INTRODUCTION

Top Previous Next References

In the 1990s, drug resistance has become an important problem in a variety of infectious diseases including human immunodeficiency virus (HIV) infection, tuberculosis, and other bacterial infections which have profound effects on human health. At the same time, there have been dramatic increases in the incidence of fungal infections, which are probably the result of alterations in immune status associated with the AIDS epidemic, cancer chemotherapy, and organ and bone marrow transplantation. The rise in the incidence fungal infections has exacerbated the need for the next generation of antifungal agents, since many of the currently available drugs have undesirable side effects, are ineffective against new or reemerging fungi, or lead to the rapid development of resistance. Although extremely rare 10 years ago, antifungal drug resistance is quickly becoming a major problem in certain populations, especially those infected with HIV, in whom drug resistance of the agent causing oropharyngeal candidiasis is a major problem (60, 203). For instance, 33% of late-stage AIDS patients in one study had drug-resistant strains of Candida albicans in their oral cavities (101). There are no large-scale surveys of the extent of antifungal drug resistance, which has prompted requests for an international epidemiological survey of this problem (133).

Since the initial studies of antifungal drug resistance in the early 1980s, we have accumulated a body of knowledge concerning the clinical, biochemical, and genetic aspects of this phenomenon. Recently, an elucidation of the molecular aspects of antifungal drug resistance has been initiated. This review will summarize the current knowledge of the clinical and cellular factors that contribute to antifungal drug resistance and will focus on the molecular mechanisms that have recently been described. The elucidation of the molecular mechanisms of drug resistance is still a work in progress. An attempt will be made to highlight research areas that have not yet been investigated.

This review will focus on the pathogenic yeast Candida albicans, since a large body of work on the factors and mechanisms associated with antifungal drug resistance in this organism has recently been reported. When appropriate, antifungal drug resistance in other common medically important fungi, such as Cryptococcus and Aspergillus, will be discussed. This review will not deal extensively with dermatophytes or with medically unimportant fungi. However, it is likely that many of the mechanisms described for C. albicans will be applicable to other fungi, especially since the currently available antifungal drugs are used for a variety of fungal infections.

With the exception of 5-flucytosine, the antifungal drugs in common usage are directed in some way against ergosterol, the major sterol of the fungal plasma membrane, which is analogous to cholesterol in mammalian cells. Ergosterol in the fungal membrane contributes to a variety of cellular functions. It is important for the fluidity and integrity of the membrane and for the proper function of many membrane-bound enzymes, including chitin synthetase, which is important for proper cell growth and division (77, 204). The mechanisms of action of antifungal drugs discussed below have been reviewed extensively (204).

The polyenes are a class of antifungal drugs that target membranes containing ergosterol. These drugs, which include amphotericin B and nystatin, are amphipathic, having both hydrophobic and hydrophilic sides. The drugs are thought to intercalate into membranes, forming a channel through which cellular components, especially potassium ions, leak and thereby destroying the proton gradient within the membrane (201). The specificity of the drugs for ergosterol-containing membranes is thought to be due to the interactions of the amphotericin B channel with ergosterol in the membrane, although the mechanism of this interaction is unknown. The specificity of amphotericin B for ergosterol-containing membranes may also be associated with phospholipid fatty acids and the ratio of sterol to phospholipids (201). The drugs are less likely to interact with membranes containing cholesterol. It has also been suggested that amphotericin B causes oxidative damage to the fungal plasma membrane (201, 203, 204). However, recent evidence suggests that amphotericin B has an antioxidant effect in vivo, protecting fungal cells against oxidative attack from the host (135).

Several inhibitors of the ergosterol biosynthetic pathway have been developed for use against medically important fungi, including allylamines and thiocarbamates, azoles, and morpholines (Table 1). All of these drugs interact with enzymes involved in the synthesis of ergosterol from squalene, which is produced from acetate through acetyl coenzyme A, hydroxymethylglutaryl coenzyme A, and mevalonate. Ergosterol is an important sterol for fungi, since it is the predominant or "bulk" sterol in fungal plasma membranes. In addition, it has an essential "sparking" function in which trace amounts of ergosterol are necessary for the cells to progress through the cell cycle. This sparking function is independent of the bulk sterol in the fungal membranes, since certain sterols can replace the bulk sterol of the membrane without supplying the sparking function for the cell cycle (67). The ergosterol biosynthesis inhibitors disrupt both the bulk sterol of fungal membranes and the sparking function in the cell cycle.

The allylamines (i.e., naftifine and terbinafine) and thiocarbamates (i.e., tolnaftate and tolciclate) inhibit the conversion of squalene to 2,3-oxidosqualene by the enzyme squalene epoxidase (44, 123, 163). This enzyme is the product of the ERG1 gene (gene designations are given in Table 2), which has recently been cloned in C. albicans (Table 1) (161). This class of drugs may inhibit the epoxidase through a naphthalene moiety common to both types of drugs (204). The drugs are noncompetitive inhibitors, and cells accumulate squalene in their presence.

The azoles, including both imidazoles (ketoconazole and miconazole) and triazoles (fluconazole, itraconazole, and voriconazole), are directed against lanosterol demethylase in the ergosterol pathway. This enzyme is a cytochrome P-450 enzyme containing a heme moiety in its active site (66, 199). The azoles act through an unhindered nitrogen, which binds to the iron atom of the heme, preventing the activation of oxygen which is necessary for the demethylation of lanosterol (77). In addition to the unhindered nitrogen, a second nitrogen in the azoles is thought to interact directly with the apoprotein of lanosterol demethylase. It is thought that the position of this second nitrogen in relation to the apoprotein may determine the specificity of different azole drugs for the enzyme (66, 199). At high concentrations, the azoles may also interact directly with lipids in the membranes (71, 77). Two other antifungal drug classes, the pyridines (buthiobate and pyrifenox) and the pyrimidines (triarimol and fenarimol), inhibit lanosterol demethylase and are used extensively as antifungal agents in agriculture but are not used in medicine (201). While many fungal species are sensitive to the azoles, Mucor spp. are intrinsically resistant to these drugs (112). Candida krusei and Aspergillus fumigatus are intrinsically resistant to fluconazole and ketoconazole, but both are usually sensitive to itraconazole (112) and voriconazole (118).

The morpholines (i.e., fenpropimorph and amorolfine) inhibit two enzymes in the ergosterol biosynthetic pathway, C-14 sterol reductase and C-8 sterol isomerase. The genes for these two enzymes, ERG24 and ERG2, have not yet been cloned from any medically important fungus, but have been cloned from S. cerevisiae. However, little is known of the interaction of the morpholines with ERG24 or ERG2.

5-Flucytosine (5-FC) has an entirely distinct mode of action from the azoles. 5-FC is taken up into the cell by a cytosine permease and deaminated into 5-fluorouracil (FU) by cytosine deaminase. 5-FC is fungus specific since mammalian cells have little or no cytosine deaminase (203, 204). FU is eventually converted by cellular pyrimidine-processing enzymes into 5-fluoro-dUMP (FdUMP), which is a specific inhibitor of thymidylate synthetase, an essential enzyme for DNA synthesis, and 5-fluoro-UTP (FUTP), which is incorporated into RNA, thus disrupting protein synthesis.

Drugs from the polyene class of antifungal agents, specifically amphotericin B, have long been considered the most effective of the systemically administered antifungal agents. The fungicidal antifungal drug amphotericin B has in vitro and in vivo activity against yeasts (Candida spp., Cryptococcus neoformans), molds (Aspergillus spp., Zygomycetes, dematiaceous fungi), and dimorphic fungi. Unfortunately, infusion-related toxicities, the frequent association of renal dysfunction, and the intravenous formulation of amphotericin B have limited the utility of this drug. Similarly, 5-FC, although generally active against Candida spp., C. neoformans, and some molds, has limited clinical utility owing to the frequent association of hematological toxicity and the rapid development of resistance, particularly when it is used as a single agent (discussed in more detail below) (17). The azole antifungal agents, because of their relative safety and ease of delivery, have subsequently become a critical component in the antifungal armamentarium.

Inconsistent levels in blood and high toxicities limited the utility of the first azoles, until ketoconazole was developed and used to treat chronic mucocutaneous candidiasis in the early 1980s (72). Fluconazole, found to be active against C. albicans in vitro and in vivo (162), has since become the drug of choice for the treatment of the most common AIDS-associated opportunistic infection, oropharyngeal candidiasis. Convenient administration and extended activity against non-Candida fungal infections (cryptococcosis, histoplasmosis, and coccidioidomycosis) have made azoles attractive as systemically administered drugs for the therapy and prophylaxis of AIDS-associated opportunistic infections (29, 51). Fluconazole is the preferred azole owing to a high oral bioavailability and good safety profile, although the less expensive ketoconazole is still frequently used for antifungal treatment in other countries (38). A recent study suggests that high doses of fluconazole might be effective therapy for mild blastomycosis in HIV-negative patients (137). Itraconazole and some of the newer azoles such as voriconazole have in vitro activity against various nonyeast fungal infections (molds, dimorphic fungi), but the clinical utility of the newer compounds is still being defined (17, 118).

Over the last several decades, aggressive anticancer therapy has resulted in another patient population that has become increasingly exposed to azole drugs for antifungal treatment and prophylaxis. An increased incidence of invasive Candida infections and attributable mortality have been documented in patients undergoing chemotherapy and bone marrow transplantation (193). Prophylactic fluconazole has been shown to decrease the incidence of local and invasive candidiasis in patients receiving chemotherapy and bone marrow transplantation, and today it is used in some centers for the prevention of candidiasis in neutropenic patients (59, 182, 214).

The role of azole antifungal agents in preventing and treating yeast infections in nonneutropenic patients has been expanding. Fluconazole has been shown to be effective presumptive therapy in surgical intensive care unit patients colonized with Candida spp. (143, 188) and as treatment for candidemia in nonneutropenic patients (156). In addition, azole antifungal agents have been found to be an effective single-dose therapy for vaginal candidiasis (183, 190).

Resistance to each of the antifungal agents has been reported in clinical isolates under specific conditions. However, the increased use of azoles in immunocompromised patients has had the largest effect on the frequency, morbidity, and response to therapy of fungal infections in the hospital and in the outpatient setting. Recently, advances in clinical microbiology have enabled us to detect and define azole resistance. The epidemiology of antifungal resistance, including the incidence, prevalence, and related host risk factors, will be discussed in more detail below.

Before the development of susceptibility testing of yeasts as outlined by the National Committee for Clinical Laboratory Standards (NCCLS), MIC determinations were inconsistent and varied up to 50,000-fold in different laboratories (49). NCCLS document M27, first published in 1992, has recently been revised (M27-A) and subsequent studies have demonstrated that the interlaboratory reproducibility of MIC determination approximates that of antibacterial testing (159). The macrodilution method has set the groundwork for the development of less cumbersome methods adapted for the clinical laboratory. These methods, relying on microdilution with or without colorimetric indicators or agar diffusion, have been shown to be reproducible and consistent with the standardized macrodilution method according to preliminary studies (141). One complication of the NCCLS protocol is that for certain isolates, the timing of the end point determination can have a major effect (up to 128-fold) on the MIC. A recent study suggests that determining the end point after 24 instead of 48 h ensures that the MICs for these isolates correlate with the in vivo response to azoles (157). In addition, there have been significant advances in susceptibility testing for filamentous fungi (34), as well as in vitro testing of amphotericin B susceptibility. Technical advances of antifungal susceptibility testing have been recently reviewed (141).

Historically, clinical resistance has been defined as persistence or progression of an infection despite appropriate antimicrobial therapy. A successful clinical response to antimicrobial therapy typically not only depends on the susceptibility of the pathogenic organism but also relies heavily on the host immune system, drug penetration and distribution, patient compliance, and absence of a protected or persistent focus of infection (e.g., a catheter or abscess). This is particularly true for fungal infections (Table 3).

The in vitro resistance of an isolate can be described as either primary or secondary. An organism that is resistant to a drug prior to exposure is described as having primary or intrinsic resistance. Secondary resistance develops in response to exposure to an antimicrobial agent. Both primary and secondary resistance to antifungal agents have been observed.

A correlation of in vitro susceptibility with in vivo response has been observed for mucosal candidal infections in HIV-infected patients. Many groups have noted that the clinical outcome is generally dependent on the in vitro susceptibility of the organism. One group performed fluconazole susceptibility testing on fungal isolates obtained from 87 HIV-infected patients. They observed that persistent oropharyngeal candidiasis correlated with infections caused by yeasts for which the MICs of fluconazole were high (24). Similar trends have been documented by a number of other investigators (9, 94, 108, 126). However, host variables become important when in vivo resistance is compared with in vitro MIC testing, especially in evaluating systemic infections (Table 3). The more immunocompromised the host, the less reliable is the correlation between in vivo resistance and MIC. The ability of C. albicans to form biofilms on surfaces (catheters, teeth, and endothelial cells) has been implicated as a cause of clinical "resistance" despite microbial susceptibility in vitro. One study demonstrated an increased resistance in the cells of the biofilm compared to that in cells growing in suspension (62). The formation of layers of cells may confer protection to organisms in the inner layers, leading to failure of antifungal therapy. Alternatively, infection caused by a resistant organism does not always predict clinical failure. A clinical response to fluconazole occurred in 11 of 13 patients with oral candidiasis even though they were infected with isolates for which the MICs were 32 or 64 µg/ml (155). The correlation between in vivo response and in vitro susceptibility has been reviewed in detail, with most of the data being derived from patients infected with HIV (53).

For aid in clinical interpretation of antifungal susceptibility testing, the NCCLS Subcommittee for Antifungal Susceptibility Testing recently established interpretive breakpoints for testing of fluconazole and itraconazole for Candida infections (158). The breakpoints for fluconazole MICs are as follows: <8 µg/ml, sensitive; 8 to 32 µg/ml, susceptible dose dependent; and 64 µg/ml, resistant. The breakpoints for itraconazole MICs are as follows: 0.125 µg/ml, sensitive; 0.25 to 0.5 µg/ml, susceptible dose dependent; and 1.0 µg/ml, resistant. These breakpoints do not apply for C. krusei, which is intrinsically resistant to most azole drugs. The intermediate levels of resistance are denoted susceptible dose dependent since they are usually treatable with high doses of drug (at least 400 mg of fluconazole per day or doses of itraconazole that maintain the levels in plasma at 0.5 µg/ml). Despite the advances in testing and interpretation, the ability of susceptibility testing to guide therapeutic decision making remains controversial. In response to this dilemma, experts in the field recently proposed recommendations for antifungal testing in the clinical laboratory (141). According to these authors, susceptibility testing is not routinely indicated for any fungus. They recommend testing of Candida spp. causing oropharyngeal candidiasis in patients with AIDS who are not responding to azole therapy and in some patients with invasive candidiasis (141).

The establishment of a reproducible method of susceptibility testing not only has resulted in advances in therapeutic decision making, but it also will enable further study of the epidemiology of antifungal resistance in yeasts by allowing comparisons between studies. Recent analyses addressing the prevalence, impact, and microbiological and host risk factors are discussed below and addressed in more detail in two recent reviews (3, 206).

Flucytosine resistance. Primary resistance to 5-FC is common in certain yeasts and molds. Non-albicans Candida spp., as well as Aspergillus spp., C. neoformans, and the dimorphic fungi, have high rates of 5-FC resistance (47, 178). In addition, secondary resistance is a common development, especially in patients receiving 5-FC monotherapy. There is some indication that the severity of immunosuppression and fungal burden may be important risk factors leading to the development of resistance (47). Because 5-FC resistance develops frequently, the drug should never be used as a single agent to treat either yeast or mold infections. Specific mechanisms of 5-FC resistance in Candida spp. and C. neoformans have been described and are detailed later in this review.

Azole resistance. As azole antifungal agents have become important in the treatment of mucosal candidiasis in AIDS patients, reports of resistance have increased. In fact, azole resistance has now been found in patients not infected with HIV and, in some situations, in patients not previously exposed to antifungal agents. Several factors contribute to clinical antifungal drug resistance (Table 3).

For each fungal species, there is a distribution of MICs of each antifungal drug. The average MIC of a particular drug can be determined for each species. By using this type of analysis, it is clear that C. krusei is intrinsically more resistant to azole drugs than is C. albicans, with most strains of C. krusei having high azole MICs. In addition, many strains of C. glabrata have azole MICs that are significantly higher than those for most strains of C. albicans (reviewed in references 133 and 160). Recently, Candida norvegensis and Candida inconspicua have also been described as intrinsically resistant species, although these species are uncommon human pathogens (5, 166).

These resistant species may be present by chance in a healthy individual as commensal organisms which can then become a problem when the individual becomes immunocompromised and is administered azole therapy. Alternately, infection with these species can be acquired from the environment or from other individuals. It is thought that in an immunocompetent individual, the endogenous species would prevent these new species from becoming established. However, in an immunocompromised patient, the selective pressure of antifungal drugs may allow the replacement of endogenous species with a more resistant superinfecting species.

The intrinsic resistance of C. krusei and some strains of C. glabrata to azole drugs has resulted in an increased frequency of these species in patient populations that depend on azole drugs for treatment or prophylaxis (see, e.g., references 133, 144, 226, and 227). It is likely that the azole drugs used by these patients suppress the growth of sensitive isolates such as those of C. albicans, and allow the growth of more resistant strains or species.

Several studies have investigated the frequency of strain replacement (one strain for another) in a single patient. In one such study (13), strain replacement was seen in the development of resistance in one of four AIDS patients (25%). Sensitive and resistant isolates were obtained from these patients over time, and the isolates were analyzed by pulsed-field gene electrophoresis (molecular karyotyping) and restriction fragment length polymorphism (RFLP). Three of the four patients maintained the same strain, while the sensitive and resistant isolates from the fourth patient were clearly different by RFLP but not by karyotyping. In a second study (120), sensitive and resistant isolates from seven patients were studied by RFLP and karyotyping. In all seven (0% replacement), the sensitive and resistant isolates were the same strain. In a third study (104), multiple sensitive and resistant isolates from three patients were studied by RFLP, karyotyping, and randomly amplified polymorphic DNA. Two of the three patients maintained the same strain (33% replacement). The fourth study used a mixed-linker PCR method to determine that strain replacement accompanied an increase in resistance in only one of five patients (20% replacement) (117). In the final study (142), karyotyping analysis was used to type sensitive and resistant isolates from 10 AIDS patients. Two patients (patients 6 and 20) exhibited sensitive and resistant isolates from the same strain at different times, while several other patients exhibited mixed infections of different strains with different sensitivities at several time points. It is difficult to determine if these mixed infections represent strain replacement over time or mixed infections in which the proportion of sensitive and resistant strains was altered by azole therapy. In one of the two patients from this study, a substrain was selected as resistance developed (225), indicating that the concept of strain replacement is dependent on the technique used to determine strain relatedness. In all of these studies, the sample size is small and so a generalization on the overall frequency of strain replacement is not possible. It is clear that development of resistance within the endogenous strain is much more common than strain replacement. In conclusion, there is a need for a large prospective survey of HIV-infected patients in which the development of resistance and strain identity are closely monitored. Such a study is in progress (139, 155).

The transfer of a strain from one person to another has been documented in several instances. In one case, a strain of C. albicans was transferred orally from one HIV-infected man with oral candidiasis to another (121). In a larger study of 19 sexual partners, at least five pairs had identical or highly related strains of C. albicans (18). In another study, C. albicans strains were identical in women with vaginal candidiasis and their male sexual partner in 8 of 10 cases studied (176). Transfer was not observed in sexual partners without symptoms of disease. None of these studies tested isolates for antifungal susceptibility. Still, the documented strain similarities suggest that transfer of strains can occur, at least in cases in which one of the individuals has disease symptoms.

Within a species, intrinsically resistant strains occur as part of the normal distribution of resistance levels (primary resistance). In addition, there is the possibility that a strain can be induced to be resistant by exposure to the drug over long periods (secondary resistance) (160). In these cases, it is hypothesized that the drug itself does not cause resistance but, rather, selects for growth of the more resistant cells in the population. Genetic mutation occurs by chance in all organisms at a low frequency (usually 10⁶ to 10⁸/gene). However, in a large population of yeast cells under selective drug pressure, specific random mutations that render the cell slightly more resistant will eventually become the dominant strain in the population. Unless the genetic mutations that render a cell resistant also reduce the overall fitness of the cell (133, 202), the mutant strain will persist even in the absence of the selective pressure of the drug.

Given the potential for the transfer of strains between individuals and the genetic stability of resistance, resistant isolates can quickly have a profound impact on the efficacy of the antifungal drug used for treatment, especially within a relatively isolated population with frequent interpersonal contacts, such as HIV-infected populations.

There are recent data suggesting that strains of Candida can become transiently resistant to a drug, a phenomenon called epigenetic resistance. That is, a cell can alter its phenotype, probably through transient gene expression, to become resistant in the presence of the drug, but the resistant phenotype can revert quickly to a susceptible phenotype once the drug pressure is eliminated. Recently, several researchers have demonstrated epigenetic resistance both in vitro (23) and in vivo (67, 203). Given the transient nature of epigenetic resistance, little is known about it at this time.

C. albicans has a variety of different cell types which vary in their susceptibility to azoles. C. albicans can be divided into two serotypes, A and B, based on carbohydrate surface markers. The B serotype strains of C. albicans are more sensitive to azoles (specifically ketoconazole) but are more resistant to 5-FC than are the A serotype strains (reviewed in references 132 and 201).

C. albicans is a dimorphic fungus. The yeast-like or blastospore cell types are spherical, budding cells which are associated with commensal growth of C. albicans. Under a variety of conditions, C. albicans can form long, slender hyphal projections, which are often associated with C. albicans pathogenicity (32). Azole drugs can interfere with hyphal production at 0.1 µM, 1/10 the therapeutic concentration of the drug in vivo (132). To date, there have been no studies monitoring the ability of resistant and sensitive clinical isolates of C. albicans to form hyphae. However, it is possible that a resistant strain with the ability to form hyphae even in the presence of azole drugs would be more pathogenic than a sensitive strain that is unable to form hyphae. This possibility needs further study.

In addition to the yeast/hypha transition that can occur in C. albicans, several species of Candida have the ability to alternate between several different forms, referred to as "switch phenotypes" (reviewed in references 185 and 187). These switch phenotypes represent different states of a yeast cell where each state expresses a unique set of genes, cellular structures and cell morphologies. These switch phenotypes are defined by their ability to manifest as different colony morphologies on agar plates. At least two different laboratories have described strains of C. albicans in which two different switch phenotypes exhibit a difference in azole susceptibility (50, 54).

Antifungal drug resistance is not a constant among the fungal cells in a population. The number of cells that are present in an infection or as a commensal growth will have an effect on antifungal drug resistance, since an increased number of cells will increase the probability that a mutation can occur that confers resistance. Recently, several researchers have observed that there is a great deal of variation in MICs between individual colonies of the same strain from a single patient (75, 179). These differences may reflect genomic instability or variation within a strain. Finally, it is likely that population "bottlenecks" can randomly select for resistance (or susceptibility). A population bottleneck is a set of conditions that randomly selects for a small number of cells from a population. That small subset may contain genetic alterations that will shift the characteristics of a population in a different direction. Such a population "bottleneck" might occur when azole therapy drastically reduces the number of fungal cells in oral candidiasis. In such an example, the azole drug may select for more resistant cells. However, the bottleneck may not necessarily involve drug selection. The same phenomenon might be seen with the removal and/or sterilization of dentures. It is possible that the small subset of fungal cells that survive on the denture or in the oral cavity is more (or less) genetically predisposed to drug resistance.

Many different types of mechanisms are known to contribute to a drug-resistant phenotype in eukaryotic cells (Table 4). The most frequent resistance mechanisms include reduction in the import of the drug into the cell; modification or degradation of the drug once it is inside the cell; changes in the interaction of the drug with the target enzyme (binding, activity); changes in other enzymes in the same enzymatic pathway; and an increased efflux of the drug from the cell. Current molecular analysis of antifungal drug resistance has focused on some of these areas, but has not adequately explored other possible mechanisms.

It is important in the analysis of the molecular mechanisms of resistance to use a matched set of isolates, that is, sensitive and resistant versions of the same strain, as determined by RFLP or karyotyping. This is required because C. albicans is mostly clonal; that is, it grows vegetatively without sexual reproduction. There is good evidence from Hardy-Weinberg equilibrium experiments (150) and balanced lethal experiments (220) that C. albicans almost never undergoes a sexual cycle. Enzyme analysis shows that C. albicans is not in Hardy-Weinberg equilibrium, which results from a randomly mating population. The disequilibrium in C. albicans indicates that mating is not a common event (150). Most strains of C. albicans contain a large number of balanced lethals (recessive null mutations) in which the cell relies on one functional copy of the gene for activity (220). These recessive lethal mutations could not persist in a population that mated frequently. The clonal nature of C. albicans makes it imperative that matched sets of sensitive and resistant versions of a single strain be characterized when determining the molecular mechanisms of resistance.

RFLP analysis with the repetitive marker Ca3 demonstrates a wide variety of RFLP patterns, suggesting that only rarely are clinical isolates related to each other (105, 177). Specific mutations in a target gene can be the result of allelic variation or selection for a specific resistance phenotype. The difference can be determined only with a matched set of sensitive and resistant isolates from the same strain. To date, few studies have used matched sets of isolates (114, 142, 153, 170, 171, 222, 223, 225).

Defects in drug import are a common mechanism of drug resistance. However, it is important to emphasize the distinction between the import of a drug into a cell and the gradual accumulation of the drug in the cell, which is the result of a balance between import into the cell and efflux of the drug from the cell. Analysis of drug import is difficult at best and requires that mechanisms of import be separated from efflux mechanisms. The standard method for studying drug import is to determine the amount of drug that associates with cells in extremely short periods (less than 10 s). This method has been used extensively in studying drug import in parasitic protozoans (65). The short incubation periods are achieved by mixing the cells and the drug above a layer of mineral oil. After the short incubations, the cells are pelleted through the mineral oil and the amount of drug in the cell pellet is determined. At this time, an analysis of drug import during short periods has not been performed for any medically important fungi.

Studies of fungi have usually used labeled drug to monitor the amount of label that accumulates within the cell over several minutes. These accumulation studies have led to the identification of several fungal efflux mechanisms (see below). One study with C. albicans suggested that accumulation of [³H]ketoconazole required glycolytically derived energy and was controlled by cell viability, environmental pH, and temperature (19). This study also showed that ketoconazole accumulation at low extracellular concentrations was saturable, implying a specific facilitator of import, while accumulation at high concentrations appeared to be by passive diffusion. Curiously, this study found no evidence for export, since the addition of unlabeled drug did not lower the concentration of labeled drug in the cells. Finally, data in this study suggested that other azole drugs and amphotericin B increased the accumulation of labeled ketoconazole in the cell (19). These results are not easily explained by the currently understood molecular mechanisms, which include an energy-dependent efflux pump for fluconazole (see below). The energy requirement for ketoconazole accumulation described in this study is inconsistent with the energy-requiring efflux mechanisms. Elimination of energy levels would inhibit energy-dependent efflux pumps, resulting in increased drug accumulation, the opposite of what is seen with ketoconazole. These results suggest that active transport may be involved in ketoconazole import. It is possible that the ketoconazole and fluconazole are imported by two different mechanisms. Further study is clearly needed.

Drug import may also be affected by the sterol composition of the plasma membrane. Several studies have demonstrated that when the ergosterol component of the membrane is eliminated or reduced in favor of other sterol components such as 14-methyl sterols, there are concomitant permeability changes in the plasma membrane and a lack of fluidity (204). These changes may lower the capacity of azole drugs to enter the cell.

It is important to note that alterations in drug processing (modification or degradation) are important drug resistance mechanisms in a variety of bacterial and eukaryotic systems (20). To date, little analysis of drug modification or degradation within a resistant cell has been performed for the medically important fungi. It has been mentioned that azoles are inert to metabolism by C. albicans (67), although no studies have analyzed the sensitivity of azole drugs to metabolism in resistant strains or non-albicans species.

Another common mechanism of drug resistance is modification of the target enzyme and/or other enzymes in the same biochemical pathway. For azole drugs, that pathway is the ergosterol biosynthetic pathway. Analysis of the sterols of a cell can provide a wealth of information concerning the alterations that have occurred in a resistant strain (Table 1). Modifications in the ergosterol pathway are likely to generate resistance not only to the drug to which the cells are exposed but also to related drugs. This cross-resistance will be discussed in a later section.

A defective lanosterol demethylase (the predominant azole target enzyme) will result in the accumulation of 14-methyl sterols, especially 14-methyl fecosterol and the diol 14-methyl-ergosta-8,24(28)-dien-3,6-diol (84). In azole-treated cells, the presence of 14-methyl sterols can modify the function and fluidity of the plasma membrane. In addition, it appears that cells with 14-methyl sterols have an increased sensitivity to oxygen-dependent microbicidal systems of the host (180). The diol that accumulates is known to cause growth arrest in Saccharomyces cerevisiae (84, 199) but is thought to be tolerated in C. albicans (11). However, recent studies suggest that accumulation of the diol is correlated with growth arrest even in C. albicans (86). The toxic effects of the diol in S. cerevisiae are eliminated by a mutation in ERG3, which encodes C-5 sterol desaturase (Table 1) (11, 84, 217).

Plasmid complementation of the ERG3 mutation in S. cerevisiae suggests that the ERG3 mutation alone can cause azole resistance (4, 84). Biochemical analysis suggests that ERG3 mutations are responsible for resistance in at least two clinical isolates of C. albicans (86). The sterol composition of Cryptococcus neoformans has been studied when cells are exposed to itraconazole (200). The sterol composition suggests that the drug affects both lanosterol demethylase and the C-4 sterol demethylase enzyme, 3-ketosteroid reductase.

Fungal cell extracts and labeled mevalonate can also be used to monitor the synthesis of sterol intermediates, including squalene, oxidosqualene, lanosterol, desmethyl sterols, and ergosterol (12, 115). The analysis of cell extracts from fungi has been important in monitoring overall sterol synthesis and the level of inhibition of lanosterol demethylase by azole drugs (see, e.g., references 98, 100, and 207). For example, cell extracts from a polyene- and azole-resistant strain of C. albicans (D10) did not contain any detectable lanosterol demethylase activity whereas the revertant strain, D10R, has lanosterol demethylase activity (102). In the absence of drug, strain D10 accumulates 14-methyl sterols similar to an azole-treated wild-type cell. Strain D10 is also defective in the formation of hyphae, while the revertant forms hyphae at normal rates. This finding suggests that hyphal formation, which is an important virulence factor (32), can be affected by changes in the ergosterol pathway. To date, the genetic mutation in strain D10 has not been identified.

Biochemical analysis of an unrelated azole-resistant strain of C. albicans, the Darlington strain, has shown that lanosterol demethylase activity in this strain does not correlate with resistance (68). Surprisingly, drug accumulation in the Darlington strain is increased compared to that in azole-sensitive strains, suggesting that drug efflux (see below) is altered with resistance (69), although one would expect to see increased efflux in a resistant strain. An analysis of the sterol components of this strain shows an accumulation of fecosterol, suggesting that there may be a defect in ERG2 (not ERG3 as previously reported). The ERG3 gene was recently cloned and expressed in the Darlington strain. Overexpression of the ERG3 gene restored ergosterol as the major sterol of the transformant but did not render the cell susceptible (122).

Cell extracts from fungi other than C. albicans have been used to show that the lanosterol demethylase has become less susceptible in resistant clinical isolates. These include Cryptococcus neoformans (97, 211), A. fumigatus (35), and H. capsulatum (218). The molecular mechanisms for resistance in these isolates were not described. Cell extracts have also been used to characterize a resistant form of lanosterol demethylase from the plant pathogen Ustilago maydis (79).

In one study, a point mutation in ERG11 was identified when an azole-resistant clinical isolate was compared with a sensitive isolate from a single strain of C. albicans (223). This point mutation results in the replacement of arginine with lysine at amino acid 467 of the ERG11 gene (abbreviated R467K, where R = Arg and K = Lys). The mutation is positioned near the cysteine which coordinates the fifth position of the iron atom in the heme cofactor. The mutation is thought to cause structural or functional alterations associated with the heme. Recent genetic manipulations suggest that R467K alone is sufficient to cause azole resistance (224). However, in the series of clinical isolates, several alterations occurred simultaneously (223), so it is impossible to determine how much R467K contributes to the overall azole resistance of the isolate.

Point mutations in ERG11 have been developed in laboratory strains that result in azole resistance. The point mutation T315A (the replacement of threonine [T] with alanine [A] at position 315) was constructed in the C. albicans ERG11 gene (100) based on the current understanding of the active site of the enzyme (reviewed in reference 77). The active site represents a pocket positioned on top of the heme cofactor. Substrates or inhibitors enter the active site through a channel that is accessible only with a shift in an -helix of the apoprotein. Mutations in the active-site pocket, in the channel, and/or in the mobile helix would be predicted to affect the function of the enzyme. The T315A mutation, which is situated in the active-site pocket, was studied in S. cerevisiae, which is more amenable to genetic manipulation. T315A causes a reduction in enzymatic activity and a reduction in azole binding to the active site, resulting in fluconazole resistance. Another point mutation has been identified in the S. cerevisiae ERG11 gene from a laboratory strain that was resistant to azole drugs (73, 228). This mutation, D310G (replacement of aspartic acid [D] with glycine [G] at position 310), is located in the active site of the enzyme in close proximity to the T315A mutation and renders the enzyme inactive. However, the azole resistance of this strain is likely due to an ERG3 suppressor rather than to an inactive ERG11 gene product (84).

A survey of resistant and sensitive clinical isolates has identified seven different point mutations that are associated with azole-resistant isolates (106). However, the matched sensitive isolate was not available from these resistant isolates, which could be used to determine if the point mutations were associated with resistance or just the result of allelic variation. ERG11 genes from a variety of fungal sources have been expressed and manipulated in S. cerevisiae (98, 100, 181) and in Escherichia coli (210). A recent study used functional expression in S. cerevisiae to identify and characterize five ERG11 point mutations from matched sets of sensitive and resistant isolates of C. albicans (168a).

Overexpression of ERG11 has been described in several different clinical isolates (2, 171, 222). In each case, the level of overexpression is not substantial (less than a factor of 5). It is difficult to assess the contribution of ERG11 overexpression to a resistant phenotype, since these limited cases of overexpression have always accompanied other alterations associated with resistance, including the R467K mutation, and overexpression of genes regulating efflux pumps (see below). Overexpression has not been extensively evaluated, so it is difficult to assess its importance. It should be noted that low-level overexpression has also been documented in strains of S. cerevisiae (26, 41, 80, 88). In each case, the effect of overexpression on antifungal susceptibility has been minimal. It remains to be seen if overexpression of ERG11 alone will result in azole resistance.

A common mechanism of resistance in eukaryotic cells is gene amplification. The increase in the number of gene copies usually results in an increase in expression. Overexpression of the ERG11 genes in C. albicans has not been associated with amplification of the ERG11 gene (171, 222). However, in a clinical isolate of C. glabrata, increased levels of lanosterol demethylase were associated with a resistant phenotype (202). These increased levels of the protein were associated with ERG11 gene amplification (113, 203) and correlated with an increase in ERG11 mRNA levels (201). In the same isolate, changes in drug efflux and in ERG7 activity were also observed (203). After 159 subcultures, the gene amplification, increased mRNA levels, enzyme levels, and drug efflux all reverted to normal. However, the revertant retained partial resistance to fluconazole (201). Recently, gene amplification of ERG11 in this isolate has been linked to a chromosome duplication, which results in overexpression of ERG11, as well as altered expression of a variety of other proteins (113).

Another genetic alteration associated with ERG11 has been described in the same clinical isolates in which the R467K mutation was described (223). Most, if not all, strains of C. albicans are diploid, having two alleles of each gene (132). Clinical isolates are usually clonal (105) and contain several sequence differences between the two copies of a gene (allelic differences). While analyzing the R467K mutation, it was observed that all of the allelic differences present in the sensitive isolate were eliminated in the resistant isolates containing R467K. These allelic differences were eliminated from the region of the ERG11 gene including the promoter, coding region, and terminator region, and into the THR1 gene immediately downstream of the ERG11 gene (223). The simplest explanation is that the allelic differences were eliminated by a gene conversion or mitotic recombination involving the ERG11 gene, although other possibilities (gene or chromosome deletion or mating) could not be completely discounted. This gene conversion or mitotic recombination resulted in a cell in which both copies of the ERG11 gene contain the R467K mutation. Genetic evidence suggests that a cell with two copies of the R467K mutation is significantly more resistant to azoles than a cell in which only one allele has the R467K mutation (224). Thus, gene conversion or mitotic recombination, both of which have been found previously in C. albicans, may be an important mechanism in the generation of a resistant phenotype in diploid cells like C. albicans. Gene conversion or mitotic recombination would not be important in haploid organisms such as C. glabrata.

Deletion of the ERG11 gene in C. glabrata has been studied for its effect on enhanced oxidative killing (81). Two ERG11 deletion strains which were fluconazole resistant were more susceptible to killing by H₂O₂ and by neutrophils than two fluconazole-sensitive strains with intact ERG11 genes. This increased killing (in the absence of drug) may explain why no clinical isolates of any Candida species have yet been identified with a totally nonfunctional ERG11 gene, despite alterations in drug susceptibility due to point mutations in C. albicans.

In addition to alterations in the lanosterol demethylase, a common mechanism of resistance is an alteration in other enzymes in the same biosynthetic pathway. The ergosterol biosynthetic pathway beginning with squalene is shown in Table 1. All of the genes listed have been cloned in S. cerevisiae (10, 83, 103). To date, ERG1, ERG2, ERG3, ERG4, ERG7, and ERG11 have been cloned from C. albicans (82, 89, 92, 96, 122, 174), ERG3 and ERG11 have been cloned from C. glabrata (22, 52), and ERG11 has been cloned from C. krusei (22) and C. tropicalis (26). In addition, PRD1, the NADPH-cytochrome P-450 reductase that is important for the function of ERG11, has been cloned from C. albicans (174). Except for the R467K point mutation in ERG11 of C. albicans, no point mutations in the ergosterol pathway have been described from a matched set of azole-resistant and azole-sensitive clinical isolates. Similarly, no gene amplification has been observed for the C. albicans genes to date (171, 222).

The ERG3 and ERG11 genes of C. glabrata have been studied in some detail (52). When the ERG3 gene was deleted from a strain of C. glabrata, the cells remained viable under both aerobic and anaerobic conditions, accumulated the sterol ergosta-7,22-dien-3-ol, remained susceptible to fluconazole and itraconazole, and became hypersensitive to amphotericin B. When the ERG11 gene was deleted, the cells were viable only under anaerobic conditions, accumulated lanosterol predominantly, became resistant to azoles, and had a reduced susceptibility to amphotericin B. The double mutant (deletion of both ERG3 and ERG11) was aerobically viable, accumulated 14-methylfecosterol and lanosterol, was resistant to azoles, and had a reduced susceptibility to amphotericin B. The behavior of these mutations for aerobic and anaerobic growth and for sterol accumulation is similar to that described for S. cerevisiae (198). However, the antifungal resistance patterns are not similar, suggesting that the pattern of sterol control and antifungal resistance differs between closely related yeast species (52). This suggests that analysis of the ergosterol biosynthetic pathway in one species may not be applicable to other yeasts.

This analysis has also demonstrated that the ERG3 mRNA level was elevated in ERG11 deletions and that the ERG11 mRNA level was elevated in ERG3 deletions. This suggests that transcriptional control of at least two genes occurred in the ergosterol biosynthetic pathway, perhaps by negative feedback from ergosterol. The lack of ergosterol in the deletion strains may have induced the transcription of ERG3 and/or ERG11. This may be similar to transcriptional control of ERG genes in S. cerevisiae, where ERG11 is regulated by the carbon source, heme, and O₂ (aerobic and anaerobic growth) (198).

Sterol analysis of C. albicans clinical isolates has suggested that alterations in ERG3 may be a major cause of azole resistance (85, 86). ERG3 defects prevent the production of the diol that would cause growth arrest (see above). ERG3 defects have been suggested to be a major source of azole resistance in a number of fungi, including the plant pathogen U. maydis (79).

In Cryptococcus neoformans, the azoles appear to directly or indirectly block the C-4 sterol demethylase enzyme complex, which prevents the production of the diol without an ERG3 defect (76). Sterol analysis of fluconazole resistance in C. neoformans suggested that resistant strains may have defects in ERG2 or in ERG3 (211); these strains appear to be cross-resistant to amphotericin B.

In addition to its interactions with lanosterol demethylase, itraconazole interacts with ERG24 in C. neoformans and H. capsulatum, which explains the increased potency of itraconazole against these species (203).

Recently, the ERG5 gene was cloned from S. cerevisiae and found to be another cytochrome P-450 enzyme (83). The azole drugs bind with similar affinity to this enzyme, suggesting that both ERG5 and ERG11 may be responsible for azole tolerance (Table 1).

Reduced accumulation of radiolabeled fluconazole has been documented for a variety of resistant clinical isolates of C. albicans (2, 99, 171, 209). Accumulation in sensitive isolates was shown to be temperature dependent and was not affected by energy inhibitors such as sodium azide or 2-deoxyglucose. In azole-resistant isolates, including a laboratory-induced isolate, fluconazole accumulation was reduced. Intracellular levels of the drug in resistant isolates were then increased by the addition of sodium azide, suggesting that an energy-dependent efflux pump was associated with resistance (2, 171).

Accumulation of fluconazole was studied by using [³H]fluconazole in matched sets of azole-sensitive and azole-resistant isolates of C. glabrata (70, 138, 202). Azole-resistant isolates accumulated significantly less fluconazole than did azole-sensitive isolates from the same strain. Only small changes in ergosterol biosynthesis were detected in the resistant isolates. Metabolic or respiratory inhibitors increased the fluconazole accumulation of azole-resistant isolates, suggesting that energy-dependent efflux pumps were involved. Efflux pump modulators were tested for their effects on fluconazole accumulation. Only benomyl competed with fluconazole under these conditions (138) (benomyl resistance is discussed below).

Reduced drug accumulation has been associated with intrinsic azole resistance to fluconazole, itraconazole, and ketoconazole in C. krusei (111). An itraconazole-resistant isolate of C. krusei did not exhibit any alteration in ergosterol biosynthesis but did show a reduced accumulation of drug compared to a laboratory strain (208). However, a matching sensitive version of the same strain was not available.

Rhodamine 123 (Rh123) is a substrate for a variety of efflux pumps in many systems and has been used to study accumulation in fungal cells. Azole-resistant strains of C. albicans, C. glabrata, and C. krusei accumulate less Rh123 than do azole-sensitive strains (28). Rh123 accumulation is dependent on the growth phase and temperature. Proton uncouplers and the efflux pump modulator reserpine increase accumulation, probably by decreasing the activity of the efflux pumps. In C. glabrata, Rh123 and fluconazole accumulation is competitive, suggesting a common efflux pump, while in C. albicans, the two drugs do not complete (28).

Reduced accumulation has also been suggested as a mechanism of resistance for a C. neoformans laboratory mutant that is resistant to both azoles and polyenes (76). The strain had no detectable change in its sterol pattern. No direct measure of drug accumulation was reported.

For the last several years, there were no reports of itraconazole resistance that were not the result of cross-resistance to fluconazole (201). However, recently, itraconazole resistance in three strains of A. fumigatus was described. One of the strains had a reduced cellular accumulation of labeled itraconazole, suggesting that resistance in this strain is mediated by an efflux pump (35).

Drug accumulation has also been studied for terbinafine in Trichophyton rubrum. Accumulation is 10-fold higher in T. rubrum than in C. albicans, which may explain the high sensitivity of T. rubrum to terbinafine (44).

Eukaryotic cells contain two types of efflux pumps that are known to contribute to drug resistance: ATP binding cassette (ABC) transporters (ABCT) and major facilitators (MF) (110, 119). Both types of pumps are known to cause drug resistance in other systems. The ABCT are frequently associated with the active efflux of molecules that are toxic to cells and are relatively hydrophobic or lipophilic, as is the case with most azole drugs. The MF have not been studied as extensively as the ABCT but are also associated with relatively hydrophobic molecules such as tetracycline (110).

The ABCT are composed of four protein domains: two membrane-spanning domains (MSD), each consisting of six or seven transmembrane-spanning segments, and two nucleotide binding domains (NBD) (119). The NBD of ABCT bind ATP through an ABC that consists of several conserved peptide motifs, including two Walker domains, a Signature domain and a Center domain. The ATP that is bound to the ABC is used as a source of energy for the ABCT, although the mechanism by which the ATP energy causes transport of the substrate molecule is unknown.

The MF do not contain NBD. They are composed primarily of 12 to 14 transmembrane segments (110, 140). The MF use the proton motive force of the membrane (gradient of H⁺ across the membrane) as a source of energy. In general, the MF work by antiport; that is, protons are pumped into the cell and substrate molecules are pumped out.

Recently, the complete sequencing of the genome of S. cerevisiae has allowed a determination of the total number of ABCT and MF in the genome of a eukaryotic cell. Thirty ABCT genes that contain an ABC were identified (194). Twenty-three of these genes appear to encode transmembrane segments, which suggests that they could function as drug pumps and are closely related to known ABCT drug pumps. Twenty-eight MF genes were also identified in the S. cerevisiae genome by homology to the five known MF genes of fungi (57). This suggests that S. cerevisiae contains at least 51 different genes encoding efflux pumps that could contribute to the drug resistance levels of the cell. There is good reason to believe that Candida and other fungi would have a similar number of efflux pumps in their genomes (see below).