Carbapenem biosynthesis: Confirmation of stereochemical assignments and the role of CarC in the ring stereoinversion process from L-Proline

Article abstract of DOI:10.1021/ja034248a

(5R)-Carbapen-2-em-3-carboxylic acid is the simplest structurally among the naturally occurring carbapenem β-lactam antibiotics. It co-occurs with two saturated (3S,5S)- and (3S,5R)-carbapenam carboxylic acids. Confusion persists in the literature about t

Full text of DOI:10.1021/ja034248a

Published on Web 06/18/2003
Carbapenem Biosynthesis: Confirmation of Stereochemical
Assignments and the Role of CarC in the Ring
Stereoinversion Process from L-Proline
Anthony Stapon, Rongfeng Li, and Craig A. Townsend*
Contribution from the Department of Chemistry, The Johns Hopkins UniVersity,
3400 North Charles Street, Baltimore, Maryland 21218
Received January 20, 2003; E-mail: ctownsend@jhu.edu
Abstract: (5R)-Carbapen-2-em-3-carboxylic acid is the simplest structurally among the naturally occurring
carbapenem â-lactam antibiotics. It co-occurs with two saturated (3S,5S)- and (3S,5R)-carbapenam
carboxylic acids. Confusion persists in the literature about the signs of rotation and absolute configurations
of these compounds that is resolved in this paper. (3S,5S)-Carbapenam carboxylic acid was prepared
from L-pyroglutamic acid to unambiguously establish its absolute configuration as identical to the natural
product isolated from Serratia marcescens and from overexpression of the biosynthetic genes carAB in
Escherichia coli. L-Proline labeled with deuterium or tritium at the diastereotopic C-5 methylene loci was
shown to incorporate one label at the bridgehead of (3S,5S)-carbapenam carboxylic acid, but not into the
“inverted” (3S,5R)-carbapenam carboxylic acid or the final carbapenem product. CarC, the third enzyme
of the biosynthetic pathway required to synthesize the carbapenem, was demonstrated in cell-free studies
to be dependent on R-ketoglutarate and ascorbate in keeping with weak sequence identities with other
non-heme iron, R-ketoglutarate-dependent oxygenases. CarC mediated the stereoinversion of synthetic
(3S,5S)-carbapenam carboxylic acid to the (5R)-carbapenem as judged by bioassay. These findings suggest
that L-proline is desaturated to pyrroline-5-carboxylic acid prior to uptake into the biosynthetic pathway.
The loss of the bridgehead hydrogen from the (3S,5S)-carbapenam during the ring inversion process to
form the epimeric (3S,5R)-carbapenam and desaturation to the (5R)-carbapenem are proposed to be coupled
by CarC to the reduction of dioxygen to drive the formation of these higher energy products, an
unprecedented reaction for this enzyme class.
Carbapenem-3-carboxylic acid (4) is the simplest of over 50
naturally occurring carbapenem â-lactam antibiotics.1 Members
of this family and their derivatives are clinically important for
their potent, broad-spectrum antibacterial activity and their
resistance to â-lactamases.2 Early experiments established that
the â-lactam carbons of 4 are derived from acetate3 and the
fused pyrrolidine carboxylic acid arises ultimately from glutamate.3
Assembly of its primary metabolic precursors into the carbap-
enem nucleus takes place through the action of just three
enzymes, CarA, B, and C (Scheme 1).4 This efficient biosyn-
thetic process is accompanied by a remarkable stereochemical
inversion at C-5 mediated by CarC.4 While the absolute
configurations of 2, 3, and 4 were in part misassigned initially,5
stereochemical correlations by Bycroft corrected these to those
shown in Scheme 1.6 A chemical logic can be discerned in this
pathway where a natural L-amino acid is utilized in the formation
of 1 by CarB and cyclized to carbapenam 2 by CarA. CarC
then acts to invert the absolute configuration of the bridgehead
to 5R essential to the ultimate biological activity of the antibiotic
and to subtly raise the strain energy of the bicyclic system by
placing the C-3 carboxyl endo.4
Unfortunately, the amended stereochemical assignments of
Bycroft, which we relied upon in the interpretation of the
biochemical experiments summarized in Scheme 1,4 have been
called into question by Ogasawara et al.,7 and are further clouded
by reversal of the absolute stereochemistry of 2 to (3R,5R) in a
1998 review.8 We describe in this paper a third stereochemical
correlation in which 2 as its p-nitrobenzyl (PNB) ester was
prepared by total synthesis from L-pyroglutamic acid and
compared to the same product isolated from wild-type Serratia
marcescens and a recombinant Escherichia coli bearing carAB
in an overexpression vector. These findings unambiguously
confirm the corrected absolute configurations made for 2, 3,
and 4 by Bycroft as depicted in Scheme 1.4
(1) Parker, W. L.; Rathnum, M. L.; Wells, J. S. J.; Trejo, W. H.; Principe, P.
A.; Sykes, R. B. J. Antibiot. 1982, 35, 653-660.
(2) Bradley, J. S.; Garau, J.; Lode, H.; Rolston, K. V. I.; Wilson, S. E.; Quinn,
J. P. Int. J. Antimicrob. Agents 1999, 11, 93-100.
In addition, we address the issue of the cryptic stereochemical
inversion at C-5 that occurs in the CarC-catalyzed transformation
(3) Bycroft, B. W.; Maslen, C. J. Antibiot. 1988, 41, 1231-1242.
(4) Li, R.; Stapon, A.; Blanchfield, J. T.; Townsend, C. A. J. Am. Chem. Soc.
2000, 122, 9296-9297.
(5) Bycroft, B. W.; Maslen, C.; Box, S. J.; Brown, A. G.; Tyler, J. W. J. Chem.
Soc., Chem. Commun. 1987, 1623-1625.
(7) Tanaka, H.; Sakagami, H.; Ogasawara, K. Tetrahedron Lett. 2002, 43, 93-
96.
(6) Bycroft, B. W.; Chhabra, S. R. J. Chem. Soc., Chem. Commun. 1989, 1989,
423-425.
(8) McGowan, S. J.; Bycroft, B. W.; Salmond, G. P. C. Trends Microbiol.
1998, 6, 203-208.
9
8486
J. AM. CHEM. SOC. 2003, 125, 8486-8493
10.1021/ja034248a CCC: $25.00 © 2003 American Chemical Society

Carbapemem Biosynthesis
A R T I C L E S
Scheme 1
Scheme 2 a
a Reagents and conditions: (a) benzyl bromide, CH2Cl2, DIEA, 24 h; (b) (Boc)2O, DMAP, triethylamine, CH3CN, 18 h; (c) Superhydride, THF, -78 °C,
20 min; (d) tert-butyl diethylphosphonoacetate, potassium hydride, DMF, 1 h; (e) trifluoroacetic acid, 30 min; (f) tris(2,3-dihydro-2-oxobenzoxazol-3-
yl)phosphine oxide (refs 11, 12), triethylamine, CH3CN, 6 h; (g) Pd/C, H2, EtOAc, sodium bicarbonate, 15 psi, 90 min; (h) p-nitrobenzylbromide, DMF,
4 h.
of 2 to 3 and 4. L-Proline bearing first deuterium and then tritium
at C-5 was incorporated into 2, 3, and 4 by S. marcescens. By
both NMR spectroscopic means and double radioisotope
analysis, we have been able to demonstrate that one isotopic
label is retained at the bridgehead position in (3S,5S)-carbap-
enam 2, but is cleanly lost in the epimerized (3S,5R)-carbapenam
3 and its carbapenem partner 4. Apart from revealing the
probable intermediacy of pyrroline-5-carboxylate/glutamyl semi-
aldehyde as the immediate precursor from primary metabolism
in carbapenam/em formation, the outcome of these experiments
limits the possible mechanisms that can be acting in the ring
inversion process.
generated by an overexpression clone of carAB in E. coli. All
three were found to have the same absolute configuration.
Stereochemical Correlation of (3S,5S)-Carbapenam 2.
Synthesis of the PNB ester 10 of 2 employed the Horner-
Emmons-Smith reaction used previously to prepare protected
carboxymethylprolines.9 As shown in Scheme 2, N-Boc benzyl
L-pyroglutamate 510 was reduced to the hemiaminal 6 by
treatment with Superhydride.11 Coupling with tert-butyl dieth-
ylphosphonoacetate gave protected (2S,5S)-carboxymethylpro-
line 7 in a diastereomeric excess of 7:1, in accord with the
literature as determined by 1H NMR spectroscopy.9 The mixture
of diastereomers was treated with TFA to yield 5-carboxymethyl
proline R-benzyl ester 8. Cyclization of the protected â-amino
acids with a particularly efficient peptide coupling reagent12,13
formed the â-lactam ring. Separation from the minor diastere-
omer by silica gel chromatography yielded the benzyl ester of
the (3S,5S)-carbapenam 9 in 25% overall yield for 5 steps.
Hydrogenolysis of the benzyl ester in the presence of 1 equiv
of NaHCO3 yielded the carbapenam sodium salt 2 in nearly
pure form, accompanied by varying amounts of the hydrolysis
product 1 as a contaminant (e10%). Carbapenam 2 is surpris-
ingly stable in neutral solution, but is prone to hydrolysis on
exposure to protic/Bronsted acids, base, or divalent cations.
Losses were also observed upon lyophilization/concentration
after HPLC during attempts to further purify 2 (unpublished
results).
Results and Discussion
Stereochemical Analysis. The potent antibiotic activity of
the carbapenem was taken initially to establish the (5R)-
configuration at the ring junction. Catalytic hydrogenation of 4
gave a 9:1 mixture of carbapenams 3 and (in fact) the enantiomer
of 2, respectively.5 An insufficient amount of the reactive 2 was
obtained as its p-nitrobenzyl (PNB) ester to allow accurate
characterization of its absolute configuration. This ambiguity
was remedied in a subsequent stereochemical correlation in
which the antipode of 2 was prepared from D-glutamic acid and
found to be identical spectroscopically, except it displayed a
CD spectrum equal and opposite to that of the natural product,
thus establishing the true absolute stereostructure of 2 as (3S,5S)
shown in Scheme 1.6
Unfortunately, while the conclusions of the second Bycroft
paper were correct, as we confirm below, clerical errors in the
reported signs of optical rotations have led to a second
stereochemical correlation described recently by Ogasawara et
al.7 Here 3-hydroxytetrahydropyridine of known absolute ster-
eochemistry was converted to the enantiomer of 2 and, based
on its optical rotation, the absolute configuration revised to
Bycroft’s original assignment.5 We have independently reex-
amined this question and have prepared 2 by unambiguous total
synthesis from L-pyroglutamic acid and compared the optical
rotation of its PNB ester to that of both the natural product
isolated from S. marcescens and the corresponding metabolite
To complete the stereochemical correlations, 2 was treated
with p-nitrobenzyl bromide in DMF to yield the PNB ester 10,
which was used for comparison to the carbapenams isolated
from S. marcescens as well as the single carbapenam produced
by the E. coli carAB transformant described previously.4
Samples of the PNB esters from the three sources showed
1
identical H NMR spectra and HPLC/TLC chromatographic
(9) Collado, I.; Ezquerra, J.; Vaquero, J. J.; Pedregal, C. Tetrahedron Lett.
1994, 35, 8037-8040.
(10) Gosselin, F. L.; Lubell, W. D. J. Org. Chem. 2000, 65, 2163-2171.
(11) Ezquerra, J.; Pedregal, C. Tetrahedron Lett. 1994, 35, 2053-2056.
(12) Nagamatsu, T.; Kunieda, T. Chem. Pharm. Bull 1988, 36, 1249-1251.
(13) Gilchrist, T. L.; Lemos, A.; Ottaway, C. J. J. Chem. Soc., Perkin Trans. 1
1997, 3005-3012.
9
J. AM. CHEM. SOC. VOL. 125, NO. 28, 2003 8487

A R T I C L E S
Stapon et al.
Scheme 3 a
a (a) (CH2O)n, TsOH, toluene, 5 h; (b) 0.1 N NaOMe/MeOH, 18 h; (c) isobutyl chloroformate, triethylamine, N-hydroxy-2-thiopyridone, 1 h; (d)
bromotrichloromethane, 3 h; (e) sodium [13C]cyanide, DMSO, 3 h; (f) 2:1 acetic acid/HCl, 6 h; (g) ref 23; (h) methyl iodide, CH3CN, 18 h; (i) sodium
borodeuteride, MeOH, 3 h; (j) trifluoroacetic acid; (k) Pd(OH)2, H2, EtOH, 2 d.
properties. Most important, optical rotation data for the PNB
esters isolated from the three sources were nearly identical (total
synthesis [R]25D -168.2°, S. marcescens [R]25D -173.0°, E. coli
pyrroline-5-carboxylate/glutamyl semialdehyde stage, one deu-
terium would be retained at C-5 detectable as a 1:1:1 triplet by
13C NMR, and potentially detectable in all three â-lactam
compounds.
carAB [R]25 -169.6°).
D
[5,5-2H2,5-13C]-L-Proline. As outlined in Scheme 3, N-Cbz
L-glutamate was converted to its R-methyl ester according to
the procedure of Hanessian.19,20 Using the Barton decarboxyl-
ative halogenation method,21 the remaining carboxyl was
converted to the terminal bromide 11 by photolysis of the
intermediate N-hydroxy-2-thiopyridone ester in the presence of
bromotrichloromethane.22 The bromide was then displaced by
NaCN in DMSO to provide 12. Hydrolysis of the nitrile in 2:1
acetic acid/concentrated HCl23 and recrystallization from pyri-
dine/ethanol provided L-glutamic acid 13 in pure form. Fol-
lowing established procedures,24 L-glutamate was then N-ben-
zylated, cyclized to its N-benzyl pyroglutamate, protected as
tert-butyl ester, and converted to the thiolactam 14. Reaction
of 14 with 10 equiv of methyl iodide in acetonitrile, followed
by reduction with sodium borohydride,25 yielded tert-butyl
N-benzyl-L-proline 15. The fully protected amino acid was then
treated with TFA followed by catalytic hydrogenation to yield
L-proline 16.
On the basis of these findings, the absolute configuration of
2 is indeed (3S,5S) as determined earlier by Bycroft.6 The data
collected indicate, however, that Bycroft incorrectly reported
the sign of the optical rotation for the methyl ester antipode of
2 prepared in his study. Nevertheless, his assignment of the
(3S,5S) stereochemistry for 2 isolated from the natural source
was, indeed, correct.
Carbapenam(em) Stereoinversion. With the absolute con-
figurations of 2, 3, and 4 firmly in hand, attention was turned
to the striking ring inversion process carried out by CarC
(Scheme 1).4 Primary sequence alignment of CarC with other
known proteins revealed a modest 23%14 identity to clavaminate
synthase (CS2)15 and weak identities to other R-ketoglutarate-
dependent dioxygenases. These Fe(II)-requiring enzymes are
commonly associated with hydroxylation reactions at unacti-
vated carbon centers and are also known to mediate desaturation
reactions and oxidative cyclization processes.16-18 Therefore,
while introduction of the double bond in 4 is unsurprising, the
ring inversion involving no change in oxidation state in the
formation of 3 has no precedent that we are aware of among
the reactions catalyzed by this class of enzymes.
As a first step toward understanding this ring stereoinversion
process, we undertook an examination of the fate of the
bridgehead hydrogen at C-5 in 2 into 3 and 4. Previous
incorporation experiments had demonstrated that the pyrrolidine
ring is derived from L-glutamate.3 We reasoned, on the other
hand, that the more direct precursor from primary metabolism
could be L-proline on the basis of a downstream gene in the
biosynthetic cluster that appears to encode a proline dehydro-
genase.14 Whether or not hydrogen isotope remained at the C-5
position in the bicyclic products can be visualized to provide
information about the assembly of 2, 3, and 4 from their true
primary metabolic precursors. If [5,5-2H2,5-13C]-L-proline were
oxidized to an L-glutamate equivalent prior to the assembly of
2, 3, or 4, isotopic label would be lost and there would be no
spin-spin coupling resulting from a directly bonded deuterium
atom detectable by 13C NMR spectroscopy. Conversely, if
oxidation of labeled L-proline did not proceed beyond the
This sequence was repeated using sodium [13C]cyanide to
displace bromide 11 and sodium borodeuteride to reduce the
[5-13C]thiolactam 14. Deprotection of the triply labeled 15
yielded [5,5-2H2,5-13C]-L-proline in 12% overall yield from
[5-13C]-L-glutamic acid having 85-90% deuterium/site as
1
determined by H NMR and mass spectrometric analysis.
The multiply labeled L-proline was administered to a S.
marcescens fermentation, and the â-lactams 2, 3, and 4 produced
were extracted, derivatized as their PNB esters, and purified3
for 13C NMR analysis. The 13C spectra showed that one
deuterium was retained at the C-5 position of 2, JC-D ) 23.3
Hz, whereas no deuterium was detectable at the corresponding
locus in 3 or 4. Incorporation levels of [5,5-2H2,5-13C]-L-proline
into the â-lactam products were low but discernible, presumably
owing to kinetic isotope effects occurring during reactions at
this center, and the inherently inefficient spin relaxation of the
[13C-2H] labeled species suppressing signal intensities. A more
(19) Hanessian, S.; Sahoo, S. P. Tetrahedron Lett. 1984, 25, 1425-1428.
(20) Singh, G.; Mou, L. Tetrahedron Lett. 2001, 42, 6603-6606.
(21) Barton, D. H. R.; Crich, D.; Motherwell, W. B. Tetrahedron Lett. 1983,
24, 4979-4982.
(14) McGowan, S. J.; Sebaihia, M.; Porter, L. E.; Williams, P.; Stewart, G. S.
A. B.; Bycroft, B. W.; Salmond, G. P. C. Mol. Microbiol. 1996, 22, 415-
426.
(22) Barton, D. H. R.; Herve, Y.; Potier, P.; Thierry, J. Tetrahedron 1988, 44,
5479-5486.
(23) Havranek, M.; Kopecka-Schadtova, H.; Veres, K. J. Radiolabeled Com-
pounds 1970, 6, 345-354.
(15) Salowe, S. P.; Marsh, N. E.; Townsend, C. A. Biochemistry 1990, 29, 6499-
6508.
(24) Petersen, J. S. F., G.; Rapoport, H. J. Am. Chem. Soc. 1984, 106, 4539-
(16) Prescott, A. G.; Lloyd, M. D. Nat. Prod. Rep. 2000, 17, 367-383.
(17) Schofield, C.; Zhang, Z. Curr. Opin. Struct. Biol. 1999, 9, 722-731.
(18) Townsend, C. A. Curr. Opin. Struct. Biol. 2002, 6, 583-589.
4547.
(25) Sundberg, R. J.; Powers Walters, C.; Bloom, J. D. J. Org. Chem. 1981,
46, 3730-3732.
9
8488 J. AM. CHEM. SOC. VOL. 125, NO. 28, 2003

Carbapemem Biosynthesis
A R T I C L E S
Table 1. Double Isotope Analysis of [5-3H,U-14C]-L-Proline
Incorporation into 2, 3, and 4
3H/14C
3H retention (%)
[5-3H,U-14C]-L-proline
[5-3H,14C]-2
10.38
7.78
0.14
0.32
75
1
3
[5-3H,14C]-3
[5-3H,14C]-4
sensitive means to detect the retention or loss of bridgehead
label was sought to ensure the validity of this result.
Commercially available [U-14C]-L-proline and [5-3H]-L-
proline were combined to achieve a 3H/14C ratio of 10.38. The
doubly labeled amino acid was added to a growing fermentation
of S. marcescens and the â-lactams produced were isolated,
derivatized, and purified by HPLC. The results of the radio-
isotope analysis are presented in Table 1. Carbapenam 2 retained
a high proportion of tritium label relative to the 14C internal
standard, 75%. In contrast, inversion of the ring junction that
occurs in the formation of 3 and 4 was accompanied by 99%
and 97% loss of tritium, respectively. These more quantitative
findings are in complete accord with the observations made
above by 13C NMR spectroscopy. The biosynthetic path to the
pyrrolidine ring in 2, 3, and 4, therefore, proceeds by way of
the intermediate oxidation state in pyrroline-5-carboxylate/
glutamyl semialdehyde 17, presumably by the oxidation of
L-proline.
Preliminary Cell-Free Conversion of 2 to 3 and 4 by CarC.
The separate expressions of carAB and carABC in E. coli gave4
the unexpected result that (3S,5S)-carbapenam 2 precedes the
formation of the “inverted” carbapenam 3 and its unsaturated
partner 4 and implies that this stereoinversion is performed by
CarC. To gather further support for this remarkable observation,
the availability of enantiomerically pure, synthetic 2 made
possible a direct test of its conversion to 3 and 4 by a cell-free
extract containing recombinant CarC. Overexpression of carC
was carried out in pET24a and BL21(DE3) as host under
standard conditions. The cells were collected by centrifugation,
suspended in phosphate buffer and glycerol, lysed in the
presence of protease inhibitors, and treated with DNase and
RNase, and the cell debris was removed by centrifugation. The
clarified supernatant was taken as the cell-free extract (CFE).
The ability of the CFE to catalyze carbapenem synthesis was
monitored in a paper disk assay. A â-lactam supersensitive E.
coli (SC12155) was used as the test organism. As shown in
Figure 1, incubation of carbapenam 2 with the CFE and added
R-ketoglutarate and ascorbate gave a clear zone of inhibition.
Controls with just the carbapenam itself, or with the complete
incubation mixture but lacking 2, failed to give zones of
inhibition. Additional controls demonstrated antibiotic produc-
tion was dependent on added R-ketoglutarate, reduced in the
absence of added ascorbate, but was minimally affected by
addition of ferrous ammonium sulfate (data not shown). These
observations support the view from primary sequence similarities
that CarC is an R-ketoglutarate-dependent, non-heme iron
oxygenase and confirms the deduction made earlier that it
mediates the oxidative conversion of 2 to carbapenem 4.4
Figure 1. Presumed carbapenem 4 production in a cell-free extract (CFE)
of pET24a/carC BL21(DE3) E. coli visualized on a plate of â-lactam
supersensitive E. coli strain SC12155. (1) 2 mM carbapenam 2, sodium
salt (2) CarC CFE, 8 mM R-ketoglutarate, 1 mM ascorbate; (3) CarC CFE,
2 mM carbapenam 2, sodium salt, 8 mM R-ketoglutarate, 1 mM ascorbate.
To distinguish the biochemical origins of the pyrrolidine ring
of these natural products, incorporation experiments with
[5,5-2H2,5-13C]- and of [5-3H,U-14C]-L-proline were conducted
in growing cultures of S. marcescens. Analysis of the PNB esters
of 2, 3, and 4 first by 13C NMR spectroscopy and second by
two-channel scintillation counting demonstrated retention of
deuterium or tritium label at the bridgehead carbon in 2, but its
complete loss on ring inversion to 3 and 4. The paired
radioisotope analysis revealed a 75% retention of tritium label
in 2 from L-proline. Unfortunately, the distribution of radio-
isotope between the two diastereotopic C-5 L-proline positions
is not known. The high extent of tritium retention may be due
to unequal distribution of label at the C-5 loci of L-proline and/
or a substantial kinetic isotope effect in the stereochemical
inversion process mediated by CarC. Primary sequence align-
ment of CarC shows weak similarity to R-ketoglutarate-
dependent, non-heme iron oxygenases,14 suggesting that this
unusual process may be fundamentally oxidative. In the event,
the double-labeled proline experiments support the view that
the primary metabolic precursor of (2S,5S)-carboxymethylpro-
line (1), the first committed intermediate of carbapenem bio-
synthesis, is pyrroline-5-carboxylate/glutamyl semialdehyde (17,
Scheme 4) derived from L-proline.
Preliminary cell-free experiments with CarC overproduced
in E. coli demonstrate apparent carbapenem 4 production by
bioassay against a â-lactam sensitive test organism. As antici-
pated from sequence alignments, added R-ketoglutarate and
ascorbate enlarged the zones of growth inhibition in the
bioassay. Experiments to purify CarC are underway to examine
the unprecedented ring inversion process and attendant desatu-
ration. Curiously, in the unrelated biosynthesis of the potent
â-lactamase inhibitor clavulanic acid 20 an unknown, but also
likely oxidative process,26 inverts clavaminic acid 18 to the
aldehyde 19, which is reduced to 20.27 In this overall conversion,
Conclusions
(26) Townsend, C. A.; Krol, W. J. J. Chem. Soc., Chem. Commun. 1988, 1234-
1236.
(27) Nicholson, N. H.; Baggaley, K. H.; Cassels, R.; Davison, M.; Elson, S.
W.; Fulston, M.; Tyler, J. W.; Woroniecki, S. R. J. Chem. Soc., Chem.
Commun. 1994, 1281-1282.
In this work we have established with certainty the absolute
configuration of carbapenam 2 by total synthesis and confirmed
the amended stereochemical assignments of 3 and 4 by Bycroft.6
9
J. AM. CHEM. SOC. VOL. 125, NO. 28, 2003 8489

A R T I C L E S
Stapon et al.
Scheme 4
Media used for the S. marcescens feeding experiments and the
isolation, derivatization, and purification of PNB-protected â-lactams
were performed as previously described.4 [5,5-2H2,5-13C]-L-Proline was
administered to a 3 L S. marcescens fermentation at 1.3 mM total
concentration. Radiolabeled [U-14C]-L-proline and [5-3H]-L-proline were
purchased from Amersham Biosciences (Piscataway, NJ), mixed with
cold L-proline (114 mg, 1 mmol), passed through a cation exchange
column (Bond Elut, SCX, Varian, Walnut Creek, CA), and lyophilized.
A portion was purified by HPLC and the 3H/14C ratio (10.38) determined
prior to administration. Carbapenem production is analyzed by the disk
diffusion method with â-lactam supersensitive E. coli SC12155
(obtained from Bristol Myers Squibb Inc., Princeton, NJ).
â-Lactam Production and Isotopic Incorporation Experiments:
S. marcescens. S. marcescens (ATCC 39006) fermentation was
modified from the method of Bycroft et al.3 Seed medium (100 mL,
Bacto-Soytone 2%, sucrose 0.2%) was inoculated with a single colony
maintained on Nutrient Agar (Becton Dickinson, Sparks, MD). After
growing at 26 °C for 20 h with shaking at 300 rpm, 75 mL was used
to inoculate 3 L of fermentation medium3 in a MBF-500 bioreactor
(Wheaton Instruments, Willville, NJ). For the radiolabeled L-proline
incorporation experiment, the scale of the fermentation was 1 L and
all other conditions remained the same. The fermentation was carried
out for 26 h at 26 °C with 300 rpm stirring and 4 NI/min aeration. For
both incorporation experiments, the L-prolines were added at the 4th
hour of the fermentation. The pH was monitored and controlled to 7.5.
Cells were separated by centrifugation at 4000g for 10 min at 4 °C.
Genetics. CarA, carB, and carC were obtained by standard polym-
erase chain reactions (PCR) from genomic DNA isolated from Erwinia
carotoVora ssp. carotoVora (obtained from ATCC 39048).14 The PCR
products were cloned into pT7Blue-3 (Novagen, Madison, WI) and
then into pET24a (Novagen) to generate the overexpression vectors
pET24a/carA, pET24a/carB, and pET24a/carC. CarB was recovered
from pET24a/carB along with the ribosome binding sequence by XbaI-
HindIII digestion and treatment with T4 DNA polymerase, and this
fragment was inserted into the NotI site of pET24a/carA with the same
orientation as carA to give the coexpression vector pET24a/carAB.
Fermentation of E. coli Transformants. E. coli BL21 (DE3)pLysS
(pET24a/carAB) was grown at 37 °C in 50 mL of LB medium on a
rotary shaker for 15 h at 300 rpm. The resulting seed culture was used
as inoculum (0.2%) for fermentation in the modified S. marcescens
fermentation medium (L-glutamate 0.5%, NH4Cl 0.075%, K2HPO4
0.2%, MgSO4 0.05%, NaCl 0.05%, FeSO4‚7H2O 0.0025%, CaCO3
0.025% and glucose 1%; pH 7.5). Cells were grown in the bioreactor
with 3 L of fermentation medium at 37 °C. The simultaneous
overexpression of carA and carB was induced at OD600 ) 0.6 with
1 mM IPTG. The protein overexpression and carbapenam production
were carried out at 28 °C for 6 h following the induction.
Isolation of â-lactams from S. marcescens and E. coli fermenta-
tions. Extraction and derivatization of the â-lactams produced was
accomplished by the methods established by Bycroft.3
For the [5,5-2H2,5-13C]-L-proline feeding experiment, 13C NMR
analysis was carried out utilizing the crude residue isolated after silica
gel chromatography. The resonances for C-5 in the PNB esters of 2, 3,
and 4 were at δ 53.1, 53.2, and 51.3, respectively.
In the case of the radiolabeled L-proline incorporation experiment,
the similarly isolated residue was solubilized in 5 mL of 50/50
acetonitrile/water and further purified by reverse phase HPLC utilizing
a Phenomenex Prodigy 10µ C(8)-Semi-Prep column at a flow rate of
5 mL/min (60/40 acetonitrile/water eluant) with 500 µL injections. The
eluant was monitored at 320 nm and the PNB esters 2, 3, and 4 elute
individually between 12 and 16 min. The pure â-lactam containing
fractions were collected, frozen, and lyophilized for scintillation
counting.
however, the bridgehead hydrogen is retained.28 While neither
of these oxidative “enantiomerizations” is understood, both
reflect a biosynthetic strategy to create these powerfully
bioactive substances initially in their inactive, “enantiomeric”
formsswhich are carried through many steps in the case of
clavulanic acidsonly to reveal each in its true colors through a
penultimate oxidative ring inversion step. The means by which
this stereochemical change is accomplished without apparent
alteration of oxidation state in carbapenam 3 is an intriguing
problem. The formation of 3 from 2, however, is in itself an
uphill event thermodynamically (C-3 carboxylate becomes endo)
and one may speculate that loss of H-5 from 2 and its
replacement in 3 from another source is somehow coupled with
reduction of dioxygen to drive the overall reaction. How this
process is achieved is the object of current investigation.
Note Added in Proof. In a recent communication Bycroft
corrects his inadvertently misreported optical rotation of (3R,5R)-
carbapenam methyl ester.29 Unfortunately, this error has been
propagated again in a total synthesis of all four stereoisomers
of this compound.30
Experimental Section
Tetrahydrofuran was distilled from sodium/benzophenone ketyl;
acetonitrile was distilled from CaH2. Reagents were obtained from
Aldrich and used as received unless otherwise noted. Sodium [13C]-
cyanide (99%) and sodium borodeuteride (99%) were purchased from
Cambridge Isotope Laboratories, Inc. (Andover, MA). Melting points
are uncorrected. Optical rotations were recorded on a Jasco Model
P-1010 polarimeter. Scintillation analysis was performed on a Beckman
Coulter Model LS 5801 liquid scintillation counter. Flash chromatog-
raphy was performed with Bodman Industries silica gel 60 (230-400
mesh ATM). NMR spectral data were obtained on a Varian UnityPlus
400 spectrometer, unless otherwise noted. NMR spectra recorded in
CDCl3 are reported in parts per million (δ) downfield from Me4Si (1H)
or relative to CDCl3 at 77.0 ppm (13C). NMR spectra recorded in D2O
are reported in parts per million (δ) relative from D2O 4.80 ppm (1H)
or relative to dioxane at 66.0 ppm (13C).
(28) Basak, A.; Salowe, S. P.; Townsend, C. A. J. Am. Chem. Soc. 1990, 112,
1654-1656.
(29) Bycroft, B. W.; Chhabra, S. R.; Kellam, B.; Smith, P. Tetrahedron Lett.
2003, 22, 973-976.
Purification of the â-lactams for PNB derivatized stereochemical
correlations was accomplished utilizing normal phase HPLC. The
fractions containing â-lactam(s) from the silica gel chromatography
(30) Avenoza, A.; Barriobero, J. I.; Busto, J. H.; Peregrina, J. M. J. Org. Chem.
2003, 68, 2889-2894.
9
8490 J. AM. CHEM. SOC. VOL. 125, NO. 28, 2003

Carbapemem Biosynthesis
A R T I C L E S
were evaporated and resolubilized in 5 mL of 30% ethyl acetate/hexane,
and 500 µL portions were injected onto a Phenomenex Luna 5µ CN
Semi-Prep column with a flow rate of 5 mL/min (30/70 ethyl acetate/
hexanes eluant). The eluant was monitored at 320 nm with elution of
the â-lactams occurring between 8.5 and 12 min.
Transformation and carC Overexpression. pET24a/carC was
transformed into E. coli BL21 (DE3) by electroporation using a Gibco/
BRL Cell-Porator Electroporation System (Rockville, MD) according
to the manufacturer’s instructions. A 50 mL 2xYT seed culture was
inoculated with a single colony picked from freshly transformed plates
grown at 37 °C for 18 h. Cells were collected by centrifugation and
resuspended in 15 mL of 2xYT medium. Then 5 mL of the cell
suspension was used to inoculate 3 L of 2xYT medium. Upon
OD600 ) 0.6, the temperature was reduced to 20 °C and 1 mM IPTG
and 0.4 mM ferrous ammonium sulfate were added. The fermentation
was continued for 5 h and the cells were harvested by centrifugation
at 4000g for 10 min at 4 °C.
127.9, 127.8, 82.1, 81.9, 80.7, 77.3, 66.7, 66.6, 59.3, 59.0, 33.2, 32.1,
28.1, 27.9, 27.7, 26.9. IR: 3471, 3030, 2981, 1747, 1695, 1382, 1161
cm-1. HRMS calcd for C17H23NO5 [M + NH4+]: 339.1920, found
339.1931.
N-(Boc)-(2S,5S)-((tert-butyl)carboxymethyl)proline benzyl ester
(7) was synthesized on the basis of an established protocol:9 tert-butyl
diethylphosphonoacetate (4.8 mL, 17.15 mmol) was added dropwise
to a stirring suspension of 30% KH/mineral oil (2.29 g, 17.15 mmol)
in 100 mL of dry DMF. The reaction was stirred at room temperature
for 1 h under argon. At this time, carbinolamide 6 (5.5 g, 17.15 mmol)
in 100 mL of dry DMF was added to the reaction and allowed to stir
overnight. The reaction was quenched by the addition of saturated NH4-
Cl (1 L) and extracted with diethyl ether (2 × 200 mL). The organic
layers were collected, dried over MgSO4, filtered, concentrated, and
purified by silica gel chromatography with 5% ethyl acetate/hexanes
as the eluant to yield the protected carboxymethyl proline as a colorless
oil: 6.10 g, 84% yield (7:1 mixture of diastereomers); [R]25 -46.4°
D
1
(c 1, CHCl3); TLC (EtOAc/hexanes, 20/80), Rf 0.55. H NMR (major
Cell-Free Extract of CarC Overexpression. All cell lysis steps
were carried out at 0-4 °C. Assays for CarC activity were carried out
at 25 °C.
diastereomer) δ 7.28 (m, 5H), 5.09 (m, 2H), 4.31 (m, 1H), 4.23 (m,
1H), 2.85 (dd, J ) 15.4 Hz, 3.6 Hz, 0.68H), 2.66 (dd, J ) 14.6 Hz, 2.8
Hz, 0.36H), 2.14 (m, 3H), 1.86 (m, 1H), 1.73 (m, 1H), 1.40 (s, 5H),
1.37 (s, 7H), 1.26 (s, 6H). 13C NMR: δ172.6, 172.3, 170.5, 153.8,
153.3, 135.4, 128.5, 128.3, 128.3, 128.2, 128.1, 128.0, 127.9, 80.4,
80.3, 80.1, 79.9, 66.55, 59.6, 59.3, 54.9, 54.8, 54.7, 40.3, 39.1, 29.5,
28.6, 28.5, 28.1, 28.0, 27.9, 27.1; IR: 2981, 1720, 1694, 1476, 1894,
1368, 1164 cm-1. HRMS calcd for C23H33NO6 [M + H+]: 420.2386,
found 420.2392.
E. coli BL21(DE3) cell paste (4.6 g) harboring overexpressed CarC
was suspended in 10 mL of 20 mM sodium phosphate buffer containing
10% glycerol, 2 mM DTT, 1 mM benzamidine, 10 µM EDTA, and 1
mM PMSF, pH 8.0. Lysozyme (10 mg) was then added, and the solution
was mixed gently for 5 min. Brij-58 (10 mg solubilized in 1 mL of
lysis buffer) was then added and mixing continued for an additional
10 min. DNase/RNase (∼1 mg each) was added and the solution gently
mixed for 1 h. The resulting suspension was then clarified by
centrifugation at 35 000g for 30 min at 4 °C.
Assays for CarC activity in the CFE contained the following
components: CarC CFE solution, 1 mM ascorbate, 8 mM R-ketoglu-
tarate, 2 mM (3S,5S)-carbapenam 2, in a total volume of 1.00 mL
incubated at room temperature for 2.5 h. Carbapenem production was
detected by visualization of antibiosis on a plate of â-lactam supersensi-
tive E. coli strain SC12155, incubated at 28 °C for 24 h. The control
without CarC CFE solution contained lysis buffer in its place, and the
control for the CFE solution and reaction components contained lysis
buffer in place of added 2.
Carbapenam (2). (2S)-Benzyl N-(Boc)pyroglutamate (5) was
prepared according to a published procedure:10 9.42 g, 76%; mp 72-
73 °C; [R]25D -36.4° (c 1, CHCl3); (lit.10 mp 69-70 °C; [R]25D -37.8°
(c 1, CHCl3)); TLC (EtOAc/hexanes, 30/70) Rf 0.25. 1H NMR: δ 7.31
(m, 5), 5.16 (abq, J ) 12.0 Hz, 2H), 4.61 (dd, J ) 9.4 Hz, 3.2 Hz,
1H), 2.55 (m, 1H), 2.44 (m, 1H), 2.29 (m, 1H), 1.97 (m, 1H), 1.37 (s,
9H). 13C NMR: δ 173.1, 171.0, 149.0, 134.9, 128.5, 128.5, 128.4, 83.4,
67.2, 58.8, 30.9, 27.6, 21.3. IR: 2983, 2341, 2256, 1790, 1749, 1716
cm-1. HRMS calcd for C17H21NO5 [M + Na+]: 342.1312, found
342.1323.
Benzyl-(3S,5S)-carboxymethyl Prolinate (8). The fully protected
carboxymethylproline 7 (5.0 g, 11.9 mmol) was solubilized in 50 mL
of TFA and stirred for 30 min at room temperature. The reaction was
then concentrated in vacuo, diluted with distilled H2O, and lyophilized.
The residual material was taken up in 0.2 N AcOH, passed through a
cation exchange column (Bond Elut, SCX, Varian), and eluted with
distilled H2O. The H2O fractions were lyophilized to dryness to yield
the aminodiacid R-benzyl ester as a white crystalline powder: 3.02 g,
1
96% yield; mp 67-69 °C (dec); [R]25 -24.3° (c 1, H2O). H NMR
D
(major diastereomer): δ 7.44 (m, 5H), 5.28 (abq, J ) 12.0 Hz, 2H),
4.54 (t, J ) 8.4 Hz, 1H), 3.95 (m, 1H), 2.64 (dd, J ) 10.4 Hz, 2.0 Hz,
2H), 2.48 (m, 1H), 2.23 (m, 1H), 2.12 (m, 1H), 1.76 (m, 1H). 13C
NMR: 176.7, 169.2, 169.2, 134.1, 128.4, 127.9, 68.1, 58.6, 58.3, 57.7,
36.7, 36.4, 28.7, 27.8, 27.2, 26.7; IR: 2982, 1744, 1672, 1412, 1201
cm-1. HRMS calcd for C14H17NO4 [M + H+]: 264.1236, found
264.1240.
(3S,5S)-Carbapenam Carboxylic Acid Benzyl Ester (9). The
amino diacid R-benzyl ester 8 (1.78 g, 6.77 mmol), tris(2,3-dihydro-
2-oxobenzoxazol-3-yl)phosphine oxide12,13 (2.59 g, 6.77 mmol), and
TEA (1.85 mL, 15.57 mmol) in 400 mL of dry CH3CN were heated to
reflux and for 6 h. The reaction mixture was concentrated in vacuo,
absorbed onto Celite, and purified by repeated silica gel chromatography
utilizing 15% ethyl acetate/hexanes as the eluant to give a colorless
5-Hydroxy-(2S)-benzyl N-(Boc)prolinate (16) was synthesized
according to an established protocol:11 23 mL of a 1.0 M solution of
Superhydride (23 mmol) in THF was added dropwise to a solution of
pyroglutamate 5 (6.11 g, 19.16 mmol) in 200 mL of dry THF at -78
°C under an argon atmosphere and stirred for 30 min. The reaction
was quenched by the addition of saturated NaHCO3 (34 mL) and 30%
H2O2 (4 mL), the temperature was raised to 0 °C, and the solution was
stirred for 20 min. The solvent was removed in vacuo, and the aqueous
layer was extracted with CH2Cl2 (2 × 100 mL). The organic layers
were combined, dried over MgSO4, filtered, and concentrated. The
residue was purified by silica gel chromatography utilizing 40/60
EtOAc/hexanes as the eluant to afford a colorless oil: 4.54 g, 95%
oil free of any residual diastereomer: 810 mg, 38% yield; [R]25
D
-190.8° (c 1, CHCl3); TLC (EtOAc/hexanes, 40/60) Rf 0.37. 1H
NMR: δ 7.25 (m, 5H), 5.06 (s, 2H), 4.36 (t, J ) 7.6 Hz, 1H), 3.75 (m,
1H), 3.15 (dd, J ) 15.4 Hz, 5.2 Hz, 1H), 2.53 (dd, J ) 15.8 Hz, 1.6
Hz, 1H), 2.58 (m, 1H), 2.26 (m, 2H), 1.53 (m, 1H). 13C NMR: δ 175.9,
170.9, 135.1, 128.3, 128.0, 127.8, 66.6, 58.8, 52.8, 42.3, 35.2, 30.6;
IR: 2798, 2252, 1761, 1189 cm-1. HRMS calcd for C14H15NO3 [M+]:
245.1052, found 245.1054.
(3S,5S)-Carbapenam Carboxylic Acid (2). The benzyl-protected
carbapenam 9 (100 mg, 0.41 mmol) was dissolved in 10 mL of ethyl
acetate and degassed with argon for 5 min in a Parr hydrogenation
apparatus. Pd/C (20 mg) was added, followed by NaHCO3 (34.3 mg,
0.41 mmol), and hydrogenation was carried out for 90 min at 15 psi.
The suspension was then filtered through Celite, and the Celite was
washed with 50 mL of CH2Cl2, followed by 100 mL of distilled H2O.
The aqueous layer was separated and extracted with another 50 mL of
yield; [R]25 -31.8° (c 1, CHCl3); TLC (EtOAc/hexanes, 40/60) Rf
D
1
0.30. H NMR (mixture of rotamers and diastereomers): δ 7.26 (m,
5H), 5.53 (m, 0.8H), 5.44 (m, 0.3H), 5.10 (m, 2H), 4.33 (t, J ) 7.6
Hz, 0.4H), 4.20 (t, J ) 8.0 Hz, 0.8H), 3.76 (d, J ) 2.8 Hz, 0.8H), 3.28
(d, J ) 3.9 Hz, 0.2H), 2.15 (m, 2H), 1.89 (m, 2H), 1.43 (s, 3H), 1.27
(s, 6H). 13C NMR: δ 172.5, 153.6, 135.4, 128.3, 128.3, 128.1, 128.1,
9
J. AM. CHEM. SOC. VOL. 125, NO. 28, 2003 8491

A R T I C L E S
Stapon et al.
CH2Cl2. The aqueous layer was then collected and lyophilized to leave
carbapenam carboxylic acid, sodium salt with the hydrolysis product,
(2S,5S)-carboxymethylproline (1), as a contaminant (∼10%): 61.4 mg,
85% yield; mp 125 °C (dec); [R]25D -148.0° (c 1, H2O). 1H NMR: δ
4.17 (t, J ) 8.4 Hz, 1H), 3.83 (m, 1H), 3.22 (dd, J ) 16.0 Hz, 4.4 Hz,
1H), 2.66 (dd, J ) 16.4 Hz, 1.6 Hz, 1H), 2.59 (m, 1H), 2.18 (m, 1H),
2.05 (m, 1H), 1.48 (m, 1H). 13C NMR: 180.1, 179.2, 61.2, 52.9, 40.1,
35.9, 30.1; IR: 3430, 2927, 1742, 1615, 1400 cm-1. ESI-MS calcd for
C7H8NO3 [M - H+]: 154.1, found 154.2, MS/MS 112 (-ketene).
(3S,5S)-Carbapenam Carboxylic Acid p-Nitrobenzyl Ester (10).
Carbapenam, sodium salt 2 (100 mg, 0.56 mmol) was added to 10 mL
of dry DMF, followed by the addition of p-nitrobenzyl bromide (159
mg, 0.73 mmol). The reaction was stirred at room temperature for 4 h,
poured into 100 mL of distilled H2O, and extracted with EtOAc (2 ×
50 mL). The organic extracts were collected, dried over MgSO4, filtered,
and concentrated in vacuo. The residue was purified by silica gel
chromatography using 15/85 EtOAc/hexanes as eluant to yield a yellow
oil which had identical TLC, HPLC, and NMR properties as those
assigned to carbapenam isolated and derivatized from S. marcescens
and the carAB E. coli transformant: 130 mg, 80% yield; [R]25D -168.2°
L-Glutamic acid (13) was prepared by a modification of a previously
published procedure.23 4-Cyano-(2R)-Cbz-aminomethylbutanoate (12.5
g, 45.1 mmol) was added to 150 mL of acetic acid and 75 mL of
concentrated HCl and heated to reflux for 6 h. This solution was
concentrated in vacuo to afford a crude salt. Water was added to the
residue and the solution was again evaporated; this was repeated two
more times to remove any remaining volatile acids. The residue was
then dissolved in 6 mL of 2% HCl and the suspension was adjusted to
pH ) 5 with pyridine, diluted with 160 mL of ethanol, and cooled to
-20 °C overnight. The precipitate was filtered and the mother liquor
brought back to -20 °C and filtered again: 4.22 g, 63% yield; mp
198 °C (dec); [R]25 +30.6° (c 1, 6 N HCl); (lit.32 mp 200 °C (dec);
D
[R]D22.4 +31.4° (6 N HCl)). 1H NMR: δ 3.76 (t, J ) 6.4 Hz, 1H), 2.50
(m, 2H), 2.11 (m, 2H). 13C NMR: δ 182.9, 181.6, 57.0, 34.3, 29.8.
HRMS calcd for C5H10NO4 [M + H+]: 148.0604, found 148.0600.
(5-13C) [R]25D +30.8° (c 1, 6 N HCl). 1H NMR: δ 3.81 (t, J ) 6.4 Hz),
2.55 (m, 2H), 2.14 (m, 2H). 13C NMR: δ (C5) 182.8, 55.9, 34.2 (d,
J ) 51.9 Hz), 31.7. [M + H+]: 149.
(S)-N-Benzylglutamic acid was prepared according to a published
procedure:24 65% yield; mp 160-162 °C; [R]25D +18.3° (c 1, 6 N HCl);
1
1
(lit.24 mp 162-163 °C; [R]25 +18.5° (c 1, 6 N HCl)). H NMR: δ
(c 1, CHCl3); TLC (EtOAc/hexanes, 40/60) Rf 0.29. H NMR: 8.23
D
(d, J ) 8.8 Hz, 2H), 7.52 (d, J ) 8.8 Hz, 2H), 5.25 (s, 2H), 4.48 (t,
J ) 7.6 Hz, 1H), 3.87 (m, 1H), 3.30 (dd, J ) 16.0 Hz, 4.8 Hz, 1H),
2.66 (dd, J ) 16.0 Hz, 2.0 Hz, 1H), 2.61 (m, 1H), 2.29 (m, 2H), 1.56
(m, 1H). 13C NMR: 176.1, 170.8, 147.4, 142.4, 128.1, 123.6, 65.1,
58.7, 52.9, 42.3, 35.1, 30.9; IR: 3689, 2957, 2360, 2340, 2254, 1761,
1607, 1525, 1348 cm-1. HRMS calcd for C14H14N2O5 [M+]: 290.0903,
found 290.0896.
7.43 (m, 5H), 3.99 (d, J ) 12.8 Hz, 1H), 3.91 (d, J ) 12.8 Hz, 1H),
3.36 (t, J ) Hz, 6.0 1H), 2.24 (m, 2H), 1.95 (m, 2H). 13C NMR: δ
181.9, 177.3, 134.5, 129.5, 129.1, 128.9, 62.4, 50.8, 34.1, 27.8. HRMS
calcd for C12H16NO4 [M + H+]: 238.1074, found 238.1076.
(S)-N-Benzylpyroglutamic acid was prepared according to a
published procedure:24 84% yield; mp 71-75 °C; [R]25D +54.3° (c 2.32,
MeOH); (lit.24 mp 92-93 °C; [R]25D +54.6° (c 2.32, MeOH)). 1H NMR:
δ 11.7 (br. s, 1H), 7.26 (m, 5H), 5.24 (d, J ) 14.4 Hz, 1H), 4.0 (dd,
J ) 9.6 Hz, J ) 3.2 Hz, 1H), 3.97 (d, J ) 14.4 Hz, 1H), 2.63 (m, 1H),
2.51 (m, 1H), 2.27 (m, 1H), 2.17 (m, 1H). 13C NMR: δ 176.7, 173.8,
135.1, 128.7, 128.4, 127.8, 58.7, 45.6, 29.5, 22.7. HRMS calcd for
C12H13NO3 [M+]: 219.0895, found 219.0900.
L-Proline. N-Cbz-L-Glutamic acid 1-methyl ester was prepared
according to a published procedure:19,20 85% yield; [R]25 -24.5° (c
D
1, CH3Cl); TLC (EtOAc/hexanes/AcOH, 40/60/1) Rf 0.32. 1H NMR: δ
7.34 (m, 5H), 5.45 (d, J ) 8.0 Hz, 1H), 4.40 (dt, J ) 13.2, 5.2 Hz,
1H), 3.75 (s, 3H), 2.41 (m, 2H), 2.21 (m, 1H), 1.98 (m, 1H). 13C
NMR: δ 177.9, 172.3, 156.0, 136.0, 128.5, 128.2, 128.1, 67.1, 53.1,
(S)-tert-butyl N-benzylpyroglutamate was prepared according to
52.6, 29.8, 27.4. IR: 3035, 2956, 1718, 1509, 1438, 1347, 1214 cm-1
.
a published procedure:24 62% yield; mp 73-74 °C; [R]25 +16.1° (c
D
HRMS calcd for C14H17NO6 [M + H+]: 296.1135, found 296.1141.
4-Bromo-(2S)-Cbz-aminomethylbutanoate (11) was prepared ac-
cording to a published procedure:22 58% yield; mp 61 °C; [R]25D -40.3°
2.6, CH3Cl); (lit.24 mp 62-63 °C; [R]25D +16.2° (c 2.6, CH3Cl)); TLC
1
(EtOAc/hexanes, 25/75) Rf 0.17. H NMR: δ 7.25 (m, 5H), 4.97 (d,
J ) 15.2 Hz, 1H), 3.88 (d, J ) 15.2 Hz, 1H), 3.76 (dd, J ) 9.2 Hz, 3.2
Hz, 1H), 2.52-1.96 (m, 4H), 1.37 (s, 9H). 13C NMR: δ 176.1, 171.9,
137.0, 129.7, 129.5, 128.7, 83.2, 60.5, 46.5, 30.6, 28.9, 23.9. IR 2249,
1733, 1684, 1417 cm-1. HRMS calcd for C16H21NO3 [M+]: 275.1521,
found 275.1525.
(c 1, DMF); (lit.22 mp 63-64 °C, [R]25 -40.8° (c 1, DMF)); TLC
D
1
(EtOAc/hexanes, 20/80) Rf 0.34. H NMR: δ 7.35 (m, 5H), 5.39 (d,
J ) 7.3 Hz, 1H), 5.11 (s, 2H), 4.51 (dt, J ) 12.8, 4.8 Hz, 1H), 3.76 (s,
3H), 3.42 (t, J ) 6.8 Hz, 2H), 2.44 (m, 1H), 2.25 (m, 1H). 13C NMR:
δ 172.9, 156.9, 137.0, 129.7, 129.4, 129.2, 68.3, 53.9, 53.8, 36.8, 29.1;
IR 2252, 1723, 1507 cm-1. HRMS calcd for C13H16BrNO4 [M+]:
329.0263, found 329.0271.
(S)-N-Benzyl-5-thioxoproline tert-butyl ester (14) was prepared
according to a published procedure:24 82% yield; mp 75-78 °C; [R]25
D
+195° (c 1.85, CH3Cl); (lit.24 mp 78-79 °C; [R]25 +190° (c 1.85,
D
4-Cyano-(2S)-Cbz-aminomethylbutanoate (12). Sodium cyanide
(2.13 g, 4.27 mmol, 1.2 equiv) and 8.9 mL of DMSO were warmed
for 20 min at 60 °C with stirring. The bromide 11 (11.74 g, 35.56
mmol, 1 equiv) was added so that the reaction temperature did not
exceed 60 °C. After the addition was complete, the mixture was stirred
at 60 °C for 3 h. The reaction mixture was diluted with 30 mL of
distilled H2O and extracted with ethyl ether (2 × 150 mL). The organic
extract was washed with 1 N HCl (1 × 50 mL) and brine (1 × 25
mL), dried over magnesium sulfate, filtered, and evaporated. The
product was absorbed onto Celite and purified by silica gel chroma-
tography using 20/80 ethyl acetate/hexane as the eluant: 8.73 g, 88%
1
CH3Cl)); TLC (EtOAc/hexanes, 25/75) Rf 0.48. H NMR: δ 7.28 (m,
5H), 5.76 (d, J ) 14.4 Hz, 1H), 4.25 (d, J ) 14.4 Hz, 1H), 4.11 (ddd,
J ) 9.19 Hz, J ) 3.20 Hz, J ) 1.99 Hz, 1H), 3.06 (m, 2H), 2.17 (m,
1H), 2.06 (m, 1H), 1.40 (s, 9H). 13C NMR: δ 203.5, 169.1, 134.5, 128.7,
128.5, 128.0, 82.7, 66.2, 50.3, 43.3, 27.8, 24.7. IR: 2983, 2227, 1734,
1474 cm-1. HRMS calcd for C16H21NO2S; [M+] 291.1293, found
291.1298.
(S)-N-Benzylproline tert-Butyl Ester (15). The thiolactam 14 (2.75
g, 9.43 mmol) was dissolved in 23 mL of dry CH3CN, treated with
methyl iodide (2.34 mL, 37.5 mmol), and stirred until the thiolactam
was consumed as monitored by TLC (∼18 h). The solvent was then
evaporated and the residue was dried under high vacuum for 1 h. The
resulting thioiminium salt was then dissolved in 30 mL of methanol
and immediately treated with sodium borohydride (173 mg, 3.75 mmol)
and stirred at room temperature for 3 h. The reaction mixture was added
to 42 mL of 20% NaOH and extracted with ether (2 × 20 mL). The
ether extracts were washed with 1 N potassium bisulfate (2 × 15 mL).
The combined aqueous layers were adjusted to pH ) 12 with 20%
yield; mp 42 °C; [R]25 -21.3° (c 0.6, MeOH); (lit.31 mp 42-43 °C;
D
[R]25 -21° (c 0.6, MeOH)); TLC (EtOAc/hexanes), 20/80 Rf 0.25.
D
1H NMR: δ 7.35 (m, 5H), 5.56 (d, J ) 7.1 Hz, 1H), 5.11 (s, 2H), 4.44
(dt, J ) 12.3 Hz, J ) 4.8 Hz, 1H), 3.77 (s, 3H), 2.42 (m, 2H), 2.29
(m, 1H), 2.02 (m, 2H). 13C NMR: δ 171.2, 155.8, 135.8, 128.5, 128.3,
128.1, 118.6, 67.3, 52.9, 52.7, 28.5, 13.5. 13C NMR (with 13C label at
C5): δ 171.2, 155.8, 135.8, 128.5, 128.3, 128.1, (C5) 118.6, 67.2, 52.9,
52.7, 28.5, 13.5 (d, J ) 57.2 Hz). IR 2359, 2253, 1723, 1506 cm-1
.
HRMS calcd for C14H16N2O4 [M+]: 276.1110, found 276.1115.
(32) The Merck Index, 12th ed.; Merck Research Laboratories: Whitehouse
(31) Van, T. T.; Korjo, E.; Grzonka, Z. Tetrahedron 1977, 33, 2299-2302.
Station, NJ, 1996.
9
8492 J. AM. CHEM. SOC. VOL. 125, NO. 28, 2003

Carbapemem Biosynthesis
A R T I C L E S
NaOH and re-extracted with ether (3 × 15 mL). The ether extracts
were dried over anhydrous potassium carbonate, filtered, concentrated,
and purified by silica gel chromatography using 15% EtOAc/hexanes
as the eluant to yield a colorless oil: 1.65 g, 67% yield; [R]25D -68.33°
(c 4, CHCl3), (lit.33 for methyl ester, [R]25D -85.7° (c 4, CHCl3)); TLC
L-Proline (16). N-Benzyl-L-proline (971 mg, 4.7 mmol) was
dissolved in 7 mL of ethanol and sparged with argon for 15 min in a
Parr hydrogenation flask. Pearlman’s catalyst (146 mg, 15% loading)
was added ,and the mixture was hydrogenated at 60 psi for 2 d. The
reduction mixture was filtered through Celite and washed with several
portions of ethanol. The ethanol fractions were evaporated, and the
resulting residue was dissolved in distilled H2O and lyophilized. The
resulting crystalline residue was triturated in diethyl ether to yield a
fine white powder: 463 mg, 85% yield, mp 228 °C; [R]25D -82.3° (c
1
(EtOAc/hexanes, 15/85) Rf 0.22. H NMR: δ 7.27 (m, 5H), 3.96 (d,
J ) 12.8 Hz, 1H), 3.48 (d, J ) 12.8 Hz, 1H), 3.11 (dd, J ) 8.8 Hz, 6.0
Hz, 1H), 2.97 (abq, J ) 8.4 Hz, 1H), 2.33 (m, 1H), 2.05 (m, 1H), 1.90
(m, 2H), 1.73 (m, 1H), 1.45 (s, 9H). 13C NMR: δ 173.1, 138.7, 128.9,
127.9, 126.8, 80.3, 65.6, 58.3, 52.9, 28.9, 27.9, 22.8. HRMS calcd for
C16H23NO2, [M+] 261.1729, found 261.1731.
23.4
0.57, 0.5 N HCl); (lit.32 mp 220-222 °C; [R]D -85.0° (c 0.57, 0.5
N HCl)). 1H NMR: δ 4.08 (m, 1H), 3.29 (m, 2H), 2.29 (m, 1H), 2.05
(m, 3H). 13C NMR: δ 174.7, 61.3, 46.1, 29.1, 23.8. HRMS calcd for
C5H9NO2 [M + H+]: 116.0706, found 116.0705.
1
(S)-N-Benzylproline. N-Benzylproline tert-butyl ester 15 (1.65 g,
6.3 mmol) was dissolved in 10 mL of trifluoroacetic acid and stirred
at room temperature for 30 min. The TFA was evaporated and the
residue dried under vacuum for 1 h. The residue was then taken up in
distilled H2O and lyophilized. The crystalline residue was triturated
with diethyl ether to yield a fine white powder: 971 mg, 75%; mp
[5,5-2H2,5-13C]: [R]25 -84.1° (c 0.57, 0.5 N HCl). H NMR: δ
D
4.12 (m, 1H), 3.52 (m, 0.14H), 2.34 (m, 1H), 2.0 (m, 3H). 13C NMR:
δ 176.6, 61.3, 45.9 (5,5-d1, t, J ) 22.9 Hz), 45.6 (5,5-d2, p, J ) 22.4
Hz), 29.0, 23.6 (d, J ) 33.6 Hz). [M + H+]: 119.
167 °C; [R]25D -29.3° (c 1, EtOH); (lit.33 mp 163.7-165.7 °C; [R]25
-26.9° (c 1, EtOH)). H NMR: δ 7.46 (m, 5H), 4.32 (dd, J ) 30.0
D
1
Acknowledgment. We are grateful to the National Institutes
of Health for financial support (AI14937), to Dr. Brian O.
Bachmann for his advice early in this project, and to Dr. Barbara
Gerratana for frequent discussions and comments on the
manuscript.
Hz, 12.8 Hz, 2H), 3.95 (dd, J ) 9.6 Hz, 6.8 Hz, 1H), 3.60 (m, 1H),
3.23 (m, 1H), 2.45 (m, 1H), 1.99 (m, 3H). 13C NMR: δ 173.6, 130.6,
130.1, 130.1, 129.3, 68.3, 58.4, 54.7, 28.9, 22.9. HRMS calcd for
C12H15NO2 [M+]: 205.1103, found 205.1107.
(33) Rispens, M. T.; Gelling, O. J.; de Vries, A. H. M.; Meetama, A.; van
Bolhuis, F.; Feringa, B. L. Tetrahedron 1996, 52, 3521-3546.
JA034248A
9
J. AM. CHEM. SOC. VOL. 125, NO. 28, 2003 8493