Synergistic and antagonistic effects of immunomodulatory drugs on the action of antifungals against Candida glabrata and Saccharomyces cerevisiae

Strains

Five Saccharomyces cerevisiae, six Candida glabrata clinical isolates and one S. cerevisiae non-clinical isolate were selected from the Collection of Industrial Microorganisms (ZIM) at the Biotechnical Faculty, Slovenia (Table 1). These were selected from 96 clinical isolates (40 S. cerevisiae and 56 C. glabrata; full list of strains is in Supplemental Information 5) and nine non-clinical S. cerevisiae isolates based on their minimal inhibitory concentrations (MICs) obtained by the reference method for broth dilution antifungal susceptibility testing of yeasts (CLSI M27-A3) (CLSI, 2008). The criterion was to select strains with different MICs to cover several spectra of antifungal resistance. S. cerevisiae strain Sc6 from sorghum beer was selected because it displayed high tolerance (16 mg/l) towards fluconazole for a non-clinical isolate. Candida parapsilosis ATCC 22019 and Candida krusei ATCC 6258 were used as the control strains in the drug susceptibility and checkerboard assays. In addition, to explore the antagonistic mechanism of MPA and azole antifungals, we used C. glabrata ATCC 2001 and Cgpdr1Δ (CAGL0K10780g Δ::NAT1) mutant, provided by Karl Kuchler’s laboratory, Medical University of Vienna, from their deletion library (Schwarzmüller et al., 2014). Cgpdr1Δ is isogenic to C. glabrata ATCC 2001.

Media

Strains were preserved in storage media (10% glycerol, 1% NaCl, and 1% Tween 20) at −80 °C. They were revitalized and routinely grown on yeast peptone dextrose (YPD) agar plates (2% Bacto Peptone, 1% yeast extract, 2% dextrose, and 2% Bacto agar) at 35 °C, regularly sub-cultured before each experiment. Throughout the assays we also used YPD broth (2% Bacto Peptone, 1% yeast extract, 2% dextrose), Sabouraud dextrose agar (3% Sabouraud dextrose from Sigma-Aldrich, 2% Bacto agar), and RPMI (1.04% RPMI-1640 from Sigma-Aldrich, 3.453% morpholinepropanesulfonic acid from Sigma-Aldrich, pH adjusted to pH 7 with 10 M NaOH solution) (Adams et al., 1998).

Drugs

Immunomodulatory drugs used for the screening were methotrexate (MTX), mycophenolic acid (MPA) and its derivate mycophenolate mofetil, cyclosporine A (CsA), and tacrolimus (Fk506). We also included a β-lactam antibiotic amoxicillin trihydrate (AMX) because it is often administered simultaneously with immunomodulatory agents. Antifungal drugs used for the initial screenings were amphotericin B (AMB), itraconazole (ITC), and fluconazole (FLC). For further exploration of the antagonistic mechanism between MPA and azoles, we used posaconazole (POS), ketoconazole (KCT), and voriconazole (VRC). All of the drugs were obtained from Sigma-Aldrich, except Fk506, which was provided by Acies Bio.

Stock solutions of MPA, MTX, CsA, Fk506, AMB, ITC, POS, KCT, VRC, and CLO were prepared in dimethyl sulfoxide (DMSO; Sigma-Aldrich, St. Louis, MO, USA), whereas AMX and FLC were diluted directly in the medium of choice for the assay. All the final drug concentrations were made in media (RPMI for drug susceptibility and checkerboard assay, YPD for further evaluation of the antagonistic interaction). List of stock solutions is in Supplemental Information 2.

Checkerboard assay

To determine the susceptibility of the selected strains to these drugs and to observe the effects of the drug combinations on them, we used CLSI M27-A3 checkerboard microdilution assay (CLSI, 2008). Briefly, drug dilutions and combinations were prepared in RPMI medium in microtiter plates. Negative (only medium) and positive (strain and medium without drug) controls were included. Strains were grown on Sabouraud agar at 35 °C and, after 24 h, one colony was transferred in 1 ml 0.85% saline solution. Inoculum was prepared by diluting yeast cells in RPMI medium to 3–5 × 10³ cells/ml using the automatic ImageJ-counting technique (Zupan et al., 2013). A total of 100 µl of this cell suspension was then transferred to 100 µl of drug suspension prepared in microplates. When the DMSO was used for drug dilution, it comprised <1% of the final test volume in the microtiter well. Tested drug concentrations ranged from 200–2 mg/l for MTX, 400–6.25 mg/l for AMX, 120–2 mg/l for MPA, 400–0.25 mg/l for Fk506, 16–0.125 mg/l for CsA, 256–0.25 mg/l for FLC, 256–0.25 mg/l for ITC, 1–0.004 mg/l for AMB, 16–0.068 mg/l for KCT, 16–0.017 mg/l for VRC, and 16–0.25 mg/l for POS. After incubation at 37 °C for 24, 48, or 72 h, we measured the optical density at 600 nm (OD₆₀₀) with a microplate reader (Tecan, Männedorf, Switzerland). Background optical densities were subtracted from that of each well. In vitro susceptibility and drug combination tests were performed at least in biological triplicates.

When we further investigated the observed mechanism of the antagonism between MPA and additional azole antifungals versus C. glabrata ATCC 2001 and Cgpdr1Δ mutant, we used a version of the assay described above but replaced SAB and RPMI media with YPD agar plates and broth, respectively. We have confirmed that the antagonistic effect in C. glabrata ATCC 2001 is present in both RPMI and YPD media, results are in Supplemental Information 1.

Fractional inhibitory concentrations index

The data obtained from the checkerboard microdilution assays were analysed with the fractional inhibitory concentrations index (FICI) based on the Loewe additivity model (Loewe, 1928), using the following equation: FICI = FIC_A + FIC_B, where FIC = MIC_combination∕MIC_individual. For azoles and immunomodulatory drugs MIC50 was used, and MIC90 for AMB. FICI is interpreted as synergistic when ≤ 0.5, indifferent when >0.5 and <4, and antagonistic when ≥ 4 (Odds, 2003). For calculation of the FICIs when the MIC resulted in an off-scale value, the next higher concentration (e.g., >32 = 64 mg/l) was used (Moody, 2010). FICI_min was reported as the FICI in all cases unless the FICI_max was greater than 4, in which case FICI_max was reported as the FICI for that particular data set (Meletiadis et al., 2005).

Bliss independence

The expected effect of drug combinations was also calculated by a model based on the Bliss independence (BI) criterion, where we assume that the relative effect of a drug at a particular concentration is independent of the presence of the other drug (Bliss, 1939; Goldoni & Johansson, 2007; Yeh et al., 2009). We calculated the predicted decrease of relative growth (E_predicted) using the following equation: E_predicted = 1 − E_A * E_B, where E_A and E_B are individually measured relative growth inhibitions by drugs A or B, respectively. Positive or negative deviations (ΔE = E_measured − E_predicted) from this predicted decrease of relative growth describe synergistic and antagonistic interactions, respectively (Yeh et al., 2009).

To interpret and summarize the entire interaction surface calculated by the BI criterion among several different drug combination concentrations, we used previously described interpretations (Meletiadis et al., 2005). Briefly, we summed all statistically significant ΔE (ΣSSI), determined the mean percentage (MSSI), and calculated the 95% confidence interval (CI). If it did not include 0 and was positive or negative, statistically significant synergy or antagonism, respectively, was claimed for the entire data set. In addition, we also calculated the ΣSSI and MSSI for all the significant synergistic (ΣSYN and MSYN, respectively) and antagonistic (ΣANT and MANT, respectively) ΔE separately. The absolute sum of all ΣSYN or ΣANT was considered to be a weak (0–100%), moderate (100–200%), or strong (>200%) interaction.

The interaction between two drugs was considered significant, if it was confirmed with at least one model (either BI or FICI).

Gene expression

C. glabrata ATCC 2001 was grown in liquid YPD overnight at 37 °C. In the morning, we diluted it in fresh YPD to OD₆₀₀ 0.05 and let it grow at 37 °C back to OD₆₀₀ 0.1 (approximately 1.5 h). At that point, we added the following drug combinations: (a) untreated, (b) 5 mg/l FLC, (c) 5 mg/l MPA, and (d) 5 mg/l FLC and 5 mg/l MPA. The optimal concentrations were determined with preliminary growth tests and checkerboard assays. All of the samples received the same amount of DMSO.

At timepoints 0, 2, and 4 h we collected the samples with a 3 min spin down at 1,500 g, 4 °C, resuspended them in ice-cold water, and transferred them to 2 ml screw caps, where we performed a short spin down to remove the supernatant and froze the pellet in liquid nitrogen. Samples were stored at −80 °C. Time points were determined according to growth curve assays (Supplemental Information 4).

RNA isolation and qPCR analysis were performed as described previously (Tscherner et al., 2012). PGK1 and RIP1 were used as housekeeping genes (Hnisz et al., 2010; Li, Skinner & Bennett, 2012).

The primers used were CgPGK1f (5′-ACGAAGTTGTCAAGTCCTCCA-3′), CgPGK1r (5′-TTACCTTCCAACAATTCCAAGGAG-3′), CgRIP1f (5′-CTTCATGGTCGGTTCTCT AGG-3′), CgRIP1r (5′-ACAACAACGTTCTTACCCTCAG-3′), CgPDR1f (5′-TACCAATG TCTCAGATACCACCA-3′), CgPDR1r (5′-CTGTCTTTAGAATCCAACTGCGT-3′), CgCDR1f (5′-AGACTTACGCTAGACATTTAACGG-3′), and CgCDR1r (5′-CACAAATA GAGACTTCAGCAATGG-3′). Amplification curves were analysed using the Realplex Software (Eppendorf) and relative mRNA quantification was performed using the efficiency corrected ΔΔCt method (Pfaffl, 2001). Quantification was performed in Excel (Microsoft) and statistical analysis with GraphPad Prism (GraphPad Software; GraphPad, San Diego, CA, USA) using one-way ANOVA and Bonferroni’s multiple comparison test.

Drug susceptibility

Each test had MICs of the control strains (C. parapsilosis ATCC 22019, C. krusei ATCC 6258) within the expected range for the tested antifungal. MICs for selected strains against individual drugs at 48 h are summarized in Table 2. Strains showed resistance to certain antifungals; for example, Sc2 (32–64 mg/l), Cg5 (64–128 mg/l), and Cg6 (128 mg/l) against FLC and Sc1 (2 mg/l), Sc2 (2–4 mg/l), Sc3 (2–4 mg/l), Cg5 (4–16 mg/l), and Cg6 (128 mg/l) against ITC (CLSI, 2008). For most of the immunomodulatory drugs, the concentration range that was used here, and was based on the range expected in human blood after drug administration, did not obtain a MIC; exceptions were some strains with MPA (Sc3, Sc4, Sc5, Cg2 at 120 mg/l) and Fk506 (Sc4, Sc6 at 200 mg/l).

Interpretation of drug interactions

Results were obtained using the checkerboard microdilution assay. A summary of the interpretations for each strain and drug combination is found in Fig. 1. The interaction was considered significant, if it was confirmed with at least one model (FICI or BI). Calculated FICI and BI values are located in Supplemental Information 3.

The FLC + MPA combination was antagonistic in eight out of 12 strains (five S. cerevisiae, three C. glabrata) and synergistic in two C. glabrata strains. One synergistic and two indifferent interactions involved highly resistant strains (MIC of 64 mg/l or higher). The AMB + MPA combination had a synergistic effect in eight out of 12 strains (three S. cerevisiae, five C. glabrata) and antagonism in one S. cerevisiae isolate. The effect of ITC + MPA was not as uniform; four synergistic and three antagonistic interactions were observed, indicating that the response is strain-specific.

A strain-specific response was observed in the combination of antifungals and MTX or AMX as well. MTX + FLC had two synergistic and three antagonistic interactions, MTX + ITC four synergistic and one antagonistic, MTX + AMB two synergistic and five antagonistic, AMX + FLC three synergistic interactions, AMX + ITC two synergistic and one antagonistic, and AMX + AMB five synergistic and four antagonistic interactions.

Out of 72 interactions between antifungals and calcineurin inhibitors (CsA and Fk506), 67 were synergistic, even in the FLC and ITC resistant strains Cg5 and Cg6. Antagonistic interactions were observed only in the combination of CsA and FLC in three S. cerevisiae isolates.

The results from the FICI and BI models fitted well, and both showed a trend towards the same interpretation; if the interpretation with one model was synergy, the other model either showed synergy or indifference, but never the opposite, antagonism. Out of 180 combinations tested against selected strains, the FICI model showed 60 significant modulatory interactions and the BI model 124, in which most of the absolute sum values (ΣSYN or ΣANT) indicated strong interactions (>200%).

Antagonistic interaction between MPA + azole antifungals

The strain C. glabrata ATCC 2001 was used to further explore the observed antagonism between MPA and azole antifungals (Fig. 2A). Azole antifungals included FLC, ITC, KCT, VRC, and POS. Table 3 shows the MIC, FICI, and BI of this strain against azole antifungals combined with MPA (MIC for MPA ranges from 32 to 64 mg/l) in YPD at 37 °C after 48 h. All of the interactions were interpreted as antagonistic by at least one model. This confirms that an antagonistic interaction between MPA and all of the selected azole antifungals occurs in C. glabrata ATCC 2001.

The roles of PDR1 and CDR1

Gene expression of PDR1 and CDR1 was analysed in C. glabrata ATCC 2001 at two different timepoints (Fig. 2). It was calculated relative to the untreated samples for each timepoint. All of the following differences described have statistical significance (p-value < 0.05).

At 2 h, PDR1 expression was significantly higher in samples treated with FLC. It was higher than in all the other conditions, including the samples treated with the combination of FLC + MPA. At 4 h the expression of PDR1 was still higher in samples treated with FLC compared to the untreated samples and samples treated with MPA.

For CDR1, at 2 h, all treated samples (FLC, MPA, FLC + MPA) had higher expression than the untreated samples. At 2 h FLC + MPA had higher expression of CDR1 than the FLC-treated samples. At 4 h the expression of CDR1 was even higher in MPA and FLC + MPA-treated samples and was significantly higher than in the samples treated only with FLC.

The expression patterns for CDR1, which seemed like a good candidate to explain the antagonism of the drug combination, and the low expression of PDR1 in the MPA and FLC + MPA-treated samples questioned whether the PDR1 regulatory cascade has a central role in the drug response mechanism responsible for antagonism between FLC and MPA. To test this, we performed a checkerboard assay with Cgpdr1 Δ mutant against the combination of FLC and MPA. Figure 2B shows the relative growth from this assay, in which we saw increased susceptibility to FLC (MIC at 4 mg/l) and the antagonistic pattern, which was confirmed by the BI model (ΣANT = −320.79%). These results suggest that the typical PDR1 drug response is not the only pathway required for this antagonistic interaction and that the induction of CDR1 in this case could to be regulated by alternative pathways.

Synergism: calcineurin inhibitors + antifungals

Certain patterns were observed with different drug combinations (Fig. 1). The most striking one was a generally synergistic effect between most antifungals and calcineurin inhibitors (CsA and Fk506) against all strains tested. The synergistic effect of calcineurin inhibitors and antifungals is well documented (Cruz et al., 2002; Li et al., 2015; Denardi et al., 2015; Yu, Chang & Chen, 2015). Our results differ to those of (Cruz et al., 2002) who reported synergistic toxicity toward other fungal species but not S. cerevisiae. This may be due to a strain-specific response in S. cerevisiae, but further tests are required to clarify this.

Antagonism: MPA + azoles

Eight out of twelve antagonistic interactions between MPA and FLC were found. Antagonism was not observed in strains that had high resistance to FLC (Sc2 32–64 mg/l, Cg5 64–128 mg/l, and Cg6 128 mg/l), which indicates that the antagonism is probably the result of similar mechanisms that produce the high resistance. FLC resistance mechanisms include overexpression of the efflux pumps in most cases (Whaley & Rogers, 2016). The antifungal activity of MPA, which is due to the depletion of purine nucleotides (Banerjee, Burkard & Panepinto, 2014), made the antagonistic effect all the more surprising. We therefore explored whether the most common mechanism of azole resistance in C. glabrata and S. cerevisiae (Pdr1 induction of efflux pump Cdr1/Pdr5) had a role in this antagonistic interaction.

Antagonism: the roles of PDR1 and CDR1

The central role of the transcriptional factor Pdr1 in azole resistance has been described in detail (Caudle et al., 2011; Yibmantasiri et al., 2014). It is usually linked to the induction of efflux pumps (Cdr1 as a dominant example) that remove the azole antifungals from the cell. In this study, relatively high expression values of CDR1 were observed in the drug combination FLC + MPA and MPA alone compared to the untreated samples and even to the FLC-treated ones (Fig. 2D). This makes the induction of efflux pumps a good explanation for the increased resistance to azole antifungals. However, an interesting aspect arose when examining the expression values of PDR1 (Fig. 2C) because their expression pattern did not match the CDR1 induction. Normally, Pdr1 positively regulates the expression of CDR1, but this was not seen here. This suggests that Pdr1 is not the only regulatory mechanism enabling a higher expression of CDR1 with MPA or the FLC + MPA combination. This was further demonstrated by an antagonistic effect in the Cgpdr1Δ mutant, interpreted by the BI model with a ΣANT value of −320.79%, although at lower concentrations of FLC (Fig. 2B). Undoubtedly Pdr1 still plays a major role in overall resistance (because resistance to FLC did drop in the combination versus Cgpdr1Δ), but the antagonistic pattern still existed. This opens up new and interesting questions regarding which mechanisms instead of the Pdr1 regulatory cascade are responsible for higher CDR1 expression and the observed antagonistic effect. Alternative regulators of CDR1 could include transcriptional factors associated with multidrug resistance (e.g., RDR1, YRR1, YRM1, and STB5) or CAD1, MSN2, and MSN4 for general stress response, YAP1 for oxidative stress, CRZ1 from calcineurin-mediated stress response, or ECM22 and UPC2 for cell membrane composition (Monteiro et al., 2017; Teixeira et al., 2018). In addition to Cdr1, other efflux pumps associated with azole resistance should also be considered in the future, such as Cdr2 (Miyazaki et al., 1998), Snq2 (Torelli et al., 2008), Flr1 (Alarco et al., 1997), Qdr2 (Costa et al., 2013), Tpo1_2 (Pais et al., 2016), Ybt1 (Tsai et al., 2010), Yhk8 (Barker, Pearson & Rogers, 2003) and Yor1 (Vermitsky et al., 2006). Further studies at the systemic level would be required to obtain a better picture of the entire mechanism.

Antagonism: clinical environment

There are clinical implications to this discovery. MPA (and its prodrug mycophenolate mofetil) is widely used as a maintenance immunosuppressive regimen in solid organ transplant patients, for the prophylaxis and treatment of acute and chronic graft-versus-host disease, and to promote engraftment after hematopoietic stem cell transplantation (Zhang & Chow, 2016). There is no data on the frequency of clinical usage for the combination of FLC + MPA; however, antifungal prophylaxis with azoles (VRC, POS, ITC, and FLC) is commonly prescribed in immunocompromised populations, which also involve the use of MPA (Pappas & Silveira, 2009; Brizendine, Vishin & Baddley, 2011; Groll et al., 2014). Possible antifungal prophylaxis or actual treatments with azole antifungals in these cases should therefore be used with caution and considered for each individual case because induced resistance could result in a failed therapy. There is even a report of a statistically significant increase in fungal infections in the geriatric renal transplant population when receiving MPA versus azathioprine, but the specific organisms and sites of infection were not reported (Meier-Kriesche et al., 1999; Ritter & Pirofski, 2009). It can be speculated that this antagonism could be connected to this increase in fungal infections, because FLC and other azoles (VRC, POS and ITR) are used for the prophylaxis in solid organ transplantations and other therapies involving immunocompromised patients (Pappas & Silveira, 2009; Brizendine, Vishin & Baddley, 2011; Groll et al., 2014; Vazquez, 2016). The next steps should expand the number of tested strains and species against the combination of azoles and MPA, include an in vivo evaluation of the combination, and include other drugs involved in a certain therapy as well; for example, a typical cocktail of prednisolone, MPA, and cyclosporine in solid organ transplantations (Sollinger, 1995).