Isolation, structure, and biological activity of phaeofungin, a cyclic lipodepsipeptide from a Phaeosphaeria sp. Using the genome-wide Candida albicans fitness test

Article abstract of DOI:10.1021/np300704s

Phaeofungin (1), a new cyclic depsipeptide isolated from Phaeosphaeria sp., was discovered by application of reverse genetics technology, using the Candida albicans fitness test (CaFT). Phaeofungin is comprised of seven amino acids and a β,γ-dihydroxy-γ-methylhexadecanoic acid arranged in a 25-membered cyclic depsipeptide. Five of the amino acids were assigned with d-configurations. The structure was elucidated by 2D-NMR and HRMS-MS analysis of the natural product and its hydrolyzed linear peptide. The absolute configuration of the amino acids was determined by Marfey's method by complete and partial hydrolysis of 1. The CaFT profile of the phaeofungin-containing extract overlapped with that of phomafungin (3), another structurally different cyclic lipodepsipeptide isolated from a Phoma sp. using the same approach. Comparative biological characterization further demonstrated that these two fungal lipodepsipeptides are functionally distinct. While phomafungin was potentiated by cyclosporin A (an inhibitor of the calcineurin pathway), phaeofungin was synergized with aureobasidin A (2) (an inhibitor of the sphingolipid biosynthesis) and to some extent caspofungin (an inhibitor of glucan synthase). Furthermore, phaeofungin caused ATP release in wild-type C. albicans strains but phomafungin did not. It showed modest antifungal activity against C. albicans (MIC 16-32 μg/mL) and better activity against Aspergillus fumigatus (MIC 8-16 μg/mL) and Trichophyton mentagrophytes (MIC 4 μg/mL). The linear peptide was inactive, suggesting that the macrocyclic depsipeptide ring is essential for target engagement and antifungal activity.

Full text of DOI:10.1021/np300704s

Article
pubs.acs.org/jnp
Isolation, Structure, and Biological Activity of Phaeofungin, a Cyclic
Lipodepsipeptide from a Phaeosphaeria sp. Using the Genome-Wide
Candida albicans Fitness Test
Sheo B. Singh,*,† John Ondeyka,† Guy Harris,† Kithsiri Herath,† Deborah Zink,† Francisca Vicente,‡,§
Gerald Bills,‡,§ Javier Collado,‡,§ Gonzalo Platas,‡,§ Antonio Gonzalez del Val,‡,§ Jesus Martin,‡,§
́
Fernando Reyes,§ Hao Wang,⊥,∥ Jennifer Nielsen Kahn,⊥ Stefan Galuska,⊥ Robert Giacobbe,⊥
George Abruzzo,⊥ Terry Roemer,⊥,∥ and Deming Xu∥,▽
†Department of Medicinal Chemistry, Merck Research Laboratories, PO Box 2000, Rahway, New Jersey 07065, United States
‡Centro de Investigacion
§Fundacion
MEDINA, Centro de Excelencia en Investigacion
Parque Tecnologico Ciencias de la Salud, 18100 Armilla, Granada, Spain
́ ́
Basica (CIBE), Merck Sharp & Dohme de Espana, S. A., Josefa Valcarcel 38, E-28027, Madrid, Spain
̃
́
́
de Medicamentos Innovadores en Andalucía, Avenida Conocimiento 3,
́
⊥Department of Infectious Diseases, Merck Research Laboratories, PO Box 2000, Rahway, New Jersey 07065, United States
∥Center of Fungal Genetics, Merck Frosst Canada Ltd., Montreal, Quebec, H2X3Y8, Canada
S
* Supporting Information
ABSTRACT: Phaeofungin (1), a new cyclic depsipeptide isolated from Phaeosphaeria sp., was discovered by application of
reverse genetics technology, using the Candida albicans fitness test (CaFT). Phaeofungin is comprised of seven amino acids and a
β,γ-dihydroxy-γ-methylhexadecanoic acid arranged in a 25-membered cyclic depsipeptide. Five of the amino acids were assigned
with D-configurations. The structure was elucidated by 2D-NMR and HRMS-MS analysis of the natural product and its
hydrolyzed linear peptide. The absolute configuration of the amino acids was determined by Marfey’s method by complete and
partial hydrolysis of 1. The CaFT profile of the phaeofungin-containing extract overlapped with that of phomafungin (3), another
structurally different cyclic lipodepsipeptide isolated from a Phoma sp. using the same approach. Comparative biological
characterization further demonstrated that these two fungal lipodepsipeptides are functionally distinct. While phomafungin was
potentiated by cyclosporin A (an inhibitor of the calcineurin pathway), phaeofungin was synergized with aureobasidin A (2) (an
inhibitor of the sphingolipid biosynthesis) and to some extent caspofungin (an inhibitor of glucan synthase). Furthermore,
phaeofungin caused ATP release in wild-type C. albicans strains but phomafungin did not. It showed modest antifungal activity
against C. albicans (MIC 16−32 μg/mL) and better activity against Aspergillus fumigatus (MIC 8−16 μg/mL) and Trichophyton
mentagrophytes (MIC 4 μg/mL). The linear peptide was inactive, suggesting that the macrocyclic depsipeptide ring is essential for
target engagement and antifungal activity.
Candida albicans and Aspergillus fumigatus are two key fungal
pathogens responsible for the majority of life-threatening
systemic fungal infections. These infections are particularly
problematic to the patient populations that are immunocom-
promised.1 Currently, three classes of drugs are used in the
clinic for treatment of systemic fungal infections. These are
amphotericin B,2 azoles (e.g., fluconazole),3 and echinocandins
(e.g., caspofungin, micafungin, and anidulafungin).4 While these
treatment options are generally acceptable, they are not
limitless. In addition, fungal strains resistant to these treatments
are emerging, though fortunately with lower frequency.
Therefore, new treatment options are needed to augment or
replace these agents.
Natural products continue to be one of the best sources of
antifungal agents, and two of the three classes of clinically used
antifungal agents are natural product derived. However, the
natural product-based drug discovery process comes with
Special Issue: Special Issue in Honor of Lester A. Mitscher
Received: October 9, 2012
© XXXX American Chemical Society and
American Society of Pharmacognosy
A
dx.doi.org/10.1021/np300704s | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
limitations, as compared to synthetic compounds, ranging from
the high demand for resources, the specialized expertise needed
in building and maintaining high-quality libraries of natural
products, and the selection of diverse microbial strains capable
of producing natural products, to the incompatibility of natural
product extract libraries with the “blitz screen” mentality of
high-throughput screens. These limitations had been further
exacerbated by the ever-increasing rediscovery of known
compounds. Collectively, they have exhausted the conventional
“grind-and-find” approach leading to abandonment of natural
product efforts by most of the big pharmaceutical companies.
The key to increasing the probability of success in identifying
novel scaffolds from natural products is threefold: (1) The
diversity of the source microbes must be improved to increase
the probability of new natural product discovery; (2) prior to
committing chemical fractionation, those extracts (of fermenta-
tion broths) containing known compounds must be effectively
recognized and removed; and (3) extracts containing novel
biological activities (and most likely novel structures) must be
positively identified and prioritized.
To address these challenges, efficient and robust natural
product dereplication by chemical and biological means is
needed. While large-scale chemical profiling can be readily
achieved, true dereplication is resource-consuming and often
strategically prohibitive if a structure is to be specifically
associated with a specific biological activity. Dereplication of
compounds using conventional single-target-based biochemical
assays is relatively straightforward. However, parallel multitarget
(protein or whole-cell based) screenings are not amenable to
the challenge, since biological dereplication requires compre-
hensive insights of mechanisms of action of active compounds
that may have not been characterized. Chemical genomic
strategies could provide a solution to biological dereplication.
In fact, our effort in applying the chemogenomic approach to
natural products (see Roemer et al.5 for a review) was first and
foremost augmented by replenishing the natural product library
with biodiversity. We have expanded our efforts to isolate
microorganisms from different geography and habitats. This
geographic diversity was coupled with improved high-
throughput fermentation methods aimed to increase functional
chemical diversity.6
form of crude extracts. With a compendium of known natural
products (pure), the CaFT profile of any extract in question
could be readily compared. If it matched the profile of a known
compound, the extract was immediately analyzed by LC-MS to
confirm the presence/absence of such a compound. In other
cases, a CaFT profile might reflect the biological MOA of a
known natural product that is not present in the compendium.
This could be readily remedied in the CaFT and the LC-MS
analyses using the compound in question. The CaFT profiling
of natural products had been a self-reiterating process that also
grouped extracts by their biological activities. Once a novel
CaFT profile was identified, chemical resources could then be
allocated for mechanistically guided fractionation of the
representative extract.5 This approach has been proven
productive in identifying novel compounds from natural
products with distinctive biological activities.5,9−13
In this report, we present the identification, isolation,
structural elucidation, and characterization of a cyclic lip-
odepsipeptide, phaeofungin (1, Figure 1), from an extract of a
fungus, Phaeosphaeria sp. (F-167,953). The original extract was
identified for its whole-cell activity against wild-type C. albicans
and A. fumigatus, two medically important human fungal
The chemogenomic approach, known as the “fitness test”,
was first developed using baker’s yeast and adapted in the
primary human fungal pathogen, C. albicans.7,8 The assay relies
on a phenomenon known as chemically induced haploinsuffi-
ciency, where deletion of one allele of a given gene in a diploid
fungus renders the mutant hypersensitive to a compound that
inhibits the corresponding protein target at sublethal
concentrations. In fact, when a collection of heterozygous
deletion strains covering the entire or a significant portion of a
fungal genome was tested against a panel of antifungal
compounds with known modes of action (MOAs), hyper-
variable growth (hypersensitivity or resistance, hence the term
fitness test) was restricted to only a small set of strains that
correspond to aspects of their respective MOAs, including the
targets. In almost all cases, the concise fitness test profiles are
reflective of the MOAs of compounds tested.7,8 As all the
heterozygotes were barcoded at the deleted allele, only a small
volume of pooled strains for each assay is needed to determine
the relative abundance (i.e., fitness) of each strain in response
to chemical perturbation, by DNA microarray methodology.
By applying the C. albicans fitness test (CaFT), it was
possible to mechanistically annotate natural products in the
Figure 1. Structures of phaeofungin (1) and two other cyclic
depsipeptides, aureobasidin A (2) and phomafungin (3).
B
dx.doi.org/10.1021/np300704s | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Figure 2. CaFT profiles of aureobasidin A, myriocin, rustmicin, phaeofungin, and the fermentation extracts of F-167,953. The relative behavior of
each heterozygous deletion strain elicited by the antifungal compounds is expressed by the normalized z-scores, with positive values indicating
hypersensitivity, and negative values resistance. To compare the CaFT profiles, strains with absolute values of z-scores no less than 3 in at least three
independent experiments were selected. Their z-scores were grouped by hierarchical clustering using the centroid linkage method, the results of
which are displayed in TreeView, with the scale of the heat map shown on the top-right corner, together with experimental numbers and
experimental conditions. The heterozygous deletion strains are indicated by the orf19 designations of the corresponding genes, with C. albicans gene
names and S. cerevisiae orthologs/homologues (if any) (as appeared in the Candida Genome Database, www.candidagenome.org). Highlighted by
red arrows are those involved in the sphingolipid biosynthesis.
pathogens. When tested in the CaFT, it generated a profile that
contained hypersensitivities of fungal strains corresponding to
the sphingolipid biosynthesis pathway. However, the CaFT and
LC-MS analyses failed to detect the presence of known
inhibitors of the pathway (e.g., myriocin,14 rustmicin,15−17
minimoidin,16 aureobasidins (e.g., aureobasidin A, 2),18
australifungin,16 and syringomycins19). Nonetheless, the
CaFT profile overlapped with that of another set of extracts,
which were subsequently identified to contain a cyclic
lipodepsipeptide, phomafungin (3),11 indicating certain struc-
C
dx.doi.org/10.1021/np300704s | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Figure 3. CaFT profiles of phaeofungin (1) and its source extract from F-167,953, and phomafungin (3) and its source extract from F-224,939, and
other extracts identified to contain known inhibitors of the fungal glucan synthase (GS) inhibitors. The CaFT results were analyzed and displayed
using the same parameters described in Figure 2. Highlighted are genes that are involved in the calcineurin pathway (black), sphingolipid (yellow),
potential efflux (white), and glucan synthesis (red). Strains of double heterozygous deletion of FKS1 together with another gene implicated in the
glucan synthesis/cell wall are indicated by a red bracket, with blue arrows indicating three other deletion strains (RHO1+/− SMI1+/−, RHO1+/−
FKS3+/−, and RHO1+/− CHS3+/−). These three strains, together with RHO1+/− and two independently constructed GSK1+/− strains, were not
significantly hypersensitive to F-167,953 or phaeofungin. The scale of the heat map is the same as in Figure 2.
tural similarity. We proceeded to isolate and characterize the
active compound produced by F-167,953. Consistent with the
prediction of the CaFT results, this compound turned out to be
another cyclic lipodepsipeptide, termed phaeofungin (1).
Despite biological signature overlap of the two cyclic
depsipeptides, our results showed that phaeofungin (1) and
phomafungin (3) are structurally and biologically distinct. This
discovery demonstrates the power of the CaFT, which detected
and differentiated natural products in crude extracts with
distinct or overlapping (often partially) activities with distinct
structural and mechanistic attributes.
D
dx.doi.org/10.1021/np300704s | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
1
Table 1. H (500 MHz) and 13C (125 MHz) NMR Assignments of Phaeofungin (1) in C5D5N and CD3OD
AA
no.
type
δC C5D5N
δH, mult, J in Hz, C5D5N
HMBC (H→C)
δC CD3OD
δH, mult, J in Hz, CD3OD
Thr
1
2
3
4
CO
169.65
61.3
169.6
61.6
68.3
20.7
CH
4.63, m
C-1, 3, Ser1 (C-1)
C-1
4.24, d, 4
4.10, m
CH
67.9
4.16, dq, 3, 6
1.41, d, 6.0
9.82, d, 6.0
CH3
α-NH
CO
21.6
C-2, 3
1.3, d, 6.5
C-2, 3, Ser1 (C-1)
Ser1
Ser2
1
2
3
173.85
57.2
173.5
56.7
63.0
CH
5.04, dt, 5.5, 3
4.55, dd, 12, 3
4.75, dd, 12, 3
10.15, d, 5
C-1
C-1
4.39, t, 3
CH2
62.7
4.09, dd, 10, 2.5
4.25, m
α-NH
CO
C-2, Ser2 (C-1)
1
2
3
172.91
58.8
172.7
58.3
63.7
CH
5.39, dt, 9.5, 5.5
4.43, dd, 14, 6
4.39, dd, 14, 5.5
8.92, d, 9.5
C-1, 2
C-1, 2
4.55, t, 6
CH2
63.5
3.69, dd, 11.5, 6
3.67, dd, 11.5, 6
α-NH
CO
C-2, 3, Ala1 (C-1)
Ala1
Asp
1
2
3
173.85
51.0
174.1
51.0
16.8
CH
4.90, pentet, 7.0
1.70, d, 7.0
C-1, 3
4.40, q, 7.5
1.43, d, 7.5
CH3
α-NH
CO
17.4
C-1, 2
8.07, d, 7.0
Asp (C-1)
1
2
3
174.19
51.6
174.1
51.4
40.4
CH
5.27, ddd, 9, 5, 2
2.78, dd, 14, 5.5
3.41, dd, 14, 3.0
C-1, 3, 4, Ala2 (C-1)
C-1, 2, 4
4.62, dd, 5, 3
3.0, dd, 16, 3
2.48, dd, 16, 5
CH2
41.7
C-2, 4
4
CO
177.80
178.2
α-NH
CO
8.91, d, 9
C-2, 3, Ala2 (C-1)
Ala2
Ser3
1
2
3
173.94
53.2
175.3
52.5
16.4
CH
4.74, dq, 4.5, 7.5
1.42
C-1, 3
4.27, q, 7.5
1.40, d, 7.5
CH3
α-NH
CO
17.1
C-1, 2
9.69, d, 5.0
C-2, 3, Ser3 (C-1)
1
2
3
174.04
56.3
174.3
56.6
63.0
CH
5.25, dt, 9, 3
4.25, dd, 11, 2.5
4.64, m
4.48, t, 2.5
3.79, dd, 11.5, 3
4.11, m
CH2
63.3
C-1, 2
α-NH
CO
8.60, d, 9.5
C-2, DHMHDA (C-1)
DHMHDA
1
2
172.65
38.7
174.3
39.3
CH2
1.9, brd, 14
2.71, dd, 14, 10
5.90, d, 10
C-1, 3
2.69, dd, 14, 10
2.51, brd, 14
5.19, d, 10
C-1, 3
3
4
5
6
CH
Co
79.5
74.0
39.3
23.7
C-1, 2, 4, 5, 17, Thr (C-1)
79.3
74.7
39.1
24.0
CH2
CH2
1.56, m
1.60, m
1.76, m
1.25, m
1.50, m
1.42, m
7−13
CH2
30.1
30.4
1.25, m
30.4 (5)
31.1
30.8 (5)
31.4
14
15
16
17
CH2
CH2
CH3
CH3
32.5
1.25, m
1.25, m
0.88, t, 7.0
1.35, s
33.1
1.25, m
1.25, m
0.89, t, 7
1.19, s
23.5
23.7
14.9
C-14, 15
C-3, 4, 5
14.4
22.6
22.4
antifungal cyclic depsipeptide aureobasidin A), LCB1/
orf19.6438, and MIT1/orf19.4077 (Figure 2). However,
aureobasidin A (2) and other inhibitors of the sphingolipid
biosynthesis yielded CaFT profiles that shared no additional
similarity (Figure 2). Nevertheless, when compared with other
natural products from our collection, the CaFT profile of F-
167,953 overlapped with that of another class of fungal extracts
that were subsequently identified to contain the lipodepsipep-
tide phomafungin (3)11 (Figure 3). More intriguingly, F-
167,953 induced specific hypersensitivity of a group of double
RESULTS AND DISCUSSION
■
Identification of a Phaeofungin-Containing Extract in
the CaFT. Extracts of fermentation broths of microbial strains
with whole-cell antifungal activities against the wild-type C.
albicans were tested in the CaFT assay. A fungal extract from
strain F-167,953 produced a complex profile that was distinct
from any known natural products tested in the CaFT. Several
hypersensitive heterozygous deletion strains were identified
that corresponded to genes involved in the sphingolipid
biosynthesis, including AUR1/orf19.1945 (the target of the
E
dx.doi.org/10.1021/np300704s | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
heterozygous deletion strains, each of which contained
heterozygous deletions of two genes that are functionally
implicated in the fungal glucan synthesis and cell wall
biosynthesis.5 The hypersensitivity of these strains is highly
reflective of the presence of fungal glucan synthase inhibitors in
the extracts5 (Figure 3). Although the complex profiles of these
extracts did not directly suggest a discernible MOA, the
difference between F-167,953 and the phomafungin-containing
extracts indicated that their biological activities shared certain
similarities but were not completely identical. Prior to any
chemical fractionation effort, we predicted that the correspond-
ing natural products present in these extracts were structurally
and functionally different. These extracts were thus selected for
detailed chemical purification and characterization of natural
product(s) responsible for biological activities.
Producing Organisms. The fungal strain F-167,953 was
isolated from a Sedum sp., a flowering plant, collected in the
province of Albacete, Spain. Similarity searches with the 28S
and ITS rDNA sequences indicated a relatively high sequence
similarity with some species of Phaeosphaeria, Ophiosphaerella,
and various anamorph genera associated with the family
Phaeosphaeriaceae. Alignments of its 28S rDNA sequence
clearly placed this fungus within the Phaeosphaeriaceae. The
nearest match in the ITS region of the rDNA was an anamorph
species, Chaetosphaeronema hispidulum (CBS 216.75), of the
Phaeosphaeriaceae, with a similarity of 94.5%. A second strain,
F-262,327 (from the province of Cuenca), was identified
subsequently to produce phaeofungin (1). It yielded a nearly
identical 28S rDNA sequence. The 28S rDNA sequences of
these two strains were resampled from recent phylogenetic
studies on fungi of the Phaeosphaeriaceae20−22 and aligned.
Neighbor joining analysis of the alignment (Figure S1,
Supporting Information) placed F-167,953 and F-262,327 in
the Phaeosphariaceae among a group of anamorph species
including Plenodomus fuscomaculans and Chaetosphaeronema
coonsii.
and absorption bands for hydroxy and various amide/ester
carbonyls in the IR spectrum. The 13C NMR spectrum along
with the DEPT spectrum in C5D5N and CD3OD exhibited the
presence of nine carboxyl carbonyls, nine methines (seven α-
amino acid methines and two oxymethines), three oxy-
methylenes, an oxygenated quaternary, and four methyls, and
the remaining carbons were aliphatic methylenes, typical of
1
those present in a fatty chain. The H NMR spectrum in
C5D5N (Table 1) revealed seven doublets in the downfield
region of the spectrum, which were exchanged in the 1H NMR
spectrum in CD3OD (Table 1), suggesting that they are
exchangeable NH groups. This indicated that compound 1 is
peptidic, consisting of at least seven α-amino acids. The COSY,
TOCSY, and HSQC spectroscopic analysis of 1 indicated the
presence of the following moieties: three serines (Ser), a
threonine (Thr), two alanines (Ala), an aspartic acid (Asp), and
a β,γ-dihydroxy-γ-methylhexadecanoic acid (DHMHDA). The
structure of the latter unit was confirmed by HMBC
correlations of H3-17 to C-3, C-4, and C-5 and H-3 to C-4,
C-5, and C-17. Basic hydrolysis of 1 produced the acylated
heptapeptide 4 (MW 921), suggesting the presence of a lactone
ring in 1. The mass spectrometric analysis of 4 showed a
fragment ion at m/z 638 due to the loss of the fatty acid unit.
The HRMS-MS analysis of the fragment ion at m/z 638
produced b series fragment ions at m/z 519, 432, 345, 274, and
159, suggesting a Ser-Ala-Asp-Ala-Ser-Ser-Thr-OH moiety
(Figure 4). Whether Ser or Ala was present at the end of the
Prior to the CaFT analysis, these extracts of fungus F-
167,953 were observed to perturb the cell wall biosynthesis in a
distinct cell-based assay designed to detect inhibitors of the C.
albicans cell wall function/glucan synthesis (data not shown), a
bioactivity reminiscent of what was observed in the CaFT
(Figure 3). The strain was revived from the freezer and grown
in an eight-medium nutritional array to generate 1 mL extracts
for evaluation of antifungal activity.6 The fungus produced
active C. albicans activity in five of the eight media tested, and
the medium yielding the most potent extract was scaled up to
one liter.
Figure 4. HRESIMS-MS fragmentation of the acyclic acylated
heptapeptide 4.
Isolation and Structural Elucidation of Phaeofungin
(1). The fermentation broth of Phaeosphaeria sp. F-167,953 was
extracted with an equal volume of acetone. The extract was
chromatographed on an Amberchrome column and eluted with
a step gradient of aqueous MeOH. The fractions from each
chromatographic step were tested for their antifungal activity
against the wild-type C. albicans. Those fractions containing the
antifungal activity were finally purified by reversed-phase
HPLC, and the resulting active fractions were lyophilized to
afford 1 (638 mg, 638 mg/L) as an amorphous, colorless
powder. The purified material was tested in the CaFT, and it
yielded a profile indistinguishable from the original one
(Figures 2 and 3), suggesting that the purified material
represents the biological activity detected in the CaFT.
HRESIFTMS analysis of 1 provided a molecular formula of
C40H69N7O16. It displayed end absorption in the UV spectrum
N-terminus could not be differentiated from these fragmenta-
tions due to the lack of fragmentation between these two amino
acids. The amino acid sequence was corroborated by the
HMBC correlations of α-CH and NH protons to the
corresponding adjacent carbonyls and α-carbons in C5D5N
(Table 1). The unambiguous assignment of the C-1 of
DHMHDA unit was secured from the HMBC correlations of
H-2 and H-3, which were critical for the final sequence
determination. The C-1 signal of the DHMHDA unit showed
the HMBC correlation to the Ser3-NH, confirming the amide
bond between Ser3 and DHMHDA, thus the linear sequence.
The H-3 resonance of the DHMHDA unit displayed a strong
HMBC correlation to the C-1 of the threonine, confirming the
macro-lactone ring in the depsipeptide structure of 1.
1
The H NMR chemical shifts, including shifts of carbon-
bound protons in C5D5N, were highly concentration depend-
F
dx.doi.org/10.1021/np300704s | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
ent. The same concentration-dependent phenomenon was not
observed in CD3OD. This phenomenon was likely due to the
presence of a large number of hydroxylated amino acids that
interacted with C5D5N. NOESY correlations of phaeofungin in
C5D5N further corroborated the sequence and the assignment
of the structure. The Ser2-NH showed NOESY correlations to
NH groups of both the adjacent amino acids Ser1 and Ala1,
suggesting that they are proximal to each other. Similar
correlations were observed with the NH groups of other amino
acids, confirming the amino acid sequence of the depsipeptide.
Exhaustive acid hydrolysis of phaeofungin (1) with 6 N HCl
at 110 °C overnight followed by derivatization with Marfey’s
reagent (1-fluoro-2,4-dinitrophenyl-5-L-valine, L-FDVA)23 and
comparative quantitative analysis by LC-MS with correspond-
ing authentic standards confirmed the presence of 2 × D-Ser, a
L-Ser, a D-allo-Thr, a D-Ala, an L-Ala, and a D-aspartic acid
(Asp). Partial acid hydrolysis with 0.5 N HCl at 100 °C for 30
min followed by a reversed-phase HPLC purification led to the
isolation of two major peptide fragments, 5 and 6 (Figure 5).
isolated from the extract F-167,953, named phaeofungin. The
configurations of the two stereocenters of the oxygenated
carbons of the fatty acid remain unassigned.
Characterization of Biological Activity of Phaeofungin
(1). The antifungal potency and the spectrum of phaeofungin
(1) were determined in an in vitro growth assay. The data are
summarized in Table 2. Phaeofungin inhibited C. albicans with
an MIC of 16−32 μg/mL. It showed similar activity against C.
lusitaniae (MIC 32 μg/mL) and poorer activities against other
Candida species (Table 2). It exhibited slightly better activities
against the filamentous fungi A. fumigatus (MIC 8−16 μg/mL)
and the dermatophytic fungi T. mentagrophytes (MIC 4 μg/
mL). The activities of both C. albicans and A. fumigatus were
negatively impacted by the presence of 50% mouse and 50%
human sera. Compound 1 did not show any activity against
Staphylococcus aureus (tested at the highest concentration of 32
μg/mL), suggesting potential for relative specificity for the
fungal growth inhibition. The linear peptide 4 showed no
activity against any fungal strains (MIC > 32 μg/mL),
indicating the importance of the macrocyclic depsipeptide for
the antifungal activity. Phaeofungin exhibited dose-dependent
toxicity in DBA/2N mice when dosed (5% aqueous DMSO)
intraperitoneally (ip) at over 12.5 mg/kg including lethality.
The compound showed no adverse effects at 6.25 mg/kg.
When the isolated phaeofungin (1) was tested in the CaFT,
it yielded a profile that preserved characteristic features of the
original extract (Figures 2 and 3), suggesting that compound 1
corresponded to the biological activity observed in the CaFT of
the extract. After structure elucidation, it became clear that both
phaeofungin and phomafungin (3)11 are cyclic lipodepsipep-
tides. However, as predicted by the CaFT profiles, they are
structurally different from each other in both the lipid and the
cyclic peptide moieties: phaeofungin is cyclo-[DHMHDA-L-
Ser-D-Ala-D-Asp-L-Ala-D-Ser-D-Ser-D-allo-Thr], and phomafun-
gin is cyclo-[HMHDA-L-Ala-homoser-L-Glu-Gly-L-Ser-D-Asn-D-
allo-Thr-homoser]. A close analysis of the CaFT profiles
revealed several functional aspects of the activity that prompted
us to characterize the biological activities of both lip-
odepsipeptides.
Figure 5. Fragments (5 and 6) from the partial acid hydrolysis of
phaeofungin (1).
The structures of these fragments were elucidated by
HRESIMS-MS analysis (Figure S2, Supporting Information).
Similar exhaustive acid hydrolysis of 5 and 6 followed by
analogous reaction with Marfey’s reagent (L-FDVA)23 and
similar comparative quantitative analysis confirmed the
presence of L-Ala, 2 × D-Ser and D-allo-Thr; and 2 × D-Ser
and L-allo-Thr from 5 and 6, respectively. On the basis of the
combination of all the data presented above, a cyclo-
(DHMHDA-L-Ser-D-Ala-D-Asp-L-Ala-D-Ser-D-Ser-D-allo-Thr)
depsipeptide structure 1 was established for the natural product
It was noted that heterozygous deletions of genes involved in
the sphingolipid biosynthesis are hypersensitive to phomafun-
gin (3) and phaeofungin (1) in the CaFT (Figure 3). However,
a quantitative difference was noted in the relative hyper-
sensitivity. Most notably, AUR1+/− displayed pronounced
a
Table 2. Antifungal Activity (MIC, μg/mL) and Spectrum of Phaeofungin (1)
b
organism
Candida albicans
strain
phaeofungin (1)
phomafungin (3)
caspofungin
MY1055
MY1055
MY1381
ATCC22019
MY1396
ATCC6258
MY1012
MF5668
MF5668
MF7004
MB2865
32 (16)
>32
4
0.5
C. albicans +50% mouse serum
C. glabrata
>32
8 (4)
4
0.25
0.25
0.5
>32
C. parapsilosis
>32 (32)
32
C. lusitaniae
4
<0.03
1
C. krusei
>32 (32)
>32
4
C. tropicalis
>32
4
NT
32
Aspergillus fumigatus
A. fumigatus +50% human serum
Trichophyton mentagrophytes
Staphylococcus aureus
16 (8)
>32
>32
2 (1)
>32
32
4
32
>32
>32
a
The number in parentheses is MIC80 (prominent inhibition of growth, also referred to as MEC for Aspergillus). The time of incubation for all MIC
b
measurements was 24 h except for A. fumigatus, which were incubated for 48 h, and T. mentagrophytes was incubated for 96 h. MIC data of
phomafungin (3) are taken from ref 11.
G
dx.doi.org/10.1021/np300704s | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
hypersensitivity to phaeofungin, and RTA1+/− was most
sensitive to phaeofungin (Figure 3). Since AUR1 corresponds
to the target of aureobasidin A (2), we tested how the
antifungal activity of 2 was affected by either of the
lipodepsipeptides. Our results demonstrated that phaeofungin
and aureobasidin A are synergistic; that is, a combination of
both compounds at sublethal concentrations resulted in
maximal inhibition of growth of the wide-type C. albicans
strain (Figure 6A), whereas less synergy was observed between
phomafungin and aureobasidin A (Figure 6B).
Figure 7. Interaction of the antifungal activities of caspofungin with
phaeofungin (1) (A) and phomafungin (B) in the wild-type C. albicans
strain. Sublethal concentrations of either cyclic lipodepsipeptide were
mixed with caspofungin in a range of concentrations (100 to 1.6 ng/
mL). The antifungal activities of mixtures were measured after a 24 h
incubation from the OD600, which was then normalized by the growth
of untreated culture.
Figure 6. Interactions of the antifungal activities of aureobasidin A
(ABA) with phaeofungin (A) and phomafungin (B) against the wild-
type C. albicans strain used for the construction of heterozygous
deletions strains. Sublethal concentrations of either cyclic lip-
odepsipeptides (as indicated) were mixed with aureobasidin A in a
range of concentrations (100 to 0.4 ng/mL). The antifungal activities
of mixtures were measured after 24 h incubation from the OD600
which was then normalized by the growth of untreated culture.
,
In a cell-based screen for inhibitors of the fungal glucan
synthase, the phaeofungin-containing extract from F-167,953
was shown to perturb cell wall biogenesis in C. albicans (data
not shown). Consistent with this early observation, both the
extract and purified phaeofungin induced hypersensitivity of
two independently constructed GSK1+/− strains and a group of
double heterozygous deletion strains that have been implicated
in the action of inhibitors of the fungal glucan synthase (Figure
3). However, the rest of the profiles shared no similarity: the
RHO1+/− and three additional double heterozygous deletion
strains were only hypersensitive to the known glucan synthase
inhibitors (Figure 3). When tested in combination, low
concentrations of phaeofungin (with no inhibitory effect)
potentiated the activity of caspofungin (Figure 7A). Under
similar conditions, no potentiation was observed by phoma-
fungin (Figure 7B).
Two heterozygous deletion strains for genes involved in the
calcineurin pathway, CMD1/orf19.4413 and CRZ1/orf19.7359,
displayed specific hypersensitivity to phomafungin, but not to
phaeofungin (Figure 3). We tested an inhibitor of the
calcineurin pathway, cyclosporin A, in combination with either
lipodepsipeptide. Cyclosporin A, not active against C. albicans,
potentiated the potency of phomafungin (Figure 8B) but not
phaeofungin (Figure 8A), consistent with the CaFT results.
Despite these functional differences, the CaFT profiles of
both phomafungin (3) and phaeofungin (1) contained the
relative resistance of a heterozygous deletion strain of DIP5/
orf19.2942 (Figure 3). The corresponding gene encodes a
dicarboxylic amino acid permease. The relative resistance of the
Figure 8. Interactions of cyclosporin A (CaA) with phaeofungin (1)
(A) and phomafungin (3) (B) in the wild-type C. albicans strain.
Three concentration of cyclosporine A (as indicated) were mixed with
1 and 3 in a range of concentrations (10 to 0.16 μg/mL). The
antifungal activities of mixtures were measured after a 24 h incubation
by OD600, which was then normalized by the growth of untreated
culture.
heterozygous deletion strain could indicate a potential role of
the putative permease in the uptake of both lipodepsipeptides.
If so, a loss-of-function mutation of DIP5 should render the
mutant strain resistant to both phomafungin and phaeofungin.
H
dx.doi.org/10.1021/np300704s | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Figure 9. ATP release assay with phaeofungin (1) (A), phomafungin (3) (B), caspofungin (C), and amphotericin B (D). The ATP release (filled
symbols) and the whole-cell antifungal (open symbols) activities were tested in C. albicans strains 2323 (green symbols) and SC5314 (red symbols).
The solid blue line indicates the basal level of ATP release in the assay, the green dashed line the total of ATP release in strain 2323, and the red
dashed line the total ATP release in strain SC5314.
This was indeed observed in the fungal strain in which the
promoter of DIP5 is replaced with a regulatable tet promoter
under the repressing conditions (data not shown).
phomafungin are basic amino acids. The syringomycins family
of peptides consists of at least nine amino acids (several basic),
forming a 28-membered macrocyclic ring. The Pseudomonas
lipodepsipeptides (e.g., syringomycin E) are important virulent
factors for the phytopathogenic bacteria and are thought to
exert their effects by forming ion channels on the plasma
membrane, leading to cytolysis.28,29 Pseudomonas spp. also
produce large cyclic lipodepsipeptides known as the syringo-
peptins, comprised of 22 or 25 amino acids with an
octadepsipeptide ring structure and a 3-hydroxy fatty acyl
chain. They have been shown to target the yeast plasma
membrane by a lipid-dependent channel-forming mechanism of
action.30 The antifungal activity of syringomycin E was
functionally linked to the sphingolipid pathway in S. cerevisiae
that is sensitive to Ca2+ fluctuation.24,31 Our results also suggest
a functional connection between the two fungal lipodepsipep-
tides and the sphingolipid and Ca2+ homeostasis. However, as
predicted by the CaFT profiles (Figure 3), phomafungin and
phaeofungin are different in their biological activities: the
former is potentiated by cyclosporin A (Figure 8), while the
latter is potentiated by aureobasidin A (Figure 6). Furthermore,
phaeofungin potentiated caspofungin at sublethal concentra-
tions (Figure 7), consistent with the preferential hyper-
sensitivity of the double heterozygous deletion strains to only
phaeofungin in the CaFT (Figure 3). Phaeofungin, but not
phomafungin, was found to cause ATP release in two strains of
C. albicans (Figure 9).
Both cyclic lipodepsipeptides (1 and 3) were then tested for
their effects on release of ATP from the cytoplasm in two wild-
type C. albicans strains. With the reference compound
amphotericin B, no ATP release was observed even at a
concentration greater than the MIC (Figure 9D). Only minimal
ATP release was observed with caspofungin at concentrations
greater than the MIC, and this was particularly so in strain
2323, the starting C. albicans strain used for strain construction
(Figure 9C). While phomafungin (3) did not cause ATP
release in either strain (Figure 9B), phaeofungin (1) resulted in
the release of ATP in both strains. Most importantly, the
maximal release correlated with the MIC (Figure 9A).
Phaeofungin (1) and phomafungin (3) are distantly related
to a group of cyclic lipodepsipeptides, including syringomycin
E,24 syringostatin,25 syringotoxin,25 pseudomycin A,26 and
cormycin A,27 all of which are produced by plant-associated
Pseudomonas spp.11 However, these cyclic lipodepsipeptides
differ substantially from phaeofungin and phomafungin in both
the macrocyclic ring size and composition of constituent amino
acids. Both phaeofungin and phomafungin are comprised of a
β-hydroxyhexadecanoic (C16) fatty acid chain, which forms an
integral part of the macrocycle, in contrast to the syringomycin
series of peptides, which possess a shorter chain β-hydroxy fatty
acid and terminate with the acylation of the amino group of a
serine residue, of which the hydroxy group forms a macrocycle.
In addition, phaeofungin is comprised of only seven amino
acids and forms a 25-membered ring, and phomafungin is
comprised of only eight amino acids, forming a 28-membered
macrocyclic ring. None of the amino acids in phaeofungin and
Despite these functional differences, the exact mechanisms of
action of both fungal lipodepsipeptides remains elusive.
However, we speculate that phaeofungin (1) and phomafungin
(3) target the plasma membrane and that their activities are
differently affected (both qualitatively and quantitatively) by
I
dx.doi.org/10.1021/np300704s | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
changes in sphingolipid content, Ca2+ concentrations, and cell
wall integrity. In the case of 1, its interaction with the plasma
membrane could lead to release of ATP and cytolysis.
odepsipeptide-producing strains and showed that they belong to the
Phaeosphaeriaceae. To achieve further phylogenetic resolution, the
same genomic DNA samples were used to amplify and sequence the
intertranscribed spacer regions and 5.8S gene of the rDNA (ITS) with
primers 18S-3 and Nl1r.
Cyclic lipodepsipeptides are chemically and biologically
diverse secondary metabolites produced by many organisms.
Their biological properties have been exploited successfully in
antimicrobial therapeutics, e.g., echinocandins (caspofungin,
micafungin, and anidulafungin, inhibitors of cell wall syn-
thesis)32 and daptomycin (targeting bacterial plasma mem-
brane).33 Using a chemogenomic assay, the CaFT, we identified
two fungal lipodepsipeptides, phaeofungin (1) and phomafun-
gin (3). The CaFT profiles of these original extracts indicated
that these two active compounds are structurally and
biologically related, but not identical. This prediction was
confirmed upon isolation and characterization of both
compounds and subsequent biological analysis. It provides an
example in which the application of a chemogenomic approach
could be used to discern biological activities, hence the
structures of the active compounds, in a crude extract. This
serves as a powerful tool for the discovery of novel natural
product antifungal agents particularly when combined with
interesting chemical diversity.
PCR reactions were performed following standard procedures (5
min at 93 °C followed by 40 cycles of 30 s at 93 °C, 30 s at 53 °C, and
2 min at 72 °C) with Taq DNA polymerase (Q-bioGene), following
the procedures recommended by the manufacturer. The amplification
products (0.10 μg/mL) were sequenced using Bigdye Terminators
version 1.1 (Perkin-Elmer, Norwalk, CT, USA), following the
manufacturer’s recommendations. For all the amplification products,
each strand was sequenced with the same primers used for the initial
amplification. Partial sequences were assembled using Genestudio
software (Genestudio, Inc.), and consensus sequences were aligned
with Clustal X. Neighbor-joining analyses (Figure S1, Supporting
Information) were used to approximate the phylogenetic relationships
among strains.
Fermentations in Microarrays and Scale up for Isolation.
Our screening strategy relied on high-throughput generation of 1 mL
scale extracts of organisms grown under varied fermentation
parameters followed by assay for antibiosis caused by cell-penetrable
molecules using bioassays with C. albicans and S. aureus. Organism-
and-medium combinations yielding extracts with a minimum potency
and activity spectrum were scaled up to provide larger fermentations
suitable for profiling in the CaFT and further processing for an extract
library and for chemical fractionation, if needed. The strategy and
protocols for fermentation of fungi on nutritional microarrays have
been described previously.6 Each week, 160 to 240 fungal strains were
selected for fermentation. These fungi were grown 2 to 3 weeks in 60
mm Petri dishes containing YM agar (Fluka or Difco malt extract 10 g,
Difco yeast extract 2 g, agar 20 g, distilled H2O 1000 mL). Three to
four mycelial disks were cut from each 60 mm plate. Mycelia disks
were crushed in the bottom of tubes (25 × 150 mm) containing 8 mL
of SMYA medium (Difco neopeptone 10 g, maltose 40 g, Difco yeast
extract 10 g, agar 4 g, distilled H2O 1000 mL) and two cover glasses
(22 mm2). Tubes were agitated on an orbital shaker (200 rev min−1, 5
cm throw), and rotation of the cover glasses continually sheared
hyphae and mycelial disk fragments to produce homogeneous hyphal
suspensions. Tubes were agitated 4 to 6 days at 22 °C. Hyphal
suspensions from these tubes were transferred to master inoculum
plates. Master plates of fungal inoculum were used to inoculate eight-
media nutritional arrays in a 10-column × eight-row pattern.6
Nutritional arrays were grown statically for 21 days at 22 °C.
The detection of antifungal activity from strain F-167,953 originated
from a 1 mL fermentation in the medium GLUS [glucose 20 g; yeast
extract (Becton Dickinson) 1 g; monosodium glutamate 14.5 g; KCl
0.5 g; K2HPO4 0.25 g; CaCl2·2H2O 0.20 g; MgSO4·7H2O 0.02 g;
ZnSO4·7H2O 0.01 mg; CuSO4·5H2O 0.005 mg; distilled H2O 10 000
mL]. This fermentation was scaled up to 1 L by growing F-167,953 in
500 mL flasks with 150 mL of liquid GLUS agitated at 220 rpm, 22 °C
for 22 days. The liquid fermentations were extracted with an equal
volume of acetone and pooled. A 4 mL aliquot was frozen and tested
for CaFT, and the remaining extract was used for the isolation of active
metabolites.
EXPERIMENTAL SECTION
■
Geneal Experimental Procedures. All reagents were obtained
from Sigma-Aldrich and were used without further purification. The
UV spectrum was recorded on a Perkin-Elmer Lambda 35 UV/vis
spectrometer. Optical rotations were obtained on a Perkin-Elmer 241
polarimeter; IR spectral data were obtained on a Perkin-Elmer
Spectrum One spectrometer. The NMR spectra were obtained on a
Varian Inova 500 or 600 MHz spectrometer operating at 500 or 600
1
MHz for H and 125 or 150 MHz for 13C nuclei. The chemical shifts
were referenced to residual C5D5N (δH 8.74 ppm and δC 150.4 ppm)
and CD3OD (δH 3.30 ppm and δC 49.00 ppm). Data were collected
uniformly at 25 °C in 3 mm NMR tubes. A Nalorac 3 mm H{CN}
indirect Z-gradient probe was used for all samples. Varian standard
pulse sequences were used for all data collection. The 2D TOCSY data
were collected with a 4900 Hz spin-lock field held for 80 ms, using the
flopsy16 mixing scheme. The 1D TOCSY data were collected with a
6200 Hz spin-lock field using the MLEV17 mixing scheme. Mixing
times were arrayed: 20, 40, 60, 80, and 90 ms. Peak selection was
achieved using an IBURP waveform with a 50 Hz bandwidth. 1H
homonuclear correlation data were obtained with the Varian gCOSY
or gDQF-COSY pulse sequences. ROESY data were collected using a
4500 Hz spin-lock field applied for 200 ms. Single- and multiple-bond
heteronuclear connectivity data were observed using the gHSQC and
gHMBC pulse sequences, respectively. The gHMBC data were
collected using a mixing time optimized for a 7 Hz heteronuclear
coupling constant. High-resolution mass spectra were obtained on a
Thermo Finnigan LTQ-FT using electrospray ionization using a
Finnigan Ion Max source with source fragmentation on and equal to
18 V.
Candida albicans Fitness Test. The assay was reported
elsewhere.8 Its application to biological dereplication of natural
products has been published.5,8−13
Fungal Material. The fungal strain (F-167,953, E-000504901) was
isolated from living stems and leaves of a Sedum sp. (Crassulaceae)
collected in Motilleja, Albacete, Spain (Figure S1, Supporting
Information). A second strain (F-262,327, E-000531145) producing
phaeofungin (1) (Figure S1, Supporting Information) was isolated
from the stems and leaves of Teucrium sp. (Lamiaceae) collected at
Serrania de Cuenca, Cuenca, Spain. Strains were maintained as frozen
mycelia in 10% glycerol at −80 °C.
DNA Sequencing and Characterization of Fungal Strains.
Total genomic DNA was extracted from mycelia grown on YM agar.
The rDNA region containing the partial sequence of 28S rDNA
including the D1 and D2 variable domains was amplified with the
primers NL1 and NL4. Sequences of 28S rDNA were used to generate
a neighbor-joining tree (Figure S1, Supporting Information) that
demonstrated a close relationship between the two cyclic lip-
Extraction and Isolation of Phaeofungin (1). One liter of the
fermentation broth was extracted with 1 L of acetone and filtered, and
the filtrate was concentrated under reduced pressure to remove most
of the acetone. The 500 mL of the mainly aqueous solution remaining
after concentration was charged to a 60 cubic centimeter
Amberchrome column. The column was eluted by a 100 min linear
gradient of 0 to 100% aqueous MeOH at 10 mL/min, and 5 mL
fractions were collected every 0.5 min. The antifungal activity eluted in
fractions 13−19, which were concentrated under reduced pressure and
lyophilized to yield 1.3 g of a partially enriched fraction. This fraction
was dissolved in 20 mL of MeOH and chromatographed on a 1 L
Sephadex LH20 column, eluted with 100% MeOH at 10 mL/min.
J
dx.doi.org/10.1021/np300704s | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Fractions (20 mL each) were collected every 2 min. The activity eluted
in fractions 38−58. These fractions were combined and concentrated
to give 1.0 g of solid. This solid material was dissolved in 7 mL of
MeOH. One milliliter of this material was purified by preparative RP
HPLC (Zorbax Rx C8 21.2 × 250 mm) using a 40 min gradient of 10−
95% aqueous acetonitrile (0.1% TFA) at 12 mL/min. The activity
eluted at 20−26 min. The chromatography was repeated seven times,
and fractions from each run were pooled and lyophilized to afford 638
mg (638 mg/L) of phaeofungin (1) as a colorless powder: [α]23D +6.8
(c 0.45, MeOH); UV (MeOH) λmax end absorption; IR (ZnSe) νmax
flow rate of 3.6 mL/min]. Fractions eluting at 23.5, 24.5, and 25.0 min
were lyophilized to give pure phaeofungin, 5, and 6. The structures of
compounds 5 and 6 were elucidated by LCMS-MS analysis (Figure S2,
Supporting Information). 5: HRESIMS m/z 631.3922 (calcd for
C30H54N4O10 + H, 631.3913), 560.3538 (M − Ala), 473.3224 (M −
Ala-Ser), 386.2897 (M − Ala-Ser-Ser). 6: HRESIMS m/z 560.3516
(calcd for C27H49N3O9 + H, 560.3542), 473.3219 (M − Ser),
386.2878 (M − Ser).
Compounds 5 and 6 were separately hydrolyzed exhaustively,
similar to the exhaustive hydrolysis of phaeofungin (1). The
hydrolyzed mixture of amino acids was derivatized by Marfey’s
reagent in an identical manner and was analyzed in the similar manner
by reversed-phase HPLC. Compound 5 yielded 2 × D-Ser (tR 9.0 min),
L-Ala (tR 12.5 min), and D-allo-Thr (tR 12.2 min), and 6 afforded 2 ×
D-Ser (tR 9.0 min) and D-allo-Thr (tR 12.2 min).
3294, 2923, 2853, 1643, 1517, 1378, 1255, 1068 cm−1; for 1H and 13
C
NMR see Table 1; ESIMS m/z 904 [M + H]+, 886 [M − H2O + H]+,
868 [M − 2H2O + H]+; HRESIFTMS m/z 904.4865 (calcd for
C40H69N7O16 + H, 904.4879), 868.4688 (calcd for C40H65N7O14 + H,
868.4668).
Antifungal Assay. The whole-cell antifungal activity assay was
described elsewhere.13 The MIC (minimum inhibitory concentration)
against each of the strains was determined as previously described.34
Cells were inoculated at 105 colony-forming units/mL followed by
incubation at 37 °C with a 2-fold serial dilution of compound in the
growth medium for 37 °C for 20 h. MIC is defined as the lowest
concentration of an antifungal inhibiting visible growth.
ATP Release Assay. A Promega BacTiter Glo assay was used
according to the manufacturer’s instructions to measure the release of
ATP from the fungal cells in the presence of various antifungal
compounds.
Basic Hydrolysis of Phaeofungin (1). Phaeofungin (1) (20 mg)
was suspended in water (5 mL), treated with 0.1 NaOH (2 mL), and
stirred at room temperature for 90 min. The reaction mixture was
directly charged on a 5 mL Amberchrome column and washed
thoroughly with water, and the linear peptide was eluted with
methanol, which was concentrated under reduced pressure and
rechromatographed by reversed-phase HPLC using similar conditions
to those used for the purification of 1. Lyophilization of fractions gave
1
linear peptide 4 as a colorless, amorphous powder: H NMR (500
MHz, C5D5N) δ 9.51 (1H, d, J = 6 Hz), 9.37 (1H, d, J = 7.5 Hz), 9.31
(1H, d, J = 8 Hz), 9.04 (1H, d, J = 8 Hz), 8.87 (1H, d, J = 6.5 Hz),
8.86 (1H, d, J = 6 Hz), 5.43 (1H, brd, J = 5.5 Hz), 5.34 (1H, brt, J = 6
Hz), 5.32 (1H, t, J = 5.5 Hz), 5.26 (1H, dd, J = 11, 6 Hz), 5.20 (1H, t,
J = 6 Hz), 4.95 (1H, pentet, J = 7 Hz), 4.79 (1H, q, J = 7 Hz), 4.70
(1H, pentet, J = 6 Hz), 4.52 (1H, dd, J = 6, 5.5 Hz), 4.50 (1H, m),
4.44 (1H, dd, J = 11, 5.5 Hz), 4.36 (1H, dd, J = 11, 6 Hz), 4.33 (1H,
dd, J = 10.5, 5 Hz), 3.49 (1H, dd, J = 16.5, 5.5 Hz), 3.43 (1H, dd, J =
16.5, 7 Hz), 3.06 (1H, brd, 14 Hz), 2.99 (1H, dd, J = 14, 10 Hz), 1.90
(1H, dt, J = 5, 13 Hz), 1.79 (1H, dt, J = 4.5, 13 Hz), 1.67 (3H, d, J = 7
Hz), 1.63 (3H, d, J = 7 Hz), 1.58 (2H, m), 1.42 (3H, s), 1.28−124
(18H, m), 0.87 (3H, t, J = 7 Hz); HRESIFTMS m/z 922.4979 (calcd
for C40H71N7O17 + H, 922.4985); HRESIFTMS-MS (m/z) 638.2632
(calcd for C23H40N7O14, 638.2630), 519.2072 (calcd for C19H31N6O11,
519.2050), 480.1960 (calcd for C17H30N5O11, 480.1940), 432.1740
(calcd for C16H26N5O9, 432.1730), 345.1421 (calcd for C13H21N4O7,
345.1400), 274.1040 (calcd for C10H16N3O6, 274.1030), 159.0770
(calcd for C6H11N2O3, 159.0760).
ASSOCIATED CONTENT
■
S
* Supporting Information
1H and 13C NMR, neighbor joining analysis of the 28S rDNA,
and photos of the phaeofungin-producing strains. This material
is available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*Tel: 1 (732) 594-3222. Fax: 1 (732) 594-3007. E.mail: sheo.
singh@merck.com.
■
Present Address
▽WuXi AppTec Co. Ltd., Shanghai, People's Republic of
China.
Acid Hydrolysis of Phaeofungin (1). Phaeofungin (0.5 mg) was
dissolved in 0.5 mL of 6 N HCl in a Reacti-Vial and heated at 110 °C
overnight. The reaction was cooled to room temperature, and volatile
material was removed under a stream of nitrogen. To the completely
dried material was added 200 μL of a solution of 1% 1-fluoro-2,4-
dinitrophenyl-5-L-valine amide (L-FDVA) in acetone. To this solution
was added 40 μL of an aqueous solution of 1.0 M sodium bicarbonate.
This mixture in a Reacti-Vial was heated in a Reacti-Therm heating
module for 1 h. The solution was allowed to cool to room
temperature, and 20 μL of 2 M HCl was added. The solution was
degassed and analyzed by reversed-phase HPLC (Zorbax SB-C8, 2.1 ×
30 mm, 3.5 μm, elution with a 35 min linear gradient of 5−20% of
solvent A: 90:10 H2O/CH3CN containing 1.3 mM TFA and 1.3 mM
ammonium formate, and solvent B: 90:10 CH3CN/H2O containing
1.3 mM TFA and 1.3 mM ammonium formate, at a flow rate of 0.3
mL/min) and compared with the authentic samples of the Marfey’s
derivatives of the corresponding authentic amino acids. Retention
times of authentic standards were as follows: L-Ser (7.0 min), D-Ser
(9.0 min), L-Ala (12.5 min), D-Ala (23.4 min), L-Thr (7.6 min), D-Thr
(17.2 min), L-allo-Thr (8.3 min), D-allo-Thr (12.2 min), L-Asp (7.9
min), D-Asp (11.0 min).
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
We would like to thank M. Arrocho, K. Calati, and K. Furgeson
for the preliminary isolation and assay support.
■
DEDICATION
■
Dedicated to Dr. Lester A. Mitscher, of the University of
Kansas, for his pioneering work on the discovery of bioactive
natural products and their derivatives.
REFERENCES
■
(1) Hof, H. Eur. J. Clin. Microbiol. Infect. Dis. 2008, 27, 327−334.
(2) Gallis, H. A.; Drew, R. H.; Pickard, W. W. Rev. Infect. Dis. 1990,
12, 308−329.
(3) Zonios, D. I.; Bennett, J. E. Semin. Respir. Crit. Care Med. 2008,
29, 198−210.
(4) Georgopapadakou, N. H. Expert. Opin. Investig. Drugs 2001, 10,
269−280.
Partial Acid Hydrolysis of Phaeofungin (1). Phaeofungin (1) (5
mg) was dissolved in 0.2 mL of 0.5 N HCl in a Reacti-Vial and heated
at 100 °C for 30 min. The reaction mixture was allowed to cool at
room temperature, concentrated under a stream of nitrogen, and
chromatographed by reversed-phase HPLC [Zorbax RX-C8, 9.4 × 250
mm, 5 μm, elution with a 32 min linear gradient of 5−100% of solvent
A (H2O with 0.1% TFA) and solvent B (CH3CN with 0.1% TFA) at a
(5) Roemer, T.; Xu, D.; Singh, S. B.; Parish, C. A.; Harris, G.; Wang,
H.; Davies, J. E.; Bills, G. F. Chem. Biol. 2011, 18, 148−164.
(6) Bills, G. F.; Platas, G.; Fillola, A.; Jimenez, M. R.; Collado, J.;
Vicente, F.; Martin, J.; Gonzalez, A.; Bur-Zimmermann, J.; Tormo, J.
R.; Pelaez, F. J. Appl. Microbiol. 2008, 104, 1644−1658.
K
dx.doi.org/10.1021/np300704s | J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
(7) Giaever, G.; Flaherty, P.; Kumm, J.; Proctor, M.; Nislow, C.;
Jaramillo, D. F.; Chu, A. M.; Jordan, M. I.; Arkin, A. P.; Davis, R. W.
Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 793−798.
(30) Bensaci, M. F.; Gurnev, P. A.; Bezrukov, S. M.; Takemoto, J. Y.
Front. Microbiol. 2011, 2, 216.
(31) Stock, S. D.; Hama, H.; Radding, J. A.; Young, D. A.; Takemoto,
J. Y. Antimicrob. Agents Chemother. 2000, 44, 1174−1180.
(32) Cappelletty, D.; Eiselstein-McKitrick, K. Pharmacotherapy 2007,
27, 369−388.
(33) Baltz, R. H.; Miao, V.; Wrigley, S. K. Nat. Prod. Rep. 2005, 22,
717−741.
(34) Bartizal, K.; Scott, T.; Abruzzo, G.; Gill, C.; Pacholok, C.; Lynch,
L.; Kropp, H. Antimicrob. Agents Chemother. 1995, 39, 1070−1076.
(8) Xu, D.; Jiang, B.; Ketela, T.; Lemieux, S.; Veillette, K.; Martel, N.;
Davison, J.; Sillaots, S.; Trosok, S.; Bachewich, C.; Bussey, H.;
Youngman, P.; Roemer, T. PLoS Pathog. 2007, 3, e92.
(9) Jiang, B.; Xu, D.; Allocco, J.; Parish, C.; Davison, J.; Veillette, K.;
Sillaots, S.; Hu, W.; Rodriguez-Suarez, R.; Trosok, S.; Zhang, L.; Li, Y.;
Rahkhoodaee, F.; Ransom, T.; Martel, N.; Wang, H.; Gauvin, D.;
Wiltsie, J.; Wisniewski, D.; Salowe, S.; Kahn, J. N.; Hsu, M. J.;
Giacobbe, R.; Abruzzo, G.; Flattery, A.; Gill, C.; Youngman, P.;
Wilson, K.; Bills, G.; Platas, G.; Pelaez, F.; Diez, M. T.; Kauffman, S.;
Becker, J.; Harris, G.; Liberator, P.; Roemer, T. Chem. Biol. 2008, 15,
363−374.
(10) Parish, C. A.; Smith, S. K.; Calati, K.; Zink, D.; Wilson, K.;
Roemer, T.; Jiang, B.; Xu, D.; Bills, G.; Platas, G.; Pelaez, F.; Diez, M.
T.; Tsou, N.; McKeown, A. E.; Ball, R. G.; Powles, M. A.; Yeung, L.;
Liberator, P.; Harris, G. J. Am. Chem. Soc. 2008, 130, 7060−7066.
(11) Herath, K.; Harris, G.; Jayasuriya, H.; Zink, D.; Smith, S.;
Vicente, F.; Bills, G.; Collado, J.; Gonzalez, A.; Jiang, B.; Kahn, J. N.;
Galuska, S.; Giacobbe, R.; Abruzzo, G.; Hickey, E.; Liberator, P.; Xu,
D.; Roemer, T.; Singh, S. B. Bioorg. Med. Chem. 2009, 17, 1361−1369.
(12) Ondeyka, J.; Harris, G.; Zink, D.; Basilio, A.; Vicente, F.; Bills,
G.; Platas, G.; Collado, J.; Gonzalez, A.; de la Cruz, M.; Martin, J.;
Kahn, J. N.; Galuska, S.; Giacobbe, R.; Abruzzo, G.; Hickey, E.;
Liberator, P.; Jiang, B.; Xu, D. M.; Roemer, T.; Singh, S. B. J. Nat. Prod.
2009, 72, 136−141.
(13) Xu, D.; Ondeyka, J.; Harris, G. H.; Zink, D.; Kahn, J. N.; Wang,
H.; Bills, G.; Platas, G.; Wang, W.; Szewczak, A. A.; Liberator, P.;
Roemer, T.; Singh, S. B. J. Nat. Prod. 2011, 74, 1721−1730.
(14) Miyake, Y.; Kozutsumi, Y.; Nakamura, S.; Fujita, T.; Kawasaki,
T. Biochem. Biophys. Res. Commun. 1995, 211, 396−403.
(15) Takatsu, T.; Nakayama, H.; Shimazu, A.; Furihata, K.; Ikeda, K.;
Seto, H.; Otake, N. J. Antibiot. 1985, 38, 1806−1809.
(16) Mandala, S. M.; Harris, G. H. In Methods in Enzymology;
Sphingolipid Metabolism and Cell Signaling, Part A; Academic Press:
San Diego, CA, 2000; Vol. 311, pp 335−348.
(17) Mandala, S. M.; Thornton, R. A.; Milligan, J.; Rosenbach, M.;
Garcia-Calvo, M.; Bull, H. G.; Harris, G.; Abruzzo, G. K.; Flattery, A.
M.; Gill, C. J.; Bartizal, K.; Dreikorn, S.; Kurtz, M. B. J. Biol. Chem.
1998, 273, 14942−14949.
(18) Vicente, M. F.; Basilio, A.; Cabello, A.; Pelaez, F. Clin. Microbiol.
Infect. 2003, 9, 15−32.
(19) Grilley, M. M.; Stock, S. D.; Dickson, R. C.; Lester, R. L.;
Takemoto, J. Y. J. Biol. Chem. 1998, 273, 11062−11068.
(20) de Gruyter, J.; Woudenberg, J. H. C.; Aveskamp, M. M.;
Verkley, G. J. M.; Groenewald, J. Z.; Crous, P. W. Mycologia 2010, 102,
1066−1081.
(21) Zhang, Y.; Schoch, C. L.; Fournier, J.; Crous, P. W.; de Gruyter,
J.; Woudenberg, J. H. C.; Hirayama, K.; Tanaka, K.; Pointing, S. B.;
Spatafora, J. W.; Hyde, K. D. Stud. Mycol. 2009, 85−102.
(22) Zhang, Y.; Crous, P. W.; Schoch, C. L.; Hyde, K. D. Fungal
Diversity 2012, 53, 1−221.
(23) Marfey, P. Carlsberg Res. Commun. 1984, 49, 591−596.
(24) Grilley, M. M.; Stock, S. D.; Dickson, R. C.; Lester, R. L.;
Takemoto, J. Y. J. Biol. Chem. 1998, 273, 11062−11068.
(25) Fukuchi, N.; Isogai, A.; Nakayama, J.; Takayama, S.; Yamashita,
S.; Suyama, S.; Y., T. J.; Suzuki, A. J. Chem. Soc., Perkin Trans. 1 1992,
1149−1157.
(26) Di Giorgio, D.; Camoni, L.; Marchiafava, C.; Ballio, A.
Phytochemistry 1997, 45, 1385−1391.
(27) Scaloni, A.; Dalla Serra, M.; Amodeo, P.; Mannina, L.; Vitale, R.
M.; Segre, A. L.; Cruciani, O.; Lodovichetti, F.; Greco, M. L.; Fiore, A.;
Gallo, M.; D’Ambrosio, C.; Coraiola, M.; Menestrina, G.; Graniti, A.;
Fogliano, V. Biochem. J. 2004, 384, 25−36.
(28) Szabo, Z.; Budai, M.; Blasko, K.; Grof, P. Biochim. Biophys. Acta
2004, 1660, 118−130.
(29) Raaijmakers, J. M.; de Bruijn, I.; de Kock, M. J. Mol. Plant-
Microbe Interact. 2006, 19, 699−710.
L
dx.doi.org/10.1021/np300704s | J. Nat. Prod. XXXX, XXX, XXX−XXX