The biosynthesis of pramanicin in Stagonospom sp. ATCC 74235: A modified acyltetramic acid

Article abstract of DOI:10.1039/b006007k

Biosynthetic incorporations of acetate, malonate and serine precursors which had been isotopically labelled with ²H, ¹³C, ¹⁵N and ¹⁸O into pramanicin 1 in Stagonospora sp. ATCC 74235 were demonstrated. Intact incorporation of a starter acetate and six extender malonates generates the acyclic, hydrophobic tail. A further intact acetate, in preference to malonate, and a serine entity which is incorporated only as the L-enantiomer and with the O=C-CH(N)-CH₂ entity intact, provide the pyrrolidone ring and hydroxymethyl group of 1. The results are fully consistent with a biosynthetic pathway involving an acyltetramic acid (2). The olefinic precursor 3 of the epoxide in 1 is described, and is also shown to co-occur in the cultures. The ratio of 1:3 can be controlled by addition of precursors. The Royal Society of Chemistry 2000.

Full text of DOI:10.1039/b006007k

View Article Online / Journal Homepage / Table of Contents for this issue
The biosynthesis of pramanicin in Stagonospora sp. ATCC 74235:
a modified acyltetramic acid†
Paul H. M. Harrison,* Petar A. Duspara, Stephen I. Jenkins, Salima A. Kassam,
David K. Liscombe and Donald W. Hughes
1
Department of Chemistry, McMaster University, 1280 Main Street West, Hamilton, Ontario,
L8S 4M1, Canada
Received (in Cambridge, UK) 25th July 2000, Accepted 28th September 2000
First published as an Advance Article on the web 22nd November 2000
Biosynthetic incorporations of acetate, malonate and serine precursors which had been isotopically labelled with
2H, 13C, 15N and 18O into pramanicin 1 in Stagonospora sp. ATCC 74235 were demonstrated. Intact incorporation
of a starter acetate and six extender malonates generates the acyclic, hydrophobic tail. A further intact acetate,
in preference to malonate, and a serine entity which is incorporated only as the -enantiomer and with the
O᎐C–CH(N)–CH entity intact, provide the pyrrolidone ring and hydroxymethyl group of 1. The results are fully
᎐
2
consistent with a biosynthetic pathway involving an acyltetramic acid (2). The olefinic precursor 3 of the epoxide
in 1 is described, and is also shown to co-occur in the cultures. The ratio of 1:3 can be controlled by addition of
precursors.
Introduction
Pramanicin 1 is a natural product first isolated by bioassay-
directed fractionation in 1994 by Schwartz et al. from an
unidentified fungus which was subsequently shown to be a
member of the genus Stagonospora.1 The connectivity in the
structure was determined by spectroscopic analysis, and the
proton and carbon-13 NMR resonances were carefully
assigned by one- and two-dimensional NMR techniques.
Further, the relative configuration within the pyrrolidone ring
was established, and it was shown that the substituents on the
epoxide were in the trans orientation. However, none of the
absolute configurations was established. Pramanicin was shown
to exhibit modest antifungal activity against a range of
microbes including the human opportunistic pathogen Crypto-
coccus neoformans, the causative agent of cryptococcal menin-
gitis. Notably, pramanicin is active against both the acapsular
and capsular forms. Recently we have shown that pramanicin
also exhibits intriguing biological effects on vascular endo-
thelial cells, and have suggested that it may act by increasing
calcium permeability in the cell.2
We have published two preliminary reports on our studies of
further incorporation experiments designed to test this bio-
the biosynthesis of 1 in Stagonospora sp. ATCC 74235, which
have shown that the carbon skeleton originates from eight acet-
ate units and one intact -serine entity.3,4 These results suggest
that 1 is formed via an acyltetramic acid derivative (e.g., 2)
which is subsequently modified by oxidation at C-3, keto-
reduction at C-4, and formation of the epoxide from the olefin
at C-10/C-11, in an as-yet-undetermined order, to furnish the
natural product. The acyltetramic acids5 are an intriguing group
of natural products that includes, for example, cyclopiazonic
acid, and which are formed biosynthetically by condensation of
acetate units with an amino acid.6–14 Incorporation of -serine
with retention of all four bonds to the α-carbon showed that
1 has the corresponding 5S absolute configuration, a con-
clusion also reached by Barrett et al. who recently synthesised
the enantiomer of 1, thus also confirming the absolute con-
figuration of the epoxide moiety to be as depicted.15
synthetic hypothesis. In particular, this postulate has impli-
cations for the biogenesis of the oxygen atoms. These new
experiments demonstrate an oxygen-labelling pattern which is
fully consistent with the intermediacy of an acyltetramic acid,
and which excludes some reasonable alternative biosynthetic
sequences. Data to differentiate between various mechanisms
of pyrrolidone-ring formation are presented and discussed.
The isolation and characterisation of co-metabolite 3 is also
described.
Results
A. Culture of Stagonospora sp. ATCC 74235, and isolation of
1 and 3
Pramanicin 1 was readily and consistently produced by cultures
of Stagonospora sp. ATCC 74235, essentially under the growth,
isolation and purification conditions reported by Schwartz
et al.1 We found that reduction of the glucose levels in the
medium, from 75 to 40 and even to 10 g dmϪ3, had essentially
Herein, we describe these earlier results in detail, as well as
† The IUPAC name for tetramic acid is 2,5-dihydro-4-hydroxypyrrol-
2-one.
4390
J. Chem. Soc., Perkin Trans. 1, 2000, 4390–4402
This journal is © The Royal Society of Chemistry 2000
DOI: 10.1039/b006007k

View Article Online
no adverse effect on production of 1, although this modifi-
cation improved the level of incorporation in biosynthetic
experiments (see below). During the final phase of purification,
which we found could be conducted conveniently by reversed-
phase MPLC on a C8 column, a small amount of a related
natural product (3) was frequently isolated, which was slightly
more strongly retained than was 1. Subsequent biosynthetic
experiments (see below) using N-acetyl-S-[1-13C]decanoylcyste-
amine 4 and the known β-oxidation inhibitor 3-(tetradecylthio)-
propanoic acid16 5 resulted in alteration of antibiotic produc-
tion so that smaller amounts of pramanicin were produced,
while the co-metabolite level was enhanced. HPLC analysis of
aliquots of culture medium from controls, as well as flasks to
which were added 4 in ethanol; 5; or ethanol alone, showed
that 4 and 5 each cause an increase in formation of the co-
metabolite, while neither completely abolished production of 1.
The combination of 4 and 5 essentially completely inhibited
formation of 1 under these conditions, resulting in 3 as the only
significant product. This HPLC result was confirmed by pre-
parative incubations, which resulted in the exclusive isolation
of 3.
Fig. 1 Summary of results from two-dimensional NMR experiments
which support structure 3: unambiguous correlations were determined
from A: the COSY spectrum, and B: the long-range 13C–1H hetero-
nuclear shift-correlated spectrum. One-bond C–H correlations are
given in Table 1.
Preliminary analysis of this compound by proton and
carbon-13 NMR immediately suggested structure 3, a reason-
able proposal on biosynthetic grounds. Compound 3 was
reported by the Merck group as the product of treatment of
pramanicin with potassium selenocyanate,1 a reagent known to
convert epoxides stereospecifically to olefins.17 However, the
spectral data reported for the product of this reaction did not
precisely match those for material from our cultures.
The natural material clearly possessed a long aliphatic chain,
and the polar pyrrolidone head group exhibited proton reson-
ances in the NMR spectrum at essentially identical shifts, and
with similar coupling constants to 1 (Table 1). The NMR
signals for the epoxide protons were missing, but there were
resonances for four olefinic protons as opposed to the two in
pramanicin: a doublet and a doublet of doublets, attributable
to H-8 and H-9 respectively, were still present, with J8–9 15.1 Hz
(cf. 15.6 Hz in 1), supporting trans (E) stereochemistry for this
olefin. The remaining two olefinic protons (H-10 and H-11)
formed a strongly coupled overlapping multiplet in methanol-
d4, a result that seemed different to that reported by the Merck
group. It was thus not immediately possible to assign stereo-
chemistry to this second double bond. Therefore it was feasible
that the organism had produced an isomer of 3, e.g. the 10,11-
cis (Z) olefin, perhaps as a shunt metabolite, or that the KSeCN
treatment of 1 in the previously reported work had caused some
undetected effect above and beyond the expected deoxygen-
ation of the epoxide (e.g., production of cis olefin). We thus
sought further confirmation of the structure.
One-dimensional proton and carbon-13 spectra were
recorded at high resolution in methanol-d4 (Table 1), as well as
in DMSO-d6; the latter solvent revealed the presence of an NH
proton as well as protons of OH groups bonded to a primary, a
secondary and a tertiary centre. Next followed proton COSY
and inverse-detected one- and multiple-bond proton–carbon
correlated spectroscopy; the results of these experiments are
summarised in Fig. 1 and Table 1. Although all of the data were
completely consistent with structure 3, these results were still
inconclusive with respect to double-bond geometry. Several
solvents were evaluated in order to enhance the resolution
of the H-10 and H-11 protons and to measure the coupling
constant between them. The largest shift difference was found
in acetonitrile-d3, but strong coupling effects even at 500
MHz were still present. Spectral simulation allowed a good fit
for this system (Fig. 2), provided that the H-10–H-11 coupling
constant was ≈15 Hz, supporting a trans geometry at this
double bond.
Fig. 2 Experimental and calculated proton NMR spectra of natural
product 3 for H-10 and H-11. The parameters for the calculated
spectrum are ∆δ(10–11) 0.041 ppm and (in Hertz): J12a-12b Ϫ13; J11-12a
=
J11-12b = 7.0; J10-11 15.1; J11-9 Ϫ1; J10-9 10. Trace impurities are seen in the
experimental spectrum.
with identical retention time on HPLC to 3 produced by fer-
mentation. In contrast, we found that treatment of pramanicin
with NaI and toluene-p-sulfonic acid according to the pro-
cedure of Baruah et al. which is also known to retain the con-
figuration of the epoxide substituents,18 provided 3 in good
yield (72%). This material was identical in all respects (HPLC,
proton and carbon-13 NMR, MS) with the fermentation
product. These experiments established the structure of the
co-metabolite to be 3; however, we remain unable to explain the
reported results with potassium selenocyanate.
B. Labelling of acetate-derived sites
Preliminary experiments investigated the incorporation of
sodium [1-13C]acetate into pramanicin in cultures of Stagono-
spora sp. ATCC 74235 essentially under the growth conditions
reported by Schwartz et al.,1 with precursor addition on days
5–8 and isolation on day 9. The resulting 1 showed small (≈0.3–
0.4% siteϪ1) but significant incorporation of label into C-2 and
seven other sites within the acyl side-chain (Scheme 1). In order
to improve the specific incorporation rate, a systematic vari-
ation of both the time of precursor addition and the glucose
concentration was undertaken. These experiments showed that
reduction of glucose from 70 to 40 g dmϪ3, along with earlier
initiation of precursor feeding (days 3–6), improved incorpor-
ation substantially. The combination of these variations
resulted in pramanicin labelled to the extent of 1.7 to 2.3% per
site (Table 1). Label was located at each of the sites C-2, C-7,
C-9, C-11, C-13 and C-19, which had been unambiguously
assigned in the carbon-13 NMR spectrum of 1 (Scheme 1).
Chemical evidence also favoured structure 3. In our hands,
the selenocyanate reaction of 1 described by the Merck group
proved inefficient, but did furnish a small amount of product
J. Chem. Soc., Perkin Trans. 1, 2000, 4390–4402
4391

View Article Online
Table 1 Carbon-13 chemical shifts for pramanicin 1 and dienone 3; proton NMR spectral data for 3; and incorporations for precursors into 1
C #a
δ(13C) 1b (ppm)
δ(13C) 3b (ppm)
δ(1H) 3b (ppm)
Mult 3c (J in Hz)
Fedd
Incorp(s)e
7
2
9
8
3
4
11
6
5
10
12
18
14–17
14–17
14–17
14–17
13
19
20
197.91
174.93
145.13
127.88
88.10
78.99
62.90
62.12
60.33
57.78
33.02
33.02k
30.61
30.61
30.48
30.39
26.92
23.69
14.38
198.70
175.26
145.53
124.25
88.07
79.01
148.58i
62.20
60.39
130.48i
34.20
33.02
30.63
30.53
30.40
30.29
29.82
23.69
14.40
A
A
A
B
B
C; D
A
D; E
D
B
2.3
1.9
2.2
2.2
7.27
6.72
ddm (15,10)
d (15)
1.2
4.14
d (7)
m
5.2; 8f
2.3
6.2–6.3j
3.80, 3.55
3.51
dd (11,2.5), dd (11,5.5)
ddd (7,5.5,2.5)
m
m
m
m
m
m
8f; 7.5g
8f
6.2–6.3j
2.19–2.22
1.29
2.4
B
2.0h
1.29
1.29
1.29
1.29
1.44
1.29
0.89
A; B
1.5h; 1.2h
B
A
A
A
B
1.9
1.8
2.1
1.8
1.1
m
quintet (7.5)
m
t (7)
Notes: a The numbering system is based on Ref. 1. b Determined in CD3OD. c Multiplicity (standard symbols) and coupling constant(s) in Hertz.
d Precursors for 1: A: sodium [1-13C]acetate, B: diethyl [2-13C]malonate, C: -[1-13C]serine, D: -[U-13C,15N]serine, E: -[3-13C,2-2H]serine. e Percent-
age incorporation into 1 is expressed as number of fold increase in the height of the carbon-13 resonance relative to unlabelled 1, with correction by
standardising the two samples using the mean peak height for all unlabelled carbons. f Determined by dividing the sum of the coupled peak heights
by the height of the uncoupled, natural-abundance peak. g Determined by adding enrichments at both the unshifted and isotopically shifted sites.
h Values averaged over two overlapping carbons. i Carbon-13 assignments for C-10 and C-11 derived from the observed labelling pattern.
j In methanol, these resonances form a complex, strongly coupled system; see text for assignments in acetonitrile. k Assignment by analogy with
compound 3.
Fig. 3 Sections of the carbon-13 NMR spectrum of pramanicin 1
derived from sodium [1,2-13C2]acetate. Left side: the resonance at δC 33,
Scheme 1 Overview of results of biosynthetic incorporation experi-
corresponding to C-12 and C-18, exhibits a central, unenriched peak
surrounded by two doublets with different coupling constants, corre-
ments with labelled acetate and serine samples into pramanicin 1
sponding to intact incorporation at both of the different sites. The small
peaks at the extreme left and right are due to molecules which contain
interunit couplings from adjacent acetate units which are both
enriched; this is due to the pulsed feeding protocol. The level of
incorporation is 10% of that for signals from intraunit coupling only.
Right side: the resonances at δC 30.3–30.6, corresponding to carbons
from C-14 to C-18, form a complex pattern due to strong coupling. The
base peak at δC 30.6 is strongly enhanced, despite the absence of single
label at any other site, and this must thus be due to suppression of the
outside line of the satellites. In addition, natural-abundance peaks at
δC 30.5 and 30.4 are each surrounded by weakly coupled doublets
(see text).
Within the region δC 30.3–33, where the five as-yet-unassigned
nuclei C-14 to C-18 resonate, the resonance at δC 30.39 was
enhanced to the same extent as those above. The signal at
δC 30.61, which results from overlap of resonances from two
carbon atoms, was enhanced 1.5-fold, indicative that one of
these two sites was labelled to the same extent as above while
the other site was not labelled. The signal at δC 33.02, which
again results from two carbon sites, was not enhanced, so
neither of these positions was labelled. Together, the results
suggest that 1 is labelled at eight positions from sodium
[1-13C]acetate.
Sodium [1,2-13C2]acetate was next incorporated into the
Stagonospora cultures, giving labelled 1. Natural-abundance
singlet carbon resonances corresponding to each of the atoms
C-2, C-3 and C-7 to C-19 were surrounded by doublet signals
due to intact incorporation of acetate units (Fig. 3). A combin-
ation of high-resolution one-dimensional carbon-13 NMR and
two-dimensional inverse-detected gradient INADEQUATE
spectroscopy was used to demonstrate which pairs of atoms
were coupled to each other. The results showed unambiguously
that the four adjacent pairs of atoms C-2 and C-3, C-7 and C-8,
C-9 and C-10, and C-19 and C-20 were coupled (Scheme 1).
Three further coupled pairs of carbon atoms were observed in
the INADEQUATE spectrum. The assigned resonance for
C-11 was coupled with a one-bond coupling constant of 43.7
Hz to one of the two overlapping signals at δC 33 which must
therefore be C-12 and thus C-11–C-12 is an intact acetate unit.
The unambiguously assigned C-13 resonance (δC 26.9) was
coupled to one of the signals at δC 30.3–30.6 (1J 34.7 Hz), which
must therefore include C-14 and thus C-13–C-14 is an intact
acetate unit, even though C-14 cannot be unambiguously
assigned. The third correlation that was observed in the
INADEQUATE spectrum was between the second signal
at δC 33, which possessed a one-bond coupling constant (34.7
Hz) that was different from the first signal (Fig. 3), and which
4392
J. Chem. Soc., Perkin Trans. 1, 2000, 4390–4402

View Article Online
method.19 In the high-resolution carbon-13 NMR spectrum of
the resulting 1, isotopically shifted signals were observed at C-2
(∆δ 0.031 ppm) and C-7 (∆δ 0.047 ppm).19,20 These signals were
low in intensity (0.7 and 0.3% incorporation, respectively),
while the unshifted signals were enhanced (2.9% incorporation
at both sites), implying that considerable exchange of 18O
for 16O had occurred within the culture. This exchange is a
common phenomenon during biosynthetic studies of poly-
ketides.19,22
Diethyl [2-13C]malonate was then incorporated into 1 in
cultures of Stagonospora. Label was found at each of the sites
derived from C-2 of the eight intact acetate units. The extent of
incorporation ranged from 1.9 to 2.4% at C-8 and C-10, and by
inference at sites C-12 and -18 (based on the average incorpor-
ation at the two overlapping resonances at δC 33.02) as well as at
C-14 and -16 (based on the average incorporation at the two
overlapping resonances at δC 30.61, and the incorporation at the
single, but unassigned, site at δC 30.48). By contrast, incorpor-
ation levels were lower at C-20 (1.1%) and at C-3 (1.2%). Thus
the level of incorporation at the starter acetate unit (C-20) and
the terminating unit (C-3) was significantly different from that
at the other, malonate-derived extender units (C-8 to C-18).
Since extension of the starter acetate unit by four malonate
units is expected to lead to an intermediary decanoyl moiety,
sodium [1-13C]decanoate was next tested for incorporation
into 1. Although the resulting 1 was rigorously purified by the
standard procedure, the product was contaminated with a small
amount of [1-13C]decanoic acid, which co-chromatographs with
the target material. The carbon-13 NMR spectrum of this
sample showed essentially no incorporation of carbon-13 into 1
(<0.2%), either at the site expected for specific incorporation
(C-11) or at any of the sites corresponding to labelling by
β-oxidation of decanoate (i.e., those from C-1 of acetate). The
recovered decanoic acid was shown to be fully labelled with
carbon-13: within the limits of detection, all of the protons at
the α-carbon were coupled to carbon-13, and the resonance due
to C-α was a doublet without a central, uncoupled singlet.
Labelled decanoic acid was next converted to the N-acetyl-
cysteamine derivative 4 in order to improve incorporation,
following a standard protocol for conversion of fatty acids to
their SNAC thioesters (Scheme 2).23
Fig. 4 Sections of the carbon-13 NMR spectrum of pramanicin 1
derived from sodium [1-13C,2-2H3]acetate. Left side: the resonance for
C-9 exhibits a peak from a β-deuterium isotope shift, ∆δ 0.088 ppm,
due to deuterium at C-10. Right side: the resonance for C-19 exhibits
three isotopically shifted signals, ∆δ 0.088 ppm per D atom, due to the
presence of one, two and three deuterium atoms at C-20.
was coupled to one of the remaining signals at δC 30.3–30.6;
therefore there is another intact acetate unit in the region
between C-15 and C-18. No correlation was observed in the
INADEQUATE spectrum for the coupling of the two remain-
ing labelled carbons in 1 in this experiment, because these two
resonances have very similar chemical shifts (δC 30.4–30.6) and
are therefore strongly coupled. In this case analysis of the one-
dimensional carbon spectrum revealed this strong coupling of
the two carbon atoms, confirming the presence of one further
intact acetate unit between C-15 and C-18 (Fig. 3). A total of
eight intact acetate units was thus found in 1.
For each resonance for C-3, and C-7 to C-19 in this double-
labelled acetate experiment, small doublets of doublets were
observed in the 1D-carbon NMR spectrum, surrounding the
coupled signals (Fig. 3); these are sites where incorporation of
two labelled acetate molecules into two adjacent units has
occurred. These satellite signals were missing at the C-2 and
C-20 positions, as expected since these positions represent the
terminus and start, respectively, of the presumed polyketide
chain. The level of labelling for incorporation of two adjacent
units was ≈10% of that for labeling of a single unit, for every
pair. Such multiple incorporations at adjacent units are a com-
mon phenomenon in biosynthetic studies of polyketides such as
this, and are the result of the pulsed feeding of acetate.19
Sodium [1-13C,2-2H3]acetate was incorporated into 1. Anal-
ysis of the carbon-13 NMR spectrum of the derived product
showed shifted signals due to the presence of a β-deuterium
isotope shift19,20 at C-9 (∆δ 0.088 ppm), the effect of a single
deuterium at C-10; and at C-19, where three signals were
observed due to the presence of one, two and three deuterium
atoms at C-20 (∆δ 0.088 ppm per deuterium) (Fig. 4 and
Scheme 1). No significant isotopically shifted signals were
observed at any of the remaining sites which derive from C-1 of
acetate. However, this is probably the result of low incorpor-
ation of deuterium, signal overlap and/or small isotope shifts,20
because 2H-NMR of the sample revealed the presence of
deuterium in the methylene region and in the olefinic position
H-8, as well as in the methyl group as expected. Integration
showed that the sum of the levels of deuterium at methylene
sites 14, 16 and 18 was less than that at the unique olefinic site
H-8.
Scheme 2 Synthesis and results of incorporation of N-acetyl-S-
[1-13C]decanoylcysteamine 4 into dienone 3. Reagents: a, (i) EtOC-
(᎐O)Cl; (ii) HSCH2CH2NHAc (prepared in situ); b, β-oxidation
᎐
pathway in cells; c, incorporation by normal route.
Incorporation of 4 either alone or in combination with the
β-oxidation inhibitor 3-(tetradecylthio)propanoic acid
5
resulted in formation of 3 (see above). In the carbon-13 NMR
spectrum of the samples of 1 and 3 (from 4 alone) or of 3 (from
the combination), enhanced singlets were observed at each of
the sites that had been labelled by C-1 of acetate. The level of
incorporation was uniform for each experiment, ranging from
1.0 to 2.7% between experiments. No significant specific
incorporation of label was found at C-11 over and above that
observed at the other labelled sites, despite attempts to obtain
specific incorporation under a range of feeding protocols.
Next sodium [1-13C,18O2]acetate was prepared by the method
of Cane et al..21 Analysis by both mass spectrometry and
high-resolution carbon-13 NMR demonstrated an isotopic
incorporation of 99 and 70 atom% 13C and 18O per site, respect-
ively. This material was then used to determine the origin of
the oxygen atoms in 1, using the 18O-induced isotope-shift
J. Chem. Soc., Perkin Trans. 1, 2000, 4390–4402
4393

View Article Online
C. Incorporation into the serine entity
Since the results with acetate suggested that the remaining three
carbon atoms and the nitrogen of the pyrrolidone ring might
originate from serine, the incorporation of this amino acid into
pramanicin 1 was investigated. First, label from -[1-13C]serine
was found to be located exclusively into the C-4 position of 1
with a high level of incorporation (5.2%) (Scheme 1). In order
to address the issue of which enantiomer was incorporated,
commercial -[2,3,3-2H3]serine was tested. The resulting sample
of 1 was examined by 2H{1H} NMR spectroscopy in methanol.
Two broad resonances were observed, one at δ 3.72 and a
slightly more intense signal at δ 3.48. The former could be
unambiguously assigned to one of the two diastereotopic
deuterons at C-6 (δH 3.79, 0.24 ppm from the closest adjacent
resonance), while the latter corresponded to the other deuteron
at C-6 (δH 3.55) and/or deuterium at C-5 (δH 3.47). These two
signals were not expected to be resolved due to the natural
lineshape in the deuterium spectrum (w1/2 ≈ 0.1 ppm). Examin-
ation of the proton NMR spectra of pramanicin in a number of
solvents showed that the largest shift difference between the
H-5 and the H-6 protons occurred in DMSO-d6. However,
when the labelled sample was examined by 2H NMR in DMSO,
substantially increased linewidths were observed, probably due
to the greater viscosity of DMSO compared with methanol.
When the sample was warmed to 85 ЊC, the thermal reduction
in viscosity sharpened the deuterium lines so that the two
diastereotopic deuterons at C-6 were well resolved; a shoulder
on the more shielded signal was observed, indicative of deuter-
ium at C-5 as well as at both C-6 positions.
Fig. 5 Section of the nitrogen-15 NMR spectrum of pramanicin 1
derived from [1,2,3-13C3,15N]serine, acquired in the INEPT mode at 50.7
MHz in DMSO-d6. Coupling to the attached proton is seen in the pair
of inverted signals; the smaller splitting is due to the serine-derived
attached carbon-13.
With this knowledge, commercial -[1,2,3-13C3,15N]serine was
incorporated into pramanicin. The resulting sample was exam-
ined by 15N NMR, using the INEPT procedure to enhance the
signal intensity through polarisation transfer from the NH
proton.1 To avoid exchange of the NH position, DMSO-d6 was
used as solvent for this experiment. The spectrum (Fig. 5)
shows the presence of a doublet of doublets at δ Ϫ256.4 relative
to external MeNO2. The chemical shift compares with that
reported by Schwartz and co-workers (126.2 ppm relative to 2.9
M 15NH4Cl in 1 M HCl; this converts to Ϫ229.1 via correlation
through the shift of liquid NH3).1 Thus, 15N from serine is
incorporated into the nitrogen atom of 1. The observed coup-
To increase the dispersion of the three signals of interest, the
pramanicin was derivatised. The trimethylacetyl (pivaloyl)
ester, a protective group which is known to selectively protect
primary alcohols,24 was chosen in order to react selectively at
C-6, and to provide a substantive shift to the H-6 protons.
When treated with pivalic anhydride in Et3N, pramanicin was
converted to the 6-pivaloyl derivative 6a (56%) (Scheme 3) with
lings are due to (1J
93 Hz) (lit.1 92 Hz), and a further
15N–H
one-bond coupling to carbon-13 (10 Hz). Within the limit of
detection, all of the nitrogen was coupled to carbon-13, so
serine is incorporated with the C–N bond intact. The same
C–N coupling was observed in the carbon-13 NMR spectrum
of the sample. In this case, the signal at C-5 was observed to be
composed of a doublet of triplets (Jd 10 Hz to 15N, and 38 Hz),
and a doublet of doublets (J 10 Hz to 15N, and 38 Hz) surround-
ing the unenhanced natural-abundance signal. At C-4, the
natural-abundance singlet was surrounded by a doublet (38
Hz), while the resonance for C-6 was comprised of an enhanced
singlet surrounded by a doublet of lower intensity than that
at C-4.‡ These results show that 1 derived from the labelled
serine is comprised of three isotopomers: the fully labelled
[4,5,6-13C3,15N]isotopomer, as well as [4,5-13C2,15N]- and [6-13C]-
species.
Scheme 3 Incorporation of -[2,3,3-2H3]serine into pramanicin 1, and
preparation of the 6-pivaloyl derivative 6a. Reagent and conditions:
a, pivalic anhydride, TEA, 70 ЊC, 30 min.
only small amounts of other products, presumably correspond-
ing to reaction at the other alcohol groups, which could be
readily removed by chromatography. The proton NMR
spectrum of 6a exhibited all of the resonances expected for
pramanicin, except that the two H-6 protons had shifted to
δ 4.37 and 4.05, while H-5 was only slightly influenced by the
neighbouring pivaloyl group at C-6, shifting to δ 3.62. The
material derived from labelled serine was similarly converted to
6b, and the purified product was examined by deuterium NMR
spectroscopy in methanol. Three broad resonances were
observed at δ 4.30, 4.00 and 3.55, establishing that deuterium
is present at C-5 and at both positions at C-6.‡ Thus, the
-enantiomer of serine is a viable precursor to 1, although no
conclusion can be reached about the role of the -enantiomer
in the pathway at this point.
The incorporation of both - and -[2-2H,3-13C]serine was
next examined. The separate enantiomers were prepared by
exchange of commercial racemic [3-13C]serine (92 atom% 13C),
using pyridoxal and aluminium sulfate in D2O according to the
procedure initially developed for [2-2H]serine by Miles and
McPhie (Scheme 4).25 Modifications to this procedure, as
described by Townsend et al.26 and later by de Kroon et al.27
for [2-3H,1-14C]serine, were applied to enhance the yield of
recovered serine. The resulting racemic [2-2H,3-13C]serine 7a
was resolved by the method of Velluz et al.,28 as modified by
Gorissen et al.29 Thus, after conversion of the racemic [2-2H,
3-13C]serine to the 3,5-dinitrobenzoyl (3,5-DNB) amide deriv-
ative 8, the -enantiomer was preferentially crystallised as the
(Ϫ)chloramphenicol salt 9. The free acid was liberated from this
salt, and then the 3,5-DNB group was removed with aq. HCl to
‡ These spectra have been published in preliminary communications3,4
and are not reproduced here.
4394
J. Chem. Soc., Perkin Trans. 1, 2000, 4390–4402

View Article Online
Fig. 6 Partial carbon-13 NMR spectrum of pramanicin 1 derived
from [1-13C,18O2]serine, showing the isotopically shifted signal due to
intact incorporation of the C–O bond at C-4.
Scheme 4 Reagents and conditions: a, Pyridoxal, Al2(SO4)3, D2O, pD
9.4; b, 3,5-DNBCl, aq. NaOH; c, -(Ϫ)-threo-chloramphenicol (free
base), MeOH–EtOAc; d, (i) conc. HCl, (ii) aq. HCl, reflux; e, (i) extract
with HCl, (ii) quinidine, EtOH. Overall yields: 14% for 7b and 15%
for 7c.
case, the distribution of isotopomers between C-4 and C-6
(26:74) reflects the distribution in the -component of the start-
ing serine (21:79) within experimental error. At C-6, the size of
the shift for the upfield signal corresponds to that expected for a
β-deuterium isotope shift,20 indicating that an intact 13C–C–2H
unit from -serine was incorporated into 1. However, some
carbon-13 had been incorporated without the accompanying
deuterium.
give -[2-2H,3-13C]serine 7b in 14% overall yield. The material
was confirmed to be fully labelled with deuterium, and 92%
labelled with carbon-13, by NMR and electrospray MS. Optical
rotation confirmed the  configuration, while chiral TLC
according to the method of Günther30 showed the absence of
any detectable -enantiomer. The mother liquor from the
chloramphenicol crystallisation (above) was converted to the
acid, and the -enantiomer crystallised as the salt 10 with
(ϩ)-quinidine. The salt was acidified, and the 3,5-DNB group
deprotected to afford the desired -[2-2H,3-13C]serine 7c. This
material was also fully labelled with deuterium, but had the
opposite optical rotation to that of the  isomer. It migrated as
a single spot with higher Rf than the  isomer on the chiral TLC
system, in accord with the relative migrations reported in the
literature for the enantiomers of serine.30
Next, the origin of the hydroxylic oxygen at C-4 of pram-
anicin was investigated. Racemic [1-13C,1-18O]serine 11 was
prepared from -[1-13C]serine by exchange with H218O–HCl,
using the method developed for [1-18O2]serine by Murphy and
Clay;31 however, only 2 equiv. of H218O and conc. HCl in H216O
was used. After neutralisation, a mixture of serine hydro-
chloride and NaCl was obtained which was carried forward
without further purification for incorporation. The labelled
serine contained no organic impurities (by NMR). These modi-
fications conserved isotopes, and ensured a 13C–16O-labelled
reference peak in the 13C NMR spectrum of the derived 1, thus
avoiding swamping of the natural abundance peak due to the
high level of serine incorporation expected in this experiment.19
The - and -enantiomers of [2-2H,3-13C]serine were each
separately mixed with racemic [1-13C]serine as an internal
positive control for incorporation into C-4 of pramanicin. The
samples of serine were analysed by integration in the proton
NMR spectrum of the serine methylene signals that were
coupled to carbon-13 at C-3, vs. the uncoupled signals. The
latter derives from the added racemic [1-13C]serine, plus a small
amount of [2-2H]serine which had originated from the com-
mercial starting material, i.e. the racemic [3-13C]serine which
was purchased contained 8% 12C. The results showed that the
sample from the -enantiomer contained 68% [2-2H,3-13C]-, 6%
[2-2H]-, and 13% of each enantiomer of [1-13C]serine, in close
agreement to the amounts calculated by mass during mixing of
the isotopomers (23% [1-13C]). The corresponding values for the
-enantiomer were 62, 6, 16 and 16%, respectively.
The isotopomer distribution was shown to be approx. 99% 13
C
and ≈50:50 16O:18O per site by high-resolution 13C NMR
spectroscopy. When pramanicin derived from this material was
examined, the high-resolution 13C NMR spectrum showed an
enhanced singlet at the shift of C-4 (3.8% incorp.), as well as a
signal shifted upfield by ∆δ 0.013 ppm20 due to the presence of
an oxygen-18-induced isotope shift19 (2.2% incorp.) (Fig. 6).
Therefore, the results show that a significant proportion of
pramanicin molecules are formed from serine with the C–O
bond at C-4 of 1 remaining intact from the C᎐O bond of serine.
᎐
The remainder have undergone exchange of 18O for 16O.22
Discussion
These two samples of serine were separately incorporated
into pramanicin 1, which was then analysed by carbon-13
NMR spectroscopy. For the sample from -serine, the control
site was labelled (1.7% incorporation at C-4) but no enhance-
ment within experimental error, nor an isotopically shifted
signal, was observed at C-6. Therefore, only the -enantiomer
of serine is incorporated. For the -enantiomer of serine, con-
trol site C-4 was also enhanced to essentially the same level as in
the -serine experiment (2.1% incorporation), while C-6 was
both enhanced (7.6% incorporation) and furnished a signal
shifted upfield by 0.074 ppm (2.9% incorporation).‡,19 In this
Like many fatty acids32 and polyketides,33 the assembly of
pramanicin begins with a two-carbon starter unit at C-20 and
C-19 (Scheme 5). The methyl ‘tail’ of acetate provides C-20,
while the carboxylate carbon ‘head’ furnishes C-19, as shown
by the labelling experiments with [1-13C]acetate. The C–C bond
of acetate remains intact in this process, as demonstrated by
the coupling of C-20 to C-19 in pramanicin derived from
[1,2-13C2]acetate. In fatty acid biosynthesis, it has long been
thought that the starter unit is usually an acyl entity, as opposed
to the chain-extension units which are malonate-derived.
Recently, the generality of this hypothesis has come into
J. Chem. Soc., Perkin Trans. 1, 2000, 4390–4402
4395

View Article Online
atom derived from C-1 of the starter acetate unit. Long fatty
acyl chains are assembled by repetitive cycles of condensation
of the ensuing saturated acyl chain with further malonate units
(KS) and reduction of the resulting ketone carbon to a methyl-
ene group, using KR, DH and ER subunits as above. The
condensation of acetate and malonate units thus results in a
labelling pattern in fatty acids and polyketides in which acetate
units are connected in a ‘head-to-tail’ manner. PKS’s are con-
sidered to assemble acetate units in a similar manner, but with
the omission of all, some or none of the β-ketone-processing
steps (KR, DH and ER), the choice made being dependent
upon chain length.33
The pattern of labelling and the intact incorporation of
acetate into each extender unit in 1 indicate that the acyclic
carbon chain in 1 is assembled in a ‘head-to-tail’ manner: C-11/
C-12 and C-13/C-14 are head-to-tail units, while C-15/C-16 and
C-17/C-18 are also intact units which are presumably arranged
in the same way, although overlap of the 13C NMR resonances
precluded unambiguous assignment of the direction of
incorporation of the latter two acetate moieties. Both labelled
acetates and diethyl [2-13C]malonate were found to label each
extender unit through to C-11 to an equal extent. Taken
together, these results are strongly suggestive that assembly of a
fatty acyl entity by an FAS-like enzyme furnishes at least the
unfunctionalised C-20 to C-11 saturated chain in 1, and suggest
a decanoyl entity as an obligatory intermediate.
There is precedent in the aflatoxin biosynthetic pathway in
Aspergillus parasiticus for assembly of the hexanoate starter
unit by a short-chain FAS; hexanoate is then extended by a
separate PKS which furnishes the polycyclic aromatic core of
the natural product.40 In that case, Townsend et al. demon-
strated impressive specific incorporations of hexanoate and its
N-acetylcysteamine (NAC) thioester into the side-chain of
averufin, a precursor for aflatoxin B1.41 Similarly, octanoate is
assembled from acetate by an FAS, then acts as a starter for the
PKS in piliformic acid biosynthesis.42 On the basis of these
results, we reasoned that decanoate might similarly be syn-
thesised on a separate FAS enzyme, prior to loading onto a
putative PKS for the addition of further acyl units en route to 1.
The ability of the PKS to load an acyl thioester which might
well originate from a separate enzyme made this system an
attractive candidate for incorporation of the corresponding
labelled fatty acid or NAC derivative. Thus, we investigated the
incorporation of both sodium [1-13C]decanoate and S-[1-13C]-
decanoyl-NAC 4 into 1, and/or into 3, in both the presence
and absence of the β-oxidation inhibitor 3-(tetradecylthio)-
propanoic acid 5. The results failed to provide any reproducible
evidence for specific incorporation of label into either 1 or 3;
although one experiment did appear to provide a 0.2% specific
incorporation into the C-11 site of 3, we have been unable to
duplicate this result. The cells of Stagonospora were permeable
to putative precursor 4: sites corresponding to each C-1 of
acetate were labelled to the extent of up to 2.7% per site. This
non-specific incorporation indicates that the NAC derivative
was incorporated exclusively via β-oxidation to acetate.43 No
evidence for a pool of free decanoic acid was found, since
recovered decanoic acid from the sodium decanoate experiment
was still entirely labelled with carbon-13. The failure to observe
intact incorporation of 4, while being inconclusive, may suggest
that the FAS creates a chain longer than decanoate, e.g., a
fourteen-carbon species such as tetradecanoate. In that case, the
obligatory decanoyl moiety would be a transient intermediate
on the FAS surface only, and 4 might not be able to load onto
the enzyme.
Scheme 5 Proposed biosynthetic pathways to pramanicin 1
dispute from work on actinorhodin34 and tetracenomycin35
polyketide synthases (PKSs), where malonate is the starter.
Leadlay and co-workers also showed that malonate could be
decarboxylated by components of two different PKSs to fur-
nish the starter unit in situ on the enzyme surface.36 Discrimin-
ation between acetate and malonate units in polyketides has
in some cases been made by quantitative labelling experiments,
usually using diethyl malonate which acts as a convenient
source of malonyl-CoA through hydrolysis and thioesterifi-
cation in vivo.37 However, malonate may also label all sites
equally because the CoA thioesters of acetate and malonate
undergo rapid interconversion in vivo.38,39 In the current case,
the results with diethyl [2-13C]malonate show that incorporation
into the starter, compared with the extender units, is reduced by
a factor of two. This effect clearly suggests that malonate must
be converted to acetate prior to incorporation into this site. The
results compare with the acyltetramic acid erythroskyrine,
where a similar low incorporation of malonate into the starter
unit was observed.6 Further, significant retention of three
deuterium atoms in 1 derived from [1-13C,2-2H3]acetate shows
that acetate can be incorporated without the intervention of
malonate into the starter unit in this case, although some
molecules may have passed through malonate, as evidenced by
the presence of monodeuterio- and dideuterio-isotopomers.
Clearly, acetate is an effective starter unit. Again, the acyl-
tetramic acid chaetoglobosin was shown by Tamm and Probst
to retain up to three deuterons from C-2 of acetate.13
In fatty acid biosynthesis, extension from the starter acetate
unit occurs by fatty acid synthase (FAS)-mediated decarboxyl-
ative condensation with a malonyl group derived from malonyl-
CoA, using the ketosynthase (KS) subunit.32 After this initial
C–C bond-forming reaction, the resulting β-keto group is
removed by a sequence of FAS-catalysed steps: the ketone
is reduced by a ketoreductase (KR), the resulting alcohol is
dehydrated by a dehydratase (DH) to give an α,β-unsaturated
acyl chain, and finally conjugate addition of hydride by an
enoylreductase (ER) produces a methylene group at the carbon
The extension of the presumed decanoyl intermediate occurs
by ‘head-to-tail’ addition of two further intact acetate-derived
malonate units (C-10/C-9 and C-8/C-7), as shown by the label-
ling experiments with acetates and diethyl malonate. We had
reasoned that these units may be added by a PKS-like activity,
in a manner reminiscent of the biosynthesis of the polyene
4396
J. Chem. Soc., Perkin Trans. 1, 2000, 4390–4402

View Article Online
class of polyketide antibiotics,39,44 as opposed to the FAS-like
assembly of C-20 to C-11. In this case, the two cycles of
malonate decarboxylative addition would each be followed by
reduction of the resulting β-keto group by a KR and dehydra-
tion of the ensuing alcohol by DH, leading to a tetradeca-2,4-
dienoyl entity. The exceptional level of retention of deuterium
at C-8 and C-10 over that in the C-18 to C-11 portion of the
molecule in the experiment with [1-13C,2H3]acetate supports this
idea in that the acetate units in the former region seem to be
processed differently from those in the latter. However, the
assembly of a saturated tetradecanoyl moiety by a single FAS,
followed by dehydrogenation to afford the diene moiety at this
or a later stage, a process that is well established in fatty acid
β-oxidation,43 cannot be ruled out. Nonetheless, fatty acids
longer than fourteen carbon atoms, such as palmitate, can be
excluded from the pathway, since the results with oxygen-18-
by the action of SHMT on [1,2,3-13C3,15N]serine, generating
[1,2-13C2,15N]glycine and [13C]Me-THF. These then recombine
with unlabelled Me-THF and Gly, respectively, from the
cellular pool to generate [1,2-13C2,15N]serine and [3-13C]serine
which then incorporate to give the observed isotopomers of 1.
Not only is serine incorporated with the three carbon atoms
intact, but the nitrogen atom derives from serine as well; more-
over, serine is incorporated with the Cα–N bond intact, as
shown by coupling in both the 15N and 13C NMR spectra in the
intact isotopomer derived from -[1,2,3-13C3,15N]serine. This
intact incorporation of nitrogen from the amino acid into 1 is
again consistent with the acyltetramic acid route: Probst and
Tamm observed incorporation of [2-15N]tryptophan into chaeto-
globosin.13 In addition, both illicicolin H and tenellin have been
studied using nitrogen-labelled precursors, although the bio-
synthesis of these metabolites may not involve an acyltetramic
acid intermediate.45–47 The results exclude the action of trans-
aminase on serine as a required component of the pathway to 1.
Further, if serine transaminase is present and active in the
Stagonospora cells, then it must be acting in an essentially
irreversible manner, since all the nitrogen label in 1 was found to
be coupled to carbon-13.
labelled acetate show retention of the C᎐O bond at C-7; as well,
᎐
the origin of the pyrrolidone acetate unit from acetate in prefer-
ence to malonate supports this conclusion. Incorporation
experiments with other NAC derivatives, spurred on by the
successful uptake of 4 into the cells of Stagonospora, to clarify
this issue are currently in progress, and will be reported in
due course.
The formation of intermediate 2 can be rationalised mech-
anistically in the same way as has been proposed for other
acyltetramic acids; i.e., condensation of the nucleophilic C-2 of
acetate or its equivalent with the electrophilic carboxy groups
of both the preformed polyketide or fatty acid and the presum-
ably activated serine. The formation of the amide bond by
attack of the amino group of serine onto the activated carbonyl
group of the ring acetate unit also makes mechanistic sense.
These mechanisms have analogy to both polyketide-like chain-
extension steps, and to non-ribosomal peptide synthetases
(NRPS),50 but also involve the more unusual polyketide-like
C–C bond formation between acetate C-2 and an amino acid
carboxy group. Because of the intense interest and activity in
both NRPS and PKS which has evolved from genetic work and
protein studies on both groups of enzymes, there has been
much recent discussion of the similarities between these two
classes of enzymes.50 Further, although many biosynthetic
experiments have been conducted on acyltetramic acids (see
above), there is a paucity of information on the nature,
sequence and timing of the steps involved in this part of the
assembly process. For these reasons, it was of interest to investi-
gate further the details of events leading from the assembled
fourteen-carbon acyclic chain through to 2.
First, the proposed mechanism requires that the oxygen
atoms at C-2 and C-4 of 2 derive from acetate and the carboxyl
oxygen of serine, respectively. The results from incorporation of
[1-13C,18O2]acetate and [1-13C,18O2]serine into 1 confirm intact
incorporation of these C–O bonds. Thus, if 2 is an intermediate
en route to 1, it must also be labelled in accord with the
proposed mechanism. The retention of oxygen at C-7, and of
deuterium at C-5 and C-6 (below), are also fully consistent with
the intermediacy of 2. Further, the retention of O-4 excludes
some alternative pathways. One attractive possibility which can
be excluded is the formation of an α-acyl-α,β-unsaturated
lactam such as 13 (Scheme 6); epoxidation to give 14, followed
At this point, the assembly of the pyrrolidone head group of
1 begins. The results with labelled acetates show that the C-2
and C-3 atoms, as well as O-2, in 1 are derived from an intact
acetate unit which is arranged head-to-tail with respect to C-7.
The remaining three carbon atoms, C-4, C-5 and C-6, are
derived from the carboxy-, α-, and hydroxymethyl carbon
atoms of serine, respectively, as shown by the combined results
of the labelling experiments with [1-13C]-, [3-13C,2H]- and [1,2,3-
13C3,15N]-serines. This pattern of labelling of the ring in 1 is
inconsistent with an origin in glutamate, which we had con-
sidered to be a reasonable alternative precursor to N-1 through
to C-6. However, the labelling arrangement is identical to that
observed in the acyltetramic acid group of metabolites, suggest-
ing that the biosynthetic pathway proceeds through an inter-
mediate such as 2. All of the members of this class of metabol-
ites that have been investigated biosynthetically possess a
tetramic acid ring which is derived from one acetate unit and an
amino acid. Thus, the pyrrolidone moieties in erythroskyrine,6
cyclopiazonic acid,7 tenuazonic acid,8 malonomicin,9,10 strep-
tolydigin,11 ikarugamycin,12 the cytochalasans,13 and aflastatin
A14 are derived from valine, tryptophan, isoleucine, 2,3-
diaminopropanoate, methylaspartate, ornithine, tryptophan
and alanine, respectively. Tetramic acid intermediates have also
been proposed in the biosynthesis of, for example, tenellin45
and ilicicolin H46 from phenylalanine, although more recent
results have cast this hypothesis into doubt.47 To the best of
our knowledge, pramanicin represents the first example of
a (modified) acyltetramic acid which has been shown to be
derived from serine, although there are probably other
metabolites which are formed similarly such as the interesting
bistetramic acid polycephalin C from Physarum polycephalum
which controls response to light.48 Serine has been shown to
be incorporated into malonomicin,9 but in this case serves as
a precursor for 2,3-diaminopropanoate prior to acyltetramic
acid formation.
The results show that label at carbon-5 is coupled to both
C-4 and C-6 in a significant proportion of the molecules of
1 derived from -[1,2,3-13C3,15N]serine. Therefore the carbon
skeleton of serine is incorporated intact into pramanicin. Other
labelled molecules possess either a single site of enrichment at
C-6, or are labelled at both C-4 and C-5 (and with 15N, see
below), but not C-6. Hence there must be an incidental process,
which does not have to occur en route to 1, that involves cleav-
age of the C-2–C-3 bond of serine. The most likely candidate is
serine hydroxymethyltransferase (SHMT), which freely inter-
converts serine with glycine and methylenetetrahydrofolate
(Me-THF).49 Thus the observed results are readily explained
Scheme 6 A putative route through an epoxide is attractive, but
inconsistent with the retention of the C᎐O bond of serine
᎐
J. Chem. Soc., Perkin Trans. 1, 2000, 4390–4402
4397

View Article Online
by hydrolysis would have accounted for the trans stereo-
chemistry of the C-3–C-4 diol. Precedent for such an epoxid-
ised α-acyl-α,β-unsaturated lactam may be found in fusarin C.51
In order to elucidate the timing of the C-3–C-7, N-1–C-2 and
C-3–C-4 bond-formation events, diethyl [2-13C]malonate was
incorporated into 1. If the C-3–C-7 bond formed first, then it
would be expected that C-3 would derive from malonate by a
normal, PKS-catalyzed chain extension. If, on the other hand,
the N-1–C-2 and/or C-3–C-4 bonds were formed prior to
acylation by the side-chain intermediate, then C-3 might origin-
ate from acetate. Malonate was in fact found to incorporate
into all of the acetate-derived sites, including the starter, acyl-
chain extender, and C-3 sites; however, C-3 was labelled to the
same extent as the acetate-derived starter unit, and significantly
less than the extender units. This effect clearly suggests that the
C-2–C-3 unit is derived from acetate. Further, when the results
from incorporation of [13C2]acetate are examined in detail,
small doublet of doublet signals were observed due to
incorporation of multiple, adjacent labelled acetate entities.
Carbon resonances from label at C-3 to C-19 are split by both
intra- and inter-unit one-bond couplings to generate these
signals. The incorporation level for interunit coupling between
C-3 and C-7 in 1 was indistinguishable from that of any other
pair of interunit-coupled carbons. Since this interunit effect,
which has often been observed in acetate incorporations into
polyketides, is larger than statistically expected, it has been
ascribed to the pulsed feeding of the starting material which
results in temporal variations of the levels of labelled precursor
within the metabolic pool. Thus, the present result suggests that
addition of the C-2–C-3 unit occurs at a time and within a
cellular sub-compartment indistinguishable from that of
assembly of the C-7-to-C-20 acyl chain. If the pyrrolidone is
assembled from acetate and serine prior to acylation, then this
intermediate cannot accumulate to a significant extent, because
isotopic dilution would occur. While these experiments do not
provide an unambiguous answer to the question of the
sequence of events for ring assembly, they are fully consistent
with and suggestive of a distinct type of process for attachment
of the C-2–C-3 unit, perhaps addition of serine to acetate to
generate N-acetylserine and/or the tetramic acid derivative 14
prior to acylation. Unlike all the results described so far, this
experiment stands in contrast to previous observations on
acyltetramic acids, where malonate has been widely assumed to
be the precursor of the ring acetate unit. On the basis of radio-
labelling experiments with malonate, Shibata et al.6 concluded
that erythroskyrine was formed from an acetate starter, but
that malonate was used in all the extender units including
that of the pyrrolidone ring. In malonomycin, Schipper et al.,
using deuterated precursors, showed that the pyrrolidone is
formed by condensation of a preformed 3-oxoadipoyl unit with
diaminopropanoate.10 The current results provide a clear dis-
tinction between direct incorporation of malonate and indirect
incorporation through acetate, as evidenced by the incorpor-
ation rate into the starter unit; this unit was independently
shown to be acetate-derived from the [13C,2H3]acetate experi-
ment which therefore acts as a positive internal control for the
indirect incorporation rate from malonate. The observation of
an essentially identical level of labelling at C-3 as for C-20 thus
clearly indicates that the pyrrolidone ring in 1 is formed from
acetate. Further investigations to resolve the issue of sequence
of bond-forming steps are being undertaken.
the epoxide from the C-10–C-11 olefin in 3 is the last step in
the biosynthetic pathway to pramanicin. This is consistent with
the lack of an 18O-induced isotope effect at C-11 when sodium
[1-13C,18O2]acetate is incorporated into 1, a result that suggests,
albeit unsurprisingly, that the epoxide oxygen does not derive
from acetate. We are not aware of any precedent for the appar-
ent inhibition of an epoxidation reaction, which presumably is
promoted by a cytochrome P450, by the supposed β-oxidation
inhibitor 5, or by the FAS substrate analogue 4, although both
might be expected to have some competitive inhibitory activity
on the cytochrome by virtue of their passing resemblance to the
olefinic substrate 3. Compound 5 has, however, been shown to
be processed by both β- and ω-oxidation in the rat,53 and to
induce rat cytochrome P4504A1 mRNA levels.54 Experiments
designed to test incorporation of 2 into 3 are also in progress.
The results with deuterated serine samples enable conclu-
sions to be reached about which enantiomer of serine incorpor-
ates into pramanicin, as well as the absolute configuration of 1.
The -enantiomer was incorporated at high levels in the
experiments with [2,3,3-2H3]-, [13C3,15N]- and [3-13C,2-2H]-
serines. However, the -enantiomer of [3-13C,2-2H]serine is not
a viable precursor, since it was not incorporated despite concur-
rent incorporation of racemic [1-13C]serine as a positive internal
control. For the experiment with [2,3,3-2H3]serine, the results
also show significant incorporation of deuterium at both of the
diastereotopic C-6 positions. Thus, there is no oxidation state
change at C-6. An apparently smaller extent of deuterium label-
ling was observed at C-5 than at C-6, suggestive of partial
exchange and the involvement of an amino acid racemase. On
the other hand, this result could be explained by the action of
SHMT, which is active in this case (above), and which would
generate [2H1]glycine and [2H2]Me-THF. These species would
recombine under the action of the freely reversible SHMT with
unlabelled serine and Me-THF from the metabolic pools. Since
these pools may be unequal in concentration, differing amounts
of [2-2H]- and [3-2H2]-serine would be generated, which would
lead to different levels of labelling at C-5 vs. C-6 of 1. Since the
results of this experiment thus do not demonstrate that all three
deuterium atoms from a single molecule of labelled serine
incorporate into a single molecule of 1, it was necessary to
show that deuterium was incorporated from C-2 without the
intermediacy of SHMT. This result was established by the
experiment with -[3-13C,2-2H]serine; the observation of a
β-deuterium isotope shift at C-6 in 1 for approximately 25% of
the molecules that are labelled shows that the D-Cα-13C entity
can be incorporated in an intact manner; i.e., without the inter-
vention of SHMT. The enhanced, unshifted signal at C-6
corresponds to the remaining 75% of serine which has been
processed by SHMT, or by a serine racemase that, as
expected,55,56 causes loss of the α-deuteron. These results estab-
lish that such a racemase is not required in the pathway to 1. In
combination, the results show that the  enantiomer of serine is
incorporated into 1 with all four bonds to the α-carbon intact,
while the  enantiomer is not incorporated, a result that also
argues against the action of a serine racemase. Therefore, 1
must possess the same absolute configuration at C-5 as that in
-serine, and thus has the 5S absolute configuration. Since we
first published this result as a communication, Barrett et al. have
reached the same conclusion by synthesis of the enantiomer of
pramanicin.15
Finally on the pathway, transformation of 2 into 3 requires a
mechanistically unremarkable reduction at the C-4 keto group,
a hydroxylation at the highly acidic C-3 position, a process
reminiscent of chemical enolate oxidations with electrophilic
oxygen species,52 and epoxidation of the C-10,C-11 alkene, in an
as-yet-undetermined order. Surprisingly, we found that adminis-
tration of both 4 and 5 enhanced formation of 3 and reduced
that of 1; thus both compounds appear to suppress oxidation
of 3 to 1. The isolation of 3 suggests that the formation of
Conclusions
In summary, pramanicin 1 has been shown to be formed from
head-to-tail condensation of a starter acetate unit with six
malonate units to provide the acyclic, acyl tail. The pyrrolidone
ring is derived from one acetate and -serine, which was shown
to be incorporated with the three carbons, the nitrogen, the
carboxy oxygen, the α- and both β-hydrogens as an intact
unit. As expected, oxygen at C-1 of both the last unit in the
4398
J. Chem. Soc., Perkin Trans. 1, 2000, 4390–4402

View Article Online
unsaturated fatty acid side chain, and in the unit within the pyr-
rolidone ring, is retained. The results suggest the intermediacy
of N-acetylserine and/or tetramic acid 12 (Scheme 5), which
is then acylated by a preassembled fourteen-carbon moiety,
leading to acyltetramic acid 2. Modification by a reduction
at C-4 and oxidation at C-3 then ensue. Epoxidation of the
10,11-alkene is probably the last step, since the putative precur-
sor 3 was isolated as a co-metabolite. The observation of reten-
tion of all four bonds to the α-carbon of -serine shows that
pramanicin must possess the same absolute configuration at C-
5; i.e., 5S. Although the results do not yet precisely define the
sequence of all steps in this pathway, they allow the exclusion of
some of the conceptually reasonable possibilities and provide
specific limits on the pathway from simple precursors to 1.
Although decanoyl-NAC 4 was not incorporated, the results
show that this type of derivative is available to the cells. Thus
the stage is set for further ongoing experiments with advanced
precursors, which should shed further light on the pathway to 1,
and presumably by extension to other members of the import-
ant acyltetramic acid group of metabolites.
of 1 g dmϪ3 FeSO4ؒ7H2O, 1 g dmϪ3 MnSO4ؒ4H2O, 25 mg dmϪ3
CuCl2ؒ4H2O, 100 mg dmϪ3 CaCl2, 56 mg dmϪ3 H3BO3, 19 mg
dmϪ3 (NH4)2Mo7O24ؒ4H2O, and 20 mg dmϪ3 ZnSO4ؒ7H2O).
The seed flask was incubated at 26 ЊC in a rotatory incubator-
shaker at 240 rpm for 3 days with a light cycle (40 W fluorescent
bulb) of 12 h on, 12 h off. Aliquots (2 cm3) of the resulting
suspension of thick grey to black mycelium were used to inocu-
late twenty-four 500 cm3 Erlenmeyer flasks containing LCM
production medium [corn meal (20 g dmϪ3), glucose (either 10,
40 or 75 g dmϪ3), ardamine PH (5 g dmϪ3), -leucine (3.5 g
dmϪ3), MES (16.2 g dmϪ3) and water (100 cm3) adjusted to pH
6.0 with NaOH]. The production flasks were incubated at 25 ЊC
in a rotatory incubator-shaker at 240 rpm for 7 days with full
illumination.
Addition of precursors
Isotopically labelled precursors were added via a sterile syringe
as aseptic solutions in water, except for diethyl malonate and 4
which were added as ethanol solutions in a volume of 0.2 cm3
flaskϪ1 dayϪ1. Compound 5 (100 mg flaskϪ1 dayϪ1) was added as
the solid. Additions were made daily in even aliquots over days
3 to 6, except where indicated. The amounts of precursor added
to the cultures were: 15 mg flaskϪ1 dayϪ1 for all serine samples,
50 mg flaskϪ1 dayϪ1 for optimal incorporation of labelled
acetates, 50 mg flaskϪ1 dayϪ1 for S-decanoyl-NAC 4 and 20 mg
flaskϪ1 dayϪ1 for diethyl malonate. The glucose concentration
in the medium was: 75 g dmϪ3 for all serine samples, 40 g dmϪ3
for optimal incorporation of labelled acetates, and 10 g dmϪ3
for S-decanoyl-NAC 4 and diethyl malonate.
Experimental
General
Stagonospora sp. ATCC 74235 was purchased from the
American Type Culture Collection. All media and materials for
culture were autoclaved at 121 ЊC for 20 min. Chemicals were
purchased from Aldrich and Sigma and used without further
purification. Silica gel was from Silicycle (Montreal). Isotopic-
ally labelled precursors were from CIL (Andover, MA). Water
was obtained from a Nanopure water-purification system. THF
was freshly distilled under nitrogen protection from potassium/
benzophenone.
All reactions were performed in flame-dried glassware, under
a positive pressure of nitrogen, unless otherwise noted. Air- and
moisture-sensitive compounds were transferred by syringe.
Concentration of organic solutions was performed on the
rotary evaporator (≈25 mmHg), followed by evacuation to
constant weight (<0.05 mmHg).
Isolation of pramanicin 1 and dienone 3
Methyl ethyl ketone (100 cm3 flaskϪ1) was added, and shaking
continued in the rotatory incubator-shaker for 1 h. The mixture
was centrifuged (8000 rpm, 15 min), and the organic and aque-
ous layers removed to a separatory funnel. The organic layer
was collected, and treated with one volume of hexanes. The
aqueous layer was then removed, and the organic layer concen-
trated in vacuo. The resulting extract was purified by flash
chromatography on silica gel. After elution with 2% v/v MeOH–
EtOAc, the pramanicin fraction was obtained by elution with
10% v/v MeOH–EtOAc. This material was concentrated in
vacuo, dissolved in methanol, and the composition was
adjusted to 70:30 MeOH–water. The precipitate was removed,
and the solution loaded onto a commercial RP-8 extraction
column (400 mg). Pramanicin was eluted with 70:30 MeOH–
water and the eluate concentrated by removing the water as the
ethanol azeotrope. The resulting material was dissolved in a
minimum volume of 70:30 MeOH–water, and applied to a
Merck LOBAR RP-8 column which was eluted in 70:30
MeOH–water. Fractions containing 1 or the deoxy compound
3 were combined and concentrated by azeotropic distillation
with ethanol. For 1, the spectral data were identical to those
reported by Schwartz et al.1 For 3: mp 135–138 ЊC; λmax
(MeOH)/nm 288.5 (ε/dm3 molϪ1 cmϪ1 26 800); νmax/cmϪ1 3000–
Mps were measured on a Gallenkamp MF370 instrument
and are uncorrected. Proton and carbon-13 nuclear magnetic
resonance (NMR) spectra were recorded on Bruker AC 200,
Bruker AC 300 or Bruker DRX 500 NMR spectrometers. Pro-
ton chemical shifts (δ scale) refer to TMS (SiMe4) and those of
13C to CDCl3. Data are given as chemical shift (integral, multi-
plicity [s = singlet, d = doublet, t = triplet, q = quartet, m =
multiplet], coupling constant[s] [J in Hz], assignment). Fourier
transform (FT) IR spectra were recorded on a Bio-Rad SPC
3200 spectrophotometer or on the tip of the probe of an ASI in
situ reaction-monitoring system. UV spectra were obtained on
a Perkin-Elmer Lambda 9 UV/VIS/NIR spectrophotometer or
a Cary 50 machine. Mass spectra were recorded on a Finnigan
4500 mass spectrometer for electron impact (EI) and chemical
ionisation (CI) or on a Micromass Quattro LC for electrospray
(ES). Optical rotations were measured on a Perkin-Elmer 241MC
polarimeter; [α]D-values are given in units of 10Ϫ1 deg cm2 gϪ1
.
3600br (O–H), 2956, 2923, 2854 (C–H), 1692, 1650 (C᎐O),
᎐
1586, 1009; NMR data are given in Table 1; MS (ES) 353 (Mϩ).
Culture of Stagonospora
Dienone 3 from pramanicin 1
Freeze-dried cultures were suspended in water, and used to
inoculate Petri dishes of potato dextrose agar [prepared from
200 g of potatoes, diced and boiled in water (333 cm3) for 10
min, filtered through cheesecloth and the filtrate added to agar
(Difco, 10 g) and dextrose (13.3 g)]. The plates were incubated
in a moist atmosphere at 25 ЊC for 3 days with a light cycle
(40 W fluorescent bulb) of 12 h on, 12 h off. The resulting
mycelia were suspended in water and transferred to an un-
baffled 500 cm3 Erlenmeyer flask containing KF seed medium
[tomato paste (4 g), corn steep liquor (0.5 g), oat flour (1 g),
glucose (1 g), water (99 cm3) and trace element solution (1 cm3
To a solution of pramanicin (71.4 mg, 194 µmol) in anhydrous
acetonitrile (10 cm3) were added toluene-p-sulfonic acid (107.5
mg) and sodium iodide (141.3 mg). The mixture was stirred
for 90 min at room temp., and the solvent was removed. The
residue was partitioned between chloroform (5 cm3) and aq.
sodium bicarbonate (1 M; 5 cm3). After washing of the aqueous
layer with two further aliquots of chloroform, the combined
organic extracts were dried over sodium sulfate, filtered and
concentrated. The residue was purified by chromatography on
an RP-8 LOBAR column as above. The desired fractions were
J. Chem. Soc., Perkin Trans. 1, 2000, 4390–4402
4399

View Article Online
combined and evaporated with ethanol to furnish compound 3
(49.5 mg, 72%).
mmol) in NaOH (1 M; 3 cm3) was added 3,5-dinitrobenzoyl
chloride (230 mg) and the mixture was stirred for 10 min. More
NaOH (2 cm3) and acid chloride (370 mg) were added and the
mixture was stirred for 1 h. The precipitate was collected,
washed with diethyl ether (2 × 10 cm3) and dried in vacuo to
give the product 8 (364 mg, 77%) which was used immediately
without further characterisation due to poor solubility.
S-[1-13C]Decanoyl-N-acetylcysteamine 4
A solution of [1-13C]decanoic acid (430 mg, 2.5 mmol) in dry
THF (50 cm3) was cooled to 0 ЊC and triethylamine (0.34 cm3)
then ethyl chloroformate (0.24 cm3) were added dropwise via
syringe. The solution was stirred for 1 h at 0 ЊC. To degassed
nanopure water (40 cm3), through which a slow stream of nitro-
gen was continuously bubbled, was added KOH (1.4 g) and,
after cooling to room temperature, N,S-diacetylcysteamine
(1.21 g). After stirring of the mixture at room temperature for
45 min, the pH was adjusted to 7.8 with conc. HCl. This solu-
tion was then added to the above solution of mixed anhydride.
After stirring of the mixture for 1 h at 0 ЊC while the pH was
maintained at 8.0 with conc. aq. HCl or KOH solutions, the pH
was carefully adjusted to 3 using conc. HCl. Solvent was then
removed on the rotary evaporator and the remaining aqueous
solution was extracted with diethyl ether (3 × 75 cm3). The
organic layer was washed with 3% aq. HCl, dried over
anhydrous sodium sulfate, and concentrated to give crude
material as a slightly off-white solid. Flash column chromato-
graphy in EtOAc gave product 4 (620 mg, 90%) as a white solid,
3,5-Dinitrobenzoyl-D-[2-2H,3-13C]serine chloramphenicol salt 9
Product 8 (364 mg, 1.21 mmol) and chloramphenicol (154.5
mg, 729 µmol, 1.2 eq. with respect to -enantiomer) were dis-
solved in methanol (0.35 cm3) by heating. Ethyl acetate (3.5
cm3) was added, and crystallisation induced by partial evapor-
ation of methanol and addition of more EtOAc. The crystals
were collected, washed with EtOAc, and dried in vacuo to give
the salt 9 (280 mg, 89% on -enantiomer), [α]D28 Ϫ29.7 (c 1.09,
H2O) [lit.,29 Ϫ32.2 (c 1, H2O)]. The filtrate was retained for
preparation of the -enantiomer (see below).
3,5-Dinitrobenzoyl-D-[2-2H,3-13C]serine and D-[2-2H,3-13C]-
serine hydrochloride 7b
The chloramphenicol salt 8 (280 mg, 545 µmol) was dissolved in
water (4 cm3) and treated with NaOH (10 M; 0.075 cm3). Conc.
HCl (0.2 cm3) was added dropwise, and the solution kept until
no further tan precipitate appeared. After cooling on ice for 30
min, the precipitate was collected, washed with HCl (5%, 1 cm3)
and dried in vacuo over phosphorus pentoxide to give the DNB-
serine (127.5 mg, 77%). A mixture of this material (120 mg, 400
µmol), water (0.9 cm3) and conc. HCl (0.9 cm3) was heated at
reflux for 1 h. The mixture was filtered and the solid was washed
successively with HCl (5%; 1 cm3) and water (1 cm3). The filtrate
was concentrated by azeotropic distillation from ethanol, and
the residue triturated twice with acetone, then dried to give the
product 7b (49.8 mg, 87% of hydrochloride), [α]D28 Ϫ9.0 (c 1,
H2O) [lit.,29 Ϫ12 (c 1, HCl)]; 1H NMR (D2O; 200 MHz) δ 3.82
(1H, dd, J 148, 12.5, 3-Ha), 3.76 (1H, dd, J 148, 12.5, 3-Hb), 3.62
(m, 0.02H, residual 2-H); 13C NMR (D2O; 50 MHz) δ 172.4 (s),
60.3 (enhanced s), the C-α signal was not observed; m/z (NH3-
CI) 108 ([13C,2H]MHϩ); m/z (EI) 108 ([13C,2H]MHϩ), 76
([2H][MH Ϫ 13CH2O]ϩ), 75 ([CD(NH2)CO2H]ϩ), 62 ([13C,2H]-
[MH Ϫ CO2H]ϩ), 58 (unassigned), 44 ([13C,2H][CH2CH=
NH]ϩ); TLC: single spot on CHIRALPLATE in acetone–
MeOH–water 10:2:2 with detection by 0.3% ninhydrin and
heating.30
mp 59–61 ЊC; νmax/cmϪ1 3294br (N–H), 2954, 2917, 2850 (C–H),
1
1702, 1648 (C᎐O), 1553, 1468, 1360, 1295, 1148, 957, 758; H
᎐
NMR (CDCl3; 200 MHz) δ 6.24 (1H, br s, NH), 3.33 (2H, q,
J 6.0, CH2), 2.94 (2H, dt, J 6.5, 5.5, CH2), 2.48 (2H, dt, J 6.0,
7.2, CH ), 1.89 (3H, s, CH C᎐O), 1.57 (2H, m, CH ), 1.19 (12H,
᎐
2
3
2
m, 6 × CH2), 0.80 (3H, t, J 6.7, CH3); 13C NMR (CDCl3;
50 MHz) δ 199.87 (enhanced s), 175.73, 43.94 (d, J 46),
39.51, 31.66, 29.19, 29.05 (2C), 28.76 (d, J 4), 28.25, 25.49,
22.95, 22.47, 13.91; m/z (EI) 275 (M ϩ H, weak), 119, 60, 43;
m/z (NH3 Ϫ CI) 275 (M ϩ H).
6-Pivaloylpramanicin 6a and deuterated sample 6b
To pramanicin (10.4 mg, 28.2 µmol) were added pivalic
anhydride (0.02 cm3) and triethylamine (0.18 cm3). The mixture
was heated to 60 ЊC for 15 min, cooled, and quenched with a
small amount of ice. After removal of volatiles, the gummy
residue was dissolved in EtOAc, applied to a flash chromato-
graphy column, and eluted with EtOAc. The product was
collected in the first few fractions, and concentrated in vacuo to
1
give 6a as a white powder (7.1 mg, 56%); H NMR (CD3OD;
500 MHz) δ 7.06 (1H, dd, J 15.6, 1, 8-H), 6.64 (1H, dd, J 15.6,
7.0, 9-H), 4.37 (1H, dd, J 12.0, 3.0, 6-Ha), 4.17 (1H, d, J 7.4,
4-H), 4.05 (1H, dd, J 12.0, 4.4, 6-Hb), 3.62 (ddd, J 7.4, 4.5, 3.0,
5-H), 3.32 (1H, m, 10-H), 2.92 (1H, m, 11-H), 1.6 (2H, m,
12-H2), 1.4 (2H, m, 13-H2), 1.29 (12H, m, 14- to 19-H2), 1.21
(9H, s, But), 0.89 (3H, t, J 6.6, 20-H3). Pramanicin derived from
-[2,3,3-2H3]serine was similarly converted to the pivaloyl
derivative 6b: 2H NMR (CH3OH; 76.77 MHz) δ 4.30, 4.00, 3.55.
L-Enriched 3,5-dinitrobenzoyl-[2-2H,3-13C]serine and the
quinidine salt 10
The filtrate from the chloramphenicol crystallisation above was
concentrated, the resulting oil was suspended in EtOAc (2 cm3),
and the solution was washed successively with HCl (2 M, 1
cm3), then water; both aqueous layers were further extracted
with ethyl acetate. The combined organic layers were dried
(Na2SO4), filtered, and concentrated in vacuo to give a foam
(195 mg, 96%) which was then triturated with hexanes, and
collected by filtration. This material and quinidine (184 mg, 565
µmol) were heated in ethanol (2.1 cm3) until they dissolved.
After cooling for 90 min, the crystals were collected, washed
with ethanol (2 cm3), and dried (in vacuo over phosphorus
pentoxide) to give salt 10 (207 mg, 59% on quinidine).
DL-[2-2H, 3-13C]Serine 7a
-[3-13C]Serine (250.5 mg, 2.37 mmol) and Al2(SO4)3ؒ18H2O
(57.6 mg) were dissolved in D2O (3 cm3); the solution was
adjusted to pD 9.6, and lyophilised. Pyridoxal was liberated
from the hydrochloride (94 mg) according to the literature pro-
cedure,25 and mixed with the lyophilised serine in D2O (5 cm3).
The solution was stirred in the dark for 24 h, then adjusted to
pH 6.5 with HCl. Ethanol (1.5 vol) was added and the mixture
allowed to crystallise overnight at Ϫ20 ЊC. The crystals were
collected, washed with EtOH, and dried in vacuo over phos-
L-3,5-Dinitrobenzoyl-[2-2H,3-13C]serine and L-[2-2H,3-13C]-
serine hydrochloride 7c
1
phorus pentoxide to give product 7a (178 mg, 70%), H NMR
Salt 10 (207 mg, 331 µmol) was suspended in EtOAc (2 cm3),
and the mixture was washed successively with HCl (2 M; 1 cm3),
then water; both aqueous layers were further extracted with
ethyl acetate. The combined organic layers were dried (Na2-
SO4), filtered, and concentrated in vacuo. Water (0.8 cm3) and
conc. HCl (0.8 cm3) were added and the mixture heated at reflux
(D2O; 200 MHz) δ 3.77 (1H, dd, J 146, 12, 3-Ha), 3.73 (1H, dd,
J 146, 12, 3-Hb), 3.62 (0.02H, m, residual 2-H).
3,5-Dinitrobenzoyl-DL-[2-2H,3-13C]serine 8
To a chilled solution of -[2-2H,3-13C]serine 7a (178 mg, 1.66
4400
J. Chem. Soc., Perkin Trans. 1, 2000, 4390–4402

View Article Online
14 S. Sakuda, M. Ono, K. Furihata, J. Nakayama, A. Suzuki and
for 1 h. The mixture was filtered and the solid washed with HCl
(5%; 1 cm3). The filtrate was concentrated by azeotropic distil-
lation from ethanol, and the residue triturated twice with acet-
one, then dried (in vacuo over phosphorus pentoxide) to give the
product 7c (48.7 mg, 100% of hydrochloride), [α]D28 10.0 (c 1,
H2O); TLC: single spot on CHIRALPLATE at lower Rf
A. Isogai, J. Am. Chem. Soc., 1996, 118, 7855; M. Ono, S. Sakuda,
H. Ikeda, K. Furihata, J. Nakayama, A. Suzuki and A. Isogai,
J. Antibiot., 1998, 51, 1019.
15 A. G. M. Barrett, J. Head, M. L. Smith and N. S. Stock, Chem.
Commun., 1999, 133; A. G. M. Barrett, J. Head, M. L. Smith, N. S.
Stock, A. J. P. White and D. J. Williams, J. Org. Chem., 1999, 64,
6005.
16 O. Spavold and J. Bremer, Biochim. Biophys. Acta, 1989, 1003, 72;
For application in biosynthetic studies, see: Z. Li, F. M. Martin and
J. C. Vederas, J. Am. Chem. Soc., 1992, 114, 1531; Y. Liu, Z. Li and
J. C. Vederas, Tetrahedron, 1998, 54, 15937.
17 J. M. Behan, R. A. W. Johnstone and M. J. Wright, J. Chem. Soc.,
Perkin Trans. 1, 1975, 1216; D. Van Ende and A. Krief, Tetrahedron
Lett., 1975, 2709; D. L. J. Clive, Tetrahedron, 1978, 34, 1049.
18 R. N. Baruah, R. P. Sharma and J. N. Baruah, Chem. Ind., 1983,
524.
19 T. J. Simpson, in Biosynthesis: Polyketides and Vitamins, ed. F. J.
Leeper and J. C. Vederas, Springer, Berlin, 2000, pp. 1–48. For a
review of the use of oxygen-18 and deuterium isotope shifts, oxygen
exchange with medium and effects of pulsed feeding on spectra, in
biosynthetic studies, see: J. C. Vederas, Nat. Prod. Rep., 1987, 4, 277.
20 For magnitudes of deuterium and oxygen-18 isotope shifts, see:
P. E. Hansen, Annu. Rep. NMR Spectrosc., 1983, 15, 106.
21 D. E. Cane, T.-C. Liang and H. Hasler, J. Am. Chem. Soc., 1982,
104, 7274.
1
than the -enantiomer (above); H NMR, 13C NMR, CI-MS
and EI-MS gave results that were identical to those for the
-enantiomer.
DL-[1-13C,18O2]Serine 11
-[1-13C]Serine (106 mg), H218O (0.09 cm3) and conc. HCl (0.1
cm3) were sealed into a flame-dried glass tube under a nitrogen
atmosphere. The tube was heated at 90 ЊC in an oil-bath for
5 days. After cooling, the mixture was neutralised with NaOH
and lyophilised. The resulting material containing the product
and sodium chloride was used without further purification. 1H
NMR (D2O; 200 MHz) spectrum was indistinguishable from
that for -[1-13C]serine; 13C NMR (D2O; 50 MHz) spectrum
was as for the starting material but the C᎐O resonance at
᎐
δC 172.4 was enhanced and exhibited 3 peaks, ∆δ 0.027 ppm
per 18O; 24% 18O2, 48% 18O16O, 28% 16O2, corresponding to 70
atom% 18O per site.
22 For an example, see: A. K. Demetriadou, E. D. Laue and
J. Staunton, J. Chem. Soc., Chem. Commun., 1985, 764, 1125.
23 L. R. Kass and D. J. H. Brock, Methods Enzymol., 1969, 14, 696.
24 M. J. Robins, S. D. Hawrelak, T. Kanai, J.-M. Siefert and R. Mengel,
J. Org. Chem., 1979, 44, 1317.
Acknowledgements
25 E. W. Miles and P. McPhie, J. Biol. Chem., 1974, 249, 2852.
26 C. A. Townsend, A. M. Brown and L. T. Nguyen, J. Am. Chem. Soc.,
1983, 105, 919.
27 A. I. P. M. de Kroon, J. W. Timmermans, J. A. Killian and B. de
Kruijff, Chem. Phys. Lipids, 1990, 54, 33.
28 L. Velluz, G. Amiard and R. Heymes, Bull. Soc. Chim. Fr., 1954,
1015.
29 H. Gorissen, C. van der Maesen, A. Mockel, G. Journee and
V. Libert, in Synthesis and Applications of Isotopically Labelled
Compounds 1991, ed. E. Buncel and G. W. Kabalka, Elsevier,
New York, 1992, pp. 588–591.
Financial support from the Natural Sciences and Engineering
Research Council of Canada (NSERC) is gratefully acknow-
ledged. Cambridge Isotopes is thanked for a grant of isotopic-
ally labelled compounds. H218O was a generous gift from
Professor J. C. Vederas, University of Alberta. We thank
Professor A. D. Bain at McMaster for assistance with the
spectral simulation.
References
30 K. Günther, J. Chromatogr., 1988, 448, 11.
31 R. C. Murphy and K. L. Clay, Biomed. Mass Spectrom., 1979, 6,
309.
1 R. E. Schwartz, G. L. Helms, E. A. Bolessa, K. E. Wilson, R. A.
Giacobbe, J. S. Tkacz, G. F. Bills, J. M. Liesch, D. L. Zink,
J. E. Curotto, B. Pramanik and J. C. Onishi, Tetrahedron, 1994, 50,
1675.
2 C. Y. Kwan, H. Lu-Cao, T. K. Kwan, P. H. M. Harrison, P. Duspara
and E. E. Daniel, unpublished work.
3 P. H. M. Harrison, D. W. Hughes and R. W. Riddoch, Chem.
Commun., 1998, 273.
4 P. Duspara, S. I. Jenkins, D. W. Hughes and P. H. M. Harrison,
Chem. Commun., 1998, 2643.
5 For reviews of tetramic acids, see: B. J. L. Royles, Chem. Rev., 1995,
95, 1981; H.-G. Henning and A. Gelbin, Adv. Heterocycl. Chem.,
1993, 57, 139; S. J. Shimshock and P. DeShong, Stud. Nat. Prod.
Chem., 1994, 14, 97.
6 S. Shibata, U. Sankawa, H. Taguchi and K. Yamasaki, Chem.
Pharm. Bull., 1966, 14, 474.
7 C. W. Holzapfel and D. C. Wilkins, Phytochemistry, 1971, 10, 351;
R. M. McGrath, P. S. Steyn and N. P. Ferreira, J. Chem. Soc., Chem.
Commun., 1973, 812; R. M. McGrath, P. S. Steyn, N. P. Ferreira and
D. C. Neethling, Bioorg. Chem., 1976, 4, 11; A. E. de Jesus, P. S.
Steyn, R. Vleggaar, G. W. Kirby, M. J. Varley and N. P. Ferreira,
J. Chem. Soc., Perkin Trans. 1, 1981, 3292; A. A. Chalmers, C. P.
Gorst-Allman and P. S. Steyn, J. Chem. Soc., Chem. Commun.,
1982, 1367.
8 S. Gatenbeck and J. Sierankiewicz, Acta Chem. Scand., 1973, 27,
1825.
9 D. Schipper, J. L. van der Baan and F. Bickelhaupt, J. Chem. Soc.,
Perkin Trans. 1, 1979, 2017.
10 D. Schipper, J. L. van der Baan, N. Harms and F. Bickelhaupt,
Tetrahedron Lett., 1982, 23, 1293.
11 K. L. Rinehart, Jr., D. D. Weller and C. J. Pearce, J. Nat. Prod.,
1980, 43, 1; C. J. Pearce, S. E. Ulrich and K. L. Rinehart, Jr., J. Am.
Chem. Soc., 1980, 102, 2510; C. J. Pearce and K. L. Rinehart, Jr.,
J. Antibiot., 1983, 36, 1536.
32 For recent reviews, see: B. J. Rawlings, Nat. Prod. Rep., 1997, 14,
335; 1998, 15, 275.
33 J. Staunton and B. Wilkinson, in Biosynthesis: Polyketides and
Vitamins, ed. F. J. Leeper and J. C. Vederas, Springer, Berlin, 2000,
pp. 49–92. For a selection of recent reviews, see: B. J. Rawlings,
Nat. Prod. Rep., 1999, 16, 425; R. Bentley and J. W. Bennett, Annu.
Rev. Microbiol., 1999, 53, 411; B. J. Rawlings, Nat. Prod. Rep., 1997,
14, 523; C. W. Carreras, R. Pieper and C. Khosla, Top. Curr. Chem.,
1997, 188, 85 and references therein.
34 C. W. Carreras and C. Khosla, Biochemistry, 1998, 37, 2084;
T. S. Hitchman, J. Crosby, K. J. Byrom, R. J. Cox and T. J. Simpson,
Chem. Biol., 1998, 5, 35; J. Dreier and C. Khosla, Biochemistry,
2000, 39, 2088.
35 W. L. Rao, E. Wendt-Pienkowski and C. R. Hutchinson,
Biochemistry, 1998, 37, 8132.
36 C. Bisang, P. F. Long, J. Cortés, J. Westcott, J. Crosby, A.-L.
Matharu, R. J. Cox, T. J. Simpson, J. Staunton and P. F. Leadlay,
Nature, 1999, 401, 502.
37 There are many early studies of this starter effect using carbon-14.
For an example where malonate is shown to incorporate selectively
into extender units using carbon-13 and NMR analysis, see:
I. M. Chandler and T. J. Simpson, J. Chem. Soc., Chem. Commun.,
1987, 17.
38 K. Schröder and H. G. Floss, J. Org. Chem., 1978, 43, 1438.
39 H. Noguchi, P. H. Harrison, K. Arai, T. K. Nakashima, L. A.
Trimble and J. C. Vederas, J. Am. Chem. Soc., 1988, 110, 2938.
40 For a review, see: R. E. Minto and C. A. Townsend, Chem. Rev.,
1997, 97, 2537; D. W. Brown, T. H. Adams and N. P. Keller,
Proc. Natl. Acad. Sci. USA, 1996, 93, 14873 and references
therein.
41 C. A. Townsend and S. B. Christensen, Tetrahedron, 1983, 39, 3575;
C. A. Townsend, S. B. Christensen and K. Trautwein, J. Am. Chem.
Soc., 1984, 106, 3868; S. W. Brobst and C. A. Townsend, Can. J.
Chem., 1994, 72, 200.
12 H. Seto, H. Yonehara, S. Aizawa, H. Akutsu, J. Clardy, E. Arnold,
M. Tanabe and S. Urano, Koen Yoshishu–Tennen Yuki Kagobutsu
Toronkai, Kyushu University, Fukuoka, Japan, 22nd conference,
1979, p. 394 (Chem. Abstr., 1980, 92, 211459).
42 N. C. J. E. Chesters and D. O’Hagan, J. Chem. Soc., Perkin Trans. 1,
1997, 827; H. Culceth, J. Fuchser, S. J. Moss, J. Nieschalk and
D. O’Hagan, Tetrahedron Lett., 1998, 39, 1949.
13 A. Probst and C. Tamm, Helv. Chim. Acta, 1981, 64, 2065.
J. Chem. Soc., Perkin Trans. 1, 2000, 4390–4402
4401

View Article Online
43 For a recent review, see: J. K. Hiltunen and Y.-M. Qin, Biochim.
Biophys. Acta, 2000, 1484, 117.
44 C. M. McNamara, S. Box, J. M. Crawforth, B. S. Hickman, T. J.
Norwood and B. J. Rawlings, J. Chem. Soc., Perkin Trans. 1, 1998,
83.
45 A. G. McInnes, D. G. Smith, J. A. Walter, L. C. Vining and J. L. C.
Wright, J. Chem. Soc., Chem. Commun., 1974, 282; E. Leete,
N. Kowanko, R. A. Newmark, L. C. Vining, A. G. McInnes and
J. L. C. Wright, Tetrahedron Lett., 1975, 4103; J. L. C. Wright,
L. C. Vining, A. G. McInnes, D. G. Smith and J. A. Walter, Can. J.
Biochem., 1977, 55, 678.
46 M. Tanabe and S. Urano, Tetrahedron, 1983, 39, 3569.
47 M. C. Moore, R. J. Cox, G. R. Duffin and D. O’Hagan, Tetrahedron,
1998, 54, 9195.
50 For a detailed discussion, see the thematic issue of Chem. Rev., 1997,
7(7).
51 M. E. Savard and J. D. Miller, J. Nat. Prod., 1992, 55, 64.
52 See, for example: D. A. Evans, M. M. Morrissey and R. L. Dorow,
J. Am. Chem. Soc., 1985, 107, 4346; M. P. Gore and J. C. Vederas,
J. Org. Chem., 1986, 51, 3700.
53 E. Hvattum, S. Skrede, J. Bremer and M. Solbakken, Biochem. J.,
1992, 286, 879.
54 P. Tollet, M. Strömstedt, L. Frøyland, R. K. Berge and J.-Å.
Gustafsson, J. Lipid Res., 1994, 35, 248.
55 For a recent review, see: M. E. Tanner and G. L. Kenyon, in
Comprehensive Biological Catalysis, Academic Press, San Diego,
London, 1998, vol. 2, pp. 7–41.
56 For PLP-dependent enzymes, see: C. Walsh, Enzymatic Reaction
Mechanisms, W. H. Freeman, New York, 1977; Vitamin
B6, Pyridoxal Phosphate in Coenzymes and Cofactors, ed.
D. Dolphin, R. Poulson and O. Avramovicˇ, Wiley, New York, 1986,
vol. 1.
48 A. Nowak and B. Steffan, Angew. Chem., Int. Ed., 1998, 37, 3139.
49 For reviews of SHMT, see: L. Schirch, Adv. Enzymol. Relat. Areas
Mol. Biol., 1982, 53, 83; R. G. Matthews and J. T. Drummond,
Chem. Rev., 1990, 90, 1275.
4402
J. Chem. Soc., Perkin Trans. 1, 2000, 4390–4402