Probable role of clavaminic acid as the terminal intermediate in the common pathway to clavulanic acid and the antipodal clavam metabolites

Article abstract of DOI:10.1021/ja963107o

Emerging chemical and genetic evidence suggests that separate biochemical solutions have evolved to synthesize the four known classes of β-lactam antibiotics. One of these classes contains clavulanic acid (1) and a family of structurally related but antip

Full text of DOI:10.1021/ja963107o

2348
J. Am. Chem. Soc. 1997, 119, 2348-2355
Probable Role of Clavaminic Acid as the Terminal Intermediate
in the Common Pathway to Clavulanic Acid and the Antipodal
Clavam Metabolites
Laura A. Egan, Robert W. Busby, Dirk Iwata-Reuyl,† and Craig A. Townsend*
Contribution from the Department of Chemistry, The Johns Hopkins UniVersity,
Baltimore, Maryland 21218
ReceiVed September 4, 1996X
Abstract: Emerging chemical and genetic evidence suggests that separate biochemical solutions have evolved to
synthesize the four known classes of â-lactam antibiotics. One of these classes contains clavulanic acid (1) and a
family of structurally related but antipodal clavam metabolites 2-5, 7, and 8 which lack a carboxylate at C-3, have
a different oxidation state, and exhibit stereochemical features at C-2. Previous work has demonstrated the
incorporation of ornithine/arginine in the identical regiochemical sense in all of these natural products, and has
established the common intermediacy of the monocyclic â-lactam proclavaminic acid (12) as well. In this paper the
quite advanced bicyclic intermediate clavaminic acid (14) has been synthesized in doubly 13C-labeled form by
preparative incubation of recombinant clavaminate synthase. The intact and equally efficient incorporation of 14
into valclavam (7) and 2-(2-hydroxyethyl)clavam (8), together with 18O2-incorporation experiments, has been interpreted
to define clavaminic acid as the final intermediate shared in the biosynthesis of clavulanic acid and the antipodal
clavams. A mechanistic rationale of this interrelationship and the late stages of the respective biosyntheses is proposed.
Introduction
is the more direct precursor from primary metabolism.10 The
first cyclic intermediate in the anabolic process is believed to
be 10,11 which is hydroxylated by the R-ketoglutarate (R-KG)-
dependent oxygenase clavaminate synthase (CS)12,13 to give 11
(Scheme 2). Through the action of a second enzyme, procla-
vaminate amidino hydrolase (PAH),11,14,15 the guanidino group
of 11 is hydrolyzed in an arginase-like reaction15 to afford
proclavaminic acid (12) and urea. Remarkably, while 11 is not
a substrate for CS, the terminal amino compound 12 is, and a
stepwise16,17 pair of oxidative reactions ensue mediated again
by CS to give clavaminic acid (14).18 A penultimate oxidation
and “enantiomerization” 19 gives the unstable aldehyde 15,
which is reduced by the NADPH-dependent clavulanate alde-
hyde dehydrogenase (CAD) to clavulanic acid (1).20
Growing biochemical and genetic evidence suggests that
separate biosynthetic strategies have evolved to create the four
known classes of â-lactam antibiotics.1 Among these the potent
â-lactamase inactivator clavulanic acid (1) is produced by
Streptomyces claVuligerus together with the antipodal 2-ala-
nylclavam (2)2 and the stereochemically related metabolites 3-5
(Scheme 1).3 The allied strain Streptomyces antibioticus does
not produce clavulanic acid, but accumulates valclavam (7)4-7
and 2-(2-hydroxethyl)clavam (8).8 Apart from having the 5S-
configuration at their ring junction, these clavams share a
common stereochemistry at C-2 and an oxidation state more
reduced at this center than clavulanic acid, and they lack the
C-3 carboxyl found in 1. These differences notwithstanding,
experiments described in this paper establish a common
biosynthetic pathway among these structures to the probable
branch point of clavaminic acid (14), a highly advanced
intermediate.
It was initially hypothesized that 2-alanylclavam (2) might
derive from ornithine (6b) in the opposite regiochemical sense
(9) Townsend, C. A.; Ho, M.-F. J. Am. Chem. Soc. 1985, 107, 1065-
1066.
(10) Valentine, B. P.; Bailey, C. R.; Doherty, A.; Morris, J.; Elson, S.
W.; Baggaley, K. H.; Nicholson, N. H. J. Chem. Soc., Chem. Commun.
1993, 1210-1211.
Early experiments demonstrated that clavulanic acid (1) is
derived from the urea cycle amino acids ornithine and arginine,9
but subsequently, mutant studies clearly showed that arginine
(11) Elson, S. W.; Baggaley, K. H.; Davidson, M.; Fulston, M.;
Nicholson, N. H.; Risbridger, G. D.; Tyler, J. W. J. Chem. Soc., Chem.
Commun. 1993, 1212-1214.
* To whom correspondence should be addressed. Phone: (410) 516-
7444. Fax: (410) 516-8420. E-mail: Townsend@jhunix.hcf.jhu.edu.
† Current address: Department of Chemistry, Portland State University,
Portland, OR 97207.
(12) Elson, S. W.; Baggaley, K. H.; Fulston, M.; Nicholson, N. H.; Tyler,
J. W.; Edwards, J.; Holms, H.; Hamilton, I.; Mousdale, D. M. J. Chem.
Soc., Chem. Commun. 1993, 1211-1212.
(13) Baldwin, J. E.; Merritt, K. D.; Schofield, C. J.; Elson, S. W.;
Baggaley, K. H. J. Chem. Soc., Chem. Commun. 1993, 1301-1302.
(14) Aidoo, K. A.; Wong, A.; Alexander, D. C.; Rittammer, R. A. R.;
Jensen, S. E. Gene 1994, 147, 41-46.
X Abstract published in AdVance ACS Abstracts, February 15, 1997.
(1) Townsend, C. A. Biochem. Soc. Trans. 1993, 21, 208-213.
(2) Mu¨ller, J.-C.; Toome, V.; Pruess, D. L.; Blount, J. F.; Weigele, M.
J. Antibiot. 1983, 36, 217-225.
(15) Wu, T. K.; Busby, R. W.; Houston, T. A.; McIlwaine, D. B.; Egan,
L. A.; Townsend, C. A. J. Bacteriol. 1995, 177, 3714-3720.
(16) Baldwin, J. E.; Adlington, R. M.; Bryans, J. S.; Bringhen, A. O.;
Coates, J. B.; Crouch, N. P.; Lloyd, M. D.; Schofield, C. J.; Elson, S. W.;
Baggaley, K. H.; Cassells, R.; Nicholson, N. J. Chem. Soc., Chem. Commun.
1990, 617-619.
(3) Brown, D.; Evans, J. R.; Fletton, R. A. J. Chem. Soc., Chem. Commun.
1979, 282-283.
(4) Ro¨hl, F.; Rabenhorst, J.; Za¨hner, H. Arch. Microbiol. 1987, 147, 315-
320.
(5) Baldwin, J. E.; Claridge, T. D. W.; Goh, K.-C.; Keeping, J. W.;
Schofield, C. J. Tetrahedron Lett. 1993, 34, 5645-5648.
(6) Janc, J. W.; Egan, L. A.; Townsend, C. A. Bioorg. Med. Chem. Lett.
1993, 3, 2213-2216.
(7) Postels, H.-T.; Ko¨nig, W. A. Tetrahedron Lett. 1994, 35, 4535-
4538.
(17) Salowe, S. P.; Krol, W. J.; Iwata-Reuyl, D.; Townsend, C. A.
Biochemistry 1991, 30, 2281-2292.
(18) Elson, S. W.; Baggaley, K. H.; Gillett, J.; Holland, S.; Nicholson,
N. H.; Sime, J. T.; Woroniecki, S. R. J. Chem. Soc., Chem. Commun. 1987,
1736-1738.
(8) Wanning, M.; Za¨hner, H.; Krone, B.; Zeeck, A. Tetrahedron Lett.
1981, 22, 2539-2540.
(19) Townsend, C. A.; Krol, W. J. J. Chem. Soc., Chem. Commun. 1988,
1234-1236.
S0002-7863(96)03107-1 CCC: $14.00 © 1997 American Chemical Society

Role of ClaVaminic Acid in the Pathway to ClaVulanic Acid
J. Am. Chem. Soc., Vol. 119, No. 10, 1997 2349
Scheme 1
Scheme 2
Scheme 3
acid synthesis in this organism (see Scheme 1). Divergence of
the biosynthetic pathways to clavulanic acid and to the clavams
at the early stage of C5- or C6-amino acid attachment to a C3-
precursor22 was quickly disproved by experiments with [3-13C]-
ornithine21 and [2,3-13C2]proclavaminic acid (16, Scheme 3).6,21
The former was incorporated into clavulanic acid at C-2 as
expected, but was found to label C-2 of clavam-2-carboxylic
acid (5).21 This location of isotopic enrichment clearly pointed
to an orientation of arginine/ornithine identical to that occurring
in clavulanic acid, implying that a common pathway extended
beyond the initial linkage of C3- and C5-precursors. Subse-
quently, administration of [2,3-13C2]proclavaminic acid (16) to
S. claVuligerus gave equally efficient incorporation of paired
heavy isotopes into clavam-2-carboxylic acid (17) and copro-
duced clavulanic acid (18) as shown in Scheme 3.21 The
common derivation of the highly modified clavam 5/17 from
an advanced precursor of clavulanic acid (1/18) was tested
further by experiments in S. antibioticus. In a similar incor-
poration study both valclavam (19) and 2-(2-hydroxyethyl)-
clavam (20) were comparably labeled by 16.6 Therefore, it
to its utilization in clavulanic acid biosynthesis (6a, heavy lines
in 2, Scheme 1).21 Analogously, valclavam (7) could be
visualized to be derived from lysine (9, heavy lines in 7)5 and
simultaneously provide a rationale for the absence of clavulanic
(20) 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.
(21) Iwata-Reuyl, D.; Townsend, C. A. J. Am. Chem. Soc. 1992, 114,
2762-2763.
(22) Townsend, C. A.; Ho, M.-F. J. Am. Chem. Soc. 1985, 107, 1066-
1068.

2350 J. Am. Chem. Soc., Vol. 119, No. 10, 1997
Egan et al.
would appear that all clavams and clavulanic acid have a
common biosynthetic origin at least up to and including
proclavaminic acid.
Scheme 4
These findings in S. antibioticus prompted a search to detect
clavaminate synthase (CS) activity in this organism, a strain
that does not produce clavulanic acid but presumably must carry
out the hydroxylation step necessary for the formation of
proclavaminic acid (12), that is, the conversion of 10 to 11 in
Scheme 2. Two isozymes of this Fe(II)/O2/R-KG-dependent
oxygenase, CS1 and CS2, have been isolated and characterized
from S. claVuligerus.23 A third protein, denoted CS3, of
virtually identical mass and kinetic properties was indeed
purified from S. antibioticus.24 N-Terminal sequence informa-
tion clearly supported the view that all three of these proteins
were highly similar.24 These findings confirmed and extended
reports elsewhere25 in which this CS activity was observed in
S. antibioticus and established the presence of a functional PAH
enzyme as well.
The existence of CS3 in S. antibioticus with all three of its
oxidative activities intact pressed further the question21 whether
clavaminic acid (14) might be a common intermediate between
clavulanic acid (1) and the clavam metabolites represented by
structures 2-7, and 8. While the latter share the ring fusion
geometry of clavaminic acid (14), they lack the C-3 carboxyl
of clavulanic acid and have the opposite absolute configuration
at C-2 of the intermediate 13 involved in the stepwise conversion
of 12 to 14 (Scheme 2).16,17 Moreover, clavaminic acid itself
appears to be of the wrong oxidation state to serve readily as a
precursor of the clavams saturated at C-2. To determine the
extent of the common pathway, [2,3-13C2]clavaminic acid has
been prepared and its intact incorporation has been demonstrated
into valclavam (7) and 2-(2-hydroxyethyl)clavam (8). Detailed
mechanistic consideration of these findings suggests that cla-
vaminic acid (14) is the last biosynthetic intermediate common
among these natural products.
which was reduced with nBu3SnH and AIBN to 29. Deproto-
nation of 29 with LiHMDS at -78 °C and aldol reaction29,33-35
with 26 gave a mixture of the erythro-30 and threo-31 products,
favoring the former.17,34 Flash or radial chromatography gave
small amounts of pure 30 and 31 as well as mixed fractions.
The latter were combined with 30 and equilibrated with DBN34
to now significantly favor (>3:1) the desired threo product 31.
After further careful chromatographic purification, the pooled
31 was hydrogenolyzed17,29,35 to afford (()-[2,3-13C2]pro-
2
clavaminic acid (16, JCC ) 39.3 Hz).
Results
In the presence of Fe(II), R-KG, and molecular oxygen,
clavaminate synthase is known to be rapidly inactivated (t1/2
)
Clavaminic acid is a modestly stable amino acid not readily
accessible in 13C-labeled form by total synthesis. As CS2 from
S. claVuligerus has been successfully overproduced in Escheri-
chia coli,26,27 it was resolved at the outset to convert doubly
13C-labeled proclavaminic acid enzymically to the correspond-
ingly enriched clavaminic acid.28 Racemic [2,3-13C2]pro-
clavaminic acid (16) was prepared as outlined in Scheme 4.
Label was introduced from potassium [13C]cyanide and sodium
[2-13C]bromoacetate. 2-Aminoethanol was protected to give 21,
which was converted to the previously described 13C-labeled
acid 24.29 Reduction of the mixed anhydride30 of 24 gave the
alcohol 25, which was readily oxidized29 to the aldehyde 26.
Labeled bromoacetic acid was converted to its benzhydryl ester31
and reacted with 4-(phenylthio)azetidin-2-one (27)32 to give 28,
ca. 5 min).23,24 In the presence of these cofactors and substrate
the half-life increases only to about 20 min, from which it can
be calculated that CS undergoes fewer than 1000 turnovers
before inactivation.23 Therefore, the preparative use of cla-
vaminate synthase would require several hundred milligrams
of protein to produce 50-100 mg of clavaminic acid.
Overproduction of CS2 was carried out essentially as
described previously27 with one modification. The incubation
time following induction of the overexpression vector with
nalidixic acid was lengthened from 3 to 7 h and gave increased
yields of enzyme. A streamlined purification procedure was
developed to obtain protein of sufficient purity (>80%) for the
large-scale enzymatic conversion of labeled proclavaminic acid
to clavaminic acid. This shortened protocol omitted the final
gel filtration step using G-75 described in the isolation of the
wild-type enzyme.23
(23) Salowe, S. P.; Marsh, E. N.; Townsend, C. A. Biochemistry 1990,
29, 6499-6508.
(24) Janc, J. W.; Egan, L. A.; Townsend, C. A. J. Biol. Chem. 1995,
270, 5399-5404.
Optimization of Enzymic Conversion. Maximizing the
efficiency of the CS2 cyclization/desaturation reaction was
paramount owing to the value of the doubly 13C-labeled substrate
and the amount of recombinant enzyme that would be required.
(25) Baldwin, J. E.; Fujishima, Y.; Goh, K. C.; Schofield, C. J.
Tetrahedron Lett. 1994, 35, 2783-2786.
(26) Lawlor, E. J.; Elson, S. W.; Holland, S.; Cassels, R.; Hodgson, J.
E.; Lloyd, M. D.; Baldwin, J. E.; Schofield, C. J. Tetrahedron 1994, 50,
8737-8748.
(27) Busby, R. W.; Chang, M. D.-T.; Busby, R. C.; Wimp, J.; Townsend,
C. A. J. Biol. Chem. 1995, 270, 4262-4269.
(31) Stelakatos, G. C.; Paganou, A.; Zervas, L. J. Chem. Soc. C 1966,
1191-1199.
(28) Such an enzymatic conversion has been carried out previously,
although no experimental details were reported: Elson, S. W.; Baggaley,
K. H.; Gillett, J.; Holland, S.; Nicholson, N. H.; Sime, J. T.; Woroniecki,
S. R. J. Chem. Soc., Chem. Commun. 1987, 1739-1740.
(29) Townsend, C. A.; Basak, A. Tetrahedron 1991, 47, 2591-2602.
(30) Ishizumi, K.; Koga, K.; Yamada, S.-I. Chem. Pharm. Bull. 1968,
16, 492-497.
(32) Kobayashi, T.; Iwano, Y.; Hirai, K. Chem. Pharm. Bull. 1978, 26,
1761-1767.
(33) Pansare, S. V.; Vederas, J. C. J. Org. Chem. 1987, 52, 4804-4810.
(34) Baggaley, K. H.; Elson, S. W.; Nicholson, N. H.; Sime, J. T. J.
Chem. Soc., Perkin Trans. 1 1990, 1513-1520.
(35) Iwata-Reuyl, D.; Basak, A.; Silverman, L. S.; Engle, C. A.;
Townsend, C. A. J. Nat. Prod. 1993, 56, 1373-1396.

Role of ClaVaminic Acid in the Pathway to ClaVulanic Acid
J. Am. Chem. Soc., Vol. 119, No. 10, 1997 2351
Several variables were examined to establish optimal conditions
on a small scale that could be applied to a large-scale incubation.
The variables examined include temperature, pH, buffer com-
position, substrate and cofactor concentrations, and the substrate:
enzyme ratio.
Knowing the oxidative lability of the enzyme and uncertain
of the stability of the product, the conversion was examined at
0 and 25 °C. Previous experiments had shown that higher
temperatures had an adverse effect on CS stabililty. However,
in a 24 h period, little difference was observed in the net yield
and the slower, low-temperature reaction was abandoned.
Similarly, small variations of the pH around 7.0 had minimal
effect on product stability in solution, so the reaction was
conducted at this pH, which is the pH optimum of clavaminate
synthase.23 Tris, MOPS, and phosphate buffer were prepared
at pH 7, and the enzyme activity was measured. While the
organic amine buffers were less easily removed in the final
HPLC purification step, reaction in phosphate was only about
50% as efficient.
Figure 1. Conversion of (()-[2,3-13C2]proclavaminic acid (16) to (-)-
[2,3-13C2]clavaminic acid (32) using recombinant CS2 monitored as a
function of time.
Scheme 5
Significant effects on the overall enzymic conversion were
observed in variations of the substrate:cofactor and substrate:
CS2 ratios. The concentration of racemic proclavaminate was
arbitrarily set at 1 mM on the basis of earlier experience.
Periodic substoichiometric additions of R-KG were found to
be effective in prolonging the catalytic lifetime of the enzyme
and slightly improved the net yield. In the final optimization
10% of the theoretically required R-KG was added every 20
min of the incubation. Rather than a single addition of CS2, it
was determined that portionwise addition of the enzyme
benefited the overall yield. While the enzymic conversion of
proclavaminate to clavaminate can be carried out functionally
to completion, large amounts of enzyme are required. In a
typical test reaction of 40 µg of (()-12, turnover efficiency
increased with further enzyme up to approximately 70 µg of
CS2, after which additional protein improved the product yield
only marginally. A practical compromise among these variables
was achieved by hourly additions of CS2 for the first 4 h of
reaction to a final substrate:enzyme molar ratio of 80:1. The
optimized conditions were examined on a scale 200-fold greater
than the test reactions, but still on 1/5 the scale envisioned for
the labeled synthesis. A 68% yield was attained, in heartening
agreement with the small-scale test reactions.
Enzymatic Synthesis of [2,3-13C2]Clavaminic Acid. The
amount of racemic [2,3-13C2]proclavaminic acid (16) was
determined using the ninhydrin assay previously described23 and
found to be 1.53 mmol. This was converted to [2,3-13C2]-
clavaminic acid (32) in a 1.5 L reaction containing 0.1 mM
(0.15 mmol, 10 mol %) R-KG, 25 µM ferrous ammonium
sulfate, dithiothreitol, and ascorbic acid. Additional 0.15 mmol
amounts of R-KG were added every 20 min over 9 h, and
partially purified recombinant CS2 estimated to contain 1.0 g
of pure enzyme was supplied in equal portions at 0, 1, 2, and
3 h of incubation. We were gratified to find that a 71% yield
of 32 was obtained (Figure 1).
Control experiments showed that the observed concentrations
of MOPS buffer had no adverse effects on cell growth or clavam
production.
Incorporation of [2,3-13C2]Clavaminic Acid. S. antibioticus
was grown in a modified soy bean medium, and the doubly
13C-labeled clavaminic acid 32 was administered (0.2 mM) at
the onset of clavam production as determined by the imidazole
assay.23,36 After 120 h, valclavam (19) and 2-(2-hydroxyethyl)-
clavam (20) were isolated and purified (see the Experimental
Section). 13C{1H}NMR analysis revealed that the labeled
precursor had been incorporated intact and with equal efficiency
(ca. 0.2%) into both 19 and 20 (Scheme 5). The coupling
constant observed between the enriched C-2 and C-3 sites was
1JCC ) 32.6 Hz, in accord with the earlier successful incorpora-
tion of [2,3-13C2]proclavaminic acid (16).6 To ensure that the
observed incorporation of paired 13C-labels from [2,3-13C2]-
clavaminic acid (32) was not due to trivial incorporation of
unreacted [2,3-13C2]proclavaminic acid (16) carried along from
the enzyme incubation, the integrity of the purified cyclization
1
product was confirmed by analytical HPLC and H- and 13C-
spectroscopy. The relatively low but specific incorporation of
clavaminic acid was similarly observed in experiments to
investigate the biosynthesis of clavulanic acid (1).28
The reaction mixture was immediately passed through a
DEAE Sephadex A25 column to remove most of the protein.
Unlike the smaller scale test incubations, a second protein
removal step was required prior to HPLC separation. This was
successfully achieved by membrane ultrafiltration. Remaining
[2,3-13C2]-(2S)-proclavaminic acid (16) and its unreactive 2R-
enantiomer were cleanly resolved from the 3S,5S-product 32
by preparative reversed-phase HPLC. [2,3-13C2]Clavaminic acid
(32) was obtained as an off-white solid contaminated by small
amounts of MOPS buffer, but no labeled proclavaminate 16.
Discussion
The absolute configurations at C-2 and C-5 are identical
among all the known clavams2-4,8 and the more complex
clavamycins.37 Each lacks the C-3 carboxyl seen in clavulanic
(36) Bird, A. E.; Bellis, J. M.; Gasson, B. C. Analyst 1982, 107, 1241-
1245.
(37) Naegeli, H. U.; Loosli, H.-R.; Nussbaumer, A. J. Antibiot. 1986,
39, 516-524.

2352 J. Am. Chem. Soc., Vol. 119, No. 10, 1997
Egan et al.
Scheme 6
acid (1), as well as having the opposite ring fusion geometry.
As shown schematically in Scheme 6, we propose21 the
intermediacy of aldehyde 39, whose simple reduction in S.
antibioticus could be visualized to afford 2-(2-hydroxyethyl)-
clavam (8), or by Baeyer-Villiger-like oxidation in S. cla-
Vuligerus could give the formate ester 3, which in turn could
be hydrolyzed to the alcohol 4 and further oxidized to clavam-
2-carboxylate 5. Aldol-like reactions can be invoked to generate
the carbinol center at C-9 in valclavam (7)6,38 and similarly the
various clavamycins. The implied relationship between cla-
vaminic acid (14) and the aldehyde 39 poses an interesting
question for which several mechanistic rationales can be
advanced. Preservation of the stereochemistry at C-5 is required
as is a means to promote decarboxylation. Ideally, the resulting
pair of electrons should be stored in the substrate to, in effect,
reduce the clavaminate double bond and generate the stereo-
center at C-2. Among these mechanistic possibilities, pyridoxal
can be envisioned to play such a role as outlined in Scheme 6.
Initial imine formation between 14 and pyridoxal phosphate
(PLP) and isomerization to 34 are proposed to provide a ready
path for decarboxylation. Protonation at C-3 of the resulting
vinylogous enamine 35 and isomerization and a second proto-
nation are suggested to set the stereochemistry at C-2 and result
in formation of the labile azadiene 37. Hydrolysis of the latter
would restore the PLP cofactor and release the new enamine
38, which upon hydration would give the desired aldehyde 39.
While the transformations proposed in Scheme 6 must be
regarded as speculation at present, experimental observations
have been made, particularly in clavulanic acid biosynthesis,
that lend indirect support to this scheme and further define the
extent of the common pathway to the â-lactamase inactivator 1
and the antipodal clavams. It was shown a decade ago that
growth of S. claVuligerus in an 18O2-containing atmosphere gave
rise to the incorporation of heavy isotope (*) at the oxazolidine
oxygen and, unexpectedly, into the allylic hydroxyl of clavulanic
acid (1; see Scheme 7).19 The latter finding suggested enzymic
hydroxylation at C-9 to give a transient allylic carbinolamine
bearing label from molecular oxygen. This unstable intermedi-
ate was proposed to decompose to the aldehyde enantiomeric
to 15. This intermediate was thought to be stereochemically
unstable and to “racemize”, from which the mirror image
stereoisomer could be selected by a dehydrogenase and reduced
to the enantiomeric, now biologically active, clavulanic acid
(1).19 This interpretation is not fully correct as Elson and co-
workers have demonstrated.20 In admirably careful experiments
it was shown that the aldehyde 15 is unstable but isolable as a
single stereoisomer in small quantities from fermentations of
S. claVuligerus. Moreover, the Beecham group has character-
ized an NADPH-dependent dehydrogenase responsible for
carrying out the final reduction of 15 to 1.20 Two facts should
be noted at this point. First, the striking “enantiomerization”
process that occurs during the oxidative deamination of 14 to
15 must take place prior to, or concomitant with, product release.
Second, the oxygen (*) introduced in this process is not lost
and appears with undetectable exchange in the product 1. The
intermediate aldehyde 15 is evidently stereochemically stable
and, as a vinylogous ester, is slow to undergo exchange of the
labeled aldehyde oxygen relative to the rate of enzymic reduction
to 1.
An instructive comparison can be made to fermentation of
S. antibioticus in an 18O2-containing atmosphere39san organism,
it is recalled, that does not produce clavulanic acid. This
experiment was carried out as previously described19 and gave
efficient incorporation of isotope into the oxazolidine ring sites
of valclavam (7) and 2-(2-hydroxyethyl)clavam (8), as expected
on the basis of the intermediacy of clavaminic acid (14). In
neither of these metabolites, however, could 18O be detected at
(38) Baldwin, J. E.; Goh, K.-C.; Schofield, C. J. Tetrahedron Lett. 1994,
35, 2779-2782.
(39) Egan, L. A.; Townsend, C. A., Unpublished results.

Role of ClaVaminic Acid in the Pathway to ClaVulanic Acid
J. Am. Chem. Soc., Vol. 119, No. 10, 1997 2353
Scheme 7
C-9 by NMR spectroscopy40 or mass spectrometry. The
complete absence of labeled oxygen (*) at these carbinol sites
is somewhat less compelling than their presence in path A where
structural features of the intermediate 15 can be seen to retard
or prevent loss of heavy isotope. On the other hand, presuming
the intermediacy of the unconjugated aldehyde 39 in path B,
such exchange would be more greatly favored. However, the
absolute configurations of 15 and 39 are opposite, and one might
suppose that, if 39 were initially formed by oxidative deami-
nation, a portion of the 18O-label would remain at C-9 in 7 and
8 in the time regime of biosynthesis at neutral pH and a
temperature of ca. 28 °C. The absence of detectable isotope
from molecular oxygen at the carbinol centers is consistent with
these oxygens appearing in the biosynthesis from the aqueous
medium. Whether mediated by PLP, as suggested in Scheme
6, or by some other means, together these findings point to
clavaminic acid (14) being the final common intermediate to
clavulanic acid (1, path A) and all of the remaining clavam
natural products by a distinct route (path B).
S. antibioticus, an organism that produces valclavam (7) and
2-(2-hydroxyethyl)clavam (8), but no clavulanic acid (1),
contains a single form of clavaminate synthase (CS3).24 In
contrast S. claVuligerus produces four stereochemically related
clavams 2-5 as well as clavulanic acid and harbors two
isozymes of this protein (CS1 and CS2).23 It has been proposed
that this difference might imply separate genetic loci for the
biosynthetic machinery responsible for clavulanic acid and
clavam synthesis.41,42 The experiments described in this paper
demonstrate a common biosynthetic pathway of considerable
length in the formation of these natural product groups. The
extent to which the corresponding genes encoding these common
enzymes are duplicated in S. claVuligerus, beyond cs1 and cs2,
will be of interest.
beads (0.5 mm) were purchased from VWR. Collodian apparatus and
membranes were purchased from Schleicher & Schuell (Keene, NH).
All other chemicals (Aldrich Chemical Co., Milwaukee, WI) were of
reagent grade and used without further purification. Fermentation media
were obtained from Difco Laboratories (Detroit, MI).
Hydrogen and carbon-13 NMR spectra were acquired on a Bruker
AMX 300 MHz, a Varian Unityplus 400 MHz, or a Varian Unityplus
500 MHz spectrometer. Chemical shifts of hydrogen resonances are
reported as parts per million and referenced to D2O (4.78 ppm) or CDCl3
(7.26 ppm). Carbon-13 chemical shifts are reported in parts per million
and referenced to CDCl3 (77.0 ppm) or internally referenced to 1,4-
dioxane (66.5 ppm) for solutions in D2O. Melting points were
determined in open capillary tubes with a Thomas-Hoover Uni-Melt
melting point apparatus and are uncorrected. Thin layer chromatog-
raphy was performed using Analtech (Newark, DE) Uniplate TLC plates
stained using KMnO4 (1% in 6.5% Na2CO3/0.2% KOH). Low- and
high-resolution mass spectral data were obtained on a VG Instruments
70-S GC/MS at 70 eV and are tabulated as m/z (intensity expressed as
a percent of the base peak). UV-vis spectrophotometry employed a
Beckman DU 70 spectrophotometer (Fullerton, CA). All MPLC (flash)
chromatography was carried out using Merck Kieselgel 60 (230-400
mesh) silica gel. Radial chromatography was performed on a Chro-
matotron (Harrison Research, Palo Alto, CA), using rotors prepared
with silica gel PF-254 with CaSO4‚1/2H2O as binder. HPLC chroma-
tography was performed with a Waters 600 multisolvent delivery system
equipped with a Rheodyne injector and a Waters 490 programable
multiwavelength detector. Protein assays were performed by the
method of Bradford using BSA as a standard.44 SDS-PAGE used the
buffer system of Laemmli45 in a Hoefer SE 400 slab gel electrophoresis
unit (San Francisco, CA).
Expression of CS2. Cultures of E. coli JM101 cells (100 mL)
containing the recombinant pARC2P plasmid27 were prepared by
inoculating the liquid medium (2 × YT46 containing 300 µg/mL of
ampicillin) with single colonies picked from agar plates. The cell
cultures were frozen at -86 °C in a 50% glycerol suspension.
Overnight cultures were initiated by adding 5 mg of the frozen cell
suspension to 2 × YT liquid medium containing 300 µg/mL of
ampicillin and then incubated with shaking (300 rpm) at 37 °C.
Following overnight incubation, the cultures were diluted into 500 mL
of 2 × YT medium (300 µg/mL of ampicillin) to an optical density at
590 nm (A590) of 0.10 and grown at 37 °C with shaking to approximately
1 × 109 cells/mL (A590 ) 0.90). Expression was induced by the addition
of nalidixic acid at a concentration of 50 µg/mL, and expression
proceeded at 19 °C for 7 h. The cells were harvested by centrifugation
at 4000g for 10 min, frozen in liquid N2, and stored at -86 °C.
Experimental Section
General Methods. Air- and moisture-sensitive reactions were run
under an inert atmosphere (Ar or N2) in oven-dried (150 °C) glassware.
Tetrahydrofuran (THF) was freshly distilled from Na/benzophenone
ketyl, and dichloromethane (CH2Cl2) was freshly distilled from CaH2.
All other solvents for air- or moisture-sensitive reactions were used as
received or dried by standard procedures.43 Nalidixic acid, Brij-58,
Q-Sepharose resin, and G-75-50 superfine Sephadex resin were
purchased from Sigma Chemical Co. (St. Louis, MO). Glass micro-
Abbreviated CS2 Purification Procedure. All steps were con-
ducted at 0-4 °C. General enzyme buffer (GEB) consisted of 50 mM
Tris (pH 7.0), 20% glycerol, and 10 µM EDTA with 1 mM benzamidine
(40) Vederas, J. C. Nat. Prod. Rep. 1987, 4, 277-337.
(41) Marsh, E. N.; Chang, M. D.-T.; Townsend, C. A. Biochemistry 1992,
31, 12648-12657.
(42) Paradkar, A. S.; Jensen, S. E. J. Bacteriol. 1995, 177, 1307-1314.
(43) Perrin, D. D.; Armarego, W. L. F. Purification of Laboratory
Chemicals, 3rd ed.; Pergamon Press: New York, 1988.
(44) Bradford, M. M. Anal. Biochem. 1976, 72, 248-254.
(45) Laemmli, U. K. Nature (London) 1970, 227, 680-685.
(46) Maniatis, T.; Fritsch, E. F.; Sambrook, J. Molecular Cloning. A
Laboratory Manual; Cold Spring Harbor Laboratory: New York, 1982.

2354 J. Am. Chem. Soc., Vol. 119, No. 10, 1997
Egan et al.
and 1 mM DTT added just prior to use. Partial purification of CS2
was carried out using 50 g of frozen expressed E. coli JM101 cell paste
thawed in lysis buffer (50 mM Tris (pH 8.5), 20% glycerol, 100 mM
KCl, 10 mM DTT, and 10 µM EDTA) in a total volume of 75 mL
divided evenly into 3 × 50 mL Falcon tubes. Cells were lysed by the
addition of 600 µL/tube (1/50th volume) of 1% lysozyme in 10 mM
Tris buffer (pH 8.5). Simultaneously, 300 µL/tube of 1 mg/mL DNase
I in 1 × TE (10 mM Tris (pH 8), 10 µM EDTA) was added and the
cells were incubated on ice with occasional inverting. After 5 min,
600 µL/tube (1/50th) of a 5% Brij-58 solution was added and incubation
at 0 °C continued with inverting for an additional 10 min. The lysis
mixture was further divided into six Falcon tubes (15 mL/tube), and
nitric acid-washed glass microbeads (0.5 mm, 15 g) were then added
to each tube. Cell lysis was completed by vortexing each tube for two
15 s bursts, 1 min apart. The cells were diluted to 700 mL with GEB
and stirred for 20 min. A 1% streptomycin sulfate fractionation was
used to remove nucleic acids, and the protein was precipitated by the
addition of ammonium sulfate to 70% of saturation. Following
centrifugation the protein pellet was resuspended in 20 mL of GEB
(final volume approximately 30 mL). The solution was dialyzed in a
Spectra/Por porous membrane (10 000 molecular weight cut off) against
2 L of GEB for a total of 4 h (the initial GEB was replaced after 2 h).
The dialyzed protein was removed, diluted to 150 mL with GEB, and
loaded onto a Q-Sepharose column (2.5 × 44 cm, 216 mL) equilibrated
in GEB at a rate of 2.0 mL/min. The column was washed with 175
mL of GEB followed by elution with a 1.0 L linear gradient of KCl
from 0 to 400 mM in GEB, collecting 10 mL fractions. CS2 eluted at
240 mM KCl, and fractions containing CS activity were pooled and
concentrated to approximately 20 mL in an Amicon ultrafiltration cell
fitted with a PM 10 membrane. The solution was further concentrated
and dialyzed in storage buffer (50 mM MOPS (pH 7.0), 20% glycerol,
10 µM EDTA, and 1 mM DTT) using the Collodian apparatus and
stored at -20 °C.
was concentrated in Vacuo and purified by flash chromatography (80
g of silica gel; petroleum ether:EtOAc, ) 8:2 to 3:7) to give 3.528 g
(16.94 mmol, 76%) of the desired aldehyde as a fluffy white solid.
Recrystallization from EtOAc/petroleum ether provided the aldehyde
26 as white leaves: mp 59-60 °C; 1H NMR (CDCl3) δ 9.80 (d, JC-H
) 174.3 Hz, 1H, H-1), 7.35 (m, 5H, ArH), 5.12 (br, 1H, NH), 5.10 (s,
2H, CH2Ar), 3.50 (m, 2 H, NCH2), 2.75 (m, 2 H, COCH2); 13C{1H}
NMR (CDCl3) δ 201.1 (enriched), 156.2, 136.1, 128.5, 128.1, 128.0,
66.7, 44.0 (d, J ) 37.6 Hz), 34.4. Characterization of the unlabeled
aldehyde gave the following analytical data: MS m/z 207 (M+, 3),
108, 91 (100), 79, 65; accurate mass 207.0900, calcd for C11H13NO3
207.0895.
Benzhydryl [2-13C]-2-[4′-(Phenylthio)-2′-oxoazetidin-1′-yl]acetate
(28). 4-(Phenylthio)azetidin-2-one (27)32 (2.237 g, 12.48 mmol) in THF
(25 mL) was cooled to -78 °C, and LiHMDS (13.8 mL, 1.0 M in
hexane) was added dropwise. The reaction was stirred for 45 min,
and a solution of benzhydryl [2-13C]bromoacetate31 (4.246 g, 13.87
mmol) in THF (10 mL) was added over 25 min. After stirring for a
few minutes at -78 °C, the reaction was allowed to warm to 0 °C
over the course of 2 h. The mixture was filtered through a plug of
silica gel washing with EtOAc, and the filtrate was concentrated in
Vacuo. Purification by flash chromatography on silica gel gave 2.226
g (5.504 mmol, 44%) of the desired product 28 as a pale yellow oil:
1H NMR (CDCl3) δ 7.38-7.18 (m, 15H, ArH), 6.84 (s, 1H, CHAr2),
5.21 (dd, J ) 2.3, 5.0 Hz, 1H, H-4′), 4.41 (dd, J ) 18.1, 139.4 Hz,
1H, 13CH), 3.83 (dd, J ) 18.1, 141.1 Hz, 1H, 13CH), 3.41 (dd, J ) 4.7,
15.4 Hz, 1H, H-3′cis), 2.88 (dd, J ) 2.6, 15.1 Hz, 1H, H-3′trans).
Unlabeled material prepared in an analogous manner gave the following
spectral data: IR (CHCl3) 3060, 3037, 3013, 1760 (br), 1496, 1454,
1437, 1401, 1365, 1196, 1178, 967, 941, 914, 700 cm-1 13C{1H} NMR
;
(CDCl3) δ 167.1, 165.4, 139.3, 139.2, 133.7, 130.3, 129.4, 128.7, 128.6,
128.3, 128.2, 127.1, 127.0, 78.3, 59.1, 45.0, 41.3; MS m/z 167 (100),
109; accurate mass 403.1246, calcd for C24H21NO3S 403.1242.
Assay for Clavaminate Synthase Activity. Assays for the conver-
sion of 7 to 8 were conducted as previously reported.23
Benzhydryl [2-13C]-2-(2′-Oxoazetidin-1′-yl)acetate (29). The 4-(phe-
nylthio)azetidin-2-one 28 (2.020 g, 5.167 mmol) was dissolved in dry
benzene (35 mL), and the solution was thoroughly degassed. AIBN
(170.9 mg, 1.041 mmol) and tributyltin hydride (2.8 mL, 10.3 mmol)
were added, and the reaction was heated to reflux with stirring under
N2. After 10 h, additional AIBN (102.3 mg, 0.6230 mmol) and
tributyltin hydride (1.0 mL, 3.7 mmol) were added, and the reaction
was continued for another 9 h. The reaction mixture was concentrated
in Vacuo and the residue taken up in acetonitrile (100 mL) and washed
with hexane (2 × 100 mL). The acetonitrile solution was concentrated
and the residue purified by radial chromatography (2 mm silica gel;
petroleum ether:EtOAc ) 9:1 to 1:1) to give 171.9 mg (0.4250 mmol,
8.2%) of unreacted starting material and 880.5 mg (2.971 mmol, 58%)
of the reduced azetidinone 29 as a clear oil, which solidified upon
Synthesis of (()-[2,3-13C2]Proclavaminic Acid (16). [1-13C]-3-
[(Benzyloxycarbonyl)amino]propanol (25). The previously de-
scribed29 Cbz-protected [1-13C]-â-alanine (24; 6.135 g, 27.36 mmol)
in dry THF (150 mL) was treated at -10 °C with freshly distilled
triethylamine (3.85 mL, 27.6 mmol), and ethyl chloroformate (3.2 mL,
33.5 mmol) was added slowly over 20 min. After stirring for an
additional 90 min, the precipitated triethylammonium chloride was
removed by filtration through a fritted funnel and the cooled filtrate
was added over 40 min to a 5 °C solution of NaBH4 (2.1 g, 55 mmol)
in water (80 mL). When the addition was complete, more NaBH4 (1.0
g, 26 mmol) was added and the reaction mixture was stirred for 4.5 h.
After acidification with 6 N HCl, the solution was extracted with EtOAc
(2 × 350 mL) and the combined organic layers were washed with 5%
NaHCO3 (600 mL), and brine (450 mL) and dried over anhydrous
NaSO4. Concentration in Vacuo gave a white solid, which was
recrystallized from EtOAc/hexanes to give 4.809 g (22.87 mmol, 84%)
of the desired alcohol as white needles: mp 44-45 °C; IR (CHCl3)
3630, 3450, 3015, 2950, 1706, 1520, 1250, 1225, 1135, 1072, 1028
1
standing: mp 69-70.5 °C; H NMR (CDCl3) δ 7.32 (m, 10H, ArH),
6.92 (s, 1H, CHAr2), 4.11 (d, J ) 139.6 Hz, 2H, 13CH2), 3.41 (t, J )
4.2 Hz, 2H, H-4′), 3.03 (t, J ) 4.1 Hz, 2H, H-3′); 13C{1H} NMR
(CDCl3) δ 43.3 (enriched). Unlabeled material prepared in an analogous
manner gave the following spectral data: 13C{1H} NMR (CDCl3) δ
167.9, 167.4, 139.4, 128.6, 128.2, 127.0, 78.1, 43.3, 40.0, 37.7; MS
m/z 295 (M+, 0.3), 183, 168, 167, 152, 84, 77, 42; accurate mass
295.1212, calcd for C18H17NO3 295.1208.
1
cm-1; H NMR (CDCl3) δ 7.35 (m, 5H, ArH), 5.11 (s, 2H, CH2Ar),
5.02 (br, 1H, NH), 3.68 (dt, J ) 5.6, 141.6 Hz, 2H, H-1), 3.37 (m, 2H,
H-3), 2.51 (br, 1H, OH), 1.70 (m, 2H, H-2); 13C{1H} NMR (CDCl3) δ
157.4, 136.4, 128.6, 128.2, 128.1, 66.9, 59.5 (enriched), 37.7, 32.6 (d,
Benzhydryl [2,3-13C2]-5-[(Benzyloxycarbonyl)amino]-3-hydroxy-
2-(2′-oxoazetidin-1′-yl)pentanoates 30 and 31. Ester 29 (880.5 mg,
2.971 mmol) in THF (50 mL) was cooled to -78 °C, and LiHMDS
(3.4 mL, 1.0 M in hexane) was added dropwise. After stirring for 30
min, a solution of the Cbz-protected aldehyde 26 (932.8 mg, 4.480
mmol) in THF (8 mL) was added dropwise, and the reaction was stirred
for 2 h. A solution of 5 M acetic acid (600 µL) in THF was added to
quench the reaction, and after warming to room temperature, the mixture
was transferred to EtOAc (200 mL) and washed with 1 N HCl (200
mL), 5% NaHCO3 (200 mL), and brine (2 × 200 mL). The organic
layer was concentrated in Vacuo and the residue purified by radial
chromatography (2 mm silica gel; petroleum ether:ethyl acetate, ) 4:1
to 1:1) to give 201.7 mg (0.9687 mmol) of the unreacted aldehyde 26,
140.0 mg (0.4725 mmol, 16%) of the unreacted azetidinone 29, 221.6
mg (0.4392 mmol, 15%) of the threo diastereomer 31, and 1.003 g
(1.988 mmol, 67%) of a mixture of the erythro/threo diastereomers.
J ) 42.3 Hz); MS m/z 210 (M+, 1.5), 192, 151, 108, 91 (100), 79;
12
accurate mass 210.1088, calcd for
C
13CH15NO3 210.1085.
10
[1-13C]-3-[(Benzyloxycarbonyl)amino]propanal (26). A dry 250
mL round-bottomed flask equipped with a stir bar was charged with
CH2Cl2 (50 mL) and oxalyl chloride (2.1 mL, 24 mmol), and the
solution was cooled to -78 °C. DMSO (3.5 mL, 49 mmol) in CH2Cl2
(5 mL) and a solution of the labeled alcohol 25 (4.687 g, 22.29 mmol)
in CH2Cl2 (40 mL) were added. After stirring for an additional 15
min, triethylamine (15.5 mL, 111.4 mmol) was added during 7 min,
and stirring was continued for an additional 10 min, whereupon the
reaction was allowed to warm to room temperature over 45 min. Water
(100 mL) was added, and the layers were separated. The aqueous phase
was extracted with CH2Cl2 (100 mL), and the combined organic extracts
were washed with 1 N HCl (150 mL), 5% NaHCO3 (150 mL), and
brine (150 mL). After drying with anhydrous Na2SO4, the solution

Role of ClaVaminic Acid in the Pathway to ClaVulanic Acid
J. Am. Chem. Soc., Vol. 119, No. 10, 1997 2355
The latter mixture of diastereomers 30/31 (1.003 g, 1.988 mmol) in
CH2Cl2 (40 mL) was treated with DBN (235 µL, 1.90 mmol), and the
solution was stirred for 1 h. After passage through silica gel, the filtrate
was concentrated in Vacuo, and the residue was purified by radial
chromatography (2 mm silica gel; petroleum ether:ethyl acetate (4:1
to 1:1) to give 457.3 mg (0.9064 mmol, 46%) of the threo diastereomer
31 and 245.0 mg (0.4856 mmol, 24%) of a mixture of the erythro/
threo diastereomers. The threo diastereomer 31 gave the following
spectral data: 1H NMR (CDCl3) δ 7.33 (m, 15H, ArH), 6.93 (s, 1H,
CHAr2), 5.10 (m, 3H, CH2Ar and NH), 4.59 (m, 0.5H, H-3), 4.41 (m,
1.5H, OH and H-2), 4.07 (m, 0.5H, H-3), 3.98 (m, 0.5H, H-2), 3.48
(m, 1H, H-4′), 3.40 (m, 2H, H-5), 3.28 (m, 1H, H-4′), 2.99 (t, J ) 4.2
Hz, 2H, H-3′), 1.68 (m, 2H, H-4); 13C{1H} NMR (CDCl3) δ 68.9 (d,
J ) 37.9 Hz), 61.9 (d, J ) 38.0 Hz). Unlabeled material prepared in
an analogous manner gave the following spectral data: IR (CHCl3)
3448, 3010, 2955, 1725 (br), 1710, 1514, 1503, 1408, 1231, 1181, 699
1H, H-5), 4.82 (t, 1H, H-8), 4.76 (bs, 1H, H-3), 3.67 (m, 2H, H-9),
3.57 (dd, J ) 2.7, 17.1 Hz, 1H, H-6R), 3.15 (d, J ) 17.1 Hz, 1H,
H-6â); 13C{1H} NMR (D2O) δ 181.5, 175.6, 161.1, 92.3, 90.8, 74.7,
66.0, 48.2.
Growth of S. antibioticus for Feeding Experiments. Seed
Medium. D,L-Methionine (0.4 g), D-glucose (4.0 g), yeast extract (4.0
g), and malt extract (10 g) were dissolved in 1 L of tap H2O (pH 7.3).
A 500 mL Erlenmeyer flask containing 100 mL of seed medium was
inoculated using sterile techniques with 200 µL of a spore suspension
of S. antibioticus (Tu¨ 1718) prepared as described by Wanning.47 Once
inoculated, the seed medium was grown for 48 h at 27 °C with shaking
(300 rpm).
Modified Soy Fermentation Medium. In 1 L of tap H2O, 3.75
mmol each of L-alanine, L-arginine, L-aspartic acid, L-glutamate, and
L-serine were dissolved along with 30 g of D-mannitol (pH 7.2). Ten
500 mL Erlenmeyer flasks each containing 2 g of freshly ground soy
beans and 100 mL of fermentation medium were inoculated with a
1% inoculum of the 48 h seed culture.
Administration of [2,3-13C2]Clavaminic Acid (32) to S. antibioti-
cus. [2,3-13C2]Clavaminic acid (32; 26 mg) was administered to S.
antibioticus cultures (7 × 100 mL in 500 mL Erlenmeyer shake flasks)
as a filtration-sterilized, aqueous solution (pH 7.0) in pulsed feedings
of 0.1 mmol/L aliquots at 56 and 65 h. At 121 h, doubly 13C-labeled
valclavam (19) and 2-(2-hydroxyethyl)clavam (20) were isolated as
described below.
cm-1 13C{1H} NMR (CDCl3) δ 169.3, 167.8, 157.2, 139.3, 136.3,
;
128.6, 128.5, 128.3, 128.2, 128.1, 127.2, 127.0, 78.6, 68.9, 66.9, 62.1,
40.4, 37.8, 36.4, 34.5; accurate mass (MH+) 503.2179, calcd for
C29H31N2O6 503.2182.
In other experiments the erythro and threo diastereomers 30 and 31
were isolated as oils, which solidified on standing under vacuum.
Recrystallization from EtOAc/hexanes gave white crystalline solids,
erythro mp 97-99 °C and threo mp 119-121 °C.
(()-[2,3-13C2]Proclavaminic Acid (16). Protected threo-procla-
vaminate 31 (1.04 g, 2.06 mmol) was dissolved in a solution of ethanol
H2O (40 mL, 7:1) and 10% Pd/C (200 mg) was added. The mixture
was degassed, and H2 was bubbled through for 30 min. The solution
was passed through a 0.22 µm filter, washing with water, the ethanol
was removed in Vacuo, and the aqueous solution lyophilized to give
315.0 mg (1.540 mmol, 75%) of the product 16 as a fluffy, white
solid: 1H NMR (D2O/acetone) δ 4.52 (dt, J ) 4.5, 9.0 Hz, 0.5H, H-3),
4.46 (t, J ) 4.4 Hz, 0.5H, H-2), 4.01 (m, 1H, 0.5 H-2 and 0.5 H-3),
3.64 (m, 1H, H-4′), 3.55 (m, 1H, H-4′), 3.18 (m, 2H, H-5), 3.05 (t, J
) 3.9 Hz, 2H, H-3′), 1.88 (m, 2H, H-4); 13C{1H} NMR (D2O/dioxane)
δ 68.6 (d, J ) 39.3 Hz), 60.2 (d, J ) 39.3 Hz). Physical and spectral
data of unlabeled material prepared in an analogous manner were as
previously described.17
Isolation and Purification of Valclavam (7/19) and 2-(2-Hydroxy-
ethyl)clavam (8/20). Clavam isolation was carried out according to
the procedure of Wanning with some modifications.47 The fermentation
broth was harvested by centrifugation, and the supernatant was passed
through cheesecloth, filtered through a Celite pad, and brought to 2%
w/v in NaCl. The solution was loaded onto an Amberlite XAD-4
column (800 mL) and the column rinsed with 1 L of H2O. Desorption
of valclavam (7/19) was achieved by elution with 1 L of a 20% aqueous
methanol solution. The column was then eluted with 1 L of methanol
to collect 2-(2-hydroxyethyl)clavam (8/20).
After concentration of the first of these by rotary evaporation (without
heat), the aqueous residue was passed through a Sephadex-DEAE A-25
column (100 mL). The eluent was frozen and lyophilized to provide
crude valclavam as an orange powder. Further purification of valclavam
(7/19) was achieved by reversed-phase HPLC (Whatman Partisil ODS-
3, 10 µm, 9.4 × 250 mm; mobile phase H2O). Following concentration
of the 100% methanol fraction by rotary evaporation (without heat),
the remaining aqueous residue (60 mL) was diluted with an equal
volume of water (60 mL) and extracted with EtOAc (3 × 100 mL).
Sodium chloride was used to disrupt the emulsion which normally
formed between the two layers. The EtOAc extracts were combined,
dried over anhydrous Na2SO4, and concentrated in Vacuo to provide a
brown oil. The oil was further purified by flash chromatography using
a stepwise gradient of 50:50 to 25:75 hexane/EtOAc to 100% EtOAc.
The product was identified using TLC (50:50 EtOAc/hexane) where
the clavam of interest eluted with an Rf of 0.19. After product fractions
were pooled and concentrated in Vacuo, the product was obtained as a
clear, pale amber oil.
Conversion of (()-[2,3-13C2]Proclavaminic Acid (16) to (-)-[2,3-
13C2]Clavaminic Acid (32). A 2 L Erlenmeyer flask was equipped
with a stir bar containing 1.5 L of the following: 50 mM MOPS (pH
7), 0.5 mM DTT, 0.1 mM ascorbic acid, 25 µM ferrous ammonium
sulfate, 1.0 mM [2,3-13C2]-D,L-proclavaminic acid (16, 1.53 mmol), and
0.1 mM R-KG (0.15 mmol). The reaction was initiated by the addition
of 2.5 mL of a partially-pure CS2 solution (equivalent to 100 mg/mL
pure CS2) and allowed to proceed at room temperature for 9 h with
constant stirring. Throughout the entire reaction, 0.15 mmol of R-KG
(306 µL of a 500 mM stock solution) was added every 20 min.
Additional 2.5 mL aliquots of partially-pure CS2 (ca. 250 mg of pure
enzyme) were added at 1, 2, and 3 h. Aliquots of 200 µL were removed
hourly and added to Eppendorf microcentrifuge tubes containing 10
µL of a 4 mM solution of EDTA. Samples were frozen in liquid N2
until the reaction was complete. These samples were then thawed, and
derivatized with imidazole, and the A312 was read in the usual manner
in order to monitor the extent of reaction which was determined to be
71%. The theoretical yield was calculated using the amount of
L-enantiomer available.
Purification of [2,3-13C2]-(3S,5S)-Clavaminic Acid (32). The final
reaction mixture was loaded onto a DEAE Sephadex column, pH 6
(2.5 × 41 cm, 200 mL) which was subsequently washed with 2 column
volumes of double distilled H2O. The eluent was frozen and partially
lyophilized overnight. The high concentration of MOPS hampered
lyophilization and caused the concentrated sample to thaw. The final
protein removal step employed an Amicon ultrafiltration cell fitted with
a PM 10 membrane. The filtrate was then frozen and lyophilized
overnight to approximately 5 mL of a pink, viscous solution, which
was then purified by HPLC [Whatman Partisil ODS 3, 10 µm (22 ×
250 mm), mobile phase H2O (10 mL/min), monitored at λ ) 220 nm].
The entire sample was purified in this manner (200 µL/injection) with
a tr of 8.5 min for clavaminate and tr of 6.0 min for proclavaminate.
The product contained approximately 5% of the MOPS buffer present
in the initial reaction mixture: 1H NMR (D2O) δ 5.77 (d, J ) 2.7 Hz,
13C{1H} NMR analysis of valclavam (19) and 2-(2-hydroxyethyl)-
clavam (20) revealed intact incorporation of the 13C-labels at C-2 and
C-3 in both clavams with an efficiency of 0.2% (2JCC ) 32.6 Hz).
Acknowledgment. We are grateful to the National Institutes
of Health for financial support of this research (Grant AI14937)
and to the NIH (NMR: Grants RR 01934, RR 04794, and RR
06468; MS: Grant RR 02318) and NSF (NMR: Grant PCM
83-03176) for major funding to acquire the analytical instru-
mentation used.
JA963107O
(47) Wanning, M. Ph.D. Thesis, University of Tu¨bingen, Tu¨bingen,
Germany, 1980.