Angewandte
Communications
Chemie
Figure 2. Selected correlations observed by 2D NMR spectroscopy.
1
1
Bold lines: H- H COSY correlations; solid arrows: HMBC correlations;
dashed arrow: nuclear Overhauser effect (NOE) correlation.
Scheme 1. A) Synthesis of pyreudiones A and B (1 and 2) and
analogues 9–12. Reagents and conditions: a. Ac O, EtOAc, RT, ultra-
2
sound; b. DIPEA, EDCI, N,O- dimethyl hydroxylamine, CH Cl , RT;
2
2
with the hydroxy group hydrogen bonded to the lactam
c. LHMDS, THF, À788C; d. EDCI, CH Cl , CH (CH ) CO H acid,
2
2
3
2
n
2
[19]
carbonyl group. The ratio between the Z-exo-enol and E-
exo-enol form of the pyreudiones was approximately 7:2.
In addition, we identified a previously reported cyclic
DMAP, RT. Overall yields for pyrrolizidine alkaloids over 4steps from
l-proline: 9: 9%; 10: 30%; 11: 2%; (S)-1: 7%; (R)-1: 7%; 2: 9%; 12:
6
%. B) Amoebicidal activities of pyreudione analogues against D.
discoideum. The asterisk indicates that these compounds were identi-
fied in P. fluorescens HKI0770 cultures. DIPEA=diisopropylethylamine;
EDCI=1-ethyl-3-(3-dimethylaminopropyl)carbodiimide;
LHMDS=lithium bis(trimethylsilyl)amide, DMAP=4-dimethylamino-
pyridine.
[
20,21]
lipopeptide (CLP 5, Figure 1).
Bacterially produced
CLPs constitute a very diverse group of secondary metabo-
lites with equally diverse, yet often poorly understood
[22]
ecological functions.
Next, we determined the amoebicidal activity of the
isolated bacterial metabolites. While CLP 5 was modestly
active, with a half maximal inhibitory concentration (IC ) of
confirms an (S)-configuration for the pyreudiones, thus
indicating a putative biosynthesis from l-proline.
5
0
À1
9
4 mgmL , pyreudiones A–D (1–4) were responsible for
most of the amoebicidal activity, with IC50 values in the range
We tested the chain-length analogues of 1 for amoebicidal
activity and found that the short-chain-length analogues 9–11
showed no detectable activity up to a concentration of
À1
of 1–11 mgmL . Furthermore, IC values from combina-
5
0
[23]
tions of pyreudione A and CLP 5 showed additive effects,
with no observable synergism in the D. discoideum growth
inhibition assay (see the Supporting Information). The main
metabolites isolated from a liquid culture of P. fluorescens
À1
100 mgmL (see Scheme 1). The compounds with longer
chain lengths (1, 2, and 12) showed considerable activity, with
À1
IC50 values of 4, 1, and 2 mgmL , respectively. Chain-length
À1
HKI0770 were, in descending order, CLP 5 (31 mgL ),
analogues with n > 8 did not display a strong increase in
activity when compared to pyreudione (2; n = 8).
À1
À1
pyreudione A (15 mgL ), and pyreudione B (4 mgL ).
Pyreudiones C and D were produced in small concentrations
Bioassay-guided fractionation enabled us to identify the
most potent amoebicidal secondary metabolites, yet we did
not know whether these metabolites actually prevent amoe-
bal predation. In order to address this question, mutant
strains with impaired CLP and pyreudione syntheses were
required, and thus, we set out to identify the respective
biosynthetic genes. We sequenced the genomic DNA of
P. fluorescens HKI0770 using Illumina next-generation
À1
of approximately 1 mgL . Since we could not identify any
congeners with either shorter chain length (n < 6; Scheme 1)
or longer chain length (n > 10), we sought to investigate the
influence of the alkyl chain length on amoebicidal activity.
From this, we expected to gain insight into why 1 and 2
constitute the major isolated metabolites.
Enantioselective synthesis of the pyreudiones also
allowed structural verification and determination of the
absolute stereochemistry of the isolated natural products.
Both (S)- and (R)-enantiomers of the main metabolite 1 were
prepared, as well as the chain-length analogues 9–12
[
35]
sequencing.
Sequence analysis of a variety of genes
confirmed the genus and species of P. fluorescens HKI0770.
[
25]
We used AntiSmash to identify a gene locus corresponding
to multimodular nonribosomal peptide synthetase
(NRPS).
a
[26,27]
(
Scheme 1).
Bioinformatic prediction of the amino acid
Starting from l-proline, N-acetylation was performed
sequence of the respective metabolite was in excellent
agreement with the amino acid sequence of CLP 5 (see the
Supporting Information).
[24]
with acetic anhydride in ethyl acetate, and the correspond-
ing Weinreb amide 7 was formed with N,O-dimethyl
hydroxylamine and EDCI as coupling reagent. LHMDS-
mediated cyclization yielded pyrrolizidine-1,3-dione core 8.
Subsequent acylation using various linear-chain fatty acids
furnished the desired pyrrolizidine alkaloids 1, 2, and 9–12.
Identification of the biosynthetic gene(s) corresponding
to the pyreudiones proved more complex since both an NRPS
and alternative enzymatic machineries were conceivable. We
were able to locate a monomodular NRPS gene of 3.9 kb
containing putative condensation (C), adenylation (A),
thiolation (T), and thioesterase (TE) domains (Figure 3,
box). The A domain was predicted to be specific for activating
a hydrophobic aliphatic amino acid (using NRPSpredic-
1
13
The H, C NMR, and HRMS spectra of synthetic
pyreudione A (1) matched those of isolated 1, as did their
HPLC retention times. Importantly, the specific optical
rotation values were in good agreement (see the Supporting
Information). Synthetic enantiomer (R)-1 displayed a specific
optical rotation with same magnitude yet opposite sign. This
[28]
tor2). This gene thus seemed likely to code for a pyreudione
synthetase. We inactivated both the putative pyreudione (pys)
2
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 5
These are not the final page numbers!