Echinocandin tolerance and persistence in vitro are regulated by calcineurin signaling in Candida glabrata

TL;DR

This study shows that calcineurin signaling helps Candida glabrata survive antifungal treatment by increasing tolerance and persistence, and suggests using FK506 to improve treatment outcomes.

Contribution

The study identifies calcineurin signaling as a novel regulator of echinocandin tolerance and persistence in Candida species.

Findings

Calcineurin signaling increases the lifespan of C. glabrata cells during echinocandin treatment.

Persistence in C. glabrata is strongly dependent on calcineurin signaling, independent of Crz1.

Pre-activation of calcineurin with manogepix increases tolerance and persistence in C. glabrata.

Abstract

Upon exposure to echinocandins, growing yeast cells begin to accumulate cell wall damage and eventually die, resulting in therapeutic effects. While resistance to echinocandins is well studied, tolerance and persistence mechanisms that may also contribute to clinical failures and relapses remain understudied. In time-kill assays with micafungin in vitro, the opportunistic pathogen Candida glabrata exhibited biphasic kinetics of cell death. Modeling with exponential decay equations distinguished a fast-dying major population from a slow-dying minor population, indicative of persistence. A genome-wide forward-genetic screen revealed dozens of genes that appeared to regulate persistence and/or tolerance, but not resistance. Several of those genes encoded calcineurin and its upstream regulators. Using individual gene knockout mutants and FK506, we show that calcineurin signaling increases the lifespans of most C. glabrata cells through a process that is largely independent of Crz1, one of its downstream effectors. The formation of long-lived persister-like cells (i.e., persistence) was strongly dependent on calcineurin signaling, independent of Crz1. Pre-activation of calcineurin using genetic or chemical stressors, such as manogepix, strongly increased tolerance and persistence in C. glabrata, suggesting antagonism of echinocandin efficacy by this new antifungal. Calcineurin signaling was also necessary for the induction of tolerance and persistence in Candida albicans. The findings suggest that short-term administration of FK506 during the earliest stages of echinocandin treatment may improve clinical outcomes while possibly avoiding long-term immunosuppression.

Treatment of fungal infections is often unsuccessful. Potential causes of antifungal failure include tolerance and persistence, which are poorly understood processes used by fungal pathogens to survive antifungal treatment. This study utilizes detailed experimental protocols and genome-wide screens to discover how Candida glabrata induces tolerance and persistence to a major class of antifungals. The findings suggest that a clinical immunosuppressant may be repurposed to combat tolerance and persistence in this pathogenic yeast, as well as Candida albicans and perhaps other species.

Figures

Quantifying tolerance and persistence in C. glabrata in response to micafungin treatment. Eight single colonies of wild-type strain BG14 were grown to saturation for 72 h, then diluted 50-fold into fresh SCD medium containing 125 ng/mL micafungin alone (black) or micafungin plus 1 µg/L FK506 (red). Cultures were shaken at 30°C, sampled at various time points, serially diluted, and plated on drug-free YPD medium. Colony-forming units (CFU) were determined and the eight replicates averaged and charted (filled symbols). The best fit of the data to exponential decay equations is shown (solid smooth curves). Eight single colonies that appeared after 24 h of treatment were picked and retested in the same conditions (round 2; open symbols and dashed curves).

Genetic regulation of tolerance/persistence differs from that of resistance. A pool of Hermes transposon insertion mutants in the BG14 strain that was previously analyzed for resistance to low micafungin (resistance Z-score) was reanalyzed in high micafungin for tolerance and persistence to high micafungin (persistence/tolerance Z-scores) and charted. Overall, the Pearson correlation coefficient was poor (PCC = 0.05).

Calcineurin promotes tolerance and persistence. Time-kill assays of cna1∆ (blue), cnb1∆ (red), crz1∆ (green), and wild-type parent strain (black, strain BG14) were performed as described in Fig. 1. The averages of four biological replicates (symbols) were fit to exponential decay equations (smooth curves). Experiments containing FK506 (1 µg/mL) were performed in parallel (right panel).

Manogepix activation of calcineurin induces micafungin tolerance and persistence independent of Crz1. Wild-type (black, strain BG14), cnb1∆ (red), and crz1∆ (purple) strains were grown to saturation and diluted 25-fold into fresh medium containing (dashed lines) or lacking (solid lines) manogepix (0.6 µg/mL). After shaking at 30°C for 2 h, cultures were diluted 2-fold in fresh medium containing micafungin (125 ng/mL) and analyzed in time-kill experiments as described in Fig. 1. The averages of four biological replicates (symbols) were used for curve fitting (smooth curves).

Mannogepix activation of calcineurin induces micafungin resistance through Crz1 and Fks2. Growth (green) of wild-type and crz1∆ mutants (strains BG14, TJN07) was measured after incubation for 24 h in medium containing varying concentrations of mannogepix and micafungin. Similar effects were seen in two additional replicates.

Genetic activation of calcineurin increases tolerance and persistence. The fks1∆ (green), ipt1∆ (orange), and gtb1∆ (blue) mutants were generated in wild-type strain BG14 (black) or BG14-derived strain (BG87, gray symbols) and tested in time-kill experiments as described in Fig. 1 in the absence (smooth curves) and presence (dashed curves) of FK506. The averages of four biological replicates were used to generate the curve fits.

Hog1 diminishes tolerance and persistence independent of calcineurin and Crz1. The hog1∆ crz1∆ double knockout mutant (blue), the single knockout mutants (green, red), and wild-type control strain BG14 (black) were analyzed in time-kill experiments as described in Fig. 1, with the effects of FK506 illustrated separately (right panel). The averages of four biological replicates were used to generate the curve fits.

Calcineurin induces micafungin tolerance and persistence in C. albicans but has no impact on amphotericin B. Wild-type strain SC5314 of C. albicans (left) and strain BG14, a derivative of wild-type C. glabrata strain BG2 (right), were analyzed by time-kill experiments using 125 ng/mL micafungin (black lines) or 10 µg/mL amphotericin B (red lines) in the absence or presence of FK506 (dashed lines) as described in Fig. 1. The averages of four biological replicates were used to generate the curve fits.

Working model of resistance, tolerance, and persistence regulation in C. glabrata (strain BG2). Echinocandins and manogepix activate calcineurin signaling weakly and strongly, respectively, by engaging their cellular targets and generating stresses. In turn, calcineurin activates Crz1 and unknown targets that regulate echinocandin resistance and tolerance, respectively. Increased tolerance may allow time for the development of persistence. All these effects are strongly blocked by the calcineurin inhibitor, FK506. However, the HOG signaling pathway negatively regulates tolerance and persistence, independent of calcineurin signaling, through unknown effectors. HOG signaling is naturally compromised in strain CBS138 (not shown). Created in BioRender.

Full text

INTRODUCTION

At least four distinct processes contribute to antibiotic treatment failure: resistance, heteroresistance, tolerance, and persistence (1–3). Resistance occurs when microbes acquire mutations that alter the expression or function of the drug target, that alter the influx, efflux, sequestration, or metabolism of the compound, or that otherwise increase the effective dosage of the antimicrobial drug. Common measures of antibiotic resistance include the minimal inhibitory concentration (MIC) or the concentration causing a 50% inhibition of maximal growth (IC50). Resistance mutations are stably inherited by daughter cells and cause infections that require higher doses of the antibiotic or alternative antibiotics to treat. Heteroresistance occurs when a small subpopulation of cells within a clonal population acquires phenotypic resistance without any stable mutations. Heteroresistant subpopulations have a greatly increased MIC relative to the majority of cells in the population and have been shown to contribute to treatment failure in murine models of infection. Tolerance occurs when the clonal population has acquired mutations that increase the lifespan of the cells, even when exposed to excess antibiotics, without altering the MIC or IC50. To overcome tolerance, the duration of antibiotic treatment must be increased, not the dose. Tolerance is usually assayed using time-kill experiments that measure the number of “viable” cells in the population (colony-forming units, or CFU) after transient exposure to supra-MIC doses of antibiotics and estimate the half-life of the population. Persistence refers to an epigenetic regulatory process that produces a small number of highly tolerant “persister cells” amongst the large population of fast-dying susceptible cells. Persister cells are typically quiescent, at least transiently, and are therefore not killable by multiple classes of antibiotics, which can increase the likelihood of relapse. All these processes have been studied thoroughly in many infectious species of bacteria.

Antifungal resistance is well studied in some pathogenic fungi (4, 5), but heteroresistance, tolerance, and persistence are just beginning to be distinguished and individually unraveled at the molecular level (6). Tolerance and heteroresistance to azoles—fungistats targeting ergosterol biosynthesis in the ER—are now being investigated in several pathogenic species of yeasts (7–11), as there is growing evidence of the clinical relevance of these processes (12). Heteroresistance to echinocandins—fungicides targeting the cell wall—was recently associated with breakthrough infections by Candida parapsilosis (13). A retrospective study found that rare echinocandin-tolerant strains of Candida tropicalis were much more lethal than non-tolerant strains in patients treated for candidemia (14). Though the molecular mechanisms of echinocandin heteroresistance and tolerance have not been elucidated, an inhibitor of the Ca^2+^/calmodulin-dependent protein phosphatase calcineurin (FK506) abolished the tolerance of C. tropicalis and substantially increased the lifespan of infected animals undergoing echinocandin treatment (14). A better understanding of the regulatory mechanisms responsible for heteroresistance, tolerance, and persistence could potentially lead to the development of antifungal therapies that lack undesirable side effects on the patient (such as immunosuppression caused by FK506).

Candida glabrata (also known as Nakaseomyces glabratus) is the second most common cause of life-threatening candidemia and candidiasis next to C. albicans (15–17), although it belongs to a genus that is more closely related to Saccharomyces than true Candida (18). Due to its innate and easily acquired resistance to antifungals and its rising incidence, the World Health Organization has classified C. glabrata as a High Priority Threat (19). In C. glabrata, azole resistance mutations arise primarily in the target (ERG11) and in a transcription factor gene (PDR1) that regulates expression of transporters that efflux the drug (20). PDR1 also confers mild resistance to most echinocandins through expression of lipid flippases (21, 22). However, strong resistance to echinocandins arises primarily through mutations in the two targets, encoded by FKS1 and FKS2 (23, 24), the latter of which depends on calcineurin and the Crz1 transcription factor for maximal expression (25). When a resistance mutation arises in FKS2, its impact in vitro can be blocked by calcineurin inhibitors (25, 26). Even in the absence of antifungals, calcineurin also promotes virulence of C. glabrata in mouse models of invasive candidiasis through a mechanism that is potentially independent of Crz1 (27, 28). Time-kill experiments in vitro have shown that most C. glabrata cells die quickly when exposed to high doses of echinocandins, while a small and variable subpopulation dies more slowly (29, 30), suggesting the possible development of long-lived persister cells. Such persister cells serve as a reservoir for acquisition of resistance mutations (30). After engulfment into phagosomes by macrophages, C. glabrata cells typically remain viable and survive longer when exposed to echinocandins, resulting in an increased number of long-lived cells (31). Therefore, tolerance and persistence in this species may be clinically important and regulated by unknown mechanisms.

This study quantifies tolerance and persistence in C. glabrata exposed to echinocandins in vitro by fitting high-resolution time-kill data to exponential decay equations developed previously for antibiotic research (2). It also implements a genome-wide genetic screen using Tn-seq to identify specific regulators of tolerance and persistence, as well as individual gene knockout experiments. The findings suggest that tolerance and persistence are governed by processes distinct from resistance and heteroresistance. Remarkably, C. glabrata and C. albicans appeared to induce tolerance and persistence “on demand” through the activation of a calcineurin. Therefore, FK506 and other drugs that block calcineurin signaling in yeasts and other fungi may improve clinical outcomes by lowering tolerance and persister cell development, in addition to lowering resistance (expression of echinocandin targets). Conversely, the findings suggest that drugs, mutations, and host conditions that stress fungal cells and pre-activate calcineurin may promote tolerance and persistence, thereby worsening clinical outcomes.

RESULTS

Quantitation of echinocandin tolerance and persistence in C. glabrata

Time-kill assays have been used routinely to discriminate, quantify, and characterize subpopulations of cells that exhibit distinct rates of killing by antibiotics (2). Time-kill assays involve treatment of clonal populations with high doses of cidal drugs for varying lengths of time, plating the treated cells onto drug-free agar media at appropriate dilutions, and then counting the number of colonies that appear after additional incubation. The colony-forming units (CFU) per mL of starter culture are then charted over time and, if biphasic, analyzed by fitting to the sum of two exponential decay equations (2). Each exponential decay equation contains two unknown parameters—population size and half-life—that are estimated for the major and minor subpopulations (Fig. S0). When eight independent stationary-phase cultures of wild-type C. glabrata (strain BG14) were diluted into fresh medium containing 125 ng/mL micafungin, biphasic survival kinetics were observed in each after a brief lag (Fig. S1). The lag disappeared when log-phase cultures were tested, while the biphasic kinetics remained evident (Fig. S2). After excluding the lag and fitting the averaged CFUs to the summed exponential equations (Fig. 1, smooth curve), a minor population (9%) of long-lived cells (t ½ = 3.75 h) could be distinguished from the major population (91%) of fast-dying cells (t ½ = 0.75 h). To test whether stable mutants contributed to the slow-dying subpopulations, eight colonies that arose after 24-hour treatment with micafungin were picked, regrown in fresh medium, and retested individually. All eight cultures again produced similar numbers of long-lived cells (Fig. S1) with best-fit parameters resembling the original cultures (Fig. 1, open symbols and dashed curve). These findings show that clonal cultures of C. glabrata consistently and transiently produce two phenotypically distinct populations of cells, the minor one with nearly 5-fold increased half-life in micafungin.

Quantifying tolerance and persistence in C. glabrata in response to micafungin treatment. Eight single colonies of wild-type strain BG14 were grown to saturation for 72 h, then diluted 50-fold into fresh SCD medium containing 125 ng/mL micafungin alone (black) or micafungin plus 1 µg/L FK506 (red). Cultures were shaken at 30°C, sampled at various time points, serially diluted, and plated on drug-free YPD medium. Colony-forming units (CFU) were determined and the eight replicates averaged and charted (filled symbols). The best fit of the data to exponential decay equations is shown (solid smooth curves). Eight single colonies that appeared after 24 h of treatment were picked and retested in the same conditions (round 2; open symbols and dashed curves).

When micafungin was substituted with other glucan synthase inhibitors (e.g., caspofungin and ibrexafungerp), similar numbers of persister cells with similar lifespans were detected, although slight variation was evident (Fig. S3). The half-lives and the population sizes did not change even when the doses of these antifungals were varied over a broad range (Fig. S3). Additionally, the survival kinetics of several mutants that decreased (fks2∆, pdr1∆) or increased (mrp20∆) resistance to micafungin (21) were indistinguishable from those of the BG14 parent strain (Fig. S4). Altogether, these results suggest that C. glabrata can produce substantial numbers of long-lived persister-like cells through transient regulatory processes that are distinct from stable mechanisms that control resistance. Further, this mechanism occurred independent of heteroresistance, which was not present in wild-type C. glabrata strains (7). In the remainder of this study, the process of producing and maintaining such persister-like cells will be termed “persistence,” while the process governing lifespan of the remaining susceptible cells will be referred to as “tolerance,” in accordance with earlier conventions (2).

A forward genetic screen identifies regulators of tolerance and persistence

To identify genes that specifically regulate tolerance and/or persistence in C. glabrata, a genome-wide Tn-seq screen was implemented using a complex pool of transposon insertion mutants in strain BG14 that had been previously analyzed for resistance to micafungin and other antifungals (21, 32). Resistance screens employed very low doses of antifungals and long exposure times (48 h), which would not have uncovered the genes that regulate tolerance and persistence. To elucidate such genes, the transposon pool was exposed to a high, lethal dose of micafungin for a short duration (6 h), and then, the cultures were washed and regrown in drug-free fresh medium before Tn-seq analysis. Mutants with elevated tolerance or persistence would be enriched in this modified protocol (i.e., positive Z-scores), while mutants with diminished tolerance or persistence would be depleted from the general pool (i.e., negative Z-scores).

Tolerance/persistence Z-scores were calculated for 5,275 annotated genes by comparing the 6-hour and 0-hour exposures to 64 ng/mL micafungin (Table S1). When these Z-scores were charted against resistance Z-scores obtained previously with 8 ng/mL micafungin (21), the correlation was poor (Fig. 2). Hundreds of mitochondrial genes (Fig. 2, yellow symbols) that significantly increased resistance to micafungin when disrupted with transposons had little impact on tolerance/persistence. One such mitochondria-deficient mutant (mrp20∆) exhibited wild-type levels of tolerance and persistence in time-kill experiments as mentioned earlier (Fig. S4). Conversely, dozens of genes with significantly high or low tolerance/persistence Z-scores had resistance Z-scores close to zero (Fig. 2). These findings further suggest that micafungin tolerance and persistence may be controlled by a relatively small number of genes that are largely distinct from those that regulate micafungin resistance.

Genetic regulation of tolerance/persistence differs from that of resistance. A pool of Hermes transposon insertion mutants in the BG14 strain that was previously analyzed for resistance to low micafungin (resistance Z-score) was reanalyzed in high micafungin for tolerance and persistence to high micafungin (persistence/tolerance Z-scores) and charted. Overall, the Pearson correlation coefficient was poor (PCC = 0.05).

We focused first on a set of 36 genes that exhibited low tolerance/persistence (Z < −3.0) without exhibiting low resistance (Z > −3). Z-scores less than −3 and greater than +3 correspond to P-values less than 0.0027 in two-tailed normal distributions or a false discovery rate of about 13 genes in this data set. GO term analysis (33) indicated significant enrichment of only one process (P-value = 5.2E-5; false discovery rate = 6.3E-2): regulators of calcineurin (CNB1, RCN1), a well-studied Ca^2+^/calmodulin-dependent protein phosphatase. An upstream activator of calcineurin (KCH1) in S. cerevisiae and C. albicans (34, 35) was also present on this list, but was not grouped in this ontology. The CNA1 gene, encoding the catalytic subunit of calcineurin, was not significant (Z = 0.26) possibly due to the presence of an autoinhibitory domain at the C-terminus that would respond to transposon insertions in the opposite direction from insertions in the catalytic domain (see Discussion). To explore the possible involvement of calcineurin in the regulation of tolerance and persistence, micafungin time-kill experiments were performed on wild-type C. glabrata in the presence of FK506, a specific inhibitor of calcineurin. Strikingly, FK506 caused a 2.5-fold decrease in the half-life of the susceptible cells (i.e., tolerance) and a 270-fold decrease in the number of long-lived persister-like cells in micafungin (Fig. 1, red symbols and curves). Of eight cells that survived the 24-hour exposure to micafungin plus FK506 and produced colonies in drug-free media, all produced wild-type patterns in time-kill experiments upon retesting in the same conditions (Fig. 1 dashed curve, Fig. S1). These findings using FK506 confirm the Tn-seq results, suggest that calcineurin signaling drives both tolerance and persistence, and demonstrate that these behaviors are readily reversible in C. glabrata. As calcineurin signaling becomes activated in response to micafungin and other stressors of the cell wall (36), tolerance and persistence may be protective behaviors in C. glabrata that are induced in response to stresses.

Calcineurin drives tolerance and persistence in C. glabrata independent of Crz1 and Rcn2

The CNA1 and CNB1 genes were each knocked out in the BG14 parent strain and analyzed by time-kill experiments and mathematical modeling. Both the cna1∆ and cnb1∆ mutants exhibited strongly diminished tolerance and persistence relative to the wild-type control (Fig. 3, left panel). Though the number of persister-like cells decreased by more than 100-fold in both mutants relative to wild-type, the half-lives of the persister cells remained constant in all strains at approximately 3.2 h. As expected, the presence of FK506 in the time-kill experiments did not impact the behavior of cna1∆ and cnb1∆ mutants and forced the wild-type parent strain to behave like the calcineurin-deficient mutants (Fig. 3, right). These findings further suggest that calcineurin signaling drives both tolerance and persistence upon exposure to micafungin.

Calcineurin promotes tolerance and persistence. Time-kill assays of cna1∆ (blue), cnb1∆ (red), crz1∆ (green), and wild-type parent strain (black, strain BG14) were performed as described in Fig. 1. The averages of four biological replicates (symbols) were fit to exponential decay equations (smooth curves). Experiments containing FK506 (1 µg/mL) were performed in parallel (right panel).

An important effector of calcineurin signaling is the transcription factor Crz1 (27). Though the CRZ1 gene was not significant in the Tn-seq screen, a crz1∆ mutant exhibited levels of tolerance and persistence in time-kill assays that were intermediate between wild-type and the cna1∆ and cnb1∆ mutants (Fig. 3, left). In the presence of FK506, the crz1∆ mutant resembled the calcineurin-deficient strains (Fig. 3, right). Thus, Crz1 seemed to be required for a portion of the effects of calcineurin. Targets of Crz1 include FKS2 and RCN2 (25, 28, 37), neither of which were significant in the Tn-seq screen. An fks2∆ mutant was indistinguishable from wild-type in time-kill experiments (Fig. S4), suggesting that it is not required for calcineurin-induced tolerance and persistence. An rcn2∆ mutant exhibited wild-type levels of tolerance and persister-like cells, but interestingly, the half-life of the persister-like cells increased by ~1.5-fold (Fig. S5, left). In the presence of FK506, the rcn2Δ mutants were indistinguishable from wild-type (Fig. S5, left). A plasmid that overexpresses RCN2 from a strong constitutive PDC1 promoter did not alter micafungin tolerance or persistence relative to a control plasmid when introduced into cna1∆ mutants or wild-type cells (Fig. S5, right). These findings suggest that Rcn2 functions in its canonical role as a feedback inhibitor of calcineurin signaling (28, 37), rather than as an effector of calcineurin signaling in the regulation of tolerance and persistence.

To further explore the role of calcineurin in regulating tolerance and persistence, the effects of manogepix were studied. Manogepix is a preclinical antifungal that blocks GPI anchor biosynthesis in the ER (38), which produces cellular stresses that very strongly activate calcineurin signaling in C. glabrata (36). A 2-hour pre-treatment of wild-type cells with manogepix resulted in moderately increased tolerance and persistence to micafungin (Fig. 4). Though no such increases were observed in cnb1∆ mutants, the crz1∆ mutant exhibited a dramatic increase in micafungin tolerance when pre-exposed to manogepix (Fig. 4). Checkerboard assays were utilized to determine whether manogepix increases resistance to micafungin in crz1∆ mutants. Manogepix did not antagonize micafungin in crz1∆ mutants (Fig. 5). However, manogepix did antagonize micafungin in wild-type parent strain, possibly due to calcineurin and Crz1-dependent expression of target genes such as FKS2. These findings suggest that calcineurin signaling is a key driver of tolerance and persistence in C. glabrata, with a portion of its effects mediated by Crz1 and another portion mediated by unknown effectors.

Manogepix activation of calcineurin induces micafungin tolerance and persistence independent of Crz1. Wild-type (black, strain BG14), cnb1∆ (red), and crz1∆ (purple) strains were grown to saturation and diluted 25-fold into fresh medium containing (dashed lines) or lacking (solid lines) manogepix (0.6 µg/mL). After shaking at 30°C for 2 h, cultures were diluted 2-fold in fresh medium containing micafungin (125 ng/mL) and analyzed in time-kill experiments as described in Fig. 1. The averages of four biological replicates (symbols) were used for curve fitting (smooth curves).

Mannogepix activation of calcineurin induces micafungin resistance through Crz1 and Fks2. Growth (green) of wild-type and crz1∆ mutants (strains BG14, TJN07) was measured after incubation for 24 h in medium containing varying concentrations of mannogepix and micafungin. Similar effects were seen in two additional replicates.

Chronic activation of calcineurin increases tolerance and persistence

Some of the genes with positive tolerance/persistence Z-scores in the Tn-seq screen may generate chronic cellular stresses that pre-activate calcineurin when disrupted by transposons. One such gene is FKS1 (Z = 3.38), which encodes the major catalytic subunit of glucan synthase (25, 39, 40). The fks1∆ mutants are FK506-sensitive because they depend on calcineurin signaling, Crz1, and elevated expression of FKS2 for viability (25). When tested in time-kill assays, fks1∆ mutants exhibit a large increase in tolerance compared to BG14 and the parent strain (Fig. 6, Left). Thus, chronic pre-activation of calcineurin through Fks1 deficiency may increase tolerance and persistence much like the acute activator, manogepix.

Genetic activation of calcineurin increases tolerance and persistence. The fks1∆ (green), ipt1∆ (orange), and gtb1∆ (blue) mutants were generated in wild-type strain BG14 (black) or BG14-derived strain (BG87, gray symbols) and tested in time-kill experiments as described in Fig. 1 in the absence (smooth curves) and presence (dashed curves) of FK506. The averages of four biological replicates were used to generate the curve fits.

Several genes encoding ER proteins (e.g., CNE1, GTB1, KEG1, SKN1, VMS1) exhibited strongly increased tolerance/persistence Z-scores (Table S1; Fig. 2). Knockout mutants of GTB1 were generated and tested in time-kill experiments in the presence and absence of FK506. The gtb1∆ mutants exhibited elevated tolerance to micafungin that was strongly blocked by FK506 (Fig. 6, Right). The IPT1 gene, which was also highly significant in the Tn-seq screen (Z = 13.0), encodes an enzyme in the Golgi complex that synthesizes the abundant sphingolipid M(IP2)C. In time-kill experiments, an ipt1∆ mutant exhibited elevated tolerance and persistence that was likewise blocked by FK506 (Fig. 6, Right). Heteroresistance to micafungin was slightly increased in the fks1∆ mutant (Fig. S6). These findings suggest that many negative regulators of calcineurin signaling may be among the list of genes with positive Z-scores in micafungin tolerance/persistence screens. However, other gene deficiencies may impact tolerance and persistence through calcineurin-independent effects.

HOG signaling pathway negatively regulates tolerance and persistence independent of calcineurin

Calcineurin reverses the effects of serine/threonine protein kinases on shared substrates. Two genes encoding protein kinases (FPK1, SSK2) exhibited strongly positive Z-scores in the tolerance/persistence screen and therefore could produce phosphoproteins that are directly targeted by calcineurin. However, they were also strongly positive for micafungin resistance (Fig. 2). SSK2 encodes a MAPKKK that phosphorylates and activates Pbs2, a MAPKK, that in turn phosphorylates Hog1, a MAPK that responds to high-osmolarity signals (41). Though PBS2 and HOG1 were near zero in the Tn-seq screens, an upstream regulator of SSK2 (SSK1) also exhibited positive tolerance/persistence and resistance Z-scores. Knockout mutants lacking the SSK2 or PBS2 genes in the BG2 background, and HOG1 in the BG14 background, were analyzed in time-kill experiments using 8-fold higher doses of micafungin to compensate for their mild resistance to echinocandins. All three mutants exhibited elevated tolerance relative to the wild-type control in the presence of FK506 (Fig. S8; Fig. 7). These results indicate that the HOG signaling pathway normally negatively regulates tolerance and persistence, independent of calcineurin. In the absence of FK506, calcineurin signaling strongly increased tolerance and persistence in all three mutants (compare left and right panels of Fig. 7; Fig. S8), which indicates that calcineurin promotes tolerance and persistence independent of HOG signaling. The effects of hog1∆ mutations were also observed in a crz1∆ mutant background (Fig. 7). Thus, HOG signaling can negatively regulate tolerance and persistence independent of Crz1 and calcineurin.

Hog1 diminishes tolerance and persistence independent of calcineurin and Crz1. The hog1∆ crz1∆ double knockout mutant (blue), the single knockout mutants (green, red), and wild-type control strain BG14 (black) were analyzed in time-kill experiments as described in Fig. 1, with the effects of FK506 illustrated separately (right panel). The averages of four biological replicates were used to generate the curve fits.

The CBS138 strain of C. glabrata utilizes calcineurin but not HOG signaling to regulate micafungin tolerance and persistence

The CBS138 strain of C. glabrata naturally carries a mutation that inactivates the SSK2 component of the HOG signaling pathway (41). This mutation may be adaptive by providing enhanced tolerance and persistence, resulting in infections that are more difficult to treat. To begin investigating this possibility, time-kill experiments were performed on cna1∆, cnb1∆, and crz1∆ mutants in the CBS138 strain background and on pbs2∆ mutants in the 2001 strain background derived from CBS138. Unlike its effects in the BG2 strain background, the pbs2∆ mutation had no significant impact on tolerance and persistence in the CBS138-derived strain background, which exhibited somewhat elevated tolerance and persistence even in the presence of FK506 (Fig. S9). The effects of cna1∆ and cnb1∆ were conserved in CBS138; however, the effects of crz1∆ were smaller in CBS138 relative to BG2-derived strain background (Fig. S7). The natural deficiency of HOG signaling and other polymorphisms in CBS138 may contribute to its enhanced antifungal tolerance. Interestingly, a recent survey of CBS138 and dozens of additional C. glabrata strains revealed considerable variation in the number of persister-like cells observed after micafungin exposure (30).

Calcineurin signaling promotes tolerance and persistence to micafungin, but not amphotericin B, in C. albicans

To test whether calcineurin drives tolerance and persistence in C. albicans, micafungin time-kill experiments were performed on wild-type strain SC5314 in the presence and absence of FK506. Biphasic kinetics of cell death were observed in both scenarios, while the loss of calcineurin signaling decreased the half-life of susceptible cells approximately 1.4-fold and decreased the number of persister-like cells by approximately 100-fold (Fig. 8, black symbols). Similar results were obtained previously using cna1∆/∆ mutants of C. albicans instead of FK506 (42). Curiously, the half-life of persister-like cells in C. albicans was much longer than that of C. glabrata, indicating that the two species may regulate the process in different ways.

Calcineurin induces micafungin tolerance and persistence in C. albicans but has no impact on amphotericin B. Wild-type strain SC5314 of C. albicans (left) and strain BG14, a derivative of wild-type C. glabrata strain BG2 (right), were analyzed by time-kill experiments using 125 ng/mL micafungin (black lines) or 10 µg/mL amphotericin B (red lines) in the absence or presence of FK506 (dashed lines) as described in Fig. 1. The averages of four biological replicates were used to generate the curve fits.

Amphotericin B, which kills fungal cells by depleting ergosterol and creating pores in the plasma membrane (43), was also found to kill C. albicans and C. glabrata with biphasic kinetics after a brief lag (Fig. 8, red symbols). The half-life of the susceptible population was 2- to 2.5-fold shorter than observed with micafungin, and in both species, the killing kinetics were not altered by FK506 (dashed lines). These findings suggest that calcineurin specifically drives tolerance to echinocandins, but not amphotericin B in these species and conditions.

DISCUSSION

Though there is concern that tolerance and persistence will contribute to therapeutic failure, relapse, and perhaps even the acquisition of resistance (30, 44), the regulatory mechanisms behind these phenomena have been understudied in fungal pathogens of humans. This study shows that calcineurin signaling is a major driver of tolerance and persistence to echinocandins in C. glabrata (illustrated in Fig. 9) and C. albicans. Previous studies have shown that calcineurin and Crz1 also produce mild resistance to echinocandins in C. glabrata by increasing expression of the drug target, Fks2 (25) (see Fig. 9). When strong resistance mutations in FKS2 arise in patients, calcineurin inhibitors greatly diminish their impact in vitro (25, 39). Curiously, the CRZ1, CNA1, and CNB1 genes did not exhibit significant Z-scores in Tn-seq experiments assessing resistance to echinocandins (21). The failure to detect these genes may be attributed to poor coverage of the genes with transposon insertions, small effect sizes in the experimental conditions, or a combination of these factors. Nevertheless, the CNB1 gene was significant in Z-scores for tolerance/persistence, and follow-up experiments showed that CNB1 and CNA1 increase tolerance and persistence largely independent of CRZ1, FKS2, and RCN2. Although the target of calcineurin that promotes tolerance and persistence has not yet been identified, the HOG signaling pathway was not required. FK506 still lowered tolerance and persistence in ssk2∆, pbs2∆, hog1∆, and crz1∆ hog1∆ mutants. However, all these mutants still exhibited elevated tolerance and persistence in the absence of calcineurin signaling, again through unknown effectors (Fig. 7; Fig. S8). Interestingly, the HOG signaling pathway is naturally polymorphic in different strains of C. glabrata, contributing to variation in the resistance to micafungin (21) and other stressors (41), as well as variation in tolerance and persistence within the species. Multiple strains of C. glabrata may be needed to obtain a more complete picture of resistance, tolerance, and persistence mechanisms in this diverse species. As calcineurin also increases echinocandin tolerance in C. albicans and A. fumigatus (45, 46), the mechanism may be broadly conserved among diverse fungal pathogens.

Working model of resistance, tolerance, and persistence regulation in C. glabrata (strain BG2). Echinocandins and manogepix activate calcineurin signaling weakly and strongly, respectively, by engaging their cellular targets and generating stresses. In turn, calcineurin activates Crz1 and unknown targets that regulate echinocandin resistance and tolerance, respectively. Increased tolerance may allow time for the development of persistence. All these effects are strongly blocked by the calcineurin inhibitor, FK506. However, the HOG signaling pathway negatively regulates tolerance and persistence, independent of calcineurin signaling, through unknown effectors. HOG signaling is naturally compromised in strain CBS138 (not shown). Created in BioRender.

A recent study concluded that mitochondrial dysfunction decreases tolerance to echinocandins in C. glabrata (47). That conclusion is inconsistent with the genetic data reported there (47) and here (Fig. S4), showing that several mitochondria-deficient knockout mutants (cox4∆, atp1∆, atp2∆, atp10∆, mrp20∆) exhibited wild-type levels of viability after long exposure to high doses of echinocandins. Although five genes encoding mitochondrial proteins exhibited tolerance/persistence Z-scores less than −3.0 in this study, all five are likely false positives because the cutoff score was not very stringent and because all the critical partners of these five genes exhibited Z-scores that were insignificant. The five genes and their partners are KGD1 (partner of KGD2), CYT2 (partner of CYT1 and eight other genes encoding respiratory complex III), SDH5 (partner of SDH3, SDH4, and SDH9), COQ1 (partner of eight other genes required for coenzyme-Q biosynthesis), and DNM1 (partner of FIS1). The recent report (47) did not make use of time-kill experiments and extensively utilized low-resolution dose-kill experiments, in which CFU are quantified at only one time point (24 h) after exposure to varying doses of echinocandins (23). That same study also tested “mitochondrial inhibitors” that either have multiple cellular targets (sodium azide) or have no targets at all in C. glabrata (diphenyleneiodonium chloride, rotenone), producing data that did not match the mitochondria-deficient mutants. Thus, little evidence supports the conclusion that mitochondrial dysfunction decreases tolerance to echinocandins. Strong evidence shows that mitochondrial dysfunction can increase resistance to echinocandins (Fig. 2) by activating Pdr1, which induces expression of lipid flippases (Rta1, Rsb1) that somehow lower susceptibility to most echinocandins (21). This study shows the value of high-resolution time-kill experiments and mathematical modeling in distinguishing resistance mechanisms from tolerance and persistence mechanisms.

A noteworthy observation of this study is that the estimated number of persister-like cells (i.e., persistence) correlated with the half-life of susceptible cells (i.e., tolerance) in all the mutant strains tested here. By increasing tolerance in the majority of cells in the population, calcineurin signaling may buy time for persister-like cells to form de novo through other regulatory processes. In the absence of calcineurin signaling, tolerance was low, and persister-like cells were rare, but still readily detectable. When calcineurin was pre-activated by mutations or inhibitors that target the ER (e.g., manogepix), tolerance and persistence were elevated (Fig. 4). The ability of FK506 to block these effects suggests that ongoing calcineurin signaling is necessary to maintain tolerance during micafungin exposure and enable persistence to arise. In addition to increasing micafungin tolerance and persistence independent of Crz1, manogepix increased micafungin resistance through Crz1-dependent processes (Fig. 5). Manogepix, a fungistat that is currently in phase 2 clinical trials to treat a range of fungal infections (38), may therefore produce adverse drug-drug interactions with echinocandins. However, when combined with FK506, manogepix becomes fungicidal against C. glabrata and a wide range of fungal pathogens (36, 48). The role of calcineurin in manogepix tolerance may overlap with previously studied calcineurin-dependent tolerance to tunicamycin and to azole-class antifungals, which target N-glycosylation and ergosterol biosynthesis in the ER, respectively (49).

Other environmental conditions may affect calcineurin and HOG signaling pathways leading to impacts on echinocandin tolerance and persistence. Engulfment of C. glabrata cells into phagosomes of macrophages produced elevated micafungin tolerance and persistence (31). Tolerance to amphotericin B was also observed for engulfed C. glabrata cells (31). It will be interesting to determine whether calcineurin and/or HOG signaling govern these processes, or whether the longer lifespans of engulfed cells arise simply through slower growth rates or perhaps slower drug access to this intracellular compartment. Calcineurin signaling promotes virulence of C. glabrata (27, 28), C. albicans (42, 50, 51), and other fungal pathogens (45, 52–54) in mouse models of systemic candidiasis. In C. glabrata, the effects of calcineurin were independent of Crz1 (27, 28). A new interpretation of those findings is that host environments produce toxins or hostile conditions to the pathogens and generate stresses, which might normally activate calcineurin, induce tolerance and persistence behaviors, and promote fungal cell survival during infection. Manogepix treatment may further activate calcineurin and augment tolerance and persistence mechanisms, while still exerting a fungistatic effect. On the other hand, FK506 treatment may lower the defenses of C. glabrata and enable much more rapid and complete killing by host-derived assaults as well as echinocandins. Non-immunosuppressive analogs of FK506 that specifically target fungal calcineurin could be highly effective antifungals alone or in combination with existing antifungals (55). Although the immunosuppressive effects of FK506 would preclude long-term use of this compound as an antifungal co-drug, our findings raise the possibility that short-term or bolus co-administration of FK506 during standard echinocandin therapies could improve outcomes without producing sustained immunosuppression.

In a mouse model of invasive candidiasis by C. albicans, the efficacy of fluconazole was increased by co-administration of FK506 (56). Tolerance to fluconazole and other inhibitors of lanosterol demethylase (Erg11) in C. glabrata, C. albicans, and other yeast species has long been known to depend on calcineurin signaling, but not Crz1 (27, 28, 49, 52, 57). Calcineurin also promoted tolerance to terbinafine in C. albicans (58), to tunicamycin and dithiothreitol in S. cerevisiae (59, 60), and to SDZ 90-215 in all these species (61). These compounds all block essential enzymes in the ER or Golgi complex of the fungal cells and are considered fungistatic in wild-type cells but fungicidal in calcineurin-deficient cells. Because the wild-type yeasts do not lose viability in the presence of these compounds and often continue to replicate slowly for a few doublings, the degree of tolerance conferred by calcineurin signaling is difficult to quantify using the mathematical models employed in this study. The effectors of calcineurin responsible for tolerance to these fungistats remain unknown. Therefore, it is not yet possible to determine whether the same effectors govern tolerance to these fungistats as well as to the fungicidal echinocandins. Our Tn-seq screen has produced dozens of candidate genes (e.g., PHO84 with Z = −12.6) that could mediate the effects of calcineurin or Hog1. Apart from RCN1, only two genes on this list (BOI2, RGA2; both involved in polarized cell growth) encode homologs that were found to be dephosphorylated by calcineurin in S. cerevisiae (62, 63). Deeper analysis of these candidates is ongoing because they may serve as targets for development of non-immunosuppressive strategies to increase the efficacies of our most important antifungals.

Although calcineurin promoted tolerance to echinocandins and azoles, it did not seem to alter the biphasic kinetics of survival during exposure to amphotericin B in C. glabrata, C. albicans, or S. cerevisiae (64). Amphotericin B kills growing and non-growing fungal cells by sponging essential ergosterols from the plasma membrane and generating lethal pores (43). The procedures outlined in this study could be adapted for studies of amphotericin B tolerance and persistence mechanisms. As new derivatives of amphotericin B with dramatically less animal toxicity are being developed (65), a more complete understanding of polyene resistance, tolerance, and persistence mechanisms in fungi could further increase their therapeutic index.

MATERIALS AND METHODS

Candida strains and plasmids

All strains and their sources are listed in Table S2. C. glabrata strains were derived from BG14, a ura3∆ derivative of wild-type strain BG2 (66). Individual gene knockouts were constructed using the PRODIGE method (67), in which coding sequences were replaced with the coding sequence of URA3 from S. cerevisiae plasmid pRS406 (68) using oligonucleotides listed in Table S3. Colony PCR was used for screening validation. The RCN2 gene was PCR-amplified from genomic DNA of strain BG14 and cloned into the centromeric plasmid pCN-PDC1 (69) to generate pCN-PDC1-RCN2 and authenticated via Sanger sequencing.

Antifungal drugs

Micafungin (Cat. #18009) and amphotericin B (Cat. #11636) were obtained from Cayman Chemicals; caspofungin (Cat. #S3073), tacrolimus (FK506) (Cat. #S5003), and mannogepix (E1210) (Cat. #S0491) were obtained from SelleckChem; and ibrexafungerp was obtained from Scynexis.

TN-seq screen for genes that regulate micafungin tolerance and persistence

Pool-3 of Hermes-NAT1 insertion mutants in strain BG14 that had been studied previously for resistance to echinocandins (21) was thawed from storage at −80°C, grown to saturation, then diluted into fresh SCD-0 medium containing 64 µg/mL micafungin. After 0 and 6 h of shaking at 30°C, the culture was sampled, chilled on ice, washed once with chilled SCD-0 medium, resuspended in SCD-0, and shaken at 30°C for another 48 h to allow regrowth of surviving cells. The cells were pelleted, washed, and then genomic DNA was extracted from using the Quick-DNA Fungal/Bacterial Miniprep Kit from Zymo Research (Cat. #D6005). gDNA was fragmented by sonication, repaired, ligated to adapters, size-selected, and then insertion sites were PCR amplified and sequenced on an Illumina MiSeq instrument as described previously (21). FastQ files were demultiplexed using CutAdapt (70) and aligned to the BG2v1 reference genome (71) using Bowtie2 (72). Mapped reads with low quality (Q < 20) or mismatches at position 1 were discarded, and the remainder were tabulated for each site in the genome. The total number of sequence reads in each annotated gene was tabulated, and Z-scores were calculated for both time points with respect to the starting pool and with respect to each other as before (21). Gene Ontology analysis was performed on subsets of analyzed genes using GOrilla in S. cerevisiae mode (33).

Time-kill assays and analyses

At least four single colonies of each C. glabrata strain were picked and grown to saturation for 72 h in SCD-0 media at 30°C. Microscopic observations indicated that the C. glabrata cells were monodispersed and not detectably clumped or aggregated. Each replicate culture was diluted 1:50 into fresh media containing antifungals and shaken at 30°C. Aliquots were periodically removed, serially diluted in SCD-0 medium, and immediately plated onto SCD-0 agar medium using a sterile pinning device or by pipetting. Plates were incubated at 30°C until individual colonies were visible under a dissecting microscope at 10× magnification. Visible colonies were counted manually at one dilution for each sample, and CFU per mL of undiluted culture was calculated. Replicates were averaged arithmetically and charted. For curve fitting, the average CFU was log transformed and fit to the equation y = log(m1exp(-m2x) + m3exp(-m4x)) by non-linear regression in Kaleidagraph (v5.04, Synergy), after exclusion of data points within the lag phase. Parameters m1 and m3 represent sizes of the non-persister and persister cell populations, respectively. Parameters m2 and m4 represent kill constants, which were converted to half-lives (= ln2/m2, ln2/m4) of the two populations (tabulated in Table S4 along with error estimates). Smooth curves were generated using these parameters and overlayed onto the CFU data points. An illustrative example of the method is depicted in Figure S0. The same procedure was applied for analysis of C. albicans, except that YPD medium was used.

Population analysis profiling (PAP) assay

Single colonies were grown to saturation for 72 h in SCD-0 media at 30°C. The saturated cultures were diluted 1:50 into fresh SCD-0 media and then serially diluted 1:5. Two microliters of each dilution was spotted onto agar YPD media containing varying doses of micafungin (15 ng/mL to 0.234 ng/mL). Alternatively, when viability was very low, 200 µL of undiluted cultures were plated. Plates were incubated for 24 h at 30°C, and single colonies were counted manually. CFU per mL original culture was then calculated, and replicates were averaged.

Checkerboard assay

Varying concentrations of micafungin and mannogepix were prepared in SCD-0 and aliquoted to their respective columns and rows in a 96-well dish. Three biological replicates of BG14 and crz1∆ strain were grown to saturation and diluted 3:2,000 in fresh SCD-0 media before being aliquoted into 96-well plates at one-third of the final well volume for 1:2,000 cell dilution. The 96-well plates were incubated for 24 h at 30°C, mixed, and optical density at 600 nm was measured (Accuris SmartReader 96T).