On the biosynthetic origin of methoxymalonyl-acyl carrier protein, the substrate for incorporation of "glycolate" units into ansamitocin and soraphen A

Article abstract of DOI:10.1021/ja064408t

Feeding experiments with isotope-labeled precursors rule out hydroxypyruvate and TCA cycle intermediates as the metabolic source of methoxymalonyl-ACP, the substrate for incorporation of "glycolate" units into ansamitocin P-3, soraphen A, and other antibi

Full text of DOI:10.1021/ja064408t

Published on Web 10/13/2006
On the Biosynthetic Origin of Methoxymalonyl-Acyl Carrier
Protein, the Substrate for Incorporation of “Glycolate” Units
into Ansamitocin and Soraphen A
Silke C. Wenzel,† Rachel M. Williamson,‡ Christian Gru¨nanger,§ Jun Xu,§
Klaus Gerth,| Rodolfo A. Martinez,‡ Steven J. Moss,§ Brian J. Carroll,§
Stephanie Grond, Clifford J. Unkefer,‡ Rolf Mu¨ller,*,† and Heinz G. Floss*,§
Contribution from the Institute of Pharmaceutical Biotechnology, Saarland UniVersity,
P.O. Box 151150, D-66041 Saarbru¨cken, Germany, the National Stable Isotope Resource,
Bioscience DiVision, Los Alamos National Laboratory, Los Alamos, New Mexico 87545,
the Department of Chemistry, UniVersity of Washington, Seattle, Washington 98195-1700,
the German Research Center for Biotechnology (GBF), Mascheroderweg 1,
D-38124 Braunschweig, Germany, and the Institut fu¨r Organische und Biomolekulare Chemie
der UniVersita¨t Go¨ttingen, Tammannstrasse 2, D-37077 Go¨ttingen, Germany
Received June 22, 2006; E-mail: rom@mx.uni-saarland.de; floss@chem.washington.edu
Abstract: Feeding experiments with isotope-labeled precursors rule out hydroxypyruvate and TCA cycle
intermediates as the metabolic source of methoxymalonyl-ACP, the substrate for incorporation of “glycolate”
units into ansamitocin P-3, soraphen A, and other antibiotics. They point to 1,3-bisphosphoglycerate as
the source of the methoxymalonyl moiety and show that its C-1 gives rise to the thioester carbonyl group
(and hence C-1 of the “glycolate” unit), and its C-3 becomes the free carboxyl group of methoxymalonyl-
ACP, which is lost in the subsequent Claisen condensation on the type I modular polyketide synthases
(PKS). D-[1,2-13C2]Glycerate is also incorporated specifically into the “glycolate” units of soraphen A, but
not of ansamitocin P-3, suggesting differences in the ability of the producing organisms to activate glycerate.
A biosynthetic pathway from 1,3-bisphosphoglycerate to methoxymalonyl-ACP is proposed. Two new
syntheses of R- and S-[1,2-13C2]glycerol were developed as part of this work.
Polyketides represent one of the larger classes of natural
products, and many of them are endowed with useful biological
activities, such as antibiotic, antitumor, and immunosuppressive
properties.1 Many polyketide antibiotics, particularly of the
macrolide and macrolactam families, are assembled on type I
modular polyketide synthases (PKS) by repeated chain extension
of a starter unit and reductive/dehydratative modification of the
newly incorporated unit in each chain extension cycle.2 The most
common substrates for the chain extension reactions are
Figure 1. Structures of ansamitocin P-3 (AP3, 1) from Actinosynnema
pretiosum ssp. auranticum and soraphen A (2) from Sorangium cellulosum.
malonyl-CoA for the incorporation of acetate units, 2-methyl-
FK-506 and FK-520,7 concanamycin,8 bafilomycin,9 aflastatin,10
malonyl-CoA for the incorporation of propionate units, and less
oxazolomycin,11 and tautomycin12 contain another chain exten-
sion unit, called a “glycolate” unit, carrying an oxygen
(frequently methylated) at C-2 of a two-carbon chain.
In analogy to the other chain extension units, one would
frequently, 2-ethylmalonyl-CoA to incorporate butyrate units.
A number of antibiotics, such as ansamitocin P-3 (AP3, 1, Fig-
ure 1),3 soraphen A (2, Figure 1),4 geldanamycin,5 leucomycin,6
† Saarland University.
expect the substrate for incorporating this “glycolate” unit to
‡ Los Alamos National Laboratory.
§ University of Washington.
(6) Omura, S.; Tsuzuki, K.; Nakagawa, A,; Lukacs, G. J. Antibiot. 1983, 36,
| GBF.
611-613.
University of Go¨ttingen.
(7) Byrne, K. M.; Shafiee, A.; Nielsen, J. B.; Arison, B.; Monaghan, R. L.;
Kaplan, L. DeV. Ind. Microbiol. 1993, 32, 29-45.
(1) O’Hagan, D. The Polyketide Metabolites; Ellis Horwood: New York, 1991.
(2) Rawlings, B. J. Nat. Prod. Rep. 2001, 18, 190-227 and 231-281.
(3) Hatano, K.; Mizuta, E.; Akiyama, S.; Higashide, E.; Nakao, Y. Agric. Biol.
Chem. 1985, 49, 327-333.
(8) Bindseil, K. U.; Zeeck, A. Liebigs Ann. Chem. 1994, 305-312.
(9) Schuhmann, T.; Grond, S. J. Antibiot. 2004, 57, 655-661.
(10) Ono, M.; Sakuda, S.; Ikeda, H.; Furihata, K.; Nakayama, J.; Suzuki, A.;
Isogai, A. J. Antibiot. 1998, 51, 1019-1028.
(4) (a) Hill, A., M.; Harris, J. P.; Siskos, A. P. J. Chem. Soc., Chem. Commun.
1998, 2361-2362. (b) Gerth, K.; Pradella, S.; Perlova, O.; Beyer, S.; Mu¨ller,
R. J. Biotechnol. 2003, 106, 233-253.
(11) (a) Gra¨fe, U.; Kluge, H.; Thiericke, R. Liebigs Ann. Chem. 1992, 429-
432. (b) Zhao, C.; Ju, J.; Christenson, S. D.; Smith, W. C.; Song, D.; Zhou,
X.; Shen, B.; Deng, Z. J. Bacteriol. 2006, 188, 4142-4147.
(5) Haber, A.; Johnson, R. D.; Rinehart, K. L., Jr. J. Am. Chem. Soc. 1977,
99, 3541-3544.
(12) Li, W.; Ju, J.; Osada, H.; Shen, B. J. Bacteriol. 2006, 188, 4148-4152.
9
10.1021/ja064408t CCC: $33.50 © 2006 American Chemical Society
J. AM. CHEM. SOC. 2006, 128, 14325-14336
14325

A R T I C L E S
Wenzel et al.
Scheme 1 a
be 2-hydroxy- or 2-methoxymalonyl-CoA. This was, however,
found not to be the case.9,13 Instead it was discovered that the
biosynthetic gene clusters for antibiotics carrying such “glyco-
late” units contain subclusters of four or five genes which are
responsible for synthesizing the substrate for incorporating the
“glycolate” unit. In the ansamitocin cluster (asm),14 as well as
the FK-520,15 geldanamycin,16 concanamycin,17 oxazolomycin,11b
and tautomycin12 clusters, this operon of five genes encodes
two oxidoreductases (asm13 and asm15), an acyl carrier protein
(ACP) (asm14), a methyltransferase (MT, asm17), and a pro-
tein of unknown function (asm16). In the soraphen cluster (sor)
the asm16 and asm17 equivalents are replaced by a single gene,
sorC, encoding a multifunctional protein with an acyltrans-
ferase (AT), an ACP, and a MT domain.18 Mutational and
heterologous expression experiments have demonstrated that
asm13-17 are necessary and sufficient for the synthesis of the
substrate for incorporation of a “glycolate” unit, deduced to be
2-methoxymalonyl-ACP, and for its delivery to the cognate AT
module of a PKS.13,19 The crystal structure of FkbI (correspond-
ing to Asm15), an acyl-CoA dehydrogenase homologue from
the fkb methoxymalonate subcluster, supports this notion,
predicting that the enzyme uses an ACP rather than a CoA ester
as substrate.20
a Reagents and conditions: (i) TFA, DMAP, 60 °C, Br2, then BnOH;
86.5%; (ii) PPH3, toluene, RT, 3 days, 96%; (iii) Et2O/H2O, K2CO3, 35 °C,
24 h; (iv) (CH2O)n, Et2O, reflux 24 h, 91% over two steps; (v) K3Fe(CN)6,
K2CO3, (DHQ)2-PHAL, OsO4, tBuOH/H2O, 0 °C, 10 h, 83%; (vi) THF/
H2O, Pd/C, H2, RT, 3.5 h, quantitative; (vii) THF, LiBH4, 0 °C, quantitative;
(viii) K3Fe(CN)6, K2CO3, (DHQD)2-PHAL , OsO4, tBuOH/H2O, 0 °C, 10 h,
84%.
However, the primary metabolism substrate for the synthesis
of the methoxymalonyl thioester has remained elusive. Numer-
ous isotopic tracer experiments on several of the antibiotics
containing “glycolate” units have tried to explore this question,
but no consistent picture has emerged from these studies.3-10
In this paper we report the results of additional feeding experi-
ments with the ansamitocin producer, Actinosynnema pretiosum,
and the soraphen producer, Sorangium cellulosum, which further
narrow the possibilities, pointing toward 1,3-bisphosphoglycerate
as the primary substrate, and define the orientation in which
this three-carbon substrate is incorporated into the two carbons
of the “glycolate” unit.
Scheme 2
Results and Discussion
Synthesis of Labeled Precursors. To probe further the origin
of the “glycolate” units in AP3 (1) and soraphen A (2), we aimed
to feed the enantiomers of glycerate and stereospecifically
labeled glycerol, each carrying labels of two contiguous 13C
atoms. Since these were not commercially available, we adapted
a route for the synthesis of nonstereospecifically labeled [1,2-
13C2]glycerate and -glycerol published by Pitlik and Townsend21
for their preparation. Starting from glacial [1,2-13C2]acetic acid
(99+% 13C), the R- and S-[1,2-13C2]glycerols were prepared in
63% overall yield as shown in Scheme 1. The asymmetric
dihydroxylation22 of benzyl acrylate gave benzyl glycerates of
75% ee for the D (R) and 69% ee for the L (S) isomers, as deter-
mined by enzymatic phosphorylation of the derived glycerol
samples with glycerokinase/ATP and 13C NMR analysis of the
ratio of sn-[1,2-13C2]- and sn-[2,3-13C2]glycerol 3-phosphate in
the products (Scheme 2). D-[1,2-13C2]Glycerate and L-[1,2-13C2]-
glycerate were derived from the benzyl glycerates by hydro-
genolysis, and R- and S-[1,2-13C2]glycerol, by LiBH4 reduction.
Reduction of the benzyl glycerates with LiBD4 also gave the
corresponding glycerol samples carrying two deuterium atoms
at C-1.
(13) Carroll, B. J.; Moss, S. J.; Bai, L.; Kato, Y.; Toelzer, S.; Yu, T.-W.; Floss,
H. G. J. Am. Chem. Soc. 2002, 124, 4176-4177.
(14) Yu, T.-W.; Bai, L.; Clade, D.; Hoffmann, D.; Toelzer, S.; Trinh, K. Q.;
Xu, J.; Moss, S. J.; Leistner, E.; Floss, H. G. Proc. Natl. Acad. Sci. U.S.A.
2002, 99, 7968-7973.
(15) Wu, K.; Chung, L.; Revill, W. P.; Katz, L.; Reeves, C. D. Gene 2000,
251, 81-90.
Independent of these syntheses, the group at the Los Alamos
Stable Isotope Resource developed a new, more stereoselective
route to R- and S-[1,2-13C2]glycerol (Scheme 3). This route is
based on chiral sulfoxides23 with the key reaction being the
asymmetric reduction of a â-keto sulfoxide to a â-hydroxy sulf-
oxide.24 The condensation of commercially available S-methyl
(16) Rascher, A.; Hu, Z.; Viswanathan, N.; Schirmer, A.; Reid, R.; Nierman,
W. C.; Lewis, M.; Hutchinson, C. R. FEMS Microbiol. Lett. 2003, 218,
223-230.
(17) (a) Haydock, S. F.; Appleyard, A. N.; Mironenk, T.; Lester, J.; Scott, N.;
Leadlay, P. F. Microbiology 2005, 151, 3161-3169. (b) Hano, O.; Schmidt,
O.; Welzel, K.; Weber, T.; Grond, S.; Wehmeier, U. Unpublished results.
(18) Ligon, J.; Hill, S.; Beck, J.; Zirkle, R.; Molnar, I.; Zawodny, J.; Money,
S.; Schupp, T. Gene 2002, 285, 257-267.
(19) Kato, Y.; Bai, L.; Xue, Q.; Revill, W. P.; Yu, T.-W.; Floss, H. G. J. Am.
Chem. Soc. 2002, 124, 5268-5269.
(22) (a) Sharpless, K. B.; Amberg, W.; Bennani, Y. L.; Crispino, G. A.; Hartung,
J.; Jeong, K. S.; Kwong, H. L.; Morikawa, K.; Wang, Z. M.; Xu, D.; Zhang,
X.-L. J. Org. Chem. 1992, 57, 2768-2771. (b) Kolb, H. C.; Van-
Nieuwenhze, M. S.; Sharpless, K. B. Chem. ReV. 1994, 94, 2483-2547.
(23) Solladie, G. Synthesis 1981, 185-196.
(20) Watanabe, K.; Khosla, C.; Stroud, R. M.; Tsai, S.-C. J. Mol. Biol. 2003,
334, 435-444.
(21) Pitlik, J.; Townsend, C. A. J. Labelled Cmpd. Radiopharm. 1997, 39, 999-
1009.
9
14326 J. AM. CHEM. SOC. VOL. 128, NO. 44, 2006

Biosynthetic Origin of Methoxymalonyl-Acyl Carrier Protein
A R T I C L E S
Scheme 3 a
a Reagents and conditions: (i) LDA, THF, -78 °C - 0 °C, 30 min, ethyl 2O-benzyl-[1,2-13C2]glycolate, 0 °C to RT, 2 h; (ii) DIBAL, CH2Cl2, -78 °C,
30 min, 62% (over two steps); (iii) NaOAc, Ac2O, reflux, 15 h, 94%; (iv) LiAlH4, THF, 0 °C to RT, 1.5 h, 91%, (v) 10% Pd/C, H2 (45 psi), Amberlyst (H+),
EtOH, RT, 4.5 h, 98%.
p-tolyl sulfoxide 3 with ethyl 2-O-benzyl-[1,2-13C2]glycolate
gave â-keto sulfoxide 4, which was then reduced with DIBAL
in THF at -78 °C to form â-hydroxy sulfoxide 5 in 62% overall
yield from S-methyl p-tolyl sulfoxide 3 (Scheme 3). Subjecting
â-hydroxy sulfoxide 5 to Pummerer rearrangement conditions,
by heating with acetic anhydride and sodium acetate, gave (2R)-
1,2-diacetoxy-3-benzyloxy-1-(p-tolylthio)-[2,3-13C2]propane 6
in high yield. Reductive deprotection of 6 with lithium aluminum
hydride gave benzyl glycerol 7 in 91% yield, hydrogenation of
which in the presence of 10% Pd/C catalyst yielded S-[1,2-13C2]-
glycerol. Substituting R-methyl p-tolyl sulfoxide for the S enanti-
omer, this route was used to produce R-[1,2-13C2]glycerol. The
enantiomeric purity of these glycerol samples was again deter-
mined by phosphorylation with glycerol kinase and 13C NMR
analysis.
moieties and the acetate-derived carbons, also supports this
suggestion.6 Feeding experiments with glycerol point to a
product of glycolysis as the precursor, showing labeling of C-2
of the “glycolate” units from [2-13C]glycerol in soraphen4a and
leucomycin6 and C-1 from [1(3)-13C]glycerol in concanamycin.8
These observations are somewhat at variance with the labeling
pattern from C-6 of glucose, suggesting that the metabolic path
from glucose to the “glycolate” unit is complex.
The earliest studies on the origin of the “glycolate” unit, by
Rinehart and co-workers on geldanamycin, showed labeling of
C-1 of the “glycolate” moiety by C-1 of glycolic acid.5 This
would support a pathway similar to the incorporation of acetic
acid into acetate-derived units, i.e., by activation to a thioester
followed by carboxylation. Low incorporation of [1-13C]-
glycolate into C-1 of the five “glycolate” units of aflastatin10
has also been reported, but similar experiments in several other
systems (leucomycin,6 concanamycin8) gave no incorporation.
Permeability barriers have been invoked as a possible explana-
tion for this lack of incorporation. However, the absence of a
carboxylase gene in the subclusters of genes known to be
necessary and sufficient for the formation of methoxymalonyl-
ACP and the strict requirement for two oxidoreductases argues
strongly against a process of carboxylation of an activated
glycolate. Rather, the utilization of glycolate in some systems
must involve an indirect process, possibly via conversion to
glycerate.
[1-13C]Glycerate was found to label rather specifically and
efficiently C-1 of the “glycolate” units of geldanamycin,5
although it was not incorporated at all into leucomycin.6 If
glycerate were activated in the geldanamycin producer to the
3-phosphate, it would then enter the glycolytic pathway and
could be converted into methoxymalonyl-ACP either directly
or via intermediates such as hydroxypyruvate. Alternatively,
glycerate could be activated at the carboxyl group and converted
into methoxymalonyl-ACP without passing through the glyco-
lytic pathway. If glycerate enters metabolism via the glycolytic
pathway, the labeling of C-1 of the “glycolate” units, but not
of acetate-derived units, would suggest that the free carboxyl
group of methoxymalonyl-ACP, which is lost in the condensa-
tion step on the PKS, is derived from the phosphorylated carbon,
C-3. This notion is indirectly supported by feeding experiments
with [1-13C]erythrose to the FK520 producer, which resulted
in labeling of C-1 of the two “glycolate” units, in addition to
the shikimate-derived starter unit.7 Following phosphorylation,
[1,2-13C2]Glycerol samples synthesized by the second route,
R ) 96% ee, S ) 90% ee, were used in the feeding experiments
with the soraphen A producer, whereas all other feeding experi-
ments employed the 13C-labeled glycerol and glycerate samples
of lower enantiomeric purity from the first synthesis.
Previous Feeding Experiments. Earlier isotopic tracer
experiments had shown that the two carbons of the “glycolate”
units of AP3 and soraphen A are not derived from acetate,
propionate, or glycine,3,4 consistent with the results in other
systems.6-10,25,26 This is not due to lack of precursor uptake, as
these compounds labeled other parts of the respective product
molecules, such as acetate and propionate units in the polyketide
chain or, in the case of C-2 of glycine, O- and N-methyl groups.
C-2, and to a lesser extent C-1, of the “glycolate” unit, i.e.,
C-10 and C-9 of AP3, were, however, somewhat enriched by
D-[6-13C]glucose, which also labeled most other positions of
the AP3 molecule.3 This finding is mirrored in low incorporation
into the corresponding carbons in geldanamycin5 and suggests
an involvement of carbohydrate metabolism. The incorporation
of [U-13C6]glucose into leucomycin to give contiguous labeling
of C-3 and C-4 (the “glycolate” unit), in addition to the sugar
(24) (a) Solladie, G.; Demailly, G.; Greck, C. J. Org. Chem. 1985, 50, 1552-
1554. (b) Solladie, G.; Demailly, G.; Greck, C. Tetrahedron Lett. 1985,
26, 435-438. (c) Solladie, G.; Hutt, J. Tetrahedron Lett. 1987, 28, 797-
800. (d) Solladie, G.; Adamy, M.; Colobert, F. J. Org. Chem. 1996, 61,
4369-4373. (e) Solladie, G.; Colobert, F.; Denni, D. Tetrahedron Asym-
metry 1998, 9, 3081-3094. (f) Solladie, G.; Hanquet, G.; Rolland, C.
Tetrahedron Lett. 1999, 40, 177-180. (g) Solladie, G.; Hanquet, G.; Izzo,
I.; Crumbie, R. Tetrahedron Lett. 1999, 40, 3071-3074.
(25) Johnson, R. D.; Haber, A.; Rinehart, K. L., Jr. J. Am. Chem. Soc. 1974,
96, 3316-3317.
(26) Omura, S.; Nakagawa, A.; Takeshima, H.; Atsumi, K.; Miyazawa, Y.;
Piriou, F.; Lukacs, G. J. Am. Chem. Soc. 1975, 97, 6600-6602.
9
J. AM. CHEM. SOC. VOL. 128, NO. 44, 2006 14327

A R T I C L E S
Wenzel et al.
[1-13C]erythrose would give rise, via the pentose phosphate
shunt, to triose phosphates and phosphoglyceric acid (PGA)
labeled at C-1. On the other hand, feeding experiments with
R- and S-[1-D2]glycerol showed incorporation of deuterium from
the R but not the S isomer at C-11 and C-14 of soraphen, carbons
representing C-1 of a “glycolate” and C-2 of an acetate unit,
respectively. This led Hill et al.4a to conclude that the pro-R
hydroxymethyl group of glycerol, which is phosphorylated by
glycerokinase and gives rise to C-3 of triose phosphates and
PGA, is retained and labels C-1 of the “glycolate” unit, whereas
the pro-S hydroxymethyl group, corresponding to C-1 of PGA,
is lost. The authors also concluded from the incorporation of
deuterium at C-11 of soraphen that the formation of the
“glycolate” unit cannot involve a 2-hydroxy- or 2-methoxy-
malonate intermediate.
from either enantiomer of glycerate was detected. However,
feeding of the same precursors to Sorangium cellulosum
unequivocally gave intact incorporation into the two “glycolate”
units, C-3/4 and C-11/12, of soraphen A (Table 1). Consis-
tent with the low chiral purity of the precursors, incorporation
was seen from both samples of glycerate, but the substantially
higher enrichment from D-[1,2-13C2]glycerate leaves little doubt
that D-glycerate is the true precursor. According to textbook
biochemistry, only D-glycerate is phosphorylated by glycerate
kinases, and only phospho-D-glycerates are intermediates in
glycolysis. The observation of single enrichment in the carbons
derived from C-1 of acetate (C-9, C-13 and C-15) and C-2 of
phosphoenolpyruvate (C-1′) indicates that D-glycerate enters the
glycolytic pathway, presumably via phosphoglycerate.
We then established the utility of glycerol as a precursor of
the “glycolate” unit in these systems by feeding [U-13C3]glycerol
to both A. pretiosum and S. cellulosum. The resulting AP3
indeed showed 13C-13C coupling between the carbons of the
three acetate units as well as between C-9 and C-10. Thus, gly-
cerol serves effectively as a substrate for the formation of both
the acetate and the “glycolate” units in A. pretiosum, supporting
the involvement of a triose phosphate or phosphoglyceric acid
in both processes. Identical results were obtained for the incor-
poration of [U-13C3]glycerol into soraphen A by S. cellulosum.
To elucidate which two carbons of a triose phosphate or
phosphoglycerate give rise to the “glycolate” unit, we then
determined the orientation of the incorporation of glycerol by
feeding to A. pretiosum R-[1,2-13C2]glycerol (75% ee) and
S-[1,2-13C2]glycerol (69% ee), each diluted 4-fold with unlabeled
glycerol to suppress statistical coupling. In the 13C NMR analysis
of the resulting AP3 samples, the average of the signals for
C-30, C-31, and C-32, carbons derived from unlabeled valine
added to the fermentation, was used as natural abundance
reference in the normalization of the spectra. The results, shown
in Table 2, reveal that the R isomer, in which the labeled
hydroxymethyl group is phosphorylated by R-glycerokinase,
gives rise to significant coupling between the two carbon atoms
of each of the three acetate units, but only a very small amount
of coupling between C-9 and C-10. The latter reflects the small
amount (12.5%) of S isomer present in the precursor. There
was, however, significant enrichment of C-10, as well as the
three O- and N-methyl groups and the carbamate carbon.
Coupling was also observed in the aromatic moiety between
C-16 and C-17 and between C-20 and C-21. This is consistent
with the formation from the labeled R-glycerol of [2,3-13C2]-
phosphoenolpyruvate and [3,4-13C2]erythrose 4-phosphate, which
via the aminoshikimate pathway27 then give rise to the starter
unit, 3-amino-5-hydroxybenzoic acid (AHBA) (Figure 2). The
S isomer of the glycerol, in contrast, gave significant coupling
between C-9 and C-10 of AP3, but single enrichments only in
the acetate units at the carboxyl-derived carbons. Consistent with
the aminoshikimate pathway, coupling was also seen between
C-15 and C-16 and C-19 and C-20, as well as enrichment at
C-18. In addition, the carbons corresponding to C-1 of the
propionate units and the carbamate carbon were enriched (Figure
2). The results thus show that methoxymalonyl-ACP, the
precursor of the “glycolate” unit, is derived from a triose
phosphate or phosphoglycerate such that C-1 gives rise to the
New Feeding Experiments. To probe the possible involve-
ment of 3-hydroxypyruvate in the formation of the “glycolate”
units, we fed serine to the ansamitocin producer, Actinosynnema
pretiosum. This compound, which can give rise to hydroxy-
pyruvate by transamination, had not been evaluated as a pre-
cursor of “glycolate” units in any of the previous work. Feeding
of D,L-[1-13C]serine gave no significant enrichment in any of
the carbon atoms of the resulting AP3. D,L-[3-13C]Serine labeled
(2-3.5% enrichment) only the three O- and N-methyl groups
and the carbamoyl carbon C-24 of AP3. No significant enrich-
ment at C-9 or C-10 was seen in either of the experiments, ruling
out serine or 3-hydroxypyruvate as precursor of the “glycolate”
unit. The involvement of the tricarboxylic acid (TCA) cycle in
the formation of the “glycolate” unit was probed by feeding
[1,2-13C2]succinate. This gave AP3 which showed no significant
13C-13C coupling, only low enrichment and no coupling at C-9
(0.8%) and C-10 (1.2%), but single-carbon enrichment (1.9-
2.5%) at each of the three carbons of the three propionate units,
at C-24 (2.6%) and to a lesser extent at the three acetate units
(1.4-1.6%). The data argue against any prominent role of the
TCA cycle in the generation of the “glycolate” units. The label-
ing pattern of the propionate units points to their formation from
succinate via the succinyl-CoA-methylmalonyl-CoA mutase
reaction.
In view of the labeling of the “glycolate” unit of AP3 by
[6-13C]glucose,3 we explored the mode of incorporation of this
sugar further by feeding D-[1,2-13C2]glucose. The amount of
precursor available was small (120 mg), and hence, the overall
level of incorporation was quite low. However, coupling was
clearly detected between the carbon pairs (C-1/C-2, C-7/C-8,
C-11/C-12) of the acetate units, whereas no coupling was
detectible between C-9 and C-10, the carbons of the “glycolate”
unit. The level of enrichment was too low to determine with
any confidence whether there was single-carbon enrichment at
either C-9 or C-10. These data suggest metabolism of glucose
via the glycolytic pathway and are consistent with, although
do not prove, incorporation via a triose phosphate or (phospho)-
glycerate, C-2 and C-3 of which give rise to acetate units and
C-1 and C-2 possibly to “glycolate” units.
The following experiments were then carried out to probe
the role of triose phosphates or (phospho)glycerate further. First,
the incorporation of glycerate into AP3 was evaluated by feeding
D-[1,2-13C2]glycerate (75% ee) and L-[1,2-13C2]glycerate (69%
ee). However, unlike in the case of geldanamycin,5 no enrich-
ment of, nor coupling between, any carbons of AP3 derived
(27) Yu, T.-W.; Mu¨ller, R.; Mu¨ller, M.; Zhang, H.; Draeger, G.; Kim, C.-.G.;
Leistner, E.; Floss, H. G. J. Biol. Chem. 2001, 276, 12546-12555 and
references therein.
9
14328 J. AM. CHEM. SOC. VOL. 128, NO. 44, 2006

Biosynthetic Origin of Methoxymalonyl-Acyl Carrier Protein
A R T I C L E S
S
R
1.
Tbale
9
J. AM. CHEM. SOC. VOL. 128, NO. 44, 2006 14329

A R T I C L E S
Wenzel et al.
Table 2. 13C-Enrichments and 13C-13C-Couplings in Ansamitocin P-3 Derived from R- and S-[1,2-13C2]Glycerol (150.8 MHz; numbers in red
indicate significant enrichments)
thioester carbonyl group, which becomes C-1 of the “glycolate”
unit and the phosphorylated C-3 gives rise to the free carboxyl
group which is lost in the subsequent Claisen condensation on
the PKS.
between C-3 and C-4 as well as C-11 and C-12, the two carbon
pairs of each “glycolate” unit, which therefore in this system,
too, are derived from the two carbons of glycerol which are
not phosphorylated by glycerol kinase. Coupling was also
observed between C-3′/C-5′ and C-4′ of the aromatic ring,
consistent with its shikimate pathway origin. Strong single-
carbon enrichments were seen at C-9, C-13, and C-15, the
carbons directly derived from the carboxyl group of acetate,
and at C-1′. The absence of coupling between C-1′ and C-17
supports the previously deduced28 formation of the benzoate
starter unit via phenylalanine rather than by direct aromatization
of a shikimate-derived C6C1 moiety. Since the levels of enrich-
ment were rather high in this experiment, lesser degrees of
single-carbon enrichments were seen in many other positions,
e.g., the O-methyl groups, some of the propionate-derived
carbons, C-17, C-4, and C-12, undoubtedly due to more exten-
sive metabolism of the administered precursor. The feeding
experiment with R-[1,2-13C2]glycerol gave complementary
results. Coupling was observed between the two carbon atoms
of each acetate-derived chain extension unit, but not within the
The same conclusion was reached in the feeding experiments
to Sorangium cellulosum with the stereochemically purer R- and
S-[1,2-13C2]glycerol samples. The quantitative NMR analysis
of the 13C enrichments in the soraphen A samples presented a
problem, since the molecule contains no carbon atoms of an
external origin which could be used as a bona fide reference
for the normalization of the spectra and the amounts of sample
were too small for chemical derivatization. For the normalization
of the soraphen A samples from D- and L-[1,2-13C2]glycerate
and S-[1,2-13C2]glycerol, we used the signal for C-2′/C-6′ and
for the R-[1,2-13C2]glycerol-derived sample we used the C-4′
carbon signal. According to biogenetic theory, these atoms are
not expected to be labeled by the respective precursors. If they
should carry some excess of 13C after all, this would result in
an underestimation of the enrichments of the other carbons.
Calculations using other carbon atoms as reference gave lower
apparent enrichments, i.e., were clearly underestimations. In
light of the above, the data shown in Table 1 are minimum
values. As summarized in Figure 2, the soraphen A from the
S configured [1,2-13C2]glycerol sample displayed coupling
(28) (a) Ho¨fle, G.; Reichenbach, H. In Sekunda¨rmetabolismus bei Mikroorgan-
ismen; Kuhn, W., Fiedler, H.-P., Eds.; Attempto Verlag: Tu¨bingen,
Germany, 1995; pp 61-78. (b) Hill, A. M.; Thompson, B. L.; Harris, J.
P.; Segret, R. Chem. Commun. 2003, 1358-1359.
9
14330 J. AM. CHEM. SOC. VOL. 128, NO. 44, 2006

Biosynthetic Origin of Methoxymalonyl-Acyl Carrier Protein
A R T I C L E S
Figure 2. Labeling patterns of ansamitocin P-3, soraphen A, and metabolic intermediates after feeding 13C-labeled glycerols and glycerates, and proposed
biosynthesis of methoxymalonyl-ACP, the precursor of “glycolate” units. Heavy bars denote contiguous 13C labels of connected atoms, dots denote single
13C labels.
two “glycolate” units. The latter instead showed single-carbon
enrichments at C-4 and C-12. Coupling was also observed in
the aromatic moiety between C-1′ and C-2′/C-6′ as well as C-2′/
C-6′ and C-3′/C-5′, as predicted by the shikimate pathway. In
addition, single-carbon enrichments were observed for C-17,
which originates from C-3 of phosphoenolpyruvate, the three
O-methyl groups, and some of the propionate-derived carbons.
These labeling patterns are well within expectations for the
metabolism of glycerol, following stereospecific phosphorylation
by glycerol kinase, via the glycolytic pathway and its known
ancillary reactions (Figure 2).
The results clearly demonstrate that C-1 and C-2 of the
“glycolate” units of both AP3 and soraphen A arise from C-1
and C-2, respectively, of a three-carbon glycolytic intermediate.
A similar conclusion was reached for the “glycolate” units of
concanamycin and bafilomycin.29 They also disprove the
interpretation of Hill et al. of their observation that R-[1-D2]-
glycerol results in deuterium incorporation at C-11 of soraphen
A, i.e., C-1 of the “glycolate” unit.4a Since the deuterium in the
precursor is attached to a carbon which itself is not incorporated
into the “glycolate” unit, the incorporation of the deuterium
obviously must represent an indirect process via redox reactions
and the cellular pyridine nucleotide pool. This interpretation
was considered by the authors but rejected as unlikely. Con-
sequently, their conclusion that the formation of the “glycolate”
unit cannot involve hydroxy- or methoxymalonate derivatives
as intermediates is also invalid.
Conclusions
All the present and previous results are consistent with and
support the formation of “glycolate” units from 2-methoxy-
malonyl-ACP, which in turn arises from phosphoglyceric acid,
phosphoglyceraldehyde, or a closely related glycolytic inter-
mediate. 1,3-Bisphospho-D-glyceric acid is the most likely sub-
strate, based on two lines of evidence. One is the very specific
incorporation of [1-13C]glyceric acid into the C-1 carbons of
the “glycolate” units of geldanamycin5 and the intact incorpora-
tion of D-[1,2-13C2]glyceric acid into the “glycolate” units of
soraphen A demonstrated in the present work. The lack of incor-
poration of glyceric acid in some systems, such as leukomycin6
or AP3, probably reflects the ability or inability of a given
organism to phosphorylate glyceric acid, i.e., the presence or
absence of an active D-glycerate kinase.
(29) Schuhmann T.; Grond S. New Insights into the “Glycolate” Unit Biosyn-
thesis in the Plecomacrolides Bafilomycin A1, B1 and Concanamycin A.
Manuscript in preparation.
9
J. AM. CHEM. SOC. VOL. 128, NO. 44, 2006 14331

A R T I C L E S
Wenzel et al.
The other argument relates to the mode of attachment of the
carboxyl group of the glycerate unit to the thiol group of the
ACP. This could involve 3-phosphoglyceric acid as substrate
and a carboxyl activation to the adenylate or the CoA thioester.
However, there is little support for such a process at the genetic
level. There is no obvious adenylation or other carboxyl-
activation gene in the “glycolate” subclusters,14,15 and Asm16
and SorC, which have been assigned the function of attaching
the three-carbon unit to the ACP,13 show no homology to known
carboxyl-activating enzymes. The absence of a dedicated gene
encoding a carboxyl-activating enzyme from the methoxy-
malonate subclusters suggests that the primary metabolism
substrate is already activated at the carboxyl group. This points
to 1,3-bisphospho-D-glycerate, a normal glycolytic intermediate,
which would be formed from glycerol directly via glycerol
1-phosphate and the triose phosphates and, in some organisms
only, from D-glycerate via 2- and/or 3-phospho-D-glycerate by
the action of a glycerate kinase and glycolytic enzymes. 1,3-
Bisphospho-D-glycerate can be transferred to the thiol group of
an ACP, Asm14 or the ACP domain of SorC, by Asm16 or the
AT domain of SorC, respectively. The necessary cleavage of
the phosphate ester group from the resulting 3-phosphoglyceryl-
ACP may also be catalyzed by Asm16, which shows at least
25-30% identity at the protein level to members of the IIIC
subgroup of the HAD superfamily of aspartate-dependent
hydrolases (Interpro IPR010033).30,31 The functions of many
of these enzymes are unknown, but all the characterized mem-
bers of the HAD III subfamily are phosphatases31,32 apparently
using large substrates (phosphorylated proteins),33 and the family
includes the FkbH protein which shows up to 64% identity at
the protein level to Asm16 and its homologues (Figure 3).
Whether SorC also has such a function is not clear, since it
shows no homology to Asm16 or to known phosphatases; the
sor cluster also contains no other Asm16 homologue. Another
possibility, which, however, we consider less likely, would be
a reaction of 3-phosphoglyceraldehyde with the thiol group of
the ACP to form a thiohemiacetal, which would then be oxidized
to the thioester by one of the dehydrogenases encoded in the
methoxymalonate subclusters, possibly Asm15 and SorE. The
reverse of this reaction appears to be responsible for a reductive
product release mechanism from polyketide/nonribosomal pep-
tide synthetases by reductive cleavage of acyl-ACPs.34
gene product consisting of a dedicated ACP, an AT domain,
and a methyltransferase domain. It seems likely that SorC is
responsible not only for the attachment of the glyceryl moiety
to the dedicated ACP but also its subsequent O-methylation,13
i.e., that it catalyzes the formation of 2-O-methyl-D-glyceryl-
ACP from 1,3-bisphosphoglycerate. In analogy we propose that
Asm17 acts before Asm13 and Asm15. We thus propose as a
working hypothesis the pathway shown in Figure 2 for the
formation of methoxymalonyl-ACP and hence of the “glycolate”
unit.
The finding that D-glycerate, and thus presumably 1,3-
bisphospho-D-glycerate, is the precursor of the “glycolate” unit
predicts that the 2-methoxymalonyl-ACP must have R config-
uration at C-2. Since decarboxylative Claisen condensations on
the PKS are known to proceed with inversion of configuration,35
the direct utilization of (2R)-methoxymalonyl-ACP without
epimerization would result in “glycolate” units of L configura-
tion.36 A survey of the configurations of known, chiral “gly-
colate” units shows a majority of them, C-10 of AP3, C-4 of
soraphen A, C-6 of geldanamycin, C-4 of leucomycin, C-13 of
FK-520/FK-506, C-14 of bafilomycin, C-16 of concanamycin,
C-4 of oxazolomycin, and C-23 of tautomycin, to be
D
configured, whereas a minority, C-12 of soraphen A and
geldanamycin and C-15 of FK-520/FK-506, have L configura-
tion. Thus, in at least some of the “glycolate” units the chain
extension substrate must have undergone epimerization either
before or after its incorporation into the polyketide. In the
analogous case of the chain extension by propionate units, only
(2S)-methylmalonyl-CoA is used as substrate but gives rise to
both D and L configured propionate units.37 Work on module 2
of the erythromycin PKS, which generates a propionate unit
with an epimerized methyl-bearing center, has shown that the
configuration at this center is controlled by the ketoreductase
domain which reduces the adjacent carbonyl group and is
stereoselective for only the epimerized substrate.38 However,
the authors stress that different mechanisms may control the
methyl stereochemistry in other PKS modules. Such a mecha-
nism can clearly not operate in the formation of the “glycolate”
units at C-10 of AP3 and C-4 of soraphen A, because in both
cases the adjacent carbonyl group has not been reduced.
Substantially more work will be necessary to unravel the
complex stereochemical control mechanisms governing these
chain extension reactions.
The conversion of 1,3-bisphosphoglycerate into methoxy-
malonyl-ACP involves connecting C-1 to the thiol group of the
ACP to become the thioester carbonyl group of methoxy-
malonyl-ACP. Methylation of the oxygen at C-2 has been
proposed as the terminal step of the conversion sequence to
methoxymalonyl-ACP.15 However, we consider it more likely
that this methylation occurs at the stage of D-glyceryl-ACP,
before the stepwise oxidation of C-3 to a carboxyl group. This
suggestion is based on the sequence of the sorC gene of the
sor biosynthetic gene cluster18 which encodes a trifunctional
Experimental Section
1
Materials and General Methods. H- and 13C NMR spectra were
recorded on Bruker AM-300, DRX-300, ARX-400, DRX-499, and
INOVA-600 NMR spectrometers in either CDCl3, D2O, or [D6]DMSO
as solvent. For determination of 13C enrichments the peak integral from
the enriched sample was compared with that of a control sample of
natural abundance, using the average of the signals for C-30, C-31,
and C-32 (AP3), the C-2′/C-6′ or the C-4′ signal (soraphen A) as natural
abundance reference. For HPLC, a Beckman System Gold program-
mable solvent module was used with a Beckman System Gold diode
(30) Koonin, E. V.; Tatusov, R. L. J. Mol. Biol. 1994, 244, 125-132.
(31) Selengut, J. D. Biochemistry 2001, 40, 12704-12711.
(32) (a) Galburt, E. A.; Pelletier, J.; Wilson, G.; Stoddart, B. L. Structure 2002,
10, 1249-1260. (b) Parsons, J. F.; Lim, K.; Tempczyk, A.; Krajewski,
W.; Eisenstein, E.; Herzberg, O. Proteins 2002, 46, 393-404. (c) Wu, J.;
Woodard, R. W. J. Biol. Chem. 2003, 278, 18117-18123. (d) Peisach, E.;
Selengut, J. D.; Dunaway-Mariano, D.; Allen, K. N. Biochemistry 2004,
43, 12770-12779.
(33) Allen, K. N.; Dunaway-Mariano, D. Trends Biochem. Sci. 2004, 29, 495-
502.
(34) Gaitatzis, N.; Kunze, B.; Mu¨ller, R. Proc. Natl. Acad. Sci. U.S.A. 2001,
98, 11136-11141.
(35) (a) Sedgwick, B.; Cornforth, J. W.; French, S. J.; Gray, R. T.; Kelstrup,
E.; Willadsen, P. Eur. J. Biochem. 1977, 75, 481-495. (b) Weissman, K.
J.; Timoney, M.; Bycroft, M.; Grice, P.; Hanefeld, U.; Staunton, J.; Leadlay,
P. F. Biochemistry 1997, 36, 13849-13855.
(36) Defined as a Fischer projection with the carboxy terminus of the polyketide
chain at the top.
(37) Wiesmann, K. E. H.; Cortes, J.; Brown, M. J. B.; Cutter, A. L.; Staunton,
J.; Leadlay, P. F. Chem. Biol. 1995, 2, 583-589.
(38) Holzbaur, I. E.; Harris, R. C.; Bycroft, M.; Cortes, J.; Bisang, C.; Staunton,
J.; Rudd, B. A. M.; Leadlay, P. F. Chem. Biol. 1999, 6, 189-195.
9
14332 J. AM. CHEM. SOC. VOL. 128, NO. 44, 2006

Biosynthetic Origin of Methoxymalonyl-Acyl Carrier Protein
A R T I C L E S
Figure 3. Alignment of Asm16 and its equivalents from the concanamycin (Orf7), FK-520 (FkbH), geldanamycin (GdmH), tautomycin (TtmE), and
oxazolomycin (OzmB) biosynthetic gene clusters (GenBank accession numbers: AAF86387 (Asm16), AAZ94395 (Orf7), AAF86387 (FkbH), AAO06922
(GdmH), AAZ08062 (TtmE), ABA39082 (OzmB)). Based on the alignment with its homologues, the translation start of Asm16 was repositioned resulting
in an N-terminal extension of 28 amino acids (shown in bold) in comparison to the published sequence (AAF86387). The HAD-superfamily phosphatase
subfamily IIIC domain (Interpro IPR010033) is highlighted in gray.
array detector. Fermentations were carried out with shaking in an Adolf
Ku¨hner ISF-4-V temperature-controlled shaker cabinet. Chemicals were
purchased from Aldrich, biochemicals from Sigma, and D-[1,2-13C2]-
glucose, [U-13C3]glycerol, D,L-[1-13C]serine, D,L-[3-13C]serine, and
9
J. AM. CHEM. SOC. VOL. 128, NO. 44, 2006 14333

A R T I C L E S
Wenzel et al.
1
[1,2-13C2]succinate from Cambridge Isotope Laboratories. Anhydrous
[1,2-13C2]acetic acid for the synthesis of [1,2-13C2]glycerate and [1,2-
13C2]-glycerol was provided by the Los Alamos Stable Isotope Resource.
Synthesis of Labeled Glycerates and Glycerols via Asymmetric
Dihydroxylation. Benzyl [1,2-13C2]Bromoacetate. [1,2-13C2]Acetic
acid (10.0 mL, 10.5 g, 170 mmol), trifluoroacetic anhydride (40.0 mL,
59.5 g, 383 mmol) and DMAP (0.213 g, 1.74 mmol, 0.01 equiv) were
heated to 60 °C. Bromine (9.40 mL, 29.3 g, 183 mmol, 1.05 equiv)
was added at 60 °C at a rate such that the orange color was just
maintained (90 min). Stirring was continued at 60 °C for another hour.
The reaction mixture was then brought to room temperature and purged
with argon for 2 h to remove HBr and unreacted bromine. Benzyl
alcohol (100 mL, 966 mmol) was added carefully at room temperature,
and the reaction mixture was then heated at 60 °C overnight. After
cooling to room temperature it was poured into concentrated NaHCO3
solution (100 mL). Diethyl ether (150 mL) was added, the phases were
separated, and the aqueous phase was extracted three more times with
ether (100 mL each). The combined organic phases were washed with
concentrated NaHCO3 solution (100 mL), concentrated Na2S2O3 solution
(100 mL), and water (100 mL) and were dried over MgSO4. Chroma-
tography (silica gel, ethyl acetate/hexanes, 5:95) gave the title compound
glycerate (19.66 g, 84%, ee: 69%). H NMR was identical to that of
D isomer.
Sodium D- and L-[1,2-13C2]Glycerate. Benzyl D-[1,2-13C2]glycerate
(1.926 g, 9.73 mmol) was dissolved in THF (14 mL) and water (6 mL).
Pd-C (196 mg) was added, and the mixture was stirred vigorously
under an atmosphere of H2 for 3.5 h, with more Pd-C (100 mg) being
added after 2 h. When TLC showed no more educt, the reaction mixture
was filtered and the filtrate evaporated to dryness. The residue was
dissolved in water (10 mL), neutralized with solid NaHCO3 (828 mg,
9.86 mmol), and evaporated to dryness to give sodium D-[1,2-13C2]-
1
glycerate as a colorless glass (1.296 g, 9.95 mmol, 102%). H NMR
(300 MHz, D2O): δ 3.66 - 3.95 (m, 3-Hab), 4.29-4.42 (m, 2-H). 13
C
NMR (75 MHz, D2O): δ 61.4 (d, J ) 40.1 Hz, C-3), 70.6 (d, J ) 53.8
Hz, C-2, enriched), 176.1 (d, J ) 53.8 Hz, C-1, enriched).
In the same way, benzyl L-[1,2-13C2]glycerate (1.93 g, 9.75 mmol)
gave sodium L-[1,2-13C2]glycerate (1.249 g, 9.61 mmol, 98.6%). NMR
spectra were identical to those of D sample.
R- and S-[1,2-13C2]Glycerol. Benzyl D-[1,2-13C2]glycerate (7.785
g, 39.28 mmol) was dissolved in THF (130 mL). The solution was
cooled to 0 °C and stirred, and LiBH4 (2 M in THF, 69.0 mL, 138
mmol, 3.5 equiv) was added dropwise. With continued stirring, the
reaction mixture was allowed to warm to room temperature over 30
min and then neutralized with water (50 mL) and 2 N HCl. The organic
phase was evaporated, and the aqueous phase was washed with ethyl
acetate (3 × 100 mL). Cations were removed from the aqueous phase
by passage through a column of ion-exchange resin AG 50 H+ with
water as eluent. The remaining borate in the aqueous phase was
removed by evaporation to dryness followed by repeated addition of
methanol and evaporation of the resulting borate ester to give R-[1,2-
1
(34.0 g, 147 mmol, 86.5%). H NMR: δ 3.88 (dd, JCH2,2-13C ) 153.4
Hz, JCH2,1-13C ) 4.7 Hz, CH2Br), 5.21 (d, JOCH2,1-13C ) 3.3 Hz, OCH2),
7.33-7.41 (m, Ar).
Benzyl [1,2-13C2]Acrylate. Benzyl [1,2-13C2]bromoacetate (34.0 g,
147 mmol) and triphenylphosphine (42.6 g, 162 mmol, 1.1 equiv) were
added to toluene (350 mL). The reaction mixture was stirred at room
temperature for 3 days. The Wittig salt was filtered, washed with
toluene, and dried under vacuum. The Wittig salt (69.7 g, 141 mmol)
was dissolved in the two-phase system diethyl ether/water (600/300
mL), and K2CO3 (19.5 g, 141 mmol, 1.00 equiv) was added over
15 min. The reaction mixture was stirred at 35 °C for 24 h. The layers
were separated, the aqueous phase was extracted with diethyl ether
(3 × 200 mL), and the combined organic phases were dried over
MgSO4. Paraformaldehyde (4.44 g, 148 mmol, 1.05 equiv) was added
to the organic phase, and the reaction mixture was heated under reflux
for 24 h. The solvent was evaporated, and chromatography (silica gel,
20% ethyl acetate in hexanes) gave the title compound (21.02 g, 87%
based on benzyl bromoacetate). 1H NMR (500 MHz): δ 5.21 (d,
JOCH2,1-13C ) 3.1 Hz, CH2O), 5.85 (ddd, Jcis-H,2-13C ) 14.3 Hz, Jcis-H,2-H
) 10.4 Hz, Jcis-H,trans-H ) 1.4 Hz, cis-H), 6.17 (dddd, J2-H,2-13C ) 164.1
Hz, J2-H,trans-H ) 17.3 Hz, J2-H,cis-H ) 10.3 Hz, J2-H,1-13C ) 3.9 Hz,
2-H), 6.45 (dddd, Jtrans-H,2-H ) 17.3 Hz, Jtrans-H,2-13C ) 7.6 Hz,
Jtrans-H,1-13C ) 3.4 Hz, Jtrans-H,cis-H ) 1.4 Hz, trans-H), 7.30-7.42 (m,
Ar).
13C2]glycerol (3.697 g, 100%). H NMR (500 MHz, D2O): δ 3.36-
1
3.70 (m, 2 × CH2OH), 3.73-3.94 (m, CHOH). 13C NMR (125 MHz,
D2O): δ 62.5 (d, J ) 41 Hz, C-1), 72.0 (d, J ) 41 Hz, C-2).
In the same way, benzyl L-[1,2-13C2]glycerate (7.849 g, 39.60 mmol)
gave S-[1,2-13C2]glycerol (3.727 g, 100%). NMR spectra identical to
those of the R sample.
Determination of Enantiomeric Purity. R-[1,2-13C2]Glycerol (20
mg, 0.21 mmol) was dissolved in water (2 mL). HEPES buffer (500
mg, ca. 10 equiv) and ATP (300 mg, ca. 2.5 equiv) were added, and
the solution was adjusted with NaOH (1 and 0.1 N) to pH 7. Glycerol
kinase (100 units) was added, and the reaction mixture was gently
shaken for 20 h at 37 °C. The reaction mixture was then analyzed by
13C NMR. The ratio of the integral of the signals at 64.9 ppm
(CH2OPO32-) and 62.3 ppm (CH2OH) was 1:0.1461, corresponding to
75% ee R.
The same procedure was used with S-[1,2-13C2]glycerol to give a δ
64.9 vs 62.3 ratio of 0.1826:1, corresponding to 69% ee S.
Benzyl D-[1,2-13C2]Glycerate. K3Fe(CN)6 (116.1 g, 355.5 mmol,
3 equiv), K2CO3 (48.59 g, 355.5 mmol, 3 equiv), (DHQ)2-PHAL
(0.9243 g, 1.185 mmol, 0.01 equiv), and K2OsO2(OH)4 (0.0877 g, 0.237
mmol, 0.002 equiv) were dissolved in tert-butyl alcohol (600 mL) and
water (600 mL). Stirring at room temperature produced two clear
phases. The mixture was cooled to 0 °C. Benzyl [1,2-13C2]acrylate
(19.46 g, 118.5 mmol) was added and the reaction mixture was stirred
at 0 °C for 10 h. Na2SO3 (177.8 g, 1.411 mol) was added at 0 °C, the
reaction mixture was allowed to warm to room temperature and stirred
for 1 h. Ethyl acetate (500 mL) was added, and the phases were
separated. The aqueous phase was extracted with ethyl acetate (3 ×
400 mL). The combined organic phases were dried over MgSO4, and
chromatography (silica gel, EtOAc/hexanes, 1:1) gave the title com-
Synthesis of Labeled Glycerols via Chiral Sulfoxides. Ethyl 2-O-
Benzyl-[1,2-13C2]glycolate. 2-O-Benzyl-[1,2-13C2]glycolic acid (8.48
g, 50.4 mmol) was heated to reflux in EtOH (80 mL) in the presence
of Amberlyst-[H+] resin. After 10 h, the reaction mixture was allowed
to cool to RT and was filtered through Celite. The Celite was rinsed
with EtOH (8 × 25 mL), and the filtrate was concentrated in vacuo.
Purification by flash column chromatography (EtOAc/hexane, 1:6)
1
yielded the product as a clear, colorless oil (8.77 g, 89%). H NMR
(CDCl3): δ 1.27 (t, J ) 7.2 Hz, 3H), 4.09 (dd, JCH ) 143.7 Hz, J )
1
4.5 Hz, 2H), 4.21 (dq, J ) 7.2 Hz, 3 Hz, 2H), 4.64 (d, J ) 4.2 Hz,
2H), 7.27-7.39 (m, 5H). 13C NMR (CDCl3): δ 14.4, 60.9, 67.48 (str,
d, 1JCC ) 62.9 Hz), 73.6, 128.5, 128.1, 128.1, 137.3, 170.6 (str, d, 1JCC
) 62.9 Hz). IR (neat, cm-1): 2981.4, 2872.8, 1707.9, 1454.1, 1424.7,
1234.2, 1182.5, 1109.7, 1029.2. Anal. Calcd for C1513C2H19O3S: C,
68.35; H, 7.19. Found: C, 68.06; H, 7.02.
1
pound (19.47 g, 83%, ee: 75%). H NMR (500 MHz): δ 3.42 (br s,
2 × OH), 3.84 (ddd, Jab ) 11.7 Hz, J3-Ha,2-H ) 4.0 Hz, J3-Ha,2-13C
)
1.1 Hz, 3-Ha), 3.87-3.93 (m, 3-Hb), 4.30 (dddd, J2-H, ) 146.5
2-13C
Hz, J2-H,3-Ha ) 4.0 Hz, J2-H,3-Hb ) 4.0 Hz, J2-H,1-13C ) 4.0 Hz, 2-H),
S-3-Benzyloxy-1-(p-tolylsulfinyl)-2-[2,3-13C2]propanone (4). A
2.5 M solution of n-butyllithium in hexanes (5 mL, 12.5 mmol) was
added dropwise to a solution of diisopropylamine (1.8 mL, 12.8 mmol)
in THF (15 mL), under argon, and was cooled to -78 °C. The reaction
5.20 (dd, Jab ) 12.2 Hz, JHa,1-13C ) 3.1 Hz, OCHaHb), 5.24 (dd, Jab
12.2 Hz, JHb,1-13C ) 3.0 Hz, OCHaHb), 7.25-7.40 (m, Ar).
The same procedure, using (DHQD)2-PHAL as reagent and benzyl
[1,2-13C2]acrylate (19.46 g, 118.5 mmol), gave benzyl L-[1,2-13C2]-
)
9
14334 J. AM. CHEM. SOC. VOL. 128, NO. 44, 2006

Biosynthetic Origin of Methoxymalonyl-Acyl Carrier Protein
A R T I C L E S
mixture was allowed to reach 0 °C and was stirred for 30 min. The
LDA solution was then added to a solution of S-methyl p-tolyl sulfoxide
3 (0.92 g, 5.97 mmol) in THF (10 mL) at 0 °C. After stirring for 30
min, the sulfoxide anion solution was added to a solution of ethyl 2-O-
benzyl-[1,2-13C2]glycolate (1.17 g, 5.97 mmol) in THF (20 mL). The
reaction mixture was allowed to warm to RT. After stirring for 2 h,
the reaction was quenched with saturated NH4Cl (50 mL), acidified to
pH 2 with 5% H2SO4, extracted with CH2Cl2 (3 × 50 mL), and dried
over Na2SO4. The solvent was removed in vacuo to give 2.19 g of
crude product 4 as a yellow crystalline solid, which was used in the
next step without further purification.
(2R,SS)-3-Benzyloxy-1-(p-tolylsulfinyl)-2-[2,3-13C2]propanol (5).
A solution of DIBAL in CH2Cl2 (1 M, 7.2 mL, 7.2 mmol) was added
dropwise to a solution of the crude â-ketosulfoxide 4 in THF (40 mL)
at -78 °C. After stirring for 30 min, the reaction was quenched with
saturated NH4Cl (30 mL) and allowed to warm to RT, acidified to pH
4 with 5% H2SO4, extracted with CH2Cl2 (3 × 50 mL), and dried over
Na2SO4, and the solvent was removed in vacuo. The crude product
(1.85 g) was purified by flash column chromatography (EtOAc/hexanes,
S-[1,2-13C2]Glycerol. Compound 7 (515 mg, 2.80 mmol) was
dissolved in EtOH (15 mL) in a thick-walled reaction flask; 10% Pd/C
(80 mg) and Amberlyst-15 (H+) resin (50 mg) were added, and the
reaction vessel was connected to a Parr hydrogenator, with an H2
atmosphere of 45 psi. The reaction was complete after shaking for 4.5 h.
The reaction mixture was filtered through Celite and rinsed with EtOH,
and the solvent was removed in vacuo to yield (S)-[1,2-13C2]glycerol
as a clear, colorless oil (257 mg, 98%). 1H NMR (300 MHz, D2O): δ
1
1
3.55-3.74 (m, 2H), 3.65 (dm, JCH ) 142 Hz, 2H), 3.84 (dm, JCH
)
1
142 Hz, 1H). 13C NMR (75 MHz, D2O): δ 62.5 (d, JCC ) 40.9 Hz),
1
72.1 (d, JCC ) 40.9 Hz). IR (neat, cm-1): 3294.1, 2927.2, 2874.8,
1416.2, 1321.0, 1205.4, 1091.3, 1031.6. Anal. Calcd for C13C2H8O3:
C, 40.41; H, 8.57. Found: C, 40.53; H, 8.72.
By the same procedures, R-methyl-p-tolyl sulfoxide (1.16 g, 7.53
mmol) gave R-[1,2-13C2]glycerol (298 mg, 3.16 mmol) in 42% overall
yield.
Biological Experiments. Feeding Experiments with Actinosyn-
nema pretiosum. For resting cell experiments, Actinosynnema pretiosum
ssp. auranticum ATCC 31565 was grown on yeast-malt extract agar
plates at 28 °C for 4 days and then stored at 0 °C. A loop of A.
pretiosum was used to inoculate 25 mL of YMG medium (yeast extract
0.4%, malt extract 1.0%, glucose 0.4%, pH adjusted to 7.4 with NaOH
before autoclaving) in a 250-mL Erlenmeyer flask with a stainless steel
coil and incubated for 24 h at 29 °C with shaking at 225 rpm. One
milliliter of this starter culture was used to inoculate each 100 mL of
YMG medium in a 500-mL Erlenmeyer flask with a spring, all of which
were incubated for 48 h at 29 °C with shaking at 225 rpm. These were
then harvested by centrifugation to form a pellet, which was washed
three times with 20 mL each of sterile double distilled (DDI) water
and then resuspended in 20 mL of DDI water. The suspension was
added to a 500-mL Erlenmeyer flask with a spring containing 80 mL
of sterile, resting cell medium (20 mL of 0.25 M lactose, 10 mL of
0.05 M MgCl2, 10 mL of 0.05 M sodium isobutyrate, 10 mL of 0.1 M
Tris × HCl pH 8.5, and 30 mL of DDI water) and incubated at 29 °C
with shaking at 225 rpm. Labeled precursors in the total amounts
indicated were added to the number of cultures indicated in three equal
portions at times 0, 24, and 48 h and the cultures harvested at 72 h to
give the amounts of purified AP3 indicated: D,L-[1-13C]serine, 150 mg,
10 cultures, 2 mg of AP3; D,L-[3-13C]serine, 150 mg, 10 cultures, 9.4
mg of AP3; [1,2-13C2]succinate, 150 mg, six cultures, 5.1 mg of AP3;
D-[1,2-13C2]glucose, 120 mg, 10 cultures, 5.8 mg of AP3; [U-13C3]-
glycerol, 150 mg, 10 cultures, 3.0 mg of AP3. At the end of the
fermentation period the combined cultures were extracted three times
with an equal volume of ethyl acetate, the combined extracts were dried,
and the solvent was evaporated to give a viscous oil. This residue was
dissolved in methanol and passed through a C18 cartridge (2.5 cm ×
0.5 cm). The eluent was purified by HPLC with UV detection (234
and 280 nm) either on a YMC Pack ODS-AQ column (5 µm, 250 mm
× 10 mm) equipped with a YMC guard column (30 mm × 10 mm)
with a gradient of water/methanol, 1:1 to 1:9, over 25 min (flow rate
3 mL/min, tret AP3 ) 21.5 min), or on an Alltech Econosil C18 10U
column (250 mm × 22.5 mm), also equipped with a guard column,
with water/methanol, 2:3 (flow rate 20 mL/min, tret AP3 ) 11 min).
75:25) to give the product 5 as a pale-yellow crystalline solid (1.13 g,
1
62% from 3). [R]22 -166.4 (c 1.21, CHCl3). H NMR (300 MHz,
D
CDCl3): δ 2.42 (s, 3H), 2.79-2.74 (m, 1H), 3.12-3.01 (m, 1H), 3.49
1
1
(dm, JCH ) 141.9 Hz, 2H), 3.65 (br s, 1H), 4.38 (dm, JCH ) 146.1
Hz, 1H), 4.56-4.59 (m, 2H), 7.25-7.53 (m, 9H). 13C NMR (75 MHz,
CDCl3): δ 21.6, 66.1 (d, JCC ) 41.8 Hz), 66.2 (str, d, JCC ) 42.1
1
1
1
Hz), 73.4 (str, d, JCC ) 42.1 Hz), 74.1, 124.2, 127.9, 128.1, 128.7,
130.3, 141.8. IR (neat, cm-1): 3302, 2859, 1491, 1446, 1359, 1295,
1112, 1079, 1011. Anal. Calcd for C1513C2H20O3S: C, 67.29; H, 6.58.
Found: C, 67.52; H, 6.57.
(2R)-1,2-Diacetoxy-3-benzyloxy-1-(p-tolylthio)-[2,3-13C2]pro-
pane (6). A mixture of 5 (1.68 g, 5.49 mmol) and NaOAc (4.51 g,
54.94 mmol) was heated at reflux in Ac2O (60 mL) for 15 h. The acetic
anhydride was removed by azeotropic distillation with toluene (3 ×
50 mL). The residue was dissolved in CH2Cl2 (25 mL) and filtered to
remove the NaOAc. The filtrate was concentrated and purified by flash
column chromatography (EtOAc/hexanes, 20:80) to yield a mixture of
the two diastereoisomers 6 as a pale-yellow oil (2.02 g, 94%). 1H NMR
(300 MHz, CDCl3): δ 1.99 (s, 3H), 2.03 (s, 3H), 2.05 (s, 3H), 2.08 (s,
1
3H), 2.32 (s, 3H), 3.68 (dm, JCH ) 141.8 Hz, 4H), 4.43-4.59 (m,
4H), 5.30 (dm, 1JCH ) 147.6 Hz, 2H), 6.25-6.29 (m, 2H), 7.07-7.39
(18H). 13C NMR (75 MHz, CDCl3): δ 21.0, 21.1, 21.4, 68.2 (str, d,
1
1
1JCC ) 43.8 Hz), 68.3 (str, d, JCC ) 43.5 Hz), 69.3, 72.6 (str, d, JCC
1
) 43.8 Hz), 72.7 (str, d, JCC ) 43.5 Hz), 73.5, 128.0, 128.6, 128.6,
130.1, 130.1, 134.0, 134.3, 137.8, 138.9, 139.1, 170.2, 169.3. IR (neat,
cm-1): 3030.8, 2860.8, 1744.6, 1492.6, 1453.0, 1369.7, 1207.3, 1017.8.
Anal. Calcd for C1913C2H24O5S: C, 65.11; H, 6.20. Found: C, 65.20;
H, 6.12.
(2S)-1-Benzyloxy-2,3-[1,2-13C2]propanediol (7). LiAlH4 (697 mg,
18.36 mmol) was added to a solution of 6 (1.20 g, 3.06 mmol) in THF
(17 mL) at 0 °C. The reaction mixture was allowed to warm to RT.
After 1.5 h, the mixture was cooled to 0 °C and cautiously quenched
with saturated aqueous NH4Cl (50 mL). After stirring for 20 min, the
mixture was diluted with CH2Cl2 (50 mL) and acidified to pH 1 with
5% H2SO4. The aqueous fraction was extracted with CH2Cl2 (4 × 50
mL), and the combined organics were dried over Na2SO4 and concen-
trated under reduced pressure. The crude oil was purified by flash
column chromatography (EtOAc/hexanes, 75:25) to give the desired
Subsequent experiments used improved fermentation conditions. A
single colony from a YMG agar plate of Actinosynnema pretiosum ssp.
auranticum ATCC 31565 was used to inoculate 100 mL of YMG
medium in a 500-mL Erlenmeyer flask with coil, which was grown
for 48 h at 30 °C with shaking at 220 rpm. Portions of 0.5 mL of this
stock culture were mixed with 0.5 mL of sterilized 40% aqueous
glycerol and stored at -20 °C. To prepare seed cultures, 1 mL of stock
culture was used to inoculate 100 mL of YMG medium in a 500-mL
Erlenmeyer flask with coil, which was grown for 48 h at 30 °C with
shaking at 220 rpm. Eight mL of seed culture was then used to inoculate
each 100 mL of production medium39 in a 500-mL Erlenmeyer flask
with a coil; 12 mL of a filter-sterilized 3% solution of L-valine was
added to each flask, and the cultures were grown for 9 days at 30 °C
product 7 as a dark-yellow oil (604 mg, 91%). [R]25 -2.95 (c 3.73,
D
1
C6H6). H NMR (300 MHz, CDCl3): δ 2.10 (br s, 1H), 2.61 (br s,
1H), 3.28-3.39 (m, 1H), 3.61-3.85 (m, 3H), 3.89 (dm, 1JCH ) 139.8
Hz, 1H), 4.56 (d, J ) 3.6 Hz, 2H), 7.27-7.39 (m, 5H). 13C NMR (75
MHz, CDCl3): δ 64.1 (dd, J ) 38.1, 4.1 Hz), 70.97 and 71.72 (str,
ABq, 1JCC ) 41.8 Hz), 73.6, 127.9, 127.9, 128.6, 137.9. IR (neat, cm-1):
3355.2, 3062.5, 3029.9, 2857.6, 1453.1, 1362.9, 1206.1, 1060.0,
1027.9. Anal. Calcd for C813C2H14O3: C, 66.28; H, 7.66. Found: C,
66.16; H, 7.73.
9
J. AM. CHEM. SOC. VOL. 128, NO. 44, 2006 14335

A R T I C L E S
Wenzel et al.
with shaking at 220 rpm. Labeled precursors were added in the total
amounts indicated to the number of cultures indicated in eight equal
portions every 24 h starting at day 1 and gave the amounts of purified
AP3 indicated: R- And S-[1,2-13C2]glycerol, 80 mg mixed with 320
mg of unlabeled glycerol, five cultures each, 24.5 and 25.4 mg of AP3,
respectively; sodium D- and L-[1,2-13C2]glycerate, 80 mg, five cultures
each, 34.3 and 40.7 mg of AP3, respectively. AP3 was isolated by
extracting the fermentation broth 3× with an equal volume of ethyl
acetate and evaporating the combined, dried extract to dryness. The
residue was taken up in 2 mL of chloroform/methanol (9:1) and loaded
onto a column (2 cm diameter) of 50 g of neutral aluminum oxide.
The column was washed with 100 mL of chloroform, after which AP3
was eluted with chloroform containing 5% methanol. This process was
repeated once to obtain pure AP3. The production medium39 consists
of (per liter): 60 g of dextrin, 30 g of maltose, 5.25 g of cotton seed
flour (ProFlo), 4.5 g yeast extract, 5 g of CaCO3, 0.5 g of K2HPO4,
2 mg of FeSO4 × 7 H2O; the medium is brought to a boil and slowly
cooled off before autoclaving.
The pH of the medium was adjusted to 7.2 with KOH before auto-
claving. From a 4-day-old preculture, fresh, 100-mL cultures were
inoculated with 10% of the seed culture and incubated in 250-mL
Erlenmeyer flasks at 30 °C and 180 rpm on a rotary shaker. One
hundred milligrams of each of the respective labeled compounds was
dissolved in 6 mL of distilled water and sterilized by filtration. After
3 days, 6 days, and 9 days of cultivation 2 mL of each of the respective
labeled compounds was added from the stock solutions above. Fifteen
days after the start of the cultivation, the adsorber resin XAD was
harvested by sieving and eluted with 100 mL of methanol. After evap-
oration of the methanol, the resulting aqueous layer was made up to
50 mL and extracted with ethyl acetate (2 × 50 mL). Evaporation of
the solvent yielded an oily residue, which was dissolved in 500 µL of
60% methanol and subjected to sequential preparative RP-HPLC
(Nucleosil C18, solvent 50% methanol) to yield 1-2 mg of soraphen A.
Acknowledgment. We gratefully acknowledge V. Wray and
co-workers of the Department of Structural Research at the
Gesellschaft fu¨r Biotechnologische Forschung (GBF) for record-
ing NMR spectra. This work was supported by the U.S. National
Institutes of Health (Grant CA 76461 to H.G.F.), the Deutsche
Forschungsgemeinschaft (grants to R.M. and S.G.), the BMBF
(grant to R.M.), and the DAAD (fellowship to C.G.). The Los
Alamos Stable Isotope Resource is funded by Grant NIH/NIBIB
5P41 EB002166.
Feeding Experiments with Sorangium cellulosum So ce26. Sor-
angium cellulosum So ce26 Y2 was cultivated in H medium [soybean
meal (soyamine 50T Lucas Meyer, Hamburg) 0.2%; glucose (Maizena)
0.2%; starch (Cerestar SF 12618, Cerestar Deutschland Krefeld) 0.8%;
yeast extract (Marcor) 0.2%; CaCl2 × 2 H2O 0.1%; MgSO4 × 7 H2O
0.1%; ethylenediaminetetraacetic acid iron(III)-sodium salt 0.0008%;
HEPES 1.19%; XAD-16 (Rohm and Haas, Frankfurt/M) 2% (w/v)].
(39) Srinivasulu, B.; Kim, Y.-J.; Chang, Y.-K.; Shang, G.; Yu, T.-W.; Floss,
H. G. Unpublished work (manuscript in preparation).
JA064408T
9
14336 J. AM. CHEM. SOC. VOL. 128, NO. 44, 2006