Angewandte Chemie International Edition
10.1002/anie.201916005
COMMUNICATION
processes,[11] the presence of OocK homologs in diverse orphan
PKS biosynthetic gene clusters (BGCs) (Figures S1-S4) suggests
that nature employs oxygen insertion more widely to construct
unusual polyketide scaffolds. In this work, we applied a genome
mining strategy to investigate the wider scope of flavoprotein
monooxygenases acting during polyketide elongation. The data
revealed an architecturally distinct oxygen insertion module that
generates a glycolyl intermediate, as well as evidence for a third
module type involved in oxime formation. Both new modules are
used in the biosynthesis of lobatamides, potent cytotoxins
previously only known from marine tunicates, but identified here
from a culturable plant symbiont.
a)
4
kb
lbm F G
H
I
J K LMNO
A
B
0
0
C-A-PCP-KS -Ox-ACP-KS -
MT-ACP-KS-KR-ACP-KS
DH-ACP-C-A-PCP-
KS-DH-KR-ACP-KS
0
C
D
E
P Q R
KR-ACP-KS-Ox-MT-ACP-KS-DH-
KR-ACP-KS-DH-KR-MT
ACP-KS-DH-KR-ACP- KS-DH-
KS-ACP-KS-KR-ACP ACP-TE
b)
c)
8
9
We initiated our work by identifying OocK homologs and
other PKS-associated flavoprotein monooxygenases encoded in
orphan BGCs using GenBank, genome neighborhood,[12]
phylogenetic, and manual analyses (status: November 2019)
(
Figures S1-S4). These revealed 69 candidate trans-AT PKS
, molecular formula:C26H42N2O7
systems harboring such enzymes (Figures S1-S4). The
phylogram contained a large clade containing the functionally
related OocK, PedG, and TobD as well as homologs from
uncharacterized PKSs. Another functionally assigned clade bears
module-integrated oxygenase (Ox) domains for which the variant
from Burkholderia sp. FERM BP-3421 was shown to introduce an
epoxide unit in spliceostatin (Figure S1).[8] In addition, other non-
OocK clades mostly belong to Ox domains integrated in various
module types, suggesting further biosynthetic diversity. Focusing
on uncharacterized oxygenases (Figure S1), an orphan trans-AT
system (termed lbm PKS, Figure 2A) in the Gram-negative
bacterium Gynuella sunshinyii (NZ_CP007142) appeared an
intriguing candidate, since the PKS contains oxygenase modules
from two different unassigned clades (Figure S1, Table S1). G.
sunshinyii is an unusual halophilic root-associated plant
symbiont[ that was recently recognized as a talented producer
Figure 2. The lbm BGC and predicted polyketide structures. a) BGC
architecture with core PKS and accessory genes marked in dark- and light grey,
respectively. The domain architecture of the core PKSs is shown using the
following abbreviations: A, adenylation domain; C, condensation domain; DH,
0
dehydratase; ER, enoylreductase; KR, ketoreductase; KS, ketosynthase; KS ,
non-elongating KS; MT, methyltransferase; Ox, oxygenase; TE, thioesterase.
Numbers above KS domains correspond to consecutive KS and module
numbers. Ox domains are shown in grey. b) TransATor-predicted structure 8 of
the lbm product. High- and low-confidence regions are shown in black and grey,
respectively. c) Manually refined structure 9 used to find the natural product.
To search for the predicted compound, G. sunshinyii was
cultivated, and the extract was analyzed by ultra-high-
performance liquid chromatography/high-resolution spectrometry
(UHPLC-HRMS). Manual inspection of metabolite-related MS
features and their molecular formulae suggested from the high-
resolution masses showed a candidate ion peak at m/z 513.2230
13]
[14]
of diverse natural products. Its genome contains six trans-AT
PKS BGCs, the highest known number for any organism. The two
integrated Ox domains are located in the PKS proteins LmbA and
LmbC, and will be referred to as LmbA-Ox and LmbC-Ox,
+
[M+H] (Figure S5). This corresponded to a molecular formula of
respectively. Additional features in these modules are
a
C H N O (calc. 513.2231). This formula was closest to that of
2
7
32
2
8
methyltransferase (MT) in the LmbC-Ox module, suggesting α-C-
26 42 2 7
the predicted structure (C H N O ) and appeared a good
0
methylation, and a predicted non-elongating KS (KS ) in the
candidate for the lbm product. No other orphan PKSs in the
genome were predicted to account for a similar chemical formula
that contained two nitrogen atoms (i.e., PKSs that contained two
NRPS modules).[ To isolate the compound, bacterial pellets
from a 2 L culture of G. sunshinyii culture were extracted with
acetone. MS-guided fractionations using reversed phase-high
performance liquid chromatography (RP-HPLC) provided pure
LmbA-Ox module, suggesting a moiety introduced by the
upstream module is further modified. This upstream module is
located at the N-terminus of LbmA and resembles an NRPS
loading module with predicted[15] glycine specificity. A C-terminal
TE domain is encoded on the PKS gene lmbE at the downstream
end, overall suggesting collinearity between the PKS gene order
and biosynthetic events.
14]
1
compound 5. The H NMR and HSQC data showed the presence
of a doublet aliphatic methyl, one methyl connected to an sp2
carbon, one methoxy group, three oxymethines, and 11 methines
As guidance for the targeted isolation of the lbm polyketide,
we analyzed the PKS using the recently developed automated
prediction tool TransATor.[3] This web application suggests
chemical structures for trans-AT PKSs from phylogenetically
inferred[ KS substrates. In agreement with the unusual PKS
architecture, the TransATor output 8 (Figure 2B) contained
extended regions of low confidence. After removal of a methyl
group from the five-bonded C18, the structure was refined by
closer inspection of the modular architecture. Since TransAtor
2
connected to sp carbons (Figure S6). Analysis of 2D NMR data
including COSY, HMBC, and HSQC determined the structure as
that of lobatamide A (Figure 3A, Figures S7-S9). Lobatamides are
+
potent vacuolar (H )-ATPase inhibitors interfering with tumor
15]
[17]
metastasis. As they were previously known only from a marine
[18]
invertebrate, the tunicate Aplidium lobatum , G. sunshinyii
represents the first culturable source for these compounds.
+
Lobatamides belong to a diverse group of benzolactone (H )-
ATPase inhibitors (Figure 3A) with as-yet unknown PKSs and
isolated from a remarkable range of organisms, including a
can predict intermediates only for modules that have
a
downstream KS, no elongation was predicted for the terminal
module. A terminal C2 unit was therefore added based on cis-AT
PKS rules. To account for the N-terminal NRPS module of LbmA,
we suspected glycine for biosynthetic initiation. This replacement
also removed an exomethylene group in the TransATor structure
that was unlikely because polyketide β-branching components[16]
were not found in the BGC. These modifications resulted in the
hypothetical structure 9 (Figure 2C) as a basis for the analytical
work.
[19]
[20]
sponge (salicylihalamides),
a
and
fungus (CJ-12,950),
Pseudomonas strain
[21]
myxobacteria (apicularens),
a
[22]
(
oximidines). Their related structures and the identification of
bacterial sources suggest that all compounds are of prokaryotic
origin. For an expansion of this family, see the study on
[23]
necroximes conducted concurrently with the work here.
Structure 5 features an internal ester that is reversed in
comparison to oocydin, pederin, and toblerol biosynthesis. To
2
This article is protected by copyright. All rights reserved.