A dedicated glutathione S -transferase mediates carbon-sulfur bond formation in gliotoxin biosynthesis

Article abstract of DOI:10.1021/ja201311d

Gliotoxin is a virulence factor of the human pathogen Aspergillus fumigatus, the leading cause of invasive aspergillosis. Its toxicity is mediated by the unusual transannular disulfide bridge of the epidithiodiketopiperazine (ETP) scaffold. Here we disclose the critical role of a specialized glutathione S-transferase (GST), GliG, in enzymatic sulfurization. Furthermore, we show that bishydroxylation of the diketopiperazine by the oxygenase GliC is a prerequisite for glutathione adduct formation. This is the first report of the involvement of a GST in enzymatic C-S bond formation in microbial secondary metabolism.

Full text of DOI:10.1021/ja201311d

COMMUNICATION
pubs.acs.org/JACS
A Dedicated Glutathione S-Transferase Mediates CarbonÀSulfur Bond
Formation in Gliotoxin Biosynthesis
Daniel H. Scharf,† Nicole Remme,† Andreas Habel,† Pranatchareeya Chankhamjon,† Kirstin Scherlach,†
Thorsten Heinekamp,† Peter Hortschansky,† Axel A. Brakhage,†,‡ and Christian Hertweck*,†,‡
†Leibniz Institute for Natural Product Research and Infection Biology — Hans Kn€oll Institute, Beutenbergstr. 11a, 07745 Jena, Germany
‡Friedrich Schiller University, Jena, Germany
S Supporting Information
b
ABSTRACT: Gliotoxin is a virulence factor of the human
pathogen Aspergillus fumigatus, the leading cause of invasive
aspergillosis. Its toxicity is mediated by the unusual transan-
nular disulfide bridge of the epidithiodiketopiperazine
(ETP) scaffold. Here we disclose the critical role of a
specialized glutathione S-transferase (GST), GliG, in enzy-
matic sulfurization. Furthermore, we show that bishydrox-
ylation of the diketopiperazine by the oxygenase GliC is a
prerequisite for glutathione adduct formation. This is the
first report of the involvement of a GST in enzymatic CÀS
bond formation in microbial secondary metabolism.
liotoxin (1) is the prototype of the infamous family of
epidithiodiketopiperazine (ETP) toxins that are produced
G
by a wide range of microorganisms.1,2 A hallmark of ETPs is the
cyclopeptide scaffold equipped with a transannular disulfide
bridge. This outstanding structural unit is indeed responsible
for deleterious effects of the toxins, as it readily mediates intra-
cellular redox cycling and inactivates vital proteins by conju-
gation.1À4 Endowed with these adverse properties, gliotoxin
significantly contributes to the virulence of the human pathogen
Aspergillus fumigatus. Notably, invasive aspergillosis caused by
this fungus is the leading cause of death in immunocompromised
patients.5,6 Since knowledge of the molecular basis and enzymol-
ogy of ETP biosynthesis could greatly aid in developing diag-
nostics and antifungal therapy, this area has been the focus of
intense research over the past years. The full genome sequencing
of A. fumigatus has set the stage for identifying the gliotoxin (gli)
biosynthesis gene cluster (Figure 1A). Various work groups have
since independently demonstrated that the diketopiperazine core
[cyclo-L-Phe-L-Ser (2)] is assembled by GliP, a bimodular non-
ribosomal peptide synthetase (NRPS).7À11 More recently, we
unveiled by in vivo and in vitro studies that an FAD-dependent
oxidoreductase, GliT, generates the intramolecular disulfide
bond from the corresponding dithiol 3 (Scheme 1).12 Never-
theless, it has remained fully enigmatic how the sulfur is incorpo-
rated into the cyclopeptide framework. Here we report that a
CYP450 monooxygenase and a specialized glutathione S-trans-
ferase (GST) play a key role in gliotoxin CÀS bond formation.
Through bioinformatic analysis of the gli biosynthesis gene
cluster, we identified a gene (gliG) that could code for a GST.
Enzymes belonging to the diverse GST family are best known for
Figure 1. (A) Organization of the gli biosynthesis gene cluster.
(B) Southern blot analysis of gDNA of the wild type (wt) and the ΔgliG
andΔgliC mutants. (C) Metabolic profiles(extractedion chromatograms)
of (a) wt, (b) ΔgliG mutant, (c) ΔgliC mutant, and (d) references of 1 and
6. (D) Cladogram of GliG and homologous GST sequences (the detailed
phylogeny is given in the Supporting Information).
their ability to form glutathione (GSH) conjugates with xeno-
biotics and leucotrienes and to harness reactive oxygen species
and radicals.13 However, to our knowledge, despite their wide-
spread occurrence,14 they have not been implicated in fungal
biosynthetic pathways. This is particularly remarkable as we
noted that orthologues of GliG are encoded in all presently
known ETP biosynthesis gene clusters.15 Furthermore, a phylo-
genetic analysis revealed that the deduced gene products even
represent a new clade (I) in the GST dendrogram (Figure 1D).
Received: February 11, 2011
Published: July 12, 2011
r
2011 American Chemical Society
12322
dx.doi.org/10.1021/ja201311d J. Am. Chem. Soc. 2011, 133, 12322–12325
|

Journal of the American Chemical Society
COMMUNICATION
Scheme 1. Structures of Gliotoxin (1) and Its Known Bio-
synthetic Precursors, Dithiol 3 and Diketopiperazine 2
We thus set out to investigate the role of GliG in gliotoxin
biosynthesis and deleted the corresponding gene in the A.
fumigatusgenomethrough homologous recombination (Figure1).
We noted that the growth of the ΔgliG mutant is comparable to
that of the wild type upon addition of 1. The lack of sensitivity of
the mutant toward the toxin shows that GliG is not involved in
detoxification of toxins as other GSTs are. HPLCÀMS profiling
of the resulting mutant broth indicated that the lack of GliG led
to a complete abrogation of gliotoxin biosynthesis (Figure 1C).
While this result unequivocally showed that gliG is essential for
gliotoxin formation, in vitro studies were required to gain insight
into its exact function. Therefore, we amplified the gliG open
reading frame (720 base pairs) by PCR and cloned the amplicon
into an Escherichia coli expression vector. After propagation of the
plasmid and sequencing of the insert, the vector was introduced
into E. coli BL21(DE3) cells for protein overproduction. After
the cells were harvested, His6-tagged GliG was purified using a
NiSepharose column (Figure 2A). Size-exclusion chromatogra-
phy (SEC) revealed the homodimeric status of GliG (Figure 2B).
To identify conditions favoring the stability of GliG, we per-
formed melting temperature analysis (Figure 2C). On the basis
of this, we optimized the starting purification conditions of the
protein and the storage buffer and eventually established a one-
step purification that led to pure, soluble His6-tagged GliG. Inter-
estingly, like the phylogenetically related theta-class proteins of
the GST superfamily, GliG lacks the ability to bind to GSH
sepharose. Furthermore, it does not employ standard GST sub-
strates such as 1-chloro-2,4-dinitrobenzene (CDNB) and 1,2-
epoxy-3-(nitrophenoxy)propane (EPNPP). The alignment of
GliG with several homologues shows a division of the protein
into two conserved domains, as shown for other GSH-binding
proteins. The N-terminal domain should be involved in GSH
binding, whereas the second domain might be involved in
binding of the cosubstrate and protein dimerization.
Apparently, GliG does not employ any of the standard sub-
strates. To identify a candidate that could have accumulated in
the ΔgliG mutant, we investigated the culture extract of the ΔgliG
mutant. Using HPLCÀMS in the positive-ion mode, we noted
the accumulation of a new compound, 6, that forms an ion with
m/z 263 ([M + H]+). The compound was isolated from an
upscaled mutant culture and then purified by open column chro-
matography on silica, SEC, and preparative HPLC, after which its
structure was elucidated by high-resolution electrospray ioniza-
tion mass spectrometry (HR-ESI-MS) and 1D and 2D NMR
measurements. From the HR-ESI-MS data, a molecular compo-
sition of C13H14N2O4 was deduced, which was corroborated by
the 1D NMR data. The 13C NMR and 135° distortionless en-
hancement by polarization transfer (DEPT135) spectra showed
the presence of two amide carbons, an aromatic ring system, and
an O-methyl function as well as two methylene carbon atoms. H,
H correlation spectroscopy (H,H-COSY) and heteronuclear
multiple-bond correlation (HMBC) measurements established
Figure 2. Characterization and in vitro reconstitution of GliG activity.
(A) Molecular mass of His6-tagged GliG analyzed by SDS-PAGE.
(B) Native molecular size determination by gel filtration. (C) Three-
dimensional plot of GliG melting temperature as a function of pH and
NaCl concentration. (D) HLPC monitoring of the biotransformation
reaction. Extracted ion chromatograms are shown for (a) GliG-mediated
transformation of 4 from the ΔgliG mutant; (b) same as (a) but using
heat-inactivated GliG (residual activity after 10 min; longer heating
led to complete loss of activity); (c) negative control lacking GliG;
(d) negative control using the crude extract from the ΔgliC mutant.
(E) HR-MSn analysis of GT conjugate. (F) Structures of the proposed
bishydroxylated intermediates (4a and 4b) and shunt product 6, HMBC
correlations for 6, and a model of the enzyme-mediated hydroxylationÀ
sulfurization sequence.
the structure of 6 as a diketopiperazine with an exo-methylene
group. A chemical shift of 82.4 ppm for C6 disclosed two
neighboring heteroatoms. Since no coupling of the O-methyl
function was observed, its connectivity to an amide nitrogen
atom was deduced. H,H-COSY coupling of H4 and H7R/β
revealed the position of the NH function and thus established the
structure of 6 (Figure 2E). While compound 6 is obviously
derived from the gliotoxin precursor 2, its substitution pattern
indicates that it would rather be a shunt product than a substrate
for GliG. However, the finding of 6 points toward an oxygenated
12323
dx.doi.org/10.1021/ja201311d |J. Am. Chem. Soc. 2011, 133, 12322–12325

Journal of the American Chemical Society
COMMUNICATION
metabolite upstream of CÀS bond formation. Thus, we mon-
itored the metabolic profiles of the wild type and mutant by
LCÀHR-MS (Exactive) and detected [M + H]+ peaks at m/z
267.0970 (calcd for C12H15N2O5: 267.0981) that could well
correspond to a bishydroxylated diketopiperazine, 4 (Figure 1C,
traces a and b). This assumption was supported by the observed
MSn fragmentation pattern, which corresponds with the sequen-
tial loss of 2 equiv of water (see the Supporting Information).
Unfortunately, the scarcity and instability of 4 hampered its iso-
lation to clarify whether the hydroxyl groups are attached to the
amide nitrogen (4a) or the R-carbon (4b). To address this, we
prepared a synthetic reference of 4a and compared it with the
metabolite produced by the ΔgliG mutant. Interestingly, both the
HPLC retention times and MSn spectra fragmentation patterns
for these samples differed, indicating that the diketopiperazine is
bishydroxylated at the R-carbon (4b), not at the nitrogen. This
conclusion is also in line with the structure of the shunt product 6.
We next tested whether the bishydroxylated intermediate
(4b) is a key intermediate in the gliotoxin pathway. First, we
surmised that 4 would result from enzymatic bishydroxylation, a
reaction that could be catalyzed by the gliC gene product, a
putative cytochrome P450 monooxygenase. To verify this, we
generated a mutant lacking gliC and monitored its metabolite
profile. In fact, neither 6 nor 4 were detectable in the broth of the
ΔgliC mutant, unequivocally proving that GliC acts upstream of
GliG in the gliotoxin pathway (Figure 1C, trace c). Furthermore,
we found that the second putative cytochrome P450 monoox-
ygenase (GliF) encoded in the gli gene cluster is not involved in
the hydroxylation of the diketopiperazine, since deletion of gliF
did not affect the production of the bishydroxylated intermediate
(data not shown). In this way, it could be ruled out that GliF plays
a role upstream of CÀS bond formation. Finally, we found that
4b is indeed a precursor of gliotoxin, as it was consumed in the
GliG in vitro assay with formation of a novel sulfur-containing
adduct (5) (Figure 2D, trace a). Specifically, a freshly prepared
crude extract from the ΔgliG mutant was incubated with GSH
and GliG. Using LCÀHRMS, we monitored the conversion of
the bishydroxylated intermediate into a new compound, 5, whose
[M + H]+ peak appeared at m/z 845.2443 (calcd for C32H45-
N8O15S2: 845.2446). This was in perfect agreement with the
structure of a bis-GSH adduct, and the HR-MSn fragmentation
pattern confirmed the presence of two GSH units loaded onto
the diketopiperazine core of 5 (Figure 2E). In negative control
experiments without GliG or with heat-inactivated GliG, none or
only traces of the conjugate could be detected (Figure 2D, traces
b and c). The nonzero transformation of 4b into 5 could be due
to spontaneous dehydration and subsequent GSH addition,
which is known for strong electrophiles,16 but this does not take
place under physiological conditions in the ΔgliG mutant. In-
stead, we found that GliG is quite thermostable (see Figure 2C)
and that there is residual enzymatic activity after standard heat
treatment (for 10 min), with longer heating periods leading to a
completely inactive enzyme. Finally, in another control experi-
ment, we demonstrated that 5 was not formed using the crude
extract of the ΔgliC mutant broth (Figure 2D, trace d), thus
excluding the possibility that another precursor would undergo
GSH conjugation. In sum, these findings revealed that a hydro-
xylase and a new type of GST are required in order to produce an
unusual bis-GSH adduct. A mechanistically reasonable scenario
would be that the bishydroxylated intermediate undergoes
sequential elimination of water, thus yielding intermediary imine
species, which would represent suitable electrophiles for the
attacking thiolate (Figure 2F). The observed retention of con-
figuration in gliotoxin is remarkable and suggests that hydroxyla-
tion and the downstream elimination/GSH addition take place
from the same side of the molecule. On another note, it is
interesting that an oxygenation reaction mediated by a cyto-
chrome P450 enzyme (GliC) is the prerequisite of gliotoxin CÀS
bond formation, a scenario that is strikingly similar to what is
observed in phase I/II detoxification reactions. Finally, we reason
that the ultimate steps for formation of the epidithio bridge
involve the degradation of the bis-GSH conjugate by the putative
dipeptidase (GliJ) and an ACC-synthase-like enzyme (GliI) to
yield the dithiol, which is eventually converted by GliT into the
canonical ETP disulfide bridge.
The topic of enzymatic CÀS bond formation has been the
focus of various excellent recent studies, for example in the con-
text of HisÀCys cross-linking17 and lantibiotic cyclization.18 It
has been shown that enzymatic sulfurization involves γ-gluta-
mylcysteine in ergothioneine biosynthesis19 and cysteine in
ovothiol biosynthesis.20 GSH has been identified as the sulfur
donor in the biosynthesis of glucosinolates,21 yet a dedicated
GST for its incorporation has not been characterized to date.
Also, except for their contribution to double-bond isomerization
in hypothemycin biosynthesis,22 to the best of our knowledge,
GSTs have not been implicated to date in microbial pathways
yielding a sulfur-containing metabolite. Through the targeted
knockouts of gliG and gliC, metabolic profiling, and in vitro
reconstitution of GST activity, we have now provided the first
insight into the sulfurization steps in gliotoxin biosynthesis. These
results are likely to be significant for a broad range of micro-
organisms, as our phylogenetic analysis revealed that genes
coding for GliG homologues are widespread in the genomes of
ETP producers. GliG is related to theta-class GSTs but appears
to be the first representative of a new family of biosynthetic
enzymes. Thus, our work has not only unveiled key steps in the
pathway of a virulence factor of a severe human pathogen but also
outlined a new role of a microbial GST. Future studies will shed
light on the structure and exact mechanism of GliG.
’ ASSOCIATED CONTENT
S
Supporting Information. Detailed phylogenetic tree, se-
b
quence alignment, HR-MSn data for 4 and 5, full spectra for 6 and
synthetic compounds, HPLC comparison of synthetic 4a with
4b, and full experimental details. This material is available free of
charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
Christian.Hertweck@hki-jena.de
’ ACKNOWLEDGMENT
We thank A. Perner for MS measurements, M. Steinacker and
M. Cyrulies for fermentation, Dr. E. Shelest for support in phylo-
genetic analyses, and S. Fricke and C. Schult for technical assistance.
Financial support by the DFG is gratefully acknowledged.
’ REFERENCES
(1) Kwon-Chung, K. J.; Sugui, J. A. Med. Mycol. 2009, 47, S97–S103.
(2) Gardiner, D. M.; Waring, P.; Howlett, B. J. Microbiology 2005,
151, 1021–1032.
12324
dx.doi.org/10.1021/ja201311d |J. Am. Chem. Soc. 2011, 133, 12322–12325

Journal of the American Chemical Society
COMMUNICATION
(3) Chai, C. L. L.; Waring, P. Redox Rep. 2000, 5, 257–264.
(4) Moebius, N.; Hertweck, C. Curr. Opin. Plant Biol. 2009, 12,
390–398.
(5) Brakhage, A. A.; Langfelder, K. Annu. Rev. Microbiol. 2002, 56,
433–455.
(6) Dagenais, T. R. T.; Keller, N. P. Clin. Microbiol. Rev. 2009, 22,
447–465.
(7) Cramer, R. A.; Gamcsik, M. P.; Brooking, R. M.; Najvar, L. K.;
Kirkpatrick, W. R.; Patterson, T. F.; Balibar, C. J.; Graybill, J. R.; Perfect,
J. R.; Abraham, S. N.; Steinbach, W. J. Eukaryotic Cell 2006, 5, 972–980.
(8) Kupfahl, C.; Heinekamp, T.; Geginat, G.; Ruppert, T.; Hartl, A.;
Hof, H.; Brakhage, A. A. Mol. Microbiol. 2006, 62, 292–302.
(9) Sugui, J. A.; Pardo, J.; Chang, Y. C.; Zarember, K. A.; Nardone,
G.; Galvez, E. M.; Muellbacher, A.; Gallin, J. I.; Simon, M. M.;
Kwon-Chung, K. J. Eukaryotic Cell 2007, 9, 1562–1569.
(10) Spikes, S.; Xu, R.; Nguyen, C. K.; Chamilos, G.; Kontoyiannis,
D. P.; Jacobson, R. H.; Ejzykowicz, D. E.; Chiang, L. Y.; Filler, S. G.; May,
G. S. J. Infect. Dis. 2008, 197, 479–486.
(11) Balibar, C. J.; Walsh, C. T. Biochemistry 2006, 45, 15029–15038.
(12) Scharf, D. H.; Remme, N.; Heinekamp, T.; Hortschansky, P.;
Brakhage, A. A.; Hertweck, C. J. Am. Chem. Soc. 2010, 132, 10136–10141.
(13) Hayes, J. D.; Flanagan, J. U.; Jowsey, I. R. Annu. Rev. Pharmacol.
Toxicol. 2005, 45, 51–88.
(14) Sato, I.; Shimizu, M.; Hoshino, T.; Takaya, N. J. Biol. Chem.
2009, 284, 8042–8053.
(15) Fox, E. M.; Howlett, B. J. Mycol. Res. 2008, 112, 162–169.
(16) Lo, W. J.; Chiou, Y. C.; Hsu, Y. T.; Lam, W. S.; Chang, M. Y.;
Jao, S. C.; Li, W. S. Bioconjugate Chem. 2007, 18, 109–120.
(17) Fujieda, N.; Ikeda, T.; Murata, M.; Yanagisawa, S.; Aono, S.;
Ohkubo, K.; Nagao, S.; Ogura, T.; Hirota, S.; Fukuzumi, S.; Nakamura,
Y.; Hata, Y.; Itoh, S. J. Am. Chem. Soc. 2011, 133, 1180–1183.
(18) Li, B.; Yu, J. P. J.; Brunzelle, J. S.; Moll, G. N.; van der Donk,
W. A.; Nair, S. K. Science 2006, 311, 1464–1467.
(19) Seebeck, F. P. J. Am. Chem. Soc. 2010, 132, 6632–6633.
(20) Braunshausen, A.; Seebeck, F. P. J. Am. Chem. Soc. 2011, 133,
1757–1759.
(21) Geu-Flores, F.; Nielsen, M. T.; Nafisi, M.; Møldrup, M. E.;
Olsen, C. E.; Motawia, M. S.; Halkier, B. A. Nat. Chem. Biol. 2009,
5, 575–577.
(22) Reeves, C. D.; Hu, Z.; Reid, R.; Kealey, J. T. Appl. Environ.
Microbiol. 2008, 74, 5121–5129.
12325
dx.doi.org/10.1021/ja201311d |J. Am. Chem. Soc. 2011, 133, 12322–12325