Synthesis and fate of o-carboxybenzophenones in the biosynthesis of aflatoxin

Article abstract of DOI:10.1021/ja045520z

o-Carboxybenzophenones have long been postulated to be intermediates in the oxidative rearrangement of anthraquinone natural products to xanthones in vivo. Many of these Baeyer-Villiger-like cleavages are believed to be carried out by cytochrome P450 enzymes. In the biosynthesis of the fungal carcinogen, aflatoxin, six cytochromes P450 are encoded by the biosynthetic gene cluster. One of these, AflN, is known to be involved in the conversion of the anthraquinone versicolorin A (3) to the xanthone demethylsterigmatocystin (5) en route to the mycotoxin. An aryl deoxygenation, however, also takes place in this overall transformation and is proposed to be due to the requirement that an NADPH-dependent oxidoreductase, AflM, be active for this process to take place. What is known about other fungal anthraquinone → xanthone conversions is reviewed, notably, the role of the o-carboxybenzophenone sulochrin (25) in geodin (26) biosynthesis. On the basis of mutagenesis experiments in the aflatoxin pathway and these biochemical precedents, total syntheses of a tetrahydroxy-o-carboxybenzophenone bearing a fused tetrahydrobisfuran and its 15-deoxy homologue are described. The key steps of the syntheses entail rearrangement of a 1,2-disubstituted alkene bearing an electron-rich benzene ring under Kikuchi conditions to give the 2-aryl aldehyde 43 followed by silyltriflate closure to a differentially protected dihydrobenzofuran 44. Regiospecific bromination, conversion to the substituted benzoic acid, and condensation with an o-bromobenzyl alcohol gave esters 47 and 50. The latter could be rearranged with strong base, oxidized, and deprotected to the desired o-carboxybenzophenones. These potential biosynthetic intermediates were examined in whole-cell and ground-cell experiments for their ability to support aflatoxin formation in the blocked mutant DIS-1, defective in its ability to synthesize the first intermediate in the pathway, norsolorinic acid. Against expectation, neither of these compounds was converted into aflatoxin under conditions where the anthraquinones versicolorin A and B readily afforded aflatoxins B1 and B2. This outcome is evaluated further in a companion paper appearing later in this journal.

Full text of DOI:10.1021/ja045520z

Published on Web 02/19/2005
Synthesis and Fate of o-Carboxybenzophenones in the
Biosynthesis of Aflatoxin
Kevin M. Henry and Craig A. Townsend*
Contribution from the Department of Chemistry, The Johns Hopkins UniVersity,
Charles and 34th Streets, Baltimore, Maryland 21218
Received July 26, 2004; E-mail: ctownsend@jhu.edu
Abstract: o-Carboxybenzophenones have long been postulated to be intermediates in the oxidative
rearrangement of anthraquinone natural products to xanthones in vivo. Many of these Baeyer-Villiger-like
cleavages are believed to be carried out by cytochrome P450 enzymes. In the biosynthesis of the fungal
carcinogen, aflatoxin, six cytochromes P450 are encoded by the biosynthetic gene cluster. One of these,
AflN, is known to be involved in the conversion of the anthraquinone versicolorin A (3) to the xanthone
demethylsterigmatocystin (5) en route to the mycotoxin. An aryl deoxygenation, however, also takes place
in this overall transformation and is proposed to be due to the requirement that an NADPH-dependent
oxidoreductase, AflM, be active for this process to take place. What is known about other fungal
anthraquinone f xanthone conversions is reviewed, notably, the role of the o-carboxybenzophenone
sulochrin (25) in geodin (26) biosynthesis. On the basis of mutagenesis experiments in the aflatoxin pathway
and these biochemical precedents, total syntheses of a tetrahydroxy-o-carboxybenzophenone bearing a
fused tetrahydrobisfuran and its 15-deoxy homologue are described. The key steps of the syntheses entail
rearrangement of a 1,2-disubstituted alkene bearing an electron-rich benzene ring under Kikuchi conditions
to give the 2-aryl aldehyde 43 followed by silyltriflate closure to a differentially protected dihydrobenzofuran
44. Regiospecific bromination, conversion to the substituted benzoic acid, and condensation with an
o-bromobenzyl alcohol gave esters 47 and 50. The latter could be rearranged with strong base, oxidized,
and deprotected to the desired o-carboxybenzophenones. These potential biosynthetic intermediates were
examined in whole-cell and ground-cell experiments for their ability to support aflatoxin formation in the
blocked mutant DIS-1, defective in its ability to synthesize the first intermediate in the pathway, norsolorinic
acid. Against expectation, neither of these compounds was converted into aflatoxin under conditions where
the anthraquinones versicolorin A and B readily afforded aflatoxins B1 and B2. This outcome is evaluated
further in a companion paper appearing later in this journal.
Both heme and non-heme oxygenases play prominent roles
in creating the structural diversity of every natural product class.
While these enzymes typically catalyze hydroxylation reactions
at C-H bonds, they also can mediate pivotal oxidative cleavage,
insertion, desaturation, and rearrangement chemistry that is much
less well understood. In the biosynthesis of the potent environ-
mental carcinogen aflatoxin B1 (11, AFB1) from Aspergillus
parasiticus and A. flaVus, at least four of six cytochromes P450
are thought to play critical roles among the ca. 15 transforma-
tions that convert the earliest polyketide-derived intermediate,
the anthraquinone norsolorinic acid (1, NA), to the highly
modified coumarin product 11 (Scheme 1).1,2
One such P450 (AflG)3 contributes to the multistep conver-
sion of the hexanone side chain of NA to the tetrahydrobisfuran
of versicolorin B (2, VB; Scheme 1).4-10 A second monooxy-
genase (AflL) catalyzes the desaturation of VB to versicolorin
A (3, VA).11-13 Not all of the available VB is consumed by
AflL in vivo, and the resulting substrate/product mixture
constitutes the first branch point of the biosynthetic pathway.12,14
Gene disruption experiments suggest that the anthraquinones
VB and VA convert to xanthone bisfurans 4 and 5, respectively,
through the joint action of a cytochrome P450 (AflN)15 and an
NADPH-dependent ketoreductase (AflM).13,16,17 This complex
transformation is the focus of the experiments that follow.
(3) The genetic nomenclature of the aflatoxin biosynthetic pathway is designated
throughout this and the accompanying paper according to the recently
published ordering of Yu et al. (Yu, J.; Chang, P. K.; Ehrlich, K. C.; Cary,
J. W.; Bhatnagar, D.; Cleveland, T. E.; Payne, G. A.; Linz, J. E.; Woloshuk,
C. P.; Bennett, J. W. Appl. EnViron. Microbiol. 2004, 70, 1253-1262). In
this review, the aflatoxin genes, originally named according to their
respective substrate or enzymatic function, were standardized with the three-
letter code “afl” and arranged alphabetically according to their ordering in
vivo: aflG ) aVnA, aflL ) VerB, aflM ) Ver1, aflN ) aflS, aflO ) omt1,
aflP ) omtII, and aflQ ) ordA.
(4) Townsend, C. A.; Christensen, S. B.; Davis, S. G. J. Am. Chem. Soc. 1982,
104, 6154-6155.
(5) Yabe, K.; Nakamura, Y.; Nakajima, H.; Ando, Y.; Hamasaki, T. Appl.
EnViron. Microbiol. 1991, 57, 1340-1345.
(6) Townsend, C. A.; Davis, S. G.; Koreeda, M.; Hulin, B. J. Org. Chem.
1985, 50, 5428-5430.
(7) Townsend, C. A.; Isomura, Y.; Davis, S. G.; Hodge, J. A. Tetrahedron
1989, 45, 2263-2276.
(8) Townsend, C. A.; Whittamore, P. R. O.; Brobst, S. W. Chem. Commun.
1988, 726-728.
(9) Townsend, C. A.; Christensen, S. B.; Davis, S. G. J. Chem. Soc., Perkin
Trans. 1 1988, 839-861.
(1) Keller, N. P.; Watanabe, C. M. H.; Kelkar, H. S.; Adams, T. H.; Townsend,
C. A. Appl. EnViron. Microbiol. 2000, 66, 359-362.
(2) Minto, R. E.; Townsend, C. A. Chem. ReV. 1997, 97, 2537-2555.
(10) McGuire, S. M.; Townsend, C. A. Bioorg. Med. Chem. Lett. 1993, 3, 653-
656.
9
3300
J. AM. CHEM. SOC. 2005, 127, 3300-3309
10.1021/ja045520z CCC: $30.25 © 2005 American Chemical Society

o-Carboxybenzophenones in the Biosynthesis of Aflatoxin
A R T I C L E S
Scheme 1
Successive O-methylations then are catalyzed by substrate-
specific SAM-dependent O-methyltransferases (AflO followed
by AflP) during the conversion of demethylsterigmatocystins 4
and 5 to the O-methylsterigmatocystins 7 and 9.18-23 Recently,
we confirmed and extended the observation of Wolochuk et al.
that the coumarins aflatoxin B2 and B1 (10 and 11, respectively)
are formed through the action of a single P450 (AflQ) from the
O-methylsterigmatocystins (7 and 9, respectively; OMST).24 We
showed that this remarkable sequence takes place through two
oxidative cycles of this enzyme to first hydroxylate 9 at C-10
to 10-hydroxyOMST (12, Scheme 1) and then to cleave the
xanthone A-ring followed by reclosure, decarboxylation, and
demethylation.25,26 The loss of C-10 uniquely at the oxidation
state of carbon dioxide had been established earlier in a sensitive
radiochemical assay27 and provides important support to the idea
that two cycles of P450-catalyzed oxidation are required.26
Additionally, the G-series (13/14) and B-series (10/11) aflatoxins
arise in parallel from the AflQ-catalyzed oxidation of OMSTs
(7 and 9) during a final branching step of the biosynthetic
pathway. Although the mechanism for lactone formation has
not been established, Yabe conclusively demonstrated that the
B-series coumarins are not precursors of the G-series.28 Thus,
the biosynthetic pathway concludes with aflatoxins B2 (10) and
G2 (13) containing the fully saturated tetrahydrobisfuran moiety
and aflatoxins B1 (11) and G1 (14, Scheme 1) containing the
electron-rich dihydrobisfuran, the lethal seat of its activation
as a carcinogen.29-31
Although of unknown mechanism, the conversion of an-
thraquinones 2/3 to xanthones 4/5 has been characterized by
genetic and biochemical experiments. Genetic disruptions of
either AflN15 or AflM13,16 led to the accumulation of VA and
blocked the formation of sterigmatocystin (8, ST). Pathway
intermediates were not detected from these experiments, and
notably, separate disruptions of either gene yielded the same
phenotype. The corresponding in trans complementation of each
gene successfully restored xanthone formation.
Insights into the anthraquinone f xanthone conversion have
been gained in isotope labeling experiments. For example, a
single labeling pattern was observed from the incorporation of
[1,2-13C2]acetate into sterigmatocystin (8a, ST; Scheme 2).
Therefore, rings A and C from VA must never become C2-
symmetric during the course of the rearrangement of the
anthraquinone to the xanthone framework.32-34 Incorporation
of oxygen (*) from 18O2 into ring C of ST suggested a Baeyer-
(11) McGuire, S. M.; Brobst, S. W.; Graybill, T. L.; Pal, K.; Townsend, C. A.
J. Am. Chem. Soc. 1989, 111, 8308-8309.
(12) Yabe, K.; Ando, Y.; Hamasaki, T. Agr. Biol. Chem. 1991, 55, 1907-1911.
(13) Keller, N. P.; Kantz, N. J.; Adams, T. H. Appl. EnViron. Microbiol. 1994,
60, 1444-1450.
(14) Yabe, K.; Ando, Y.; Hamasaki, T. Appl. EnViron. Microbiol. 1988, 54,
2101-2106.
(15) Keller, N. P.; Segner, S.; Bhatnagar, D.; Adams, T. H. Appl. EnViron.
Microbiol. 1995, 61, 3628-3632.
(16) Skory, C. D.; Chang, P. K.; Cary, J.; Linz, J. E. Appl. EnViron. Microbiol.
1992, 58, 3527-3537.
(17) Liang, S. H.; Skory, C. D.; Linz, J. E. Appl. EnViron. Microbiol. 1996, 62,
4568-4575.
(18) Yabe, K.; Matsushima, K.; Koyama, T.; Hamasaki, T. Appl. EnViron.
Microbiol. 1998, 64, 166-171.
(19) Motomura, M.; Chihaya, N.; Shinozawa, T.; Hamasaki, T.; Yabe, K. Appl.
EnViron. Microbiol. 1999, 65, 4987-4994.
(20) Keller, N. P.; Dischinger, H. C.; Bhatnagar, D.; Cleveland, T. E.; Ullah,
A. H. J. Appl. EnViron. Microbiol. 1993, 59, 479-484.
(21) Keller, N. P.; Dischinger, H. C.; Bhatnagar, D.; Cleveland, T. E.; Ullah,
A. H. J. Faseb J. 1991, 5, A822.
(28) Yabe, K.; Nakamura, M.; Hamasaki, T. Appl. EnViron. Microbiol. 1999,
65, 3867-3872.
(29) Groopman, J. D.; Cain, L. G.; Kensler, T. W. Crit. ReV. Toxicol. 1988, 19,
113-145.
(22) Yu, J. J.; Cary, J. W.; Bhatnagar, D.; Cleveland, T. E.; Keller, N. P.; Chu,
F. S. Appl. EnViron. Microbiol. 1993, 59, 3564-3571.
(23) Yu, J. J.; Chang, P. K.; Payne, G. A.; Cary, J. W.; Bhatnagar, D.; Cleveland,
T. E. Gene 1995, 163, 121-125.
(30) Eaton, D. L.; Gallagher, E. P. Annu. ReV. Pharmacol. Toxicol. 1994, 34,
135-172.
(31) Muench, K. F.; Misra, R. P.; Humayun, M. Z. Proc. Natl. Acad. Sci. U.S.A.
1983, 80, 6-10.
(24) Prieto, R.; Woloshuk, C. P. Appl. EnViron. Microbiol. 1997, 63, 1661-
1666.
(32) Pachler, K. G. R.; Steyn, P. S.; Vleggaar, R.; Wessels, P. L. Chem. Commun.
1975, 355-356.
(25) Casillas, L. K.; Townsend, C. A. J. Org. Chem. 1999, 64, 4050-4059.
(26) Udwary, D. W.; Casillas, L. K.; Townsend, C. A. J. Am. Chem. Soc. 2002,
124, 5294-5303.
(33) Simpson, T. J.; Stenzel, D. J. Chem. Commun. 1982, 890-892.
(34) Sankawa, U.; Shimada, H.; Kobayashi, T.; Ebizuka, Y.; Yamamoto, Y.;
Noguchi, H.; Seto, H. Heterocycles 1982, 19, 1053-1058.
(27) Chatterjee, M.; Townsend, C. A. J. Org. Chem. 1994, 59, 4424-4429.
9
J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005 3301

A R T I C L E S
Scheme 2
Henry and Townsend
Scheme 3
Villiger-like addition to the anthraquinone carbonyl and re-
giospecific insertion of molecular oxygen following migration
of ring C.35,36
of VA, 6-deoxyVA43 (15, Scheme 1), was not converted to
AFB1 in either whole-cell44 or cell-free experiments45 despite
its structural similarity to 19, 22, and 27. If reduction is not the
first step in the anthraquinone f xanthone conversion, oxidation
must initiate the overall process. Potential roles for the P450
AflN can be visualized by considering the fate of other
anthraquinones in related biosynthetic pathways.
Three cases become germane, and each is thought to involve
a Baeyer-Villiger oxidation (Scheme 4). In the first, the
spirodienone (+)-geodin (26)46 arises from oxidative phenolic
coupling of sulochrin (25), a reaction known to be catalyzed
by a blue copper protein.47-49 Sulochrin would appear to arise
by Baeyer-Villiger cleavage of questin (24) and obvious
methylation.50-52 Next, secalonic acid A (29)53-55 and ravenelin
(31)56 are known to be derived from emodin (23) and chrysoph-
anol (27). The biosyntheses of these fungal metabolites are
thought to proceed by way of Baeyer-Villiger oxidation of
chrysophanol to 28. By analogy to sulochrin (25) formation,
one and two such Baeyer-Villiger oxidations of 27 are proposed
to yield intermediates 28 and 30, respectively. The free
carboxylate of 28 may be lost as carbon dioxide in the formation
of 30. In each case, aryl epoxidation of these benzophenone
A key mechanistic question is whether AflN-mediated
oxidation takes place before or after AflM-catalyzed reduction.
Important precedent for anthraquinone reduction as the first step
of xanthone formation can be found in the well-documented
NADPH-dependent conversion of emodin (23) to chrysophanol
(27, Scheme 4).37,38 One can visualize these reactions as
occurring directly on the keto-tautomer of the C-6 phenol or
by conjugate addition at C-6 stabilized by the anthraquinone
C-9 carbonyl and elimination of hydroxide. Similarly, the
NADPH-dependent reductases (99% identical to each other)39
that function on 1,3,6,8-tetrahydroxy (16, T4HN)- and 1,3,8-
trihydroxynaphthalene (19, T3HN) early in melanin biosynthesis
are 64% identical to those of AflM in a BLASTP primary
1
sequence alignment (Scheme 3).40 Interestingly, the H NMR
signals of T4HN (16) in acetone corresponded to ca. 30% of
the keto-tautomer 17 in equilibrium with the phenolic form,
while no such signals were evident in the corresponding
spectrum of T3HN.41 These observations argue that reduction
of the keto-tautomer takes place in these reactions. Consistent
with this view, the biosynthesis of actinorhodin (22) in Strep-
tomyces coelicolor includes an ActIII ketoreductase-mediated
deoxygenation of a seemingly pre-aromatic cyclohexadienone
intermediate 20.42 The predicted amino acid sequence of ActIII
is 33% identical and 53% similar to that of AflM. Against
expectation, however, the corresponding deoxygenated analogue
(43) Fukuyama, K.; Ashida, T.; Katsube, Y.; Kakudo, M. Bull. Chem. Soc. Jpn.
1979, 52, 677-683.
(44) Graybill, T. L.; Pal, K.; McGuire, S. M.; Brobst, S. W.; Townsend, C. A.
J. Am. Chem. Soc. 1989, 111, 8306-8308.
(45) Watanabe, C. M. H.; Townsend, C. A. J. Am. Chem. Soc. 1998, 120, 6231-
6239.
(46) Barton, D. H. R.; Scott, A. I. J. Chem. Soc. 1958, 1767-1772.
(47) Huang, K. X.; Fujii, I.; Ebizuka, Y.; Gomi, K.; Sankawa, U. J. Biol. Chem.
1995, 270, 21495-21502.
(48) Fujii, I.; Iijima, H.; Tsukita, S.; Ebizuka, Y.; Sankawa, U. J. Biochem.
(Tokyo) 1987, 101, 11-18.
(35) Nakashima, T. T.; Vederas, J. C. Chem. Commun. 1982, 206-208.
(36) Watanabe, C. M. H.; Townsend, C. A. J. Org. Chem. 1996, 61, 1990-
1993.
(37) Anderson, J. A.; Lin, B. K.; Williams, H. J.; Scott, A. I. Biochemistry 1988,
27, 3087.
(49) Fujii, I.; Iijima, H.; Ebizuka, Y.; Sankawa, U. Chem. Pharm. Bull. 1983,
31, 337-340.
(50) Curtis, R. F.; Parry, D. R.; Hassall, C. H. J. Chem. Soc., Perkin Trans. 1
1972, 240.
(51) Fujii, I.; Ebizuka, Y.; Sankawa, U. J. Biochem. (Tokyo) 1988, 103, 878-
(38) Anderson, J. A.; Lin, B. K.; Williams, H. J.; Scott, A. I. J. Am. Chem. Soc.
1988, 110, 1623-1624.
883.
(52) Chen, Z. G.; Fujii, I.; Ebizuka, Y.; Sankawa, U. Arch. Microbiol. 1992,
158, 29-34.
(39) Vidalcros, A.; Viviani, F.; Labesse, G.; Boccara, M.; Gaudry, M. Eur. J.
Biochem. 1994, 219, 985-992.
(53) Andersen, R.; Buchi, G.; Kobbe, B.; Demain, A. L. J. Org. Chem. 1977,
42, 352-353.
(40) Altschul, S. F.; Madden, T. L.; Schaffer, A. A.; Zhang, J.; Zhang, Z.; Miller,
W.; Lipman, D. J. Nucleic Acids Res. 1997, 25, 3389-3402.
(41) Vivani, F.; Gaudry, M.; Marquet, A. J. Chem. Soc., Perkin Trans. 1 1990,
1255.
(54) Howard, C. C.; Johnstone, R. A. W.; King, T. J.; Lessinger, L. J. Chem.
Soc., Perkin Trans. 1 1976, 1820-1822.
(55) Steyn, P. S. Tetrahedron 1970, 26, 51.
(42) McDaniel, R.; Ebert-Khosla, S.; Hopwood, D. A.; Khosla, C. J. Am. Chem.
Soc. 1994, 116, 10855-10859.
(56) Raistrick, H.; Robinson, R.; White, D. E. Biochem. J. 1936, 30, 1303-
1314.
9
3302 J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005

o-Carboxybenzophenones in the Biosynthesis of Aflatoxin
A R T I C L E S
Scheme 4
intermediates can be invoked (among a limited number of
possible mechanisms) to achieve xanthone closure. All three
metabolites 26, 29, and 31 arise from benzophenone precursors
comprising one symmetrical aromatic ring. In contrast to the
single labeling pattern observed from [1,2-13C2]acetate in
sterigmatocystin (8a, Scheme 2),32 incorporation of this tracer
into geodin, secalonic acid,57 and ravenelin58-60 (31a/31b,
Scheme 4) yielded two orientations in a 1:1 ratio, doubtlessly
arising through reaction intermediates with local C2 symmetry.
in the course of the rearrangement. In considering the possible
Baeyer-Villiger oxidations of VA, cleavage B in 2/3 (Scheme
4) would produce asymmetric A-ring o-carboxybenzophenone
intermediates 32/33, whereas the alternative reaction (cleavage
A) would afford a symmetric 2,4,6-trihydroxyphenyl intermedi-
ate from the anthraquinone A-ring and can be excluded. Neither
o-carboxybenzophenone 28 nor 30 has been tested as an
intermediate in the formation of ravenelin or the secalonic acids,
respectively. The necessary intervention of symmetrical inter-
mediates and the precedent of sulochrin (25) in geodin biosyn-
thesis, however, logically support the intermediacy of these
cleaved anthraquinone products in the pathways to 29 and 31.
Analogously, for the case of aflatoxin, no intermediates are
known between the anthraquinones 2/3 and the xanthones 4/5.
We deduce from these biochemical precedents that o-carboxy-
benzophenone 33 (Scheme 4) could be the first intermediate
en route to demethylsterigmatocystin (5).33,34,36,45,61
In the formation of all three of these metabolites, a mecha-
nistic pattern can be seen that requires multiple oxidative cycles.
The geodin (26) spirodienone results from Baeyer-Villiger and
oxidative phenolic coupling reactions, whereas genesis of the
xanthone-like structure of secalonic acid (29) and of ravenelin
(31) can be invoked from Baeyer-Villiger oxidation and aryl
epoxidation of chrysophanol (27, Scheme 4). The biosynthesis
of ST from VA also requires two rounds of oxidation and a
reduction to account for the overall redox changes that take place
We have chosen to prepare the reduced tetrahydrobisfuran
32 on two accounts to test the potential intermediacy of this
hypothetical anthraquinone cleavage product in the biosynthesis
of AFB2. First, the tetrahydrobisfuran-containing products are
significantly more stable and readily synthesized than their
dihydrobisfuran counterparts. Second, in the aflatoxin-producing
organism, A. parasiticus, AFB2 (10) appears in much smaller
(57) Kurobane, I.; Vining, L. C.; McInnes, A. G.; Walter, J. A.; Wright, J. L.
C. Tetrahedron Lett. 1978, 1379-1382.
(58) Birch, A. J.; Simpson, T. J.; Westerman, P. W. Tetrahedron Lett. 1975,
4173-4177.
(59) Birch, A. J.; Baldas, J.; Hlubucek, J. R.; Simpson, T. J.; Westerman, P. W.
J. Chem. Soc., Perkin Trans. 1 1976, 898-904.
(60) Hill, J. G.; Nakashima, T. T.; Vederas, J. C. J. Am. Chem. Soc. 1982, 104,
1745-1748.
9
J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005 3303

A R T I C L E S
Scheme 5 a
Henry and Townsend
a Reagents and conditions: (a) 35, THF, n-BuLi, -78 f -60 f -78 °C, then add 34, -78 f 25 f -78 °C; (b) n-BuLi, -78 °C, 2 h; (c) ethyl
bromoacetate, -78 f 25 °C; (d) tartaric acid (1 M), 24% from 34; (e) TIPSTfl, DIPEA, THF, 0 °C, 78%. MOM ) methoxymethyl, TIPSTfl ) triisopropylsilyl
trifluoromethanesulfonate, DIPEA ) diisopropylethylamine.
Scheme 6 a
a Reagents and conditions: (a) 41, n-BuLi, THF, -78 f -25 f -78 °C, then add 34, -78 f 0 °C, 80%; (b) Ag2O, I2, dioxane, water, -78 f 25 °C,
4 h; (c) TIPSTfl, DIPEA, THF, 0 °C, 69% over two steps; (d) NBS, CHCl3, 0 °C, 93%; (e) pentane, -45 °C, n-BuLi, 2 min, then add DMF, AcOH, 83%;
(f) NaClO2, THF, water, t-BuOH, NaH2PO4, 25 °C; (g) PPh3, DEAD, THF, benzyl alcohols 55 or 57, 25 °C, 76% from 46; (h) n-BuLi, THF, -100 °C; (i)
TPAP, NMO, CH2Cl2, 0 f 25 °C; (j) NaClO2, THF, water, t-BuOH, NaH2PO4, 2-methyl-2-butene, 0 f 25 °C; (k) PPh3, DEAD, THF, BnOH, 25 °C, 62%
over four steps; (l) BCl3, CH2Cl2, 0 f 25 °C, 93%; (m) NaH, BnBr, DMF, 0 f 25 °C, 75%; (n) repeat steps l and m, 72%; (o) TBAF, THF, 0 f 60 °C,
then HCl (3 M), 25 °C, 86%; (p) H2 (30 psi), Pd black, EtOAc, C18 chromatography, 67%. TBDMS ) tert-butyldimethylsilyl, NBS ) N-bromosuccinimide,
DEAD ) diethyl azodicarboxylate, TPAP ) tetrapropylammonium perruthenate, NMO ) 4-methylmorpholine N-oxide, TBAF ) tetrabutylammonium
fluoride.
amounts than the unsaturated AFB1 (11). That is, incorporation
experiments to detect the formation of AFB2 will be more
sensitive by enhancing the amount of the minor metabolite rather
than the major (AFB1) and, as so, confer an analytical advantage
without recourse to radiolabeled substrates.
Addition of n-butyllithium gave the lithioenamine 37, which
was quenched with ethyl bromoacetate to alkylate specifically
on carbon.63,64 Unfortunately, hydrolysis of the resulting imine
38 gave poor yields of the desired phenylacetaldehyde 39 under
even mild conditions, presumably owing to the intrinsic instabil-
ity of the electron-rich product in acid.
Results
The strong electron donor properties of the tri-O-methoxy-
methyl phenyl ring were turned to advantage by preparing the
branched phenylacetaldehyde 43 (Scheme 6) in a Pinacol-like
rearrangement of a linear precursor under functionally neutral
conditions. Wittig reaction of benzaldehyde 34 with the readily
prepared propylphosphonium bromide 4165 smoothly afforded
the styrene 42 in 80% yield (5:2 trans:cis). When treated with
I2/Ag2O in dioxane/water,8,66 the initially formed iodohydrin
rearranged to the phenylacetaldehyde 43 in excellent yield and
was readily scalable. The desired product was found to be prone
to polymerization and decomposition. As a consequence, 43 was
A multistep synthesis of the tetrahydrobisfuran-containing
o-carboxybenzophenone 32 was undertaken from 2,4,6-trihy-
droxybenzoic acid. Following esterification and methoxymethyl
ether formation, LAH reduction and oxidation with TPAP/NMO
readily afforded the protected aldehyde 34 (Scheme 5). Initial
elaboration of the fused tetrahydrobisfuran was attempted by
modification of a method originally developed by Martin et al.62
and extended for dihydro- and tetrahydrobisfuran syntheses.44,61
Benzaldehyde 34 was converted to azadiene 36 following
Horner-Emmons condensation with iminophosphonate 35.
(63) Wender, P. A.; Eissenstat, M. A. J. Am. Chem. Soc. 1978, 100, 292-294.
(64) Wender, P. A.; Schaus, J. M. J. Org. Chem. 1978, 43, 782-784.
(65) Boutellier, M.; Wallach, D.; Tamm, C. HelV. Chim. Acta 1993, 76, 2515-
2527.
(61) Graybill, T. L.; Casillas, E. G.; Pal, K.; Townsend, C. A. J. Am. Chem.
Soc. 1999, 121, 7729-7746.
(62) Martin, S. F.; Phillips, G. W.; Puckette, T. A.; Colapret, J. A. J. Am. Chem.
Soc. 1980, 102, 5866-5872.
(66) Kikuchi, H.; Kogure, K.; Toyoda, M. Chem. Lett. 1984, 341-344.
9
3304 J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005

o-Carboxybenzophenones in the Biosynthesis of Aflatoxin
A R T I C L E S
Scheme 7. Preparation of 2-Bromobenzyl Alcohols 55 and 57a
lyophilization of the aqueous fractions furnished the pure acid
32.
With o-carboxybenzophenone 32 in hand, whole-cell fer-
mentation experiments were conducted with the A. parasiticus
DIS-1 mutant.71 The DIS-1 strain is characterized by a 2.6 kb
insertion into hexA, one of two genes encoding specialized fatty
acid synthase subunits associated with the first step of the
aflatoxin biosynthetic pathway.72 Therefore, aflatoxin biosyn-
thesis is disrupted, and only trace amounts of aflatoxins are
synthesized relative to wild-type producers. The availability in
DIS-1, however, of all the other required biosynthetic enzymes,
coupled with only background levels of aflatoxins, enabled any
increases to be readily detected.45 Owing to their coumarin
nuclei, 10, 11, 13, and 14 are highly fluorescent, and their
formation is easily visualized. Furthermore, the aflatoxins 10,
11, 13, and 14 are readily separated by HPLC, and their relative
amounts can be easily quantified.
a Reagents and conditions: (a) NBS, CHCl3, 0 f 25 °C, 92%; (b) n-BuLi,
toluene, -78 f -5 °C, 1 h, -78 °C, then 1,2-dibromotetrachloroethane,
-78 f 25 °C, 12 h, 52%.
partially purified by rapid filtration through a plug of silica gel
and reacted with triisopropylsilyl triflate (TIPSTfl) to induce
selective methoxymethyl deprotective cyclization8 yielding the
stable siloxyacetal 44.
The successful conversions of known, earlier precursors
versicolorin B (2) and A (3) to aflatoxin products were
implemented as positive controls. The use of both of these
substrates ensured that the relevant biosynthetic enzymes were
present and, indeed, compatible with both tetrahydro- and
dihydrobisfuran-containing substrates. The fermentations were
conducted according to the method of Bhatnagar et al.73 Adye
and Mateles (AM) medium was inoculated with a spore
suspension of DIS-1 and shaken for 48 h in the dark at 30 °C.
The resulting mycelia were collected on cheesecloth, washed
with water, and resuspended in minimal Replacement Medium
(RM).74 One gram portions of cells were alternatively incubated
with acetone solutions of VA, VB, o-carboxybenzophenone 32,
and acetone in the absence of added substrate. DIS-1 incubations
of VA and VB resulted in markedly increased aflatoxin
production relative to the incubations of the pure acetone
negative control. Much to our disappointment, however, repeated
incubation of o-carboxybenzophenone 32 resulted in no detect-
able increase in aflatoxin production (data not shown).
Once convinced of the nonincorporation of o-carboxyben-
zophenone 32 into aflatoxin, we chose to synthesize and incubate
the corresponding 15-deoxybenzophenone 53 (Scheme 6). This
experiment was conceived to confirm the earlier result that
6-deoxyversicolorin A (15, Scheme 1) was not converted to
aflatoxin under conditions where versicolorin A was readily
transformed to the mycotoxin.44 That is, this experiment was
intended to independently verify the previous interpretation that
C-6 reduction is not the initial step of the anthraquinone to
xanthone conversion.
Electrophilic bromination of 44 with NBS was regiospecific
and high-yielding. Metal-halogen exchange and reaction with
DMF generated the aldehyde 46 in superior yield to direct
carbonylation with CO2. Pinnick oxidation67 gave the corre-
sponding labile carboxylic acid, which was coupled with
2-bromo-3,5-bis(O-benzyl)benzyl alcohol under Mitsunobu
conditions to give the ester 47. The o-bromobenzyl alcohol 55
used in this reaction was easily prepared by reaction of the
symmetrically substituted precursor alcohol 54 (Scheme 7) with
NBS. Treatment of the ester 47 at -100 °C with n-BuLi
according to the protocol of Lampe and others in their syntheses
of balinol68-70 proceeded well to the unstable o-hydroxymethyl
benzophenone. The latter was, therefore, oxidized in two steps
to the o-carboxylic acid and converted to the stable benzyl ester
48 (R′ ) MOM). At this point, the O-methoxymethyl ethers
on the B-ring were exchanged for benzyl groups (48, R′ ) Bn)
to allow final deprotection by hydrogenolysis. Parallel reactions
in which the MOM groups were retained to the final deprotec-
tion gave low yields of the fused tetrahydrobisfuran correspond-
ing to 49, presumably owing to side reactions with the
formaldehyde released on hydrolysis. Attempts to cleave both
MOM ethers simultaneously under a variety of protic and Lewis
acidic conditions led to extensive decomposition and traces of
singly cleaved phenols. However, a single ether was reliably
cleaved with just 1 equiv of BCl3. In this way, we obtained a
50/50 mixture of regioisomeric phenols, which co-eluted on
silica gel. The phenolic pair was converted to the corresponding
benzyl ethers with NaH and benzyl bromide in DMF. We then
subjected the resulting mixture to a second round of BCl3-
mediated methoxymethyl cleavage followed by O-benzylation
and obtained the fully benzyl-protected benzophenone 48.
Complete benzyl protection having been achieved, the TBDMS
and TIPS groups were removed with TBAF,61 and acidic workup
was accompanied by intramolecular ether formation to afford
49 in excellent overall yield. Hydrogenolysis over palladium
black, reverse-phase chromatography on C18 silica gel and
The synthesis of 15-deoxybenzophenone 53 was easily
adapted from the route described above following substitution
of the A-ring precursor, 2-bromo-3-benzyloxybenzyl alcohol
(57, Scheme 7), for the analogous 2-bromo-3,5-dibenzyloxy-
benzyl alcohol (55). The two bromides were prepared by
different means. 3,5-Dibenzyloxybenzyl alcohol was sufficiently
electron-rich to undergo bromination with NBS. In contrast, the
comparatively electron-deficient 3-benzyloxybenzyl alcohol (56)
was o-metalated with n-BuLi in toluene, and the resulting
(67) Bal, B. S.; Childers, W. E.; Pinnick, H. W. Tetrahedron 1981, 37, 2091-
2096.
(71) Mahanti, N.; Bhatnagar, D.; Cary, J. W.; Joubran, J.; Linz, J. E. Appl.
EnViron. Microbiol. 1996, 62, 191-195.
(68) Lampe, J. W.; Hughes, P. F.; Biggers, C. K.; Smith, S. H.; Hu, H. J. Org.
Chem. 1994, 59, 5147-5148.
(72) Hitchman, T. S.; Schmidt, E. W.; Trail, F.; Rarick, M. D.; Linz, J. E.;
Townsend, C. A. Bioorg. Chem. 2001, 29, 293-307.
(69) Lampe, J. W.; Hughes, P. F.; Biggers, C. K.; Smith, S. H.; Hu, H. J. Org.
Chem. 1996, 61, 4572-4581.
(73) Bhatnagar, D.; McCormick, S. P.; Lee, L. S.; Hill, R. A. Appl. EnViron.
Microbiol. 1987, 53, 1028-1033.
(74) Adye, J.; Mateles, R. I. Biochim. Biophys. Acta 1964, 86, 418-420.
(70) Nicolaou, K. C.; Koide, K.; Bunnage, M. E. Chem.sEur. J. 1995, 1, 454-
466.
9
J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005 3305

A R T I C L E S
Henry and Townsend
lithio-anion was quenched with 1,2-dibromotetrachloroethane
in modest yield.68 2-Bromo-3-benzyloxybenzyl alcohol (57) and
the benzoic acid derived from 46 were then reacted under
Mitsunobu conditions. The resulting ester 50 was smoothly
converted to 15-deoxy-o-carboxybenzophenone 53 in compa-
rable overall yield in the manner described for the preparation
of the 15-hydroxy analogue 32 (Scheme 6).
Acetone solutions of VA, VB, and 15-deoxybenzophenone
53 were administered to A. parasiticus DIS-1 whole-cell
fermentations, as described above. Again, VA and VB were
converted to aflatoxins, while the 15-deoxybenzophenone 53
was clearly not (data not shown).
Neither of the benzophenones 32 or 53 was converted to
aflatoxins upon whole-cell fungal incubation. We turned our
attention, therefore, to the solubility of our benzophenone
substrates relative to those of the versicolorin anthraquinones.
Versicolorins A and B are planar, neutral, and relatively
hydrophobic. As such, they seem comparatively well suited to
the hydrophobic environment of membrane-bound eukaryotic
P450 monooxygenases.75 By contrast, the o-carboxybenzophe-
nones 32 and 53 are water-soluble, charged at physiological
pH, and rotationally flexible. As a consequence, we were
concerned that the hydrophilic benzophenones might be in-
capable of penetrating the fungal cell wall and/or reaching the
membrane-anchored sites thought to house the AflS monooxy-
genase.
We explored this possibility in the following manner. DIS-1
cells were grown, as described, in 48 h in AM medium,
collected, ground with a mortar and pestle, suspended in buffer,
and centrifuged. The resulting filtrate was separated from the
cellular pellet, and the pellet was resuspended in fresh buffer.
The extract and pellet suspensions were independently inocu-
lated with acetone solutions of VA and pure acetone controls.
A measurable increase in aflatoxin production from VA was
obtained upon incubation with the pellet suspension, but no
comparable increase in AFB1 (11) was detected upon incubation
of VA with the membrane-free preparation. Therefore, a simple
“ground-cell” procedure was employed in a final attempt to
convert either of the benzophenone substrates 32 and 53 to
aflatoxin. Cells were collected, ground with sand in a mortar
and pestle, and suspended with buffer. The sandy suspension
was used directly for substrate incubations.
The results of the ground-cell incubations are depicted by
the representative TLC plate in Figure 1. Lanes A-E correspond
to small molecule standards as indicated, and the lanes I-V
represent chloroform extracts of the ground-cell incubations.
The experiments demonstrate that the known aflatoxin inter-
mediates versicolorin A (3) and B (2) were successfully
incorporated into aflatoxins AFB1 (11) and AFB2 (10),
respectively (blue spots, lanes II and III), while the benzophe-
none substrates 32 and 53 were not (lanes IV and V). Lane I is
the negative control, an extract of the fungal cells incubated
with acetone in the absence of added substrate. The absence of
incorporation of benzophenones 32 and 53 was confirmed in
additional trials and by HPLC analysis.
Figure 1. Thin-layer chromatogram of DIS-1 ground-cell incubations
developed in 60% CHCl3, 30% EtOAc, and 10% HCO2H, photographed
under short-wave UV light. Lanes A-E standards: A, versicolorin A (3);
B, versicolorin B (2); C (top to bottom), AFB1 (11), AFB2 (10), AFG1
(14), AFG2 (13); D, 15-deoxybenzophenone 53; E, 15-hydroxybenzophe-
none 32. Lanes I-V: I, chloroform extract of DIS-1 control; II, DIS-1
with versicolorin A (3); III, DIS-1 with versicolorin B (2); IV, DIS-1 with
15-deoxybenzophenone 53; V, DIS-1 with 15-hydroxybenzophenone 32.
Discussion
Separate disruptions of the aflatoxin pathway genes, aflM and
aflN, led to the accumulation of VA and VB, suggesting
essential roles for both encoded proteins in the cryptic oxidative
cleavage, rearrangement, and reduction to demethylsterigma-
tocystins 4 and 5. The failure of 6-deoxyversicolorin A (15) to
incorporate into AFB1 (11) under conditions where the closely
related versicolorin A (3) is readily transformed to the myc-
otoxin, and the inability of the reduced, that is, 15-deoxy-o-
carboxybenzophenone 53, to support its synthesis all indicate
that AflM-catalyzed reduction is not the first step in the
conversion of 2/3 to 4/5. An AflN-mediated oxidation must,
therefore, initiate this process. Consideration of other known
anthraquinone f xanthone conversions led us to explore the
possibility that the first chemical step was a Baeyer-Villiger
oxidation and ring opening to yield o-carboxybenzophenone 32.
If cleavage was to occur in sense A shown in Scheme 4, a
symmetrical 2,4,6-trihydroxybenzophenone would be generated
and the asymmetric labeling pattern from [1,2-13C2]acetate
incorporation would be lost in subsequent biosynthetic steps.
This outcome is contrary to experiment, and therefore, path A
can be omitted from further consideration. Analogous cleavage
in sense B, however, opens the anthraquinone to an asymmetri-
cally substituted A-ring in keeping with skeletal mapping
experiments using doubly 13C-labeled acetate.32-34 Notwith-
standing apparent biochemical precedent and chemical logic
pointing to o-carboxybenzophenone 32 as the next probable
intermediate beyond versicolorin A/B, experimental evidence
could not be obtained to support this hypothesis despite the
(75) Ortiz de Montellano, P. R. Cytochrome P450: Structure, Mechanism, and
Biochemistry, 2nd ed.; Plenium Press: New York, 1995; Vol. p.
(76) Marvel, C. S.; Porter, P. K. Organic Synthesis 1941, 1, 377-379.
(77) Linderman, R. J.; Jaber, M.; Griedel, B. D. J. Org. Chem. 1994, 59, 6499-
6500.
(78) Bailey, A. J.; Griffith, W. P.; Mostafa, S. I.; Sherwood, P. A. Inorg. Chem.
1993, 32, 268-271.
9
3306 J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005

o-Carboxybenzophenones in the Biosynthesis of Aflatoxin
A R T I C L E S
Hz, 1H), 5.81, 5.78 (t, J ) 6.8 Hz, 1H), 5.14, 5.12, 5.10 (s, 6H), 3.70,
3.61 (t, J ) 7.2 Hz, 2H), 3.45 (m, 9H), 2.42, 2.21 (dq, J ) 2.0, 6.8 Hz,
2H), 0.88, 0.85 (s, 9H), 0.05, 0.00 (s, 6H). 13C NMR (100 MHz,
CDCl3): δ 157.8, 157.0, 156.5, 156.0, 130.5, 130.1, 121.9, 121.5, 111.5,
97.7, 97.6, 95.0, 94.9, 94.7, 94.7, 63.5, 62.8, 56.2, 56.2, 56.1, 38.6,
33.8, 26.1, 26.0, 18.4, 18.4, -5.1, -5.2. IR (CHCl3): ν 2955, 1607,
1473, 1393, 1156, 1048, 925, 836 cm-1. MS-EI m/z (rel inten): 442
(M+, 13), 297 (14), 265 (12), 89 (12), 73 (18), 45 (100). HRMS-FAB
(m/z): M+Na calcd for C22H38O7Si, 465.2279; found, 465.2314.
{3-[2-(tert-Butyldimethylsilanyloxy)ethyl]-4,6-bis-(O-methoxy-
methyl)-2,3-dihydrobenzofuran-2-yloxy}triisopropylsilane (44). Sty-
rene 42 (9.4 g, 21.2 mmol), silver(I) oxide (5.2 g, 22.3 mmol), dioxane
(100 mL), and water (30 mL) were cooled to 0 °C. A dioxane (80 mL)
solution of iodine (5.4 g, 21.2 mmol) was gradually added, and the
mixture was stirred 1 h at 0 °C and 2 h at room temperature. The
mixture was passed through a plug of silica. The filtrate was diluted
with Et2O and washed with 10% aqueous sodium bisulfite. The etherial
layer was dried and concentrated. The crude aldehyde 43 was eluted
through a short column of silica gel with 20% EtOAc in hexanes. The
resulting clear oil was prone to decomposition and taken immediately
to the next reaction. TLC Rf 0.37 (20% EtOAc in hexanes).
ability to detect aflatoxin production with exquisite sensitivity.
Moreover, positive controls with known intermediates readily
afforded aflatoxins under the assay conditions.
The evidence leads inescapably to oxidation of VA and VB
as the first step toward xanthone formation. The experiments
described here, however, exclude addition of the cytochrome
P450-activated oxygen of AflN to the anthraquinone carbonyl
followed by Baeyer-Villiger rearrangement. The only alterna-
tive becomes attack of the activated oxygen species directly on
the aryl rings themselves. Deeper consideration of the order,
identity, and effects of two oxidative and one reductive step
carried out by AflN and AflM is taken up in a companion paper
later in this journal. An additional set of experiments has led to
a new hypothesis of xanthone biosynthesis from anthraquinones
that could prove general for related acetogenin natural products.
Experimental Section
Only the key synthetic steps are described below. The remaining
full synthetic details are available in Supporting Information.
Crude aldehyde 43, DIPEA (5.8 mL, 33.5 mmol), and THF (250
mL) were cooled to 0 °C under Ar. Triisopropylsilyl triflate (7.7 mL,
28.7 mmol) was added dropwise over 5 min. The mixture was stirred
1.5 h and warmed to room temperature. Following quenching with N,N-
dimethylethanolamine (2.5 mL, 25 mmol), the mixture was diluted with
Et2O and washed with cold aqueous 5% HCl, aqueous NaHCO3, and
brine. The Et2O layer was dried and concentrated, and the resulting
residue was eluted through a column of silica gel with 5% EtOAc in
hexanes. The dihydrobenzofuran 44 was isolated (8.4 g, 14.7 mmol,
General Materials and Methods: Melting points were determined
with a Thomas-Hoover oil bath apparatus in open capillaries and are
uncorrected. 1H and 13C NMR spectra were recorded on a Varian
UnityPlus 400 MHz spectrometer and are referenced to CDCl3 (7.27
and 77.0 ppm) and d6-acetone (2.04 and 29.9 ppm) as indicated. High-
and low-resolution mass spectra were recorded by Dr. Joseph Kachinski
of the Department of Chemistry, The Johns Hopkins University, using
a VG Instruments 70-S 250 GC/MS at 70 eV in EI+, CI+, or FAB
operating modes. Additional high-resolution mass spectra were per-
formed by Dr. Christopher Hadad of The Ohio State University and
by Dr. Ronald Cerny of The University of Nebraska. Infrared spectra
were recorded on a Perkin-Elmer 1600 series FTIR spectrophotometer.
Combustion analyses were performed by Atlantic Microlab, Inc.
(Norcross, GA). Flash chromatography was performed using EM silica
gel 60 (230-400 mesh). Thin-layer chromatography (TLC) was
performed on Analtech Uniplate glass plates containing fluorescent
indicator. TLC plates were photographed by James Van Rensselaer,
The Johns Hopkins University. Reagents and aflatoxin standards were
purchased from Aldrich, Inc. Media components were purchased from
Difco (Detroit, MI) or Fisher (Pittsburgh, PA). Nonaqueous reactions
were performed in flame-dried glassware under an atmosphere of N2
or Ar. Solvents were distilled immediately prior to use (THF and Et2O
from sodium/benzophenone ketyl, and CH2Cl2 and CH3CN from CaH2).
During workup, organic solutions were dried over anhydrous Na2SO4.
HPLC was performed with a Perkin-Elmer 235C diode array detector
and series 410 LC pump using a Phenomenex (Torrance, CA) reverse-
phase column (Prodigy 5µ ODS(3) 100Å, 250 × 4.60 mm, 5 micron).
Mini-prep columns were purchased from Qiagen (Valencia, CA).
1
69% yield) as a clear oil. TLC Rf 0.63 (10% EtOAc in hexanes). H
NMR (400 MHz, CDCl3): δ 6.32 (d, J ) 2.1 Hz, 1H), 6.24 (d, J )
2.1 Hz, 1H), 5.83 (d, J ) 1.6 Hz, 1H), 5.10 (m, 4H), 3.68 (m, 2H),
3.45 (s, 6H), 3.29 (dq, J ) 0.8, 4.4 Hz, 1H), 1.98 (m, 1H), 1.70 (m,
1H), 1.13 (m, 21H), 0.88 (s, 9H), 0.02, 0.00 (s, 6H). 13C NMR (100
MHz, CDCl3): δ 160.2, 159.0, 154.4, 111.1, 106.8, 96.0, 95.0, 94.4,
92.9, 61.1, 56.3, 56.2, 47.1, 34.7, 26.1, 18.3, 18.0, 17.9, 12.3, -5.2.
IR (CHCl3): ν 2949, 2867, 1609, 1498, 1465, 1389, 1255, 1217, 1156,
1132, 1052, 1021, 949, 882, 836 cm-1. MS-CI m/z (rel inten): 571 (M
+ 1, 17), 397 (23), 363 (9), 339 (10), 73 (9), 45 (100). Anal. Calcd for
C29H54O7Si2: C, 61.01; H, 9.53. Found: C, 61.27; H, 9.54.
5-Bromo-{3-[2-(tert-butyldimethylsilanyloxy)ethyl]-4,6-bis-(O-
methoxymethyl)-2,3-dihydrobenzofuran-2-yloxy}triisopropyl-
silane (45). Acetal 44 (6.0 g, 10.5 mmol) and CHCl3 (150 mL) were
cooled to 0 °C under Ar. NBS (2.1 g, 11.5 mmol) was added, and the
ice bath was removed. One hour later, the mixture was washed with
water and brine, concentrated, and passed through a column of silica
with 5-10% EtOAc in hexanes. Pure bromide 45 (6.4 g, 9.8 mmol,
93% yield) was isolated as a faintly yellow oil. TLC Rf 0.67 (10%
EtOAc in hexanes). 1H NMR (400 MHz, CDCl3): δ 6.45 (s, 1H), 5.90
(d, J ) 0.8 Hz, 1H), 5.2 (m, 4H), 3.8 (m, 2H), 3.58 (s, 3H), 3.45 (s,
3H), 3.40 (dq, J ) 0.8, 4.4 Hz, 1H), 1.96 (m, 1H), 1.68 (m, 1H), 1.1
(m, 21H), 0.87 (s, 9H), 0.01, 0.00 (s, 6H). 13C NMR (100 MHz,
CDCl3): δ 159.3, 155.1, 151.6, 116.6, 106.4, 98.8, 98.4, 95.5, 95.4,
61.7, 60.7, 58.2, 57.9, 56.5, 55.4, 48.3, 34.6, 26.1, 18.3, 18.0, 18.0,
17.9, 12.3, -5.3. IR (CHCl3): ν 2949, 2867, 2360, 1616, 1588, 1464,
1388, 1256, 1212, 1156, 1041, 926, 882, 836, 777 cm-1. MS-CI m/z
(rel inten): 649 (M + 1, 82), 591 (20), 475 (100), 132 (18), 90 (23),
45 (43). HRMS-EI (m/z): M+ calcd for C29H53BrO7Si2, 648.2513;
found, 648.2500. Anal. Calcd for C29H53BrO7Si2: C, 53.60; H, 8.22.
Found: C, 53.99; H, 8.32.
Synthesis of 5-Keto-[(2′,4′-dihydroxy-6′-carboxy)phenyl)]-4,6-
dihydroxy-[2,3-b]-benzo-3a,8a-dihydrobisfuran (32). [4-(2,4,6-Tris-
(O-methoxymethyl)phenyl)but-3-enyloxy]-tert-butyldimethylsilyl Ether
(42). A slurry of 3-trimethylsiloxypropyltriphenylphosphonium bromide
(41) (12.0 g, 23.3 mmol) in THF (100 mL) was cooled to -78 °C. A
hexane solution of 2.43 M n-BuLi (25.6 mmol) was gradually added.
The resulting yellow solution turned red following 40 min of stirring
at -78 °C, then 40 min at -25 °C. The ylide solution was then cooled
to -78 °C, and benzaldehyde 34 (6.2 g, 21.7 mmol) in THF (50 mL)
was slowly cannulated under Ar. The solution was stirred 30 min at
-78 °C and 30 min at 0 °C. The mixture was then diluted with hexanes
and filtered through a silica plug. The resulting filtrate was concentrated
and eluted through a column of silica gel with 10-20% EtOAc in
hexanes. Styrene 42 (7.6 g, 17.4 mmol, 80%, trans:cis isomers ) 2.4:
1) was isolated as a clear oil. TLC Rf 0.50 (20% EtOAc in hexanes).
1H NMR (400 MHz, CDCl3): δ 6.53 (m, 2H), 6.21, 6.18 (t, J ) 1.6
5-Formyl-{3-[2-(tert-butyldimethylsilanyloxy)ethyl]-4,6-bis-(O-
methoxymethyl)-2,3-dihydrobenzofuran-2-yloxy}triisopropyl-
silane (46). Arylbromide 45 (7.3 g, 11.3 mmol) in pentane (125 mL)
was cooled to -45 °C under Ar. n-BuLi (16.1 mL, 1.4 M in hexanes)
was added over 90 s, and the resulting slurry was stirred an additional
9
J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005 3307

A R T I C L E S
Henry and Townsend
30 s. Dry DMF (4.4 mL, 57 mmol) was added in one portion. Stirring
was continued for 1 h at -45 °C. An Et2O solution of acetic acid (24
mL, 1 M) was added at -45 °C. The mixture was diluted with Et2O
(300 mL) and washed with aqueous NaHCO3, water, and brine, then
dried, concentrated, and passed through a column of silica gel with
10% EtOAc in hexanes. Benzaldehyde 46 (5.6 g, 9.4 mmol, 83% yield)
was isolated as a yellow oil. TLC Rf 0.28 (10% EtOAc in hexanes). 1H
NMR (400 MHz, CDCl3): δ 10.32 (s, 1H), 6.46 (s, 1H), 5.95 (s, 1H),
5.21, 5.07 (m, 4H), 3.70 (m, 1H), 3.60 (m, 1H), 3.50 (s, 3H), 3.47 (s,
3H), 3.42 (m, 1H), 2.01 (m, 1H), 1.71 (m, 1H), 1.07 (m, 21H), 0.85 (s,
9H), -0.01, -0.03 (s, 6H). 13C NMR (100 MHz, CDCl3): δ 187.9,
165.3, 163.1, 156.4, 116.2, 112.9, 107.4, 100.5, 95.2, 93.9, 60.7, 57.7,
56.6, 47.4, 34.2, 26.0, 18.3, 17.9, 17.8, 12.2, -5.3. IR (CHCl3): ν 2949,
2870, 1680, 1613, 1467, 1390, 1236, 1118, 1080, 1042, 926, 834, 778
cm-1. MS-EI m/z (rel inten): 599 (M + 1, 100), 541 (8), 45 (20).
HRMS-EI (m/z): M+ calcd for C30H54O8Si2, 598.3357; found, 598.3341.
Anal. Calcd for C30H54O8Si2: C, 60.16; H, 9.09. Found: C, 60.67; H,
9.15.
tech. grade, 3.0 g, 26.5 mmol) was added and stirring continued until
the aldehyde was consumed. The mixture was diluted with water (100
mL) and extracted with EtOAc. The organic layer was washed with
brine, and then dried, concentrated, and combined with THF (25 mL),
PPh3 (1.18 g, 4.5 mmol), and benzyl alcohol (0.6 mL, 6.0 mmol) at
room temperature. DEAD (0.75 g, 4.5 mmol) was gradually added,
and the mixture was stirred for 10 min. The mixture was then
concentrated and passed through a column of silica gel with 10-20%
EtOAc in hexanes. Benzophenone 48 (R′ ) MOM; 2.2 g, 2.3 mmol,
62% yield from bromide 47) was isolated as a clear oil. TLC Rf 0.45
1
(20% EtOAc in hexanes). H NMR (400 MHz, CDCl3): δ 7.3 (m,
13H), 6.99 (d, J ) 2.1 Hz, 1H), 6.95 (d, J ) 6.7 Hz, 2H), 6.62 (d, J
) 2.4 Hz, 1H), 6.35 (s, 1H), 5.94 (s, 1H), 5.20 (dd, J ) 8.2, 21.0 Hz,
2H), 5.07 (s, 2H), 4.88 (s, 4H), 4.49 (m, 2H), 3.66 (m, 2H), 3.42 (m,
1H), 3.27 (b, 3H), 2.95 (s, 3H), 2.04 (b, 1H), 1.60 (b, 1H), 1.10 (m,
21H), 0.88 (s, 9H), 0.01 (d, 6H). 13C NMR (100 MHz, CDCl3): δ
191.1, 167.8, 162.3, 160.0, 159.0, 157.1, 154.9, 136.2, 135.9, 135.7,
133.7, 128.7, 128.5, 128.3, 128.3, 128.1, 127.8, 127.6, 127.0, 117.8,
115.7, 107.1, 106.7, 103.1, 99.9, 94.8, 93.6, 70.5, 70.4, 67.4, 65.9, 60.9,
57.0, 55.6, 47.8, 34.3, 26.0, 18.3, 17.9, 17.8, 12.2, -5.4. IR (CHCl3):
ν 2948, 2867, 1730, 1599, 1465, 1389, 1326, 1258, 1215, 1143, 1044,
924, 836 cm-1. MS-FAB m/z (rel inten): 1043 (M + Na, 100), 953
(8), 181 (13), 115 (33). HRMS-EI (m/z): M+Na calcd for C58H76O12-
Si2, 1043.4768; found, 1043.4766. Anal. Calcd for C58H76O12Si2: C,
68.20; H, 7.50. Found: C, 68.22; H, 7.62.
5-[3′,5′-Bis-(O-benzyl)-2′-bromobenzyl]benzoyl-{3-[2-(tert-bu-
tyldimethylsilanyloxy)ethyl]-4,6-bis-(O-methoxymethyl)-2,3-dihy-
drobenzofuran-2-yloxy}triisopropylsilane (47). (Procedure A) Ben-
zaldehyde 46 (3.0 g, 5.0 mmol), THF (25 mL), t-BuOH (25 mL), H2O
(10 mL), 2-methyl-2-butene (9 mL, 2 M in THF), and NaH2PO4 (20
mL, 1 M in H2O) were cooled to 0 °C. Sodium chlorite (80% tech.
grade, 2.5 g, 22.1 mmol) was added, and the ice bath was removed.
After 40 min, the mixture was diluted with EtOAc and water. The
organic layer was washed with water and brine, dried, and concentrated.
The resulting oily residue was combined with 3,5-dibenzyloxy-2-
bromobenzyl alcohol (55) (2.6 g, 6.4 mmol), PPh3 (3.1 g, 11.6 mmol),
and THF (100 mL) and cooled to 0 °C under Ar. DEAD (1.84 mL,
11.6 mmol) was gradually added; the ice bath was removed, and the
mixture was stirred 30 min, and then concentrated and passed through
a column of silica gel with 10% EtOAc in hexanes. Ester 47 was
isolated as a clear oil (3.8 g, 3.8 mmol, 76% yield, over two steps).
Microbiological Procedures
Organisms. The wild-type aflatoxin producing strain of A. para-
siticus (SU-1, ATCC 56775) was purchased from the American Type
Culture Collection (ATCC) of Manassas, VA. The averufin-accumulat-
ing strain (AVR-1, SRRC-165) of A. parasiticus used for whole-cell
incubations was generously provided by Shannon Beltz and Maren
Klich of the USDA-ARS-SRRC. The HexA disruption mutant (DIS-
1) was provided by Professor John Linz of Michigan State University.
Versicolorin A biosynthesis was achieved with the accumulating strain
(Wh-1, ATCC 36537) of A. parasiticus as described.45 Sterigmatocystin
was isolated from the accumulating strain (ATCC 28286) of A.
Versicolor as described.26
AVR-1 and DIS-1 stock cultures were maintained on plates of
potato-dextrose agar (PDA) containing potatoes (300 g) diced, boiled,
strained through cheesecloth, and combined with glucose (20 g) and
agar (15 g). The mixture was diluted to 1 L with distilled water and
autoclaved.
Media. Adye and Mateles (AM) medium74 contained (per liter) 50
g of sucrose, 10 g of KH2PO4, 3 g of (NH4)2SO4, 1 g of anhydrous
MgSO4, and 2 mL of trace metals solution. The trace metals solution
contained (per liter) 0.35 g of Na2B4O7 × 10 H2O, 0.25 g of (NH4)6-
Mo7O24 × 4 H2O, 5.0 g of Fe2(SO4)3 × 6 H2O, 0.15 g of CuSO4 × 5
H2O, 0.055 g of MnSO4 × H2O, and 8.8 g of ZnSO4 × 7 H2O.
Replacement Medium (RM) contained (per liter) 1.62 g of glucose,
5 g of KH2PO4, 0.5 g of KCl, 0.25 g of anhydrous MgSO4, and 2 mL
of trace metals solution. The medium was autoclaved at 140 °C, 20
psi, for 20 min.
30% Glycerol Buffer (0.5 L). Glycerol (150 mL), potassium
phosphate (pH 7.5, 0.05 M), EDTA (1 mM), benzamidine HCl (100
µM, 8 mg), phenylmethylsulfonyl fluoride (100 µM, 8 mg), and
mercaptoethanol (2 mM, 0.07 mL) were diluted to 500 mL with distilled
water and stored at 4 °C.
1
TLC Rf 0.80 (20% EtOAc in hexanes). H NMR (400 MHz, CDCl3):
δ 7.35 (m, 10H), 6.96 (d, J ) 2.8 Hz, 1H), 6.58 (d, J ) 2.8 Hz, 1H),
6.51 (s, 1H), 5.89 (s, 1H), 5.47 (dd, J ) 9.4, 24.2 Hz, 2H), 5.15 (m,
6H), 5.02 (s, 2H), 3.74 (m, 2H), 3.67 (m, 1H), 3.48 (s, 3H), 3.40 (s,
3H), 2.0 (m, 1H), 1.65 (m, 1H), 1.1 (m, 21H), 0.92 (s, 9H), 0.05 (s,
6H). 13C NMR (100 MHz, CDCl3): δ 166.2, 161.5, 158.9, 156.3, 155.8,
152.0, 137.6, 136.5, 136.4, 128.7, 128.7, 128.3, 128.1, 127.7, 127.1,
114.4, 111.2, 107.6, 106.3, 103.8, 101.1, 98.6, 95.0, 93.9, 71.0, 70.4,
66.5, 60.7, 57.4, 56.2, 47.7, 34.5, 26.0, 18.4, 18.0, 17.8, 12.2, -5.4.
IR (CHCl3): ν 2949, 2866, 1734, 1596, 1465, 1327, 1259, 1158, 1045,
834 cm-1. MS-FAB m/z (rel inten): 995 (M + 1, 3), 597 (33), 553
(17), 495 (31), 383 (23), 157 (32), 115 (100). HRMS-EI (m/z): M+Na
calcd for C51H71BrO11Si2, 1017.3610; found, 1017.3613. Anal. Calcd
for C51H71BrO11Si2: C, 61.49; H, 7.18. Found: C, 61.48; H, 7.22.
5-Keto-[2′,4′-bis-(O-benzyl)-6′-carboxybenzyl]-{3-[2-(tert-butyldi-
methylsilanyloxy)ethyl]-4,6-bis-(O-methoxymethyl)-2,3-dihydro-
benzofuran-2-yloxy}triisopropylsilane (48, R′ ) OMOM). (Proce-
dure B) Bromide 47 (3.6 g, 3.7 mmol) in THF (30 mL) was cooled to
-100 °C under Ar, and n-BuLi (3.0 mL, 1.3 M in hexanes) was added.
The bath (N2/pentane) temperature was then warmed to -78 °C and
stirring continued for 30 min. Saturated aqueous NaHCO3, then Et2O
(150 mL), and H2O (50 mL) were added. The etherial layer was washed
with water and brine, then dried, concentrated, and passed through a
plug of silica gel with 20% EtOAc in hexanes. The filtrate was
concentrated, and the resulting crude alcohol product was immediately
combined with CH2Cl2 (100 mL), NMO (0.48 g, 4.1 mmol), and 4 Å
sieves under Ar. TPAP (0.08 g, 0.24 mmol) was added and stirring
continued for 30 min. The green mixture was concentrated, then diluted
with EtOAc, and passed through a plug of silica gel. The filtrate was
concentrated, combined with THF (24 mL), H2O (6 mL), t-BuOH (24
mL), 2-methyl-2-butene (12 mL, 2 M in THF), and aqueous NaH2PO4
(9 mL, 1 M), and stirred at room temperature. Sodium chlorite (80%
Whole-Cell Feeding Experiments. AM medium (500 mL) in a 2
L Erlenmeyer flask was inoculated with a 20% glycerol spore
suspension (50 µL). After incubation in the dark for 48 h at 175 rpm
and 28 °C, the mycelial spheres were collected on cheesecloth, washed
with RM, and separated into portions of approximately 1.0-1.2 g of
wet cells. Each portion was transferred to a 50 mL Erlenmeyer flask
containing RM medium (10 mL). All substrates (1 mg/mL) were
administered as acetone solutions (20 µL). Incubation continued in the
9
3308 J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005

o-Carboxybenzophenones in the Biosynthesis of Aflatoxin
A R T I C L E S
dark for an additional 48 h at 175 rpm and 28 °C. The medium was
then decanted from the mycelial spheres, and the cells were steeped in
acetone. Acetone washings were combined with the medium solution,
diluted with water, and extracted with chloroform. The extract was
dried over Na2SO4, concentrated, and examined by TLC (6:3:1
chloroform/ethyl acetate/formic acid). Subsequent HPLC analysis was
performed with a Phenomenex reverse-phase column (Prodigy 5µ ODS-
(3) 100 Å, 250 × 4.60 mm, 5 micron) eluting at 0.8 mL/min of water:
methanol (55:45) and monitoring at 360 nm.
column (2 mL, QIAGEN, Inc.) and centrifuged at 14 000 rpm for 1
min to filter sand and cellular debris. Chloroform (50 µL) was then
added to the spin column filter, and the centrifugation step was repeated.
The filter was discarded, and the filtrate was transferred to a fresh
Eppendorf tube (1.5 mL). Following brief vortexing and centrifugation,
the chloroform layer was analyzed by TLC or HPLC (as described
above).
Ground-Cell Feeding Experiments. AM medium (100 mL) in a
500 mL Erlenmeyer flask was inoculated with a 20% glycerol spore
suspension (50 µL) of the A. parasiticus DIS-1 mutant. After incubation
in the dark for 48 h at 175 rpm and 28 °C, the mycelial spheres were
collected on cheesecloth and washed with distilled water. Approximately
5 g of wet cells, 5 g of sand, and 3-5 mL of 30% glycerol buffer were
ground in a mortar for 5 min at room temperature.
Acknowledgment. We are grateful to the National Institutes
of Health for financial support of this research (ES01670) and
for major funding to purchase the analytical instrumentation used
(NMR: RR 01934, RR 04794, RR 06468, MS: RR 02318).
We acknowledge with pleasure discussions of the early synthetic
work with Professor Amit Basak (IIT, Kharagpur).
To each incubation tube (Eppendorf, 1.5 mL) were added the sandy
ground-cell mixture (1 mL), aqueous SAM (Cl) and NADPH (30 µL
of an approximately 1 mg/mL stock solution), and each of the
designated aflatoxin precursors (10 µL, in acetone, of an approximately
1 mg/mL stock solution).
Supporting Information Available: Complete synthetic de-
tails of o-carboxybenzophenones 32 and 53. This material is
available free of charge via the Internet at http://pubs.acs.org.
Incubations were shaken in the dark for 8 h at 175 rpm and 28 °C.
The contents of each Eppendorf tube were loaded onto a QIAprep spin
JA045520Z
9
J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005 3309