Synergism of Antifungal Activity between Mitochondrial Respiration Inhibitors and Kojic Acid

Co-application of certain types of compounds to conventional antimicrobial drugs can enhance the efficacy of the drugs through a process termed chemosensitization. We show that kojic acid (KA), a natural pyrone, is a potent chemosensitizing agent of complex III inhibitors disrupting the mitochondrial respiratory chain in fungi. Addition of KA greatly lowered the minimum inhibitory concentrations of complex III inhibitors tested against certain filamentous fungi. Efficacy of KA synergism in decreasing order was pyraclostrobin > kresoxim-methyl > antimycin A. KA was also found to be a chemosensitizer of cells to hydrogen peroxide (H2O2), tested as a mimic of reactive oxygen species involved in host defense during infection, against several human fungal pathogens and Penicillium strains infecting crops. In comparison, KA-mediated chemosensitization to complex III inhibitors/H2O2 was undetectable in other types of fungi, including Aspergillus flavus, A. parasiticus, and P. griseofulvum, among others. Of note, KA was found to function as an antioxidant, but not as an antifungal chemosensitizer in yeasts. In summary, KA could serve as an antifungal chemosensitizer to complex III inhibitors or H2O2 against selected human pathogens or Penicillium species. KA-mediated chemosensitization to H2O2 seemed specific for filamentous fungi. Thus, results indicate strain- and/or drug-specificity exist during KA chemosensitization.


Introduction
The mitochondrial respiratory chain (MRC) can serve as a valuable molecular target for control of fungal pathogens (Figure 1a). Chemical inhibitors of MRC, such as antimycin A (AntA) or strobilurins (e.g., Pyraclostrobin (PCS), Kresoxim-methyl (Kre-Me), mucidin, etc.), interfere with cellular energy (e.g., ATP) production in fungi [1,2], weakening fungal viability. Coinciding with this interference is an abnormal leakage of electrons from MRC. The escaped electrons can cause oxidative damage to vital components in fungal cells, such as chromosomes, lipid membranes and proteins, resulting in apoptosis or necrosis [1,2] (see Figure 1b for scheme). The antioxidant system in fungi, e.g., glutaredoxins, cytosolic or mitochondrial superoxide dismutases (Cu, Zn-or Mn-SOD), glutathione reductase, plays a protective role in such cases, maintaining cellular homeostasis/integrity from toxic oxidative species [3,4]. Fungi can also overcome the toxicity of MRC inhibitors by expressing alternative oxidase (AOX) (Figure 1a), rendering the completion of electron flow via MRC [5,6]. AOX is insensitive to MRC inhibitors [5,6]. With respect to other targets of conventional antifungal drugs already identified (e.g., cell wall/membrane integrity pathway, cell division, signal transduction, and macromolecular synthesis, etc.) [8], MRC is a relatively unexploited target in human fungal pathogens. However, the MRC has been actively used as a drug target for control of malarial parasites, e.g., Plasmodium. For example, the antimalarial drug atovaquone disrupts the mitochondrial electron transport as well as the inner mitochondrial membrane potential (ΔΨ m ) in parasites [9]. Atovaquone is also used to treat fungal infections such as Pneumocystis jirovecii (pneumonia) [10].
Co-application of certain types of compounds with commercial antimicrobial drugs can increase the effectiveness of drugs through a mechanism termed "chemosensitization" [11][12][13][14]. For example, a prior study showed that the 4-methoxy-2,3,6-trimethylbenzensulfonyl-substituted D-octapeptide chemosensitized cells to the antifungal drug fluconazole (FLC), countering FLC resistance of clinical isolates of Candida pathogens, and of strains of the model yeast Saccharomyces cerevisiae overexpressing multidrug efflux pumps/drug transporter or a lanosterol 14α-demethylase (Erg11p, molecular target of FLC) [11]. Similarly, in bacterial pathogens, application of sub-inhibitory concentrations of squalamine enhanced the antibiotic susceptibility of various Gram-negative bacteria, in both antibiotic-resistant and susceptible strains [12]. Squalamine is thought to modify membrane integrity by increasing permeability of drugs [12].
Meanwhile, co-application of proguanil, which modulates mitochondria in protozoan parasites, resulted in an increased antimalarial activity of atovaquone [15]. Of note is that proguanil-based chemosensitization was specific for atovaquone, i.e., proguanil did not enhance the activities of other MRC inhibitors, such as myxothiazole or AntA [15]. Results indicate "drug-chemosensitizer specificity" exists in the process. Collectively, these studies showed that chemosensitization could ultimately lead to lowering dosages of conventional drugs necessary for effective control of pathogens. It would also lead to preventing development of pathogen resistance to conventional drugs [16].
Kojic acid (KA, Figure 2a) is a natural product of some filamentous fungi, mainly certain species of Aspergillus or Penicillium. KA is widely used as a depigmenting agent due to its ability to inhibit the activity of tyrosinase, a key enzyme responsible for melanogenesis in melanoma and melanocytes [17][18][19][20]. From a clinical perspective, KA can potentially inhibit pathogen infection since: (1) it enhances host immunity by stimulating phagocytosis, generating reactive oxygen species (ROS) in macrophages, and potentiating phytohemagglutinin-based proliferation of lymphocytes [21,22]; (2) KA or its structural derivatives directly exert antimicrobial activity against fungal/bacterial pathogens [23]. For instance, KA functions as an antifungal agent against Cryptococcus neoformans (cryptococcosis), where KA also inhibits melanin synthesis necessary for fungal infectivity [24].
We previously showed that KA could act as a chemosensitizing agent when co-applied with the polyene antifungal drug amphotericin B (AMB) or hydrogen peroxide (H 2 O 2 ) against various filamentous fungal or yeast pathogens [25]. The mechanism of antifungal chemosensitization appeared to be modulation of the function of the antioxidant system in the fungus. Noteworthy is that the degree/efficacy of KA-mediated antifungal chemosensitization was related to the kinds of fungal strain and/or drug examined [25]. This tendency is similar to the "drug-chemosensitizer specificity" found in atovaquone-mediated chemosensitization (see above).
In this study, we further investigated if KA, as a chemosensitizer, could improve the activities of complex III inhibitors of MRC (i.e., AntA, Kre-Me, PCS; see Figure 2b-d for structures and 2e for scheme), and thus, possess potential as an active pharmaceutical/agroceutical ingredient, against various filamentous fungi. We included a number of human and plant pathogens, as well as model fungal strains, in our tests (see Table 1; Figure 2e). We observed that human fungal pathogens, i.e., Aspergillus fumigatus, A. terreus, Acremonium sp., and Scedosporium sp., were the most sensitive strains to KA-mediated chemosensitization to complex III inhibitors.

Enhancing Antifungal Activity of H 2 O 2 or Complex III Inhibitors with KA against Aspergillus or Penicillium Strains: Agar Plate Bioassay
Hydrogen peroxide acts similarly to host-derived ROS, as a host defense response against infecting pathogens. For example, patients with chronic granulomatous disease (CGD) experience high susceptibility to invasive infections by Aspergillus [28]. The phagocytic immune cells of CGD patients cannot induce an oxidative burst because they lack NADPH oxidase, necessary to generate superoxides, the precursor to the antimicrobial ROS H 2 O 2 [28]. Although the infecting fungi rely on their cellular antioxidant system for protection from host ROS, application of KA further enhances host immunity by stimulating phagocytosis and generation of ROS in macrophages (see Introduction) [21,22].
We previously examined KA-mediated chemosensitization to H 2 O 2 and AMB [25]. Besides disrupting fungal plasma membranes, AMB also induces fungal oxidative damage [29][30][31][32] by stimulating ROS production [33]. Thus, we surmised that the effect of KA + AMB would be similar to KA + H 2 O 2 . However, unlike with KA + AMB, chemosensitization did not occur with KA + H 2 O 2 in any of the yeast pathogens tested. We concluded that the effectiveness of KA-mediated chemosensitization was fungal strain-and/or drug-specific [25].
Since complex III inhibitors, like AMB, also trigger cellular oxidative stress in fungi (see Introduction), we also compared the effect of KA + complex III inhibitors with that of KA + H 2 O 2 in this study.

Filamentous Fungi
Our initial agar bioassays were performed with the human pathogenic fungi. Co-application of KA  Table 2). For example, co-application of H 2 O 2 and KA at 5 mM, each, completely inhibited the growth of A. fumigatus AF10 (i.e., no visible germination on plates), whereas independent application of either H 2 O 2 or KA, alone, did not achieve this level of antifungal activity. A similar level of chemosensitization was also observed in other fungi tested, i.e., A. terreus, Acremonium, and Scedosporium, by KA + H 2 O 2 (Figure 3a; see Table 2 for summary). Next we found that KA-mediated chemosensitization could also be achieved with complex III inhibitors in most of the human pathogens tested (Figure 3b; Table 2 . Whereas, independent application of KA or PCS, alone, did not result in such a level of antifungal activity. Levels of enhancement of antifungal activity also depended upon types of complex III inhibitors co-applied. PCS exerted the highest activity, followed by Kre-Me and AntA. Similar trends were also observed in other pathogens, such as A. terreus, Acremonium and Scedosporium (Figure 3a; see Table 2 for summary). The only exceptions were A. terreus UAB698 (no enhancement of sensitivity by KA + any of the complex III inhibitors) and A. terreus UAB673/680 (no enhancement of sensitivity by KA + AntA), respectively. Therefore, sensitivity of fungal strains to KA-mediated chemosensitization with complex III inhibitors ranged, from highest to lowest, as follows: Acremonium, Scedosporium > A. fumigatus > A. terreus. Of note is that, although human pathogens were also sensitive to KA + H 2 O 2 , levels/degrees of their sensitivity were generally not parallel to that of KA + complex III inhibitors (see Table 2).
-+ + a +, Enhancement of antifungal activity after co-application (reduced radial growth of fungi); ++, Enhancement of antifungal activity after co-application (no germination of fungi); -, No enhancement of antifungal activity after co-application. b [25]. c n/t, Not tested due to no growth of fungi w/ PCS (25 μM) alone (i.e., hypersensitivity to PCS alone). d n/t, Not tested due to no growth of fungi w/ Kre-Me (25 μM) alone (i.e., hypersensitivity to Kre-Me alone).
Agar bioassays performed on Penicillium strains, mostly plant pathogens, showed that co-application of KA with H 2 O 2 resulted in enhancement of antifungal activities of both compounds (KA and H 2 O 2 ), except P. griseofulvum 2300, P. italicum and P. glabrum, which were insensitive to this chemosensitization (Table 2; Figure data not shown).
KA-mediated chemosensitization was also performed using the complex III inhibitors on the Penicillium strains. Unlike the human pathogens tested, chemosensitization was more limited with the Penicillium strains, being effective only in strains, P. expansum FR2 and FR3 (both being fludioxonil (FLUD) resistant strains), P. digitatum, P. italicum and P. glabrum with KA + PCS ( Table 2; PCS was the most effective complex III inhibitor in this test). Levels of strain sensitivity in decreasing order with KA + PCS were: P. digitatum > P. italicum, P. expansum FR2 > P. glabrum, P. expansum FR3. P. digitatum, P. italicum, and P. glabrum were also sensitive to KA + Kre-Me or AntA. However, Penicillium strains were generally not as sensitive to KA-mediated chemosensitization with complex III inhibitors as human pathogens. As observed in human pathogens, levels/degrees of fungal sensitivity to KA + H 2 O 2 were not parallel to that of KA + complex III inhibitors (see Table 2).
Agar bioassays were performed on six other strains of Aspergillus, mainly plant pathogens or model strains (A. flavus: pathogenic to both plants and humans). These assays showed that co-application of KA with H 2 O 2 or complex III inhibitors resulted in no enhancement of antifungal activity of any compound tested (KA, H 2 O 2 or complex III inhibitors), except A. nidulans, which showed sensitivity to KA + PCS or Kre-Me ( In our previous study, KA-mediated chemosensitization with H 2 O 2 was not effective in any of the yeast pathogens tested [25]. Therefore, in the present study, we attempted to examine how the treatment of KA + H 2 O 2 was related to various functions of antioxidant system of yeasts using S. cerevisiae as a model. For this study, we used a yeast dilution bioassay (see Experimental Section) and tested a wild type and four antioxidant mutant (gene knock-out) strains of S. cerevisiae as follows: (1) yap1 (Yap1p, a transcription factor, regulates the expression of four downstream genes within the antioxidant system, i.e., GLR1 (glutathione reductase), YCF1 (a glutathione S-conjugate pump), TRX2 (thioredoxin), and GSH1 (γ-glutamylcysteine synthetase [34,35]); (2) sod1 (Cu,Zn-SOD); (3) sod2 (Mn-SOD); and (4) glr1 (Glr1p, glutathione reductase; see Saccharomyces Genome Database [27]). These representative mutants play key roles in maintaining cellular redox homeostasis in both enzymatic (e.g., ROS radical-scavenging) and non-enzymatic (e.g., glutathione homeostasis) aspects.
Worth noting is that S. cerevisiae has also been developed as a model system for studying atovaquone resistance [36].
To our surprise, in these yeast strains, KA mainly acted as an antioxidant, but not as an antifungal chemosensitizer ( Figure 4). For example, when wild type or mutants were treated with 1 mM of H 2 O 2 alone, all yeast strains showed sensitive responses, as reflected in no growth of cells at 10 −2 to 10 −5 dilution spots ( Figure 4). As expected, yap1, which regulates the expression of four downstream genes in the antioxidant system, was more sensitive to H 2 O 2 (i.e., no growth at 10 −1 dilution spot) than any other yeast strains. However, as shown in Figure 4, co-application of KA with H 2 O 2 ameliorated the H 2 O 2 -triggered oxidative stress, resulting in enhancement of the growth of all yeasts tested. For example, the wild type showed growth recovery up to 100,000-fold dilution (the 10 −5 dilution spot), revealing this strain fully recovered from oxidative stress induced by H 2 O 2 when KA was co-applied. Additionally, the sod1, sod2 and glr1 mutants grew up to 10 −3 to 10 −4 dilution spots and yap1 grew up to 10 −1 dilution spot when H 2 O 2 was co-applied with 5 mM KA. The antioxidant capacity of KA was also commensurate with KA concentrations. Although yeast strains showed increased sensitivity to 2 mM of H 2 O 2 , similar trends of antioxidation activity by KA were also observed ( Figure 4). Thus, overall, the results indicate KA has a different effect, depending on types of fungi examined. That is, KA functions as an antioxidant in S. cerevisiae, while it acts as an antifungal chemosensitizer in certain of the filamentous fungi tested. KA may induce different transcriptional programs in S. cerevisiae than in filamentous fungi. Further studies, such as genome-wide gene expression profiling, are warranted to determine the precise mechanism of antioxidation in and/or insensitivity of yeast to KA + H 2 O 2 chemosensitization.

Co-Application of KA and PCS
Synergistic Fractional Inhibitory Concentration Indices (FICIs; see Experimental Section for calculations) were found between KA and PCS for most human pathogens (A. fumigatus, A. terreus, Acremonium sp., Scedosporium sp.) and A. nidulans (Table 3). Despite the absence of calculated "synergism", as determined by "indifferent" interactions [38] (Table 3), there was enhanced antifungal activity of KA and PCS (i.e., chemosensitization) in Acremonium, which was reflected in lowered Minimum Inhibitory Concentrations (MICs) of each compound when combined. However, synergistic Fractional Fungicidal Concentration Indices (FFCIs) (at the level of ≥ 99.9% fungal death) between KA and PCS occurred only in Acremonium (Table 3), indicating the KA-mediated chemosensitization with PCS is fungistatic, not fungicidal, in most strains tested.  Synergistic FICIs between KA and PCS also occurred in four Penicillium strains (Table 3). Despite the absence of calculated "synergism" [38] (Table 3), there was enhanced antifungal activity of KA and PCS (i.e., chemosensitization) also in P. expansum FR3 (FLUD resistant strain), which was reflected in lowered MICs of each compound when combined. However, synergistic FFCI (at the level of ≥99.9% fungal death) between KA and PCS was not achieved in any of the Penicillium strains examined (Table 3), indicating that, as in the human pathogens/A. nidulans (see above), the KA-mediated chemosensitization with PCS is mostly fungistatic, not fungicidal, in Penicillium strains (Lowered Minimum Fungicidal Concentrations (MFCs), although not "synergistic" level, were observed in P. glabrum and P. italicum at the level of ≥99.8% fungal death; see Table 3).

Strains
Hypersensitive to Complex III Inhibitors: Testing Acremonium, Scedosporium, P. digitatum with Kre-Me KA + Kre-Me was also examined in Acremonium, Scedosporium and P. digitatum, which were the most sensitive strains to complex III inhibitors (see Table 2). We tried to determine the level of sensitivity of these strains to Kre-Me, which is less potent than PCS (see Figure 3 and Table 2). Consistently, synergistic FICIs between KA and Kre-Me occurred in all strains tested (Table 4). However, synergistic FFCIs (at the level of ≥99.9% fungal death) between KA and Kre-Me were not achieved in any of the strains examined (Table 4), while lowered MFCs for both KA and Kre-Me were observed in Acremonium (FFCI = 0.6). Acremonium sp. is the only strain with low FFCI values for both PCS and Kre-Me, i.e., 0.5 PCS and 0.6 Kre-Me , respectively (see Tables 3 and 4). Results further confirmed the sensitive responses of Acremonium, Scedosporium and P. digitatum to complex III inhibitors (both PCS and Kre-Me).  The results of all CLSI-based checkerboard (chemosensitization) tests (i.e., KA + PCS or Kre-Me in filamentous fungi) are summarized in Table 5. As shown in the Table, the FICIs for thirteen strains (out of fifteen strains) w/PCS and for three fungi (the most sensitive strains to complex III inhibitors) w/Kre-Me were synergistic. Whereas, FFCI for only Acremonium sp. was synergistic, indicating the KA-mediated chemosensitization with complex III inhibitors exerted mostly fungistatic (but not fungicidal) effects.

Chemicals
Antifungal chemosensitizing agent [kojic acid (KA)], antifungal drugs [antimycin A (AntA), kresoxim methyl (Kre-Me), pyraclostrobin (PCS)] and oxidizing agent [hydrogen peroxide (H 2 O 2 )] were procured from Sigma Co. Each compound was dissolved in dimethyl sulfoxide (DMSO; absolute DMSO amount: <1% in media), except H 2 O 2 , which was dissolved in water, before incorporation into culture media. In all tests, control plates (i.e., "No treatment") contained DMSO at levels equivalent to that of cohorts receiving antifungal agents, within the same set of experiments.

Agar Plate Bioassay: Filamentous Fungi
In the agar plate bioassay, measurement of sensitivities of filamentous fungi to the antifungal agents was based on percent (%) radial growth of treated compared to control ("No treatment") fungal colonies (see text for test concentrations.) [40]. Minimum inhibitory concentration (MIC) values on agar plates were determined based on triplicate bioassays, and defined as the lowest concentration of agents where no fungal growth was visible on the plate. For the above assays, fungal conidia (5 × 10 4 CFU/mL) were diluted in phosphate-buffered saline (PBS) and applied as a drop onto the center of PDA plates with or without antifungal compounds. Growth was observed for three to seven days to determine cellular sensitivities to compounds. . Interactions were defined as: "synergistic" (FICI or FFCI ≤ 0.5) or "indifferent" (FICI or FFCI > 0.5-4) [38]. Statistical analysis was based on [39].

Agar Plate Bioassay: S. cerevisiae
Petri plate-based yeast dilution bioassays were performed on the wild type and antioxidant mutants (yap1Δ, sod1Δ, sod2Δ, glr1Δ) to assess effects of KA + H 2 O 2 on the antioxidant system. Yeast strains were exposed to 1 to 5 mM of KA, w/o or w/H 2 O 2 (1 or 2 mM) on SG for 5 to 7 days. These assays were performed in duplicate on SG agar following previously described protocols [41].

Conclusions
In this study, KA enhanced antifungal activities of MRC inhibitor(s) or H 2 O 2 as follows: (1) Most human pathogens tested (i.e., A. fumigatus, A. terreus, Acremonium sp., Scedosporium sp.) were sensitive to both KA + complex III inhibitors and KA + H 2 O 2 , except A. terreus UAB698 (no chemosensitization w/all complex III inhibitors tested) and A. terreus UAB673/680 (no chemosensitization w/AntA); (2) Most of the plant pathogenic Penicillium species were sensitive to KA + H 2 O 2 , except P. griseofulvum 2300, P. italicum and P. glabrum (no chemosensitization); (3) Some Penicillium species (i.e., P. digitatum, P. italicum, P. glabrum, and FLUD-resistant P. expansum FR2/FR3) were sensitive to KA + at least one of the complex III inhibitors. However, all other Penicillium species were insensitive to KA + complex III inhibitors (no chemosensitization); (4) All other Aspergillus species (i.e., A. flavus, A. parasiticus, A. oryzae, A. niger, A. ochraceous, A. nidulans) were insensitive to KA + complex III inhibitors and/or KA + H 2 O 2 (no chemosensitization), except A. nidulans, which was sensitive to KA + PCS or Kre-Me (chemosensitization). Further studies are required to determine the mechanism(s) governing the variability of these Aspergillus strains to KA-mediated chemosensitization; (5) Most compound interactions at MIC level (i.e., FICI) between KA and PCS or Kre-Me, determined by CLSI method, resulted in synergism, except Acremonium sp. (KA + PCS) and P. expansum FR3 (KA + Kre-Me), which resulted in a certain level of positive interaction between compounds, but not synergism; (6) The antifungal chemosensitizing capacity of KA appears to be fungal strain-specific (i.e., specific for certain human pathogens or Penicillium species only) as well as fungal isolate-dependent (i.e., A. terreus). KA mainly functions as an antioxidant in yeasts; and (7) Strain sensitivity to KA + complex III inhibitors or H 2 O 2 varied as follows (in decreasing order): human fungal pathogens > Penicillium species > all other Aspergillus species.
In conclusion, KA, which is a relatively safe, natural compound to humans [42], shows some potential to serve as an antifungal chemosensitizing agent in combination with complex III inhibitors. This potential appears to be greatest with those filamentous fungi tested that are mainly pathogenic to humans. Chemosensitization can lower dosage levels of antifungal drugs necessary for effective control of fungi. Thus, use of safe chemosensitizing agents that selectively debilitate the fungal pathogen may be a viable approach to circumvent potential side-effects commonly associated with antimycotic therapy.