Insights into the Biosynthetic Origin of 3-(3-Furyl)alanine in Stachylidium sp. 293 K04 Tetrapeptides

Article abstract of DOI:10.1021/acs.jnatprod.6b00601

The marine-sponge-derived fungus Stachylidium sp. 293 K04 produces the N-methylated peptides endolide A (1) and endolide B (2), showing affinity for the vasopressin receptor 1A and serotonin receptor 5HT_2B, respectively. Both peptides feature the rare amino acid 3-(3-furyl)alanine. Isotope labeling experiments, employing several ¹³C-enriched precursors, revealed that this unprecedented heterocyclic amino acid moiety in endolide A (1) is synthesized from a cyclic intermediate of the shikimate pathway, but not from phenylalanine. Two new tetrapeptide analogues, endolides C and D (3 and 4), were characterized, as well as the previously described hirsutide (5).

Full text of DOI:10.1021/acs.jnatprod.6b00601

Article
pubs.acs.org/jnp
Insights into the Biosynthetic Origin of 3‑(3-Furyl)alanine in
Stachylidium sp. 293 K04 Tetrapeptides
Fayrouz El Maddah,†,‡ Stefan Kehraus,† Mamona Nazir,† Celso Almeida,†,§ and Gabriele M. Konig*,†
̈
†Institute for Pharmaceutical Biology, University of Bonn, Nussallee 6, D-53115 Bonn, Germany
‡Department of Pharmacognosy and Tissue Culture, National Organization for Drug Control and Research, Cairo, Egypt
§UCIBIO-REQUIMTE, Departamento de Cien
̂
cias da Vida, Faculdade de Cien
̂
cias e Tecnologia, Universidade Nova de Lisboa,
Caparica 2829-516, Portugal
S
* Supporting Information
ABSTRACT: The marine-sponge-derived fungus Stachylidium sp. 293 K04
produces the N-methylated peptides endolide A (1) and endolide B (2),
showing affinity for the vasopressin receptor 1A and serotonin receptor
5HT2B, respectively. Both peptides feature the rare amino acid 3-(3-
furyl)alanine. Isotope labeling experiments, employing several 13C-enriched
precursors, revealed that this unprecedented heterocyclic amino acid moiety
in endolide A (1) is synthesized from a cyclic intermediate of the shikimate
pathway, but not from phenylalanine. Two new tetrapeptide analogues,
endolides C and D (3 and 4), were characterized, as well as the previously
described hirsutide (5).
he marine sponge-derived fungus Stachylidium sp. 293
T
K04 produces a plethora of secondary metabolite classes
including peptides.1−3 In a former study we characterized the
novel cyclic N-methylated peptides endolides A and B (1 and
2), which showed affinity for the vasopressin receptor 1A and
serotonin receptor 5HT2B, respectively.4
The most intriguing feature of these supposedly non-
ribosomal peptides is the incorporation of the nonproteino-
genic amino acid N-methyl-3-(3-furyl)alanine, to date only
reported in the heptapeptides rhizonins A and B and the
pentapeptide bingchamide B.5,6 Rhizonin A was initially
considered as a mycotoxin metabolite isolated from the fungus
Rhizopus microsporus, but was later found to be produced by the
bacterial endosymbiont Burkholderia endofungorum.7 Bingcha-
mide B was isolated from a soil-dwelling strain of Streptomyces
bingchenggensis.
Due to the novel nature of N-methyl-3-(3-furyl)alanine, the
current study set out to investigate its metabolic origin,
employing classical isotope tracer experiments. During the
course of this study, we obtained, apart from 1, three further
cyclic N-methylated peptide analogues, two of which (3, 4)
were structurally unprecedented, whereas hirsutide (5) was
already described.8 A further observation during our studies was
the production of two thiodiketopiperazines, upon adding
phenylalanine to a liquid culture of Stachylidium sp. 293 K04.
These are bisdethiobis(methylthio)acetylaranotin (BDA) and
bisdethiobis(methylthio)acetylapoaranotin (BDAA), belonging
to the aranotin group of metabolites.9
RESULTS AND DISCUSSION
■
For studying the biosynthesis of the unusual N-methyl-3-(3-
furyl)alanine moiety in Stachylidium 293 K04 peptides using
isotope-labeled precursors, the major metabolite endolide A (1)
was targeted. First experiments were initiated to establish an
isotope incorporation protocol. Because the peptides are not
produced in liquid media, solid biomalt sea-salt-containing agar
media were essential for the investigation. Applying LC-MS,
production of endolide A (1) was detectable around 10 days
Here we report the investigation of the biosynthetic origin of
the N-methyl-3-(3-furyl)alanine moiety of endolide A (1) and
the description of the new structural analogues 3 and 4.
Received: June 29, 2016
© XXXX American Chemical Society and
American Society of Pharmacognosy
A
DOI: 10.1021/acs.jnatprod.6b00601
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Table 1. Results of 13C-Labeling Experiments for Endolide A
(1)
after inoculation onward and up to day 55 (Table S1,
Supporting Information). From this time scale experiment,
we concluded that the labeled precursors should be added
twice, i.e., on day 5 and 10 with a total cultivation time of 30
days (see Experimental Section and Table S1).
The first hypothesis on the biosynthesis of 3-(3-furyl)alanine
was based on its structural relationship to the phenylalanine
skeleton. Indeed, the oxidative cleavage of a benzene ring as
required in this case would not be unusual for fungi. Prominent
examples of this are seen in the mycotoxin patulin,10 the
betalain pigments,11 and the pyrrolobenzodiazepine antitumor
antibiotics.12 Also, the metabolites BDA and BDAA (Figure S1,
Supporting Information), isolated from Stachylidium sp. 293
K04 in this study, both have an oxidatively cleaved aromatic
ring, pointing toward such enzymatic capabilities in the
fungus.13
enriched carbon atoms in endolide A
N-Me-FurAla
N-Me-FurAla
no. fed precursor
(FurAla A)
Val
n.e.
(FurAla B)
Leu
a
1
L-[1-13C]
n.e.
n.e.
n.e.
phenylalanine
2
[U-13C]
glycerol
C-1/C-2/C-
3,
C-9/C- C-14/C-15/
C-22/C-
23,
10,
C-16,
C-5/C-4/C-
7,
C-11/
C-12,
C-18/C-17/
C-20,
C-25/C-
27,
C-6
C-13
C-19
C-24
3
4
5
D-[1-13C]
glucose
C-3, C-5
C-12,
C-13
C-16, C-18
C-23, C-
26, C-27
[1-13C]sodium C-1, C-6, C-7 n.e.
acetate
C-14, C-19,
C-20
C-22
L-[Me-13C]
methionine
C-8
n.e.
C-21
n.e
Consequently, a labeling experiment with L-[1-13C]-
phenylalanine was initiated. Incorporation of L-[1-13C]-
phenylalanine was monitored using LC-MS and further verified
after isolation of endolide A (1) and other peptides and their
13C NMR spectroscopic analysis. No shift in the molecular
mass for endolide A (1) was observed in the LC-MS spectra as
compared to the control without labeled phenylalanine (Figure
S2). Additionally, no 13C-enriched signals were observed in the
13C NMR spectrum of isolated endolide A (1) from the same
labeling experiment (Figure S3). Validity of this experiment was
confirmed by the simultaneous isolation of hirsutide (5), a
cyclic peptide previously reported from the entomopathogenic
fungus Hirsutella sp.8 Hirsutide (5) was found to incorporate
three L-[1-13C]phenylalanine units, as evident from three
enhanced 13C NMR signals for the carbonyl carbons C-1 (δC
170.1), C-16 (δC 171.0), and C-26 (δC 172.1) of the
phenylalanine residues (Figure S4). Along with that, a 3 Da
shift of the molecular mass of hirsutide (5) was observed in the
LC-MS spectrum of the extract of the Stachylidium sp. 293 K04
culture supplemented with L-[1-13C]phenylalanine; that is, 13C-
labeled hirsutide (5) showed a signal at m/z 572 [M + H + 3]+
as compared to the unlabeled control with m/z 569 [M + H]+
(Figure S2). Thus, in this experiment it was clearly
demonstrated that phenylalanine is not the precursor for 3-
(3-furyl)alanine.
In a further set of experiments, general carbon source
precursors such as 13C-labeled glycerol or glucose, which are
involved in several biosynthetic pathways, were employed.
These building blocks or metabolites derived thereof give
specific labeling patterns upon incorporation and help in
delineating the involved biosynthetic pathway. Incubation with
fully labeled [U-13C]glycerol gave a complex 13C NMR
spectrum for the isolated endolide A (1) (Figure S5). Detailed
analysis of the signal pattern and 13C−13C coupling constants in
the inverse gated proton decoupled 13C NMR spectrum of
endolide A (1) showed the presence of a coupled assembly of
three carbon atoms, i.e., C-1/C-2/C-3, evidenced by a 52.5 Hz
coupling between C-1 and C-2 and a 38.4 Hz coupling between
C-2 and C-3. Another coupled three-carbon assembly was
observed for C-5/C-4/C-7, as evidenced by a 49.5 Hz coupling
between C-4 and C-5 and a 72.8 Hz coupling between C-4 and
C-7. A single enriched noncoupled 13C signal was detected for
C-6 of the first 3-(3-furyl)alanine moiety (FurAla A, Table 1).
This labeling pattern was mirrored in the second 3-(3-
furyl)alanine moiety (FurAla B) as shown in Table 1 (for
detailed chemical shifts and coupling constants of [U-13C]
glycerol-derived endolide A (1) see also Table S4).
a
n.e.: not enriched.
It is well established that during glycolysis [U-13C]glycerol
gives rise to an intact, 13C-coupled three-carbon unit in the
form of phosphoenolpyruvate.14 Concurrently, [U-13C]glycerol
is converted in the reductive pentose phosphate cycle, via
dihydroxyacetone phosphate and phosphoglyceraldehyde, into
fructose-6-phosphate, containing two 13C-coupled three-carbon
units, which with the help of a transketolase gives erythrose-4-
phosphate with one intact 13C-coupled three-carbon unit and a
single noncoupled carbon (Figure S6).15 In the shikimate
pathway, phosphoenolpyruvate and erythrose-4-phosphate are
the two principle building blocks, the 13C labeling of which can
also be found in any further metabolites derived thereof (Figure
1). Thus, one observed three carbon atom coupled assembly,
i.e., C-5/C-4/C-7 (FurAla A), forming one part of the furan
ring, could arise from phosphoenolpyruvate, while the second
three carbon atom coupled assembly, i.e., C-1/C-2/C-3 (FurAla
A), forming the amino acid backbone, along with the single
enriched noncoupled C-6 (FurAla A), incorporated also in the
furan ring, could arise from erythrose-4-phosphate (Figure S6).
Interestingly, the labeling observed for the 3-(3-furyl)alanine
moieties in endolide A (1) after adding [U-13C]glycerol
resembles one of two patterns observed for the dihydrofur-
anacrylic acid moiety of reductiomycin produced by Strepto-
myces xanthochromogenus, following similar biosynthetic studies
using [U-13C]glycerol.16,17 In this context, it is important to
note that dihydrofuranacrylic acid shares the basic carbon
skeleton with 3-(3-furyl)alanine. Analogous to reductiomycin,
we thus conclude that this labeling pattern could be explained
only through cleavage of a cyclic structure formed in the
shikimate pathway. However, because we observe only one
distinct labeling pattern, the intermediate in the case of 3-(3-
furyl)alanine must not be symmetrical (Figure S7).
In the labeling study using D-[1-13C]glucose, we observed
enriched 13C signals for C-3 (δC 24.5) and C-5 (δC 111.0) of
FurAla A and C-16 (δC 24.5) and C-18 (δC 110.9) of FurAla B
of endolide A (1) (Figure S8). In this way, the involvement of
the shikimate pathway was further supported, because it was
indeed expected that two carbons should be 13C enriched.
These originate from [3-13C]phosphoenolpyruvate via glycol-
ysis and [4-13C]erythrose-4-phosphate via the pentose
phosphate pathway, following conversion of [1-13C]glucose to
[6-13C]glucose-6-phosphate or [1,6-13C]glucose-6-phosphate
via glycolysis and gluconeogenesis (Figure S9).18−20
B
DOI: 10.1021/acs.jnatprod.6b00601
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Figure 1. Summary of 13C enrichment patterns in N-methyl-3-(3-furyl)alanine from the biomalt salt agar culture of Stachylidium sp. 293 K04 labeled
with [U-13C]glycerol, D-[1-13C]glucose, and L-[Me-13C]methionine in separate experiments. Bold lines indicate 13C-labeled isotopomers with directly
adjacent 13C atoms from [U-13C]glycerol. Black dots indicate 13C-enriched atoms from [U-13C]glycerol. Red dots indicate 13C-enriched atoms from
D-[1-13C]glucose. Asterisk indicates 13C-enriched atoms from L-[Me-13C]methionine. The involvement of phenylalanine as a precursor is excluded.
13C-Enrichment after incorporating [1-13C]sodium acetate
can also be explained by the involvement of phosphoenolpyr-
uvate and erythrose-4-phosphate in the biosynthesis of 3-(3-
furyl)alanine (Figure S10). After metabolism in the tricarbox-
ylic acid cycle, [1-13C]sodium acetate enters gluconeogenesis as
[1,4-13C]oxaloacetate.21 A carboxykinase catalyzes the trans-
formation of oxaloacetate into [1-13C]phosphoenolpyruvate,22
which labels C-7 (δC 141.0) of FurAla A and C-20 (δC 141.1)
of FurAla B, respectively. [1-13C]Phosphoenolpyruvate is
further metabolized to [1,2-13C]erythrose-4-phosphate as
already described above (Figure S11).15 This leads to the
labeling of C-1 (δC 170.3) and C-6 (δC 144.5) for FurAla A and
C-14 (δC 170.7) and C-19 (δC 144.4) of FurAla B.
Finally, for elucidating the biosynthetic origin of the N-
methyl substituent in 3-(3-furyl)alanine, we added L-[Me-13C]-
methionine, a universal donor of methyl groups frequently used
in biosynthetic studies.23 This resulted in high 13C-enrichments
for C-8 (δC 30.3) and C-21 (δC 30.5) of endolide A (1) (Figure
S12).
Based on these results (Table 1), a scheme for the
biosynthesis of N-methyl-3-(3-furyl)alanine is proposed for
the first time (Figure 1). This involves both phosphoenolpyr-
uvate and erythrose-4-phosphate as precursor molecules and
evidences that the shikimate pathway is the biosynthetic route
for the 3-(3-furyl)alanine moiety in endolide A (1), with the N-
methyl group being provided from methionine. The labeling
pattern also requires that a cyclic intermediate of the shikimate
pathway is involved in the biosynthesis.
In the LC-MS analyses performed during our labeling
experiments we noticed that further new cyclic peptide
analogues, endolides C and D (3 and 4), as well as the
known tetrapeptide 5 were produced by the fungus on solid
biomalt salt medium supplemented with phenylalanine.
1
The H and 13C NMR spectra of the cyclic N-methylated
peptide analogues 3 and 4 were quite similar to those of the
previously reported endolides A and B (1 and 2).4 They
1
showed four H NMR resonances at δH 4−5 characteristic for
α-protons of amino acid residues, along with two singlets at
around δH 2.8 for N-methyl protons and two characteristic
broad doublets at around δH 7.8 indicative of amide protons. In
addition 13C NMR spectra included four signals at δC 170−173
for four carbonyls, with α carbon signals at δC 50−70. All of
these data resulted in the assignment of peptide-like molecules,
composed of four amino acid residues, two of which are N-
methylated (Table 2).
Endolide C (3) was assigned a molecular formula of
C32H38N4O5 using HRESIMS, indicating 16 degrees of
unsaturation. The 13C NMR and DEPT135 spectra revealed
32 resonances, resulting from four methyl groups, three
methylenes, five sp3 methines, 13 sp2 methines, and seven
nonprotonated carbons. The COSY spectrum revealed
correlations of two closely related spin systems, each consisting
of five aromatic methine protons, one spanning from H-5
through to H-9 and a second from H-20 through to H-24.
HMBC correlations of H2-3 (δH 3.61 and 2.92) to the aromatic
carbon C-4 (δC 138.9) as well as to C-2 (δC 63.5) established
the first aromatic residue as phenylalanine (Phe A). The second
C
DOI: 10.1021/acs.jnatprod.6b00601
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
1
Table 2. H and 13C NMR Spectroscopic Data for Endolides C and D (3 and 4)
3
4
a
a
a
a
amino acid
pos
δC, mult.
δH (mult., J in Hz)
amino acid
pos
δC, mult.
169.8, C
δH (mult., J in Hz)
N-Me-L-Phe
1
2
3
170.4, C
N-Me-
1
2
3
(A)
63.5, CH
34.7, CH2
4.50, dd (3.3, 11.7)
a 3.61, dd (3.3, 15.0)
b 2.92, dd (11.7,15.0)
L-FurAla (A)
62.5, CH
24.3, CH2
4.25, dd (3.3, 11.7)
a 3.22, dd (3.3, 15.6)
b 2.73, m
4
5
6
7
8
9
138.9, C
4
5
6
7
8
121.9, C
129.2, CH
129.5, CH
127.5, CH
129.5, CH
129.2, CH
30.5, CH3
171.8, C
7.27, m
7.30, m
7.23, m
7.30, m
7.27, m
2.72, s
110.9, CH
144.4, CH
140.9, CH
30.5, CH3
6.31, dd (0.9, 1.7)
7.50, t (1.7)
7.38, br s
2.71, s
10
L-Val
11
L-Val
9
172.1, C
12
56.2, CH
30.4, CH
20.7, CH3
18.4, CH3
4.26, dd (7.7, 9.2)
2.04, m
10
56.2, CH
30.4, CH
20.8, CH3
18.4, CH3
4.36, m
13
11
2.11, m
14
0.68, d (6.6)
0.82, d (7.0)
7.79, d (9.2)
12
0.82, d (6.6)
0.84, d (6.6)
7.77, d (9.5)
15
13
NH-12
16
NH-10
14
N-Me-L-Phe
171.0, C
63.6, CH
35.0, CH2
N-Me-
170.7, C
62.7, CH
24.7, CH2
(B)
17
4.55, dd (3.3, 11.7)
a 3.67, dd (3.3, 15.0)
b 3.02, dd (11.7, 15.0)
L-FurAla (B)
15
4.34, m
18
16
a 3.36, dd (3.3, 15.6)
b 2.95, dd (11.3, 15.6)
19
20
21
22
23
24
25
26
27
28
138.9, C
17
18
19
20
21
122.0, C
129.1, CH
129.5, CH
127.5, CH
129.5, CH
129.1, CH
30.9, CH3
172.2, C
7.27, m
7.30, m
7.23, m
7.30, m
7.27, m
2.79, s
111.0, CH
144.4, CH
141.0, CH
30.6, CH3
6.30, dd (0.9, 1.7)
7.44, t (1.7)
7.37, br s
2.78, s
L-FurAla
L-Phe
22
23
24
172.5, C
52.2, CH
38.5, CH2
51.4, CH
28.0, CH2
4.78, dt (9.5, 6.6)
4.96, m
a 2.90, dd (6.6, 15.0)
b 2.52, dd (6.6, 15.0)
a 3.20, dd (6.8, 13.9)
b 2.72, m
29
121.8, C
25
139.0, C
30
112.7, CH
143.2, CH
141.1, CH
6.20, br s
26
130.4, CH
128.7, CH
126.9, CH
128.7, CH
130.4, CH
7.18, m
31
7.38, t (1.8)
7.14, br s
27
7.24, m
32
28
7.17, m
NH-27
7.93, d (9.5)
29
7.24, m
30
7.18, m
NH-23
7.93, d (9.5)
a
Acetone-d6, 300/75.5 MHz for peptide 3 and 600/150 MHz for peptide 4
aromatic ring showed similar correlations and hence led to the
conclusion that the second aromatic amino acid is also
(Phe B), H-17 (PheB) to C-26 (FurAla), and H-27 (FurAla) to
C-1 (Phe A). The N-methyl substituents were positioned on
the relevant amino acid residues Phe A and Phe B, based on
HMBC correlations with their adjacent carbonyl carbons and
α-protons, that is, H3-10 to C-11 and C-2 and H3-25 to C-26
and C-16. This led to the peptide sequence, as cyclo-(N-
methyl-phenylalanyl, valyl, N-methyl-phenylalanyl, furylalanyl).
The 13C NMR and DEPT135 spectra of endolide D (4)
revealed 30 resonances, resulting from four methyls, three
methylenes, five sp3 methines, 11 sp2 methines, and seven
nonprotonated carbons. The 1H and 13C NMR spectra
displayed characteristic proton and carbon resonances for two
magnetically equivalent 3-(3-furyl)alanine moieties, which were
N-methylated. The remaining proton and carbon resonances in
the aromatic region were indicative of a phenylalanine residue.
COSY correlations between NH-23 (δH 7.93) and H-23 (δH
4.96) proved that the phenylalanine residue was not N-
methylated. The last residue was ascribed as valine based on
1
phenylalanine (Phe B). A further H−1H spin system was
assigned to a valine, i.e., using correlations from the methyl
protons H3-14 and H3-15 to NH-12 (δH 7.79), through H-13
and H-12, supported by HMBC correlations from H-13 and H-
12 to C-11. The last residue displayed carbon resonances
characteristic for a 3-substituted furyl moiety, i.e., δC 143.2,
141.1, 121.8 and 112.7. This is confirmed with HMBC
correlations from H-30 (δH 6.20) and H-32 (δH 7.14) to the
furyl carbon C-29 (δC 121.8). COSY correlations evidenced a
spin system from H-27 (δH 4.78) to NH-27 (δ 7.93). HMBC
correlations of H-27 and H-28 with the C-26 (δC 172.2)
finalized the assignment of the last residue as 3-(3-furyl)alanine.
The sequential relationship of the four amino acid residues was
deduced from HMBC correlations of the α-protons of the
amino acid residues and the carbonyl carbon of the adjacent
amino acid, i.e., H-2 (Phe A) to C-11 (Val), H-12 (Val) to C-16
D
DOI: 10.1021/acs.jnatprod.6b00601
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Figure 2. Proposed formation of 3-(3-furyl)alanine from a cyclic intermediate of the shikimate pathway. The last step is analogous to the
bioconversion of cinnamic acid to phenylalanine.34 The exact intermediate and sequence of steps are currently not known.
COSY and HMBC correlations. COSY correlations between
NH-10 (δH 7.77) and H-10 (δH 4.36) showed that the valine
residue was also not N-methylated. The sequential relationship
of the four amino acid residues was deduced from HMBC
correlations of the α-protons of the amino acid residues and the
carbonyl carbon of the adjacent amino acid, i.e., H-2 (FurAla A)
to C-9 (Val), H-10 (Val) to C-14 (FurAla B), H-15 (FurAla B)
to C-22 (Phe), and H-23 (Phe) to C-1 (FurAla A), and was
supplemented with HMBC correlations of the N-methyl
substituents. Thus, endolide D (4) was assigned as cyclo-[N-
methyl-3-(3-furyl)-alanyl, valyl, N-methyl-3-(3-furyl)-alanyl,
phenylalanyl].
in this fungus but previously reported from the fungi
Arachniotus aureus and Aspergillus terreus, were isolated.9,13 It
has been previously established that the biosynthesis of BDA
and BDAA involves a phenylalanine intermediate.25 Through
addition of phenylalanine to the liquid cultivation media,
biosynthesis of these metabolites was stimulated.26,27
The Stachylidium sp. 293 K04 is a biochemically versatile
fungus, as evident from the secondary metabolite classes
(polyketides, phenylpropanoids, peptides) obtained from it so
far.1−4 The exogenous supply of phenylalanine led to the
isolation of three additional peptide analogues (3−5) differing
in terms of the incorporated amino acids (phenylalanine, 3-(3-
furyl)alanine, and valine), their sequence, and the position of
N-methylation. This is probably attributed to an increased
supply of the precursor amino acid phenylalanine, as seen in
precursor-directed biosynthetic studies.27,28 Nevertheless these
peptide analogues were also detected in the LC-MS analyses of
the time scale studies. Furthermore, production of endolide A
(1) was negatively affected in incorporation experiments
employing phenylalanine (Figure S2, Table S2), presumably
as a result of feedback inhibition where both the primary and
secondary metabolites are derived from the same biosynthetic
pathway28 or as a result of an increased pool size of
phenylalanine leading to the abundant production of peptides
3, 4, and 5. Overall, this indicates a biosynthetic relationship
between the here isolated peptides.
The peptides isolated so far from the Stachylidium sp. 293
K04 are proposed to be synthesized by the nonribosomal
peptide biosynthetic machinery. This is supported by their
structural architecture incorporating unusual amino acid motifs
with additional N-methylation and cyclization, as frequently
encountered in other nonribosomal peptides, e.g., cyclo-
sporins.29 Contradictory to the number of isolated peptides
(1−5), a single nonribosomal peptide synthetase (NRPS) is
High-resolution ESIMS measurements along with further
spectroscopic data and literature searches, assigned peptide 5 as
hirsutide (Table S5 and Figures S19, S20). The configuration of
5 is identical to that reported, as the specific rotations are
comparable (reported specific rotation [α]20 −192 (c 0.2,
D
CH2Cl2), isolated compound: [α]23D −122 (c 0.125, CH2Cl2)).
The absolute configurations of the amino acids in the
isolated peptides 3 and 4 (as well as 5 for comparison) were
assigned after acidic hydrolysis followed by derivatization of the
amino acids with Marfey’s reagent and HPLC analyses
(Advanced Marfey’s method).24 On the basis of comparisons
of retention times to standard amino acids we could assign the
L-configuration to the amino acids valine and phenylalanine
(Figures S21, S22) for all three compounds. For the 3-(3-
furyl)alanine moiety, which decomposes upon acid treatment
and for which no standard is available, an L-configuration is
implied, based on biogenetic considerations. The recently
determined X-ray crystal structure of endolide A (1) clearly
established the L-configuration of 3-(3-furyl)alanine.4
Additionally, from a liquid biomalt salt medium supple-
mented with phenylalanine, the thiodiketopiperazines BDA and
BDAA (Figures S23−S26, Table S6), hitherto not encountered
E
DOI: 10.1021/acs.jnatprod.6b00601
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
believed to be involved in their biosynthesis. Promiscuity of the
NRPS to alternative amino acid residues is generally
encountered, rationalized by the microheterogeneity of the
involved adenylation domains accepting different building
blocks during peptide synthesis.30 This was previously observed
for structurally related peptides from fungal species, such as 25
naturally occurring cyclosporins14 and as many as 29 isolated
enniatins.31 Four of these Stachylidium sp. 293 K04 peptides
(1−4) contain the rare, nonproteinogenic amino acid N-
methyl-3-(3-furyl)alanine. From a series of isotope labeling
experiments, we outlined the biosynthesis of this rare building
block in secondary metabolite peptides, by analyzing the 13C
labeling pattern in endolide A (1). Incubation with 13C-labeled
glycerol was most decisive and resulted in a labeling signature
of two assemblies of three coupled 13C atoms, with distinct
coupling constants, in addition to a single enriched but not
coupled carbon atom, in the 3-(3-furyl)alanine moiety. This is
clearly implicative of a biosynthesis via the shikimate pathway
from phosphoenolpyruvate (PEP) and erythrose-4-phosphate
(E-4-P).
It should be noted that cyclization of 3-deoxy-D-arabino-
heptulosonate-7-phosphate (DAHP), the condensation product
of the precursor molecules PEP and E-4-P, could not be
envisioned as a direct source for the 3-(3-furyl)alanine moiety,
as deduced from the observed labeling pattern for 3-(3-
furyl)alanine. Instead, it is proposed that the ring cleavage of a
later stage cyclic intermediate such as shikimic acid or a
derivative thereof, e.g., 3-dehydroshikimic acid, 3-hydroxyben-
zoic acid, or 3,4-dihydroxybenzoic acid, and subsequent
cyclization would lead to the formation of the furan moiety,
as disclosed from biosynthetic studies for the dihydrofurana-
crylic moiety of reductiomycin.17 In the latter case two labeling
patterns were observed indicative of a symmetric intermediate.
In our study, however, the involvement of such a symmetrical
intermediate is ruled out, as we observe only one distinct
pattern. In the case of reductiomycin, 4-hydroxybenzoic acid
undergoes two plausible ring cleavage reactions leading to the
two labeling patterns observed for the dihydrofuranacrylic acid
moiety.17 Finally, for 3-(3-furyl)alanine the introduction of an
amino group, probably from glutamic acid catalyzed by a
transaminase, would lead to the basic skeleton of 3-(3-
furyl)alanine.14 The methyl group is provided from S-
adenosyl-L-methionine, with the help of a methyltransferase,23
presumably during peptide assembly as seen in cyclosporine
biosynthesis.32
bingchamide. No experimental proof is, however, provided
for this.33
As to date no bacteria could be detected in Stachylidium sp.
293 K04, the endolides are thus considered the first fungal
peptides harboring a 3-(3-furyl)alanine moiety. Hence, the here
proposed biosynthetic mechanism for the 3-(3-furyl)alanine
supports only fungal biosynthesis, and yet a different pathway
could be operative for the synthesis of this moiety in those
peptides of bacterial origin, i.e., rhizonins A and B and
bingchamide B.
Thus, the information gained from the here applied labeling
studies provide us with important data necessary for further
biosynthetic investigations, i.e., to determine the exact
shikimate intermediate for 3-(3-furyl)alanine biosynthesis.
Genomic approaches, such as the currently ongoing sequencing
of the Stachylidium sp. 293 K04 genome, will result in a detailed
view of the assembly of the amino acid building blocks for the
respective nonribosomal peptides and may also explain the
structural diversity observed for the latter.
EXPERIMENTAL SECTION
■
General Experimental Procedures. Optical rotations were
measured with a Jasco DIP 140 polarimeter. UV and IR spectra
were obtained employing PerkinElmer Lambda 40 and PerkinElmer
Spectrum BX instruments, respectively. All NMR spectra were
recorded in acetone-d6 or methanol-d4 referenced to residual solvent
signals with resonances at δH/C 2.04/29.8 and δH/C 3.35/49.0,
respectively, using either a Bruker Avance 300 DPX spectrometer
operating at 300 MHz (1H) and 75 MHz (13C) or a Bruker Ascend
600 spectrometer operating at 600 MHz (1H) and 150 MHz (13C).
LCESIMS measurements were conducted employing an Agilent 1100
Series HPLC including DAD, which was coupled with an API 2000,
Triple Quadrupole LC/MS/MS, Applied Biosystems/MDS Sciex, and
an ESI source. HRESIMS measurements were conducted using a
Bruker Daltonics micrOTOF-Q mass spectrometer with an ESI
source. HPLC was performed on (a) an HPLC system composed of a
Waters 515 pump with a Knauer K-2300 differential refractometer,
using a Knauer column (250 × 8 mm, 5 μm, Eurospher II-100 Si), and
(b) a Merck-Hitachi system equipped with an L-6200A pump, an L-
4500A photodiode array detector, and a D-6000 interface, using
Macherey-Nagel columns (Nucleodur C18 EC Isis with 250 × 4.6 mm,
5 μm).
Fungal Material. The marine-derived fungus Stachylidium sp. 293
K04 was isolated from the sponge Callyspongia sp. cf. C. flammea,
collected from the coral reef in Bare Island, New South Wales,
Australia. The fungal strain was identified through the Belgian
coordinated collections of microorganisms of the Catholic University
of Louvain (BCCM/MUCL) by Dr. P. Massart and Dr. C. Decock. A
fungal specimen is deposited at the Institute for Pharmaceutical
Biology, University of Bonn, isolation number 293 K04.
Cultivation. Solid and liquid biomalt salt (BMS) media were used
for preparation of precultures and cultures. Biomalt salt medium: 20 g
L−1 biomalt extract, 15 g L−1 agar (for solid medium), and 1 L of
artificial seawater: 0.1 g L−1 KBr, 23.48 g L−1 NaCl, 10.61 g L−1 MgCl·
6H2O, 1.47 g L−1 CaCl2·2H2O, 0.66 g L−1 KCl, 0.04 g L−1 SrCl2·
6H2O, 3.92 g L−1 Na2SO4, 0.19 g L−1 NaHCO3, and 0.03 g L−1
H3BO3. For the first preculture, the fungal strain was inoculated on
BMS agar Petri dishes and incubated at 25 °C for 4 weeks. A seed
inoculum from this solid fungal preculture was then used to inoculate
1 L Erlenmeyer flasks (3 flasks), each containing 300 mL of liquid
BMS media, which were then shaken at 121 rpm at 25 °C for 10 days
(liquid fungal preculture), which was later used to inoculate solid and
liquid BMS media.
Moreover, our results exclude other metabolic pathways,
such as a polyketide origin, a mixed acetate-glycerol pathway, or
terpenoid biosynthesis being involved in the formation of 3-(3-
furyl)alanine (Figures S27, S28). Therefore, the shikimate
pathway accounts for all the carbon atoms of the N-methyl-3-
(3-furyl)alanine, apart from the N-methyl group, which is
provided by the methyl group of methionine (Figure 1). The
exact nature of the cyclic intermediate that serves as a precursor
for 3-(3-furyl)alanine is currently still elusive (Figure 2).
Concerning the only other known peptides with a 3-(3-
furyl)alanine substructure, i.e., rhizonins A and B and
bingchamide B, hardly any information is available on their
biosynthesis. Genome sequencing of the bingchamide B
producing S. bingchenggensis revealed a biosynthetic gene
cluster containing four genes encoding for nonribosomal
peptide synthetases with adenylation domain substrate
specificities supposedly correlating with the structure of
For the time scale experiment, 1 L of BMS agar medium was
prepared and transferred to 20 Petri dishes (30 mL each). From the
liquid fungal preculture, 0.5 mL of a liquid seed inoculum was used to
inoculate the Petri dishes. Petri dishes were incubated at room
F
DOI: 10.1021/acs.jnatprod.6b00601
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
temperature for 60 days and at defined time intervals (days 5, 10, 15,
20, 30, and 55), and representative Petri dishes were extracted (with
EtOAc) and analyzed using LC-MS. LC-MS measurements were
performed with the following conditions: RP C18 column (Macherey-
Nagel Nucleodur 100, 125 × 2 mm, 5 μm) and a 2 mmol NH4Ac
buffered MeOH/H2O gradient elution system (flow rate 0.25 mL
min−1), starting from 10% to 100% MeOH over a 20 min period, then
isocratic for 10 min. Endolide A (1) (m/z 515.5 [M + H]+)
production was monitored at retention time 19−20 min. Results are
summarized in Table S1.
For labeling studies, the fungus was cultivated on solid BMS
medium in Fernbach flasks (250 mL each) using a liquid seed
inoculum from the liquid fungal preculture (5 mL for each Fernbach
flask). Sterile aqueous solutions of labeled precursors (5 mL per
feeding per flask) were added twice on day 5 and 10 during a 30-day
cultivation period and incubated at room temperature. See Table S2
for concentrations and conditions of labeled precursors. Stable isotope
labeled compounds L-[1-13C, 99%]phenylalanine, [U-13C, 99%]-
glycerol, D-[1-13C, 99%]glucose, [1-13C, 99%]sodium acetate, and L-
[Me-13C, 99%]methionine were obtained from Cambridge Isotope
Laboratories.
For preliminary studies on solid media, 1.5 L of BMS medium was
prepared and transferred to six Fernbach flasks (250 mL each). From
the liquid fungal preculture, a liquid seed inoculum (5 mL) was used
to inoculate each flask. Sterile aqueous solution of D/L-phenylalanine
(1 mg mL−1) was added to each flask after 3 days. Cultures were
incubated at room temperature for 30 days. For preliminary studies in
liquid media, 3 L of BMS medium was prepared and transferred to
three 5 L Erlenmeyer flasks (1 L medium each). From the liquid fungal
preculture a liquid seed inoculum (10 mL) was used to inoculate each
flask. A sterile aqueous solution of D/L-phenylalanine (1 mg mL−1) was
added to each flask after 3 days. Liquid cultures were shaken at 121
rpm at room temperature for 15 days.
Extraction and Isolation. a. Endolide A (1) for Labeling Studies.
At the end of the 30-day cultivation period, the homogenized fungal
biomass and cultivation media were exhaustively extracted with EtOAc
and concentrated under vacuum (using a vacuum rotatory evaporator,
40 °C) to yield an organic extract. Vacuum liquid chromatography
(VLC) was then used for fractionation of the extract using Merck silica
gel 60 M (0.040−0.063 mm, 230−400 mesh size) as sorbent. A
gradient solvent system of increasing polarity was used for sample
elution, starting with 100% petroleum ether to 100% CH2Cl2 to 100%
EtOAc to 100% acetone to 100% MeOH, yielding nine fractions. VLC
fraction 4 (50% CH2Cl2/50% EtOAc) was then subjected to NP-
HPLC fractionation (Knauer column; 250 × 8 mm, 5 μm, Eurospher
II-100 Si; mobile phase: petroleum ether/acetone (3:1); flow rate 2
mL min−1) to yield 10 subfractions. Subfraction 4.6 was further
purified using RP-HPLC (Macherey-Nagel column; Nucleodur C18
EC Isis, 250 × 4.6 mm, 5 μm; mobile phase: 65% MeOH/H2O; flow
rate 1 mL min−1). Endolide A (1) was collected at retention time (tR)
15 min. For yields see Table S2.
collected at tR 20 min, yield 1.4 mg L−1. Hirsutide (5) was isolated
from subfraction 4.7 using RP-HPLC (Macherey-Nagel column;
Nucleodur C18 EC Isis, 250 × 4.6 mm, 5 μm; mobile phase: 63%
MeOH/H2O; flow rate 1 mL min−1) and collected at tR 20 min, yield
1.0 mg L−1.
Endolide C (3): white solid, [α]23 −130 (c 0.2, MeOH); UV
D
(MeOH) λmax (log ε) 204 (3.45), 266 (2.05) nm; IR (ATR) νmax 3330
1
(br), 2961, 1704, 1660, 1514, 1362, 1091 cm−1; H NMR and 13C
NMR (Table 2); LRESIMS m/z 559.4 [M + H]+, m/z 557.6 [M −
H]−; HRESIMS m/z 581.2731 [M + Na]+ (calcd for C32H38N4NaO5,
581.2740).
Endolide D (4): white solid, [α]23 −76 (c 0.058, MeOH); UV
D
(MeOH) λmax (log ε) 204 (3.62), 268 (3.16) nm; IR (ATR) νmax 3331
1
(br), 2926, 1660, 1505, 1386, 1090 cm−1; H NMR and 13C NMR
(Table 2); LRESIMS m/z 549.4 [M + H]+, m/z 547.5 [M − H]−;
HRESIMS m/z 571.2532 [M + Na]+ (calcd for C30H36N4NaO6,
571.2533).
Hirsutide (5): white solid, [α]23 −153 (c 0.125, MeOH), [α]23
D
D
−122 (c 0.125, CH2Cl2); IR (ATR) νmax 3329 (br), 2961, 1703, 1659,
1512, 1362, 1090 cm−1; 1H NMR and 13C NMR (Table S5);
HRESIMS m/z 569.3119 [M + H]+ (calcd for C34H41N4O4,
569.3128). Data are in accordance with those in the literature.8
Advanced Marfey’s Method. Peptides 3, 4, and 5 (0.5 mg each)
were separately dissolved in 6 N HCl (0.5 mL) and heated at 110 °C
for 16 h in closed glass vials. After cooling, the solvent was removed
using a nitrogen stream and redissolved in 50 μL of H2O. The peptide
hydrolysate and 1 mg of each respective standard L- and D-amino acid
were resuspended in 50 μL of H2O, and 100 μL of 1% (w/v) L-FDAA
(in acetone) and 40 μL of 1 M NaHCO3 were added. The mixture was
heated at 70 °C for 40 min. After cooling to room temperature, the
reaction was quenched by adding 20 μL of 2 N HCl, and solvents were
evaporated to dryness. Samples were resuspended in MeOH (1 mg
mL−1) for HPLC-MS analyses. The retention times for FDAA
derivatives of standard L-valine, D-valine, L-phenylalanine, and D-
phenylalanine were 14.4, 18.2, 16.8, and 19.0 min, respectively.
Accordingly, the amino acids were assigned in peptide 3 as L-Val (15.0
min) and L-Phe (17.0 min), in peptide 4 as L-Val (14.5 min) and L-Phe
(17.0 min), and in peptide 5 as L-Val (14.6 min) and L-Phe (17.1 min).
ASSOCIATED CONTENT
* Supporting Information
■
S
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.jnat-
prod.6b00601.
1H and 13C NMR spectra for compounds 3−5, BDA, and
BDAA as well as for labeling experiments; results of
advanced Marfey’s method; biosynthetic schemes;
extraction and isolation of BDA and BDAA (PDF)
b. Endolides C and D (3 and 4) and Hirsutide (5). A 1.5 L amount
of BMS agar medium, supplemented with phenylalanine, was
exhaustively extracted with EtOAc and concentrated under vacuum
(using a vacuum rotatory evaporator, 40 °C) to yield a 315 mg of
extract. VLC was then used for fractionation of the extract using Merck
silica gel 60 M (0.040−0.063 mm, 230−400 mesh size) as sorbent. A
gradient solvent system of increasing polarity was used for sample
elution, starting with 100% petroleum ether to 100% CH2Cl2 to 100%
EtOAc to 100% acetone to 100% MeOH, yielding nine fractions. VLC
fraction 4 (50% CH2Cl2/50% EtOAc) was then subjected to NP-
HPLC fractionation (Knauer column; 250 × 8 mm, 5 μm, Eurospher
II-100 Si; mobile phase: petroleum ether/acetone (3:1); flow rate 2
mL min−1) to yield 10 subfractions. Endolide C (3) was isolated from
subfraction 4.8 using RP-HPLC (Macherey-Nagel column; Nucleodur
C18 EC Isis, 250 × 4.6 mm, 5 μm; mobile phase: 60% MeOH/H2O;
flow rate 1 mL min−1) and collected at tR 20 min, yield 1.5 mg L−1.
Endolide D (4) was isolated from subfraction 4.9 using RP-HPLC
(Macherey-Nagel column; Nucleodur C18 EC Isis, 250 × 4.6 mm, 5
μm; mobile phase: 60% MeOH/H2O; flow rate 1 mL min−1) and
AUTHOR INFORMATION
Corresponding Author
*Tel: +49228733747. Fax: +49228733250. E-mail: g.koenig@
uni-bonn.de.
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Financial support came from the German Centre for Infection
Research (DZIF). A scholarship for F.E.M. from the Ministry of
Higher Education of the government of Egypt is gratefully
acknowledged.
REFERENCES
(1) Almeida, C.; Kehraus, S.; Pruden
Org. Chem. 2011, 7, 1636−1642.
■
̂
cio, M.; Konig, G. M. Beilstein J.
̈
G
DOI: 10.1021/acs.jnatprod.6b00601
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
(2) Almeida, C.; Hemberger, Y.; Schmitt, S. M.; Bouhired, S.;
Natesan, L.; Kehraus, S.; Dimas, K.; Gutschow, M.; Bringmann, G.;
̈
Konig, G. M. Chem. - Eur. J. 2012, 18, 8827−34.
̈
(3) Almeida, C.; Part, N.; Bouhired, S.; Kehraus, S.; Konig, G. M. J.
̈
Nat. Prod. 2011, 74 (1), 21−25.
(4) Almeida, C.; El Maddah, F.; Kehraus, S.; Schnakenburg, G.;
Konig, G. M. Org. Lett. 2016, 18 (3), 528−531.
̈
(5) Steyn, P. S.; Tuinman, A. A.; van Heerden, F. R.; van Rooyen, P.
H.; Wessels, P. L.; Rabie, C. J. J. Chem. Soc., Chem. Commun. 1983,
47−49.
(6) Xiang, W. S.; Wang, J. D.; Wang, X. J.; Zhang, J. J. Antibiot. 2009,
62 (9), 501−505.
(7) Partida-Martinez, L. P.; de Looss, C. F.; Ishida, K.; Ishida, M.;
Roth, M.; Buder, K.; Hertweck, C. Appl. Environ. Microbiol. 2007, 73
(3), 793−797.
(8) Lang, G.; Blunt, J. W.; Cummings, N. J.; Cole, A. L.; Munro, M.
H. J. Nat. Prod. 2005, 68, 1303−1305.
(9) Neuss, N.; Boeck, L. D.; Brannon, D. R.; Cline, J. C.; DeLong, D.
C.; Gorman, M.; Huckstep, L. L.; Lively, D. H.; Mabe, J.; Marsh, M.
M.; Molloy, B. B.; Nagarajan, R.; Nelson, J. D.; Stark, W. M.
Antimicrob. Agents Chemother. (Bethesda) 1968, 8, 213−219.
(10) Puel, O.; Galtier, P.; Oswald, I. P. Toxins 2010, 2 (4), 613−631.
(11) Gandía-Herrero, F.; García-Carmona, F. Trends Plant Sci. 2013,
18 (6), 334−343.
(12) Gerratana, B. Med. Res. Rev. 2012, 32 (2), 254−293.
(13) Guo, C.-J.; Yeh, H.-H.; Chiang, Y.-M.; Sanchez, J. F.; Chang, S.-
L.; Bruno, K. S.; Wang, C. C. J. Am. Chem. Soc. 2013, 135, 7205−7213.
(14) Dewick, P. M. Medicinal Natural Products: A Biosynthetic
Approach, 3rd ed.; John Wiley & Sons Ltd.: The Atrium, Southern
Gate, Chichester, UK, 2009.
(15) Hochuli, M.; Patzelt, H.; Oesterhelt, D.; Wuthrich, K.;
̈
Szyperski, T. J. Bacteriol. 1999, 181 (19), 3226−3237.
(16) Beale, J. M.; Lee, J. P.; Nakagawa, A.; Omura, S.; Floss, H. G. J.
Am. Chem. Soc. 1986, 108, 331−332.
(17) Cho, H.; Beale, J. M.; Cynthia, G.; Mocek, U.; Nakagawa, A.;
Omura, S.; Floss, H. G. J. Am. Chem. Soc. 1993, 115, 12296−12304.
(18) Hansson, D.; Menkis, A.; Olson, Å.; Stenlid, J.; Broberg, A.;
Karlsson, M. Phytochemistry 2012, 84, 31−39.
(19) Katayama, K.; Miyoshi, M.; Nabeta, K.; Tazaki, H. Phytochem.
Lett. 2012, 5, 761−765.
(20) Yamazaki, Y.; Kitajima, M.; Arita, M.; Takayama, H.; Sudo, H.;
Yamazaki, M.; Aimi, N.; Saito, K. Plant Physiol. 2004, 134, 161−170.
(21) McMurry, J.; Begley, T. The Organic Chemistry of Biological
Pathways; Roberts and Company Publishers, 2005; pp 285−286.
(22) Matte, A.; Tari, L. W.; Goldie, H.; Delbaere, L. T. J. Biol. Chem.
1997, 272 (13), 8105−8108.
(23) Roje, S. Phytochemistry 2006, 67, 1686−1698.
(24) Bhushan, R.; Bruckner, H. Amino Acids 2004, 27, 231−247.
̈
(25) Kamata, S.; Sakai, H.; Hirota, A. Agric. Biol. Chem. 1983, 11,
2637−2638.
(26) Kennedy, J. Nat. Prod. Rep. 2008, 25, 25−34.
(27) Thiericke, R.; Rohr, J. Nat. Prod. Rep. 1993, 10 (3), 265−289.
(28) Adrio, J. L.; Demain, A. L. Int. Microbiol. 2003, 6, 191−199.
(29) Schwarzer, D.; Finking, R.; Marahiel, M. A. Nat. Prod. Rep. 2003,
20, 275−287.
(30) Challis, G. L.; Ravel, J.; Towensend, C. A. Chem. Biol. 2000, 7
(3), 211−224.
(31) Sy-Cordero, A. A.; Pearce, C. J.; Oberlies, N. H. J. Antibiot.
2012, 65 (11), 541−549.
(32) Dittmann, J.; Wenger, R. M.; Kleinkauf, H.; Lawen, A. J. Biol.
Chem. 1994, 269 (4), 2841−2846.
(33) Wang, X.-J.; Yan, Y.-J.; Zhang, B.; An, J.; Wang, J.-J.; Tian, J.;
Jiang, L.; Chen, Y.-H.; Huang, S.-X.; Yin, M.; Gao, A.-L.; Liu, C.-X.;
Zhu, Z.-X.; Xiang, W.-S. J. Bacteriol. 2010, 192, 4526−4527.
̈
̧ S.; Akay, B.; Ozdamar, T. H. Enzyme Microb. Technol.
(34) Takac,
1995, 17 (5), 445−452.
H
DOI: 10.1021/acs.jnatprod.6b00601
J. Nat. Prod. XXXX, XXX, XXX−XXX