The Metabolism of Susceptibility: Clearing the FoG Between Tolerance and Resistance in Candida albicans

Abstract

Purpose of Review

Failure of antifungal treatment is alarmingly common in patients infected with Candida albicans isolates that test as susceptible in vitro. This means that clinical susceptibility tests have limited predictive value for treatment success. To guide the improvement of patient outcomes, we must understand the effects of environmental and metabolic states on drug responses.

Recent Findings

Lab conditions often deviate from host environments, and current susceptibility testing standards ignore slow-growing, tolerant phenotypes; both factors may contribute to antifungal treatment failure. Metabolomic studies reveal that strain background, nutrient availability, and drug exposure influence the metabolic state of C. albicans cells; similarly, the metabolic state influences drug susceptibility.

Summary

Identifying tolerant strains in the clinic may improve patient outcomes. Studies that analyze the effects of essential but limited nutrients have the potential to improve the avoidance of persistent candidiasis and to reduce the frequency of antifungal treatment failures. Here, we highlight literature that explores the effect of drug exposure and antifungal drug resistance status on the C. albicans metabolome. Similar analyses need to be carried out relative to antifungal drug tolerance. Additionally, we focus on the biological relevance of four essential small molecules—iron, zinc, phosphate, and sphingolipids—to antifungal tolerance and resistance.

Introduction

Invasive fungal infections are an underappreciated threat to human health. Candida species, most prominently Candida albicans, are the ninth most common causative agent of bloodstream infections in the USA; such infections have an associated mortality rate estimated at around 20% [1, 2]. Currently, a limited number of antifungals are available, and a meta-analysis of clinical trials found only a 67.4% treatment success using these antifungals [3]. Yet, in vitro levels of antifungal drug resistance remain very low: less than 1% of C. albicans isolates test as resistant to fluconazole, the most commonly used antifungal drug [3]. Mechanisms of azole resistance directly affect drug-target interactions: drug uptake, drug efflux, and synthesis of the drug target (which in the case of fluconazole and other azoles is lanosterol dem-ethylase, a central enzyme in ergosterol biosynthesis). Furthermore, drug resistance is usually due to heritable genetic mutations, and all mutant progeny exhibits a similar level of drug resistance.

Antimicrobial susceptibility tests are used to predict the clinical success of antimicrobial therapies and generally measure the minimal inhibitory concentration (MIC) after 24 h of growth in the presence of a drug. Such tests are performed under standardized conditions proscribed by the European Committee for Antimicrobial Susceptibility Testing (EUCAST) or the Clinical and Laboratory Standards Institute (CLSI) [4–6]. Fungal infections follow the “90/60” rule in predicting therapeutic outcomes: ~ 90% of susceptible isolates and ~ 60% of resistant isolates respond to treatment [7, 8]. Given the high levels of treatment failure, the MIC determined in vitro is only weakly correlated with the clinical outcome; thus, we need a clearer understanding of how C. albicans overcomes drug stress beyond the standard measures of resistance.

“Tolerance” or “trailing growth” (termed tolerance in this review) [8–10] is detected in a large number of clinical isolates (20 to 60%) [11–13] and has the potential to explain some of the disparity between treatment failure and the low level of antifungal drug resistance, especially since clinical MIC measurements are designed to avoid measuring residual fungal growth. Clinical assays are read at 24 h, yet tolerant cells grow slowly, appearing above the MIC after 24 to 48 h of growth [13, 14]. Thus, tolerance is largely ignored in clinical assays. Antifungal drug tolerance, measured either as average supra-MIC growth after 48 h in a broth microdilution assay or as the fraction of growth (FoG) inside the zone of inhibition in a disk diffusion assay (Fig. 1), may be clinically relevant: tolerance is associated with persistent fungal infections, and high tolerance levels tend to predict fluconazole efficacy if antifungal treatment is initiated within 24 h of diagnosis [13, 15].

Figure 1. Distinguishing tolerance and resistance.

Example of disk diffusion assays (DDA) measuring fluconazole resistance and tolerance in two C. albicans strains after 24 h (top row) and 48 h of growth (bottom row). The zone of inhibition border is indicated in white and is determined using diskImageR, which identifies the region exhibiting ≥ 20% growth inhibition relative to the outer part of the plate [14]. Resistance is measured as the radius from the disk to the edge of this ≥ 20% inhibition zone after 24 h (RAD (pink line). Tolerance is measured after 48 h and quantifies the fraction of growth (pixel intensity within the zone of inhibition (FoG) relative to pixel intensity at distal parts of the plate). In this example, strain 1 and strain 2 have similar levels of resistance (RAD), but strain 1 has high tolerance, while strain 2 has very low tolerance (FoG)

Factors required for tolerance include VMA11, a component of vacuolar ATPase required for hemoglobin-iron utilization; the chaperone Hsp90 and its client, calcineurin; and the target of rapamycin (TOR) kinase [16, 17]. Adjuvant drugs are compounds that in combination with fluconazole (or another antifungal) clear tolerance while minimally affecting resistance. The effect of an adjuvant in combination with an antifungal on susceptibility also reveals mechanisms behind tolerance of that antifungal. In addition to inhibitors of Hsp90, calcineurin, or Vma11, tolerance is cleared by inhibitors of TOR, sphingolipid biosynthesis, the unfolded protein response, and indeed drugs without obvious fungal targets, such as fluphenazine or fluoxetine [13].

In C. glabrata and S. cerevisiae, loss of mitochondrial DNA is associated with increased drug resistance [18, 19]. C. albicans is canonically considered petite negative, meaning it cannot survive the loss of mitochondrial DNA. Nevertheless, there are C. albicans mutants that have respiratory defects, and they show a corresponding fluconazole sensitivity [20]. This suggests that mitochondrial function is connected to azole susceptibility in C. albicans and provides intriguing opportunities to understand which mitochondrial processes are required for azole resistance or tolerance.

The relationship between resistance and tolerance has been blurred in past literature, such that it is often difficult to clearly distinguish between the two processes. For example, altered stress response pathways have been listed as a mechanism of antifungal resistance [21] and may cause some resistance in addition to having a major effect on tolerance. Furthermore, there is clearly some overlap in the mechanisms behind resistance and tolerance, like in the upregulation of efflux pumps. In C. albicans, fluconazole tolerance does not correlate with MIC levels and appears to be regulated by a number of factors that do not affect fluconazole resistance [13, 14, 22]. Additionally, unlike resistance, tolerance is often detected in only a subset of the cell population. Thus, tolerance is largely distinct from resistance or susceptibility.

Tolerant cells might promote a higher chance of acquiring resistance mutations simply because they survive longer and thus undergo more mitotic divisions than non-tolerant cells. However, under the selective pressure of fluconazole, changes in tolerance levels also emerge without accompanying alterations in resistance [23]. Thus, tolerance is not simply a milder version of resistance.

In C. albicans, aneuploidy and shorter-range copy number variations can promote either tolerance or resistance, depending on the identity of the chromosome and the selective pressure being applied[24]. When loss of the aneuploid chromosome is accompanied by a return to the parental phenotype (prior to acquisition of the aneuploidy), it is likely that the aneuploidy is the cause of the improved growth in drug [25, 26]. However, aneuploidy is not required for antifungal tolerance; in most lab and clinical strains, tolerance levels differ despite most of the strains being euploid [13, 27]. Because inhibition of many different pathways clears tolerance to baseline levels, and because different clinical isolates usually differ by tens of thousands of SNPs and INDELS, we assume that strain-specific genetic backgrounds contribute to minor shifts in the many stress pathways that influence metabolism and antifungal tolerance [13, 27, 28]. Notably, while specific aneuploids can give rise to antifungal resistance, many resistant isolates are not ane-uploid. In contrast to the subpopulation effect of tolerance, aneuploidy-driven resistance affects many cells in a population similarly and appears to involve mutations that directly affect drug-target interactions (e.g., [29]).

The metabolic plasticity of C. albicans likely contributes to its success as an opportunistic pathogen. Gene expression affects metabolic flux, and both have downstream effects on the proteome [30, 31]. In this review, we highlight recent metabolomic studies that improve our understanding of how metabolism affects antifungal drug response in C. albicans and how antifungal drug exposure alters metabolomes. We focus on several components of metabolism in C. albicans: iron, a nutrient that is limited in niches colonized by C. albicans; zinc, a component of the fungal-specific zinc cluster proteins; phosphate, a macronutrient whose metabolism differs between fungi and humans; and sphingolipids, a class of lipids important for cell membrane functions as well as signaling. We note potential adjuvant therapies to the existing three classes of antifungals: azoles, which target ergosterol biosynthesis required for fungal membrane integrity; echinocandins, which inhibit the synthesis of β-glucan in the fungal cell wall; and polyenes, which disrupt cell membranes by interacting with sterols.

Metabolomic Monitoring of C. albicans: Fingerprinting the Culprit

Metabolism is built of anabolic (building up) and catabolic (breaking down) pathways that intersect and overlap, forming highly interconnected, complex networks. These networks operate within a cell and in the extracellular milieu, mediating nutrient sharing and signaling between cells in the population [31, 32]. Extracellular exchange can facilitate communication between species and even across kingdoms [33].

Metabolomics includes the large-scale study of metabolites within a cell (the intracellular metabolome) and the metabolite composition surrounding a cell (the extracellular metabolome, e.g., the space between cells of a biofilm). Mass spectroscopy coupled with gas chromatography (GC-MS) and liquid chromatography (LC-MS) is the most common approach to metabolite quantification; other techniques include nuclear magnetic resonance (NMR) and capillary electrophoresis–mass spectrometry (CE–MS) [34, 35]. Approximately 16,000 metabolites are listed in the Yeast Metabolome Database: more than 2600 predicted water-soluble metabolites and more than 13,000 predicted lipids [36]. Although the most comprehensive metabolomic methods can only identify between several hundred metabolites for a given sample, metabolomic data provide valuable snapshots of the cellular state, highlight the adaptability of C. albicans to different conditions, and reveal the biology behind drug resistance [32]. For example, comparative lipidomics of clinical isolates of C. albicans revealed connections between the mitochondria, cell wall integrity, and azole resistance [37].

Amino acids are one of the most informative classes of metabolites: they are sensitive reporters of metabolism within the mitochondrion and the cytoplasm and can indicate a metabolic response to stress [38]. For example, glutamate, cysteine, and glycine are required for the synthesis of glutathione, which is an antioxidant. Accordingly, depletion of these amino acids may indicate activation of the oxidative stress response [39]. Additionally, energy derived from amino acid catabolism induces and maintains hyphal growth in vitro and in response to mammalian immune cells such as macrophages [40]. Carbon sources affect resistance/suscepti-bility, and amino acids promote virulence, yet little is known about how specific amino acids contribute to antifungal drug resistance or tolerance [33, 41–45] (for a current review on amino acid metabolism in virulence, see [46]). Importantly, metabolomic studies in C. albicans reveal meaningful differences in the distribution of certain amino acids in response to drug exposure and as a function of susceptibility status.

C. albicans metabolomes have been compared across clinical isolates [47], including azole-resistant and azole-susceptible isolates derived from a common ancestor [48], with a range of growth conditions [49], and as a consequence of drug exposure, including fluconazole, micafungin, and amphotericin B [48, 50–52]. A comparison of the extra- and intracellular metabolomes of 49 Candida isolates collected from different human niches during routine checkups (C. albicans (61%), C. glabrata (20%), and C. parapsilosis (12%)) revealed that the strongest biofilm producers had the highest intracellular levels of glutamate and lysine [47]. Why these two amino acids are enriched has not been experimentally tested. Nonetheless, given that increased lysine uptake is an antioxidant strategy in S. cerevisiae that allows NADPH to be diverted to glutathione metabolism rather than to lysine biosynthesis [53], it is tempting to speculate that a similar mechanism could be involved in the establishment of C. albicans biofilms.

Trehalose accumulation promotes heat stress survival in S. cerevisiae [54, 55]. Interestingly, clinical isolates that are fluconazole-resistant (4 C. albicans, 1 C. parapsilosis, 2 C. tropicalis, and 10 C. glabrata strains) or caspofungin-resistant (1 C. dubliniensis and 5 C. parapsilosis) had high levels of intracellular trehalose relative to susceptible strains [47]. This suggests that drug-resistant isolates across different Candida species may survive drug stress by increasing their intracellular trehalose concentration and may indicate that trehalose accumulation provides cross-resistance to heat stress and drug exposure; supporting this notion, Hsp90, a chaperone protein induced by heat shock, is required for azole tolerance in C. albicans [13, 17]. Resistant isolates also showed increased levels of α-ketoglutarate, glutamate, glutamine, ornithine, and proline; biosynthesis of these molecules at least partially relies on mitochondrial function [56]. This suggests respiration may be upregulated in resistant strains, which supports prior work showing increased activity of mitochondrial complex I and V in two fluconazole-resistant strains [57]. Drug-resistant strains also had reduced ethanol levels, suggesting that, when they are exposed to antifungal drug stress, Candida species may prefer to generate ATP via mitochondrial respiration, rather than via glycolysis followed by fermentation [47].

Metabolomes are highly dynamic and respond to antifungal drug exposure. Fluconazole exposure caused major alterations in phospholipid metabolomics in both sensitive and resistant isolates [48], with fluconazole exposure increasing the pools of α-ketoglutarate [48, 50]. The mechanisms contributing to this pool are currently unknown. Nonetheless, both intra- and extracellular metabolomes were statistically different between susceptible and resistant strains even in the absence of fluconazole, implying that metabolic “fingerprints” are associated with a strain’s susceptibility status [47, 48]. Future studies will need to determine if metabolic fingerprints are associated with tolerant populations. If so, the development of an efficient assay that identifies resistant, susceptible, as well as tolerant isolates via metabolic fingerprinting would have advantages over existing rapid diagnostic tests of antifungal resistance alone.

Iron Affects Azole Activity and the Unfolded Protein Response

The essentiality of iron is linked to its mitochondrial function: iron-sulfur cluster formation is an essential function of mitochondria, including in mitochondria with respiratory defects. Iron availability is extremely limited in niches commonly inhabited or invaded by C. albicans. Reviews of how C. albicans and mammalian hosts compete for iron can be found elsewhere [58–62]. However, the role of iron in drug responses is less well-characterized. In early studies, iron deprivation increased azole susceptibility via (1) reduced expression of ERG11, which encodes lanostrol demethylase, the target of azole antifungals; (2) increased membrane fluidity; and (3) increased passive diffusion of drug into the cells [63, 64]. Iron starvation also increased azole susceptibility in vivo in a murine vaginitis model [65]. Surprisingly, calcineurin signaling, which is essential for normal resistance and tolerance to azoles and caspofungin [13, 66–68], is inhibited by iron starvation in an uncharacterized mechanism independent from the cell wall integrity pathway [69]. We posit that the relationship between calcineurin and iron starvation might be exploited to increase the efficacy of azole or echinocandin treatments.

Recently, connections between iron availability and unfolded protein response (UPR) functions were discovered. Ire1, the endoplasmic reticulum (ER) stress-sensor and an activator of the UPR, is essential for growth during iron starvation [70, 71]. Notably, deletion of IRE1 or of HAC1, its downstream target, increased azole and caspofungin susceptibility in C. albicans and in other pathogenic fungi [71–75]. Inactivation of the UPR also affects fluconazole tolerance: tunicamycin, which induces ER-stress, significantly reduces tolerance to fluconazole and renders fluconazole cidal [13]. Sterol production also influences ER activity; Ire1 clustering is abolished when the ergosterol biosynthetic pathway is genetically disrupted in Saccharomyces cerevisiae [76]. The mechanistic basis of interactions between sterol biosynthesis, iron starvation, and drug response remains to be uncovered.

Functional calcineurin signaling is required for ER stress responses (e.g., UPR) and is inhibited by iron starvation [77–79]. However, how iron starvation influences UPR activity is not known. Connections between iron, sterol biosynthesis, and ER activity abound: (1) Erg11 requires the iron-containing metalloporphyrin cofactor heme to function, and (2) the inhibition of ergosterol biosynthesis affects ER activity, and (3) fluconazole treatment rapidly causes the incorporation of free iron into structures such as heme [80]. Thus, it seems logical that fluconazole inhibits both ergosterol biosynthesis and iron availability which causes the iron starvation effects on ER function. Testing the hypothesis that iron availability, sterol biosynthesis, ER function, and calcineurin signaling converge to influence drug response via both resistance and tolerance will be an exciting topic for future work that could provide routes novel therapeutic strategies.

Zinc Modulates Cell Size and Growth Dynamics

Zinc-cofactor utilizing DNA-binding proteins make up ~ 50% of C. albicans transcription factors, and many of them regulate drug responses such as azole efflux (e.g., Tac1, Mrr1) and ergosterol biosynthesis (e.g., Upc2) [81]. Yet few studies have linked zinc starvation with antifungal drug resistance in Candida species [82]. Scattered evidence, primarily from S. cerevisiae, suggests that zinc is a potentially important player in drug responses. For example, in S. cer-evisiae, zinc depletion leads to ER stress, inducing the UPR, which is dependent on iron, as discussed above [83]. Yet, the effects of iron and zinc starvation elicit opposite effects on UPR activation: iron starvation increases susceptibility to azole treatment and prevents UPR activation, while zinc starvation has the opposite effect on UPR activity. UPR activation upon zinc starvation suggests a potentially antagonistic effect with azole treatment, but this remains to be tested.

Other observations relating to zinc starvation have indirect implications for drug responses. In C. albicans and several related species, zinc starvation induces the formation of very large “goliath” cells with roughly 5-fold increased volume [84]. In goliath cells, cell wall chitin exposure and chitin content are increased relative to cell size, a property linked to higher resistance to echinocandins [84, 85]. Fluconazole treatment leads to increased chitin content in C. albicans, C. krusei, C. tropicalis, and C. parapsilosis [86]. Interestingly, azole-resistant Candida auris have higher chitin levels relative to susceptible isolates, although it is unclear if increased chitin content contributes to the intrinsically elevated level of azole resistance of most C. auris isolates, as the relative contributions of multiple factors, including the amplification of several classic resistance genes, remain to be established [87]. Changes to cell wall composition and architecture and the resulting effect on drug response need to be investigated in more detail.

Zinc-starved goliath cells adhere to plastic surfaces with higher affinity than normal yeast cells [84]. This may facilitate the formation of biofilms, dense mixtures of yeast, and pseudohyphae and hyphae embedded in a network of extracellular matrix, because substrate adhesion is a prerequisite for biofilm formation. Furthermore, extracellular matrix components reduce the efficacy of antifungal drugs, in part by adsorbing drug and thus reducing drug access to the cells [88]. Biofilm cells also communicate via extracellular vesicles that carry extracellular matrix proteins that can partially cross-complement drug response defects in different mutants [89]. While the goliath phenotype has not been directly connected with altered drug responses, the effect of zinc starvation on surface adhesion and cell wall architecture suggests an antagonistic role for zinc starvation in azole responses, perhaps via the UPR activation.

The goliath phenotype was observed in multiple Candida species, including C. albicans, C. dubliniensis, and C. tropicalis, while goliath cell formation has not been seen in C. parapsilosis, C. lusitaniae, and Debaryomyces hansenii. Interestingly, unusually large cells were observed during zinc starvation of S. cerevisiae, where zinc starvation induced an intriguing asymmetric growth phenotype: mother cells increased in size and produced daughter cells that arrested in G1 [90]. It is unclear whether the enlarged, zinc-starved S. cerevisiae mother cells are equivalent to goliath cells, but the unequal distribution of cell contents between goliath mother cells and their daughters may benefit the large cell. Accordingly, unequal cell sizes generate a mixed population in which zinc-starved mother cells can divide while simultaneously limiting the number of actively growing and dividing daughter cells.

In some ways, such mixed populations of cells are reminiscent of antifungal drug tolerance, where a subpopulation of tolerant cells grows slowly, increasing their population size relative to those cells that do not grow in drug [13, 90]. A similar slow growth mode also was observed during iron and copper starvation, although the goliath phenotype is seen only with zinc starvation [84, 90]. It will be interesting to determine if zinc starvation-induced goliath cells produce G1 arrested small daughters, if goliath cells are also produced during azole drug exposure, and if zinc starvation promotes increased fluconazole tolerance.

ZAP1, which encodes a transcriptional activator of zinc transporters in C. glabrata, is repressed during fluconazole exposure, suggesting that fluconazole reduces zinc uptake [91, 92]. Yet, deleting ZAP1 resulted in increased fluconazole sensitivity in C. glabrata, but had no effect in S. cerevi-siae [91]. This lack of functional conservation reflects the previously mentioned lack of conservation of the goliath phenotype and suggests that zinc sensing and response are very diverse among yeasts. While the authors did not differentiate between resistance and other mechanisms of drug response such as tolerance and heteroresistance (in which a subpopulation of isogenic cells exhibits a higher MIC than the rest of the population) [91], the species-specific roles of Zap1 in azole responses remain to be understood. Clearly, there are more questions than answers about how zinc affects cell size, cell wall architecture, cell cycle progression, and population dynamics and how these affect cellular drug responses.

Targeting Phosphate Metabolism for Fungal-Specific TOR Inhibition

Phosphate, an essential nutrient central to nucleotide and membrane phospholipid production, is also the substrate of kinase and phosphatase signaling [93–96]. Yet, like iron and zinc, excess phosphate is associated with increased morbidity and mortality in humans and mice [97]. Mammals obtain phosphate from protein sources, while fungi have additional mechanisms for sensing and importing inorganic phosphate, including phosphate transceptors, which are membrane transporters that also have a receptor function and can be found throughout eukaryotes [98, 99].

Pho84 is a high-affinity phosphate transceptor that has been well-characterized in S. cerevisiae and, more recently, in C. albicans. C. albicans Pho84 is essential for cell wall integrity signaling and contributes to cell wall stress resistance [100], as well as phosphate-dependent growth [101], morphogenesis [101], and virulence [102]. Importantly, Pho84 has no human homolog. Pho84 may be a useful drug target because it simultaneously affects two biological components that differ between fungi and humans: phosphate sensing/transport and the fungal cell wall.

Inhibition of Pho84 with foscarnet or phosphonoacetic acid potentiates the antifungal activity of amphotericin B (a polyene) and micafungin (an echinocandin) in C. albicans [101]. Inhibition of Pho84 indirectly inhibits TORC1 (target of rapamycin complex I) [101], a central regulator of translation, amino acid metabolism, antifungal drug tolerance, and drug resistance [13, 103, 104]. Notably, AZD8055, a TOR inhibitor, reversed azole resistance in several Candida species. For example, posaconazole and AZD8055 synergized in 5 of 6 C. albicans and 9 of 10 C. auris isolates [103]. In a separate study, rapamycin, the classic TOR inhibitor, cleared tolerance (growth measured after 48 h) to baseline levels and rendered fluconazole fungicidal [13]. A remaining question is whether Pho84 inhibitors can synergize with azoles as adjuvant therapy; if they can, then Pho84 inhibitors have the potential to increase the efficacy of all three major classes of antifungal drugs, possibly without increasing host side effects.

Sphingolipids: Connecting Membranes and Signaling

Sphingolipids are broad-range bioactive signaling molecules and essential components of cell membranes [105]. Several genes encoding enzymes involved in sphingolipid biosynthesis have emerged from screens for azole-resistant or azole-tolerant strains [106]. In C. albicans, sphingolipid biosynthesis is governed by the mitochondrial retrograde pathway.

Mitochondrial retrograde (RTG) signaling is a stress response pathway found throughout eukaryotes. Various biological signals originate from the mitochondria as a product of metabolism—Ca²⁺, AMP/ATP, ROS, NADH/NAD+, and Krebs’ cycle metabolites like α-ketoglutarate, for example—which can then go on to affect nuclear gene expression [107]. Mitochondrial stress activates RTG signaling through the heterodimeric transcription factor Rtg1/3, which controls the expression of a subset of nuclear genes [108]. The precise mitochondrial stressors that activate RTG signaling differ among species, but in S. cerevisiae include glutamate starvation and TOR inhibition (in contrast, TOR inhibition in C. albicans leads to the degradation of Rtg1/3). Metabolomic analysis of C. albicans mutants lacking either rtglΔΔ or rtg31ΔΔ revealed high levels of sphingolipids and ceramides, the direct precursors of sphingolipids; glutamate was not affected [109]. In parallel, transcriptomic data showed that the RTG response in C. albicans negatively regulates PHO84, the phosphate transceptor described above, as well as three genes necessary for synthesizing sphingolipid precursors: ACC1, FAS2, and BIO2 [109]. Thus, there is a defined connection between mitochondria and sphingolipid homeostasis.

Sphingolipid homeostasis is crucial for C. albicans virulence [110]. Furthermore, C. albicans can gain azole resistance by altering sphingolipid composition [106]. It was also shown that adding phosphorylated phytosphingosine during miconazole treatment enhanced efflux activity—a classic method of drug resistance—by increasing the expression of efflux pump genes [111]. Loss of C. albicans RTG1 or RTG3 leads to increased calcium sensitivity, and in S. cerevisiae, pharmacological inhibition of sphingolipid synthesis leads to altered intracellular vesicle trafficking and endocytosis [105, 109, 110]. Taken together, this suggests that sphin-golipid homeostasis, much like ergosterol biosynthesis, is a critical component of metabolism in C. albicans.

Supporting this notion, myriocin, an inhibitor of sphin-gosine biosynthesis, and aureobasidin A, which inhibits the protein product of AUR1 (a gene with no human homolog), were tested against different C. albicans and C. glabrata strains, including clinical isolates resistant to fluconazole. Susceptibility tests found that all Candida isolates were sensitive to both inhibitors, and both inhibitors were synergistic with fluconazole [112]. Additionally, a screen of ~ 20,000 natural compounds and their derivatives identified NPD827 as being synergistic with fluconazole. NPD827 affects membrane homeostasis by causing the accumulation of long-chain sphingoid bases and the depletion of modified ceramides; additionally, it reduces multi-drug efflux and causes abnormal vacuolar morphology and activates the UPR in a calcineurin-dependent manner [113]. Thus, fungal-specific inhibition of sphingolipid synthesis has the potential to be a successful adjuvant therapy to azoles.

Conclusion

Our understanding of the interplay between metabolism and drug exposure is still in its infancy. By design, standardized antimicrobial susceptibility tests ignore slow growth; this is to reduce sensitivity to environmental conditions, despite environmental conditions clearly affecting susceptibility and tolerance. Thus, until recently, tolerant phenotypes have been ignored, and “resistance” and “tolerance” have been used interchangeably in much of the literature. This leaves a gap in our understanding of the biology behind persistent antifungal infections.

Niches colonized and infected by C. albicans vary considerably in temperature, pH, gas content, and the availability of a multitude of nutrients. Human blood glucose levels are quite low (0.05–0.1%), yet most in vitro experiments are conducted in rich or minimal media with 2% glucose. Critical nutrients that are limited in niche environments, like iron or zinc, are abundant in media used in most laboratories. Furthermore, the importance of specific amino acids in stress responses and drug exposure was recently discovered in S. cerevisiae [53, 114, 115], warranting a deeper evaluation of amino acid metabolism in the context of antifungal drug resistance and tolerance. The influence of such molecules discussed here is summarized in Fig. 2. To better understand why some treatments fail, it will be important to consider nuances in nutrient availability when designing experiments.

Figure 2.

Drug responses, including resistance and/or tolerance, are affected by a variety of mechanisms including biofilm formation, alteration of membrane lipid composition, and respiration, among others. The influence or potential influence of iron or zinc starvation, uptake of inorganic phosphate, or altered metabolomes on drug responses discussed in this review is summarized here

The relationships between metabolites, virulence factors, and drug susceptibility are only understood in broad strokes. A better understanding of these interactions and transactions will be needed to guide the improvement of therapeutic strategies. For example, metabolic profiles could be used as biomarkers to detect antifungal tolerance or resistance and, in parallel, could guide the choice of pharmacologic intervention. Currently, the identification of fungal species and antifungal resistance can be rapidly determined using matrix-assisted laser desorption ionization–time-of-flight mass spectrometry (MALDI-ToF MS) and its successor MS-AFST (MALDI-ToF MS antifungal susceptibility testing) [116–118]. However, MS-AFST does not measure tolerance, and fungal resistance is determined by measuring levels of gene expression coupled to MIC measurements. Yet there are no well-defined gene expression markers for tolerance, and MIC results have a minimal predictive value for therapy success. Developing diagnostics that use metabolism-based assays to predict drug response is an attractive idea, but first research is needed to determine if metabolite quantification alone or in combination with MIC testing is sensitive enough to predict drug responses.

While current research has made progress in characterizing the metabolic fingerprints of susceptible and resistant cells—i.e., increased levels of trehalose, α-ketoglutarate, and mitochondrially-linked amino acids in resistant isolates [47, 48]—it will be critical to recapitulate similar analyses with tolerant cells. Expanding our understanding of how the metabolism of tolerant cells differs from that of susceptible and resistant isolates has the potential to improve the clinical outcomes of patients suffering from systemic fungal infections and persistent candidiasis.