Isolation, structure elucidation, and biological activity of virgineone from Lachnum virgineum using the genome-wide Candida albicans fitness test

Article abstract of DOI:10.1021/np800511r

A glycosylated tetramic acid, virgineone (1), was isolated from saprotrophic Lachnum virgineum. The antifungal activity of the fermentation extract of L. virgineum was characterized in the Candida albicans fitness test as distinguishable from other natura

Full text of DOI:10.1021/np800511r

136
J. Nat. Prod. 2009, 72, 136–141
Isolation, Structure Elucidation, and Biological Activity of Virgineone from Lachnum Wirgineum
Using the Genome-Wide Candida albicans Fitness Test
John Ondeyka,† Guy Harris,† Deborah Zink,† Angela Basilio,‡ Francisca Vicente,‡ Gerald Bills,‡ Gonzalo Platas,‡
Javier Collado,‡ Antonio Gonza´lez,‡ Mercedes de la Cruz,‡ Jesus Martin,‡ Jennifer Nielsen Kahn, Stefan Galuska,†
Robert Giacobbe, George Abruzzo, Emily Hickey, Paul Liberator, Bo Jiang,§ Deming Xu,§ Terry Roemer,§,† and
Sheo B. Singh*,§
Natural Products Chemistry, Merck Research Laboratories, Rahway, New Jersey 07065, Centro de InVestigacio´n Ba´sica, Merck Sharp &
Dohme de Espan˜a, S. A., Madrid, Spain, Infectious Diseases, Merck Research Laboratories, Rahway, New Jersey 07065, and Center of Fungal
Genetics, Merck Frosst Canada Ltd., Montrel, Quebec, Canada
ReceiVed August 16, 2008
A glycosylated tetramic acid, virgineone (1), was isolated from saprotrophic Lachnum Virgineum. The antifungal activity
of the fermentation extract of L. Virgineum was characterized in the Candida albicans fitness test as distinguishable
from other natural products tested. Bioassay-guided fractionation yielded 1, a tyrosine-derived tetramic acid with a
C-22 oxygenated chain and a ꢀ-mannose. It displayed broad-spectrum antifungal activity against Candida spp. and
Aspergillus fumigatus with a MIC of 4 and 16 µg/mL, respectively. Virgineone was also identified in a number of
Lachnum strains collected from diverse geographies and habitats.
Infections caused by pathogenic fungi (e.g., Candida albicans
and Aspergillus fumigatus) are life-threatening, particularly to
immunocompromised populations.1 There are three main therapeutic
options for such infections: azoles (e.g., fluconazole),2 macrocyclic
polyenes (e.g., amphotericin),3 and candins (e.g., caspofungin,
micafungin, and anidulafungin).4 Each treatment option has limita-
tions to its utility, thus creating a need for new antifungal agents.
sources.9,10 In the reported version of the CaFT, a pool of ∼2900
heterozygous deletion strains was used in the assay. In each of the
heterozygotes, specific barcodes are introduced in the deleted allele
that enable strain identification by DNA microarrays and phenotypic
assay en masse.
CaFT profiling can be used in a number of ways to aid natural
product research. By matching CaFT profiles of crude extracts to
those in a compendium of characterized compounds, it is possible
to remove extracts containing known actives prior to chemical
fractionation. On the other hand, if a profile suggests a novel target
and/or mechanism of action, the corresponding extract could be
prioritized for chemical dereplication and fractionation, as in the
case of parnafungin, in which the original extract produced a CaFT
profile similar to that of cordycepin, but containing additional
information indicative of mechanistic novelty.9 Occasionally,
antifungal actives yield CaFT profiles that do not sufficiently reflect
the targets or the precise MOAs. Some of them nevertheless contain
enough biological information to indicate that the bioactivities are
distinguishable from known compounds and other extracts. If LCMS
analyses fail to identify known actives in these extracts, they could
then be fractionated. The CaFT provides a means to verify the
purified bioactivities.
Natural products have provided many of the antifungal leads and
drugs and continue to be the most prolific source for novel and
chemically diverse leads. Rediscovery of known compounds,
however, poses a significant challenge in natural product research
and calls for ingenious dereplication strategies enabling efficient
discovery of new bioactive compounds. Critical to this is the fusion
of new sources of natural products and bioactivity detection in crude
extracts. In a two-prong approach, we expanded our efforts to isolate
microorganisms from diverse geographical regions and habitats in
combination with improved high-throughput fermentation methods.5
In parallel, we adapted a chemogenetic assay in C. albicans, the
fitness test, that profiles the biological activities (targets and/or
mechanisms of action, in many cases) of antifungal compounds
and extracts.6-8 Combining these two resources, we discovered,
from fungi, a new class of novel antifungal compounds, parnafun-
gins, that inhibit mRNA polyadenylation with a broad spectrum of
antifungal activity with in ViVo efficacy in a murine model of
candidiasis.9,10
Applying the CaFT to crude extracts from microbial sources,
we identified a number of natural product extracts that were
predicted to possess novel bioactivities. Here we report the isolation,
structural elucidation, and biological activity of a new antifungal
compound, virgineone (1), produced by the fungus Lachnum
Virgineum.
The C. albicans fitness test (CaFT) exploits the diploid nature
of the pathogen. For the majority of genes, deletion of one allele is
phenotypically neutral. However, antiproliferative compounds, at
sublethal concentrations, may induce variations in growth that are
restricted to only a subset of the heterozygotes. As demonstrated,
target- or mechanism-specific inhibitors elicit phenotypic variations
(i.e., hypersensitivity and resistance) in heterozygotes that are
indicative of mechanisms of action (MOAs).8 Conversely, the
collective responses (i.e., “signature” profiles) of heterozygotes may
reflect different aspects of MOAs of synthetic compounds11 and
bioactive compounds present in the crude extracts from natural
Results and Discussion
* Corresponding author. Fax: 1(732) 594-6880. E-mail: sheo_singh@
Identification of a Novel Bioactivity Produced by Lachnum
Wirgineum and Related Lachnum Strains. Screening of micro-
bial natural product extracts against wild-type C. albicans identified
a large number of extracts that showed growth inhibition of C.
albicans. While testing these broths in the CaFT, we identified
merck.com.
† NPC, Rahway.
‡ CIBE, Spain.
Infectious Diseases, Rahway.
§ Center of Fungal Genetics, Merck Frosst, Canada.
10.1021/np800511r CCC: $40.75
2009 American Chemical Society and American Society of Pharmacognosy
Published on Web 12/30/2008

Biological ActiVity of Virgineone
Journal of Natural Products, 2009, Vol. 72, No. 1 137
several extracts that yielded highly related CaFT profiles that were
distinct from the larger screening set and consisted largely of
heterozygous deletion strains for genes involved in multiple cellular
processes (Figure 1). Of particular interest, it is likely that the two
extracts, F-230,270 and F-230,308, elicited the calmodulin-depend-
ent stress response, as suggested by induced hypersensitivity of
heterozygotes for calmodulin (CMD1), calcineurin (the regulatory
subunit of the phosphatase, CMP1), heat-shock protein 90 (HSP90),
and its associated cochaperones (SIS1, HSP104, and CPR6).
Although the components of this stress response pathway are
targeted by known compounds derived from microbial sources,
including FK506 (produced by Streptomyces tsukubaensis), cy-
closporine A (Tolypocladium inflatum), and radicicol (Pochonia
chlamydosporia), all of them and rapamycin (S. hygroscopicus,
targeting the TOR pathway) generated largely nonoverlapping CaFT
profiles (Figure 1 and data not shown). Furthermore, the collective
hypersensitivity of strains for the aforementioned genes, together
with two C. albicans specific genes (orf19.2049 and orf19.3405),
and the resistance of HCH1 (for another HSP90 cochaperone)
heterozygote are restricted to these three and other related extracts
(see below). They showed no responses to other compounds or
natural products tested in the CaFT (data not shown). These data
suggested that the biological activities and the compounds within
these extracts were likely unique within the screening set examined.
The extract of fungus F-230,270 was selected for bioassay-guided
isolation of the active principle.
The producer fungal strain F-230,270 was isolated from an
unidentified plant in the province of Santa Cruz, Argentina. It was
further identified as Lachnum Virgineum by ITS and 28S rDNA
analysis (Supporting Information, Figures S1 and S2). The anti-
fungal activity of its metabolites was first detected in a fermentation
extract grown in a microfermentation array employing eight media,5
only one of which yielded significant antifungal activity against
wild-type C. albicans (data not shown). Further screening encoun-
tered at least nine other highly related Lachnum strains that produced
the same active component (Supporting Information, Table S1).
Isolation and Structure Elucidation of Virgineone (1). A
fermentation of F-230,270 was scaled up to 1 L and extracted with
acetone. A portion of the acetone extract, equivalent to 50 mL of
broth, was subjected to a standardized, prioritization scale frac-
tionation on Mitsubishi CHP20Y. A single antifungal activity
eluting in fractions 10-15 (of 28 in total) was obtained. Extracts
from the related culture F-230,308 exhibited similar behavior when
subjected to prioritization scale fractionation with activity eluting
in fractions 13-15. Interestingly, small-scale elution-extrusion
countercurrent fractionation (EECCC)10 of the rich cuts from both
extracts resulted in stationary phase disruption and hence no
separation. Scale-up isolation was accomplished by capture on
Amberchrom followed by successive chromatographies of the
antifungal active fractions on Sephadex LH20 and reversed-phase
C8 HPLC to afford 326 mg (326 mg/L) of virgineone (1) as a
colorless powder.
Figure 1. CaFT profiles of crude fungal extracts (F-230,270 and
F-230,308), virgineone (1), radicicol, and rapamycin. Two inde-
pendent CaFT experiments with comparable inhibitory concentra-
tions (ICs) were selected for each active. The normalized z-scores
(statistic appraisal of the responsiveness,8 hypersensitivity with
z-score >0, and resistance <0) were compiled and analyzed by
Cluster 3.0. The strains were selected with the absolute value of
z-scores no less than 3.5 in at least two experiments. They (74 in
total) were grouped by hierarchical clustering using the centroid
linkage method and displayed in TreeView with the scale of the
heat map (z-scores) indicated in the top right corner. The hierarchi-
cal linkage of each experiment is shown on the top. The highlighted
strains include those that are specific to these extracts and virgineone
(red boxes; see text for detailed description), radicicol (green box,
described in ref 8, with the exception of STI1, encoding a
cochaperone of HSP90), and rapamycin (blue box, the TOC
complex, * indicating an independent heterozygote in which the 5′
portion of the ORF was deleted). Also highlighted, in blue (bold),
are components of the INO80-chromatin remodeling complex. Other
independently constructed strains are indicated by two asterisks (**).
The UV spectrum of 1 showed absorption maxima at λmax 201,
229, 242 (sh), and 281 nm. HRESIFTMS produced a formula of
C40H63NO12. The 13C NMR spectrum displayed many overlapping
signals particularly in the aliphatic region of the spectrum but
supported the overall formula. The IR spectrum exhibited absorption
bands for hydroxy groups, an aliphatic chain, and a ketone. The
13C NMR spectrum in DMSO-d6 displayed 28 (12 overlapping)
resonances, which were eventually assigned to 20 methylenes, 12
methines, and one methyl by a DEPT spectrum. The remaining
1
seven were quaternary carbons (Table 1). The H NMR spectrum
supported these assignments, and careful integration indicated the
presence of 63 protons. The 1H NMR spectrum showed the presence
of four aromatic protons as two pairs of doublets at δH 6.56 (2H,
d, J ) 8.0 Hz) and 6.92 (2H, d, J ) 8.0 Hz), which displayed
DFQ-COSY cross-peaks to each other (Table 1). The methine at

138 Journal of Natural Products, 2009, Vol. 72, No. 1
Ondeyka et al.
Table 1. 1H (500 MHz) and 13C (125 MHz) NMR Assignments of Virgineone (1) and Aglycone (2)
1 δH
(mult, J in Hz)
DMSO-d6
1 δH
(mult, J in Hz)
CD3OD
2 δH
(mult, J in Hz)
DMSO-d6
1 δC
DMSO-d6
HMBC (HfC)
DMSO-d6
2 δC
DMSO-d6
position
1
2
3
4
NH
CH
C°
7.33 (s)
3.57 (brt, 5.5)
C-2, 3, 4
C-3, 5, 2′
60.8
193.6
99.6
3.74 (dd, 7.5, 4.5)
62.1
191.7
100.4
175.5
194.5
35.8
33.0
26.6
28.5-29.2
23.2
4.04 (m)
C°
5
6
7
8
177.1
197.5
38.4
33.0
27.0
29.0
23.3
41.8
C°
C°
CH
CH2
CH2
CH2
CH2
CH2
3.65 (m)
C-6, 8, 9, 28
3.67 (m)
3.42 (m)
1.4-1.5 (m)
1.2 (m)
1.2 (m)
1.42 (m)
2.36 (m)
1.05 (m) 1.50 (m)
1.20 (m)
0.95 (m) 1.45 (m)
1.24-1.29 (m)
1.24-1.29 (m)
1.52 (m)
9
10-13
14
15
1.20 (m)
1.40 (m)
2.34 (apparent q,
7.0)
C-14, 15, 10-13
C-16, 12-13, 15
C-13, 14, 16
2.43 (m)
41.8
16
17
210.0
41.8
C°
CH2
210.5
41.8
2.34 (apparent q,
7.0)
C-16, 18, 19
2.43 (m)
2.36 (m)
18
19-24
25
26
27
23.3
29.0
33.4
69.0
73.4
CH2
CH2
CH2
CH
1.40 (m)
1.20 (m)
C-16, 17, 19
1.52 (m)
23.2
28.5-29.2
33.4
71.0
66.0
1.42 (m)
1.2 (m)
1.4-1.5 (m)
3.36 (m)
1.24-1.29 (m)
1.18 (m) 1.65 (m)
3.73 (m)
3.43 (dd, 10, 6.0)
3.86 (dd, 10, 2.5)
1.20 (m) 1.35 (m)
3.53 (m)
3.28 (dd, 10.0, 7.0)
3.62 (dd, 10.0, 4.0)
0.82, (d, 7.0)
2.59 (dd, 13.5, 6.0)
2.78 (dd, 13.5, 3.5)
C-26, 23-24, 27
C-25, 26, 1′′ C-25, 1′′
CH2
3.20 (2H,m))
28
1′
17.6
36.8
CH3
CH2
C-6, 7, 8
C-2, 3, 2′, 3′ C-2, 3, 2′, 3′
17.0
35.8
0.95 (d, 7.0)
2.82 (brd, 5.0)
2.98 (brd, 14.5)
2.74 (br)
2′
127.6
130.4
114.6
155.5
100.6
70.4
73.6
67.1
77.5
C°
125.6
130.6
114.7
155.9
3′/7′
4′/6′
5′
1′′
2′′
3′′
4′′
5′′
CH
CH
C°
CH
CH
CH
CH
CH
6.92 (d, 8.0)
6.56 (d, 8.0)
C-1′, 7′/3′, 4′/6′, 5′
C-2′, 5′, 6′/4′
7.00 (d, 8.5)
6.62 (d, 8.5)
6.97 (t, 8.5
6.65 (t, 8.5)
4.36 (brs)
3.64 (br)
3.23 (br)
3.26 (m)
3.0 (ddd, 9.0, 6.0,
2.0)
C-27, 2′′, 5′′
C-1′′, 3′′, 4′′
C-1′′, 3′′, 5′′
C-1′′, 3′′, 5′′
C-1′′, 3′′, 4′′, 6′′
4.55 (d, 1.0)
3.88 (dd, 3.0, 1.0)
3.45 (dd, 9.5, 3.0)
3.55 (t, 9.5)
3.20 (ddd, 9.5, 6.0,
2.5)
6′′
61.4
CH2
3.42 (m) 3.66 (m)
3.70 (dd, 12.0, 6.0)
3.87 (dd, 12.5, 3.0)
4′-OH
2′′-OH
3′′-OH
4′′-OH
6′′-OH
OH
OH
OH
OH
OH
9.95 (br)
4.22 (br)
4.5 (br)
4.67 (d, 5.0)
4.41 (br)
C-4′/6′, 5′
9.15 (brs)
C-2′′, 3′′
C-3′′, 4′′, 5′′
δH 3.57 (1H, brt, J ) 5.5 Hz) exhibited DFQ-COSY correlations
to the benzylic methylene protons resonating at δΗ 2.78 (1H, dd, J
) 13.5, 3.5 Hz) and δΗ 2.59 (1H, dd, J ) 13.5, 6.0 Hz). The
benzylic methylene protons displayed HMBC correlations to the
methine carbon C-2 (δC 60.8), C-3 (δC 193.6), and the aromatic
carbons C-2′ (δC 127.6) and C-3′ (δC 130.4); the phenolic OH group
resonated at δH 9.95, which showed HMBC correlation to C-4′/6′
and C-5′, thus establishing the presence of a para-hydroxy benzyl
group connected to a methine.
δC 100.6), five oxymethines, and two oxymethylenes. DQF-COSY
correlations of these protons suggested the presence of a hexose
moiety and confirmed the presence of the remaining oxymethine
and oxymethylene at the terminal end of the aliphatic chain. The
2′′ proton of the hexose moiety showed small couplings with H-1′′
(J ) 1 Hz) and H-3′′ (J ) 3 Hz), consistent with the presence of
an axial hydroxy group at C-2′′ and, therefore, suggesting that the
hexose moiety was ꢀ-mannose. This conclusion was corroborated
by coupling constants measured in CD3OD (Table 1) and is
consistent with values reported for mannosyl tetramic acids.13,14
Like many tetramic acids, the NMR spectrum of 1 exhibited signal
broadenings12 for a large number of resonances and required both
solvents (DMSO-d6 and CD3OD) for critical assignments. The
anomeric proton (δH 4.36) displayed HMBC correlations with C-2′′
and C-5′′ and C-27, confirming the glycosidation at C-27.
HRESIFTMS analysis showed a major fragment ion at m/z
588.3881 due to loss of the mannose moiety. The exact location of
the ketone group in the chain could not be assigned by NMR
methods due to degeneracy of the 1H and 13C NMR chemical shifts,
but was assigned at C-16 by HRESIFTMS-MS fragmentation of
the aglycone 2, which was prepared by acid hydrolysis of 1. The
HRESIFTMS analysis of 2 displayed a protonated parent ion at
m/z 588.3886 and key fragment ions at m/z 372.2172 and 400.2121
due to cleavages of the C-C bonds before (C15/C16) and after
(C16/C17) the keto group (Figure 2). The origin of these fragment
ions was confirmed by MS-MS analysis. On the basis of the
spectroscopic data, the acyl tetramic acid structure 1 was assigned
for virgineone.
The 13C NMR spectrum showed three keto carbons (δC 193.6,
177.1, and 197.5) and a upfield shifted quaternary carbon (δC 99.6),
which are generally present in tetramic acid moieties.12 The methine
(C-2) resonance at δC 60.8 and a nitrogen in the formula of 1
corroborated the presence of a tetramic acid moiety, which was
confirmed by HMBC correlations of NH (δH 7.33) to C-2, C-3 (δC
193.6) and C-4 (δC 99.6). The methine proton H-2 exhibited HMBC
correlations to C-3, C-5 (δC 177.1) and C-2′, confirming the
presence of a para-hydoxy benzyl tetramic acid moiety. The
downfield shifted methine proton H-7 (δH 3.65, m) showed DQF-
COSY correlations to the methyl group (δH 0.82, d, J ) 7 Hz) and
a pair of methylene proton multiplets (δH 1.05 and 1.50). These
methylene protons displayed DQF-COSY correlations to the
methylene protons that resonated as a part of the methylene
envelope. The methyl group H3-28 showed HMBC correlations to
C-6 (δC 197.5), C-7 (δC 38.4), and C-8 (δC 33.0), and H-7 displayed
HMBC correlations to C-6, C-8, C-9, and C-28, thus establishing
the connectivity of the tetramic acid moiety with the aliphatic chain.
Analysis of the remaining resonances suggested the presence of
an aliphatic ketone (δC 210.0), an anomeric methine (δH 4.36 and

Biological ActiVity of Virgineone
Journal of Natural Products, 2009, Vol. 72, No. 1 139
Table 2. Antifungal Activity (MIC, µg/mL) of Virgineone (1)
and Caspofungin
test organism
C. albicans
C. albicans + 50% mouse serum MY1055
C. glabrata
C. parapsilosis
C. lusitaniae
C. krusei
strain no. virgineone caspofungin
MY1055
8
>32
8
8
4
0.5
0.25
0.25
0.5
<0.03
1
MY1381
ATCC22019
MY1396
ATCC6258
MF5668
16
8
A. fumigatus
32
A. fumigatus + 50% human serum MF5668
>32
32
32
>32
T. mentagrophytes
S. aureus
MF7004
MB2865
8
16
S. cereVisiae Lys14p (CGD, www.candidagenome.org) but not
essential for viability.17 Another group of genes are for components
of the INO80 chromatin remodeling complex, including INO80,
ARP8, ARP4, ACT1, RVB1, and RVB2, some of which are more
responsive to the extracts (Figure 1). Although none of these genes
likely correspond to the target of virgineone, they seem to reflect
the stress responses (at different levels, e.g., transcription, heat-
shock, and drug tolerance responses) elicited by the antiproliferative
effect of this compound such that heterozygosity at these loci results
in reduced growth.
Virgineone (1) is likely derived from condensation of tyrosine
and a C-24 ꢀ-keto acid. The left-hand portion of 1 has been reported
as a structural part of several natural products exemplified by
militarinone C (3).15 The dihydroxy C-22 long acyl alkyl chain
with or without a ketone group has not been reported. Epicocca-
mides A (4) and D (5), C-18 and C-20 ꢀ-keto acid and alanine
derived tetramic acids, have been reported.13,14 Militarinone C was
isolated from the fungus Cordyceps militaris and has been shown
to have very weak neuritogenic activity in PC-12 cells.15 Epicoc-
camide A (4) was isolated from jellyfish-derived strains of
Epicoccum purpurascens, and epicoccamide D (5) was isolated from
an Epicoccum sp. associated with the wood-decay fungus Pholiota
squarrosa.13,14 These compounds demonstrated cytotoxic activities
against the HeLa cell line.
Antifungal Activity of Virgineone (1). The CaFT profile of
purified virgineone (Figure 1) was consistent with that originally
observed with crude acetone extracts. Thus the in Vitro C. albicans
growth inhibition assay was an appropriate surrogate bioassay for
isolation of the active component. Virgineone exhibited broad-
spectrum antifungal activity against a number of key pathogenic
fungal strains (Table 2). There was little variation in potency among
the Candida spp. examined, with MIC values of 4-16 µg/mL.
Virgineone was equally potent against two filamentous fungi,
Aspergillus fumigatus and Trichophyton mentagrophytes. The in
Vitro antifungal activity of 1 was reduced at least 4-fold in the
presence of mouse or human serum. It was tested for efficacy in
an in ViVo disseminated C. albicans MY1055 mouse (DBA/2) model
at 100 and 50 mg/kg dosed twice daily intraperitonially (ip). It
showed 100% mortality at 100 mg/kg and 40% mortality at 50 mg/
kg dose levels. No reduction in fungal cfu was observed in surviving
mice compared to the control, suggesting lack of efficacy. Similar
dosing of uninfected mice did not show any mortality perhaps due
to immune modulation. Caspofungin was used as a control. The
aglycone (2) did not show any antifungal activity at 32 µg/mL.
CaFT profiling mechanistically differentiates virgineone from
other natural products (radicicol, cyclosporine A, FK506, and
rapamycin) that target the calmodulin-calceneurin-HSP90 path-
way16 (Figure 1 and data not shown). This was supported by the
lack of potentiation of fluconazole by 1, which was achieved in
the combination of fluconazole with cyclosporine A (data not
shown). However, it is not possible to predict the precise target or
MOA of 1 based solely on the CaFT profiles. Nonetheless, they
contain additional genes that are of interest. Noticeably, one of the
C. albicans specific genes, orf19.2049, encodes for potential
membrane protein, while the other, orf19.3405, has been recently
described as a zinc-finger transcription factor (Zcf18p) related to
Production of 1 by Lachnum Wirgineum and Other Lach-
num Strains. The presence of 1 in the other extract (F-230,308)
was confirmed by HRLCMS (data not shown), as predicted by the
almost identical CaFT profiles (Figure 1), demonstrating the
dereplication potential of the assay to correctly predict the major
bioactive components within complex natural product mixtures.
Once the chemical properties of 1 were characterized, database
matching was performed on new antibiotic extracts using an in-
house application where the DAD, retention time, and POS and
NEG mass spectra of the active samples were compared to the
UV-LC-MS data of the authentic standard stored in a proprietary
database. An additional seven strains from the United States, the
Republic of Georgia, and Spain were recognized as producing the
same tetramic acid (1) (Supporting Information, Table S1). DNA
barcoding with the 28S rDNA indicated that these strains were all
highly related and possibly conspecific, although they originated
from widely different geographic locations. This group of strains
also included an eighth apparently conspecific strain (F-262,581)
that showed antibiotic activity against S. aureus, but in which 1
was not detected (Supporting Information, Table S1, Figure S2).
None of these strains sporulated in agar culture. They were generally
characterized as relatively slow growing, with waxy to appressed,
pale buff to whitish gray mycelial colonies, and their identifications
were not obvious. For the most part, they were associated with
living plant or decaying plant material. Only F-265,955 was
obtained from an unidentified lichen thallus (Supporting Informa-
tion, Table S1). The relative placement in the 28S barcode tree
indicated that these fungi are members of the Helotiales and belong
to the genus Lachnum (Supporting Information, Figure S2).
Similarity searches with ITS sequences from F-230,270, the original
producing strain from Argentina, indicated that it may be conspecific
with strains identified as Lachnum Virgineum.
A large number of structurally diverse tetramic acid derivatives
have been isolated from a variety of organisms. They have
demonstrated antibacterial, antiviral, and antineoplastic activities;
however, the precise mechanisms of action have not been deter-
mined for most of them.18 Antifungal activities have been described
for only a few tetramic acids.19-21 Isolation of tetramic acid 1 was
based on its intrinsic antifungal activity and the distinct mechanistic
profile in the C. albicans fitness test. Virgineone (1) contains a
tyrosine-derived tetramic acid, a C-22 oxygenated chain, and a
ꢀ-mannose, and is active against a broad spectrum of fungal
pathogens. Although the molecular mechanism of virgineone has
not been established, the CaFT profiles suggest that its antiprolif-
Figure 2. Mass fragmentation of 1 and 2.

140 Journal of Natural Products, 2009, Vol. 72, No. 1
Ondeyka et al.
fermentations were extracted with an equal volume of acetone and
pooled. A 4 mL aliquot was frozen and tested in the CaFT, and the
remaining extract was used for isolation of active metabolites.
Genome-Wide Candida albicans Fitness Test. The C. albicans
fitness test was performed as described previously.8 Briefly, 5 mL
cultures of the pool (at the initial OD600 of 0.025) were treated with
antifungal actives (compounds and extracts) at multiple concentrations
together with the mock treatment. The active inhibitory concentration
of each culture was determined after 15 h at 30 °C. Those with desirable
ICs were retained and diluted to OD600 0.05 with medium containing
the antifungal active at the original concentrations. After another 23 h
of growth, cell pellets were collected, and total genomic DNA was
prepared. DNA samples of treated and mock cultures were PCR
amplified using common primers that flank the strain-specific barcodes
and labeled. Mixtures of labeled barcodes were hybridized against DNA
microarrays. The relative responses (hypersensitivity and resistance)
of each strain in the treated culture were appraised using an error-
modeling statistic framework that involves a set of ∼50 reference
compounds and expressed by normalized z-scores of both up- and
downstream barcodes (see ref 8 for details). For each strain, the z-score
with higher absolute value was selected to compile the final list. Cluster
3.0 was used to analyze multiple fitness test experiments, the result of
which was displayed with TreeView. Both are available at http://
bonsai.ims.u-tokyo.ac.jp/∼mdehoon/software/cluster/software.htm.
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 D2 variable domains was amplified with primers NL1 and NL4
or LROR and LR6. Sequences of 28S rDNA were used to generate a
neighbor-joining tree (Supporting Information, Figure S1) that dem-
onstrated the close relationships among the virgineone-producing strains
and that they likely belong to the genus Lachnum. To achieve further
phylogenetic resolution, the same genomic DNA samples were used
to generate sequences of the intertranscribed spacer regions and 5.8S
gene of the rDNA (ITS).
erative activity elicits multiple stress responses in C. albicans.
Furthermore, the cytotoxicity of virgineone is not restricted to fungi.
The discovery of 1 was achieved by chemogenetic profiling of
crude natural products, in the form of extracted fermentation broths.
In the CaFT, the biological activities in the unfractionated extracts
are detected on the basis of induced growth variation in heterozy-
gous deletion strains that cover a large fraction of the C. albicans
genome. By comparing the profiles of natural products in question
with those of known bioactive molecules, it is possible to discern
the signature chemotypes present in the crude extract based on
deduced bioactivities. This approach is particularly fruitful when
combined with natural products from new sources. As demonstrated
in this report, distinct CaFT profiles can reflect distinct biological
activities, which, in turn, suggest distinct chemotypes and, likely,
chemical entities in the extracts. We believe that the genome-wide
fitness test affords a solution to biologically dereplicate crude
extracts such that chemical resources can be allocated to those that
are most likely to contain new chemical entities.
Experimental Section
General Experimental Procedures. Optical rotations were recorded
with a Perkin-Elmer 241 polarimeter. UV spectra were recorded on a
Perkin-Elmer Lambda 35 spectrometer. IR spectra were recorded with
a Perkin-Elmer Spectrum One FT-IR spectrophotometer. All NMR
spectra were recorded with a Varian Unity 500 (1H, 500 MHz; 13C,
125 MHz) spectrometer in DMSO-d6 or CD3OD. Chemical shifts are
reported in δ (ppm) using residual solvent signals (for DMSO-d6: δH
2.53 and δC 39.51 ppm; for CD3OD: δH 3.30 and δC 49.00) as internal
standards. 1H, 13C, COSY, DQF-COSY, DEPT, gHSQC, gHMBC, and
TOCSY spectra were measured using standard Varian pulse sequences.
LRMS data were recorded on an Agilent 1100 MSD with ES ionization,
and HREIMS were obtained on a Thermo Finnigan LTQ-FTMS
spectrometer. An Agilent HP 1100 instrument was used for analytical
HPLC.
Fungal Material. The fungal strain F-230,270 (E-000509065, Merck
Research Laboratory) was isolated from an unidentified plant collected
in the province of Santa Cruz, Argentina (Supporting Information, Table
S1). Additional Lachnum strains that produced virgineone are listed in
Table S1 (Supporting Information). Strains were maintained as frozen
mycelium in 10% glycerol at -80 °C.
Fermentations in Nutritional 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 Staphylococcus 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, 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.5 Each week, 160 to 240 fungal strains were
selected for fermentations. 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 discs were cut from each 60 mm plate. Mycelia discs
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 rpm, 5 cm
throw), and rotation of the cover glasses continually sheared hyphae
and mycelial disc fragments to produce homogeneous hyphal suspen-
sions. 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 8-media nutritional arrays in a ten-column × eight-row
pattern.5 Nutritional arrays were grown statically 21 days at 22 °C.
The detection of antifungal activity from strain F-230,270 originated
from a 1 mL fermentation in the medium SCAS (soluble starch 40 g;
casein hydrolysate 5 g; KH2PO4 0.5 g; MgSO4 · 7H2O 0.5 g;
FeSO4 ·7H2O 0.01 g; distilled H2O 10 000 mL). This fermentation was
scaled up to 1 L by growing F-230,270 in 500 mL flasks with 150 mL
of liquid SCAS agitated at 220 rpm, 22 °C for 22 days. The liquid
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 the Bigdye Terminators
version 1.1 (Perkin-Elmer, Norwalk, CT) 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 (Gene-
studio, Inc.), and consensus sequences were aligned by the same
software. Neighbor-joining analyses were used to approximate phylo-
genetic relationships among strains.
Extraction and Isolation. An 1 L fermentation (pH 6.5) was
extracted with 1 L of acetone and filtered through Celite, the filtrate
was concentrated under reduced pressure to remove most of the acetone,
and the extract was loaded onto a 50 cm3 Amberchrom column packed
in H2O. The column was eluted with a 100 min linear gradient of H2O/
MeOH at a flow rate of 5 mL/min followed by 20 min elution with
MeOH. The fractions eluted with 80-100% MeOH showed antifungal
activity. These fractions were combined, concentrated under reduced
pressure to remove organic solvents, and lyophilized to yield 650 mg
of material. This material was fractionated on a 500 cc Sephadex LH20
column packed and eluted with MeOH. The activity was spread and
eluted in 0.5 to 1 cv (column volume), which upon concentration
produced 650 mg of material, suggesting no separation. This material
was purified by RP-HPLC using a Zorbax C8 (250 × 21.2 mm) column
at a flow rate of 12 mL/min with a 45 min linear gradient of 10-90%
aqueous CH3CN with 0.1% TFA. Fractions 22-25 were pooled from
six independent HPLC runs and lyophilized to afford 326 mg (326
mg/L) of virgineone (1) as a colorless powder: [R]23 -95.7 (c 0.3,
D
MeOH); UV (MeOH) λmax 201.4 (log ꢀ 4.36), 229 (4.13), 242 sh (4.07),
281 (4.22) nm; IR (ZnSe) νmax 3340, 2925, 2853, 1604, 1515, 1479,
1
1228, 1073, 1026 cm-1; H and 13C NMR data, see Table 1; ESIMS
m/z 750 [M + H]+; HRESIFTMS m/z 750.4423 (calcd for C40H63NO12
+ H, 750.4384), 588.3881 (calcd for C34H53NO7 +H, 588.3900).
Hydrolysis of Virgineone. A solution of 1 (6 mg) in 6 N methanolic
HCl (1 mL) was heated at 60 °C for 2 h. The reaction mixture was
concentrated to dryness, redissolved in MeOH, and separated on a
reversed-phase preparative HPLC using a Zorbax C8 (250 × 21.2 mm)
column at a flow rate of 12 mL/min with a 45 min linear gradient of

Biological ActiVity of Virgineone
Journal of Natural Products, 2009, Vol. 72, No. 1 141
(6) Haselbeck, R.; Wall, D.; Jiang, B.; Ketela, T.; Zyskind, J.; Bussey,
H.; Foulkes, J. G.; Roemer, T. Curr. Pharm. Des. 2002, 8, 1155–72.
(7) Roemer, T.; Jiang, B.; Davison, J.; Ketela, T.; Veillette, K.; Breton,
A.; Tandia, F.; Linteau, A.; Sillaots, S.; Marta, C.; Martel, N.;
Veronneau, S.; Lemieux, S.; Kauffman, S.; Becker, J.; Storms, R.;
Boone, C.; Bussey, H. Mol. Microbiol. 2003, 50, 167–81.
(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, 363–74.
(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, 7060–6.
(11) Rodriguez-Suarez, R.; Xu, D.; Veillette, K.; Davison, J.; Sillaots, S.;
Kauffman, S.; Hu, W.; Bowman, J.; Martel, N.; Trosok, S.; Wang,
H.; Zhang, L.; Huang, L. Y.; Li, Y.; Rahkhoodaee, F.; Ransom, T.;
Gauvin, D.; Douglas, C.; Youngman, P.; Becker, J.; Jiang, B.; Roemer,
T. Chem. Biol. 2007, 14, 1163–75.
(12) Singh, S. B.; Zink, D. L.; Goetz, M. A.; Dombrowski, A. W.;
Polishook, J. D.; Hazuda, D. J. Tetrahedron Lett. 1998, 39, 2243–46.
(13) Wangun, H. V. K.; Dahse, H.-M.; Hertweck, C. J. Nat. Prod. 2007,
70, 1800–03.
(14) Wright, A. D.; Osterhage, C.; Konig, G. M. Org. Biomol. Chem. 2003,
1, 507–10.
(15) Schmidt, K.; Riese, U.; Li, Z.; Hamburger, M. J. Nat. Prod. 2003,
66, 378–83.
(16) Imai, J.; Yahara, I. Mol. Cell. Biol. 2000, 20, 9262–70.
(17) Nobile, C. J.; Mitchell, A. P. Curr. Biol. 2005, 15, 1150–55.
(18) Schobert, R.; Schlenk, A. Bioorg. Med. Chem. 2008, 16, 4203–21.
(19) Gunasekera, S. P.; Gunasekera, M.; McCarthy, P. J. Org. Chem. 1991,
56, 4830–33.
10-90% aqueous CH3CN with 0.1% TFA. The fraction eluting at 20
min was lyophilized to afford 3 mg of 2 as a colorless powder: [R]23
D
-60 (c 0.3, MeOH); UV (MeOH) λmax 201.4 (log ꢀ 3.94), 229 (3.72),
242 sh (3.66), 281 (3.88) nm; IR (ZnSe) νmax 3248, 2928, 285X, 1616,
1511 cm-1; HRESIFTMS m/z 588.3886 (calcd for C34H53NO7 + H,
588.3900), 400.2121 (calcd for C23H30NO5, 400.2124), 372.2172 (calcd
for C22H30NO4, 372.2175).
Antifungal Assay. The MIC (minimum inhibitory concentration)
against each fungal strain was determined as previously described.22
Cells were inoculated at 105 colony-forming units/mL followed by
incubation at 37 °C for 20 h with a 2-fold serial dilution of compound
in the growth medium.
In ViWo Efficacy Assay. Mice were infected iv (lateral tail veins)
with Candida albicans MY1055 at 2.51 × 104 cfu/mouse. Five DBA/2
mice in each group were treated with virgineone (1) formulated in 5%
DMSO in sterile distilled water and treated at 100 and 50 mg/kg twice
daily by ip administration. Mice were observed for mortality and general
health for 4 days after challenge. At day 4 after challenge mice were
euthanized, both kidneys were aseptically removed, placed in sterile
Whirl-Pak bags, weighed, and then homogenized in 5 mL of sterile
saline. Kidney homogenates were then serially 10-fold diluted in
sterile saline and plated on SDA. Plates were incubated at 35 °C and
counted after 30 to 48 h. Colony forming units (cfu) per gram of kidney
were determined and counts from treatment groups compared to counts
from sham-treated controls using a paired two tailed t-test (Excel).
Acknowledgment. We thank M. Arocho, K. Calati, K. Ferguson,
and J. Occi for the preliminary isolation and assay support.
Supporting Information Available: 1H and 13C NMR, description
of other producing strains (Tables S1), and phylogenetic tree (Figures
S1 and S2). This material is available free of charge via the Internet at
http://pubs.acs.org.
References and Notes
(1) Hof, H. Eur. J. Clin. Microbiol. Infect. Dis. 2008, 27, 327–34.
(2) Zonios, D. I.; Bennett, J. E. Semin. Respir. Crit. Care Med. 2008, 29,
198–210.
(20) Wang, C. Y.; Wang, B. G.; Wiryowidagdo, S.; Wray, V.; van Soest,
R.; Steube, K. G.; Guan, H. S.; Proksch, P.; Ebel, R. J. Nat. Prod.
2003, 66, 51–6.
(21) Hellwig, V.; Grothe, T.; Mayer-Bartschmid, A.; Endermann, R.;
Geschke, F. U.; Henkel, T.; Stadler, M. J. Antibiot (Tokyo) 2002, 55,
881–92.
(3) Gallis, H. A.; Drew, R. H.; Pickard, W. W. ReV. Infect. Dis. 1990,
12, 308–29.
(4) Georgopapadakou, N. H. Expert Opin. InVestig. Drugs 2001, 10, 269–
80.
(5) 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, 1644–58.
(22) Bartizal, K.; Scott, T.; Abruzzo, G.; Gill, C.; Pacholok, C.; Lynch,
L.; Kropp, H. Antimicrob. Agents Chemother. 1995, 39, 1070–76.
NP800511R