Epidithiol formation by an unprecedented twin carbon-sulfur lyase in the gliotoxin pathway

Article abstract of DOI:10.1002/anie.201205041

Two in one go: The elucidation of a key step in the biosynthesis of gliotoxin, the infamous virulence factor of the human pathogen Aspergillus fumigatus, provides insight into the formation of an epidithiol. Isolation of a bis(cysteine) S-conjugate from a ΔgliI mutant and in vitro studies show that GliI concomitantly cleaves two C-S bonds, along with the formation of ammonia and pyruvate (see scheme). Copyright

Full text of DOI:10.1002/anie.201205041

Angewandte
Chemie
DOI: 10.1002/anie.201205041
Natural Sulfur Compounds
Epidithiol Formation by an Unprecedented Twin Carbon–Sulfur Lyase
in the Gliotoxin Pathway**
Daniel H. Scharf, Pranatchareeya Chankhamjon, Kirstin Scherlach, Thorsten Heinekamp,
Martin Roth, Axel A. Brakhage, and Christian Hertweck*
Sulfur, the tenth most abundant element in the universe, has
been a spearhead in the development of life on earth. It was
not only present in the first organic compounds that
originated from hydrogen sulfide,[1] but, according to the
autotrophic theory of the origin of life, sulfur was also crucial
for mediating amino acid, peptide, and RNA synthesis.[2,3]
Whereas cysteine- and methionine-derived thiols and thio-
ethers are essential components of proteins and peptides,[4–6]
other sulfur-containing natural products are much less
abundant and only little is known about their biosynthesis.
Even so, when present, sulfur is often the key to biological
activity of secondary metabolites. Important examples are the
highly reactive glucosinolate-derived isothiocyanates[7,8] and
polysulfides that act as plant defense agents,[9,10] thiazole-
bearing cytotoxic compounds,[11] and polythioamide pharma-
cophores in antibacterial Clostridium metabolites.[12,13]
A
structurally and functionally outstanding sulfur-containing
moiety is the transannular disulfide bridge in epidithiodi-
ketopiperazines (ETP), a family of microbial toxins.[14] The
prototype of these compounds, gliotoxin (1; Scheme 1), is an
infamous virulence factor of the human pathogenic fungus
Aspergillus fumigatus, which is notable as a leading cause of
death in immunocompromized patients.[15–17] Endowed with
the reactive epidithio moiety, the toxin exerts deleterious
effects in host cells through redox-cycling and inactivation of
vital proteins by conjugation.[17–19] The cell-damaging impact
of dithiol species 2 is illustrated by the mandatory presence of
a disulfide forming oxidoreductase, GliT, to provide the
Scheme 1. Structure of gliotoxin (1), a virulence factor of the human
pathogen Aspergillus fumigatus, and its biosynthesis starting from
diketopiperazine 3.
gliotoxin producer with self-resistance.[20] A number of
studies showed that the diketopiperazine core (3) of gliotoxin
is assembled by a non-ribosomal peptide synthetase.[21–24]
However, downstream pathway steps have remained elusive,
mainly because of the scarcity and instability of pathway
intermediates produced.[19] However, several gliotoxin con-
geners or shunt products could be detected in crude
extracts.[25] Recently, we demonstrated by in vivo and in vitro
studies that the diketopiperazine is activated by an oxygenase,
GliC, to allow glutathione transfer catalyzed by a dedicated
glutathione S-transferase, GliG, yielding bis-adduct 4.[26] The
function of GliG has also been studied by another group.[27]
Herein, we elucidate the key step of epidithiol formation in
gliotoxin biosynthesis. We unveil the structure of an unusual
bis(cysteinyl) adduct and demonstrate that a specialized C–S
lyase, GliI, promotes a dual carbon–sulfur cleavage to
generate the notorious epidithiol moiety of the toxin.
Our previous discovery of the bis(glutathione) adduct 4
and deduced gene functions were highly suggestive of
a scenario involving hydrolytic degradation of the peptide
side chains of 4 followed by carbon–sulfur cleavage. A
bioinformatic analysis of the gliotoxin biosynthesis gene
cluster revealed a candidate gene, gliI, that could potentially
code for an enzyme catalyzing the latter step. According to
BLAST analyses, the deduced gene product of gliI was
annotated as a 1-aminocyclopropane-1-carboxylic acid syn-
[*] D. H. Scharf,[+] P. Chankhamjon,[+] Dr. K. Scherlach,
Dr. T. Heinekamp, Dr. M. Roth, Prof. Dr. A. A. Brakhage,
Prof. Dr. C. Hertweck
Depts. of Biomolecular Chemistry, Molecular and Applied Micro-
biology, and Bio Pilot Plant, Leibniz Institute for Natural Product
Research and Infection Biology, HKI
Beutenbergstrasse 11a, 07745 Jena (Germany)
E-mail: christian.hertweck@hki-jena.de
Prof. Dr. A. A. Brakhage, Prof. Dr. C. Hertweck
Friedrich Schiller University, Jena (Germany)
[+] These authors contributed equally to this work.
[**] We thank A. Perner for MS measurements, S. Fricke and C. Schult
for technical assistance, P. Hortschansky for artwork, and C. Heiden
and P. Berthel for performing fermentations. This work was
supported by the “Pakt fꢀr Forschung und Innovation” of the Free
State of Thuringia and the BMBF, and the International Leibniz
Research School for Biomolecular and Microbial Interactions
(ILRS), as part of the excellence graduate school Jena School for
Microbial Communication (JSMC).
Supporting information for this article, including experimental
details, is available on the WWW under http://dx.doi.org/10.1002/
anie.201205041.
Angew. Chem. Int. Ed. 2012, 51, 1 – 6
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

.
Angewandte
Communications
thase. To elucidate the role of the putative enzyme we aimed
at constructing a gliI deletion mutant (DgliI) of A. fumigatus.
We succeeded in generating the targeted knockout by using
a split-fragment PCR-based strategy. By this reaction, over-
lapping ends to the pyrithiamine resistance cassette were
introduced at the 3’ end of the upstream flanking region and
at the 5’ end of the downstream flanking region of the gliI
gene, thus warranting the identity of the desired mutant
(Figure 1A). HPLC-MS monitoring of the DgliI mutant
culture revealed that gliotoxin biosynthesis was completely
abrogated (Figure 1B). We carefully examined mutant broth
and mycelium by HPLC-HRMS and detected trace amounts
DEPT135 spectra of 5 showed signals for two amide carbons,
an aromatic ring system, two carboxyl functions, four
methylene carbon atoms, and two methine carbon atoms.
The chemical shifts of 28.2 and 28.6 ppm for C1’’ and C1’,
respectively, indicated the proximity of two methylene
functions to the sulfur atoms. This was corroborated by
marked HMBC couplings of H1’’ to C2 and C3’’, and H1’ to
C5 and C3’. Moreover, we observed correlations of H8 to C9,
C10, and C14 as well as to the amide carbon C6 (d =
165.6 ppm), thus establishing the assignment of the phenyl-
alanine partial structure. The hydroxymethyl protons (H7)
were seen to have HMBC correlations to the amide carbon
C3 and the quaternary carbon C2. Furthermore, the structure
of the unprecedented bis(cysteinyl) diketopiperazine 5 was
fully supported by high-resolution MS analyses of the
daughter ions generated by tandem mass spectrometry
employing an Exactive (Orbitrap) mass spectrometer. The
molecular composition of each fragment ion is in full agree-
ment with the proposed structure (Supporting Information,
Figure S8).
To clarify the biochemical function of GliI in vitro, we
PCR-amplified gliI, cloned the amplicon into an E. coli
expression vector, and introduced the construct into E. coli
BL21(DE3) cells for protein overproduction. MBP-tagged
GliI, which had been harvested from the biomass of an E. coli
culture (1 L), was purified using a dextrin column and treated
with TEV protease for tag removal (Figure 2A). Static light
scattering experiments revealed the homodimeric status of
GliI. The theoretical molecular mass of a GliI dimer is
97 kDa, which is in excellent accordance with the measured
mass of 96.3 Æ 2 kDa (Figure 2C). To identify the cofactor of
the putative C–S lyase we searched for consensus motifs in the
amino acid alignment and identified a conserved pyridoxal
5’-phosphate (PLP) binding domain within the GliI amino
acid sequence. This was supported by spectrophotometric
characterization, which indicated a 1:1 stoichiometry for GliI–
PLP (Figure 2B). To elucidate the fate of the bis(cysteinyl)
adduct in the presence of GliI, a solution of 5 (5 mm) was
incubated with GliI-PLP and heat-inactivated GliI-PLP as
a negative control. The course of the reaction was monitored
by LC-HRMS. Whereas no biotransformation of 5 in the
presence of inactivated GliI could be detected (Figure 2D,
trace a), we observed that active Gli-PLP readily converted 5
into two new products, 6 and 7 (Figure 2D, trace b), with m/z
[M+H]À 297.0379 (calcd. for C12H13N2O3S2 297.0373) and m/z
295.0220 [M+H]À (calcd. for C12H11N2O3S2 295.0217), respec-
tively. Through analysis of an oxidation reaction, we deter-
mined that disulfide 7 is the oxidation product of the dithiol 6,
which explains the formation of 7 under the aerobic
conditions of the assay. We could unequivocally assign the
structure of 7 by HPLC-MS comparison with a fully charac-
terized reference compound (Figure S10).
Figure 1. Genotype and phenotype of the DgliI mutant. A) Southern
blot showing successful gliI gene deletion in the genome of A.
fumigatus; D=DgliI mutant, wt=wild type. B) HPLC-MS profiles (SIM
mode) of the wild type and the DgliI mutant. C) Structure of
bis(cysteinyl) adduct 5 and key HMBC correlations. The absolute
configuration was inferred from 1.
of a new compound (5) with a molecular weight of m/z 473
[M+H]+. A molecular formula of C18H24N4O7S2 was deduced
from HRESI-MS data, which is in full agreement with the
proposed composition of the tentative pathway intermediate
resulting from glutathione side chain cleavage of 4. MS/MS
fragmentation patterns and HRMS of daughter ions provided
further support for the identity of 5. To unequivocally confirm
the structure of the bis adduct, a full characterization of 5 was
desired. However, this proved to be a major challenge
because of the extremely low amounts produced and the
instability of the metabolite. Furthermore, we noted that 5 is
highly water soluble and mainly retained in the fungal cytosol,
which prevented its extraction from the fermentation broth.
To obtain sufficient material of 5 we pooled, sheared, and
extracted mycelia from a total of 140 L of mutant cultures.
Through size exclusion chromatography and repeated prep-
arative LC-MS, 1.49 mg of 5 could be isolated. 13C NMR and
The finding that dithiol 6 is the immediate product of the
GliI-mediated biotransformation of 5 is intriguing as it
À
implies that both cysteinyl C S bonds were cleaved enzymati-
cally. A plausible reaction mechanism would involve a
b-elimination to liberate the thiol with concomitant formation
of an imine that would hydrolyze to yield pyruvate and
ammonia. To verify the postulated release of NH3 in the GliI-
2
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 1 – 6
These are not the final page numbers!

Angewandte
Chemie
ammonia. In analogy to other PLP-dependent reactions the
GliI-mediated transformation can be dissected into three
stages: 1) transaldimination, 2) a-proton abstraction (both
promoted by the w-amino group of a Lys residue) and 3) b-
elimination, followed by hydrolysis (Scheme 2). As no
monocysteine intermediate was detected in the enzyme
assay, and considering that GliI is a homodimer, one may
conceive of a simultaneous C–S cleavage in both substrate
halves.
Figure 2. Characterization of GliI and results from in vitro assays.
A) Purity of GliI analyzed by SDS-PAGE stained with Coomassie
Brilliant Blue R-250. B) UV/Vis absorbance spectra of native (c) and
denatured (c; NaOH, 958) PLP-saturated GliI. C) Native absolute
molecular mass determination by static light scattering; UV (c;
ultraviolet light), LS (c; light scattering), MW (c; molecular
weight). D) HPLC-MS profiles (SIM mode) of 5 (c; m/z=473), 6
(c; m/z=297), and 7 (c; m/z=295) of biotransformation experi-
ments using a) heat-inactivated GliI (negative control) and b) active
GliI-PLP. Structures of reaction products 6 and 7 (reduced and
oxidized forms, respectively) E) Detection of ammonia using Nessler’s
reagent. F) HPLC-MS analysis of pyruvate-derived hydrazone 8 (assay
treated with 4-fluorophenylhydrazine); a) negative control, b) assay
using active GliI-PLP, and c) synthetic reference.
Scheme 2. Model for the pyridoxal-dependent b-elimination reaction
catalyzed by GliI.
From an evolutionary point of view it is remarkable that
cysteine conjugate b-lyases have only been known from
detoxification reactions and primary metabolism.[28] Whereas
related C–S lyase activities have been implicated in a number
of secondary metabolite pathways, so far only indirect
evidence for their involvement has been given. In glucosino-
late biosynthesis, a C–S lyase (Sur1) has been tentatively
assigned to the pathway, but its activity has only been probed
using an artificial surrogate substrate.[29] Another C–S lyase
homologue, EtgE, seems to play a role in the biosynthesis of
ergothioneine. However, the insolubility of the heterolo-
gously produced enzyme prevented the collection of firm
biochemical evidence.[30] Thus, GliI represents the first func-
tionally proven C–S lyase involved in a secondary metabolite
pathway. Furthermore, this is the first twin C–S cleavage to
yield an epidithiol. Another interesting observation is the
relaxed substrate specificity of GliI. The enzyme mediates
both C–S disconnections in 5, irrespective of the different
substitution pattern and configuration of the diketopiperazine
substructures. To shed light on the phylogenetic relationship
catalyzed reaction we utilized Nesslerꢀs reagent (K2HgI4 in
a KOH solution). Indeed, in the positive enzyme assay the
occurrence of the typical brown-orange color verified the
formation of ammonia (Figure 2E). To prove the formation
of pyruvate as a side product we trapped the predicted oxo
compound by addition of 4-fluorophenylhydrazine (4-FPH)
to the enzyme assay. By HRESI-MS we detected a peak
corresponding to the expected hydrazine
8 with m/z
[M+H]À = 195.0569 (calcd. for C9H8N2O2F 195.0575). The
identity of 8 was further corroborated through comparison
with a synthetic reference (Figure 2F). These results unequiv-
ocally show that GliI functions as a C–S lyase to generate
a dithiol with concomitant liberation of pyruvate and
Angew. Chem. Int. Ed. 2012, 51, 1 – 6
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

.
Angewandte
Communications
of this unusual enzyme, we constructed a neighbor-joining
tree based on the amino acid sequences of GliI and related
enzymes (Figure 3). To our surprise, GliI and yet uncharac-
a dual C–S cleavage of 5, yielding the corresponding
epidithiol. Moreover, through a series of further analyses
we identified pyruvate and ammonia as side products of the
lyase reaction. To the best of our knowledge, this is the first
complete characterization of a thiol-forming C–S lyase
reaction in a secondary metabolic pathway. Furthermore,
the GliI-catalyzed reaction is unparalleled in forming two
thiol groups concomitantly, possibly by the dual action of the
GliI homodimer. A phylogenetic analysis granted first
insights into the evolution of this unusual enzyme and
revealed that GliI is the prototype of a new subfamily of
C–S lyases that are encoded in ETP biosynthesis clusters.
Thus, our findings not only hold lessons for a deeper under-
standing of gliotoxin biosynthesis, but will also facilitate the
elucidation of many related ETP pathways. From a broader
perspective the GliI-mediated C–S lyase reaction sheds new
light on the formation of organosulfur compounds that have
evolved as highly reactive biomolecules.
Received: June 27, 2012
Published online: && &&, &&&&
Keywords: Aspergillus fumigatus · gliotoxin · lyases · mycotoxins ·
.
organosulfur compounds
[1] E. T. Parker, H. J. Cleaves, M. P. Callahan, J. P. Dworkin, D. P.
Glavin, A. Lazcano, J. L. Bada, Orig Life Evol. Biosph. 2011, 41,
569 – 574.
[2] L. Leman, L. Orgel, M. R. Ghadiri, Science 2004, 306, 283 – 286.
[3] A. D. Keefe, S. L. Miller, G. McDonald, J. Bada, Proc. Natl.
Acad. Sci. USA 1995, 92, 11904 – 11906.
[4] G. Bierbaum, H. G. Sahl, Curr. Pharm. Biotechnol. 2009, 10, 2 –
18.
Figure 3. Phylogenetic analysis of GliI and related enzymes. Clade I:
cysteine-conjugate lyases from different organisms. Clade II: alliinases
from plants. Clade III: 1-aminocyclopropane-1-carboxylate synthases
from plants. Clade IV: GliI homologues encoded in other fungal ETP
clusters (for full details, see the Supporting Information).
[5] J. O. Melby, N. J. Nard, D. A. Mitchell, Curr. Opin. Chem. Biol.
2011, 15, 369 – 378.
[6] J. M. Willey, W. A. van der Donk, Annu. Rev. Microbiol. 2007,
61, 477 – 501.
[7] B. A. Halkier, J. Gershenzon, Annu. Rev. Plant Biol. 2006, 57,
303 – 333.
[8] I. E. Sønderby, F. Geu-Flores, B. A. Halkier, Trends Plant Sci.
2010, 15, 283 – 290.
[9] U. Mꢁnchberg, A. Anwar, S. Mecklenburg, C. Jacob, Org.
Biomol. Chem. 2007, 5, 1505 – 1518.
[10] I. D. Nwachukwu, A. J. Slusarenko, M. C. Gruhlke, Nat. Prod.
Commun. 2012, 7, 395 – 400.
[11] Z. Jin, Nat. Prod. Rep. 2011, 28, 1143 – 1191.
[12] S. Behnken, T. Lincke, F. Kloss, K. Ishida, C. Hertweck, Angew.
Chem. 2012, 124, 2475 – 2478; Angew. Chem. Int. Ed. 2012, 51,
2425 – 2428.
[13] T. Lincke, S. Behnken, K. Ishida, M. Roth, C. Hertweck, Angew.
Chem. 2010, 122, 2055 – 2057; Angew. Chem. Int. Ed. 2010, 49,
2011 – 2013.
[14] D. M. Gardiner, P. Waring, B. J. Howlett, Microbiology 2005,
151, 1021 – 1032.
[15] D. H. Scharf, T. Heinekamp, N. Remme, P. Hortschansky, A. A.
Brakhage, C. Hertweck, Appl. Microbiol. Biotechnol. 2012, 93,
467 – 472.
[16] T. R. T. Dagenais, N. P. Keller, Clin. Microbiol. Rev. 2009, 22,
447 – 465.
[17] K. J. Kwon-Chung, J. A. Sugui, Med. Mycol. 2009, 47, S97 – S103.
[18] N. Mçbius, C. Hertweck, Curr. Opin. Plant Biol. 2009, 12, 390 –
398.
terized homologues encoded in various other fungal ETP
biosynthesis gene clusters form a new clade (IV) that is clearly
distinct from related enzymes. In particular, lyases that are
involved in primary metabolism and xenobiotic detoxification
in bacteria, fungi, plants, and humans fall into one clade (I)
that is only distantly related. Furthermore, it is particularly
noteworthy that the GliI clade (IV) is more closely related to
alliinases, for example, in garlic (clade II), and 1-amino-
cyclopropane-1-carboxylate synthases from plants (clade III).
It appears that the C–S lyase activity of GliI and its
homologues has evolved independently in secondary metab-
olism. Finally, on biotechnological grounds, it will be inter-
esting to study the specificities of the yet uncharacterized C–S
lyases that appear to be functionally related to GliI (clade
IV).
In summary, we have elucidated a key step in the
biosynthesis of gliotoxin, an infamous virulence factor of
the human pathogen A. fumigatus. From the large-scale
fermentation of an engineered DgliI mutant we isolated an
unparalleled bis(cysteine) conjugate 5, a missing link between
the diketopiperazine precursor and the mycotoxin. We also
succeeded in producing active GliI and verified through an
in vitro assay that the PLP-dependent enzyme catalyzes
4
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 1 – 6
These are not the final page numbers!

Angewandte
Chemie
[19] D. H. Scharf, T. Heinekamp, N. Remme, P. Hortschansky, A. A.
Brakhage, C. Hertweck, Appl. Microbiol. Biotechnol. 2012, 93,
467 – 472.
[20] D. H. Scharf, N. Remme, T. Heinekamp, P. Hortschansky, A. A.
Brakhage, C. Hertweck, J. Am. Chem. Soc. 2010, 132, 10136 –
10141.
[21] C. J. Balibar, C. T. Walsh, Biochemistry 2006, 45, 15029 – 15038.
[22] R. A. Cramer, M. P. Gamcsik, R. M. Brooking, L. K. Najvar,
W. R. Kirkpatrick, T. F. Patterson, C. J. Balibar, J. R. Graybill,
J. R. Perfect, S. N. Abraham, W. J. Steinbach, Eukaryotic Cell
2006, 5, 972 – 980.
[25] R. R. Forseth, E. M. Fox, D. Chung, B. J. Howlett, N. P. Keller,
F. C. Schroeder, J. Am. Chem. Soc. 2011, 133, 9678 – 9681.
[26] D. H. Scharf, N. Remme, A. Habel, P. Chankhamjon, K.
Scherlach, T. Heinekamp, P. Hortschansky, A. A. Brakhage, C.
Hertweck, J. Am. Chem. Soc. 2011, 133, 12322 – 12325.
[27] C. Davis, S. Carberry, M. Schrettl, I. Singh, J. C. Stephens, S. M.
Barry, K. Kavanagh, G. L. Challis, D. Brougham, S. Doyle,
Chem. Biol. 2011, 18, 542 – 552.
[28] A. J. Cooper, B. F. Krasnikov, Z. V. Niatsetskaya, J. T. Pinto, P. S.
Callery, M. T. Villar, A. Artigues, S. A. Bruschi, Amino Acids
2011, 41, 7 – 27.
[23] C. Kupfahl, T. Heinekamp, G. Geginat, T. Ruppert, A. Hartl, H.
Hof, A. A. Brakhage, Mol. Microbiol. 2006, 62, 292 – 302.
[24] J. A. Sugui, J. Pardo, Y. C. Chang, K. A. Zarember, G. Nardone,
E. M. Galvez, A. Muellbacher, J. I. Gallin, M. M. Simon, K. J.
Kwon-Chung, Eukaryotic Cell 2007, 9, 1562 – 1569.
[29] M. D. Mikkelsen, P. Naur, B. A. Halkier, Plant J. 2004, 37, 770 –
777.
[30] F. P. Seebeck, J. Am. Chem. Soc. 2010, 132, 6632 – 6633.
Angew. Chem. Int. Ed. 2012, 51, 1 – 6
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!

.
Angewandte
Communications
Communications
Natural Sulfur Compounds
D. H. Scharf, P. Chankhamjon,
K. Scherlach, T. Heinekamp, M. Roth,
A. A. Brakhage,
C. Hertweck*
&&&& — &&&&
Epidithiol Formation by an
Unprecedented Twin Carbon–Sulfur
Lyase in the Gliotoxin Pathway
Two in one go: The elucidation of a key
step in the biosynthesis of gliotoxin, the
infamous virulence factor of the human
pathogen Aspergillus fumigatus, provides
insight into the formation of an epidithiol.
Isolation of a bis(cysteine) S-conjugate
from a DgliI mutant and in vitro studies
show that GliI concomitantly cleaves two
À
C S bonds, along with the formation of
ammonia and pyruvate.
6
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 1 – 6
These are not the final page numbers!