A nonribosomal peptide synthetase-derived iron(III) complex from the pathogenic fungus aspergillus fumigatus

Article abstract of DOI:10.1021/ja311145n

Small molecules (SMs) play central roles as virulence factors of pathogenic fungi and bacteria; however, genomic analyses suggest that the majority of microbial SMs have remained uncharacterized. Based on microarray analysis followed by comparative metabolomics of overexpression/knockout mutants, we identified a tryptophan-derived iron(III)-complex, hexadehydro-astechrome (HAS), as the major product of the cryptic has nonribosomal peptide synthetase (NRPS) gene cluster in the human pathogen Aspergillus fumigatus. Activation of the has cluster created a highly virulent A. fumigatus strain that increased mortality of infected mice. Comparative metabolomics of different mutant strains allowed to propose a pathway for HAS biosynthesis and further revealed cross-talk with another NRPS pathway producing the anticancer fumitremorgins.

Full text of DOI:10.1021/ja311145n

Communication
pubs.acs.org/JACS
A Nonribosomal Peptide Synthetase-Derived Iron(III) Complex from
the Pathogenic Fungus Aspergillus fumigatus
Wen-Bing Yin,†,§,# Joshua A. Baccile,∥,# Jin Woo Bok,† Yiming Chen,† Nancy P. Keller,*,†,‡
and Frank C. Schroeder*,∥
†
‡
Departments of Medical Microbiology and Immunology and Bacteriology, University of WisconsinMadison, Wisconsin 53706,
United States
∥Boyce Thompson Institute and Department of Chemistry and Chemical Biology, Cornell University, Ithaca, New York 14853,
United States
S
* Supporting Information
transcription factors. Because hasA was the third most down-
ABSTRACT: Small molecules (SMs) play central roles as
regulated gene in the ΔlaeA array and strongly induced in
conidia during neutrophil exposure,6,12−14 we chose to
virulence factors of pathogenic fungi and bacteria;
however, genomic analyses suggest that the majority of
microbial SMs have remained uncharacterized. Based on
microarray analysis followed by comparative metabolomics
of overexpression/knockout mutants, we identified a
tryptophan-derived iron(III)-complex, hexadehydro-aste-
chrome (HAS), as the major product of the cryptic has
nonribosomal peptide synthetase (NRPS) gene cluster in
the human pathogen Aspergillus fumigatus. Activation of
the has cluster created a highly virulent A. fumigatus strain
that increased mortality of infected mice. Comparative
metabolomics of different mutant strains allowed to
propose a pathway for HAS biosynthesis and further
revealed cross-talk with another NRPS pathway producing
the anticancer fumitremorgins.
overexpress (OE) this gene to facilitate chemical character-
ization of has-associated metabolites.18,19 Replacement of the
hasA promoter with the constitutive gpdA promoter (OE::ha-
sA) resulted in induction of the other seven genes (Figure 1b)
and yielded a pink pigmented strain (Figure S1a). Three cluster
flanking genes (AFUB_036320, AFUB_036310, and
AFUB_036220, Figure 1a,b) were not regulated by hasA and
thus likely defined the boundaries of the cluster. To probe the
functions of the individual has genes, we additionally
constructed knockout mutants (KO) for hasB-hasH in the
OE::hasA background (Figure S2).
Previous biochemical characterization of the DMATS HasE
had proven its capacity to prenylate a variety of indole
derivatives in position 7, with L-tryptophan as a preferred
substrate;17 however, no cluster-associated metabolites were
identified. We employed HPLC-UV-MS (Figure 1c) and NMR-
based comparative metabolomics8,19,20 to probe for has-
dependent metabolites in the OE::hasA strain. Analysis of the
1H NMR spectrum from the OE::hasA extract revealed signals
indicative of a 3-substituted indole moiety that were absent in
the WT extract spectrum (Figure S3). All signals of this
compound suffered from intense line broadening, suggestive of
a paramagnetic species such as an iron complex. Comparison of
high-resolution (HR)-HPLC-MS data from the OE::hasA and
WT extracts showed a OE::hasA-specific ion cluster eluting at
46.8 min whose m/z values and isotopic distribution suggested
a molecular formula of C60H60FeN9NaO9 (m/z calcd.
1129.3761; obsvd. 1129.3751), suggesting an Fe(III) complex.
Supporting the requirement of iron for production of this
metabolite, OE::hasA cultures in iron-deficient medium did not
produce the peak at 46.8 min or the pink color (Figure. S1b).
Because extreme line-broadening prevented full spectroscopic
characterization of the has-dependent iron complex, we treated
the isolated starkly pink compound with strong base to
precipitate iron as the hydroxide. Following iron removal, we
obtained the O-methylated diketopiperazine derivative 9, whose
structure was determined using a routine set of 2D NMR
Aspergillus fumigatus is a ubiquitous filamentous fungus and
causative agent of invasive aspergillosis (IA), a life-threatening
disease. IA associated mortality ranges from 50% to 90%.1 One
of the few proteins that have been shown to be required for A.
fumigatus’ pathogenicity is LaeA, a conserved virulence factor in
filamentous fungi.2−5 LaeA is a global regulator of secondary
metabolism in fungi and effects virulence, in part, by positive
regulation of multiple small-molecule gene clusters.3−6 For
example, LaeA controls the biosynthesis of the epidithiodike-
topiperazine gliotoxin,7−10 which contributes to A. fumigatus’
success as a pathogen.11
We reasoned that SMs most likely to have a role in virulence
would be positively regulated by both laeA and exposure to
host environments (host exposure itself or hypoxic environ-
ments). By comparing microarrays of genes down-regulated in
laeA deletants but up-regulated by host exposure/hypo-
xia,6,12−14 we identified an eight-gene SM cluster named has
meeting these criteria. The eight cluster genes encode putative
C6 transcription factors (HasA and HasF), one transporter
(HasB), one O-methyltransferase (HasC), one nonribosomal
peptide synthetase (NRPS, HasD, previously denoted as Nrps
515 and PesF16), one 7-dimethylallyltryptophan synthase
(DMATS)17 (HasE), one putative FAD binding protein
(HasG) and one cytochrome P450 (HasH) (Figure 1a and
Tables S1−S3). Often SM gene clusters are regulated by C6
Received: November 13, 2012
Published: January 29, 2013
© 2013 American Chemical Society
2064
dx.doi.org/10.1021/ja311145n | J. Am. Chem. Soc. 2013, 135, 2064−2067

Journal of the American Chemical Society
Communication
Figure 1. hasA controls has cluster gene expression and metabolite production. (a) has gene cluster and putative assignments. hasA and hasF, C6
transcription factors; Trs, transporter; MT, methyltransferase; NRPS, nonribosomal peptide synthetase; DMATS, dimethylallyltryptophan synthase;
FAD, FAD-binding domain protein; P450, cytochrome P450. Flanking genes are shown in white. (b) Northern analysis of has cluster gene
expression in the WT and OE::hasA strains. (c) HPLC-UV chromatograms for A. fumigatus mutant strains and structures of has-dependent
metabolites; (*) denotes HAS decomposition products. Ergosterol amounts were highly variable between replicates and are not has-dependent.
spectra and HR-MS (Figure 1c). On the basis of the combined
spectroscopic data, the iron-containing metabolite was
identified as a trimeric complex of 9 with Fe(III), similar to
the previously described Aspergillus terreus metabolite, aste-
chrome (1113.1, [M+H]+).21 Accordingly, we named the has-
dependent metabolite we identified from A. fumigatus
hexadehydroastechrome “HAS” (1; Figure 1c).
abolished HAS production and instead yielded large quantities
of a single shunt metabolite, the prenylated Trp-Ala
diketopiperazine, terezine D (7),24 small amounts of which
were also detected in the parent OE::hasA strain. In the
OE::hasA strain, but not the P450 knock out, terezine D was
accompanied by its hydroxy derivative, 8.24 Deletion of the
putative O-methyltransferase (OE::hasAΔhasC) unexpectedly
led to production of large quantities of L-7-dimethylallyl-
tryptophan (3), as well as the trans- and cis-isomers of L-7-(3-
methylbutadienyl)tryptophan (4a and 4b, respectively), where-
as production of all diketopiperazine derivatives was completely
abolished. L-configuration of compounds 3, and 4a/b was
confirmed by converting them to their corresponding Mosher
amides (Figure S4).25 In contrast, the NRPS deletant
(OE::hasAΔhasD) did not accumulate prenylated tryptophan
derivatives such as 3 or 4a/b, suggesting that prenylation of
tryptophan may not precede formation of NRPS-tethered
tryptophan. Lastly, the FAD deletant (OE::hasAΔhasG) yielded
the previously described A. terreus metabolite, astechrome (2)
instead of HAS (Figure 2b and Table S4), suggesting that HasG
converts the prenyl to a methylbutadienyl side chain.
Bioinformatic examination of the A. terreus genome revealed a
gene cluster with close homology to the A. fumigatus has
cluster; however, the A. terreus cluster lacks a gene encoding a
homologue of the FAD HasG (Figure S5), supporting that
HasG is required for introducing the diene moiety in the side
chain (Figure 2a). We further noted that isolated samples of the
cis-isomer 4b slowly isomerized to 4a, suggesting that HasG
produces cis-7-(3-methylbutadienyl)indole derivatives, which
To gain insight into HAS biosynthesis, we examined the
metabolomes of the seven double mutants carrying deletions of
the other has genes in OE::hasA background (Figure S2) by 2D
NMR and HPLC-UV-MS for the presence of pathway
intermediates or shunt metabolites, focusing on compounds
whose UV and NMR spectra suggested incorporation of
tryptophan. Putative pathway-related metabolites were charac-
terized by HR-MS and purified as needed to complete NMR-
spectroscopic assignments (Tables S4−S9). Loss of the second
transcription factor (OE::hasAΔhasF) did not result in any
detectable metabolomic change compared to the OE::hasA
strain, indicating that under the growth conditions used in this
study HasF does not affect the has pathway (Figure 1c and
Figure S2). In contrast, deletion of the transporter (HasB) led
to complete abolishment of HAS production and no
intermediates or shunt metabolites were observed (Figure
1c), which is not unexpected given that deletions of SM cluster
transporter genes have previously been shown to inhibit SM
production.22,23
The metabolite profiles of several other double mutant
strains provided important clues for the HAS biosynthetic
pathway. Deletion of the cytochrome P450 (OE::hasAΔhasH)
2065
dx.doi.org/10.1021/ja311145n | J. Am. Chem. Soc. 2013, 135, 2064−2067

Journal of the American Chemical Society
Communication
Figure 2. Model for HAS biogenesis and crosstalk with the ftm gene cluster. (a) Putative has biosynthetic pathway (A = adenylation, T = thiolation,
C = condensation domains16,17), (*) order of enzyme activity unclear. (b) ftm gene cluster expression in two has mutants.
then slowly convert to the corresponding trans-isomers such as
4a and 9 during extraction and isolation. Therefore, isolated
hexadehydroastechrome (1) is shown as trans in Figure 1.
These results are consistent with a biogenesis model in which
the NRPS is loaded with tryptophan and alanine, followed by
prenylation of the tethered tryptophan or the resulting tethered
Trp-Ala-dipeptide (Figure 2a). Deletion of the P450 hasH
appears to stall further processing of the tethered dipeptide,
which then undergoes spontaneous cyclization to form the
diketopiperazine, terezine D, the only has pathway metabolite
detected in the OE::hasAΔhasH strain. Our observation that
the OE::hasAΔhasC (MT) mutant strain does not accumulate
diketopiperazines, but instead yields large quantities of
prenylated tryptophan derivatives, suggests a model in which
HAS biosynthesis continues via tethered intermediates as
shown in Figure 2a, as opposed to an alternative pathway
involving further modification of terezine D-derived diketopi-
perazines. In this model, the presence of prenylated tryptophan
derivatives 3 and 4a/b could result from enzymatic hydrolysis
of a tethered, partially oxidized dipeptide, for example by the
DMATS HasE, which has previously been reported to
hydrolyze linear dipeptides such as of H-L-Trp-L-Ala-OH, but
not diketopiperazines.26 Deletion of the prenyltransferase
(DMATS, HasE) led to complete abolishment of HAS
production and no intermediates or shunt metabolites were
observed.
Surprisingly, we found that the NRPS deletant (OE::ha-
sAΔhasD), although producing no HAS intermediates, yielded
large amounts of metabolites derived from the partially
characterized ftm (fumitremorgin) NRPS cluster, including
brevianamide F, fumitremorgin C and demethoxy-fumitremor-
gin C (5a, 5b, 6, Figure 1c). Fumitremorgin biosynthesis also
involves prenylation of tryptophan derivatives;18 however, the
ftm cluster does not contain a pathway-specific transcription
factor. Examination of ftm gene expression in wild type (WT),
OE::hasA, and OE::hasAΔhasD strains showed greatly
increased ftm cluster gene expression in the latter strain
(Figure 2b), consistent with the observation of increased ftm-
metabolite production in OE::hasAΔhasD. These results
suggest that tryptophan and prenylation activity is shunted
toward fumitremorgin biosynthesis when the has pathway is
shut down in OE::hasA background. Interestingly, fumitre-
morgin production was also increased in NRPS deletants which
abolished fumigaclavine C production,27 possibly due to a
similar mechanism.
To gain initial insight into the possibility that HAS might
contribute to virulence, WT, OE::hasA and OE::hasAΔhasD
strains were compared in a neutropenic murine pulmonary
model. Infection with the OE::hasA strain resulted in
significantly higher mortality than infection with WT and
OE::hasAΔhasD (Figure 3a). Murine lung tissue from all
Figure 3. The has pathway affects survival in a mouse model of IA. (a)
Mantel-Cox survival analysis of mice infected with WT, OE::hasA, or
OE::hasAΔhasD A. fumigatus in a neutropenic pulmonary mouse
model of IA. OE::hasA is significantly more virulent than WT or
OE::hasAΔhasD (P < 0.05, Log-Rank Test). There was no significant
difference between WT or OE::hasAΔhasD strains. (b) MS ion
chromatograms for terezine D (7) standard and lung extracts from
mice infected with WT, OE::hasA, or OE::hasAΔhasD.
treatments were examined for the presence of has metabolites.
Whereas none of the identified has-dependent metabolites
could be detected in the WT- and OE::hasAΔhasD-infected
samples, tissue samples from OE::hasA-inoculated mice
revealed the presence of terezine D (Figure 3b).24 HAS was
not detected in the OE::hasA lung extracts; however, based on
our experience with isolating the iron(III) complex from fungal
cultures, this main product of the has pathway is likely too
2066
dx.doi.org/10.1021/ja311145n | J. Am. Chem. Soc. 2013, 135, 2064−2067

Journal of the American Chemical Society
Communication
(2) Bok, J. W.; Balajee, S. A.; Marr, K. A.; Andes, D.; Nielsen, K. F.;
Frisvad, J. C.; Keller, N. P. Eukaryotic Cell 2005, 4, 1574.
(3) Amaike, S.; Keller, N. P. Eukaryotic Cell 2009, 8, 1051.
(4) Wiemann, P.; Brown, D. W.; Kleigrewe, K.; Bok, J. W.; Keller, N.
P.; Humpf, H. U.; Tudzynski, B. Mol. Microbiol. 2010, 77, 972.
(5) Wu, D.; Oide, S.; Zhang, N.; Choi, M. Y.; Turgeon, B. G. PLoS
Pathog. 2012, 8, e1002542.
(6) Perrin, R. M.; Fedorova, N. D.; Bok, J. W.; Cramer, R. A.;
Wortman, J. R.; Kim, H. S.; Nierman, W. C.; Keller, N. P. PLoS Pathog.
2007, 3, 508.
(7) Scharf, D. H.; Remme, N.; Habel, A.; Chankhamjon, P.;
Scherlach, K.; Heinekamp, T.; Hortschansky, P.; Brakhage, A. A.;
Hertweck, C. J. Am. Chem. Soc. 2011, 133, 12322.
(8) Forseth, R. R.; Fox, E. M.; Chung, D.; Howlett, B. J.; Keller, N.
P.; Schroeder, F. C. J. Am. Chem. Soc. 2011, 133, 9678.
(9) Scharf, D. H.; Chankhamjon, P.; Scherlach, K.; Heinekamp, T.;
Roth, M.; Brakhage, A. A.; Hertweck, C. Angew. Chem., Int. Ed. 2012,
51, 10064.
(10) Nicolaou, K. C.; Lu, M.; Totokotsopoulos, S.; Heretsch, P.;
Giguere, D.; Sun, Y. P.; Sarlah, D.; Nguyen, T. H.; Wolf, I. C.; Smee,
D. F.; Day, C. W.; Bopp, S.; Winzeler, E. A. J. Am. Chem. Soc. 2012,
134, 17320.
unstable to permit isolation from animal tissue. Our
observation that in laboratory culture OE::hasAΔhasD
produces large amounts of fumitremorgins suggests the
possibility that crosstalk with other metabolic pathways may
in part be responsible for the increased virulence of OE::hasA
in mice. However, fumitremorgins could not be detected in any
of the mouse lung tissue samples, and production of gliotoxin,
another metabolite contributing to A. fumigatus virulence,11 was
not increased in either the OE::hasA or OE::hasAΔhasD
samples relative to WT (Figure S6).
In conclusion, we show that microarray analysis coupled with
comparative metabolomics using knockout and overexpression
A. fumigatus mutants enables identification of potential
virulence factors of a pathogenic fungus and proposing a
tentative biosynthetic pathway. Examination of available fungal
genomes indicates that has-related gene clusters are found only
in a few opportunistic human pathogenic Aspergilli and related
dermatophytic fungi of the Trichophyton and Arthroderma
genera (Figure S5), providing additional motivation to further
investigate the impact of HAS and related metabolites on
virulence. Because of its iron-binding properties, we considered
the possibility that 9 could be acting as a siderophore, as
siderophores are well-known virulence factors in A. fumigatus
and other fungi.28 However, for entropic reasons, it seems
unlikely that the hydroxamic acid 9 is able to compete with
potent multidentate siderophores such as triacetylfusarinine C
(TAFC) or ferricrocin.28 Correspondingly, we found that
neither 9 nor 1 was produced by OE::hasA cultures grown in
iron-deficient medium, whereas TAFC production was strongly
upregulated (Figure S7). Nevertheless, the iron requirement for
HAS biosynthesis suggests that the has pathway and side-
rophore production may interact. In addition, the highly
lipophilic trimer HAS may function as a vehicle for the delivery
of its much more polar monomer 9 and/or serve to inflict
oxidative damage.
(11) Dagenais, T. R. T.; Keller, N. P. Clin. Microbiol. Rev. 2009, 22,
447.
(12) Sugui, J. A.; Kim, H. S.; Zarember, K. A.; Chang, Y. C.; Gallin, J.
I.; Nierman, W. C.; Kwon-Chung, K. J. PLoS One 2008, 3, e2655.
(13) Willger, S. D.; Puttikamonkul, S.; Kim, K. H.; Burritt, J. B.;
Grahl, N.; Metzler, L. J.; Barbuch, R.; Bard, M.; Lawrence, C. B.;
Cramer, R. A., Jr. PLoS Pathog. 2008, 4, e1000200.
(14) Vodisch, M.; Scherlach, K.; Winkler, R.; Hertweck, C.; Braun, H.
P.; Roth, M.; Haas, H.; Werner, E. R.; Brakhage, A. A.; Kniemeyer, O.
J. Proteome Res. 2011, 10, 2508.
(15) Cramer, R. A., Jr.; Stajich, J. E.; Yamanaka, Y.; Dietrich, F. S.;
Steinbach, W. J.; Perfect, J. R. Gene 2006, 383, 24.
(16) Stack, D.; Neville, C.; Doyle, S. Microbiology 2007, 153, 1297.
(17) Kremer, A.; Westrich, L.; Li, S. M. Microbiology 2007, 153, 3409.
(18) Winter, J. M.; Behnken, S.; Hertweck, C. Curr. Opin. Chem. Biol.
2011, 15, 22.
(19) Forseth, R. R.; Schroeder, F. C. Curr. Opin. Chem. Biol. 2011, 15,
38.
ASSOCIATED CONTENT
(20) Pungaliya, C.; Srinivasan, J.; Fox, B. W.; Malik, R. U.; Ludewig,
A. H.; Sternberg, P. W.; Schroeder, F. C. Proc. Natl Acad. Sci. U.S.A.
2009, 106, 7708.
(21) Arai, K.; Sato, S.; Shimizu, S.; Nitta, K.; Yamamoto, Y. Chem.
Pharm. Bull. 1981, 29, 1510.
(22) Wiemann, P.; Willmann, A.; Straeten, M.; Kleigrewe, K.; Beyer,
M.; Humpf, H. U.; Tudzynski, B. Mol. Microbiol. 2009, 72, 931.
(23) Lee, S.; Son, H.; Lee, J.; Lee, Y. R.; Lee, Y. W. Curr. Genet. 2011,
57, 343.
(24) Wang, Y.; Gloer, J. B.; Scott, J. A.; Malloch, D. J. Nat. Prod.
1995, 58, 93.
(25) Dale, J. A.; Mosher, H. S. J. Am. Chem. Soc. 1973, 95, 512.
(26) Kremer, A.; Li, S. M. Chem. Biol. 2008, 15, 729.
(27) O’Hanlon, K. A.; Gallagher, L.; Schrettl, M.; Jochl, C.;
Kavanagh, K.; Larsen, T. O.; Doyle, S. Appl. Environ. Microbiol.
2012, 78, 3166.
■
S
* Supporting Information
Methods and details regarding enzyme assays and analytical
methods. This material is available free of charge via the
Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
npkeller@wisc.edu; schroeder@cornell.edu
Present Address
§Department of Chemical and Biomolecular Engineering,
University of California, Los Angeles, CA, United States.
Author Contributions
#These authors contributed equally.
(28) Haas, H. Front. Microbiol. 2012, 3, 28.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This research was funded by NIH 1 R01 Al065728-01 to N.P.K
and a DuPont Young Investigator Award to F.C.S. We thank
Maciej Kukula for assistance with high-resolution mass
spectrometry.
REFERENCES
(1) Latge, J. P. Clin. Microbiol. Rev. 1999, 12, 310.
■
2067
dx.doi.org/10.1021/ja311145n | J. Am. Chem. Soc. 2013, 135, 2064−2067