Early Oxidative Transformations During the Biosynthesis of Terrein and Related Natural Products

Article abstract of DOI:10.1002/chem.202101447

The mycotoxin terrein is derived from the C₁₀-precursor 6-hydroxymellein (6-HM) via an oxidative ring contraction. Although the corresponding biosynthetic gene cluster (BGC) has been identified, details of the enzymatic oxidative transformations are lacking. Combining heterologous expression and in vitro studies we show that the flavin-dependent monooxygenase (FMO) TerC catalyzes the initial oxidative decarboxylation of 6-HM. The reactive intermediate is further hydroxylated by the second FMO TerD to yield a highly oxygenated aromatic species, but further reconstitution of the pathway was hampered. A related BGC was identified in the marine-derived Roussoella sp. DLM33 and confirmed by heterologous expression. These studies demonstrate that the biosynthetic pathways of terrein and related (polychlorinated) congeners diverge after oxidative decarboxylation of the lactone precursor that is catalyzed by a conserved FMO and further indicate that early dehydration of the side chain is an essential step.

Full text of DOI:10.1002/chem.202101447

Accepted Article
Accepted Article
Title: Early Oxidative Transformations During the Biosynthesis of
Terrein and Related Natural Products
Authors: Lukas Kahlert, Darlon Bernardi, Maurice Hauser, Laura
P. Ioca, Roberto G. S. Berlinck, Elizabeth J. Skellam, and
Russell John Cox
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.202101447
Link to VoR: https://doi.org/10.1002/chem.202101447
01/2020

10.1002/chem.202101447
Chemistry - A European Journal
FULL PAPER
Early Oxidative Transformations During the Biosynthesis of
Terrein and Related Natural Products
Lukas Kahlert,[a] Darlon Bernardi,[a,b] Maurice Hauser,[a] Laura P. Ióca,[b] Roberto G. S. Berlinck,[b]
Elizabeth J. Skellam[a,c] and Russell J. Cox*[a]
[a]
Mr L. Kahlert, Mr D. Bernardi, Mr M. Hauser, Dr E. J. Skellam and Prof R. J. Cox,
Institute for Organic Chemistry and BMWZ
Leibniz Universität Hannover
Schneiderberg 38, 30167, Hannover, Germany
E-mail: russell.cox@oci.uni-hannover.de; @RussellCox_Chem
Mr D. Bernardi, Dr Laura P. Ióca and Prof R. G. S. Berlinck
Instituto de Química de São Carlos,
[b]
Universidade de São Paulo,
CP 780, CEP 13560-970, São Carlos, SP, Brazil
Dr E. J. Skellam
[c]
Current Address: Department of Chemistry & BioDiscovery Institute
University of North Texas, 1155 Union Circle 305220
Denton, Texas, 76203, USA
Supporting information for this article is given via a link at the end of the document.
Abstract: The mycotoxin terrein is derived from the C10-precursor 6-
hydroxymellein (6-HM) via an oxidative ring contraction. Although the
corresponding biosynthetic gene cluster (BGC) has been identified,
details of the enzymatic oxidative transformations are lacking.
Combining heterologous expression and in vitro studies we show that
the flavin-dependent monooxygenase (FMO) TerC catalyzes the
initial oxidative decarboxylation of 6-HM. The reactive intermediate is
further hydroxylated by the second FMO TerD to yield a highly
oxygenated aromatic species, but further reconstitution of the
pathway was hampered. A related BGC was identified in the marine-
derived Roussoella sp. DLM33 and confirmed by heterologous
expression. Our studies demonstrate that the biosynthetic pathways
of terrein and related (polychlorinated) congeners diverge after
oxidative decarboxylation of the lactone precursor that is catalyzed by
a conserved FMO and further indicate that early dehydration of the
side chain is an essential step.
could be isolated from individual knockout strains.[3] The group’s
alternative attempt to reconstitute the biosynthetic pathway by
heterologous expression in A. niger also met with failure as native
enzymes were thought to interfere with formation of 2.
Structurally related (poly)chlorinated polyketides include
the families of cyclohelminthols
velutinum),[11]
palmaenones
3
(Helminthosporium
4
(Lachnum
palmae),[12]
cryptosporiopsin 5 and cryptosporiopsinol 6 (Cryptosporiopsis
sp., Periconia macrospinosa and Phialophora asteris f. sp.
Helianthi)[13–15] as well as the putative Diels-Alder adduct of 5,
roussoellatide 7 (Roussoella sp. DLM33, Scheme 1).[16] These
compounds are expected to utilize chlorinated congeners of 2,
such as the double chlorinated 8, as biosynthetic precursors. All
respective BGC share genes homologous to terABC and terR, but
no homologs of terDEF of the terrein BGC are present in any of
these systems (Scheme 1B).[17]
Remarkably, labelling studies showed that the cyclopent-
2-en-1-one moieties of terrein 1 and cryptosporiopsin 5 display
opposite isotopic incorporation at the α-hydroxyl carbon indicating
a different course of the crucial ring contraction in each case
(Scheme 1A).[16,18,19] It therefore appears that the biosynthetic
pathways of 1 and its chlorinated congeners diverge at some point
in the pathway, possibly after the action of TerC which is
annotated as a flavin dependent monooxygenase (FMO). TerD
and TerE are also likely oxidative enzymes (FMO and Cu-
dependent oxygenase, respectively), while TerF (a so-called
Kelch protein) is of unknown function. The role of each enzyme
beyond the formation of 2 is therefore speculative or unknown.
We,[20,21] and others,[22] have shown that reconstruction of
cryptic fungal biosynthetic pathways in the host organism
Aspergillus oryzae NSAR1 is an effective way to understand
complex transformations. Here, we deploy those methods and in
vitro studies of isolated enzymes aiming to overcome the hurdles
that have previously hampered characterization of pathway
intermediates during the biosynthesis of these intriguing
metabolites.[3]
Introduction
The mycotoxin terrein 1, first isolated in 1935 from the
ascomycete Aspergillus terreus,[1] and related fungi such as
Aspergillus lentulus,[2] has been the focus of many studies due to
its diverse biological activities. These include phytotoxic[3] and
anti-inflammatory properties,[4] inhibition of tumour angiogenesis[5]
and melanogenesis[6] as well as inhibition of biofilm formation in
Pseudomonas aeruginosa.[7] Isotope labelling[8,9] and genetic
experiments[3] revealed that the oxygenated cyclopent-2-en-1-
one moiety, present in 1, is derived via ring contraction of the
precursor R-6-hydroxymellein 2 (6-HM, Scheme 1A). Although
the terrein biosynthetic gene cluster (BGC) was discovered in
2014, and the biosynthesis of 2 by collaboration of the non-
reducing PKS (nr-PKS) TerA and the PKS-like multidomain
protein TerB has been investigated in detail,[10] the later steps
remain obscure.
By gene-knockout experiments in A. terreus Brock et al.
identified that the late-stage genes terCDEF (and terR encoding
a transcriptional activator) are essential for production of terrein 1
in vivo. However, no pathway intermediates following 6-HM 2
1
This article is protected by copyright. All rights reserved.

10.1002/chem.202101447
Chemistry - A European Journal
FULL PAPER
TerB produce 6-HM 2[25] that is the putative substrate for TerC.[3,10]
Expression of the truncated terCt with terAB led to no change over
expression of terAB alone. However, transformants coexpressing
the extended sequence of terC with terAB exhibited a distinct red
colouration of the fermentation media that intensified with
prolonged incubation (Figure 1), similar to that of ∆terD-∆terF
strains of A. terreus SBUG844.[3] In addition to TerA-derived
polyketide products 9-11 and the tetraketide pyrone 12, LCMS
analysis revealed that 2 was present in much lower quantities
compared to transformants solely expressing terAB (Figures 1B
and 1C).
We initially identified two new peaks during early stages
of cultivation of terABC transformants corresponding to the very
polar compounds 13 (C9H12O4; ([M+H]+ calculated 185.0814,
found 185.0802) and 14 (C9H10O4). Based on their molecular
weights, both 13 and 14 could be possible pathway intermediates.
Compound 13 was purified to homogeneity (~12 mg from 1 L
fungal culture) and its structure was revealed by full NMR analysis
(Figures S34-S38). The R-configuration of 13 was assumed from
the known configurations of the precursor 6-HM 2 and the off-
pathway lactone 12. Compound 14 proved even more unstable
than 13 and could not be purified. Based on its mass, retention
time and distinctive uv spectrum we propose 14 to be a quinone
derived from 13.
Interestingly, 13 is not red and no obvious new peaks in
the respective LCMS chromatograms of terABC-transformants
were observed concomitantly with 13. We therefore increased the
cultivation time to five days and separated the organic extract into
20 fractions by preparative HPLC. Although almost all fractions
displayed a light red-brownish hue, three fractions in particular
exhibited a deep red colouration. Two isomers 15a and 15a* were
identified in this extract that show the same UV-absorption and
molecular weight (C18H16O7; ([M+H]+ calculated 345.0974, found
345.0977).
A
third peak 15b (C18H18O8; [M-H]- calculated
361.0923, found 361.0904) is more highly oxidised. However,
despite many attempts the structure of these dimeric compounds
could not be unambiguously assigned.
F
E
C
G
H
A
B
D
I
R
J
In order to corroborate that 2 is the precursor of 13 and
15ab, three individual A. oryzae transformants solely expressing
terC were pulse fed with 2. Each 50 ml fermentation was fed with
9 mg of 2 each day for 3 days. At the end of the experiment
extracts were examined by LCMS. Very low levels of 13 and 15a
were detected, but only in extracted ion chromatograms.
Meanwhile, the peak for 2 completely disappeared in the
supplemented terC-strains, but not in the supplemented control
strain lacking terC (Figure S3). This observation emphasizes the
in vivo instability of the new intermediate 13. In addition, only the
fermentation media of the terC-strain supplemented with 2
displayed a red colouration, but not the supplemented control
strain (Figure S4).
Scheme 1: A, Labelling patterns in terrein 1 and cryptosporiopsin 5. Related
compounds are boxed; B, Homology analysis[23] between the terrein 1 BGC in
A. lentulus / A. terreus and the putative cryptosporiopsin 5 /-ol 6 BGC in
Roussoella sp. DLM33 and P. macrospinosa, respectively. Confirmed or
expected pathway end-products are boxed. For more detailed analysis of
indicated genes within the terrein 1 BGC see Table S1.
Results
In order to monitor the conversion of 6-HM 2 into 13 in
detail, TerC was produced as soluble protein in Escherichia coli
(Figure S5). Incubation of 2 with TerC in vitro, initially leads to
production of a more polar compound 16 with 16 additional mass
units (C10H10O5; ([M-H]- calculated 209.0450, found 209.0457)
that we propose to be the C-2 hydroxyl-congener of 2 (Figure 2B).
Lactone 16, in turn, readily converts to 13. Upon prolonged
incubation, the peak for 16 is completely absent and 14 and 15b
are formed at the expense of 13 (Figure 2C). In contrast to
previous in vivo experiments neither 15a nor 15a* are formed in
vitro.
Characterization of the FMOs TerC and TerD
The BGC of terrein 1 and related natural products each harbor
one homologous FMO encoding gene (terC). Bioinformatic
analysis suggests that the sequence of TerC deposited at NCBI
is truncated at the N-terminus, with the true terC sequence
featuring an unusual intron starting after the first two 5’ codons
(Figure S2). We therefore generated A. oryzae transformants co-
expressing either the truncated (terCt) or the extended sequence
of terC with terAB following an established protocol.[24] TerA and
2
This article is protected by copyright. All rights reserved.

10.1002/chem.202101447
Chemistry - A European Journal
FULL PAPER
the TerD-derived compounds 17-19 were not observed in vivo,
indicating an unknown incompatibility with our host organism.
Attempts to use alternative start sites for terD-terF, cloning
individual genes from gDNA instead of cDNA, using different
fermentation media and cultivation protocols, also met with failure
to generate new pathway intermediates or the end product
terrein 1. Additional introduction of the four genes terG-terJ
(previously shown to be non-essential for production of 1 in
A. terreus),[3] also with known strong A. oryzae promoters, did not
lead to any change in the metabolic output (Figure S8). Since 1
shows antifungal activity,[27] and its production could conceivably
kill transformants successfully expressing the complete pathway,
we tested its toxicity against A. oryzae. No inhibition of growth at
a concentration up to 10 mg∙mL-1 was observed on solid agar
plates.
A
OH
O
NSAR1
O
HO
HO
HO
HO
2
O
2
O
B
+terAB
9
OH
O
O
OH
OH
10
10
OH
11
9
10
C
+terABC
11
13
Earlier studies showed that the transcriptional regulator
TerR induces high-level transcription of all essential terrein
pathway genes terA-terF and is a useful tool for the construction
of new fungal expression systems.[28] To exclude the possibility
that cloning mistakes or uneven transcription levels between
terC-terF caused by the use of different Aspergillus promotors
(PamyB, Peno, Padh, PgdpA) hampered production of 1, we constructed
a terR-based expression vector. In contrast to the previously
12
11
O
14
O
9
2
O
OH
HO
HO
O
D
12
+terAB
+terRCDEF
OH
10
OH
OH
15b
applied
modular
expression
system,
this
construct
11
13
OH
14
(pTYGSade-terRCDEF) features terR cloned downstream of the
starch-inducible PamyB, followed by its native terminator and a
contiguous A. terreus genomic fragment covering the promotor
region of terC up to the terminator region of terF (Figure S9). This
approach should ensure that terCDEF are expressed as closely
as possible to the native situation.
15a*
9
15a
12
2
O
OH
1
2
3
4
5
6
7
8
14
tR [min]
In former A. oryzae transformants terA-C are equally
functional when cloned from either gDNA or cDNA, showing that
A. terreus introns are generally spliced correctly in A. oryzae. The
vector pTYGSade-terRCDEF was co-transformed with
pTYGSarg-terAB and positive transformants that integrated the
desired fragments were confirmed by PCR (Figure S10).
However, the terABRCDEF transformants only produced the
dimeric compounds related to 15 in elevated titres compared to
previous transformants. Once again neither 1 nor any new
pathway intermediates were observed (Figure 1D).
Unsuccessful reconstitution of the terrein pathway in vivo
prompted us to revisit our in vitro strategy. However, attempts to
obtain the multicopper oxidase TerE and the Kelch protein TerF
as soluble proteins in E. coli met with failure (Figures S11-S12).
We tried to simulate the general oxidative ability of TerE by
addition of Cu2+ ions to the TerCD + 2 in vitro assay. In the
absence of copper this reaction gives a faint red reaction mixture,
but upon presence of copper (CuCl2, 5 mM; Figure S13) it
immediately adopts an intense red colouration. Yet, no new
products were observed by LCMS.
Unlike the final pathway product terrein 1, the
intermediates 13/14 and 18/19 possess a secondary hydroxyl-
group at C-9 that has to be eliminated at some stage. The only
candidate enzyme encoded in the terrein BGC that could
conceivably catalyse this reaction is the SDR TerH. Yet, TerH has
been shown non-essential for production of 1 in A. terreus.[3] TerH
was obtained as soluble protein from E. coli and added to either
TerC + 2 or TerCD + 2 in vitro. No new compounds were
observed. Addition of the N-terminal catalytically inactive DH-
domain of TerB also yielded no new intermediates (data not
shown, Figures S14-S15).
Figure 1: LCMS chromatograms (DAD 210-600 nm) of A. oryzae transformants
expressing the indicated genes with example images of fungal liquid cultures.
Fungal strains depicted in lane A-C were grown for 3 days; lane D was grown
for 5 days. Note that the chromatogram in C is expanded 4-fold vs A and
unlabelled peaks in C are also present in the WT A.
TerD is also an FMO, but is specific to the terrein BGC. We
prepared TerD as recombinant protein in E. coli (Figures S6-S7).
Incubation of TerD with 6-HM 2 yields a single new product 17
(Figure 2D) that was identified as the C-4-hydroxyl-congener of 2
by 2D-NMR analysis (Figures S54-S58).
Co-incubation of TerC and TerD with 2, as well as pre-
incubation of 6-HM 2 with TerC followed by addition of TerD,
yields another very polar new minor product 18 (Figure 2E) in
addition to the major product 13. Compared to the TerC-derived
tetra-ol 13 compound 18 shows additional 16 mass units
(C9H12O5; ([M-H]- calculated 199.0606, found 199.0605). Based
on UV, mass spectra, retention time and comparison to previously
observed compounds, 18 appears to constitute the C-4 hydroxyl-
congener of 13. Compound 18 slowly oxidises to 19 (Figure 2F)
similar to the oxidation of 13 to 14. UV analysis of 19 is consistent
with the quinone structure.[26]
Late steps during the biosynthesis of terrein 1
In order to reconstitute and decipher the late-stage biosynthesis
of terrein 1 we individually co-expressed all of the proposed
essential genes terA-terF in A. oryzae. In this experiment each
structural gene was expressed from a promoter known to be
functional in A. oryzae.[24] However, none of the transformants
showed production of
1 and no other putative pathway
intermediates other than 13/14 and the corresponding dimeric
forms including 15a could be identified (Figure S8). Unexpectedly,
3
This article is protected by copyright. All rights reserved.

10.1002/chem.202101447
Chemistry - A European Journal
FULL PAPER
elimination, and then BBr3-mediated demethylation. This afforded
a ca 2:1 mixture of inseparable E:Z isomers of 20.
2
OH
O
A
standard
O
HO
HO
O
2
OH
2
OH
OH
B
TerC
1h
13
13
16
OH
O
14
C
TerC
3h
OH
14
15b
13
O
O
2
2
O
OH
D
TerD
1h
17
HO
16
OH
O
HO
HO
O
13
17
E
pre-incubation
TerC
then TerD
2h
Scheme 2. Synthesis of alternative substrates. A, Synthesis of 20; B, synthesis
of 6-methoxymellein 2b; C, possible product of enzymatic conversion of
substrate 20.
OH
HO
HO
OH
OH
18
16
18
F
pre-incubation
TerC
then TerD
12h
19
Neither TerC nor TerD alone show any conversion of 20, but
simultaneous incubation with both enzymes results in a new small
peak in LCMS-chromatograms (Figures 2G and 2H) of assay
components. HRMS confirmed a composition of C9H10O3 ([M-H]-
calculated 165.0552, found 165.0549) that would match the
structure of tri-ol 21 that has been proposed as a potential
biosynthetic intermediate.[17,19] However, very low production of
21 thwarted structural elucidation by NMR. In addition, the
methoxy-derivative 6-methoxymellein 2b (Scheme 2B) was
probed as an alternative substrate, but neither TerC nor TerD
exhibited any conversion (Figure S16).
14
OH
HO
O
O
O
13
OH
OH
19
20
OH
G
standard
HO
20
E/Z
mixture
20
Identification a homologous BGC in Roussoella sp. DLM33
Inability to progress understanding of the terrein system beyond
the activity of TerC led us to consider the related biosynthetic
systems which lack TerD and the later proteins. We hypothesized
that such pathways may use more tractable rearrangement
systems and that results from such studies might help to draw
conclusions about the functions of TerE and TerF that were
previously shown to be essential for formation of 1.[3]
H
TerCD
18h
OH
OH
HO
21?
21
(possible product)
5.5
tR [min]
6.5
1.5
2.5
3.5
4.5
7.5
Figure 2: LCMS-chromatograms (DAD 210-600 nm) of in vitro assays with
TerC/D using substrate 6-HM 2 or 20 under conditions as indicated. Colour
changes of selected enzymatic conversions are given next to the
chromatograms.
FungiSMASH analysis[29] identified a candidate BGC in
the genome of Roussoella sp. DLM33, a producer of the double
chlorinated lactone
8 and the cryptosporiopsin 5 derived
roussoellatide 7 (for details of genome sequencing see ESI).[16]
Refinement using the gene prediction tools AUGUSTUS[30] and
FGENESH[31] associated nine genes with the proposed rsl BGC
(~36 kb). The central genes encode RslA and RslB which are
homologs of TerA and TerB and are expected to produce 6-HM
2. Two additional genes encode an FMO (RslC, homologous to
TerC) and a transcriptional activator (RslR, homologous to TerR)
We considered it possible that dehydration of any intermediate
might have to occur before the oxidative steps, and that the
dehydration might be catalysed by proteins encoded outside the
known terrein BGC. We therefore prepared 20 in a three-step
synthesis from 2 as an alternative substrate for in vitro studies
(Scheme 2A). O-methylation of 2 was followed by base catalysed
4
This article is protected by copyright. All rights reserved.

10.1002/chem.202101447
Chemistry - A European Journal
FULL PAPER
and are also conserved among related BGC (Scheme 1B).
Further genes in the rsl BGC encode: two flavin-dependent
halogenases (RslK and RslN); a short-chain dehydrogenase/
reductase (SDR, RslO); an O-methyltransferase (RslP); and
another putative incomplete transcription factor. Seven of these
genes are also conserved in the putative cryptosporiopsinol 6
BGC of Periconia macrospinosa (Scheme 1B).
The putative rsl BGC was confirmed by heterologous
expression of rslA in A. oryzae. This yields the same three
polyketides 9-11 as previously reported for expression of terA, in
a combined titre of 125 mg∙L-1 (Figure 3B).[10] Addition of rslB
results in formation of the key intermediate 6-HM 2 in high titres
of approx. 500 mg∙L-1 (Figure 3B).
A
NSAR1
O
A
O
NSAR1
HO
HO
HO
10
B
+rslA
9
11
B
OH
O
O
+rslA
9
OH
OH
2
10
C
C
+rslAB
OH
+rslAB
Compatibility between TerA/RslA and TerB/RslB
TerB is a tridomain protein with an N-terminal catalytically inactive
11
O
10
10
11
dehydratase (ψDH),
methyltransferase- (ψC-MeT) and
a
central catalytically inactive C-
functional C-terminal
a
2
OH
O
O
ketoreductase (KR).[10] Interestingly, RslB, that is ca. 300 amino
acid residues shorter than TerB, does not contain the central ψC-
MeT domain (InterPro[32] analysis). Since TerA and TerB interact
during the biosynthesis of 6-HM 2 in vivo we were interested
whether any of these enzymes could be exchanged with its
homologous protein RslA or RslB, respectively.[10] We therefore
generated two transformant strains of A. oryzae expressing either
terA + rslB or rslA + terB. Both transformant strains produce 2
very efficiently (ca. 400 mg∙L-1 and 450 mg∙L-1 vs. 500 mg∙L-1;
Figure S17).
D
D
+rslABKN
+rslABKN
O
O
22
23
HO
8
2
25
11
OH
24
Cl
HO
2
Cl
8
E
F
+rslABK
+rslABK
OH
O
23
Cl
O
O
Characterization of the flavin-dependent halogenases RslK
and RslN
HO
10
11
22
Addition of the two annotated halogenase genes (rslK and rslN)
to the A. oryzae transformants leads to production of the di-
chlorinated 8 in vivo, that was also isolated from cultures of
Roussoella sp. DLM33.[16] The mono-chlorinated congeners 22
and 23 are also produced (Figure 3D; NMR in ESI). Individual
expression revealed regioselectivity for RslK at the C-6 and for
RslN at the C-4 position, respectively (Figures 3E and 3F). These
results are in agreement with recent in vitro characterization of the
homologous halogenases ChmKN (cyclohelminthols) and PloKN
(palmaeonones).[17]
2
OH
O
F
E
+rslABN
+rslABN
22
HO
23
Cl
10
25
24
11
OH
O
Cl
O
2
G
G
HO
+rslABNK
+rslABKN
+rslC
24
OH
Both 22 and 23, but not 8, are slowly converted into their
C-8 hydroxyl-derivatives including 24 by endogenous enzymes in
A. oryzae, supported by feeding studies (Figure S18, S81-S85).
6-HM 2 in turn is hydroxylated at C-6 to give 25 upon prolonged
+rslC
23
OH
O
10
11
O
22
22
incubation
which
is
also
consistent
with
previous
HO
2
observations.[10,17] Supplementation of the fermentation medium
with sodium bromide (50 µg∙mL-1) results in the successful
formation of mono- and dibrominated derivatives of 2 as well as
mixed chloro-bromo species based on distinct isotope ratios in the
respective mass spectra (Figure S19). However, iodinated
derivatives were not formed in the presence of NaI.
H
OH 25
G
+rslABNK
+rslABKN
+rslCOP
+rslCLM
OMe
Cl
OH
OH
10
11
26
23
HO
Cl 26
1
2
3
4
5
6
7
8
tR [min]
Characterization of the flavin-dependent monooxygenase
RslC and the O-methyltransferase RslP
Figure 3: LCMS chromatograms (DAD 210-600 nm) of A. oryzae transformants
expressing the indicated genes from the rsl BGC of Roussoella sp. DLM33.
Example images of fungal liquid cultures are given next to the chromatograms.
Fungal strains depicted in lane A-G were grown for 4 days. The fungal strain
depicted in lane H was grown for 2 days to show a visible peak for 26.
The TerC/RslC-type FMO are encoded by all related BGC
(Scheme 1B). In fact, rslC appears to exhibit the same unusual
intron pattern located directly at the 5’- end of the corresponding
mRNA analogous to terC (Figure S2).
5
This article is protected by copyright. All rights reserved.

10.1002/chem.202101447
Chemistry - A European Journal
FULL PAPER
Similar to previous in vivo expression of terABC, rslABCKN-
transformants display a distinct reddish-brown colouration of the
fermentation medium. However, no new compounds could be
detected by LCMS analysis (Figure 3G). Shortened or prolonged
cultivation also revealed no new products by LCMS.
All attempts to obtain RslC as soluble and active protein
in E. coli met with failure. Alternatively, TerC was incubated with
8 in vitro and although no turnover was visible by LCMS analysis
the samples displayed a reddish-brown colouration as observed
in vivo (data not shown), indicative of oxidation.
the proposed pathway end-product cryptosporiopsin 5. Since no
gene encoding an enzyme that would catalyze dehydration at this
position is conserved among homologous BGC (Scheme 1B) this
step remains obscure. Despite heterologous expression of seven
genes (rslABCKNOP) in vivo, no compounds were identified that
feature a re-arranged cyclopentenone/-ol skeleton.
Discussion
Since potential RslC-derived products may also be
chemically unstable or prone to degradation in A. oryzae (e.g. as
observed for 13), we proceeded to add the two ancillary genes
rslP (O-methyltransferase) and rslO (SDR) to generate “full
cluster” transformants. A new pale brownish di-chlorinated
compound 26 was found in the fermentation media of respective
transformants exclusively on the third day of cultivation (before
the culture medium adopted a dark colouration, Figure 3H).
Compound 26 is two mass units heavier than the expected
pathway end product cryptosporiopsin 5 (C10H12O4Cl2; ([M-H]-
calculated 265.0034, found 265.0037). 2D-NMR analysis
revealed that 26 is a di-chlorinated derivative of the aromatic TerC
product 13 with an additional O-methylation at the C-3 hydroxyl
6-HM 2 is an intermediate during the biosynthesis of terrein 1 and
related (poly-chlorinated) natural products such as 5.[3,17] In the
terrein pathway lactone 2 is formed by collaboration between an
nrPKS and a multidomain protein with catalytically active KR-
domain in vivo.[10] Bioinformatic analysis allowed the identification
of a related BGC in the marine-derived Roussoella sp. DLM33
that shares the two corresponding genes rslAB, amongst others.
Heterologous expression of either rslA alone or in combination
with rslB revealed the same qualitative and quantitative product
manifold reported for the homologous proteins of the terrein 1
BGC.[10] In contrast to TerB, RslB does not feature a central ψC-
MeT domain. Despite this altered domain architecture combining
homologous genes from two different species (terA + rslB and
rslA + terB) yielded the product 6-HM 2 in titres nearly as high as
obtained for the native system (Figure S17). These results show
that collaboration is also facilitated between homologous proteins
originating from different fungal species. Furthermore, since RslB
physically lacks ψC-MeT it appears that this domain is neither
involved in protein-protein interactions, or fulfils a catalytic role, in
line with previous findings.[10]
group (isolated
6
mg∙L-1; Figures S87-S91). To exclude
spontaneous methylation all analytical steps were repeated in the
absence of MeOH whereby 26 was still produced. Interestingly,
26 has been previously isolated from Periconia macrospinosa, a
producer of cryptosporiosinol 6.[33] This suggests that O-
methylation is introduced by RslP. No obvious function could be
attributed to the SDR RslO.
In analogy to the TerC-derived 13, the RslC-derived 26
features a secondary hydroxyl-group at C-9 that is not present in
Scheme 3. Summary of observed and hypothetical reactions on the pathway to terrein 1 and cryptosporiopsinol 5. A, Reactions and intermediates observed in vivo
and in vitro. All compounds feature the C-9 hydroxyl which is not observed in 1 and 5; B, hypothetical pathways to 1 and 5 involving early dehydration and late
benzilic acid rearrangements. See ESI for DFT calculations of the respective rearrangements in each case.
6
This article is protected by copyright. All rights reserved.

10.1002/chem.202101447
Chemistry - A European Journal
FULL PAPER
It is likely that collaboration during the biosynthesis of 6-HM 2 is a
general feature of such homologous systems (e.g. also ChmAB
from H. velutinum),[17] but further in-depth investigations are
required to characterize the interacting elements in greater detail.
The rsl BGC encodes two flavin-dependent halogenases
RslK and RslN that halogenate 6-HM 2 at the C-4 (RslN) and C-6
(RslK) positions (Scheme 3). The observed regioselectivity
corresponds with previous reports on homologous enzymes.[17]
The formation of brominated congeners in the presence of NaBr
reveals that both halogenases show halide promiscuity. This is a
potentially useful approach to generate novel brominated species.
The O-methylated congener 26 was identified only in the
presence of the O-methyltransferase RslP. Based on labelling
studies O-methylation within 26 and cryptosporiopsin 5 shares the
same position at C-3 (Scheme 1A). It remains elusive whether 26
constitutes a true pathway intermediate or an off-pathway shunt.
Despite elevated titres of 6-HM 2 in the respective transformants
(Figure 3G), the non-chlorinated 13 was not observed. This
indicates that RslC requires chlorination at C-4 and C-6 for
substrate recognition, and rules out the activity of RslC before
chlorination.
Characterization of both FMOs TerC and RslC provides
experimental evidence that the biosynthetic pathways towards
terrein 1 and related (poly-chlorinated) congeners branch after
oxidative lactone cleavage. The subsequent ring contractions
follow a different pathway in each case as indicated by distinct
labelling patterns (Scheme 1A). However the pathway beyond
this decarboxylation remains unclear in both cases.
Of primary interest was the elucidation of the function of the flavin-
dependent monooxygenases that are common to all the related
pathways (e.g. TerC and RslC, Scheme 1B) and that are likely to
catalyse the final common step before divergence of the terrein
and other pathways. Characterization of TerC shows that this
enzyme initially hydroxylates 6-HM 2 adjacent to the carbonyl-
position at C-2 to yield intermediate 16. TerC thus matches the
A key difficulty revealed by our results is the timing of the
dehydration required to provide the E-propenyl side chain of
terrein 1, cryptosporiopsin 5 and the majority of other compounds
in this class. The fact that this moiety is widely present suggests
that the dehydration should occur early in the pathway. Since 13
is also produced by TerC in vitro, we can exclude that the
secondary hydroxyl-group at C-9 is (re)introduced by native
enzymes of our in vivo host A. oryzae.[43–46]. Likewise 18, that is
derived from sequential catalysis of TerC and TerD in vitro, does
not feature the E-propenyl side chain.
Furthermore, previous studies[3] imply that the TerD-
derived lactone 17 is not an intermediate but a shunt metabolite:
Feeding of 17 to a ∆terB strain of A. terreus did not restore
production of terrein 1. Similarly, the ∆terD strain of A. terreus did
not show production of 1 in the presence of 17.
However, neither TerAB nor TerC (or the respective
homologs) appear capable of catalysing the required dehydration.
An attractive possibility would be coupling the oxidative
decarboxylation catalysed by TerC/RslC with the dehydration
(Scheme 3B). The fact that we did not observe this in vitro or in
vivo may indicate that TerC/RslC requires another protein to
achieve such a transformation. However, since there are no more
homologous genes in the two BGC a candidate is hard to identify.
In addition, full expression of both clusters in A. oryzae could not
progress either pathway further, suggesting that if such a protein
exists, it is not encoded within the ter or rsl BGC. In other systems,
such as those responsible for azaphilone biosynthesis in
Monascus species, a specific O-acetylation and elimination
sequence (MrPigM and MrPigO) provides the prop-2-enyl
moiety.[47] However homologs of these proteins are not encoded
within the terrein or related BGC.
For this reason, it appears unlikely that 9-hydroxylated
compounds such as 18 and 19 are true intermediates. Although
we did not observe compounds such as 28 (the C-8/C-9
dehydrated homolog of 19), this compound could be envisaged
as a potential direct precursor of 1 via tautomerisation to 29 and
then an α-ketol, or related benzilic acid rearrangement (Scheme
3B).[48] Such a mechanism is supported as feasible by DFT
calculations (see ESI for details). Ring contractions in natural
product biosynthesis have been previously proposed to proceed
following such rearrangements. Examples include the
biosynthetic pathways of xenovulene A,[49] preuisolactone A[50]
(both C7 to C6) and fredericamycin A (C6 to C5).[51] In synthetic
function
expected
for
a
classical
16
flavin-dependent
decarboxylates
monooxygenase.[34,35]
Intermediate
spontaneously to the tetra-ol 13 in vitro (Scheme 3A). An
analogous mechanism was proposed by Townsend et al. for the
FMO-dependent oxidative lactone opening during the
biosynthesis of cercosporin and related perylenequinones.[36,37]
Intermediate 13 is unstable both in vivo and in vitro and it is
converted into the corresponding quinone 14 by autoxidation[38]
that readily dimerises to 15b that exhibits a strong red colouration
(Figure S13). Tetra-ol 13 appears to be a substrate for TerD which
hydroxylates at C-4. But the product, 18, does not proceed to ring-
contracted species, giving only dimers by autoxidation (Scheme
3A). TerD appears inactive in A. oryzae expression experiments
for unknown reasons. However, TerD is a potential tool for the in
vitro stereoselective hydroxylation of isocoumarin/ phenol-
derivated compounds. Further investigations and screening of a
substrate library will be necessary to evaluate its substrate
promiscuity.
A significant difficulty in this study has been the instability
of compounds beyond 13 and our inability to satisfactorily identify,
isolate or characterise them. Dimers such as 15b were identified
by HRMS in vitro and in vivo, but do not appear to be on the
pathway to 1. Formation of 15a exclusively in vivo suggests that
a native enzyme in A. oryzae is involved in its formation, adding
another layer of complexity to efforts to detect and identify
pathway intermediates. Oxidative coupling of phenol-related
compounds in fungi is often mediated by laccases or cytochrome
P450-enzymes.[39,40] A recently described laccase catalyses the
coupling of two subunits structurally very similar to 14, yielding the
bibenzoquinone oosporein that also exhibits a strong red
colour.[41]
As anticipated, the homologous FMO RslC catalyses the
same fundamental reaction, i.e. oxidative lactone opening of the
di-chlorinated 8, that we could only observe in vivo. However, the
direct product 27 defied detection. Similar to previous
transformants expressing terABC, A. oryzae rslABCKN
transformants adopted a distinct dark colouration (cf. Figures 3G
and 3H) indicative of further oxidation. Alternatively, production of
27 might have induced production of fungal melanins that function
as
a defence mechanism against chemical environmental
stressors.[42]
7
This article is protected by copyright. All rights reserved.

10.1002/chem.202101447
Chemistry - A European Journal
FULL PAPER
chemistry such reactions are often performed in the presence of
metal-ion catalysts, pointing out a potential function of the
predicted copper-dependent TerE.[52,53] Decarboxylation after
rearrangement would give terrein 1 after tautomerisation.
Keywords: 6-hydroxymellein • oxidative decarboxylation • flavin-
dependent monooxygenase • terrein • cryptosporiopsin
A similar rearrangement mechanism could be envisaged
References
during the biosynthesis of cryptosporiopsin 5 (Scheme 3B).
Chlorination of C-4, however, in this case, prevents an identical
mechanism. Here, 3-methoxy hydroquinone 30 could oxidise to
benzoquinone 31. Rearrangement of 31 by a benzilic acid
reaction would also lead to ring-contraction, but via extrusion of
C-3. This is consistent with the observed labelling pattern. Early
[1]
[2]
H. Raistrick, G. Smith, Biochem. J. 1935, 29, 606–11.
T. O. Larsen, J. Smedsgaard, K. F. Nielsen, M. A. E. Hansen, R. A.
Samson, J. C. Frisvad, Med. Mycol. 2007, 45, 225–232.
C. Zaehle, M. Gressler, E. Shelest, E. Geib, C. Hertweck, M. Brock,
Chem. Biol. 2014, 21, 719–731.
J. C. Lee, M. K. Yu, R. Lee, Y. H. Lee, J. G. Jeon, M. H. Lee, E. C.
Jhee, I. D. Yoo, H. K. Yi, J. Endod. 2008, 34, 433–437.
M. Arakawa, T. Someno, M. Kawada, D. Ikeda, J. Antibiot. (Tokyo).
2008, 61, 442–448.
[3]
[4]
[5]
[6]
methylation of O-3 would create
a
methyl ester after
rearrangement and thereby prevent the decarboxylation which
must occur during the rearrangement on the route to 1. Again,
DFT calculations (ESI section 2.14) suggest this type of
mechanism would be feasible.
S. H. Park, D. S. Kim, W. G. Kim, I. J. Ryoo, D. H. Lee, C. H. Huh,
S. W. Youn, I. D. Yoo, K. C. Park, Cell. Mol. Life Sci. 2004, 61,
2878–2885.
[7]
B. Kim, J. S. Park, H. Y. Choi, S. S. Yoon, W. G. Kim, Sci. Rep.
2018, 8, 8617.
[8]
A. J. Birch, A. Cassera, A. R. Jones, Chem. Commun. 1965, 9,
167–168.
[9]
R. A. Hill, R. H. Carter, J. Staunton, J. Chem. Soc. Perkin Trans. 1
1981, 2570–2576.
L. Kahlert, M. Villanueva, R. J. Cox, E. J. Skellam, Angew. Chemie
Int. Ed. 2021, 60, 11423–11429.
Y. Honmura, S. Uesugi, H. Maeda, K. Tanaka, T. Nehira, K. ichi
Kimura, M. Okazaki, M. Hashimoto, Tetrahedron 2016, 72, 1400–
1405.
Conclusion
[10]
[11]
The biosynthesis of terrein 1 and related compounds such as 5
and 6 has remained mysterious for more than 85 years. Despite
the recent discovery of the BGC and molecular studies involving
gene knockouts that confirmed the intermediacy of 6-
hydroxymellein 2, no more details on their biosynthesis could be
found. Here we have advanced knowledge in this area by another
small step. The FMO TerC (and its homologs) are involved in the
oxidative decarboxylation which removes C-1 early in the
biosynthesis, and this seems to be the last step in common
between the terrein and (poly)chlorinated pentenone pathways.
However we have not identified the origin of the required
dehydration, but our studies suggest this must also occur early in
the pathways. The highly oxygenated intermediates produced by
TerC and TerD are unstable and difficult to characterise, but if 1
or 5 had been formed they would have been detected. The lack
of these compounds suggests the likely importance of early
dehydration. It is noteworthy that while our approach to
[12]
[13]
T. Matsumoto, T. Hosoya, H. Tomoda, M. Shiro, H. Shigemori,
Chem. Pharm. Bull. 2011, 59, 1559–1561.
G. M. Strunz, A. S. Court, J. Komlossy, M. A. Stillwell, Can. J.
Chem. 1969, 47, 2087–2094.
D. Giles, W. B. Turner, J. Chem. Soc. C Org. 1969, 0, 2187–2189.
R. J. J. C. Lousberg, Y. Tirilly, M. Moreau, Experientia 1976, 32,
331–332.
E. L. F. Ferreira, D. E. Williams, L. P. Ióca, R. P. Morais-Urano, M.
F. C. Santos, B. O. Patrick, L. M. Elias, S. P. Lira, A. G. Ferreira, M.
R. Z. Passarini, et al., Org. Lett. 2015, 17, 5152–5155.
T. Ugai, A. Minami, S. Tanaka, T. Ozaki, C. Liu, H. Shigemori, M.
Hashimoto, H. Oikawa, ChemBioChem 2020, 21, 360–367.
J. S. E. Holker, K. Young, J. Chem. Soc. Chem. Commun. 1975, 0,
525–526.
L. O. Zamir, C. C. Chin, Bioorg. Chem. 1982, 11, 338–349.
D.-S. Tian, E. Kuhnert, J. Ouazzani, D. Wibberg, J. Kalinowski, R. J.
Cox, Chem. Sci. 2020, 11, 12477–12484.
C. Schotte, L. Li, D. Wibberg, J. Kalinowski, R. J. Cox, Angew.
Chemie - Int. Ed. 2020, 59, 23870–23878.
L. Kahlert, C. Schotte, R. J. Cox, Synthesis, 2021, DOI 10.1055/a-
1401-2716.
T. Carver, S. R. Harris, M. Berriman, J. Parkhill, J. A. McQuillan,
Bioinformatics 2012, 28, 464–469.
K. A. K. Pahirulzaman, K. Williams, C. M. Lazarus, in Methods
Enzymol., Academic Press Inc., 2012, pp. 241–260.
M. C. Tang, Y. Zou, K. Watanabe, C. T. Walsh, Y. Tang, Chem.
Rev. 2017, 117, 5226–5333.
T. Wilke, M. Schneider, K. Kleinermanns, Open J. Phys. Chem.
2013, 03, 97–102.
J. Goutam, G. Sharma, V. K. Tiwari, A. Mishra, R. N. Kharwar, V.
Ramaraj, B. Koch, Front. Microbiol. 2017, 8, 1334.
M. Gressler, P. Hortschansky, E. Geib, M. Brock, Front. Microbiol.
2015, 6, 184.
K. Blin, S. Shaw, K. Steinke, R. Villebro, N. Ziemert, S. Y. Lee, M.
H. Medema, T. Weber, Nucleic Acids Res. 2019, 47, W81–W87.
M. Stanke, B. Morgenstern, Nucleic Acids Res. 2005, 33, W465–
W467.
V. Solovyev, P. Kosarev, I. Seledsov, D. Vorobyev, Genome Biol.
2006, 7, S10.
P. Jones, D. Binns, H.-Y. Chang, M. Fraser, W. Li, C. McAnulla, H.
McWilliam, J. Maslen, A. Mitchell, G. Nuka, et al., Bioinformatics
2014, 30, 1236—1240.
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
sequentially
reconstitute
fungal
secondary
metabolite
biosynthetic pathways by heterologous expression in A. oryzae
NSAR1 is usually very expedient and reliable, it was not able to
decipher the final transformation during the biosynthesis of 1 and
5.[20,43,49,54,55] Inexplicably TerD seems inactive in heterologous
expression experiments in A. oryzae, although it is active in vitro.
We propose that polyhydroxylated intermediates probably
form quinones and then benzilic acid type rearrangements could
connect the biosynthesis of terrein 1 and the related compounds
such as 5. However, we have so-far failed to discover the
catalysts involved in the key dehydration and rearrangement
steps and it is clear that significantly more work will have to be
done to finally elucidate these steps.
Acknowledgements
[33]
[34]
[35]
[36]
[37]
J. Liu, M. Chen, R. Chen, K. Xie, D. Chen, S. Si, J. Dai, J. Chinese
Pharm. Sci. 2020, 29, 244–251.
E. Romero, J. R. Gómez Castellanos, G. Gadda, M. W. Fraaije, A.
Mattevi, Chem. Rev. 2018, 118, 1742–1769.
A. H. Westphal, D. Tischler, W. J. H. van Berkel, Arch. Biochem.
Biophys. 2021, 702, 108820.
A. G. Newman, C. A. Townsend, J. Am. Chem. Soc. 2016, 138,
4219–4228.
J. Hu, F. Sarrami, H. Li, G. Zhang, K. A. Stubbs, E. Lacey, S. G.
Stewart, A. Karton, A. M. Piggott, Y. H. Chooi, Chem. Sci. 2019, 10,
1457–1465.
Matthias Brock is thanked for the gift of Aspergillus terreus
SBUG844. LK was funded by DFG (CO 1328/5-1). MH is funded
by DFG (CO 1328 / 9-1). DFG is thanked for the provision of
LCMS (DFG, INST 187/621-1,) and NMR (INST 187/686-1)
instrumentation. RGSB, DIB and LPI were funded by FAPESP
(2013/50228-8, 2019/17721-9, 2016/21341-9, 2020/11691-8,
2018/10742-8, 2016/05133-7).
[38]
T. H. James, J. M. Snell, A. Weissberger, J. Am. Chem. Soc. 1938,
8
This article is protected by copyright. All rights reserved.

10.1002/chem.202101447
Chemistry - A European Journal
FULL PAPER
60, 2084–2093.
[39]
C. Gil Girol, K. M. Fisch, T. Heinekamp, S. Günther, W. Hüttel, J.
Piel, A. A. Brakhage, M. Müller, Angew. Chemie - Int. Ed. 2012, 51,
9788–9791.
[40]
[41]
[42]
L. S. Mazzaferro, W. Hüttel, A. Fries, M. Müller, J. Am. Chem. Soc.
2015, 137, 12289–12295.
P. Feng, Y. Shang, K. Cen, C. Wang, Proc. Natl. Acad. Sci. U. S. A.
2015, 112, 11365–11370.
R. J. B. Cordero, A. Casadevall, Fungal Biol. Rev. 2017, 31, 99–
112.
[43]
[44]
Y. He, R. J. Cox, Chem. Sci. 2016, 7, 2119–2127.
R. Fujii, T. Ugai, H. Ichinose, M. Hatakeyama, T. Kosaki, K. Gomi, I.
Fujii, A. Minami, H. Oikawa, Biosci. Biotechnol. Biochem. 2016, 80,
426–431.
[45]
[46]
[47]
J. Feng, F. Surup, M. Hauser, A. Miller, J. P. Wennrich, M. Stadler,
R. J. Cox, E. Kuhnert, Chem. Commun. (Camb). 2020, 56, 12419–
12422.
Z. Song, W. Bakeer, J. W. Marshall, A. A. Yakasai, R. M. Khalid, J.
Collemare, E. Skellam, D. Tharreau, M. H. Lebrun, C. M. Lazarus,
et al., Chem. Sci. 2015, 6, 4837–4845.
W. Chen, R. Chen, Q. Liu, Y. He, K. He, X. Ding, L. Kang, X. Guo,
N. Xie, Y. Zhou, et al., Chem. Sci. 2017, 8, 4917–4925.
A. Burke, C. Marques, Mini. Rev. Org. Chem. 2007, 4, 310–316.
R. Schor, C. Schotte, D. Wibberg, J. Kalinowski, R. J. Cox, Nat.
Commun. 2018, 9, 1963.
[48]
[49]
[50]
[51]
[52]
[53]
[54]
A. J. E. Novak, C. E. Grigglestone, D. Trauner, J. Am. Chem. Soc.
2019, 141, 15515–15518.
Y. Chen, Y. Luo, J. Ju, E. Wendt-Pienkowski, S. R. Rajski, B. Shen,
J. Nat. Prod. 2008, 71, 431–437.
H. Brunner, H. B. Kagan, G. Kreutzer, Tetrahedron Asymmetry
2003, 14, 2177–2187.
A. J. Burke, A. Moutayakine, Chem. Commun. 2021, 57, 2585–
2590.
R. Nofiani, K. de Mattos-Shipley, K. E. Lebe, L. C. Han, Z. Iqbal, A.
M. Bailey, C. L. Willis, T. J. Simpson, R. J. Cox, Nat. Commun.
2018, 9, 3940.
[55]
L. Kahlert, E. F. Bassiony, R. J. Cox, E. J. Skellam, Angew. Chemie
- Int. Ed. 2020, 59, 5816–5822.
9
This article is protected by copyright. All rights reserved.

10.1002/chem.202101447
Chemistry - A European Journal
FULL PAPER
Entry for the Table of Contents
Heterologous expression of two related biosynthetic gene clusters from Aspergillus terreus and Roussoella sp. DLM33 confirmed a
conserved flavin-dependent monooxygenase that initiates the oxidative decarboxylation of the key intermediate (dichloro-) 6-
hydroxymellein. Further in vitro studies imply that both pathways diverge after this initial oxidation, whereas additional dehydration
appears essential for downstream processing.
10
This article is protected by copyright. All rights reserved.