Whole-Genome Approach to Understanding the Mechanism of Action of a Histatin 5-Derived Peptide

The incidence of opportunistic fungal infections that threaten immunocompromised patients, along with the limited arsenal of antifungal drugs, calls for renewed efforts to develop novel antifungal therapies. Antimicrobial peptides have garnered interest as potential therapeutics. Among naturally occurring peptides, histatin 5 is a well-characterized 24-amino-acid peptide with strong antifungal activity.

ABSTRACT

The incidence of opportunistic fungal infections that threaten immunocompromised patients, along with the limited arsenal of antifungal drugs, calls for renewed efforts to develop novel antifungal therapies. Antimicrobial peptides have garnered interest as potential therapeutics. Among naturally occurring peptides, histatin 5 is a well-characterized 24-amino-acid peptide with strong antifungal activity. Our lab has identified a smaller histatin derivative, KM29, with stronger activity against multiple Candida spp., prompting us to investigate its fungicidal mechanism. A genetic screen was developed to test the Saccharomyces cerevisiae genomewide deletion collection for mutants with increased or decreased peptide sensitivity. The goal was to identify genes that would reveal insights into the mechanism of action of KM29, to be assessed in Candida albicans. Several biological processes yielded increased sensitivity, with endosomal transport and vacuolar function appearing at high frequencies. Among the pathways involved in increased resistance, mitochondrial function showed the highest normalized genome frequency; hence, we focused on characterizing this pathway. KM29 localizes to mitochondria, and the killing activity depends on a functional electron transport chain. In addition, KM29 triggered reactive oxygen species (ROS) production, which was responsible for some cell death but insufficient to account for the complete killing activity. In agreement with this finding, we found that KM29 induced mitochondrial fragmentation and a mild loss of mitochondrial membrane potential. Furthermore, respiratory mutants exhibited severely diminished KM29 uptake. We confirmed this behavior in a C. albicans respiratory mutant. Taking our findings together, this work delineates the mitochondrial functions associated with KM29 fungicidal activity and provides additional pathways for further characterization in Candida spp.

INTRODUCTION

Fungal infections represent a growing human health concern. Candida species are the causative agents of the most prevalent fungal infections, causing superficial and systemic candidiasis, the latter associated with high mortality rates (1). Candida albicans is the predominant pathogen, although non-albicans Candida species have gained an alarming presence among clinical isolates over recent years (1). Normally present as nonthreatening commensal organisms in many individuals, Candida spp. are kept in check by the host immune system; however, they can readily cause vaginal and oral mucosal infections under favorable conditions. Immunocompromised patients and those with altered mucosal protective barriers are at risk of disseminated candidemia and invasive candidiasis, the leading nosocomial fungal infections (1, 2). The therapeutic agents used to treat superficial and systemic candidiasis are mainly limited to three classes of compounds: the polyenes, azoles, and echinocandins (3). The azoles are the most prescribed antifungal agents, even though some strains of C. albicans and C. glabrata have acquired resistance to this class of compounds, and other Candida species, such as C. krusei, are innately resistant to azoles at therapeutic concentrations (4, 5). In contrast, the polyenes remain highly effective, but drug toxicity has limited their usage (1, 2). Echinocandins, the most recent addition to the antifungal arsenal, have been shown to be effective with minimal drug toxicity and thus a promising alternative to the former classes (1, 2). However, an increase in acquired resistance has been reported for C. glabrata, likely due to increased drug usage (6–8). The limitations of each class of compounds and the present and future concerns of drug resistance have brought to the forefront efforts to develop new antifungals effective against Candida species and other pathogenic fungi.

One area of research that has garnered interest as a source for novel antimicrobial agents is the utilization of naturally occurring antimicrobial peptides (AMPs). AMPs have been identified in a wide range of organisms, including plants, invertebrates, and vertebrates (9). These peptides often serve as the first line of defense against invading pathogens and are a vital component of innate immune systems (10, 11). One characteristic that has led to the excitement surrounding these peptides is the broad range of activity they display. AMPs can have antibacterial, antifungal, antiviral, and antitumor activities. In addition to the broad range of activities, AMPs display a variety of mechanisms of action (12). For AMPs with antifungal properties, several mechanisms of action have been proposed, including membrane permeabilization, mitochondrial perturbations, initiation of apoptosis, osmotic stress, and DNA damage (13–17). The unique mechanisms of action presented by antimicrobial peptides raise the question of whether some of these could have therapeutic synergism with traditional antifungal agents during combination therapy. Indeed, some synergistic effects between AMPs and antifungal agents have been reported (18, 19).

AMPs share the issues complicating the development of many promising pharmaceuticals, including cytotoxicity, expensive or laborious production, low activity, and low stability (20, 21). One increasingly popular technique used to minimize or circumvent these issues is the design and synthesis of AMP analogs (20, 22). Common ways of manipulating natural AMPs include shortening of the peptide, amino acid substitution, and lipid tagging (20, 21, 23). Shortening of AMPs can not only reduce the cost of production but also increase activity. A notable example of increasing activity by shortening an AMP is dhvar4, a histatin 5-derived peptide (24). Amino acid substitution has shown to be able to improve activity and stability (24–26): activity is usually improved by substituting amino acids to increase hydrophobicity or cationicity, whereas stability is improved by the introduction of d-amino acids or bulky side chains. It is thought that lipid tagging improves activity by increasing membrane association.

Studies from our laboratory have identified a small histatin 5-derived analog, KM29 (Y-K-R-K-F-K-R-K-Y), that yields significant fungicidal activity and is active against a variety of Candida species (27). Structure-function experiments were performed to identify the minimal sequence (Y-K-R-K-F; KM motif) needed to retain fungicidal activity. Moreover, a review of the literature revealed the presence of this motif in nearly every published histatin 5-derived analog. This sequence was used to generate peptide variants with enhanced antifungal activity. Additionally, KM29 displayed minimal red blood cell hemolysis (27). These data suggest the potential for KM29 to be developed into an antifungal therapeutic agent.

In the present study, we used Saccharomyces cerevisiae as a genetically tractable model with which to elucidate the mechanism of action used by KM29 in Candida albicans. We initially developed and performed an unbiased screen to identify S. cerevisiae mutations that confer increased resistance or increased sensitivity to KM29. This screen allowed us to uncover genetic pathways involved in the mechanism of action of KM29. We hypothesized that any biological process implicated in increased resistance to KM29 may represent an intracellular target of the peptide, a way of entry into the cell, or alterations in the cell wall and membranes. The genetic screen revealed several cellular pathways involved in KM29 tolerance, with a preponderance of genes related to mitochondrial function. The possibility that mitochondria serve as a cellular target of KM29 was further investigated. Localization studies showed the association of KM29 with yeast mitochondria. Additionally, and likely as a consequence of KM29 targeting the mitochondria, we found increased production of reactive oxygen species (ROS) and mitochondrial fragmentation upon KM29 treatment. Finally, we found that nonrespiring yeast cells were protected against KM29, predominantly due to decreased KM29 accumulation. The relationship between respiration and KM29 accumulation was observed in C. albicans as well, indicating that S. cerevisiae serves reliably as a surrogate species for the study of the killing mechanism of KM29 in C. albicans. Taken together, these findings suggest that the role of functional mitochondria in the fungicidal activity of KM29 is twofold: they regulate the uptake of KM29 into cells and serve as a site for KM29-induced perturbations.

RESULTS

Genomewide screen for mutants with increased resistance or increased sensitivity to KM29.

In an effort to discern the mechanism of the antifungal action of KM29, we took advantage of the S. cerevisiae nonessential deletion library in the BY4741 background and devised a modified MIC assay that would allow us to identify mutants conferring either increased sensitivity or increased resistance to the peptide. Mutants with increased resistance were identified by the ability to grow in the presence of a concentration of KM29 that completely inhibited wild-type growth, while mutants with increased sensitivity were identified by the inability to grow in the presence of a KM29 concentration that did not inhibit wild-type growth. Of the ∼4,900 mutants tested, 1,360 displayed increased resistance in duplicate, while 508 displayed increased sensitivity in duplicate. These mutants represented ∼27% and ∼10% of the total number of mutants screened, respectively.

To establish potential patterns associated with increased sensitivity or resistance to KM29, we carried out Gene Ontology (GO) analysis. The mutants were grouped on the basis of biological processes; the normalized genome frequency (cluster number/genome number × 100) was calculated for each biological process; and the processes were then ranked in descending order. Gene Ontology analysis of mutants with increased sensitivity to KM29 (Table 1) revealed that the process of invasive growth in response to glucose limitation had the highest normalized genome frequency (27%). The five processes with the next highest frequencies were organelle inheritance (23%), cytoplasmic translation (22%), endosomal transport (21%), vesicle organization (20%), and protein glycosylation (19%). To assess the consistency of this bioinformatics analysis, we used a separate software tool, FunSpec. Analysis of mutants with increased sensitivity to KM29 by FunSpec revealed overlapping annotations between these two techniques relating to endosomal transport and vacuolar function (Table 2). This overlap may indicate the importance of these two processes in the mechanism of action of KM29. Interestingly, the FunSpec analysis revealed a new insight: the divalent cation and heavy metal sensitivity phenotype was significantly enriched in mutants that displayed increased sensitivity to KM29.

TABLE 1.

Gene Ontology analysis of functional pathways involved in increased sensitivity to the antifungal activity of KM29

Biological process	Normalized genome frequency (%)
Invasive growth in response to glucose limitation	27
Organelle inheritance	23
Cytoplasmic translation	22
Endosomal transport	21
Vesicle organization	20
Protein glycosylation	19
46 other processes	10–19
49 other processes	0–9

TABLE 2.

FunSpec analysis of functional pathways involved in increased sensitivity to the antifungal activity of KM29

Category and functiona	P value
GO biological process
tRNA wobble uridine modification	4.98e−07
Protein urmylation	1.32e−06
Arginine biosynthetic process	1.50e−06
GO cellular component
Endosome	2.45e−08
Cytosolic small ribosomal subunit	9.26e−08
Endosome membrane	3.56e−06
MIPS functional classification
Vacuolar/lysosomal transport	2.59e−09
Ribosomal proteins	1.45e−06
Biosynthesis of arginine	1.75e−05
MIPS phenotypes
Other vacuolar mutants	5.70e−08
Slow growth	5.26e−07
Divalent cation and heavy metal sensitivity	2.24e−06
MIPS subcellular localization
Endosome	1.70e−05
Cytoplasm	7.14e−05
Vacuole	1.95e−04

^aGO, Gene Ontology; MIPS, Munich Information Center for Protein Sequences.

Gene Ontology analysis of mutants that displayed increased resistance to KM29 showed enrichment in several functional categories (Table 3). The biological process with the highest normalized genome frequency was mitochondrial translation (37%). The six processes with the next highest frequencies were amino acid transport (36%), cellular respiration (35%), sporulation (35%), mitochondrion organization (31%), generation of precursor metabolites and energy (30%), and “biological process unknown” (30%). Interestingly, four of the seven processes with the highest normalized genome frequencies are related to mitochondrial function (mitochondrial translation, cellular respiration, mitochondrion organization, and generation of precursor metabolites and energy), strongly suggesting that mitochondria are a target of KM29.

TABLE 3.

Gene Ontology analysis of functional pathways involved in increased resistance to the antifungal activity of KM29

Biological process	Normalized genome frequency (%)
Mitochondrial translation	37
Amino acid transport	36
Cellular respiration	35
Sporulation	35
Mitochondrion organization	31
Generation of precursor metabolites and energy	30
Biological process unknown	30
31 other processes	20–29
63 other processes	0–19

It is also interesting that the amino acid transport process had a high normalized genome frequency. This may indicate a way of entry into the cell by KM29. To analyze whether there is specificity for the transport mechanisms that KM29 can utilize, we looked more closely at the genes categorized in this group. Many different amino acid transporters were present, including a leucine permease (BAP2), an arginine permease (CAN1), a proline permease (PUT4), a tryptophan and tyrosine permease (TAT2), a cysteine transporter (YCT1), and a general amino acid transporter (VBA5). The broad representation of transporter mutants that confer resistance on KM29 suggests that the entry of KM29 into cells may be promiscuous and lack specificity.

The presence of the sporulation process among the highest normalized genome frequencies is surprising and difficult to rationalize at this time. The mutants grouped into this process may be generally stress tolerant, a characteristic that could contribute to KM29 resistance; however, this does not preclude the possibility that genes involved in the process of sporulation are also involved in the mechanism of action of KM29.

As shown in Table 4, strains with mutations relating to mitochondrial function were the most overrepresented mutants in almost every category. Many of the mitochondrion-related annotations enriched by FunSpec were the same as those enriched by Gene Ontology analysis. This showed that the two techniques used to analyze the results of the genetic screen are in agreement regarding the importance of mitochondria and mitochondrial function in the mechanism of action of KM29. Given this relevance, we chose to further explore the significance of mitochondrial function for the fungicidal activity of KM29. This is the focus of the rest of the studies presented here.

TABLE 4.

FunSpec analysis of functional pathways involved in increased resistance to the antifungal activity of KM29

Category and functiona	P value
GO biological process
Mitochondrial translation	1.00e–14
Positive regulation of mitochondrial translation	2.67e–07
GO cellular component
Mitochondrion	1.00e–14
Mitochondrial inner membrane	1.95e–08
Mitochondrial small ribosomal subunit	9.20e–08
Mitochondrial large ribosomal subunit	3.80e–07
MIPS functional classification
Mitochondrion	1.76e–10
Aerobic respiration	2.14e–05
MIPS phenotypes, respiratory deficiency	1.00e–14
MIPS subcellular localization
Mitochondria	1.00e–14
Mitochondrial inner membrane	1.36e–06
Mitochondrial matrix	2.62e–05

^aGO, Gene Ontology; MIPS, Munich Information Center for Protein Sequences.

KM29 localizes to mitochondria in S. cerevisiae and C. albicans.

Among the mutants that displayed increased KM29 resistance, there was a prevalence of mitochondrial annotations that prompted us to investigate the relationship between mitochondrial function and the mechanism of action of KM29. To study whether there was a direct association between KM29 and mitochondria, the internal localization of KM29 was investigated by fluorescence microscopy. S. cerevisiae and C. albicans cells were incubated with fluorescein isothiocyanate (FITC)-labeled KM29 (KM29-FITC) at 5 μM and 3 μM, respectively, and concurrently with 500 nM MitoTracker Deep Red. Preliminary work showed that labeling with FITC did not affect the fungicidal activity of KM29, as assessed by MIC assays (data not shown). KM29-FITC localized in a punctate pattern, with some cells displaying increased localization near the cell periphery in S. cerevisiae and C. albicans (Fig. 1). In both organisms, partial colocalization occurred between KM29-FITC and the mitochondrion-specific stain MitoTracker Deep Red. These data suggest that once internalized by S. cerevisiae or C. albicans, KM29 is directly associated with mitochondria, even though some KM29 remains cytosolic, perhaps due to interaction with other cellular targets.

FIG 1.

Intracellular localization of KM29-FITC in S. cerevisiae and C. albicans. Shown are representative fluorescence microscopy images from a double-labeling experiment. Wild-type cells were treated concurrently with 500 nM MitoTracker Deep Red (Mito Tracker DR) and either 5 μM KM29-FITC (S. cerevisiae) or 3 μM KM29-FITC (C. albicans).

The fungicidal activity of KM29 is dependent on ETC function.

The results from the genomewide screen and the KM29 localization study suggested that mitochondria play a role in the mechanism of action of KM29 and led us to hypothesize that the process of aerobic respiration plays an important role in the fungicidal activity of KM29. To test this hypothesis, as well as to confirm the results of the genetic screen, we performed microdilution killing assays on S. cerevisiae mutants with deletions in the electron transport chain (ETC) complexes (Fig. 2). NADH:ubiquinone oxidoreductase (Ndi1) is the only subunit of the yeast complex I that is responsible for transferring electrons from NADH to ubiquinone in the respiratory chain (Saccharomyces Genome Database [SGD] [http://www.yeastgenome.org/]). The ndi1Δ strain did not show significantly increased resistance to KM29 in either the microdilution killing assays or the genomewide screen (Fig. 2). We next assayed Sdh2, a subunit of succinate dehydrogenase and a constituent of complex II, which receives electrons from reduced flavin adenine dinucleotide (FADH₂) (SGD). The sdh2Δ strain showed significantly increased resistance to KM29, in agreement with its identification in the genomewide screen. It should be noted that limited respiratory activity can still occur in the absence of complex I or II, which may explain the ability of KM29 to exert some fungicidal activity against mutants with single deletions in these complexes.

FIG 2.

Mutations in the electron transport chain affect the fungicidal activity of KM29. The viability of cells of wild-type (WT) S. cerevisiae and the indicated strains with mutations in each complex of the ETC was determined by a microdilution assay with 2.5 μM KM29. Ndi1, NADH:ubiquinone oxidoreductase; Sdh2, succinate dehydrogenase subunit; Qcr7, ubiquinol cytochrome c oxidoreductase subunit; Qcr10, nonessential subunit of ubiquinol cytochrome c oxidoreductase; Cox9 and Cox12, cytochrome c oxidase subunits. Error bars represent standard errors of the means. Asterisks indicate statistically significant differences from the wild-type strain (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

The next two complexes in the ETC series, complex III (cytochrome bc₁ complex) and complex IV (cytochrome c oxidase), are essential for respiration (SGD [http://www.yeastgenome.org/]). Deletion of a gene encoding an essential subunit in either of these complexes (the qcr7Δ, cox9Δ, or cox12Δ mutation) results in significantly increased resistance to KM29, with the mutants almost reaching 100% survival (Fig. 2). These three mutants also displayed high resistance in the genomewide screen. This increased resistance is presumably due to the complete lack of respiration in these mutants. QCR10 encodes a nonessential subunit in the cytochrome bc₁ complex; it is the last subunit to be assembled into the complex but is not required for respiratory growth (28). According to our hypothesized relationship between aerobic respiration and the mechanism of action of KM29, we would expect a qcr10Δ strain not to be resistant to KM29, since it is still able to respire. Indeed, qcr10Δ cells were sensitive to KM29 and did not differ significantly from the wild-type strain. Surprisingly, the qcr10Δ mutant was also identified in the screen for increased resistance to KM29, likely as a result of the extreme sensitivity of the MIC assay in detecting slight perturbations in growth rates. Taken together, these results confirm our hypothesis that the fungicidal activity of KM29 is dependent on ETC function and that cells defective in respiration display increased resistance to KM29.

ROS production in S. cerevisiae.

The results presented above, implicating the ETC in KM29 activity, raise the possibility that perturbations of the ETC may be involved in initiating/causing cell death. One of the most common outcomes of aberrant ETC function is ROS formation, which can lead to cellular damage. To study whether KM29 induces ROS formation as part of its mechanism of action, we used the common fluorescent probe 2′,7′-dichlorofluorescin diacetate (DCFDA). When S. cerevisiae cells were treated with 333 μm KM29, there was a time-dependent increase in ROS levels, as indicated by the time-dependent increase in relative fluorescence (Fig. 3A). These levels of ROS are close to the maximum that can be formed by KM29, since treating cells with 833 μm KM29 did not produce a significant increase in ROS production. When cells were treated concurrently with N-acetyl cysteine (NAC), a known ROS scavenger, and KM29, ROS accumulation was completely abolished, as indicated by the lack of increase in relative fluorescence. When a variation of this experiment was performed by pretreating cells with NAC instead of treating them with KM29 and NAC concurrently, the same effect was seen (data not shown). The latter experiment also served as a control, eliminating the possibility of competition between NAC and KM29 for entry into cells. This pattern was also seen when C. albicans cells were used (Fig. 3B), showing consistent responses for these two fungal species.

FIG 3.

Time-dependent formation of ROS by KM29. (A and B) Fluorescence spectrophotometry analysis of wild-type S. cerevisiae (A) or C. albicans (B) cells preloaded with 10 μm DCFDA and then treated with KM29 alone or KM29 plus NAC. (C and D) Survival of S. cerevisiae (C) and C. albicans (D) cells treated with KM29 alone (filled bars) or KM29 plus NAC (shaded bars). (E) Comparison of KM29-induced ROS production in S. cerevisiae and C. albicans. Error bars represent standard errors of the means.

To assess whether there is a correlation between KM29-induced ROS formation and KM29-induced cell death, the fungicidal activity of KM29 was assessed when cells were treated concurrently with NAC and KM29 (Fig. 3C and D). As shown above, NAC treatment completely abolished KM29-induced ROS production, and concomitantly, NAC treatment led to significant increases in survival at each KM29 concentration in both C. albicans and S. cerevisiae (Fig. 3C and D). At 333 μm KM29, survival was increased by 26% in S. cerevisiae and 77% in C. albicans, while at 833 μm KM29, survival was increased by 60% in S. cerevisiae and 62% in C. albicans. It should be noted that NAC treatment did not completely protect cells from the fungicidal activity of KM29, indicating that ROS formation may not be the only source of damage caused by KM29. It is also noteworthy that when KM29-induced ROS production in S. cerevisiae was compared with that in C. albicans, treatment of S. cerevisiae cells with KM29 clearly produced more fluorescence (Fig. 3E). Whether this increased fluorescence is due to increased ROS production or simply to increased DCFDA uptake remains to be elucidated.

KM29 causes mitochondrial fragmentation and mild MMP depolarization in S. cerevisiae.

In order to explore other mitochondrial perturbations possibly caused by KM29, we employed the MitoLoc system to assay mitochondrial membrane potential (MMP) and mitochondrial network morphology in vivo. A plasmid expressing green fluorescent protein (GFP)-tagged pre-SU9 and mCherry-tagged pre-COX4 was introduced into wild-type and qcr7Δ strains. pre-SU9 localizes to mitochondria independently of the MMP, thus robustly labeling mitochondrial structures, while pre-COX4 localization to mitochondria is MMP dependent. Colocalization of the two markers allows for single-cell quantification of MMP and is indicated by the Pearson correlation coefficient (PCC) (29). Transformed wild-type cells were treated with either 333 μM KM29 or 15 μM carbonyl cyanide m-chlorophenylhydrazone (CCCP) (as a positive control for MMP disruption). As expected, the PCC was significantly decreased when cells were treated with CCCP, indicating a decrease in MMP (Fig. 4A). Cells treated with KM29, however, showed a mild decrease in MMP, not enough to appear significantly different from untreated cells (Fig. 4A). We included qcr7Δ cells as an additional control. Since these cells are respiration deficient, we expected the PCC to be decreased, as indeed was the case. With regard to mitochondrial network morphology, KM29-treated cells displayed severe mitochondrial fragmentation, in contrast to untreated cells (Fig. 4B). Also noteworthy is the fact that mitochondrial fragmentation did not occur in either CCCP-treated cells or transformed qcr7Δ cells. These data show that the structure of mitochondria is significantly affected by KM29 treatment, in agreement with KM29 localization to mitochondria.

FIG 4.

Effects of KM29 on mitochondrial membrane potential and mitochondrial morphology. Wild-type S. cerevisiae cells transformed with pMitoLoc were treated with 166 μm KM29 or 15 μm CCCP for 3 h, and single cells were analyzed using the ImageJ plugin yeast_correlation. Respiration-deficient qcr7Δ cells carrying pMitoLoc served as a positive control. (A) Loss of colocalization of the two markers is reflected by a decrease in the Pearson correlation coefficient (coeff.). Error bars represent standard errors of the means. Asterisks indicate statistically significant differences from the wild-type strain (*, P < 0.05; **, P < 0.01). (B) Fluorescence microscopy images of a representative cell from each experimental condition.

Respiratory status inversely regulates KM29 uptake in S. cerevisiae.

The results thus far revealed mitochondria as one of the sites for KM29-induced perturbations; however, these findings had not clarified the mechanism by which respiratory deficiency confers resistance to KM29 on cells. To investigate whether the resistance associated with respiratory deficiency is a consequence of a cytoplasmic effect or of peptide uptake, we measured KM29-FITC accumulation in the ETC mutants presented above. Surprisingly, the respiratory capability of each mutant correlated inversely with KM29 accumulation: mutants that were completely respiration deficient had minimal KM29 accumulation, and mutants with partially decreased respiratory capabilities had intermediate levels of KM29 accumulation (Fig. 5A and B). The decrease in KM29 uptake is reflected in the killing assays presented in Fig. 2. We also looked at the accumulation and fungicidal activity of KM29 in ATP synthase mutants. Since ATP synthase is the final complex involved in the production of energy from aerobic respiration, we hypothesized that these mutants would be deficient in KM29 uptake. We chose strains carrying deletions of ATP5, a subunit of the stator stack domain, and ATP17, a subunit of the F_o coupling factor of the F₁F_o ATP synthase. Confirming our hypothesis, both atp5Δ and atp17Δ mutants showed minimal KM29 accumulation and close to 100% survival when treated with 2.5 μM KM29 (Fig. 5C and D).

FIG 5.

Involvement of the ETC in the uptake and killing activity of KM29. (A) Fluorescence spectrophotometry analysis of KM29-FITC uptake in wild-type S. cerevisiae and ETC mutants. (B) Fluorescence microscopy images of KM29-FITC uptake in wild-type S. cerevisiae (respiration capable) (left), qcr7Δ (respiration deficient) (center), and qcr10Δ (respiration capable) (right) cells. (C) Fluorescence spectrophotometry analysis of KM29-FITC uptake in wild-type S. cerevisiae and ATP synthase mutants. (D) Viability (determined by a microdilution assay) of wild-type S. cerevisiae and ATP synthase mutant cells treated with 2.5 μM KM29. Error bars represent standard errors of the means. Asterisks indicate statistically significant differences from the wild-type strain (**, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

Respiratory status correlates inversely with KM29 uptake and killing in C. albicans.

The purpose of the genetic screen of the available S. cerevisiae genomewide deletion collection was to use this amenable model organism as a surrogate species with which to learn about the killing mechanism of KM29 in Candida species. Therefore, we asked if our findings in S. cerevisiae would extend to the clinically relevant species Candida albicans. To do this, we analyzed KM29 uptake and killing in a coq3Δ mutant of C. albicans. COQ3 encodes an enzyme involved in the synthesis of mitochondrial ubiquinone. Ubiquinone shuttles electrons from complexes I and II of the ETC to complex III (30). The absence of ubiquinone results in respiratory deficiency; thus, we would expect the levels of KM29 uptake and killing to be lower in a coq3Δ mutant than in the wild type. The coq3Δ mutant had significantly decreased KM29 uptake, and this is reflected in the survival rates after both the mutant and the wild type were treated with 1 μM KM29 (Fig. 6A). In addition, the S. cerevisiae coq3Δ mutant shared this phenotype (Fig. 6C). Interestingly, when we compared KM29 uptake in wild-type C. albicans and S. cerevisiae cells (Fig. 6B and D), C. albicans had substantially higher KM29 uptake, which correlates with the increased sensitivity of C. albicans to KM29. These data indicate that although differences in tolerance to KM29 exist between these organisms, the correlation between respiration and KM29 uptake is conserved between the two. Our findings thus provide a solid foundation for further analyses of additional pathways revealed by the S. cerevisiae screen.

FIG 6.

Effects of respiratory deficiency on KM29 uptake and killing in C. albicans and S. cerevisiae. (A) Microdilution assay showing the viability of C. albicans LLF100 (wild-type) and C. albicans LLF088 (coq3Δ; respiration deficient) cells treated with 1.0 μM KM29. (B) Fluorescence spectrophotometry analysis showing the uptake of 5 μm KM29-FITC in C. albicans LLF100 (wild-type) and C. albicans LLF088 (coq3Δ; respiration deficient) cells. (C) Microdilution assay showing the viability of wild-type and coq3Δ S. cerevisiae cells treated with 2.5 μM KM29. (D) Fluorescent spectrophotometry analysis showing the uptake of 5 μm KM29-FITC in wild-type S. cerevisiae and the coq3Δ mutant. Error bars represent standard errors of the means. Asterisks indicate statistically significant differences from the wild-type strain (*, P < 0.05; ***, P < 0.001).

DISCUSSION

The goal of this study was to elucidate the mechanism of fungicidal action of the novel antifungal peptide KM29, derived from naturally occurring histatins. We initially developed and performed an unbiased screen of the S. cerevisiae haploid deletion collection in order to identify mutations that confer increased resistance or increased sensitivity to KM29, in an effort to uncover genetic pathways involved in the mechanism of action of KM29. Of the ∼4,900 mutants tested, 1,360 (27% of total mutants) displayed increased resistance, while 508 (10% of total mutants) displayed increased sensitivity. We hypothesized that any biological process implicated in increased resistance to KM29 may represent an intracellular target of the peptide, a way of entry into the cell, or alterations in the cell wall and membranes. In contrast, we hypothesized that any biological process implicated in increased sensitivity may represent a defense system against KM29 fungicidal effects, or alterations in the cell wall and membranes. Although these are broad hypotheses, they served as our basis for analyzing the results of the genetic screen.

With regard to the mutants that conferred increased sensitivity to KM29, we found endosomal and vacuolar mutants to be at the intersection of the two independent bioinformatic analyses that we performed. It is interesting that these two systems are implicated in sensitivity to KM29, since they both have functions in the degradation of cell membrane proteins (31). This process begins with endocytosis and the formation of vesicles carrying membrane components. These vesicles coalesce to form early endosomes. From here, cargo in early endosomes is sorted into those components destined for degradation in the vacuole and those destined for recycling in the Golgi apparatus. It is possible that upon contact with the cell membrane, KM29 induces damage in cell membrane components, which signals these components to be endocytosed and degraded. Impairment of the endocytosis system may result in the accumulation of damaged molecules on the cell membrane, thus resulting in the collapse of cell membrane integrity and, eventually, cell death. Likewise, impairment of the system that transports these damaged molecules to the vacuole, or impairment of the vacuole itself, may result in the intracellular accumulation of damaged membrane components and may cause further damage. This explanation is particularly interesting given the cationic nature of KM29 and the numerous reports of cationic molecules interacting with membranes on various cells to induce pore formation or membrane damage (32–34).

Vacuoles also play an important role in cellular homeostasis by storing various ions and molecules (35). In a scenario where KM29 induces osmotic stress, the vacuole would serve to counter this stress. Impairment of vacuolar function would make the cell less able to cope with the osmotic stress caused by KM29, thus causing increased sensitivity. Confirmation of this hypothesis would indicate a conservation of the mechanism of action between KM29 and histatin 5, the peptide from which KM29 was designed, since a study found that histatin 5 induces osmotic stress in C. albicans (16). Taken together, these findings led us to hypothesize that KM29 induces cellular membrane damage and/or alterations in cellular homeostasis. Further work will be needed to analyze the specific contributions of these pathways to the fungicidal activity of the peptide.

With regard to the mutants that conferred increased resistance to KM29, we initially hypothesized that any biological process implicated in increased resistance to KM29 may represent a cellular target, a way of entry into the cell, or an alteration in the cell membrane or cell wall. We believe that our results support this hypothesis by identifying amino acid transport as a possible way of entry into the cell and mitochondria as a possible target for the peptide.

Amino acid transport had the second highest normalized genome frequency in our screen. We interpret this result as showing that KM29 can hijack various amino acid permeases and transporters as a way of entry into the cell. It would make sense for a deletion of one of the permeases to confer increased resistance to KM29, since there would be less cytoplasmic accumulation of KM29 and therefore less cellular damage. As stated above, we did not find any pattern or specificity for the type of transporter that KM29 can utilize based on the results of the genetic screen. These data open up the intriguing possibility that this 9-amino-acid cationic peptide interacts with highly specific but unrelated membrane transporters in a nonspecific manner to enter the cell. Alternatively, the effect may be indirect; cells lacking certain transporters may undergo changes on the plasma membrane that affect permeability to KM29.

Also, as part of our initial hypothesis, we postulated that biological processes implicated in increased resistance to KM29 would represent a cellular peptide target. Our reasoning was that if a cellular peptide target were deleted or impaired, KM29 would be unable to induce damage, thereby increasing survival. Our screen identified the mitochondria as a drug target of KM29, because deletion mutations of a myriad of genes related to mitochondria and mitochondrial function were found to increase resistance to KM29. The screen, however, did not identify a specific molecular target of KM29 similar to what others have found for the antifungal agents auranofin and honokiol (36, 37). This likely indicates that KM29 has a generalized mode of action with regard to mitochondria, such as disturbing mitochondrial integrity or MMP. The consequences of this generalized mode of action could include ROS production, a decrease in ATP levels, or cell death signaling, all of which would have adverse effects on cells.

The possibility that mitochondria serve as the cellular target of KM29 is exciting. Several studies have highlighted the potential of developing antifungals that target this organelle and the advantages associated with such development (38, 39). These advantages include the crippling damage that mitochondrial perturbations cause to yeast cells due to the critical cell functions in which this organelle is involved. Additionally, many of the fungal mitochondrial factors that could be used as potential drug targets do not have close orthologs in humans, a feature that would reduce drug toxicity (38). Mitochondrial dysfunction has also been implicated in hypersusceptibility to fluconazole, stressing the relevance of this organelle in the mechanism of action of antifungals (39).

To follow up on our hypothesis, we focused on the role of mitochondrial function in the fungicidal activity of KM29. Our observation of the localization of KM29 to mitochondria, as well as the increased resistance to KM29 resulting from impairment of the electron transport chain, supports our proposed mechanism. The finding that KM29 localizes to mitochondria is interesting, since an independent study found that the cellular target of histatin 5 is the mitochondrion (40). Although that group proposes that the localization of histatin 5 to mitochondria is due to structural resemblances between histatin 5 and mitochondrial presequences, we do not believe that this is the case for KM29. It is not clear why KM29 localizes to mitochondria, but we believe that the cationic nature of this peptide likely plays a role in this process through electrostatic forces. These electrostatic forces might include the attraction of KM29 to the negatively charged phospholipids in the mitochondrial outer membrane (41). Attraction to and interaction with negatively charged phospholipids is considered to be a mechanism common to most AMPs (42). Additionally, it has been reported that protein translocation across lipid bilayers can be potential dependent (43). The proton gradient formed across the inner mitochondrial membrane in actively respiring cells may serve as an attractant for KM29 and may contribute to its translocation across the mitochondrial membrane.

Adding to the implication of mitochondrial function in the mechanism of action of KM29, we also observed that KM29 induces time-dependent ROS formation. Although it has been reported that histatin 5 could induce ROS formation, and this ROS formation was strongly correlated with cell death (14), there is controversy surrounding the role of ROS in the mechanism of action of histatin 5, since a separate group did not observe ROS formation in response to histatin 5 exposure (44). In our case, KM29-induced ROS formation was strongly linked with cell death, as evidenced by the fact that concurrent treatment with a ROS scavenger significantly increased survival. However, this treatment did not completely protect cells from the fungicidal activity of KM29; therefore, we cannot conclude that ROS production is the only source of KM29-induced damage. Our analyses of mutants from each complex of the ETC showed various levels of resistance. Therefore, it is unlikely that KM29-induced ROS production is the result of disruption of a specific site in the ETC; rather, it is likely a consequence of generalized disruption of mitochondria. Considering the numerous types of molecules that ROS can damage, including nucleic acids, lipids, and proteins, it is reasonable that ROS production plays an important role in the mechanism of action of KM29. Additionally, ROS can induce apoptosis and necrosis (45). For further development of KM29 into a therapeutic agent, it will be useful to know the downstream effects of KM29-induced ROS production, including the molecules damaged and the type of cell death initiated.

In order to gain a deeper understanding of how KM29 may disrupt mitochondria, we sought to determine the health status of mitochondria after KM29 treatment. Given the relationship between the health status of mitochondria and mitochondrial network morphology, we analyzed the effect of KM29 on this parameter. We also analyzed MMP, since this parameter can influence mitochondrial network morphology. We observed severe KM29-induced mitochondrial fragmentation independent of significant alterations in MMP. This could be a sign of mitochondrial damage, since the process of mitochondrial fission has been shown to protect healthy mitochondria and to precede the degradation of damaged mitochondria (46, 47). Damaged mitochondria will ultimately be eliminated through mitophagy, a process that serves as an internal quality control for the organelle (46). Additionally, the mitochondrial fragmentation observed could be one of the effects of KM29-induced ROS production, since ROS can induce mitophagy (48). In an independent study, H₂O₂, a known inducer of oxidative stress, has been shown to cause the same phenotype, mitochondrial fragmentation independent of MMP alterations (29); this finding supports our characterization of KM29 as an oxidative-stress inducer. Overall, our findings indicate that KM29 induces high levels of mitochondrial stress, consistent with studies that have found mitochondrial fission to be associated with high levels of stress, while mitochondrial fusion is associated with low levels of stress (47).

The findings presented above provided a partial elucidation of the role of mitochondria in the mechanism of action of KM29, but it remained unclear why respiratory deficiency conferred near-complete resistance to KM29. The phenomenon of respiratory impairment rendering cationic peptides and proteins less effective is longstanding, as evidenced by various reports (49–51). Several explanations for this phenomenon have been proposed. One group hypothesized that respiratory deficiency conferred resistance to histatin 5 due to a reduction in histatin 5-induced ROS production (14). This hypothesis may be applicable to KM29, given the apparent importance of ROS production in KM29 activity; however, our results suggest that the tolerance of the mitochondrial mutants that we tested is due instead to reductions in KM29 accumulation. This is in agreement with the report suggesting that the histatin 5 resistance of C. albicans respiratory mutants is due to reductions in histatin 5 accumulation (50). Given the numerous reports of respiratory impairment conferring resistance to cationic peptides and proteins, it is likely that this is a general mechanism of peptide resistance by fungal cells. The molecular mechanism behind this phenomenon is not readily clear. A reasonable explanation might involve the energy availability in these mutants. It has been reported that the plasma membrane proton ATPase is the major facilitator in establishing the plasma membrane potential of the yeast cell (52). This raises the question of whether respiration-deficient mutants, because of the decrease in ATP output, have decreased proton ATPase activity and, therefore, a depolarized plasma membrane potential. Given the apparent importance of electrostatic forces in the mechanisms of cationic agents, this could explain the inability of KM29 to enter these mutants (33, 34, 42). This explanation is contested, though, by reports showing that plasma membrane potential is not severely affected by changes in respiratory activity (53). Other explanations suggest that resistance might stem from changes in plasma membrane composition and rigidity (51). Not only are mitochondria the “powerhouse” of the cell, but they are also involved in lipid synthesis (54). Since lipids are a major constituent of the plasma membrane, alterations in lipid synthesis due to respiratory deficiency might cause alterations in plasma membrane composition, leading, in turn, to alterations in KM29 accumulation. Additionally, the reduced energy availability might lead to physical rigidification of the plasma membrane through cytoskeletal mechanisms, thus affecting the accumulation of KM29. This explanation was originally presented by a study showing that a cytoskeletal inhibitor could mimic the effects of mitochondrial inhibitors on histatin 5 activity (51).

In summary, we show that S. cerevisiae serves reliably as a surrogate species with which to study the killing mechanism of KM29 in C. albicans, and we present initial steps in the elucidation of the mechanism of action of KM29 in fungal cells. Based on the data, we propose that KM29 has a multistep mechanism of action with the following sequence of events. (i) KM29 enters the cell via interaction with the plasma membrane, resulting in (ii) pore formation and nonspecific opportunistic use of plasma membrane transport mechanisms, as well as (iii) cellular damage caused by the interaction of KM29 with the plasma membrane, followed by (iv) targeting to the mitochondria. Once KM29 is inside the intracellular compartment, (v) mitochondrial function becomes perturbed, in part by alteration of the mitochondrial network and to a lesser extent by alteration of the membrane potential. The possible multifaceted mechanism by which KM29 induces cell death makes this peptide an interesting antifungal agent, since this feature would make it more difficult for pathogenic yeasts to acquire drug resistance. By understanding the mechanism of action of KM29 in S. cerevisiae, we can now begin to understand how this peptide works against C. albicans and other Candida spp. and to provide parameters for continuing to develop KM29 as a potential therapeutic agent for the treatment of mucosal and systemic candidiasis.

MATERIALS AND METHODS

Yeast strains, culture conditions, and peptides.

The genetic screen and all experiments involving S. cerevisiae were carried out with the MATa haploid deletion library in the BY4741 background (55, 56). Unless otherwise indicated, strains were grown on standard YPD medium (1% yeast extract, 2% peptone, 2% glucose [wt/vol]). Plates contained 2% (wt/vol) agar (Difco). S. cerevisiae was grown on a solid medium for 48 h at 30°C and in a liquid medium for 24 h at 30°C. C. albicans was grown on a solid medium for 24 h at 37°C and in a liquid medium for 24 h at 30°C.

KM29 (Y-K-R-K-F-K-R-K-Y), synthesized with N-terminal acetylation and C-terminal amidation, was obtained from GenScript. The peptide was dissolved in deionized water to a final concentration of 11 μM (54 μg/ml). The concentration was determined using Beer’s law with the extinction coefficient of KM29 (E₂₈₀ = 2,980 M⁻¹ cm⁻¹).

KM29-FITC, synthesized with an N-terminal FITC–γ-aminobutyric acid (GABA) linker and C-terminal amidation, was obtained from LifeTein.

Fluorescence microscopy.

For standard fluorescence microscopy, cells were examined with a Zeiss AxioImager M1 microscope using Zeiss filter sets 38 (GFP), 50 (Cy5), and 20 (rhodamine). Confocal fluorescence images were obtained with a Leica SP5 microscope with FITC and tetramethyl rhodamine isocyanate (TRITC) filter sets.

Genomewide screen for mutants with increased resistance or increased sensitivity to KM29.

Previous characterization of the KM29 peptide indicated promising antifungal properties. KM29 showed strong antifungal activity against C. albicans, with a MIC range of 4.2 to 8.4 μg/ml, compared to 1.4 to 2.1 μg/ml for fluconazole, by use of a broth microdilution susceptibility test based on the Clinical and Laboratory Standards Institute (CLSI) method (27, 57). Since cell growth was monitored with a colorimetric bioassay (resazurin; see description below), the MIC was equivalent to the minimum fungicidal concentration. In order to perform a genomewide screen in S. cerevisiae, we used the CSLI guidelines (57) to develop a MIC assay with the following modifications required for the growth conditions of S. cerevisiae. Mutant strains from the S. cerevisiae MATa haploid deletion library were collected from frozen stocks, inoculated on OmniTray (Nunc) YPD plates, and grown for 48 h at 30°C. Deletion strains were then transferred to 150 μl of sterile distilled water (dH₂O) in 96-well plates via 96-well pin replicators. The optical density at 600 nm (OD₆₀₀) of the cells in each well was measured with a spectrophotometer (Eon; BioTek), and the average OD₆₀₀ of the cells in all the wells was calculated. The cell density in individual wells was adjusted by adding more cells or more dH₂O to achieve the average OD₆₀₀. Cells were then added to a 96-well plate containing 0.125× SD+Trp (synthetic dextrose plus 400 μM tryptophan [pH 6]) supplemented with 1% (wt/vol) glucose, 4 μM (5.4 μg/ml) or 2 μM (2.7 μg/ml) KM29, and 50 μg/ml resazurin (Sigma) to a final concentration of 5.25 × 10⁴ cells/ml. A 4 μM concentration of KM29 was used to test for increased resistance, since wild-type cells display no growth at this concentration, and growth of a deletion mutant would indicate increased resistance. A 2 μM concentration of KM29 was used to test for increased sensitivity, since wild-type cells display growth at this concentration, and lack of growth of a deletion mutant would indicate increased sensitivity. Each plate contained one well with a wild-type control (BY4741) and one well with no cells as a negative control. Plates containing 4 μM KM29 were incubated for 43 to 48 h, and plates containing 2 μM KM29 were incubated for 29 to 32 h, at 30°C. After incubation, plates were analyzed for a color change of the pH indicator, resazurin, a compound commonly used to assess cell growth and drug susceptibility (58, 59). Resazurin has a blue color at pH 6.0 and, as the pH drops, in this case due to by-products of fermentative growth, transitions to pink/red. Resorufin, the acid form of resazurin, can also be fully reduced to form the colorless compound dihydroresorufin. Wells were analyzed for a transition from blue to pink/red/colorless and were labeled as having growth or no growth. The testing of each plate was repeated in duplicate for each concentration of KM29, and deletion mutants that showed growth or no growth in both trials were considered to exhibit increased resistance or increased sensitivity, respectively, to KM29.

Bioinformatic analysis of genes identified in the genomewide screen.

Two clustering methods were carried out on the mutants identified in the genetic screen in order to categorize any patterns associated with increased resistance or increased sensitivity. The first clustering analysis method employed was Gene Ontology (GO) annotation from the Saccharomyces Genome Database (SGD [http://www.yeastgenome.org]). Open reading frame (ORF) lists for mutants that displayed increased resistance or increased sensitivity were input into the Yeast GO Slim Mapper and grouped by “Process.” The second clustering analysis method employed was FunSpec (http://funspec.med.utoronto.ca/). ORF lists for mutants that displayed increased resistance or increased sensitivity were input into FunSpec and queried against all available yeast databases, with a P value cutoff of 0.01.

KM29-FITC localization and accumulation in S. cerevisiae and C. albicans.

S. cerevisiae or C. albicans cells from an overnight culture were pelleted and resuspended in either 5 mM or 10 mM sodium phosphate (NaP) buffer (pH 7.4), respectively, to a final concentration of 1 × 10⁸ cells/ml. For localization studies, either KM29-FITC, MitoTracker Deep Red (Molecular Probes), or both concurrently were added to cell suspensions to final concentrations of 5 μM or 3 μM KM29-FITC and 500 nM MitoTracker Deep Red. For accumulation studies, KM29-FITC was added to cell suspensions to a final concentration of 5 μM. After incubation for 30 min at 30°C, cells were washed twice in 1× phosphate-buffered saline (PBS). Localization was then visualized with fluorescence microscopy, and accumulation was quantified on a 96-well fluorescence microplate reader (Varioskan LUX; Thermo Scientific).

Fungicidal activity of KM29 against S. cerevisiae and C. albicans.

The fungicidal activity of KM29 was probed using a microdilution plate assay (killing assay) as described previously (60) with the following modifications. Single colonies of S. cerevisiae were resuspended in 1 ml of 5 mM NaP buffer (pH 7.4). Cell counts were then calculated using a hemocytometer. Afterwards, cells were diluted in the same buffer to reach a density of 1.5 × 10⁵/ml. Cell suspensions were made by mixing 20 μl cells and 20 μl of the peptide dissolved in 5 mM NaP buffer. The final peptide concentrations used were 1, 2.5, and 5 μM. These suspensions were incubated in a shaking incubator at 220 rpm for 2 h at 30°C. Peptide reactions were stopped by adding 360 μl of yeast nitrogen base (YNB) (0.34% yeast nitrogen base without amino acids and ammonium sulfate, 1% ammonium sulfate [wt/vol]). A 1:10 dilution of the cell suspension was spread onto YPD plates and incubated for 48 h at 30°C. CFU were counted, and the percentage of survival was calculated as (colonies from suspension with peptide)/(colonies from suspension without peptide) × 100.

This protocol was modified slightly to determine the fungicidal activities of KM29 against C. albicans LLF100 and LLF088. Initially, C. albicans was grown for 24 h on YPD at 37°C. Due to the increased sensitivity of C. albicans to KM29, 10 mM NaP buffer and final peptide concentrations of 0.5, 1, and 2.5 μM were used. These changes allowed us to maintain a survival curve that displayed approximately 75, 50, and 25% survival, respectively. Plates were incubated for 24 h at 37°C.

When respiration-deficient mutants were analyzed using microdilution assays, all incubation times for plates were extended by 24 h to account for the slow growth of these mutants.

Measurement of ROS production and ROS-induced cell death.

The production of ROS was measured using 2′,7′-dichlorofluorescin diacetate (DCFDA; Sigma) (61), with the following modifications. S. cerevisiae or C. albicans cells from an overnight culture were pelleted and resuspended in 1× PBS to a final concentration of 1 × 10⁸/ml. DCFDA (Sigma) was added to cell suspensions to a final concentration of 10 μM. DCFDA is a fluorescent probe that, upon oxidation by ROS, transitions to the highly fluorescent molecule 2′,7′-dichlorofluorescein (DCF). After incubation for 1 h at 30°C in the dark, cells were washed twice and resuspended in 5 mM NaP buffer. A 125-μl volume of cells preloaded with DCFDA was added to wells of a black 96-well plate (Corning) that contained 125 μl of 5 mM NaP buffer. Individual wells also contained either KM29 alone or KM29 plus 10 mM N-acetyl cysteine (NAC; Sigma) as a ROS scavenger. Final concentrations of KM29 were 0, 333, and 833 μM. Reactions were maintained for 2 h at 30°C in a shaking fluorescence microplate reader (Varioskan LUX; Thermo Scientific), and the transition from DCFDA to DCF was measured every 5 min with an excitation wavelength (λ_ex) of 504 nm and an emission wavelength (λ_em) of 529 nm. Afterwards, 1 μl of cells was taken from each well and was added to 360 μl YNB plus 39 μl 5 mM NaP to stop the reactions. An aliquot of this solution was plated onto YPD and was incubated at 30°C. Cell survival was calculated from the CFU obtained.

Assessment of mitochondrial network morphology and mitochondrial membrane potential.

S. cerevisiae cells were transformed with pMitoLoc (29) using a standard yeast transformation protocol (62) with the addition of a 4-h recovery in YPD and were then plated on YPD containing 100 μg/ml of nourseothricin (clonNAT; Werner) for plasmid selection. One of the resulting transformants was used to inoculate 5 ml of liquid YPD plus clonNAT and was grown overnight. Cells were pelleted and were resuspended in 5 mM NaP buffer (pH 7.4) to a density of 1 × 10⁸/ml. Additionally, either carbonyl cyanide m-chlorophenylhydrazone (CCCP; Sigma) or KM29 was added to the cell suspensions to final concentrations of 15 μM and 166 μM, respectively. After incubation for 3 h at 30°C, cells were washed twice in 1× PBS and were fixed with 10% (vol/vol) formaldehyde (Mallinckrodt) for 30 min at room temperature. Afterwards, cells were washed once with 1× PBS, mounted on a polylysine slide with SlowFade Gold antifade reagent mounting medium (Molecular Probes), and viewed with a Leica SP5 confocal fluorescence microscope. Image stacks using z-spacing of 200 nm were acquired with Leica image analysis software. Image stacks were then analyzed using the ImageJ plugin yeast_correlation and the “whole area” mapping method (29).

Statistical analysis.

All experiments were conducted a minimum of three times from independent biological replicates. Two-sample t tests were used for testing the significance of differences between individual mutants and the wild type. One-way analysis of variance was used for multiple comparisons. Data were analyzed using GraphPad Prism.