Exploiting enzymatic promiscuity to engineer a focused library of highly selective antifungal and antiproliferative aureothin analogues

Article abstract of DOI:10.1021/ja102751h

Aureothin is a shikimate-polyketide hybrid metabolite from Streptomyces thioluteus with a rare nitroaryl moiety, a chiral tetrahydrofuran ring, and an O-methylated pyrone ring. The antimicrobial and antitumor activities of aureothin have caught our interest in modulating its structure as well as its bioactivity profile. In an integrated approach using mutasynthesis, biotransformation, and combinatorial biosynthesis, a defined library of aureothin analogues was generated. The promiscuity of the polyketide synthase assembly line toward different starter units and the plasticity of the pyrone and tetrahydrofuran ring formation were exploited. A selection of 15 new aureothin analogues with modifications at the aryl residue, the pyrone ring, and the oxygenated backbone was produced on a preparative scale and fully characterized. Remarkably, various new aureothin derivatives are less cytotoxic than aureothin but have improved antiproliferative activities. Furthermore, we found that the THF ring is crucial for the remarkably selective activity of aureothin analogues against certain pathogenic fungi.

Full text of DOI:10.1021/ja102751h

Published on Web 07/13/2010
Exploiting Enzymatic Promiscuity to Engineer a Focused
Library of Highly Selective Antifungal and Antiproliferative
Aureothin Analogues
Martina Werneburg,† Benjamin Busch,† Jing He,† Martin E.A. Richter,†
Longkuan Xiang,§ Bradley S. Moore,§ Martin Roth,† Hans-Martin Dahse,† and
Christian Hertweck*,†,‡
Leibniz Institute for Natural Product Research and Infection Biology, HKI, Dept. of Biomolecular
Chemistry, Beutenbergstr. 11a, 07745 Jena, Germany, the Friedrich Schiller UniVersity,
Jena, Germany, and the Scripps Institution of Oceanography and Skaggs School of Pharmacy and
Pharmaceutical Sciences, UniVersity of California at San Diego, La Jolla, California, 92093-0204
Received April 1, 2010; E-mail: christian.hertweck@hki-jena.de
Abstract: Aureothin is a shikimate-polyketide hybrid metabolite from Streptomyces thioluteus with a rare
nitroaryl moiety, a chiral tetrahydrofuran ring, and an O-methylated pyrone ring. The antimicrobial and
antitumor activities of aureothin have caught our interest in modulating its structure as well as its bioactivity
profile. In an integrated approach using mutasynthesis, biotransformation, and combinatorial biosynthesis,
a defined library of aureothin analogues was generated. The promiscuity of the polyketide synthase assembly
line toward different starter units and the plasticity of the pyrone and tetrahydrofuran ring formation were
exploited. A selection of 15 new aureothin analogues with modifications at the aryl residue, the pyrone
ring, and the oxygenated backbone was produced on a preparative scale and fully characterized.
Remarkably, various new aureothin derivatives are less cytotoxic than aureothin but have improved
antiproliferative activities. Furthermore, we found that the THF ring is crucial for the remarkably selective
activity of aureothin analogues against certain pathogenic fungi.
(mutasynthesis),7,8 using individual enzymes for biotransfor-
mations,9 and swapping biosynthesis genes in a mix-and-match
Introduction
Natural products are known to represent the major source
for novel therapeutics, in particular in the field of antiinfectives
and antitumoral agents.1,2 This is not surprising in light of the
hypothesis that many of these compounds contribute toward
defending the producer’s ecological niche and have undergone
evolutionary selection and optimization.3,4 Nonetheless, natural
products as such are not necessarily suited as therapeutics. To
reduce undesired side effects, improve bioavailability issues,
and modulate their bioactivity profile, natural products typically
undergo a chemical post-evolution through semisynthesis (i.e.,
derivatization) or total synthesis of analogues prior to clinical
trials.5,6 During the past two decades, the portfolio of methods
for structural variation has been greatly increased by the
knowledge of biosynthetic pathways and their rational repro-
gramming. Synthetic methods can now be complemented by
feeding synthetic building blocks to suitable mutant strains
fashion (combinatorial biosynthesis).10-12 These techniques have
been predominantly applied to the large and diverse group of
(7) Weist, S.; Su¨ssmuth, R. D. Appl. Microbiol. Biotechnol. 2005, 68,
141–150.
(8) Recent examples include the following. Enterocin: Kalaitzis, J. A.;
Izumikawa, M.; Xiang, L.; Hertweck, C.; Moore, B. S. J. Am. Chem.
Soc. 2003, 125, 9290–9291. Rapamycin: Gregory, M. A.; Petkovic,
H.; Lill, R. E.; Moss, S. J.; Wilkinson, B.; Gaisser, S.; Leadlay, P. F.;
Sheridan, R. M. Angew. Chem., Int. Ed. 2005, 44, 4757–4760. Bor-
relidin: Moss, S. J.; Carletti, I.; Olano, C.; Sheridan, R. M.; Ward,
M.; Math, V.; Nur, E. A. M.; Brana, A. F.; Zhang, M. Q.; Leadlay,
P. F.; Mendez, C.; Salas, J. A.; Wilkinson, B. Chem. Commun. 2006,
2341–2343. Ansamitocin: Taft, F.; Bru¨njes, M.; Floss, H. G.; Czem-
pinski, N.; Grond, S.; Sasse, F.; Kirschning, A. ChemBioChem 2008,
9, 1057–1060. Geldanamycin: Kim, W.; Lee, J. S.; Lee, D.; Cai, X. F.;
Shin, J. C.; Lee, K.; Lee, C. H.; Ryu, S.; Paik, S. G.; Lee, J. J.; Hong,
Y. S. ChemBioChem 2007, 8, 1491–1494. Eichner, S.; Floss, H. G.;
Sasse, F.; Kirschning, A. ChemBioChem 2009, 10, 1801–1805. Myx-
alamid: Bode, H. B.; Meiser, P.; Klefisch, T.; Cortina, N. S.; Krug,
D.; Go¨hring, A.; Schwa¨r, G.; Mahmud, T.; Elnakady, Y. A.; Mu¨ller,
R. ChemBioChem 2007, 8, 2139–2144. Pikromycin: Gupta, S.; Lak-
shmanan, V.; Kim, B. S.; Fecik, R.; Reynolds, K. A. ChemBioChem
2008, 9, 1609–1616. Spinosyn: Sheehan, L. S.; Lill, R. E.; Wilkinson,
B.; Sheridan, R. M.; Vousden, W. A.; Kaja, A. L.; Crouse, G. D.;
Gifford, J.; Graupner, P. R.; Karr, L.; Lewer, P.; Sparks, T. C.; Leadlay,
P. F.; Waldron, C.; Martin, C. J. J. Nat. Prod. 2006, 69, 1702–
1710. Salinosporamide: Eusta´quio, A. S.; Moore, B. S. Angew. Chem.,
Int. Ed. 2008, 47, 3936–3938. Nett, M.; Gulder, T. A.; Kale, A. J.;
Hughes, C. C.; Moore, B. S. J. Med. Chem. 2009, 52, 6163–6167.
Liu, Y.; Hazzard, C.; Eusta´quio, A. S.; Reynolds, K. A.; Moore, B. S.
J. Am. Chem. Soc. 2009, 131, 10376–10377.
† HKI.
§ University of California at San Diego.
‡ Friedrich Schiller University.
(1) Newman, D. J.; Cragg, G. M. J. Nat. Prod. 2007, 70, 461–477.
(2) Cragg, G. M.; Grothaus, P. G.; Newman, D. J. Chem. ReV. 2009, 109,
3012–3043.
(3) Osada, H.; Hertweck, C. Curr. Opin. Chem. Biol. 2009, 13, 133–134.
(4) John, J. E. Drug DiscoVery Today 2010. epub ahead of print, DOI:,
10.1016/j.drudis.2010.02.008.
(5) Newman, D. J.; Cragg, G. M.; Snader, K. M. Nat. Prod. Rep. 2000,
17, 215–234.
(6) von Nussbaum, F.; Brands, M.; Hinzen, B.; Weigand, S.; Ha¨bich, D.
Angew. Chem., Int. Ed. 2006, 45, 5072–5129.
(9) Mu¨ller, M. Curr. Opin. Biotechnol. 2004, 15, 591–598.
(10) Floss, H. G. J. Biotechnol. 2006, 124, 242–257.
9
10.1021/ja102751h  2010 American Chemical Society
J. AM. CHEM. SOC. 2010, 132, 10407–10413 10407

A R T I C L E S
Werneburg et al.
of p-aminobenzoate (PABA) and its subsequent transformation
into p-nitrobenzoate (PNBA) by a novel nitro group forming
N-oxygenase (AurF).29-31 PNBA is then activated by the acyl-
CoA ligase AurE and loaded onto the first module of a non-
colinear modular type I polyketide synthase (AurA-C), which
catalyzes five elongation cycles (Figure 2).32 After off-loading
of the pentaketide with lactonization, the resulting pyrone ring
is methylated by AurI,33 and finally the cytochrome P450
monooxygenase AurH introduces the tetrahydrofuran ring.34,35
The pathway thus exhibits three general levels that could be
envisaged for diversification: incorporation of the starter unit
building block,36 formation of the polyketide backbone, and
enzymatic tailoring (i.e., alkylation and oxidative heterocycliza-
tion). As initial attempts to alter the aureothin polyketide
synthase failed (data not shown), we concluded that the
thiotemplate itself is not readily amenable to manipulations. In
contrast, mutational analyses of the aur gene cluster suggested
that both the choice of starter unit and post-PKS tailoring
reactions would in principle leave some room for genetically
engineering structural variants. While an aurF null mutant could
be used for mutasynthesis approaches, variants lacking aurH
or aurI would result in deoxy-/hydroxyl- or desmethylaureothin
derivatives, respectively. Considering x possible starter units,
three variants in backbone functionalization (methylene, hy-
droxyl, tetrahydrofuran), and three different pyrone rings
(nonmethylated, R-methlyated, γ-methylated), at least in prin-
ciple 9 x aureothin derivatives could be generated.
Tolerance of the aur PKS toward Alternative Starter
Units. A mutant lacking the N-oxygenase gene aurF is suitable
for the mutasynthesis of aureothin derivative aureonitrile.37
However, apart from cyanobenzoate we could not observe the
incorporation of further alternative starter units. Surprisingly,
the situation changed dramatically when we used another genetic
construct in a different expression host. To generate deoxyau-
reothin derivatives, we deleted the N-oxygenase gene aurF as
well as the cytochrome P450 monooxygenase gene aurH from
the aur biosynthesis gene cluster. Both genes were subsequently
excised from a cosmid bearing the entire aur locus by a
combination of PCR-targeting and cloning strategies, respec-
tively. The resulting plasmid, pMZ01, was introduced into the
host strain S. albus, and the mutant strain was fermented in
liquid production media. HPLC-MS monitoring of the extracts
revealed that aureothin biosynthesis was fully abolished (Figure
3d). To complement the double knockout mutant, we supple-
mented the medium with PNBA and introduced a compatible
expression plasmid (pMZ04), a pHJ110 derivative that contains
the aurF gene downstream of the constitutive ermE promoter.
Because these conditions fully restored aureothin biosynthesis,
polar effects could be excluded. Using the ∆aurFH mutant
supplemented with PNBA, we yielded 44.8 mg L-1 deoxyau-
Figure 1. Structures of aureothin (1), N-acetylaureothamine (2), neoau-
reothin (spectinabilin, 3), and SNF4435C (4).
polyketide metabolites,13 mainly macrolide antibiotics.14-18 Yet
in comparison, polyketide-derived pyrone compounds, which
are considered as a privileged group of bioactive natural
products,19-21 have been less thoroughly investigated to date.
The densely functionalized shikimate-polyketide hybrid me-
tabolite aureothin (1) is the prototype of a series of related
pyrone compounds from actinomycetes. Whereas aureothin (1)
is known as a cytotoxic agent from the soil bacterium Strep-
tomyces thioluteus,22 N-acetylaureothamine (2) and neoaureothin
(spectinabilin, 3) were identified because of their potent anti-
Helicobacter23 and antiviral activities, respectively.24 SNF4435C
(4) and D are known as immunosuppressants25 derived from
neoaureothin by a photoinduced electrocyclic rearrangement26
(Figure 1). To gain an insight into the biological assembly lines
involved in the formation of these metabolites and to harness
the biosynthetic potential for the generation of novel analogues,
we previously cloned and sequenced the aur and nor biosyn-
thesis gene loci.27,28 Here, we report an integrated approach
using the combination of mutasynthesis, biotransformation, and
combinatorial biosynthesis for the generation of a focused library
of aureothin analogues with remarkably selective antifungal and
antiproliferative activities.
Results and Discussion
Levels of Diversification in the Enzymatic Aureothin
Assembly Line. According to our previous functional analyses,
the aureothin biosynthetic pathway is initiated by the formation
(11) Menzella, H. G.; Reeves, C. D. Curr. Opin. Microbiol. 2007, 10, 238–
245.
(12) Zhang, W.; Tang, Y. J. Med. Chem. 2008, 51, 2629–2633.
(13) Hertweck, C. Angew. Chem., Int. Ed. 2009, 48, 4688–4716.
(14) Khosla, C.; Keasling, J. D. Nat. ReV Drug DiscoV. 2003, 2, 1019–
1025.
(15) Wilkinson, B.; Moss, S. J. Curr. Opin. Drug. DiscoVery DeV. 2005,
8, 748–756.
(16) Wilkinson, B.; Micklefield, J. Nat. Chem. Biol. 2007, 3, 379–386.
(17) Kirschning, A.; Taft, F.; Knobloch, T. Org. Biomol. Chem. 2007, 5,
3245–3259.
(18) Zhou, H.; Xie, X.; Tang, Y. Curr. Opin. Biotechnol. 2008, 19, 590–
596.
(27) He, J.; Hertweck, C. Chem. Biol. 2003, 10, 1225–1232.
(28) Traitcheva, N.; Jenke-Kodama, H.; He, J.; Dittmann, E.; Hertweck,
C. ChemBioChem 2007, 8, 1841–1849.
(19) McGlacken, G. P.; Fairlamb, I. J. Nat. Prod. Rep. 2005, 22, 369–
385.
(20) Wilk, W.; Waldmann, H.; Kaiser, M. Bioorg. Med. Chem. 2008, 17,
2304–2309.
(29) He, J.; Hertweck, C. J. Am. Chem. Soc. 2004, 126, 3694–3695.
(30) Winkler, R.; Hertweck, C. Angew. Chem. 2005, 117, 4152–4155.
(31) Winkler, R.; Richter, M.; Knu¨pfer, U.; Merten, D.; Hertweck, C.
Angew. Chem., Int. Ed. 2006, 45, 8016–8018.
(21) Busch, B.; Hertweck, C. Phytochemistry 2009, 70, 1833–1840.
(22) Hirata, Y.; Nakata, H.; Yamada, K.; Okuhara, K.; Naito, T. Tetrahe-
dron 1961, 14, 252–274.
(32) He, J.; Hertweck, C. ChemBioChem 2005, 6, 908–912.
(33) Mu¨ller, M.; He, J.; Hertweck, C. ChemBioChem 2006, 7, 37–39.
(34) He, J.; Mu¨ller, M.; Hertweck, C. J. Am. Chem. Soc. 2004, 126, 16742–
16743.
(23) Taginuchi, M.; Watanabe, M.; Nagai, K.; Suzumura, K. I.; Suzuki,
K. I.; Tanaka, A. J. Antibiot. 2000, 53, 844–847.
(24) Cassinelli, G.; Grein, A.; Orezzi, P.; Pennella, P.; Sanfilippo, A. Arch.
Microbiol. 1967, 55, 358.
(35) Richter, M. E. A.; Traitcheva, N.; Knu¨pfer, U.; Hertweck, C. Angew.
Chem., Int. Ed. 2008, 47, 8872–8875.
(25) Takahashi, K.; Tsuda, E.; Kurosawa, K. J. Antibiot. 2001, 54, 548–
553.
(36) Moore, B. S.; Hertweck, C. Nat. Prod. Rep. 2002, 19, 70–99.
(37) Ziehl, M.; He, J.; Dahse, H.-M.; Hertweck, C. Angew. Chem., Int.
Ed. 2005, 44, 1202–1205.
(26) Mu¨ller, M.; Kusebauch, B.; Liang, G.; Beaudry, C. M.; Trauner, D.;
Hertweck, C. Angew. Chem., Int. Ed. 2006, 45, 7835–7838.
9
10408 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010

Focused Library of Highly Selective Aureothin Analogues
A R T I C L E S
Figure 2. Organization of the aur biosynthesis gene cluster and model of the aureothin biosynthetic assembly line. AurG: PABA synthase, AurF: N-oxygenase,
AurE: ligase, AurA-C: type I PKS, AurI: methyl transferase, AurH: Cyp450 monooxygenase.
even benzoic acid is loaded onto the aur PKS and processed to
a phenyl analogue of deoxyaureothin (14). The most surprising
observation, however, was that even feeding 2-naphthoic acid
yielded the corresponding pyrone (15). Another unexpected
finding is that m-nitro-BA was tolerated as a surrogate, too,
giving access to 16. However, feeding alternative PNBA
surrogates equipped with ortho- or meta-substituents or multiple
functionalizations were not accepted as priming units. Likewise,
no polyketide production was detected when supplementing the
mutant with p-hydroxy-BA, p-amino-BA, p-N-methyl-amino-
BA, p-N,N′-dimethyl-amino-BA or p-N-formamido-BA. Con-
Figure 3. Metabolic profile of ∆aurFH null mutant (S. albus::pMZ01)
sidering the tolerated substitutions, it appears that mutasynthons
supplemented with mixtures of mutasynthons (a) lacking PNBA (p-
with electron-acceptor groups result in a higher production rate
nitrobenzoate) and PCBA (p-cyanobenzoate), (b) lacking PNBA, (c)
of the corresponding deoxyaureothin derivative than those with
electron donor groups.
complete mixture, and (d) without supplement. (* indicates no relation to
aureothin biosynthesis)
To evaluate the viability of a parallel biosynthesis of aureothin
derivatives, we added mixtures of substituted benzoates to the
∆aurFH mutant (Figure 3a-c) and monitored the resulting
metabolic profiles. Interestingly, mainly deoxyaureothin and
deoxyaureonitrile are produced when adding the entire variety
of possible starter units (Figure 3c). We next omitted PNBA
and noted an increased production of deoxyaureonitrile (7) and
several halogenated analogues (8-11) in low amounts (Figure
3b). Feeding a mixture lacking p-nitro-BA and p-cyano-BA led
to a similar pattern but increased the yields of halogen- and
methyl-substituted analogues (Figure 3a). These results clearly
demonstrate the competition between the various possible starter
units and that the formation of products derived from less-
favored primers is repressed in the presence of p-nitro-BA and
p-cyano-BA.
reothin on a 50 mL scale. This result encouraged probing starter
unit surrogates. For testing the substrate tolerance of the PKS,
the mutant S. albus::pMZ01 was supplemented with a series of
substituted benzoates (BA, substitutions see below) in 25% aq
DMSO (c ) 100 µg mL-1) on a 50 mL scale. Initially, we tested
the impact of chemical starter unit activation. For this purpose,
we synthesized N-acetyl cysteamine (NAC) adducts that mimic
CoA thioesters and readily diffuse into bacterial cells. However,
when applying mutasynthons it proved to be irrelevant whether
the analogues were supplied as SNAC thioester or as free acids.
Apparently, the endogenous acyl-CoA ligase AurE is highly
promiscuous and does not represent a bottleneck for engineering
experiments.
The successful incorporation of the mutasynthons was
monitored by HPLC and HRMS. The expected masses and
isotope patterns were recorded for all deoxyaureothin derivatives
(7-11) resulting from the incorporation of all p-halogenated
BAs, and the pseudohalide p-cyanobenzoate. Likewise, we
observed the incorporation of p-methyl-BA and p-methoxy-
substituted benzoate, yielding 12 and 13. It is remarkable that
The p-cyanophenyl- and naphthoate-substituted derivatives
were obtained from two fermentations in liquid R2YE media
(20 L), yielding 2 mg L-1 of 7 and 1.8 mg L-1 of 15,
respectively. For the large scale production of p-halogenated
derivatives, we selected liquid E1 medium, which proved to be
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10409

A R T I C L E S
Werneburg et al.
Figure 5. Regiospecific pyrone O-methylation by AurI and EncK. HPLC
profiles of extracts from (a) ∆aurFHI triple mutant (S. albus::pMZ02),
complemented with encK (S. albus::pMZ02/pHJ72) supplemented with
PNBA, (b) ∆aurFH mutant (S. albus::pMZ01) supplemented with PNBA,
and (c) S. albus::pMZ02/pHJ72, d) S. albus::pMZ01.
Figure 4. Structures of deoxyaureothin derivatives.
superior and less expensive, providing deoxyaureothin deriva-
tives 8-11 in higher yields (R ) F, 5.4 mg L-1; R ) Cl, 3.1
mg L-1; R ) Br, 7.8 mg L-1; R ) I, 18 mg L-1). Taken
together, we have shown that the aur PKS is rather promiscuous
with regard to the starter units and that besides PNBA, ten
different starter units can be employed for the preparative-scale
precursor-directed biosynthesis of aureothin derivatives (Figure
4).
Figure 6. Regiospecific pyrone O-methylation by AurI and EncK.
Exploiting Plasticity in Pyrone Formation through Muta-
tional and Combinatorial Biosynthesis. At the pyrone moiety,
three different substitution patterns (R-methylated, γ-methylated,
and nonmethylated variants) are conceivable. For the production
of nordeoxyaureothin derivatives, a triple mutant lacking aurF,
aurH, and the O-methyltransferase gene aurI was constructed.
This was achieved by a PspXI restriction digest of cosmid
pMZ01, which harbors a variant of the aur gene cluster lacking
aurF and aurH. Because only two PspXI sites are present in
pMZ01, one within aurI and the other downstream of the aur
locus, the target gene could be excised through restriction and
religation. After verification of the correct construct, the resulting
cosmid, pMZ02, was introduced into S. albus by conjugation.
The exoconjugant S. albus::pMZ02 was tested at a 50 mL scale
for the ability to produce desmethylated analogues. Although
no polyketide production was observed for the mutant in liquid
R2YE complex medium, supplementation with PNBA yielded
nordeoxyaureothin (5), which features the more stable nonm-
ethylated R-pyrone ring.
PNBA surrogates were analogously accepted by the triple
mutant, yielding analogues such as nordeoxyaureonitrile (7) with
different aryl substitutions. Complementation of the triple mutant
with aurI restored the formation of the methylated γ-pyrone.
To expand the scope of substitutions, we aimed at generating
an O-metylated R-pyrone. As yet, the only known enzyme
catalyzing an O-methyl transfer of a pyrone metabolite is EncK
from the enterocin pathway in Streptomyces maritimus.38,39 To
test whether this enzyme has a sufficiently broad specificity to
methylate the aureothin pyrone ring, we planned to complement
the ∆aurFHI null mutant with encK. For this purpose, encK
was recovered from cosmid 1A9 as 1.8 kb EcoRI fragment and
cloned downstream of the constitutive ermE promoter of
pWHM4*. The resulting construct, pHJ72, was introduced
together with pMZ02 into S. albus. The resulting strain (S.
albus::pMZ02/pHJ72) was cultured with apramycin and thios-
treptone as selection markers, and its metabolic profile was
monitored by HPLC/MS. In fact, supplementation of the mutant
with PNBA yielded a new product (17). While 6 and 17 share
the same molecular mass, they have a markedly different
retention time in the HPLC profile, thus suggesting they
represent two regioisomers (Figure 5). This was firmly cor-
roborated by NMR data. The successful methylation of 5 using
EncK is surprising in light of the different polyketide core
structures of aureothin and enterocin (Figure 6). Notably, only
encK can complement the ∆aurI mutant and not vice versa (data
not shown), indicating that EncK has broader substrate specific-
ity than AurI. In sum, the ∆aurFHI mutant allows access to
aureothin derivatives with modified pyrone moieties. Upon
supplementation with PNBA or other synthetic analogues, it is
possible to yield various homologues.
(38) Piel, J.; Hertweck, C.; Shipley, P.; Hunt, D. S.; Newman, M. S.; Moore,
B. S. Chem. Biol. 2000, 7, 943–955.
(39) Cheng, Q.; Xiang, L.; Izumikawa, M.; Meluzzi, D.; Moore, B. S. Nat.
Chem. Biol. 2007, 3, 557–558.
9
10410 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010

Focused Library of Highly Selective Aureothin Analogues
A R T I C L E S
Figure 7. Biotransformation of R- and γ-pyrones by AurH.
Figure 8. Survey of conceivable aureothin derivatives.
three different types of pyrone moieties can be engineered
through biotransformation and combinatorial biosynthesis,
respectively (Figure 8). The only restriction is that the tetrahy-
drofuran ring is not readily formed in the presence of R-pyrones.
Accordingly, at least in principle, 77 aureothin derivatives would
be accessible through the combination of all methods presented.
However, not all compounds appeared to be equally profitable;
we found that nonmethylated R-pyrones and hydroxyl-substi-
tuted aureothin analogues (e.g., 20, 27) are prone to acid-
catalyzed degradation and rearrangement, which renders them
less favorable from a pharmacological point of view. Further-
more, initial biological assays indicated that a deoxyaureothin
derivative featuring an R-pyrone (17) moiety exhibit a markedly
lower antiproliferative effect (Huvec GI50 >20 µg mL-1)
compared to the γ-pyrone 6 (Huvec GI50 9.3 µg mL-1).
Furthermore, the R-pyrone proved to be slightly more cytotoxic,
which is against the desired disposition. In addition, the
methylated R-pyrone compounds do not show any antifungal
activity (Figure 10). We thus focused on the large scale
production of all p-cyano-, p-halogen-, and naphthoate-
substituted deoxyaureothin (7-11, 14, 15) and aureothin deriva-
tives (22-26) featuring the methylated γ-pyrone group. The
R-pyrone 17 and the 7-hydroxy substituted analogues 20 and
27 were also produced on a preparative scale for comparison.
The mutasynthesis experiments for the production of the
methylated γ-pyrones were performed by multiple 20 or 60 L
fermentations. The culture broths were neutralized with 1 N
HCl, and mycelia and culture filtrates were individually extracted
with ethyl acetate. Crude extracts were subjected to open column
chromatography (silica gel, normal phase) using a CHCl3/MeOH
gradient. Final purification was achieved by preparative reverse
phase HPLC. For subsequent biotransformations, the deoxyau-
reothin derivatives were transformed into the corresponding
aureothin derivatives using AurH in a whole cell approach as
described above, yielding 74% conversion.
Promiscuity of the Bifunctional Monooxygenase AurH in
Polyketide Hydroxylation and Heterocyclization. The sequential,
AurH-catalyzed hydroxylation-heterocyclization represents the
last step in the aureothin pathway. From initial mutasynthesis
experiments using the ∆aurF mutant, it appeared that various
aureothin derivatives are toxic for the producer strain, thus
limiting their continuous formation. For this reason, we chose
to generate aureothin derivatives containing the characteristic
chiral tetrahydrofuran moiety by AurH-mediated biotransfor-
mation of the deoxyaureothin derivatives. While the in vitro
biotransformation with AurH is in principle viable,35 an in vivo
approach using whole cells of a host expressing aurH40 proved
to be more efficient on the preparative scale. For this purpose,
the aureothin cytochrome P450 monooxygenase AurH was
heterologously expressed in S. albus and S. liVidans using
expression plasmid pHJ110. HPLC-MS monitoring of 50 mL
cultures of S. liVidans/pHJ110 supplemented with deoxyau-
reothin derivatives indicated in all cases the formation of novel
peaks with complete disappearance of the starting material
(Figure 7). AurH proved to be remarkably tolerant toward
artificial substrates. The oxygenase is capable of transforming
deoxyaureonitrile (7), all four p-halogenated derivatives (8-11),
and even the unusual naphthoate (15) into the corresponding
aureothin analogues 21-26. However, when R-pyrones like 17
were administered, no heterocycle formation was observed.
Instead, 7-hydroxy derivatives such as 20 were produced.
O-Heterocyclization obviously only takes place in the presence
of a methylated γ-pyrone moiety, where an exact enzyme fit is
warranted. If desired, it would also be possible to introduce a
7-hydroxyl moiety into methylated γ-pyrones and circumvent
THF ring formation. This is generally feasible when comple-
menting the aurH null mutant with norH from the neoaureothin
pathway, since NorH is not capable of catalyzing the full
reaction sequence when an aureothin backbone is supplied. In
this way, appreciable amounts of the 7-hydroxylated γ-pyrone
27 were obtained.
It should be highlighted that the structures of all new
compounds (7-11, 14, 15, 17, 20, 22-27) were fully elucidated
by 1D and 2D NMR experiments, high resolution mass
spectrometry, and infrared spectroscopy. NMR shifts of the core
structures were similar to those of 1 and 6, respectively. The
influence of the groups exerting different I- and M-effects was
clearly detected in shifts of the neighboring atoms. Signals of
the fluoro-substituted analogues showed typical C-F coupling
constants (ranging from 259 to 3 Hz) up to 6 atom bondings in
the 13C NMR. Two-dimensional NMR as well as NOESY
experiments fully support the proposed structures. HR-MS data
confirmed the expected molecular masses, and chloro- and
The scale-up for preparative in vivo biotransformations was
optimized using p-iododeoxyaureothin (11) dissolved in DMSO
(6.5 µmol, c ) 1 mg 25 µL-1) in 250 mL M10 media. As a
result of its reduced metabolic background S. albus/pHJ110
proved to be superior to S. liVidans. Varying amounts of 11
(6.5, 13, c. 26, and 52 µmol) were tested, revealing an optimum
of production when using the lowest quantity of the correspond-
ing aureothin analogue 25 (74%, 52%, 52%, 40%, respectively)
as quantified by HPLC.
Production of a Focused Library of Aureothin Derivatives
by Fermentation. The mutasynthesis experiments revealed that
at least 11 different starter units are accepted by the aureothin
PKS. Furthermore, three modes of backbone substitutions and
(40) Werneburg, M.; Hertweck, C. ChemBioChem 2008, 9, 2064–2066.
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10411

A R T I C L E S
Werneburg et al.
Figure 9. Antiproliferative effects of p-fluoro-aureothin (0, 22) and
p-fluoro-deoxyaureothin (], 8) on human K-562 leukemia cells displayed
in a dose-response curve; concentration vs proliferation (% vs control,
bold line).
bromo-derivatives showed the typical isotopic pattern (“A +
2”) in the mass spectra. By IR-spectroscopy the vibration of
the nitrile group as well as the C-Hal vibrations could be clearly
observed. The IR fingerprint (ca. 850-810 cm-1) indicated the
para-substitution of the phenyl headgroup.
Figure 10. Color-coded survey on antimicrobial (inhibition diameter in
mm), cytotoxic (IC50 µg mL-1) and antiproliferative (GI50 µg mL-1) activities
of aureothin (1) and its derivatives. Less effective activities indicated with
blue and color-staggered to dark red for the most effective ones. Test strains:
Bacillus subtilis ATCC 6633; Staphylococcus aureus SG511; Pseudomonas
aeruginosa K799/61; Enterococcus faecalis (VRSA) 1528; Mycobacterium
Vaccae 10670; Candida albicans BMSY 212; Sporobolomyces salmonicolor
549; Penicillium notatum JP36; Glomerella cingulata; Aspergillus fumigatus
ATCC 46645; Aspergillus terreus DSM 826; Aspergillus niger DSM 737.
Ny: nystatin.
Identification of Novel Aureothin Derivatives with Selective
Antifungal and Improved Antiproliferative Activities. The
biological activities of all new aureothin derivatives were
analyzed in a panel of assays with the cell lines human cervix
carcinoma HeLa, mouse fibroblast cells L-929, human umbilical
vein endothelial cells Huvec, as well as human K-562 leukemia
cells. Notably, almost all new compounds show increased
antiproliferative and cytostatic activities compared to those of
aureothin (1) and have significantly lower cytotoxicity. For
deoxyaureonitrile (7) and the halogenated deoxyaureothin
derivatives (8-11) a dramatic increase in antiproliferative
activities (up to 50-fold compared to 1) could be determined,
with GI50 values (K-562) as low as 0.5 µg mL-1. Furthermore,
the antiproliferative compounds show intriguing dose-response
curves with a plateau over several orders of magnitude (Figure
9). Such a flat curve is favorable considering a broad therapeutic
window. A combination of both a flat curve and low IC50 values
may be beneficial for potential therapeutic applications.
phytopathogenic fungus and causative agent of apple bitter rot.
Among all tested compounds, p-chloroaureothin 23 proved to
be the most active antifungal agent.
Comparing the activity profiles of deoxyaureothin and au-
reothin derivatives, it is obvious that the presence of the
tetrahydrofuran unit is critical for the antifungal activity, in
particular against yeasts. It should be highlighted that 22 and
24 are significantly more active against Aspergillus fumigatus,
Aspergillus terreus and Aspergillus niger than the standard
references nystatin and amphotericin B (not shown). The
selective activities against fungal strains are striking.
The most remarkable finding in our biological assays was
the selective antifungal activity of the new aureothin derivatives.
Using a panel of bacteria, yeasts and fungi (Figure 10), the novel
aureothin derivatives were compared to the known substances
deoxyaureothin (6), aureothin (1) and aureonitrile (21). Except
for deoxyaureonitrile (7), deoxyaureothin analogues do not show
any notable antibacterial activities. Compared to deoxyaureothin
(6), deoxyaureonitrile (7) exhibits a higher activity against
Sporobolomyces salmonicolor and Penicillium notatum. In
contrast, the new aureothin derivatives 22-25 show activities
similar to those of 1 and 21 against these test strains as well as
the important pathogen Candida albicans. Because of the
remarkable activity of various aureothin derivatives against the
yeast Sporobolomyces salmonicolor, the facultative human
pathogen Candida albicans, and the mold Penicillium notatum,
an extended array of fungal test strains including human
pathogenic Aspergillus spp. was probed. Deoxyaureonitrile (7)
shows selective antifungal activity against Aspergillus niger and
Aspergillus fumigatus, the major cause for human aspergillosis.
Compounds 22-25 are even more active against aspergilli, in
particular the pathogen Aspergillus terreus. Although 22 and
24 do not show any activity against Candida glabrata, they
effectively inhibit the growth of Glomerella cingulata, the
Conclusion
Aureothin is a densely functionalized natural product with
intriguing biological activities, yet toxic and pharmacologically
unfavorable due to the nitroaryl substitutent. The aim of our
study was to generate a series of aureothin analogues for
structure-activity relationship (SAR) studies. Because of the
relatively small size of both molecule and the corresponding
biosynthesis gene cluster, this target was ideally suited as a
model for combining genetic, biochemical, and chemical
methods to generate structural diversity. Furthermore, the
aureothin pathway exhibits three principal levels that could be
envisaged for diversification: the aryl starter unit, the oxygenated
backbone, and the pyrone headgroup. To efficiently replace the
nitro substituent, we chose the concept of bioisosterism that is
well-known in drug design.41 This goal was achieved through
mutasynthesis using an engineered ∆aurF/aurH null mutant that
exhibits a broad substrate tolerance toward 10 alternative PKS
starter units. Furthermore, in a combinatorial biosynthesis
approach, we found that the endogenous γ-regiospecific pyrone
methyltransferase (AurI) can be replaced by an R-regiospecific
(41) Lima, L. M. A.; Barreiro, E. J. Curr. Med. Chem. 2005, 12, 23–49.
9
10412 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010

Focused Library of Highly Selective Aureothin Analogues
A R T I C L E S
homologue from the enterocin pathway (EncK), yielding the
corresponding iso-deoxyaureothin derivatives. Finally, the au-
reothin cytochrome P450 monooxygenase, AurH, was used as
a tool for whole cell enzymatic hydroxylation and heterocy-
clization, repectively, providing a convenient access to the chiral
tetrahydrofuran moiety. By this highly integrative approach in
total 15 aureothin analogues were generated on a preparative
scale. A biological evaluation of these compounds revealed new
insights into structure-activity relationships and identified
various promising candidates. Some of the new deoxyaureothin
derivatives are less cytotoxic but stronger antiproliferative agents
than the parent natural product. Furthermore, some of the new
aureothin derivatives hold promise as highly selective antifun-
gals, in particular anti-Candida, and it seems that the THF ring
markedly increases antifungal activity. Our results demonstrate
that exploiting enzymatic promiscuity may well complement
synthetic approaches to generate structural diversity. Thus, our
results are another illustration for the viability and potential of
biosynthetic engineering.
Acknowledgment. This work has been financially supported
by the DFG within the framework of SPP1152 and the NIH
(AI47818). We thank C. Heiden and M. Steinacker for performing
fermentation and downstream processings, U. Wohlfeld and M. G.
Schwinger for antimicrobial assays, and E.-M. Neumann for
antiproliferative and cytotoxic assays, and A. Perner and F. A.
Gollmick are acknowledged for MS and NMR measurements,
respectively.
Supporting Information Available: Experimental details,
physicochemical data, biological assays, and NMR spectra. This
material is available free of charge via the Internet at http://
pubs.acs.org.
JA102751H
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10413