Unexpected effects of azole transporter inhibitors on antifungal susceptibility in Candida glabrata and other pathogenic Candida species

The pathogenic fungus Candida glabrata is often resistant to azole antifungal agents. Drug efflux through azole transporters, such as Cdr1 and Cdr2, is a key mechanism of azole resistance and these genes are under the control of the transcription factor Pdr1. Recently, the monoamine oxidase A (MAO-A) inhibitor clorgyline was shown to inhibit the azole efflux pumps, leading to increased azole susceptibility in C. glabrata. In the present study, we have evaluated the effects of clorgyline on susceptibility of C. glabrata to not only azoles, but also to micafungin and amphotericin B, using wild-type and several mutant strains. The addition of clorgyline to the culture media increased fluconazole susceptibility of a C. glabrata wild-type strain, whereas micafungin and amphotericin B susceptibilities were markedly decreased. These phenomena were also observed in other medically important Candida species, including Candida albicans, Candida parapsilosis, Candida tropicalis, and Candida krusei. Expression levels of CDR1, CDR2 and PDR1 mRNAs and an amount of Cdr1 protein in the C. glabrata wild-type strain were highly increased in response to the treatment with clorgyline. However, loss of Cdr1, Cdr2, Pdr1, and a putative clorgyline target (Fms1), which is an ortholog of human MAO-A, or overexpression of CDR1 did not affect the decreased susceptibility to micafungin and amphotericin B in the presence of clorgyline. The presence of other azole efflux pump inhibitors including milbemycin A4 oxime and carbonyl cyanide 3-chlorophenylhydrazone also decreased micafungin susceptibility in C. glabrata wild-type, Δcdr1, Δcdr2, and Δpdr1 strains. These findings suggest that azole efflux pump inhibitors increase azole susceptibility but concurrently induce decreased susceptibility to other classes of antifungals independent of azole transporter functions.


Introduction
The pathogenic fungus Candida glabrata is the second most common cause of candidemia and is relatively resistant to azole antifungal agents [1]. A key mechanism of azole-resistance is the reduction in the intracellular drug concentration, which is achieved by activation of azole transporters, such as Cdr1 and Cdr2 (formerly denoted Pdh1) [2][3][4]. Inhibition of these transporters is one effective way for combating azole-resistance. A recent study discovered that the monoamine oxidase A (MAO-A) inhibitor clorgyline inhibits the activity of azole efflux pumps, such as Cdr1, Cdr2, and Mdr1, leading to increased azole susceptibility in C. glabrata and Candida albicans [5]. Clorgyline suppresses the oxidation of human MAO-A and has been used as an antidepressant drug [6], and may be useful for the treatment of heart failure [7] and prostate cancer [8], suggesting that clorgyline might be a candidate for combination with azole antifungals.
Antifungals of the echinocandin class or polyene class have been suggested to be effective for treatment of candidiasis caused by C. glabrata [9]. However, recent epidemiological surveys have revealed the emergence of increasing numbers of isolates that have decreased susceptibility to echinocandins, particularly among fluconazole resistant isolates [10][11][12]. Furthermore, it has also been reported that polyene susceptibility is attenuated in azole-resistant isolates [11].
In the present study, we have evaluated the effects of clorgyline on susceptibility of C. glabrata to not only azoles, but also to echinocandin and polyene antifungals, using wild-type and several mutant strains, including Δcdr1 and Δcdr2 strains.

Strains and culture conditions
The C. glabrata, C. albicans, C. parapsilosis, C. tropicalis, and C. krusei strains used in this study are listed in Table 1. Cells were grown at 30˚C in YPD medium (1% yeast extract, 2% peptone, and 2% dextrose; Difco Laboratories, Detroit, MI), minimal medium (MIN) (0.67% yeast nitrogen base without amino acids [Difco Laboratories], 2% dextrose), synthetic complete medium (SC), or SC lacking tryptophan (SC-trp) [13]. SC5314 Candida albicans wild type [42] Plasmid and strain construction The primers used in this study are listed in Table 2. Sequence information of C. glabrata genes was obtained from the Candida genome database (http://www.candidagenome.org). To construct an overexpression plasmid for CDR1, a 4.5-kb BamHI-SalI PCR fragment containing C. glabrata CDR1 open reading frame was inserted into the BamHI-SalI site of pCgACT-P [14] and expressed under the Saccharomyces cerevisiae PGK1 promoter (pCgACT-PC1). C. glabrata deletion strains were constructed using the one-step PCR-based technique, described previously [15,16]. Briefly, a deletion construct was amplified from either pBSK-HIS (pBluescript II SK+ [Stratagene, La Jolla, CA] containing C. glabrata HIS3 at the XhoI site) or pBSK-TRP (pBluescript II SK+ containing C. glabrata TRP1 at the XhoI site) using primers tagged with 100-bp sequences homologous to the flanking regions of the target open reading frame. C. glabrata parent strains were subsequently transformed with the deletion construct, and the resulting transformants were selected by tryptophan or histidine prototrophy, as appropriate [15]. Successful homologous recombination was verified by diagnostic PCR and the absence of mRNA expression of the target genes was also confirmed by real-time quantitative reverse transcription PCR (real-time qRT-PCR) (data not shown). Transformation of C. glabrata was performed using the lithium acetate protocol, as described previously [17]. CDR1-overexpressed and its control strains were constructed by transformation of 2001T with pCgACT-PC1 and pCgACT-P, respectively, selected by tryptophan prototrophy, and verified by immunoblotting.

Susceptibility test
To examine the dose-dependence of the effects of clorgyline on fluconazole susceptibility, a checkerboard assay was performed in triplicate, in three independent experiments. MIN was used instead of RPMI, because RPMI broth mixed with a high concentration of clorgyline became turbid. The effects of clorgyline on micafungin and amphotericin B susceptibility were also examined by checkerboard assays using SC broth buffered to pH7.0. Two-fold dilutions of each compound were aliquoted into a 96-well plate. Each well was then inoculated with cell suspension at the final concentration of 5 × 10 2 cells/well. Plates were incubated at 35˚C for 48 h. Cell growth was monitored using an absorption spectrometer at 620 nm. Minimum drug concentration that inhibits growth by more than 80% relative to drug free control was defined as minimum inhibitory concentration (MIC). The fractional inhibitory concentration (FIC) was calculated by the following formula: FIC for drug A = MIC of drug A in combination with drug B/MIC of drug A alone. The sum of FIC for drug A and FIC for drug B was defined as FIC index (FICI). Drug interaction was classified as synergistic if its FICI was 0.5 or less than 0.5, and was classified as antagonistic if its FICI was greater than 4, as described previously [18].
A spot dilution test was carried out as described previously [14]. Briefly, logarithmic-phase cells grown in MIN broth or SC broth were harvested and adjusted to the concentration of 2 × 10 7 cells/ml. Serial 10-fold dilutions were then prepared, and 5 μl of each dilution was Table 2. Primers used in this study.
For gene deletion Target gene Sequence (5'-3') (sequences homologous to flanking regions of the target gene are shown in italics. Sequences shown in bold exist in pBSK-HIS, or pBSK-TRP)

Immunoblotting
Anti-Cdr1 and Cdr2 antibodies were kindly provided by Dr. Dominique Sanglard. Anti-Pgk1 (OriGene EU, Herford, Germany) and rabbit IgG-hrp (GE Healthcare, Pittsburgh, PA) antibodies were purchased. Liquid-cultured C. glabrata strains at logarithmic-phase were harvested and lysed using Minute Total Protein Extraction Kit for Microbes with Thick Cell Walls (Invent Biotechnologies, Plymouth, MN) according to its instruction. Lysates were separated by SDS-PAGE, and transferred to a polyvinylidene difluoride membrane (BIO-RAD, Hercules, CA). Each protein was detected using the indicated antibodies, an enhanced chemiluminescent substrate (Thermo Fisher Scientific, Waltham, MA) and ChemiDoc Touch imaging system (BIO-RAD).

Real-time qRT-PCR
To examine expression levels of the genes associated with azole efflux, as well as FMS1, logarithmic-phase cells grown in YPD broth were adjusted to 2 × 10 7 cells/ml and then further incubated in the presence of 80 μg/ml of clorgyline for up to 4 h. Total RNA was extracted at the indicated time points, using a FastRNA Pro Red Kit (Qbiogene, Carlsbad, CA). Real-time qRT-PCR was then performed as described previously [19]. Briefly, first strand cDNA was synthesized with a QuantiTect Reverse Transcription kit (Qiagen, Valencia, CA), using 1 μg of total RNA as template, in a final volume of 20 μl, and 3 μl of the resulting cDNA was then used as the template for individual PCRs. These reactions were performed using gene-specific primers (Table 2) and a QuantiTect SYBR Green PCR kit (Qiagen). Real-time qRT-PCR was performed using a 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA). Extraction of RNA and real-time qRT-PCR were performed in triplicate, in three independent experiments.

Results and discussion
Clorgyline improves fluconazole-susceptibility in C. glabrata In a previous study, the chemosensitizing effect of clorgyline to fluconazole in C. glabrata was demonstrated by a disk diffusion assay, and the clorgyline MIC in C. glabrata was reported to exceed 40 μg/ml [5]. To test the effect of clorgyline at a concentration higher than 40 μg/ml on growth in C. glabrata, and to test the effect of clorgyline on the fluconazole MIC, we performed a checkerboard assay using serial 2-fold dilutions of these two compounds. At 48 h, clorgyline did not inhibit the growth of C. glabrata at a concentration of up to 160 μg/ml, but at 320 μg/ ml, clorgyline inhibited the growth slightly; thus, the MIC of clorgyline may exceed 320 μg/ml. The MIC of fluconazole for C. glabrata was 256 μg/ml, whereas this decreased to 64 μg/ml in the presence of 160 μg/ml clorgyline (Fig 1A). On the other hand, the MIC of clorgyline for C. glabrata decreased to 80 μg/ml in the presence of 128μg/ml fluconazole. According to these data, the FICI of these two drugs are less than 0.5, indicating synergism. This effect of clorgyline on fluconazole susceptibility was also supported by spot dilution test (Fig 1B).
The lowest concentration of clorgyline that elicits a chemosensitizing effect on fluconazole was 80 μg/ml. This concentration was sufficient to inhibit azole transporters in the C. glabrata wild-type strain. We therefore used 80 μg/ml clorgyline in the subsequent experiments, although further evaluation may be needed to detect a more appropriate dose in other Candida species.

Micafungin-and amphotericin-B-susceptibility is markedly decreased in the presence of clorgyline in Candida strains
As reported previously, echinocandin and amphotericin B MIC levels for fluconazole-resistant Candida isolates tend to be higher than those for fluconazole-susceptible isolates [11,12]. In some reports, it was suggested that overexpression of an azole transporter, particularly Cdr2p, may slightly decrease echinocandin susceptibility in C. albicans [20][21][22]. Therefore, we examined whether the azole transporter inhibitor clorgyline affects susceptibility to micafungin and amphotericin B in main Candida species including C. glabrata, C. albicans, C. parapsilosis, C. tropicalis, and C. krusei, which account for about 95% of clinical isolates of Candida species [23]. Addition of 80 μg/ml of clorgyline to a MIN plate did not inhibit any of the strains tested in this study. Contrary to our expectations, clorgyline markedly decreased micafungin-and amphotericin B-susceptibility (Fig 2). In some strains, antagonistic effects of clorgyline with micafungin (Table 3) and amphotericin B (Table 4) were observed as high FICI that exceeded 4 in checkerboard assays, while the other strains also showed slightly decreased susceptibility to micafungin or amphotericin B in the presence of clorgyline. The attenuation of micafungin and amphotericin B activity by this agent has never been reported. If this phenomenon is mediated via inhibition of an azole efflux pump, the strategy of chemosensitization by other efflux pump inhibitors may lead to an adverse effect when treatment is switched from azoles to other antifungals, such as echinocandins and polyenes.
The effect of clorgyline on micafungin-and amphotericin B-susceptibility may not be caused directly by azole transporter inhibition in C. glabrata A previous report has shown that deletion of genes encoding azole efflux pumps, such as CDR1, CDR2, and MDR1, does not alter caspofungin susceptibility in C. albicans [24]. However overexpression of these pumps slightly decreases echinocandin susceptibility [21,22]. It is still unclear whether the azole transporters Cdr1 and Cdr2 are involved in the efflux of echinocandins. It has previously been shown that clorgyline inhibits rhodamine 6G efflux in a recombinant Saccharomyces cerevisiae strain expressing C. albicans CDR1 and CDR2 [5]. Although rhodamine 6G is a known substrate of Cdr1 and Cdr2 in both C. albicans and C. glabrata [25], the precise mechanism by which clorgyline inhibits these azole transporters is not clear. To investigate the clorgyline-mediated inhibition of azole transporters at the mRNA expression level, we performed real-time qRT-PCR. As shown in Fig 3A, clorgyline did not inhibit mRNA expression of C. glabrata CDR1, CDR2, or PDR1, but markedly elevated the expression of these genes.
It may be possible that clorgyline alters the substrate specificity of azole transporters, and that activation of Cdr1 or Cdr2, or Pdr1 are involved in transport of echinocandins and polyenes. To investigate this hypothesis, we constructed C. glabrata Δcdr1, Δcdr2, and Δpdr1 strains and compared the micafungin-and amphotericin B-susceptibility of these strains in the  presence and absence of clorgyline. The Δcdr1 and Δpdr1 strains showed increased fluconazole susceptibility, but the susceptibility of Δcdr2 strain was equal to the wild-type strain (data not shown), in accordance with previous reports [26,27]. As shown in Fig 3B, deletion of C. glabrata CDR1, CDR2, and PDR1 did not change the clorgyline effect on micafungin-and amphotericin B-susceptibility. We also performed western blot analysis to investigate the clorgyline-mediated effect on Cdr1 and Cdr2 protein expression. The treatment with clorgyline increased not only CDR1 mRNA expression (Fig 3A) but also the amount of Cdr1 protein (Fig 3C) in C. glabrata. On the other hand, the amount of Cdr2 protein was decreased in the presence of clorgyline for unknown reasons (Fig 3C). In C. albicans, Cdr1 plays a role in maintaining membrane asymmetry and alterations in membrane lipid composition affects Cdr1 functions [28,29]. In addition, altered sphingolipid biosynthesis resulting in the accumulation of certain long-chain bases affects echinocandin susceptibility in C. albicans and C. glabrata [30,31]. Therefore, a possible hypothesis is that decreased susceptibility to micafungin and amphotericin B in the presence of clorgyline may be due to downstream effects of Cdr1 overproduction on membrane composition. To evaluate the effect of overproduction of Cdr1 protein on micafungin and amphotericin B susceptibility, we constructed a CDR1-overexpressing strain in C. glabrata and an increase in the amount of Cdr1 protein was confirmed by western blot analysis ( Fig  3D). As shown in Fig 3E, overexpression of CDR1 did not affect the micafungin-and amphotericin B-susceptibility.
These results suggest that clorgyline decreases the micafungin-and amphotericin B-susceptibility independent of the functions of Cdr1, Cdr2, and Pdr1.
Function of C. glabrata Fms1, a putative ortholog of human MAO-A, does not affect the decreased susceptibility to micafungin and amphotericin B in the presence of clorgyline The molecular target of clorgyline in humans is MAO-A, and its homolog in C. albicans is CBP1, encoding a corticosteroid-binding protein [5,32]. It has previously been suggested that clorgyline may act on Cbp1 and affect plasma membrane proteins, such as efflux pumps [5]. S. cerevisiae Fms1 is a homolog of C. albicans Cbp1, and is involved in sterol synthesis [33]. The pairwise alignment of deduced amino acid sequences of the C. glabrata FMS1 ortholog (CAGL0M07612g) and S. cerevisiae FMS1 exhibited 42% identity and 57% similarity (ClustalW alignment with MacVector software version 15.1). Real-time qRT-PCR showed that addition of clorgyline increased mRNA levels of FMS1 in the C. glabrata wild-type strain (Fig 4A). Next, we generated a strain lacking C. glabrata FMS1 and evaluated its susceptibility to micafungin and amphotericin B in the presence and absence of clorgyline. As shown in Fig 4B, the effect of clorgyline on micafungin-and amphotericin B-susceptibility was not attenuated in   Effects of azole transporter inhibitors on antifungal susceptibility in Candida the Δfms1 strain, compared to its parental strain, suggesting a possibility that clorgyline may have other target molecules in C. glabrata.
Azole transporter inhibitors other than clorgyline also decrease micafungin susceptibility of C. glabrata The above findings indicated that the effect of clorgyline on micafungin susceptibility may not be caused by efflux of micafungin by an azole transporter. It is possible that clorgyline may directly interfere with the activity of micafungin, without a cellular response. To investigate this, we employed an efflux pump inhibitor other than clorgyline, such as milbemycin [25,34] and CCCP [35][36][37][38], and tested their effects on micafungin susceptibility in C. glabrata. Neither of these agents affected the control growth of any of the strains. Similarly to clorgyline, these two agents also decreased micafungin susceptibility, regardless of the presence or absence of CDR1, CDR2, and PDR1 ( Fig 5). As all three azole transporter inhibitors attenuated the micafungin susceptibility, it seems unlikely that all three these compounds may similarly interact with micafungin. We cannot completely discount the involvement of azole transporters in micafungin susceptibility, because we have not yet investigated phenotype of a Δcdr1 Δcdr2 double deletion strain. Furthermore, the function of upstream regulators of azole transporters remains to be investigated.

Conclusion
In this study, we observed unexpected effects of clorgyline on antifungal susceptibility in the 5 most common species of Candida. Fluconazole susceptibility was increased but susceptibilities to micafungin and amphotericin B were decreased in the presence of azole transporter inhibitors, including clorgyline, CCCP, and milbemycin A4 oxime, in C. glabrata, independently of functions of azole efflux pumps. Although the details of the underlying molecular mechanism and clinical relevance remain to be studied, our results may provide a clue to the discovery of a novel mechanism of micafungin-and amphotericin B-resistance in pathogenic fungi.