OUP user menu

Genome-wide investigation of cellular targets and mode of action of the antifungal bacterial metabolite 2,4-diacetylphloroglucinol in Saccharomyces cerevisiae

Danielle M. Troppens, Ruslan I. Dmitriev, Dmitri B. Papkovsky, Fergal O'Gara, John P. Morrissey
DOI: http://dx.doi.org/10.1111/1567-1364.12037 322-334 First published online: 1 May 2013


Saccharomyces cerevisiae is a proven model to investigate the effects of small molecules and drugs on fungal and eukaryotic cells. In this study, the mode of action of an antifungal metabolite, 2,4-diacetylphloroglucinol (DAPG), was determined. Applying a combination of genetic and physiological approaches, it was established that this bacterial metabolite acts as a proton ionophore and dissipates the proton gradient across the mitochondrial membrane. The uncoupling of respiration and ATP synthesis ultimately leads to growth inhibition and is the primary toxic effect of DAPG. A genome-wide screen identified 154 DAPG-tolerant mutants and showed that there are many alterations in cellular metabolism that can confer at least some degree of tolerance to this uncoupler. One mutant, ydc1, was studied in some more detail as it displayed increased tolerance to both DAPG and the uncoupler carbonylcyanide m-chlorophenylhydrazone (CCCP) and appears to be unconnected to other tolerant mutant strains. Deleting YDC1 alters sphingolipid homoeostasis in the cell, and we suggest here that this may be linked to reduced drug sensitivity. Sphingolipids and their derivatives are important eukaryotic signal molecules, and the observation that altering homoeostasis may affect yeast response to metabolic uncoupling agents raises some intriguing questions for future studies.

  • respiration
  • uncoupler
  • genetic screen
  • CCCP
  • proton ionophore


The global microbiota synthesize a vast array of secondary metabolites, and although many of these metabolites remain of unknown function, others can be classified as antibiotics, toxins or signals. A large number of secondary metabolites with antimicrobial activity have been described, and such molecules are generally considered to be antibiotics produced to enhance survival prospects in competitive niches. That concept has been challenged by some workers who suggest that at physiologically relevant concentrations, some so-called antibiotics are actually signal molecules (Davies, 2006, 2009; Yim et al., 2007). Soil microorganisms, for example bacteria in the genera Streptomyces, Bacillus and Pseudomonas, are particularly prolific producers of secondary metabolites. In some cases, these molecules are clinically exploited as antibiotics, whereas in others there is interest in using the secondary metabolite-producing microorganism as a biological control agent to control plant pathogens (Morrissey et al., 2002; Couillerot et al., 2009; Babalola, 2010). For this latter application, it is important to have detailed knowledge about the metabolite, its regulation and synthesis and its effects on target and nontarget organisms.

The polyketide 2,4-diacetylphloroglucinol (DAPG) is an example of a secondary metabolite that has attracted much attention for its biocontrol potential. This metabolite is produced by certain strains of Pseudomonas fluorescens and has broad-spectrum growth inhibitory activity against bacteria, fungi, oomycetes and nematodes (Vincent et al., 1991; Fenton et al., 1992; Laville et al., 1992; Mazzola et al., 1995; Keel et al., 1996; Cronin et al., 1997a b; de Souza et al., 2003; Kwak et al., 2009). Furthermore, strains of P. fluorescens that synthesize DAPG have demonstrable biocontrol activity against phytopathogenic fungi such as Gaeumannomyces graminis, Thielaviopsis basicola and Verticillium dahliae (Laville et al., 1992; Raaijmakers et al., 1997, 1999; Berg et al., 2006; Sanguin et al., 2009). The biosynthetic and regulatory pathways that control the synthesis of DAPG in P. fluorescens are well studied (Haas & Defago, 2005; Bottiglieri & Keel, 2006; Moynihan et al., 2009; Wu et al., 2010), but less is known about how it affects other organisms. At high concentrations, DAPG certainly has antimicrobial action, but there are also reports of signalling-type effects on bacteria (Combes-Meynet et al., 2011) or plants (Weller et al., 2012) at lower concentrations.

Assessing the potential of DAPG-producing P. fluorescens strains as biocontrol agents requires knowledge of how DAPG affects eukaryotes, especially fungi, and how these fungi may respond to DAPG to gain tolerance. Several studies have addressed these questions using different oomycete, fungal and yeast species. DAPG treatment of the oomycete Pythium ultimum inhibited zoospore swimming and hyphal growth and led to plasma membrane alterations, vacuolization and cell content disintegration (de Souza et al., 2003). This is suggestive of targeting of an important metabolic process. More recent studies in the yeast Saccharomyces cerevisiae supported this conclusion and identified the mitochondrion as a target of DAPG, although the mechanism underpinning this was not established (Gleeson et al., 2010). A separate yeast study assessed the sensitivity of a library of mutants to DAPG and found that genetic lesions in a plethora of processes, including membrane function, reactive oxygen regulation and cell homoeostasis, could increase DAPG susceptibility (Kwak et al., 2011). These data can be partly explained by the increase in reactive oxygen species (ROS) that occurs following prolonged DAPG treatment and again point to disruption of core cellular functions. An alternative approach is to assess how tolerant organisms deal with DAPG, and studies of the fungi Botrytis cinerea and Fusarium oxysporum found that tolerant strains coped with DAPG either through efflux using an ABC transporter (B. cinerea) or via enzymatic degradation (F. oxysporum; Schouten et al., 2004, 2008). Importantly, however, it was also clear that some tolerant strains had other undescribed mechanisms that reduced their sensitivity to DAPG.

The strategy of using S. cerevisiae to study the effects and mode of action of small molecules in eukaryotes has been very successful in the past, for example in determining the mode of action of the immunosuppressive drugs rapamycin and FK506 (Heitman et al., 1991, 1993; Lorenz & Heitman, 1995). Saccharomyces cerevisiae has also proven useful as a model to investigate the antifungal activities of plant defensins such as DmAMP1 (Thevissen et al., 2003, 2005; Aerts et al., 2006) and plant saponins such as α-tomatine (Simons et al., 2006). In addition, the S. cerevisiae deletion libraries are useful and widely used tools for chemical screenings to perform drug sensitivity–related genome studies. Commonly, toxicogenomic studies employ either the haploid gene deletion collections with no gene expression (Parsons et al., 2004; Arita et al., 2009; Zhou et al., 2009; Costanzo et al., 2010; Emadi et al., 2010; Fujii et al., 2010; Stefanini et al., 2010; Kwak et al., 2011) or the heterozygous diploid collection in which gene expression is reduced (also referred to as haploinsufficiency profiling; Giaever et al., 1999; Baetz et al., 2004; Hillenmeyer et al., 2008; Bendaha et al., 2011; Hoepfner et al., 2012). The aim of these approaches is to establish chemical–genetic profiles of specific compounds such as arsenic or naphthoquinones (Zhou et al., 2009; Emadi et al., 2010) or on a larger scale to develop genetic interaction maps (Parsons et al., 2004; Costanzo et al., 2010), both of which can lead to the identification of the compound's target pathways or proteins. By comparing the chemical–genetic profiles of various metabolites, multidrug-resistant genes have been identified which helps in determining compound-specific cellular responses (Parsons et al., 2004). Genomic profiling can in some cases even identify the direct target of a drug as exemplified by tunicamycin and its target gene ALG7 (Giaever et al., 1999). Furthermore, a recent study illustrates the potential of genomic profiling to facilitate the discovery of novel antifungal drugs for a known target, in this case Erg11p (Hoepfner et al., 2012). In general, a major advantage of a chemical–genetic approach is that it is unbiased and may therefore point to novel and unexpected drug targets, side effects or modes of action by covering the whole genomic response. In this study, the physiological, molecular and genomic tools that are available in S. cerevisiae were exploited to assess the range of different tolerance mechanisms available in fungi and to identify precisely the effects of DAPG on yeast mitochondria.

Materials and methods

Yeast strains, culture conditions and chemicals

Saccharomyces cerevisiae strains used in this study were derived from BY4741. The yeast knockout collection (4848 Mata haploid clones; Winzeler et al., 1999) was obtained from Research Genetics. In cases where particular mutants were studied in detail, PCR analysis was used to verify that the correct open reading frame was disrupted. For this, a colony PCR was performed using primers targeting the deletion cassette and a region upstream of the deleted gene (primer sequences were obtained from Saccharomyces Genome Deletion Project, http://www-sequence.stanford.edu/group/yeast_deletion_project/deletions3.html).

Yeast mutant strains from the knockout collection were grown and maintained in either rich medium YPD (1% yeast extract, 2% peptone, 2% glucose, pH 6.5) in the presence of 200 μg mL−1 G418 or minimal medium SD (0.67% yeast nitrogen base without amino acids, 2% glucose, amino acids as required, pH 5.5) or malt extract medium ME (0.6% malt extract, 0.6% glucose, 0.18% maltose, 0.12% yeast extract, pH 5.5). To test for growth in the presence of a nonfermentable carbon source, cells were grown in either rich medium YPGE (1% yeast extract, 2% peptone, 2% glycerol and 2% ethanol, pH 6.5) or minimal medium SGE (0.67% yeast nitrogen base without amino acids, 2% glycerol, 2% ethanol and required amino acids, pH 5.5).

2,4-Diacetylphloroglucinol (DAPG) was synthesized by the Chemistry Department, University College Cork, and verified by HPLC. The chemical was stored as a dry powder at −20 °C and, when required, was dissolved in methanol or DMSO to make a stock concentration of 10 mg mL−1 (equivalent to 48 mM). Stocks of rapamycin (1 mg mL−1), carbonylcyanide m-chlorophenylhydrazone (CCCP; 100 mM) and antimycin A (10 mM) were made in DMSO, and a stock of cycloheximide (2 mg mL−1) was made in distilled water. All chemicals except DAPG were obtained from Sigma.

Genome-wide resistance screen to identify DAPG-tolerant mutants

A genome-wide screen of the yeast deletion collection was performed to identify mutants with increased tolerance to DAPG. The screen was performed using an automated robotic spotting system (2QPix; Genetix). Library plates were first replicated into 96-well plates containing fresh YPD supplemented with G418 and grown for 2 days without shaking to resuscitate mutant strains for maximum viability. Mutants were then spotted on solid ME medium supplemented with 70 μg mL−1 DAPG (concentration was selected due to complete growth inhibition of wild type on particular medium) or the same volume of methanol and incubated for up to 10 days at 30 °C. This was carried out using vented square bioassay trays (Qtray; Genetix) that fit up to twelve 96-well plates (1152 colonies). Growth was monitored and mutants where both duplicates showed growth in the presence of DAPG were picked and further analysed. For primary selection, any growth (visible colony) of mutants within the incubation time was regarded as tolerance. Tolerant mutants were individually restreaked from tolerant colonies on ME agar containing 70 μg mL−1 DAPG and directly compared to wild-type growth, and false positives were excluded from further analysis.

Sensitivity screens of tolerant mutants

To test for cross-resistance to other stresses in tolerant mutant strains, cells of mutant strains were picked from YPD agar plates and inoculated into wells of a 96-well plate containing YPD and grown overnight without shaking. Using disposable replicators, cells were then spotted onto square agar plates (Greiner) supplemented with 1.25 M NaCl, 2.5 μg mL−1 cycloheximide or 70 ng mL−1 rapamycin. In addition, to identify the most DAPG-tolerant strains, mutant strains were also spotted onto agar plates containing 70, 100 and 120 μg mL−1 DAPG. Only mutants growing on the highest concentration were considered most tolerant. Verification of DAPG tolerance and sensitivity testing of the most tolerant mutants was carried out in serial dilutions. Cells were grown overnight in medium used for the assay. Cells were diluted to an A600 of 0.2 in the same medium overnight and diluted 10-fold to 10−3 in a 96-well plate. All dilutions were then spotted on test plates using a metal replicator. Dilutions were spotted in duplicate except for the control plate. The control plate contained the same volume of the solvent as used for the highest treatment concentration. If the chemical was dissolved in distilled water, no supplement was added to control plate. Plates were incubated for 3–4 days at 30 °C. Sensitivity to DAPG and CCCP was tested in the presence of a fermentable carbon source using SD medium (glucose) and in the presence of a nonfermentable carbon source using YPGE medium (glycerol/ethanol). Nonfermentable carbon source was tested to explore the role of respiration/mitochondria in the tolerance of mutants. The selected concentrations (as indicated) resulted from IC50 values determined for each tested medium, and 2 × IC50 was used to illustrate tolerance of mutants. Mutants that grew at wild-type inhibitory concentrations were considered tolerant, whereas mutants not growing at subinhibitory concentrations for wild type were considered sensitive. When DAPG was tested in YPGE, the pH was lowered to 5.5 to adjust it to the pH of SD medium.

In silico analysis of tolerant mutants

The Saccharomyces genome database (SGD; http://www.yeastgenome.org) was used to identify gene functions of tolerant mutants and to group genes according to their gene ontology (GO) annotation for molecular function by similar GO terms. In some cases where GO terms were very specific, a close but more generic parent GO term was selected to group several mutants using the GO website AmiGO (http://amigo.geneontology.org/cgi-bin/amigo/go.cgi). To identify significantly over-represented categories, the FunSpec (functional specification) tool of the University of Toronto (http://funspec.med.utoronto.ca/), which uses the MIPS database and GO database, was used.

To categorize functional groups of genes, a network analysis was performed using Osprey version 1.0.1 (Breitkreutz et al., 2002) and displayed using the layout ‘Concentric circles’ including all genetic and physical interactions. Osprey visualizes complex interaction networks using GO-annotated interactions that are maintained by BioGRID (http://thebiogrid.org/). Subnetworks were identified by filtering interactions for physical interactions and applying the layout ‘Global spoked dual ring’. To display interactions for most tolerant mutants, all single interactions, that is, interactions that were not shared by two or more of the tolerance genes, were deleted. Shared genes linking four tolerance genes were placed in the centre of the network. ‘Unknown’ comprises all gene functions that were not known yet at the time of analysis, whereas ‘others’ groups all functions with singular occurrence to simplify diagram. For reasons of consistency, the same labelling was used in Supporting Information (Table S1).

Measurement of oxygen consumption

Oxygen consumption of yeast cells following chemical treatment was assessed with a phosphorescent oxygen-sensitive probe using a variation of a method developed for similar studies in mammalian cells (Zhdanov et al., 2011; Dmitriev & Papkovsky, 2012). For the oxygen consumption experiments, cells were grown in YPGE overnight and then diluted to A600 0.4 (± 0.03) in YPGE at pH 5.5 and incubated at 30 °C for 2 h. The glycerol/ethanol carbon source was important to ensure that the cells were generating ATP via respiration as opposed to fermentation. An accurate cell concentration was then determined using a haemocytometer, and cells were diluted to 5 × 106 cells mL−1 in YPGE at pH 5.5 in eppendorf tubes, and the O2-sensitive probe MitoXpress™ (Luxcel Biosciences, Ireland) was added to a final concentration of 0.1 μM. Depending on the experiment, CCCP (5 μM), antimycin A (1 μM), DAPG (10, 25, 50 and 100 μg mL−1) or the equivalent amount of DMSO (solvent for all the chemicals) was added to the cells. The range of DAPG concentrations was selected to provide a broad image of the effects. Following addition of the chemical/solvent, 5 × 105 cells (100 μL) supplemented with the probe and substance to be tested were added to the wells of a flat-bottom clear 96-well plate (Sarstedt), and all wells containing cell suspension were covered with 150 μL of mineral oil (Type 37; Cargille) to prevent further oxygen diffusion. The plate was monitored on a TR-F reader Victor X4 (PerkinElmer Life Sciences) at 30 °C shaking before each measurement using a Samarium filter set (340-nm excitation and 642-nm emission). Each sample well was measured repetitively every 1–5 min over 60–90 min by taking two intensity readings at delay times of 30 and 70 μs and gate time of 100 μs. Measured TR-F intensity signals for each sample well were converted into phosphorescence lifetime (τ) values as follows: τ = (t1t2)/ln(F1/F2), where F1 and F2 are the TR-F intensity signals at delay times t1 and t2. From the resulting τ profiles, the initial slopes were calculated (μs min−1) that reflect sample oxygen consumption rates (OCR; Dmitriev & Papkovsky, 2012). Because metabolites could differentially affect yeast growth, it was necessary to include a growth correction when comparing OCR under different conditions. To calculate the growth correction factor for each sample, a replicate experiment was set up, where A600 was separately measured at the time points used to calculate the increase in oxygen consumption. A600 (t2)/A600 (t0) yielded this factor that was used to correct the rates of initial slopes and allow accurate comparisons of respiration in equal numbers of cells. Results were processed using Microsoft Excel, and statistical significance of results was determined by one-way anova (Holm–Sidak method) with a P-value cut-off of 0.001 using SigmaStat 3.5.


Multiple mechanisms are involved in the acquisition of tolerance to DAPG

To determine whether deleting specific genes in yeast confers tolerance to DAPG, the haploid S. cerevisiae gene deletion library (4848 mutants) was screened for growth in the presence of an inhibitory concentration of DAPG (70 μg mL−1 on ME as described in Materials and methods). An initial set of 181 mutants grew under these conditions, with some mutants growing rapidly and others only showing growth after 10 days of incubation. Of these 181 DAPG-tolerant mutants, 27 carried deletions in unverified regions and only the 154 mutants carrying deletions in known or predicted open reading frames were included in subsequent analyses (Table S1). Initially, two different analytical methods were employed to try to determine whether any patterns could be ascertained within these 154 genes that would explain why strains lacking these particular genes were DAPG tolerant. These methods were GO annotation from the SGD database (Fig. 1a) and network analysis using the Osprey package (Fig. 1b). The GO method groups genes by functional categories, whereas the network analysis looks at how genes are connected to each other using data from multiple physical and genetic interaction data sets (see Materials and methods). As is typically the case, the largest group of genes in each analysis were those of ‘unknown function’: 33% (GO) and 23% (network). From GO analysis (Fig. 1a), mutants were grouped into 20 different categories that included functions such as DNA binding (10%), ribosomal protein (7%), transferase activity (5%), hydrolase activity (5%), protein binding (3%), kinase activity (3%), oxidoreductase activity (3%), ATP binding/ATPase activity (3%) and acetyltransferase activity (3%). Studies of interacting genes (Fig. 1b) identified networks in categories such as metabolism (21%), cell organization and biogenesis (19%), DNA-related functions (DNA repair, DNA damage response, DNA metabolism, DNA replication: 10%) and protein biosynthesis genes (6%). Within the network, at the time of analysis, there were 163 genetic and physical interactions identified connecting less than half of the genes (73) to each other, which means that some genes were highly connected, for example RAD52, SLA1, XRS2 or CKB1, whereas many others had few, if any, connections. It must be borne in mind, however, that well-studied genes may feature more strongly in network-type analyses. To gain a different insight, the FunSpec software tool was used to search within the data set to identify categories of genes that were enriched in our collection of 154 mutants (see Materials and methods). This analysis indicated that a number of categories and phenotypes were over-represented in the collection (Table S2). These include ribosomal proteins, sequence-specific DNA binding, slow growth (33/154 slow growers), cytoplasmic location, protein ubiquitination and round morphology. Broadly speaking, all three in silico analytical methods are in agreement and suggest that alterations in a number of different cellular processes can lead to DAPG tolerance. They do not, however, identify any clear DAPG targets or DAPG tolerance mechanisms. The broad representation of different types of genes indicates that there are different ways in which a cell might respond to the inhibitory effects of DAPG and become tolerant.


Informatic analysis of 154 DAPG-tolerant mutants. The 154 DAPG-tolerant mutant strains identified in a phenotypic screen were analysed using informatic methods. (a) Distribution of GO annotations for molecular function according to SGD database including mutants with uncharacterized or verified open reading frames. Functional groups represented by ≥ 4% of mutants are directly labelled. * indicates functional groups represented by 3% of mutants: protein binding, kinase activity, oxidoreductase activity, ATP binding/ATPase activity, acetyltransferase activity. ** indicates functional groups represented by ≤ 2% of mutants: transcription regulator activity, nuclease activity, ubiquitin-protein ligase activity, RNA binding, ligase activity, kinase regulator activity, hydro-lyase activity, amine transmembrane transporter activity. (b) Network analysis of tolerant mutants using Osprey. Network displays physical and genetic interactions to identify shared cellular functions. Arrow indicates group of stress-response genes. Colours correspond to categories placed outside network.

Identification and assessment of the most DAPG-tolerant mutant strains

To address the issue of whether mutant strains were generally ‘stress tolerant’ or more specifically tolerant to DAPG, the collection of 154 mutants was tested for tolerance to 2.5 μg mL−1 cycloheximide, 70 ng mL−1 rapamycin and 1.25 M NaCl. In addition, all mutants were tested for growth at a higher concentration of DAPG (120 μg mL−1). Phenotypically, the majority of the mutants displayed a similar pattern, whereby cycloheximide and rapamycin sensitivity resembled the wild-type strain BY4741, but salt (NaCl) tolerance was increased (Table S1). There were, however, some differences in drug sensitivity observed in a minority of mutants compared to wild type: seven were tolerant to cycloheximide and one was more tolerant and 15 were more sensitive to rapamycin. In summary, no over-riding pattern to stress tolerance was observed in the collection of mutants (Table S1). The analysis at 120 μg mL−1 DAPG showed that some mutants grew better at this concentration than others, although in some cases, mutants were slow growing and it was several days before significant growth was achieved (data not shown). Based on these analyses, six deletion mutants, namely rad52, ydc1, met18, ckb1, ate1 and msh6, were identified as being robustly tolerant to a high concentration of DAPG (Table S1). It is noteworthy that the rad52, met18, ate1 and msh6 mutants also showed cross-resistance to cycloheximide. Only met18 was more sensitive than WT to rapamycin. The rad52, ydc1, ckb1 and ate1 mutants were able to grow at NaCl concentrations that inhibited WT growth (Table S1). These mutants did not fall into a single GO class; therefore, network analysis was carried out to determine if and how the tolerant mutants were linked to each other and to determine possible shared interactions and cellular processes (Fig. 2). Interestingly, four of the genes, RAD52, MET18, CKB1 and MSH6, are highly connected to each other via shared interactions in cellular processes. These processes are mainly, but not exclusively, cell organization and biogenesis, protein synthesis and DNA repair/metabolism/replication, and thus, these data resemble those of the broader analysis of the 154 mutants (Fig. 1). Two genes stand out, however, ATE1, which is weakly connected, and YDC1, which has no connections at all within this subnetwork. Ate1p is an arginine transferase involved in turnover of short-lived proteins via the N-end rule/ubiquitin pathway and is therefore potentially involved in multiple processes in the cell (Balzi et al., 1990; Mogk et al., 2007). Ydc1p is an alkaline dihydroceramidase, involved in sphingolipid metabolism, and although there are no connections in this network, it is noted that one other gene with a function in sphingolipid metabolism, ORM2, was also present in the larger (154 mutants) data set in a group of genes categorized as ‘stress response’ (Fig. 1b). The separation of YDC1 from other genes could indicate that tolerance in ydc1 occurs by a mechanism that is different to other mutants and that this mutant is worth considering further.


Deletion mutants with highest DAPG tolerance are linked by shared genetic and physical interactions. Informatic analysis shows that the RAD52, MET18, CKB1 and MSH6 genes are highly connected to each other and share interactions to similar cellular functions. ATE1 shows few and YDC1 shows no connections to other genes linked to high DAPG tolerance. The network was created using Osprey and only interactions shared by at least two tolerant mutants (bold) are displayed. Single interactions of mutant strains were removed to simplify network.

As part of the verification process of the six mutants identified as being most tolerant to DAPG, strains were exposed to DAPG when growing under different conditions (Fig. 3). It was previously reported that sensitivity was influenced by growth medium (Gleeson et al., 2010) and this was also observed here with yeast more sensitive to DAPG on defined minimal medium (SD) than the malt extract medium used for the initial screens (Fig. 3a). Even on this medium, however, all six mutants are clearly more tolerant to DAPG, continuing to grow at a concentration of 100 μg mL−1. Because it was previously suggested that DAPG might affect mitochondrial function (Gleeson et al., 2010), sensitivity of the mutants was also tested on a nonfermentable carbon source, glycerol/ethanol (SGE: Fig. 3b). Some of the strains grew slowly on this medium, even in the absence of DAPG, and therefore, it was somewhat difficult to accurately compare sensitivities. Nonetheless, with the exception of met18, which grew too poorly to assess, all the mutants were still more DAPG tolerant on SGE (compare strains at 80 μg mL−1 in Fig. 3b). In another effort to test the association with mitochondrial function, the sensitivity of the DAPG-tolerant mutants to CCCP, a proton ionophore that dissipates the proton gradient across the mitochondrial membrane, was assessed. Strikingly, all of the six mutant strains showed a significant cross-resistance to CCCP when tested on SD medium (Fig. 3c). Although all 154 mutants were not tested for tolerance to CCCP, the fact that this was the only inhibitor/stress that gave the same profile as DAPG with these six mutants is an indication that CCCP and DAPG may have similar targets/mode of action.


Tolerance to DAPG and CCCP in yeast deletion mutants. All strains were diluted in a 10-fold series starting with OD 0.2 and spotted on (a) SD plates containing glucose as the sole carbon source and DAPG, (b) SGE plates containing glycerol and ethanol as carbon sources and DAPG and (c) SD plates containing glucose and CCCP, a mitochondrial inhibitor.

CCCP and DAPG have similar physiological effects on mitochondrial function

Previous data showed that DAPG treatment led to the loss of mitochondrial staining with the fluorophore Mitotracker Green™, and our results here indicate that DAPG and CCCP might have similar effects on the cell. On this basis, it was decided to directly measure oxygen consumption in yeast cells treated with mitochondrial inhibitors or with DAPG (Fig. 4). The two mitochondrial inhibitors, antimycin A, which inhibits complex III of the electron transport chain and completely blocks respiration, and CCCP, which dissipates the proton motive force, preventing ATP synthesis, were used (Brand & Nicholls, 2011). These inhibitors have very different physiological effects on the yeast cell: CCCP leads to an increase in the rate of oxygen consumption as the cell tries to compensate for the loss of ATP synthesis by increasing respiration, that is, there is a significant difference between the oxygen consumption in the control and CCCP (Fig. 4, shaded bar). This result reflects previously shown effects of protonophoric uncouplers (Cunarro & Weiner, 1975; Pozniakovsky et al., 2005) and was expected. Therefore, it served as an internal positive control for the assay and verified that calculations for OCR (Table S3) were correct and could be applied for other conditions. In contrast, treatment with antimycin A leads to a decrease in the rate of oxygen consumption as respiration is blocked (Fig. 4, clear bar). Treatment of cells with DAPG at a concentration of 10 and 25 μg mL−1 (Fig. 4, bars with vertical and horizontal hatching) had an identical effect to CCCP in wild-type cells – increased rate of oxygen consumption. It should be noted that absorbance measurements over the course of the experiment showed that growth was slightly inhibited at 10 μg mL−1 and completely inhibited at 25 μg mL−1 (see OD readings in Table S3). Thus, respiration increased prior to growth inhibition, suggesting that it is a primary rather than secondary effect of DAPG. At higher concentrations of DAPG (Fig. 4; 50 μg mL−1, diagonal hatching and 100 μg mL−1, solid fill), there were no growth and also a progressive reduction in the rate of increase in oxygen consumption, although oxygen consumption at 50 μg mL−1 still resembled untreated cells. Even at high concentrations of DAPG, however, respiration was not decreased unlike after treatment with antimycin A. Because ydc1 stood out as a mutant that appeared distinct and unconnected to other DAPG-tolerant strains, this mutant was also assessed with the same treatments (Fig. 4). When the ydc1 mutant was treated with CCCP, there was no difference observed between respiration levels of the untreated and treated cells, indicating that CCCP does not activate respiration at the same concentration as seen in the WT. Furthermore, DAPG treatment of the ydc1 mutant at the lowest concentrations of 10 and the highest concentration of 50 or 100 μg mL−1 did not lead to a significant increase in the rate of oxygen consumption, whereas there was an increase at a concentration of 25 μg mL−1. At the lowest concentration, ydc1 does not seem to be significantly affected by protonophoric action, illustrating its increased tolerance, whereas at the next highest concentration an increased respiration due to uncoupling was observed. At the higher concentrations, respiration is likely to be increased as well, but this effect is masked by growth inhibition. In contrast to CCCP and DAPG, antimycin A had similar effects on ydc1 respiration compared to the wild-type strain.


Oxygen consumption increases in the presence of CCCP and DAPG in wild-type, but not in ydc1, and is inhibited in the presence of antimycin A. Oxygen consumption (respiratory activity) was measured in the wild-type and ydc1 strains using the O2-sensitive probe MitoXpress™. Treatments and shading is as follows: control, dark grey; 5 μM CCCP, light grey; 1 μM antimycin A, white; 10 (vertical lines), 25 (horizontal lines), 50 (diagonal lines) and 100 (black) μg mL−1 DAPG. Oxygen consumption was normalized to cell density, for details, see Materials and methods and Table S3. Error bars indicate SD of mean of four biological replicates. * indicates significant difference to control condition (dark grey; Holm–Sidak, P < 0.001).


DAPG acts as a proton ionophore in S. cerevisiae

In this study, we found that the bacterial metabolite DAPG activates respiration in the yeast S. cerevisiae in a similar manner to the uncoupler CCCP. Uncouplers such as CCCP stimulate respiration in direct correlation with their protonophoric activity across the mitochondrial membrane, a phenotype also observed with DAPG. The basis of uncoupler activity is the presence of an acid-dissociable group, which stabilizes the uncoupler anion in membranes and acts as an efficient proton exchanger (Terada, 1981). In response to the loss of the proton gradient, oxygen consumption (respiration) increases but ATP synthesis cannot take place and so growth is inhibited. The presence of three hydroxyl groups in DAPG illustrates its potential to act as an uncoupler. Data in this study support the contention that DAPG acts as a proton ionophore dissipating the proton gradient across the mitochondrial membrane and are entirely consistent with a previous study showing that DAPG treatment leads to the loss of mitochondrial membrane potential (Gleeson et al., 2010). Dissipation of the proton gradient leads to a loss of mitochondrial function and is now believed to be the primary mode of action of DAPG, although additional effects of ionophoric activity remain possible. In addition, enhanced respiratory activity in mitochondria can lead to the generation of ROS that may compound the inhibitory effects of DAPG. Previous studies found that short (30-min) treatment with DAPG did not generate oxidative stress (Gleeson et al., 2010), but that extended DAPG exposure did generate superoxide (O2) and hydrogen peroxide (H2O2; Kwak et al., 2011). This pattern has previously been observed, and a number of studies illustrate that long exposure of various cell types to uncouplers, including 2,4-dinitrophenol and FCCP, causes an increase in ROS production and reduces glutathione levels (Stockl et al., 2007; Han et al., 2008, 2009; Han & Park, 2011), whereas short-term treatment with uncouplers appears not to generate ROS (Loschen et al., 1971; Tretter & Adam-Vizi, 2007). Although it had not been predicted that DAPG is a proton ionophore, the production of ionophores is not uncommon among bacteria and other examples are valinomycin (produced by a Streptomyces sp.) and cereulide (produced by a Bacillus sp.; Perlman & Bodanszky, 1971; Butaye et al., 2003; Kroten et al., 2010). Following many years of study of DAPG and biocontrol strains of P. fluorescens, this study has now established that the antibiotic activity of DAPG arises from proton ionophoric action. This finding provides a convincing explanation for the broad-spectrum activity of DAPG that does not require postulation of multiple different modes of action. In yeast, and probably in eukaryotes in general, DAPG primarily affects mitochondrial function, whereas in bacteria, it is likely that activity is at the cell membrane. Toxicity to plants may occur at the level of the mitochondrion and also at the chloroplast where inhibition of photosynthesis would be a consequence of loss of the proton gradient.

Reprogramming multiple cellular functions can lead to DAPG tolerance

Tolerance to uncouplers in eukaryotic cells has been previously reported for yeast (Lancashire & Griffiths, 1975; Dupont et al., 1984) and mammalian cells (Lancashire & Griffiths, 1975; Wiseman et al., 1985), but specific molecular mechanisms remain unknown. It is important to know how fungi respond to DAPG and yeast screens are a powerful method to address this on a systems level. The data from this genome-wide study, and from a previous one focusing on sensitivity rather than tolerance (Kwak et al., 2011), demonstrate that alterations in multiple cellular processes can give rise to changes in DAPG sensitivity/tolerance. Although when viewed individually, it can be difficult to explain why particular mutants are tolerant, when considered as reprogrammed metabolism allowing cell functions to continue in the context of disruption of the function of a key organelle, the mitochondrion, the data are plausible. Two key questions are ‘What are the implications of this study for the application of DAPG-producing strains as fungal biocontrol agents?’ and ‘What is the likelihood of fungal tolerance emerging in target organisms?’ In this regard, it is relevant to consider other phenotypes that may be associated with the tolerant mutants (Table S2). One notable feature was that a significant percentage (c. 21%) of mutants tolerant to DAPG were characterized as slow growers according to the Yeast (Table S2), indicating that a decreased metabolic rate may facilitate tolerance. One would have to conclude, however, that tolerance related to slow growth is unlikely to occur in nature due to the overall fitness disadvantage of such variants. It is possible that natural variants in cellular metabolism that confer some degree of DAPG tolerance may exist or arise in fungal populations, and this may explain why some fungi seem naturally tolerant, even in the absence of obvious detoxification or efflux mechanisms (Schouten et al., 2004, 2008). Typically, tolerance to antifungal metabolites arises due to point mutations in the target gene, for example cytochrome b or 14-α-demethylase, the molecular targets of strobilurins and azoles, respectively (Sanglard et al., 1998; Fernandez-Ortuno et al., 2008; Torriani et al., 2009; Cools et al., 2010; Becher & Wirsel, 2012). The absence of a specific target protein for DAPG suggests that tolerance due to single point mutations in a particular gene is unlikely to be as important an issue with this metabolite. Nonetheless, the finding that deletion of YDC1 confers tolerance (discussed below) shows that alternative possibilities do exist. A further aspect of tolerance that was not considered in this particular study was the role that drug efflux systems might play. As outlined in the introduction, previous data from yeast and fungi have not allowed a consistent conclusion as to whether increased expression of ABC transporters, such as Pdr5p or its orthologues, can confer tolerance. The nature of our study of deletions does not allow this question to be addressed, but it was noted that rdr1 was one of the mutants identified in the screen (Table S1). This gene encodes a putative negative regulator of ABC transporters in yeast (Hellauer et al., 2002) so perhaps a mutant has higher basal levels of ABC transporters and can better tolerate DAPG due to export of either DAPG or of toxic metabolites generated by the mitochondrial damage inflicted by DAPG. This is a topic that would need to be further explored, however.

DAPG tolerance in the ydc1 mutant may be linked to sphingolipid homoeostasis

The network analysis suggests that DAPG tolerance in the ydc1 mutant is quite distinct from the tolerance mechanism in most of the other 154 mutants identified in this screen. It was also notable that the ydc1 mutant is tolerant to both DAPG and CCCP (Fig. 3). Ydc1p is an alkaline dihydroceramidase that hydrolyses dihydroceramide to dihydrosphingosine and a fatty acid in yeast, therefore playing a role in the regulation of yeast ceramide levels (Mao et al., 2000). Ceramides are short-lived intermediates in the synthesis of complex sphingolipids that not only play an important role in the structure of eukaryotic cell membranes but are also involved in cellular processes such as apoptosis and stress responses (for reviews, see Hannun, 1996; van Blitterswijk et al., 2003). Yeast cells lacking Ydc1p have been reported to accumulate increased levels of complex sphingolipids due to less breakdown of ceramide (Mao et al., 2000). Another study found that overexpression of YDC1 leads to increased sensitivity to oxidative stress and pH stress and also to a decrease in ceramide levels (Aerts et al., 2008). One possible way for ydc1 to be more tolerant to DAPG may be an alteration in the cell wall or cell membrane with regard to its sphingolipid composition, leading to less accumulation of the drug in the cell. It may also be significant that one other DAPG-tolerant mutant identified in this study, orm2, is reported to have changed levels of ceramide in the cell (Han et al., 2010). It is plausible that tolerance to uncoupling agents like CCCP and DAPG in the ydc1 mutant is due to membrane alterations that reduce uptake of the agent. In general support of this, there is a report linking Ydc1p to the global response of S. cerevisiae to weak acids (Mira et al., 2010). Sphingolipids and their derivatives play another very important role in eukaryotic cells, however, where they act as signal molecules linking many processes in the cell (Spiegel et al., 1998; van Blitterswijk et al., 2003; Liu et al., 2005; Daquinag et al., 2007; Cowart et al., 2010; Vandenbosch et al., 2012). It is intriguing to consider the possibility that sphingolipids may play a more substantial role in response to DAPG. There is a precedent for this as the antifungal plant defensin RsAFP2 interacts with ceramides in the fungal cell membrane and cell wall and induces apoptosis in Candida albicans (Thevissen et al., 2012). Furthermore, it was found that complex sphingolipids (or in fact the lack of them in ipt1 and skn1 mutants) are linked to tolerance to certain antifungal plant defensins in S. cerevisiae. Tolerant strains were, in addition, also more resistant to oxidative stress and showed increased chronological lifespan (Thevissen et al., 2005; Aerts et al., 2006). There is a growing awareness of the importance of sphingolipid homoeostasis in yeast as well as in other eukaryotes, and although the data from this study only provide a hint that there may be links between tolerance to mitochondrial uncouplers and sphingolipid signalling in yeast, this is worthy of further investigation.

Supporting Information

Fig. S1. Oxygen consumption in the presence of DAPG increases over time.

Table S1. Gene names, GO function and cross-resistances of 154 DAPG-tolerant deletion mutants.

Table S2. Funspec analysis showing enriched categories among 154 tolerant mutants.

Table S3. Details for calculation of increase of oxygen consumption rate (ci) in wildtype (wt) and ydc1.


Primary funding support for this project was provided by a grant from the Science Foundation Ireland (08/RFP/GEN1295) to J.P.M. Additional support came from grants awarded by the European Commission (FP7-KBBE-2012-6, CP-TP-312184; FP7-KBBE-2012-6, 311975; OCEAN 2011-2, 287589; MTKD-CT-2006-042062, O36314), SFI (07/IN.1/B948; 08/RFP/GEN1319; 09/RFP/BMT2350), the Department of Agriculture and Food (DAF RSF 06 321; DAF RSF 06 377; FIRM 08/RDC/629; FIRM 06/RDC/459), IRCSET (RS/2010/2413), the Health Research Board (RP/2006/271; RP/2007/290; HRA/2009/146), the Environmental Protection Agency (EPA2006-PhD-S-21; EPA2008-PhD-S-2), the Marine Institute (Beaufort award C2CRA 2007/082), the European Science foundation, and the Higher Education Authority of Ireland (PRTLIIV; PRTLIV). Informative discussions with Nick Read and technical support from Marc McCarthy and Pat Higgins are gratefully acknowledged.


  • Editor: Jens Nielsen


View Abstract