flask equipped with an overhead stirrer, thermocouple, and
nitrogen was charged with 7 (3.50 kg) and heptane (42 L). To this
slurry was added triethyl phosphite (3.50 L), and the reaction mixture
was heated to 92 °C over 1 h. Heating was continued for 72 h and the
reaction was allowed to cool to 22 °C over 14 h. pH 6 phosphate buffer
(20 L) was charged to the vessel followed by 30% hydrogen peroxide
(1.2 L) over 2 h with the temperature maintained below 30 °C. MTBE
(20 L) was added, and after 5 min the layers were allowed to settle.
The lower aqueous phase was removed and the organic layer was
washed with water (4 × 20 L) and dried with sodium sulfate (300 g).
The organic layer was passed through a filter pot containing silica gel-
60 (300 mm dia × 10 mm) and the silica gel was washed with
20%MTBE/heptane (15 L). The solvent was switched into 2-butoxy-
ethanol (7.3 L). Water (2.5 L) was added slowly and the batch was
seeded. After 30 min the batch crystallized, water (20 L) was added,
and the slurry was aged overnight. The solids were collected by filtration
and washed twice with 20% 2-butoxyethanol/water (2 × 5 L) and then
four times with water (4 × 5 L). The solids were dried under a nitrogen
sweep to afford 2.82 kg of the 6 as an off white solid (85% yield, 93
area %, 95 wt %): mp 53-58 °C; [R]25D ) +52.6° (5.20, CHCl3); 1H
NMR (400.25 MHz, DMSO-d6) δ 8.19 (d, J ) 8.6 Hz, 2H), 7.72 (d,
J ) 8.6 Hz, 2H), 5.46-5.31 (m, 3H), 4.83 (d, J ) 14.2 Hz, 1H), 4.25-
4.21 (m, 2H), 3.89 (d, J ) 6.5 Hz, 2H), 3.48-3.46 (m, 1H), 3.32-
3.26 (m, 2H), 1.88 (m, 1H), 1.15 (d, J ) 6.2 Hz, 3H), 1.12 (d, J ) 7.4
Hz, 3H), 0.89 (t, J ) 8.0, 9H), (d, J ) 6.4, 6H), 0.54 (q, J ) 8.0, 6H);
13C NMR (100.64 MHz, DMSO-d6) δ 176.0, 160.6, 154.9, 147.6, 145.3,
143.8, 128.8, 128.3, 123.8, 74.1, 65.7, 65.3, 61.9, 60.7, 55.1, 40.3, 27.7,
22.3, 19.1, 15.4, 7.1, 4.9; IR (KBr) 2963, 1791, 1746, 1716 cm-1; MS
(EI) m/e 399, 429, 443, 461, 517, 561. Anal. Calcd for C29H42N2O9Si:
C, 58.96; H, 7.17; N, 4.74; Si, 4.75. Found: C, 59.10; H, 7.17; N,
4.68; Si, 4.74.
tert-Butyldimethylsilyl Ether Deprotection: Free Hydroxyl (8b).
The toluene solution containing 8a from the previous step was
concentrated to 20 L under vacuum. Acetonitrile (50 L) was charged
to the vessel, and with stirring aqueous HCl (2 N; 2.5 L) was added.
After 3.5 h the reaction was complete and solid sodium bicarbonate
(2.5 kg) was added to quench the reaction. Stirring was continued for
30 min. Solka floc was added, and the slurry was filtered. The filtrate
was evaporated with the addition of toluene until ∼30 L of solution
was left. The product 8b precipitated during the solvent switch. The
organics contained 10% acetonitrile. The slurry of the product was
cooled to 4 °C and aged for 3 h. The solids were isolated by filtration
and then washed with the liquors. The wet cake was slurry-washed
with 20% MTBE/hexanes (3 × 8 L) and then surface washed with
20% MTBE/hexanes (4 L). The cake was dried under a nitrogen sweep
for 18 h to furnish 2.64 kg of 8b as a white solid (63.4%, containing
0.5 A% R isomer): mp 136.0-138.5 °C; [R]25 -39.96° (c 2.5,
D
1
MeOH); H NMR (400.25 MHz, CDCl3) δ 6.40 (s, 1H), 4.78 (d, J )
16.9 Hz, 1H), 4.74 (d, J ) 16.9 Hz, 1H), 4.11 (m, 1H), 3.97 (d, J )
6.8 Hz, 2H), 3.83 (dd, J ) 2.4, 6.8 Hz, 1H), 2.93-2.85 (m, 2H), 2.63
(broad s, 1H), 1.98 (m, 1H), 1.30 (d, J ) 6.4 Hz, 3H), 1.25 (d, J ) 7.2
Hz, 3H), 0.96 (d, J ) 6.8 Hz, 6H); 13C NMR (100.64 MHz, CDCl3) δ
206.0, 168.1, 154.9, 75.0, 69.7, 65.7, 62.2, 52.6, 45.8, 27.8, 21.4, 18.8,
12.6; IR (KBr) 3500, 3200, 1734, 1703 cm-1. Anal. Calcd for C14H23-
NO6: C, 55.63; H, 7.62; N, 4.97. Found: C, 55.91; H, 7.30; N, 4.70.
Triethylsilyl Ether (8c). A 100-L cylindrical vessel equipped with
internal coils, a thermocouple, a nitrogen inlet, and an overhead stirrer
was charged with 8b (4.66 kg), imidazole (1.37 kg), toluene (30 L),
and acetonitrile (6 L). TES chloride was added slowly over 1 h. After
a 1 h age the reaction was quenched by the addition of water (30 L)
and the organic layer was collected and washed again with water (20
L). The solvent was switched into toluene (25 L) and this solution was
used in the subsequent step. Evaporation gave the triethylsilyl ether 8c
as a thick oil; 1H NMR (400.25 MHz, CDCl3) δ 5.91 (s, 1H), 4.76 (d,
J ) 16.9 Hz, 1H), 4.70 (d, J ) 16.9 Hz, 1H), 4.21-4.14 (m, 1H), 3.97
(d, J ) 6.8 Hz, 2H), 3.89 (dd, J ) 2.4, 4.8 Hz, 1H), 2.93-2.89 (m,
2H), 2.06-1.96 (m, 1H), 1.23 (d, J ) 6.8 Hz, 3H), 1.22 (d, J ) 7.2
Hz, 3H), 0.97 (d, J ) 6.4 Hz, 6H), 0.95 (t, J ) 8.0 Hz, 9H), 0.60 (q,
J ) 8.0 Hz, 6H); 13C NMR (100.64 MHz, CDCl3) δ 205.7, 167.9,
154.9, 70.4, 69.7, 65.7, 61.8, 51.4, 44.6, 27.8, 22.7, 18.8, 11.3, 6.8,
5.0; exact mass [M + H] calcd 416.2468 found 416.2441.
Triethylsilyl-p-nitrobenzyl-Protected 1 (16). A 100-L round-bottom
flask equipped with an overhead stirrer, temperature probe, and nitrogen
inlet was charged with 6 (1500 g, 2.54 mol). Next, 6.93 L of 1-methyl-
2-pyrrolidinone was added, followed by 2 (1600 g, 2.3 mol) and 2,6-
lutidine (122.4 g, 1.15 mol). This solution was aged at ambient
temperature for 5 min. The mixture was degassed at ambient temper-
ature with two nitrogen/vacuum cycles. Pd (OAc)2 (17.4 g, 0.076 mol))
was added followed by tributyl phosphite (57.2 g, 0.229 mol). The
reaction mixture was heated to 35 °C, and aged 12-16 h until reaction
was complete by HPLC. The reaction mixture was cooled back to
ambient temperature and quenched with water (0.7 L). Isopropyl alcohol
(10.8 L) was added, and the reaction mixture was seeded and aged at
ambient temperature for 40 min, at which time the product crystallized.
An additional 16.2 L of isopropyl alcohol was then added slowly over
a 40-min period, and the mixture was aged for 30 min at ambient
temperature. The mixture was cooled to 5 °C and aged for 1 h. The
solid was isolated by filtration and washed with cold isopropyl alcohol
(2 × 5.4 L). A total of 2.98 kg (97% yield) of off-white solid was
Oxalimide (7). A 100-L cylindrical vessel equipped with internal
coils, a thermocouple, a nitrogen inlet and an overhead stirrer was
charged with p-nitrobenzyl oxalyl chloride (4.90 kg) and toluene (30
L) and the temperature was adjusted to 10 °C. Pyridine (3.25 L) was
slowly added over 15 min to the slurry. The cooling was stopped and
the reaction mixture aged for 30 min. The solution of 8c in toluene
was added over 10 min. After 17 h the reaction was quenched by the
addition of aqueous citric acid (20 L), and the organic layer was
collected. After hot extraction of the organics (60 °C) with aqueous
sodium bicarbonate (20 L) then water (2 × 20 L) Ecosorb-C (400 g)
was added and the batch was filtered. The solvent was switched into
heptane (25 L). The solids were isolated by filtration and the wet cake
was washed with heptane. The solids were allowed to dry under a
obtained as the di-2-propanoate solvate of 16: mp 143 °C (dec); [R]25
D
) +28.65° (9.98, MeOH); 1H NMR (400.25 MHz, acetone-d6) δ 8.25
(m, 2H), 8.14 (d, J ) 7.4 Hz, 1H), 8.01 (d, J ) 7.5 Hz, 1H), 7.88 (m,
3H), 7.74 (br s, 1H), 7.66 (m, 1H), 7.29 (br s, 1H), 7.01 (d, J ) 7.4
Hz, 1H), 5.62 (d, J ) 14.1 Hz, 1H), 5.45 (d, J ) 14.1 Hz, 1H), 5.40
(d, J ) 17.1 Hz, 1H), 4.83 (d, J ) 17.1 Hz, 1H), 4.71 (s, 2H), 4.64 (s,
12H), 4.30 (om, 4H), 4.01 (m, 2H), 3.89 (m, isopropyl alcohol O-CH-

11266 J. Am. Chem. Soc., Vol. 121, No. 49, 1999
Humphrey et al.
(CH3)2), 3.46 (om, 2H), 3.44 (isopropyl alcohol OH), 1.33 (d, J ) 7.4
Hz, 3H), 1.24 (d, J ) 6.1 Hz, 3H), 1.10 (d, isopropyl alcohol CH3),
0.94 (t, J ) 7.8 Hz, 9H), 0.62 (q, J ) 7.8 Hz, 6H); 13C NMR (100.64
MHz, acetone-d6) δ 174.3, 164.5, 161.1, 147.7, 146.2, 143.6, 137.7,
137.2, 130.3, 129.8, 129.7, 129.6, 129.2, 128.4, 123.4, 121.1 (q, J )
320.1 Hz), 120.1, 119.2, 115.2, 104.5, 65.4, 65.3, 64.2, 62.9, 62.3, 60.4,
55.2, 52.2, 51.5, 40.7, 38.1, 24.9, 24.8, 21.5, 15.0, 6.25, 4.6; IR (KBr)
2960, 1778, 1701, 1669 cm-1; MS (EI) m/e calcd for 874.4 found 873.1.
Anal. Calcd for C46H58F6N6O15S3Si: C, 47.09; H, 4.98; N, 7.16; F,
9.72; S 8.20; Si, 2.39. Found C, 47.12; H, 5.03; N, 7.13; F, 9.72; S,
8.43; Si, 2.38.
aqueous isopropyl alcohol. The filtrate was immediately cooled to 5
°C to improve the stability of the free cation. The filtrate was partitioned
with toluene (4 L), and the layers were separated. The aqueous solution
was added to a solution of sodium benzenesulfonate (1.35 kg) in 4 L
water at 20 °C. The reaction mixture was seeded and aged 1 h until
crystallization occurred. The resulting slurry was cooled to 5 °C and
filtered, and the slurry was washed with 0.5 L of 50% aqueous isopropyl
alcohol. The solid was dried under nitrogen at ambient temperature to
give 255 g of 1 (99.3 area % pure, 73.4% overall yield): mp 150-168
°C (dec); [R]25 ) +136° (0.5, H2O, after anion exchange to Cl-);
436
1H NMR (400.25 MHz, DMSO-d6) δ 8.34 (br s, 1H), 8.18 (d, J ) 7.4
Hz, 1H), 7.88 (br s, 1H), 7.84 (d, J ) 7.4 Hz, 1H), 7.72 (d, J ) 8.6
Hz, 1H), 7.61-7.59 (m, 2H), 7.43-7.39 (m, 1H), 7.32-7.28 (m, 3H),
7.11 (d, J ) 7.4 Hz, 1H), 5.72 (d, J ) 15.9 Hz, 1H), 4.99 (m, 1H),
4.55-4.46 (m, 3H), 4.42-4.24 (m, 12H), 4.05-3.80 (m, 4H), 3.75-
3.55 (m, 2H), 3.1-3.0 (m, 1H), 2.99-2.85 (m, 1H), 1.09 (d, J ) 6.4,
3H), 1.06 (d, J ) 7.2, 3H); 13C NMR (100.64 MHz, DMSO-d6) δ 174.6,
165.2, 164.7, 148.4, 139.7, 138.7, 136.7, 132.6, 130.4, 129.8, 129.3,
129.2, 129.1, 128.2, 126.0, 120.8, 118.7, 115.6, 105.0, 64.7, 63.1, 62.1,
58.9, 55.3, 51.9, 50.9, 39.2, 37.2, 24.8, 22.1, 15.8; IR(KBr) 3426, 1751,
1698 cm-1; MS (EI) cation m/e 401.2, 624.1, 625.1. Anal. Calcd for
C37H49N5O13S2: H, 5.91; C, 53.16; N, 8.38; S, 7.67. Found H, 6.08; C,
52.94; N, 8.38; S, 7.66.
Preparation of Benzenesulfonate of 1. To a 12-L round-bottom-
flask equipped with an overhead stirrer, N2 inlet, and temperature probe
was charged 16 (600 g). In a separate flask, 2 mL triflic acid was added
to 2 L water (pH 1.95), and then added to 6 L of isopropyl alcohol,
(pH 2.6). The resulting solution was added to the TES penultimate,
and the slurry (pH ) 2.9) was stirred at ambient temperature until
reaction was complete (∼18 h).
A buffered solution of 4-morpholinepropanesulfonic acid was
prepared by dissolving 356 g in 6 L water followed by addition of
∼168 mL 5 N NaOH, resulting in a final solution pH of 7.17. The
MOPS buffered solution (4.8 L) was added to neutralize the reaction
slurry (final pH 6.85). The mixture was degassed, 150 g 5% Pd/C (dry
reduced) was added and the system was placed under hydrogen (40
psi) until completion of the H2 uptake (30 min). The catalyst was
removed by filtration and the cake slurry was washed with 7 L of 50%
JA992486T