Two Plant-Derived Aporphinoid Alkaloids Exert Their Antifungal Activity by Disrupting

28 Pages·2017·2.07 MB·English

JBC Papers in Press. Published on August 18, 2017 as Manuscript M117.781773
The latest version is at http://www.jbc.org/cgi/doi/10.1074/jbc.M117.781773
Novel antifungal mechanism of aporphinoid alkaloids

Two Plant-Derived Aporphinoid Alkaloids Exert Their Antifungal Activity by
Disrupting Mitochondrial Iron-Sulfur Cluster Biosynthesis

Siddharth K. Tripathi‡, Tao Xu‡1, Qin Feng‡, Bharathi Avula‡, Xiaomin Shi‡‡2, Xuewen Pan‡‡3,
Melanie M. Mask§, Scott R. Baerson§, Melissa R. Jacob‡, Ranga Rao Ravu‡, Shabana I. Khan‡║,
Xing-Cong Li‡║, Ikhlas A. Khan‡║, Alice M. Clark‡║, and Ameeta K. Agarwal‡¶*

From the ‡National Center for Natural Products Research, the ║Division of Pharmacognosy and the
¶Division of Pharmacology, Department of Biomolecular Sciences, School of Pharmacy, University of
Mississippi, University, MS 38677, the ‡‡Verna and Marrs McLean Department of Biochemistry and
Molecular Biology, Baylor College of Medicine, Houston, TX 77030, and the §United States Department
of Agriculture, Agricultural Research Service, Natural Products Utilization Research Unit, University,
MS 38677

1Present address: Pathology Department, Medical School, University of Michigan, Ann Arbor, MI 48109
2Present address: First Affiliated Hospital, Sun Yat-Sen University, No. 58 Zhongshan Er Road,
Guangzhou 510080, China
D
3Present address: 250 Massachusetts Avenue, Novartis Institutes for Biomedical Research, Cambridge, o
w
n
MA 02139 lo
a
de
d
*To whom correspondence should be addressed: Ameeta K. Agarwal, National Center for Natural fro
m
Products Research, School of Pharmacy, University, MS 38677. Tel: [phone], Fax: [phone], h
E-mail: [email protected] ttp://w
w
w
Running Title: Novel Antifungal Mechanism of Aporphinoid Alkaloids .jb
c
.o
rg
Key Words: drug action, small molecule, iron-sulfur protein, yeast, infectious disease b/
y
g
u
e
s
t o
ABSTRACT Third, treatment of wild-type yeast cells with n
A
derivEeudp oalpaourrpidhiinneo ida naldk alloiirdiosd tehnaitn ee xhairbei t ppolatennt-t Em9im59ic1k eodr LdMefiTc iegnecnieesra teind cmeliltuolcahro ndderfieaclt s Fteh-aSt pril 6
, 2
inhibitory activity against the opportunistic fungal cluster synthesis including an increase in 01
9
pathogens Candida albicans and Cryptococcus mitochondrial iron levels, a decrease in the
neoformans. However, the molecular mechanism activities of Fe-S cluster enzymes, a decrease in
of this antifungal activity is unknown. In this respiratory function, and an increase in oxidative
study, we show that eupolauridine 9591 (E9591), a stress. Collectively, our results demonstrate that
synthetic analog of eupolauridine, and liriodenine E9591 and LMT perturb mitochondrial Fe-S
methiodide (LMT), a methiodide salt of cluster biosynthesis; thus, these two compounds
liriodenine, mediate their antifungal activities by target a cellular pathway that is distinct from the
disrupting mitochondrial iron-sulfur (Fe-S) cluster pathways commonly targeted by clinically used
synthesis. Several lines of evidence supported this antifungal drugs. Therefore, the identification of
conclusion. First, both E9591 and LMT elicited a this pathway as a target for antifungal compounds
transcriptional response indicative of iron has potential applications in the development of
imbalance, causing the induction of genes that are new antifungal therapies.
required for iron uptake and for the maintenance
of cellular iron homeostasis. Second, a genome- INTRODUCTION
wide fitness profile analysis showed that yeast Eupolauridine and liriodenine are plant
mutants with deletions in iron homeostasis-related secondary metabolites that belong to the
genes were hypersensitive to E9591 and LMT. aporphinoid class of alkaloids which contain a
1

Copyright 2017 by The American Society for Biochemistry and Molecular Biology, Inc.
Novel antifungal mechanism of aporphinoid alkaloids

benzylisoquinoline backbone and are widely compounds will (a) facilitate the pharmacological
distributed in a large number of plant families development of these compounds, (b) potentially
(reviewed in Ref. 1). Eupolauridine has been reveal new target pathways for drug development,
isolated from Annonaceae and Eupomatiaceae and (c) improve our understanding of how these
families, while liriodenine has been isolated from compounds provide chemical defense to the
not only these two families, but also from producing plant species.
Magnoliaceae, Menispermaceae, Monimiaceae, In the present study, using the yeast
and Ranunculaceae families (reviewed in Ref. 2). Saccharomyces cerevisiae as a model organism,
As with many plant secondary metabolites, both we have conducted genomic, genetic, and
eupolauridine and liriodenine possess a diversity biochemical studies to gain insight into the
of pharmacological properties. We have mechanism of action of two compounds that are
previously isolated eupolauridine from the West derivatives of eupolauridine and liriodenine.
African tree Cleistopholis patens and shown that it Eupolauridine 9591 (E9591)4 is a
exhibits antifungal activity against several human benzylnapthyridinium analog, and liriodenine
fungal pathogens (3-5). Pan et al. have isolated methiodide (LMT) is a methiodide salt of
eupolauridine from the Madagascan plant liriodenine. Both compounds exhibit stronger
Ambavia gerrardii and demonstrated its antifungal activities compared to the respective
antiproliferative activity against human ovarian parent compounds. In this work, we show that
D
and lung cancer cell lines (6). Liriodenine has both E9591 and LMT elicit a transcriptional ow
n
been isolated by our group from C. patens, response indicative of iron depletion, and both lo
a
d
Liriodendron tulipifera, and Guatteria multivenia, compounds show increased activity against ed
and we have shown that it possesses antibacterial mutants lacking genes involved in iron uptake and fro
m
and antifungal activities (5,7-10). It has also been mitochondrial iron-sulfur (Fe-S) cluster synthesis. h
ttp
isolated by several other groups from a variety of Elemental profile analysis revealed that, in ://w
plant species, and has been shown to exhibit contrast to iron chelators which decrease cellular w
w
antibacterial, antifungal, antiviral, antitumor, and iron levels, E9591 and LMT caused a significant .jb
c
antiarrhythmic activities (reviewed in Ref. 11). increase in intracellular iron levels. Because an .o
rg
The mechanism by which eupolauridine and upregulation of the iron regulon accompanied by b/
y
liriodenine mediate their biological activities has an increase in intracellular iron levels is a key g
u
e
not been studied in great detail. Previous studies feature of yeast cells with deficiencies in st o
on eupolauridine have indicated that it targets mitochondrial Fe-S cluster biogenesis, we further n A
DbyN Aw htoicpho isito minehriabsiet sI I,t hbius t ethnez ypmreec ishea sm encoht anbiesemn epxapthlowraeyd. Wif e Ec9o5n9f1ir maendd thLaMt TE 95d9i1sr uapntedd LMthiTs pril 6, 2
0
established (4). Inhibition of DNA topoisomerase produced several cellular effects that were 1
9
II has also been reported for liriodenine by some consistent with the phenotypes of mitochondrial
groups; but at least one group has reported that Fe-S cluster synthesis deletion mutants.
liriodenine does not cause DNA damage in yeast Collectively, our results indicate that E9591 and
cells (reviewed in Ref. 1). Additional mechanisms LMT disrupt mitochondrial Fe-S cluster
proposed for the antiproliferative effects of biosynthesis, a pathway not known to be targeted
liriodenine include cell cycle arrest, increased by current antifungal drugs.
production of the tumor suppressor p53, and
induction of nitric oxide synthase expression RESULTS
(reviewed in Ref. 11). Thus, the primary Transcriptional Responses to E9591 and LMT
mechanisms of action of eupolauridine and are Indicative of Iron Depletion – In broth
liriodenine remain largely unknown. It is microdilution assays, E9591 and LMT exhibited
interesting to note that both of these compounds improved antifungal activities compared to the
may be derived from the same precursor respective parent compounds. In a previous report,
metabolite, liriodendronine, and thus, it is possible the minimum inhibitory concentration (MIC) of
that they target the same biological pathways eupolauridine against the fungal pathogens
(reviewed in Ref. 12). A detailed characterization Candida albicans and Cryptococcus neoformans
of the mechanism of action of this class of was reported to be 61.2 M and 244.8 M
2
Novel antifungal mechanism of aporphinoid alkaloids

respectively (4). In comparison, the MIC of E9591 S1 and Fig. 1A, cluster 1). Because these genes are
against these two pathogens was found to be 2.6 also known to be induced in the presence of iron
M and 5.3 M respectively (Table 1). This result chelators (19,20), we also generated
is consistent with a previous report in which transcriptional profiles for the iron chelating
E9591 exhibited a 32-fold improvement in activity compounds 1,10-phenanthroline (PHEN), and
against these two pathogens compared to the 2,2’-bipyridyl (BIPR). The gene expression
parent compound (13). Similarly, LMT showed profiles of all four compounds were highly similar.
improved activity against C. albicans and C. All four compounds commonly upregulated 43
neoformans in comparison to the parent genes, the majority of which are involved in
compound. In previous reports, the MIC of maintaining cellular iron homeostasis, further
liriodenine against these two pathogens was found confirming that E9591 and LMT cause an iron
to be 22.5 M and 45.4 M respectively (8,10); in depletion response (Fig. 1A, see cluster 1).
contrast, LMT exhibited an MIC of 0.9 M and Interestingly, there were also several notable
dissimilarities between the four profiles. For
7.5 M respectively (Table 1). This result is in
example, 61 genes commonly downregulated by
agreement with a previous report in which LMT
the iron chelators PHEN and BIPR were not
demonstrated an 8-fold improvement in activity
affected by E9591 and LMT (cluster 4 in Fig. 1A).
against C. albicans compared to the parent
Many of these genes are involved in cellular
compound (8). The activities of E9591 and LMT D
against C. albicans are comparable to the respiration and are known to be downregulated ow
n
clinically used antifungal drug amphotericin B under conditions of iron deficiency (reviewed in loa
d
(AMB). Ref. 14,15). Similarly, another set of 61 genes ed
In order to understand the mechanism behind commonly upregulated by the iron chelators from
the antifungal effects of E9591 and LMT, we PHEN and BIPR were not induced by E9591 and http
conducted transcriptional profiling studies in the LMT (cluster 3 in Fig. 1A). Many of these genes ://w
model yeast S. cerevisiae. Yeast cells were are known to be induced under low oxygen w
w
exposed to E9591 and LMT at their respective conditions and are also affected by cellular heme .jb
c
IC50 concentrations for a period of one doubling levels (reviewed in Ref. 21). It is possible that a .org
time (~ 4 h). Genes exhibiting significant longer exposure to E9591 and LMT may be b/
y
differential expression between compound-treated required in order to cause an upregulation of gu
e
and solvent-treated cells (p value of  0.001, fold anaerobic genes and a downregulation of st o
change of  2) were identified. E9591 treatment respiration genes. Finally, a set of 93 genes that n A
rLeMsuTlt etdre aintm tehnet ruepsruelgteudla itnio tnh eo uf p1r6eg8u glaetnioens, owf 3h6il2e wweerree ncootm imndouncleyd u bpyre tghuel airteodn bchye Ela9to5r9s1 P aHnEd NL ManTd pril 6, 2
0
BIPR (see cluster 2 in Fig. 1A). Many of these 1
genes and the downregulation of 86 genes (see 9
genes are involved in mitochondrial functions
Supplemental Table S1).
including mitochondrial Fe-S cluster synthesis
The major transcriptional response to E9591
(ISU1), monocarboxylic acid metabolism (CAT2,
and LMT consisted of the upregulation of genes
CIT3, CRC1, PDH1, YAT1), and amino acid
that are known to be induced when yeast cells are
metabolism (AGX1). Collectively, these results
grown in iron-limiting conditions (Fig. 1A, see
indicate that E9591 and LMT cause additional
cluster 1). These genes belong to the iron regulon
metabolic changes in yeast cells distinct from
and are regulated by the transcription factor Aft1
those caused by iron chelators. In addition, the
(reviewed in Ref. 14,15). Among the Aft1-
chemical structures of both compounds indicate
dependent genes known to be induced under iron
that they lack the functional group(s) required for
deficiency (15-18), E9591 and LMT commonly
chelation with iron (see Table 1).
induced 11 genes. These included genes involved
To further validate the transcriptional profiling
in iron uptake (ARN1, CCC2, FIT1, FRE2, FRE3,
data, the expression of 7 genes commonly induced
FRE5), iron mobilization from the vacuole (FET5
by E9591 and LMT representing iron homeostasis-
and FTH1), mitochondrial iron transport (MRS4),
related functions was analyzed by quantitative
and genes involved in metabolic adaptation to low
real-time RT-PCR (Fig. 1B). For all seven genes,
iron (HMX1 and VHT1) (See Supplemental Table
there was consistent correlation between
3
Novel antifungal mechanism of aporphinoid alkaloids

transcriptional profiling and real-time RT-PCR screen performed with individual deletion mutants
data, with similar fold change values observed for (24). These results further suggest that E9591 and
the two assays (compare values in Fig. 1B with LMT disrupt iron homeostasis. However, as was
those in Supplemental Table S1). observed in the transcriptional profiles, the fitness
profiles of E9591 and LMT were not identical to
Genome-Wide Fitness Profiles of E9591 and that of BPS. Several additional mutants that were
LMT Indicate Iron Homeostasis Disruption – To found to be hypersensitive to BPS in the previous
further delineate the molecular pathways targeted studies were not observed in our study. Thus, it is
by E9591 and LMT, we conducted a genome-wide possible that iron depletion may not be the primary
fitness profile analysis (22,23) in which we mechanism of action of E9591 and LMT.
screened a whole-genome collection of pooled
haploid yeast deletion mutants against the two E9591 and LMT Treatment Increases
compounds. Upon individual validation, we Intracellular Iron Levels – Because the transcript
identified 42 haploid deletion mutants that were and fitness profiles of E9591 and LMT were
significantly more sensitive to either E9591 or indicative of a disruption in iron homeostasis, we
LMT in comparison with the wild-type strain (see investigated whether exposure to the two
Supplemental Table S3). Highly enriched among compounds leads to an alteration in intracellular
these were mutants harboring mutations affecting iron levels in yeast cells. Using the same
D
iron ion homeostasis, Fe-S cluster assembly, experimental conditions as employed for transcript ow
n
vesicle-mediated transport, and oxidative stress profiling, yeast cells were exposed to E9591 and lo
a
d
(sFenigs.i tiv2itAy )t.o EM95u9ta1n atsn d LexMhiTb iitnincglu detdh et hosheig wheitsht LsuMbjTec taetd t htoei re lreemspeencttailv ea nIaCly50s isc obnyc eInCtrPa-tiMonSs. aFnodr ed fro
m
mutations affecting high affinity iron uptake comparison, cells were also exposed to the iron h
ttp
(FET3 and FTR1), mitochondrial Fe-S cluster chelator PHEN. In addition to iron, 7 additional ://w
synthesis (ISU1 and ISA2), oxidative stress elements (Cu, Mn, Na, K, Mg, Ca, and Zn) were w
w
(SOD2), tricarboxylic acid cycle (ACO1), and quantifiable under these conditions. .jb
c
copper uptake (CTR1) (see Fig. 2B). Four As expected, cellular iron levels were .o
rg
additional mutants with mutations in decreased (by approximately 51%) in PHEN- b/
y
mitochondrial Fe-S cluster synthesis (ISA1, GRX5, treated cells compared to DMSO-treated cells g
u
e
SSQ1, and CAF17) were hypersensitive to E9591 (Fig. 3C). In contrast, an increase in intracellular st o
or LMT (Fig. 2A and Supplemental Table S3). iron content was observed for E9591-treated n A
geneIst iinsv owlvoerdth i nn hoitginhg aftfhianti tyw ihriolne umptuatkaeti o(nFsE Tin3 ((aapppprrooxxiimmaatteellyy 1.6-2f-oflodl)d )c ellsa nredl ativeL MtoT D-tMreaStOed- pril 6, 2
0
and FTR1) resulted in increased sensitivity to treated cells (Fig. 3A and 3B). For Cu, Mn, K, Mg, 1
9
E9591 and LMT, these two genes did not appear and Ca, no significant changes occurred
to be induced in the 4 h time point transcript consistently in response to E9591 and LMT. We
profiling results (Fig. 1A). One possible observed an increase in Cu levels in PHEN-treated
explanation for these observations could be the cells (Fig. 3C), which is consistent with the
differences in the drug exposure times used in the upregulation of copper transporters under iron-
two experiments (cells used in the fitness profiling limiting conditions (reviewed in Ref. 14,15). An
analyses were continuously exposed to E9591 or increase in Zn levels and a decrease in Na levels
LMT for 2 days; discussed further below). were observed in cells treated with E9591, LMT,
Of the 42 mutants identified in our study, 12 and PHEN, suggesting that these changes were not
mutants with mutations in FET3, FTR1, CTR1, specific to E9591 and LMT. These results further
ERG4, LEM3, RIM8, RIM9, SNF7, UBP3, VPS20, support the hypothesis that exposure to E9591 and
WHI2 and YPK1 were found to be hypersensitive LMT is not associated with a direct depletion of
to the iron chelator bathophenanthrolinebisulfonic iron from yeast cells and that these two
acid (BPS) in a genome-wide fitness profiling compounds do not function as iron chelators.
study (19). Mutants with mutations in FET3, The increase in intracellular iron along with an
FTR1, and CTR1 were also found to be increase in the expression of iron regulon genes is
hypersensitive to BPS in another genome-wide highly reminiscent of the effects observed when
4
Novel antifungal mechanism of aporphinoid alkaloids

yeast cells experience a deficiency in the synthesis E9591 and LMT Treatment Mimics Deficiency
of Fe-S clusters (reviewed in Ref. 25,26-28). Fe-S in Mitochondrial Fe-S cluster Synthesis – To
clusters are cofactors found in proteins that play further confirm that E9591 and LMT have an
important functions in respiration, amino acid inhibitory effect on the mitochondrial Fe-S cluster
biosynthesis, DNA repair and protein translation synthesis pathway, we investigated whether
(reviewed in Ref. 25,26-28). The assembly of Fe-S treatment with these compounds would mimic
clusters requires a complex pathway that first additional deficiencies in this pathway. We
generates Fe-S clusters on scaffold proteins, and focused on three phenotypes of mutants in this
then transfers the clusters to acceptor sites within pathway – a decrease in the activities of the Fe-S
recipient apoproteins (reviewed in 25,26-28) (see cluster-containing enzymes, a deficiency in
Supplemental Fig. S1 for an overview). Both respiration, and an increase in oxidative stress
mitochondrial and cytosolic assembly systems (reviewed in Ref. 25,26-28). To maintain
exist in eukaryotic organisms, and the former consistency with the elemental profile analysis,
critically influences iron regulation in yeast these experiments were conducted with yeast cells
(reviewed in 25,26-28). Defects in the that were exposed to E9591 and LMT for ~15
mitochondrial Fe-S cluster synthesis pathway hours (4.5 doublings).
results in the disruption of iron homeostasis in First, we investigated whether treatment with
yeast cells and an upregulation of the iron regulon E9591 and LMT affected the enzyme activities of
D
(reviewed in Ref. 14,26). This induction results in mitochondrial and cytosolic Fe-S proteins. The ow
n
an increase in iron uptake ultimately leading to an mitochondrial Fe-S enzymes aconitase and lo
a
d
increase in intracellular iron levels (reviewed in succinate dehydrogenase, and the cytosolic Fe-S ed
Ref. 26,28). The excess iron gets distributed to the enzyme isopropylmalate isomerase (Leu1) have fro
m
mitochondria so that the iron demands of this been extensively studied to monitor the effects of h
ttp
pathway can be met (reviewed in Ref. 25,26-28). deficiencies in the Fe-S cluster synthesis pathway. ://w
Mitochondrial iron overload has been reported for Decreases in the activities of all three enzymes w
w
several mutants in this pathway including those have been reported in yeast mutants with defects .jb
c
with defects in Isa1, Isa2, Isu1, Isu2, Grx5, Nfs1, in several Fe-S cluster synthesis proteins including .o
rg
Ssq1, Yah1, and Yfh1 (29-37). Isd11, Nfs1, Yah1, and Grx5 (32,33,36,39,40). We b/
y
To determine whether treatment with E9591 observed that when yeast cells were treated with g
u
e
and LMT also caused an over-accumulation of E9591 and LMT, aconitase activity was st o
iron in the mitochondria, we monitored diminished by 35% and 64% respectively (Fig. n A
meaictohc choomndproiauln de.l eWmeen otabls elrevveedls aanf taepr peroxxpiomsuartee 7to- 5anAd) . SLimMiTla rlyc,a utrseeadtm ean t orefd yuecatsiot nc ellisn wisthu cEc9in5a9t1e pril 6, 2
0
fold increase and an approximate 13-fold increase dehydrogenase activity by 70% and 40% 1
9
in mitochondrial iron levels in E9591-treated and respectively (Fig. 5A). In contrast, the activity of
LMT-treated cells respectively compared to mitochondrial malate dehydrogenase as a non-Fe-
DMSO-treated cells (Fig. 4). For the other 7 S control enzyme was not affected (Fig. 5A). We
elements, no dramatic changes were observed to also observed that the activity of the cytosolic Fe-
occur commonly in response to E9591 and LMT. S protein Leu1 was decreased by 65% and 56%
It is worth noting that the increase in total and respectively in cells treated with E9591 and LMT,
mitochondrial iron levels was observed when yeast whereas the activity of the control enzyme alcohol
cells were exposed to E9591 and LMT for a period dehydrogenase was not strongly influenced by the
of ~15 hours (4.5 doublings). This result is two compounds (Fig. 5B).
consistent with the observed delay in cellular iron Second, we determined if cells treated with
accumulation that had been previously noted in E9591 and LMT showed a deficiency in
Yfh1-depleted cells (38). Taken together, the respiration. Respiratory failure is a common trait
results shown in Figs. 3 and 4 provide further of mutants deficient in the mitochondrial Fe-S
evidence that the inhibitory effects of E9591 and cluster synthesis pathway due to the fact that
LMT are likely associated with defects in the several Fe-S proteins participate in the respiratory
mitochondrial Fe-S cluster synthesis pathway. chain. A decline in respiratory function has been
observed in mutants with defects in Yfh1, Isu1,
5
Novel antifungal mechanism of aporphinoid alkaloids

Isa1, and Isa2 (29,31,37,38). To determine oxidative stress. Mutants with deletions in YAP1
whether E9591 and LMT treatment causes (encodes a transcription factor required for
respiratory deficiency in yeast cells, we monitored oxidative stress tolerance), SOD2 (encodes
the formation of petite (respiration-deficient) superoxide dismutase), and GSH1 (encodes the
mutants upon exposure to these two compounds. first enzyme in glutathione biosynthesis) showed
This assay is based on the principle that when increased sensitivity to E9591 and LMT (Fig. 7B).
yeast cells are grown on medium containing It is worth noting that the sod2 mutant was also
glycerol (a non-fermentable carbon source), they identified to be hypersensitive to both compounds
are dependent upon mitochondrial respiration, and in our genome-wide fitness profile analysis (see
when grown on medium containing glucose Fig. 2B). We also observed that the iron chelator
(fermentable carbon source), they can survive PHEN did not cause an increase in the levels of
without it (reviewed in Ref. 41). We exposed yeast carbonylated proteins and the three mutants tested
cells to E9591 and LMT, and grew them on media above were not strongly hypersensitive to PHEN
containing glycerol vs. glucose to determine the (see Supplemental Fig. S3). This further suggests
number of respiration competent cells. We that the cellular effects of E9591 and LMT are not
observed an approximate 75% reduction in the due to a general depletion of cellular iron, but
percentage of respiration competent cells in rather due to redistribution of iron to the
E9591-treated and LMT-treated cells compared to mitochondria. Taken together, the results shown in
D
DMSO-treated cells (Fig. 6A). Petite (small-size) Figures 5, 6, and 7 show that exposure to E9591 ow
n
colonies were also clearly visible on the glucose- and LMT produces effects that are similar to those lo
a
d
containing plates for cells treated with E9591 and observed in yeast mutants with defects in the ed
LMT (Fig. 6B). In contrast, when yeast cells were mitochondrial Fe-S cluster synthesis pathway. fro
m
exposed to the iron chelator PHEN under similar h
ttp
experimental conditions, there was no reduction Haploinsufficient Mutants of Mitochondrial ://w
observed in the number of respiration competent Fe-S Cluster Synthesis Genes are Hypersensitive w
w
cells (see Supplemental Fig. S2). Thus, even to E9591 and LMT – Our genome-wide fitness .jb
c
though the transcriptional responses to E9591 and profile analysis revealed 6 haploid mutants with .o
rg
LMT are indicative of iron depletion, the cellular deletions in the mitochondrial Fe-S cluster b/
y
effects of E9591 and LMT are not reflective of a synthesis pathway that showed increased g
u
e
general impairment in iron metabolism. sensitivity to E9591 and LMT (see Supplemental st o
Finally, we determined whether E9591 and Table S3). To explore additional mutants in this n A
LOMxidTa tiivned usctreeds s oixs idkantoivwen tsotr ebses inind ucyeeda sitn cyeelalsst. pwahtihcwh ayc, awrriee da nahleytzeerdo zyhgaopulosi nsdueflfeitciioennst minu tatnetns pril 6, 2
0
mutants with defects in the mitochondrial Fe-S different genes required for the biosynthesis and 1
9
cluster synthesis proteins Yfh1 and Grx5 (36,38). maturation of Fe-S cluster proteins. Because many
This induction is most likely due to the over- of the components of the mitochondrial Fe-S
accumulation of iron in the mitochondria, which cluster synthesis pathway are essential for the
can promote Fenton-mediated formation of viability of yeast cells, mutants containing haploid
reactive oxygen species. To determine if E9591 deletions in them are inviable. Thus, heterozygous
and LMT induced oxidative stress in yeast cells, deletion mutants allowed us to investigate multiple
we monitored protein carbonylation levels after components of the pathway. As can be seen in Fig.
exposure to the two compounds. Oxidative stress 8, all of the mutants tested showed a strong
results in the formation of carbonyl groups on increase in sensitivity to E9591 compared to the
proteins which can be immunodetected by wild-type strain. A clearly discernable increase in
Western analysis. As can be seen in Fig. 7A, sensitivity to LMT was also observed in all
E9591-treated and LMT-treated cells contained mutants, except for the mutant with a
increased levels of carbonylated proteins heterozygous deletion in ISA2, which appeared to
compared to DMSO-treated cells. We also made show reduced sensitivity to LMT. The reason for
use of yeast mutants with deletions in genes this mutant’s lack of sensitivity to LMT is unclear
required for oxidative stress responses, which at this time. For comparison, we also analyzed
exhibit hypersensitivity to chemicals promoting these ten mutants for sensitivity to the iron
6
Novel antifungal mechanism of aporphinoid alkaloids

chelator PHEN, and the mutants did not exhibit In agreement with the observed long-term
strong hypersensitivity to PHEN (see effects of E9591 and LMT on Fe-S cluster
Supplemental Fig. S4). These results further enzymes (see Fig. 5), 15 h exposures to E9591 and
indicate that E9591 and LMT likely cause LMT resulted in the induction of iron regulon
perturbations within the mitochondrial Fe-S genes (see Supplemental Table S4). Interestingly,
cluster synthesis pathway. the high affinity iron transporter genes FTR1 and
It is worth noting that although three of the ten FET3, which were not induced at the 4 h time
genes tested above are not essential genes (ISA1, point, were significantly induced (p  0.001), at
ISA2, and NFU1), deletions in them resulted in the 15 h time point. FTR1 was induced 8.5-fold by
increased sensitivity to E9591 and LMT (with the E9591 and 3.9-fold by LMT at the 15 h time point
exception of ISA2 which only affects sensitivity to (see Supplemental Table S4). FET3, although not
E9591). Interestingly, although there is a tendency as strongly affected as FTR1, was induced 1.9-fold
for haploinsufficient genes to be essential, there by E9591 and 1.5-fold by LMT at the 15 h time
are indeed several examples in the literature where point (see Supplemental Table S6 which shows
both essential and non-essential genes exhibit significant iron regulon genes induced by  1.5-
haploinsufficiency (e.g., 42,43,44). It has been fold; p  0.001). Thus, under the experimental
suggested that essential genes may represent direct conditions we have employed, a longer exposure
targets of a drug, whereas non-essential genes may to E9591 and LMT, which would likely result in D
exhibit synthetic interactions with the drug targets the prolonged disruption of iron homeostasis, may ow
n
or the target pathways. In fact, ISA1, ISA2, and be necessary to elicit the induction of the high loa
d
NFU1 show physical and genetic interactions with affinity iron transporter genes FTR1 and FET3. ed
genes involved in the mitochondrial Fe-S cluster While several iron regulon genes were fro
m
synthesis pathway. For example, Isa1 physically induced by E9591 and LMT at the 15 h time point, http
interacts with Nfu1 (45). In addition, ISA2 the major over-represented functional categories ://w
genetically interacts with GRX5 (36), and NFU1 among the upregulated genes included translation, w
w
genetically interacts with ISU1 and SSQ1 (37). ribosome biogenesis, rRNA processing, amino .jb
c
Based on this as well as the results shown in acid metabolism, nucleotide metabolism, and .org
Figures 4, 5, 6, and 7, it is highly likely that the sterol metabolism (Fig. 9A). The upregulation of b/
y
increased sensitivity of these three mutants to genes involved in amino acid and ergosterol gu
e
E9591 and/or LMT is associated with a disruption metabolism has been previously observed in st o
in mitochondrial Fe-S cluster synthesis caused by mutants defective in Yah1 and Atm1 (46), and n A
t he two compounds. tfhoer ir sienvdeuraclt iohne mmea-yc oonctcauirn idnuge eton ztyhme erse quoirr emFee-nSt pril 6, 2
0
Long-Term Exposure to E9591 and LMT cluster-containing enzymes in these two processes 19
Reveals Prolonged Effects of Mitochondrial Fe-S (reviewed in Ref. 15,28). Similarly, an
Cluster Synthesis Disruption – Because the results upregulation in nucleotide metabolism genes has
shown in Figures 3 through 7 were observed after been reported in a mutant defective in Grx5 (47),
yeast cells were exposed to E9591 and LMT for and could be attributed to the requirement for the
~15 h, we were also interested in determining the iron-containing enzyme ribonucleotide reductase
long-term effects of E9591 and LMT. We, in nucleotide biosynthesis (reviewed in Ref. 48).
therefore, conducted a transcript profiling study on In addition, a large number of genes that
yeast cells that were exposed to E9591 and LMT participate in protein translation were induced
for ~15 h grown under the same culture conditions upon long-term exposure to E9591 and LMT (Fig.
as described above. A total of 956 genes were 9A). This induction could occur due to the fact
commonly upregulated and 652 genes were that several proteins required for protein
commonly downregulated by both E9591 and translation contain Fe-S clusters in their structures
LMT (see Supplemental Table S4). The including Rli1, Elp3 and Twy1 (reviewed in Ref.
responding genes were organized into GO-based 28).
functional categories and over-represented GO As expected, “respiration” was the major
categories were identified (see Supplemental functional category for genes downregulated in
Table S5 and Fig. 9). response to long-term exposure to E9591 and
7
Novel antifungal mechanism of aporphinoid alkaloids

LMT (Fig. 9B). A large number of genes encoding iron levels, a decrease in the activities of Fe-S
components of the mitochondrial respiratory chain cluster enzymes, an increase in respiratory
and the TCA cycle are known to be downregulated deficiency, and an increase in oxidative stress, and
in yeast mutants with defects in Yah1, Atm1, and (iv) haploinsufficient mutants with deletions in ten
Grx5 (46,47). Unexpectedly, we also observed that different genes involved in mitochondrial Fe-S
long-term exposure to E9591 and LMT resulted in cluster synthesis showed increased sensitivity to
the downregulation of genes involved in E9591 and LMT. Thus, these two compounds
peroxisome protein import and fatty acid beta- target a cellular pathway that is distinct from the
oxidation, an important peroxisomal process in pathways targeted by current clinically used
yeast cells (Fig. 9B). Downregulation of antifungal drug classes, which are known to work
peroxisomal genes has not been previously by interacting with membrane ergosterol (e.g.,
observed in mutants with defects in the polyenes), disrupting ergosterol biosynthesis (e.g.,
mitochondrial Fe-S cluster synthesis pathway, and azoles), and inhibiting the synthesis of cell wall
could be attributed to differences in culture glucans (e.g., echinocandins) (reviewed in Ref.
conditions between our study and previous studies. 49).
The implications of this downregulation are Because Fe-S proteins play important roles in
discussed below. many different cellular processes including
In summary, the majority of gene expression respiration, TCA cycle, amino acid biosynthesis,
D
changes due to long-term effects of E9591 and DNA synthesis and repair, and protein translation, ow
n
LMT are similar to those observed in mutants a defect in the synthesis of Fe-S proteins has a lo
a
d
defective in Fe-S cluster synthesis. They are also severe impact on diverse cellular functions ed
consistent with the known involvement of Fe-S (reviewed in Ref. 25,26,28). In addition, most of fro
m
cluster proteins in diverse biochemical pathways the components of the mitochondrial Fe-S cluster h
ttp
including heme biosynthesis, amino acid and synthesis pathway are essential for the viability of ://w
nucleotide metabolism, and DNA repair. It is yeast cells (reviewed in Ref. 25,26,28). Thus, this w
w
important to note that the possibility cannot be pathway could serve as a highly effective target .jb
c
discounted at present that in addition to affecting for inhibiting fungal growth. However, due to the .o
rg
the activities of Fe-S enzymes, E9591 and LMT importance of Fe-S clusters in all eukaryotes, a b/
y
may also affect the activities of non-Fe-S enzymes therapeutic drug that targets Fe-S cluster synthesis g
u
e
that use iron as a co-factor including heme- could most likely lack specificity for the fungal st o
containing proteins (e.g., cytochrome complexes) pathogen. Nevertheless, specificity could be n A
arisb onwuecllle otiadse rendounc-thaesme)e. iron proteins (e.g., amcahmiemveadli adnu ep rtoot eidnisff etrheant caerse bceotwmepeonn efnutns goafl athnids pril 6, 2
0
pathway. For example, the transfer of Fe-S 1
9
DISCUSSION clusters from the Isu1 scaffold protein to target
Using a combination of genomic, genetic, and apoproteins requires a dedicated Hsp70 chaperone.
biochemical approaches, we have shown that the Mammalian cells make use of the multifunctional
antifungal activity of the plant-derived Hsp70 of the mitochondrial matrix, while yeast
aporphinoid alkaloids E9591 and LMT is likely cells utilize a specialized Hsp70 chaperone, Ssq1
mediated by a disruption in the mitochondrial Fe-S (reviewed in Ref. 26). Unlike multifunctional
cluster synthesis pathway. This conclusion is Hsp70s that bind to a large variety of hydrophobic
based on several lines of evidence from the present substrates, Ssq1 selectively recognizes and
work: (i) the transcriptome response to E9591 and interacts with a conserved peptide loop in Isu1
LMT showed an induction in iron regulon genes, (50). Also, in yeast cells an additional scaffold
(ii) a genome-wide fitness profile analysis showed protein, Isu2 is used for Fe-S cluster synthesis that
that yeast mutants with deletions in iron is not present in mammalian cells (reviewed in
homeostasis-related genes were hypersensitive to Ref. 26,28). Yeast cells with a deletion in either
E9591 and LMT, (iii) treatment with E9591 and Ssq1 or Isu2 exhibit respiratory deficiency, an
LMT caused cellular defects that mimicked increase in mitochondrial iron, and a loss of
deficiencies in mitochondrial Fe-S cluster mitochondrial aconitase activity (30,37). Thus,
synthesis including an increase in mitochondrial Ssq1 and Isu2 have the potential to serve as
8
Novel antifungal mechanism of aporphinoid alkaloids

fungal-specific targets for new antifungal toxicity in a Yfh1-deficient mutant is attenuated
therapies. by a deletion in the CIT2 gene which encodes a
By demonstrating that E9591 and LMT peroxisomal citrate synthase (53). It is, therefore,
function as disruptors of the mitochondrial Fe-S possible that a shut-down in peroxisomal functions
cluster synthesis pathway, this work has identified could cause a reduction in intracellular citrate
a promising new pharmacological tool for the levels, allowing the cells to cope with the excess
further characterization of this pathway in iron that accumulates in Fe-S cluster synthesis
eukaryotic organisms. Given that many of the mutants. Thus, it is conceivable that under
genes encoding components of this pathway are conditions of Fe-S cluster synthesis deficiency,
essential genes, previous studies have made use of coping with iron toxicity takes precedence over
conditional mutants that require the mutant cells to providing metabolites to the TCA cycle. Further
be grown under restrictive conditions that impose studies will be required to understand the
additional stress on the cells. A chemical inhibitor molecular mechanisms potentially involved in the
of this pathway will facilitate these studies and cross-talk between mitochondrial Fe-S cluster
will prevent the occurrence of secondary synthesis and peroxisomal function.
metabolic effects exerted by restrictive growth It is worth noting that, although increased
conditions. In addition, chemical inhibitors will be intracellular zinc levels were observed following a
of great value in improving our understanding of 15 h exposure to E9591 and LMT (see Fig. 3),
D
why defects in mitochondrial Fe-S cluster increased expression of the zinc transporter genes ow
n
synthesis result in various human diseases. Human ZRT1 and ZRT2 was not observed. Instead, we lo
a
d
diseases such as Friedreich’s ataxia (Yfh1 observed a downregulation of ZRT2 and also a ed
deficiency), microcytic anaemia (Grx5 downregulation of ZAP1, ZRG7, ZRG8, and fro
m
deficiency), cerebellar ataxia (Atm1), and others ZRG17 (see Supplemental Table S4), all of which h
ttp
are associated with defects in this pathway, and are required for maintaining zinc homeostasis in ://w
the direct functional connection to this pathway yeast cells (e.g., 54,55). The transcription factor w
w
remains to be determined for some of these Zap1 and its target genes are activated under zinc .jb
c
diseases (reviewed in Ref. 25,26,51). Elucidating deficient conditions in yeast cells, and inactivated .o
rg
the physiological consequences of Fe-S cluster under zinc excess conditions (reviewed in 14,56). b/
y
synthesis defects will provide a better Thus, it is possible that the excess zinc g
u
e
understanding of how these diseases occur. accumulating during the 15 h exposure to E9591 st o
Of further interest, the present work also and LMT led to the downregulation of zinc n A
icnludsicteart essy nththate sdiiss rpuapthtiwona yo fc athuese ms iat odcohwonnrdergiaull aFteio-Sn hzionmc etorasntasspios rgteernse asr. eA odndliyt itornaanlsliye,n ittl yis i npdoussciebdl,e tthheant pril 6, 2
0
of peroxisomal genes. Previous studies have subsequently downregulated once intracellular 1
9
indicated that perturbations in mitochondrial zinc levels increase. A more detailed time course
functions such as those caused by petite mutations study will be required to examine the possible
cause an upregulation of peroxisomal genes (52). temporal effects of E9591 and LMT on zinc
It was suggested that to compensate for the transporter expression.
absence of a complete TCA cycle, respiratory- While yeast cells appear to respond to the
deficient cells reconfigure their metabolism by over-accumulation of zinc, it is not likely that the
activating peroxisomal activities. This ensures that excess zinc contributes to the phenotypic effects
metabolites such as acetyl-CoA that are generated generated by E9591 and LMT. First, long-term
by the beta-oxidation of fatty acids in the exposure of yeast cells to excess zinc has been
peroxisomes are made available to the TCA cycle. shown to cause a decrease in intracellular iron
However, an interruption in mitochondrial Fe-S content (57). In contrast, E9591 and LMT
cluster synthesis causes not only respiratory increased intracellular iron levels. Second, a
deficiency but also an over-accumulation of genome-wide mutant analysis in the presence of
intracellular iron. Given that citrate is known to excess zinc primarily identified genes involved in
enhance iron toxicity, one way to control this vacuolar function (57), and did not identify any
toxicity would be to reduce intracellular citrate iron uptake genes or Fe-S cluster synthesis genes,
levels. This is supported by the fact that the iron many of which were identified in the fitness
9
Novel antifungal mechanism of aporphinoid alkaloids

profile of E9591 and LMT. Interestingly, zinc culture of S. cerevisiae S288C was used to
suppresses the phenotypes of a yeast mutant inoculate 50 ml of SD medium to an OD of 0.1.
600
deficient in the Fe-S cluster synthesis protein Yfh1 Three replicate cultures were started for each
(58). Excess zinc in the medium prevented treatment. After one doubling, each culture was
mitochondrial iron accumulation in the mutant, treated with E9591, LMT, PHEN, or BIPR at a
and also increased its growth rate as well as concentration equivalent to the IC value (0.053
50
resistance to oxidative stress (58). Thus, it is M, 2.6 M, 6.2 M, and 65.9 M respectively).
possible that zinc over-accumulation in response Control cultures were simultaneously treated with
to E9591 and LMT may be associated with a 0.25% (v/v) DMSO. The cultures were allowed to
mechanism that allows the cells to cope with grow until an OD of 0.5 was reached (~4 h). For
600
disruptions in the Fe-S cluster synthesis pathway. long-term exposure experiments, all experimental
This study lays the groundwork for future conditions were the same except cells were
studies to determine the precise mechanism of exposed to drug treatments for 4.5 doublings (~15
action of E9591 and LMT. While our data strongly h). Cells were harvested by centrifugation, flash
suggest that these compounds disrupt the frozen in liquid nitrogen and stored at -80°C.
mitochondrial Fe-S cluster synthesis pathway, RNA isolation, target preparation, and
further analysis will be required to determine the hybridizations were performed as described
precise step(s) in the pathway they inhibit and the previously (60). The Affymetrix GeneChip Yeast
D
mechanism or mechanisms involved in this Genome 2.0 array was used in all experiments. ow
n
inhibition. Image analysis, scaling and probe-set-level data lo
a
d
analysis was performed using the Affymetrix ed
EXPERIMENTAL PROCEDURES GeneChip Operating Software. Differentially fro
m
Yeast Strains, Media, and Chemicals - S. expressed genes were identified using BRB Array h
ttp
cerevisiae strain S288C was used for all Tools software (61), and genes with p  0.001 ://w
experiments. Haploinsufficient mutants lacking were considered to be significant. Gene w
w
Fe-S cluster synthesis genes were obtained from annotations were obtained from Saccharomyces .jb
c
Open Biosystems (Huntsville, AL). Synthetic Genome Database (SGD). Hierarchical cluster .o
rg
dextrose (SD) medium consisted of 2% (w/v) analysis was performed with Gene Cluster 3.0 b/
y
dextrose and 0.7% (w/v) yeast nitrogen base (62), and the data were visualized with Java Tree gu
e
without amino acids. The medium was buffered View (63). The BinGO plugin in Cytoscape st o
with 0.2 M MOPS and the pH was adjusted to 7.0. software was used for Gene Ontology (GO) n A
Y2%PD ( wm/evd)i upme pctoonnsei,s t2ed% o f( w1%/v )( wd/evx)t ryoesaes.t eLxatcrtaactte, a0n.0a5ly) siws, eraen di deonvteirf-ieredp r(e6s4en).t edT hGe Ot ratnesrcmrsip t(iopn a<l pril 6, 2
0
medium for mitochondria isolation was prepared profiling data described in this article are 1
9
as described by Amutha et al. (59). For the accessible through accession no. GSE101749 at
genome-wide fitness profiling test, the haploid NCBI’s Gene Expression Omnibus (GEO;
selection synthetic medium was prepared as http://www.ncbi.nlm.nih.gov/geo/).
described previously (22). Dimethyl sulfoxide
(DMSO), 1,10-phenanthroline (PHEN), and 2,2’- Quantitative Real-Time RT-PCR - To confirm
bipyridyl (BIPR) were obtained from Sigma- the transcriptional profiling results, quantitative
Aldrich (St. Louis, MO). LMT and E9591 were real-time RT-PCR was performed using the same
synthesized as described previously (8,13). The RNA preparations that were used in the
purity of the two compounds was greater than 95% transcriptional profiling experiments. DNAse-
based on TLC and NMR analysis. treatment of RNA samples, design of gene-
specific primers, and quantitative real-time PCR
Transcriptional Profiles of E9591, LMT, reactions were performed as described previously
PHEN, and BIPR - S. cerevisiae strain S288C was (60). The primer sequences for each gene selected
used in the transcriptional profiling experiments, for the assays are listed in Supplemental Table S7.
and all procedures including IC determinations Data were normalized to an internal control (18S
50
were performed as previously described (60). For rRNA) and the C method was used to obtain
T
the transcriptional profiling study, an overnight the relative expression level for each gene.
10