Antifungal drug screening: thinking outside the box to identify novel antifungal scaffolds

Abstract

Invasive fungal infections are responsible for a significant disease burden worldwide. Drugs to treat these infections are limited to only four unique classes, and despite these available treatments, mortality rates remain unacceptably high. In this review, we will discuss antifungal drug screening and how the approach to identifying novel compounds needs move away from traditional growth-based assays in order to meet the demand for new drugs. We highlight specific examples of creative screening strategies that increase the likelihood of identifying compounds with desired activities and provide perspective to inspire development of novel screens for the identification of first-in-class antifungals.

Introduction

Invasive fungal infections (IFIs) are a significant cause of morbidity and mortality worldwide, leading to an estimated one and a half million deaths per year [1]. In addition to the importance of invasive infections, the role of fungi and the mycobiome in the development of other diseases is becoming an area of interest [2]. Currently, there are four classes of antifungal drugs used clinically; of these, only three of which are effective as monotherapies [3]. The development of antifungal drugs has been slow compared that of other types of drugs. For example, from 2000 to 2015, 18 first-in-class drugs were approved for the use of solid tumor cancers [4]. In contrast, only a single novel class of antifungal drugs, the echinocandins, were introduced during this time; specifically, caspofungin was approved for invasive aspergillosis in 2001 [5]. The slow pace of antifungal drug development is due in part to a variety of factors: a shrinking interest of big pharma in antimicrobials [6,7]; the conservation of many biological pathways between human and fungi; and the difficulty and expense of doing properly powered clinical trials [3]. High mortality rates of IFIs, toxicity of available antifungal, and intrinsic and emergent drug resistance highlight the urgent need for new antifungal drugs.

The ‘golden age’ of antibacterial discovery has been followed by a fallow stretch characterized by low-yield screening efforts. An important reason for this lull in discovery is that many of the ‘low hanging fruit’-type compounds have been identified. As a result, application of the same growth assay-based screening strategies has led to the repeated rediscovery of the same classes compounds [8,9]. Antifungal drug development is coming to the same fate, with the same compound families and targets being identified repeatedly. Screening experiments have essentially two variables: library content and screening assay readout. Either of these variables could contribute to the ‘discovery bottle-neck’. Here we propose that moving beyond simple cell density/growth-based assays may improve our ability to identify new chemical matter within old chemical libraries.

Growth-based assays

The use of culture optical density (OD) as a readout of fungal cell growth is quick, cheap, and convenient for screening compounds against yeast and has been frequently used in drug discovery. However, these assays can be less sensitive than alternative methods and are not amenable to screening organisms with filamentous growth. Molds are a particular challenge for high throughput screening. Hyphal cultures are heterogenous, cannot be inoculated after germination, and present a high risk for contamination of equipment and facilities. OD measurements of filamentous cultures can be unreliable for two reasons. First, cells are not homogenously distributed within the well. Second, these cultures frequently form biofilms on the surface of the liquid. Because of these technical issues, traditional OD assays can only identify compounds that completely inhibit germination or growth of filamentous fungi and, thus, have poor sensitivity.

Alternative measures of cell growth can provide a more robust measurement of inhibition of filamentous cultures. The blue resazurin molecule is metabolically reduced to the pink, fluorescent resorufin and has been used to screen Aspergillus fumigatus [10] as well as the biofilm stage of Candida albicans [11–13]. Similarly, metabolism of the tetrazolium salt XTT has been used in screens for compounds active against C. albicans biofilms [14]. Quantification of total ATP in a sample has also been used withC. albicans biofilms [15,16], A. fumigatus, and the dematiaceous mold Exserohilum rostratum [17]. While these alternatives to OD provide methods to detect growth inhibition in a wider range of organisms or biological states, it is important to consider that the readouts can be altered by changes in metabolism that don’t necessarily reflect growth inhibition (Table 1). Ultimately, the desired effect of an antifungal is the inhibition of growth; however, measuring other cellular responses or coupling these growth assays with other approaches can provide more sensitive detection of molecules with antifungal activity.

Table 1.

Strengths and limitations of screening assays

Assay	Strengths	Limitations
Optical density	No additional reagents Uses standard plate reader Quick and convenient	Low sensitivity Not applicable to filamentous organisms Confounded by compound solubility and color
Metabolic reporters	Works with yeast and filamentous organisms More sensitive than OD	Does not measure growth or viability directly Can be altered by metabolic changes not only growth Narrow limit of detection
Reporter strains	Identifies compounds specific to a pathway of interest Often more sensitive than growth assays	Bias in screening will miss compounds with activity in other pathways Identified compounds may not have growth inhibitory properties
Adenylate kinase assay	Measures loss of viability directly Positive signal increases limit of detection Extremely sensitive	Does not detect fungistatic molecules Potential for false negatives due to adenylate kinase inhibitors
In vivo screening	Compound identification in context of infection Simultaneous counter screen for toxicity	Requires special equipment/facilities Laborious for large scale screens

Assays designed to target specific pathways in fungal cells

As a result of decades of research on the biology of pathogenic fungi, a wealth of knowledge about pathways required for growth and virulence is available. Designing assays to specifically detect molecules that interfere with these pathways is an effective approach to identifying mechanistic novel molecules, particularly if these assays are tailored to whole cell screening. For example, C. albicans encodes 115 glycosylphosphatidylinositol (GPI)-anchored proteins that are both critical to cell wall integrity and adhesion to host cells [18]. As such, many of the genes involved in GPI protein biosynthesis are essential and, thus, attractive as an antifungal drug target. To identify a first-in-class lead compound that inhibits GPI biosynthesis, a Saccharomyces cerevisiae reporter strain expressing a fusion protein of bacterial cephalosporinase to the GPI-anchored cell wall protein, Cwp2, was engineered [19]. Compounds were screened for those that reduced cephalosporinase activity in the cell wall fraction. This screen identified 1-[4-butylbenzyl]isoquinoline (BIQ) which was then optimized to the current clinical candidate APX001A. APX001A is a potent, broad spectrum antifungal with activity against yeast and mold, including difficult to treat organisms such as Scedosporium, Rhizopus [20,21], and the emerging drug resistant pathogen, Candida auris [22,23]. This compound is currently in phase 2 clinical trials for invasive candidiasis including C. auris (NCT03604705; NCT04148287). Notably, this screening methodology detected a 50% decrease in cephalosporinase activity at an 8-fold lower concentration than that which inhibits growth. This exemplifies how non-growth-based assays can identify novel compounds that would be missed in growth-based screens. The use of other reporter strains has been applied to identify molecules that interfere with critical pathways such as zinc and iron homeostasis [24,25] and the sterol regulatory transcription factor Upc2 [26]. These screening strategies bias compound identification to those that act on specific pathways and, thereby, reduce the likelihood of the ‘rediscovery’ phenomenon.

Distinguishing between fungicidal and fungistatic molecules

IFIs largely affect patients with immune deficiencies[1].This places a large dependence on drug treatment to clear infections without the assistance of immune cells available in normal hosts. Thus, fungicidal drugs which kill cells, rather than inhibit growth, are ideal for antifungals. In support of this, clinical studies of cryptococcal meningitis show that early clearance of organism from the cerebral spinal fluid using fungicidal drugs is associated with better patient outcomes [27–29]. Traditional growth-based assays do not distinguish between fungicidal and fungistatic molecules. To address this limitation in specificity, the adenylate kinase (AK) assay has been used to specifically identify fungicidal molecules in the high throughput settings. The AK assay measures the release of the ubiquitous cytosolic enzyme, AK, as a readout of cell lysis [30]. The AK assay is extremely sensitive and can detect cell lysis at eightfold lower concentrations of fungi-lytic drugs than traditional growth assays using OD [30]. Furthermore, the detection of a positive signal increases the linear range of the assay versus most growth assays which detect a reduction in signal. This sensitivity allows for detection of fungi-lytic molecules that would be missed by other growth assays; however, it will not detect compounds such as azoles, which do not induce cell lysis [30]. The AK assay identified the estrogen receptor antagonists, tamoxifen, and related compounds as promising antifungals [30]. While tamoxifen had previously identified antifungal activity against S. cerevisiae [31] and C. albicans [32], the identification of this drug in a screen to validate the AK assay reinvigorated interest in this drug [30]. Recently tamoxifen entered phase II clinical trial as an adjunct therapy with amphotericin B and fluconazole for cryptococcal meningitis (NCT03112031). Coupling the AK assay with secondary screens to identify cell wall active molecules led to the identification of a novel class of anti-cryptococcal molecules that block signaling through the cell wall integrity pathway [33,34].

Screening under physiologically relevant conditions

The use of whole animals in drug screening offers the unique ability to counter-screen for toxicity while screening in an actual infection setting. Antifungal screens have been implemented using the nematode, Caenorhabditis elegans [35,36]. In this model, worms are co-inoculated into plates with C. albicans, then wells are imaged with an automated imaging system, and the viability of the worms is assessed manually [36]. Notably, this screening technique can identify compounds that inhibit in vitro but not in vivo growth of C. albicans, potentially saving a lot of time following up on molecules that will fail in preclinical models of disease. The larval zebra fish, Danio rerio, has been well developed as a model system for many human fungal pathogens (reviewed in Ref. [37]). Although high throughput screens have not been performed for anti-infectives, large-scale compound screens have been performed on mutant fish to find therapeutics for specific genetic deficiencies [38,39]. Fungal infection models require a microinjection inoculation technique which is more laborious than co-culture; however, automated methods provide more potential for larger, more rapid screens [40].

While whole animal screens are not always feasible and present their own limitations (Table 1), in vitro screens can be designed to better represent in vivo conditions. Designing screens that target specific metabolic states is an example of emulating in vivo conditions in vitro. Using lactate as the sole carbon source can identify inhibitors of isocitrate lyase, an enzyme required for assimilation of C2 carbon sources and critical for pathogenesis [41,42]. Metabolic manipulations with the growth conditions have been successfully used to identify several antibiotics [43,44] and has the same potential for fungi if applied to larger scale screens. Another important aspect of an infection is the gas environment. The ability of Cryptococcus neoformans to adapt to physiological carbon dioxide correlates with the ability to cause disease in mice [45]. Similarly, low oxygen is an important characteristic of pulmonary aspergillosis [46], and the ability to adapt to low oxygen is correlated with disease severity in murine models of aspergillosis [47]. These conditions can be explored in drug screens to better represent the environment encountered at the site of infection and could identify compounds that would be missed in screens under standard lab growth conditions.

Repurposing FDA approved compounds

The screening of FDA-approved drug libraries for drug repurposing has become an attractive approach to drug development. These drugs have already passed initial safety screening and data on in vivo pharmacokinetics and pharmacodynamics are already available [48]. Repurposing screens have been performed against many major fungal pathogens, including C. albicans [49–54], C. neoformans [49,55], A. fumigatus [50,56], and the emerging multidrug resistant C. auris [14,50,57,58]. A screen against Aspergillus nidulans using the Johns Hopkins Clinical Compounds library identified the antidepressant, Sertraline, as having modest anti-Aspergillus activity [56]. Follow-up studies established more potent activity against C. neoformans [59] and a clinical trial was performed for cryptococcal meningitis with sertraline as an adjunctive therapy to Amphotericin B and fluconazole [60]. Although this trial failed to show a survival benefit with adjunctive sertraline, bringing an antifungal to clinical trial is a success that should not be overlooked. This example, and tamoxifen described above, highlight how repurposing can expedite the development of antifungals. However repurposing drug libraries are generally small and have been screened many times over, so choosing a library carefully can avoid wasting resources on screening the same set of compounds over and over again.

Combination screens and antifungal resistance

Screening for drugs which enhance the activity of current antifungals is another approach to improving antifungal therapy. Combination therapy offers the possibility of using lower doses, thus lower toxicity, of current antifungals, providing more broad spectrum coverage and shifting fungistatic activity to fungicidal activity [61]. Robbins et al. [62] screened six antifungal drugs at sub-inhibitory concentrations, in combination with ~3600 small molecules, against four fungal species. This study identified several small molecules, including FDA-approved drugs that increase the efficacy of antifungals, and even potentiate the activity of antifungals in resistant strains [62]. Performing combination screens can also be done using a resistant strain directly, to identify compounds that can be used to treat resistant organisms and to reduce the identification of cross-resistant molecules. A screen using echinocandin resistant C. albicans identified the metal chelator DTPA as a potentiator of caspofungin activity [63] and treatment with DPTA and caspofungin together significantly increased survival of mice inoculated with echinocandin resistant C. albicans compared to each drug alone.

Future perspectives

Increasing the diversity of antifungal screening assays will allow investigators to cover more chemical space and accordingly improve the chances that novel antifungal scaffolds will be discovered. While the idea of an ‘ideal’ broad-spectrum antifungal is attractive, the search for the perfect can hinder the identification of the good. Although growth-based screens are simple to perform, we must consider the old adage ‘you get what you screen for’. Repeating the same assay will yield the same results. To reuse the famous definition of insanity widely mis-attributed to Albert Einstein, “the definition of insanity is repeating the same thing again and expecting a different result”. It is time that antifungal drug discovery diversifies its approaches and methods.