INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The management of disseminated Candida lusitaniae infection is complicated because isolates are frequently resistant to amphotericin B (AmB). Moreover, many Candida lusitaniae isolates seemingly develop AmB resistance in vivo during therapy (21, 22, 41). The ability to transform from AmB susceptibility to resistance as well as the expression of phenotypic switching has not been previously described in C. lusitaniae isolates. Phenotypic or colony morphology switching in Candida albicans is, however, a well-described phenomenon. It involves high-frequency switching of either individual colonies or cells from one morphology or phenotypic state to another (46, 47). Switching may be a nonsexual mechanism which certain Candida species use to introduce variability to cells. Phenotypic changes appear to increase the chance that a daughter cell can survive immunological or other microenvironmental host defenses (60, 61).

There are several switching "systems" in C. albicans; all are believed to have an overlapping mechanism (57). Switching is characterized as reversible, pleiotropic, of high frequency (about one switch per 10⁴ cells), occurring at different rates in different clinical isolates, and inducible (up to 200-fold) by low doses of UV light, temperature, or other environmental stimuli (23, 39, 54, 55, 59). At this rate, any individual colony of more than 10⁶ cells will be heterogeneous. Switching systems have been studied primarily in C. albicans but are also known to exist in other Candida species, such as C. tropicalis (55, 56).

There is no evidence that gross chromosomal alterations are responsible for switching in C. albicans (56). Instead, recent molecular genetic analysis suggests that there are a number of genes that are transcribed in a colony-specific manner and are activated by type-specific transcriptional activators that act in a combinatorial fashion at conserved sites upstream of the affected genes. At least one gene is regulated by both yeast-to-hyphal phase transition and white-to-opaque switching (11, 56, 57).

In vitro identification of spontaneous mutants resistant to AmB in C. albicans is extremely rare. Similarly, clinical Candida isolates that have high-level resistance to AmB are not common (69). Despite this, there are numerous examples of clinical failure during AmB therapy which are thought to reflect drug failure only. AmB resistance is, however, described more frequently among the non-albicans Candida species than among C. albicans. In particular, C. lusitaniae has frequently been reported to be resistant to AmB (8, 21, 22, 38, 41, 43, 44).

In the present study, we characterized a new form of switching discovered in clinical C. lusitaniae isolates. The phenotypic switch, which occurs at a relatively high frequency, is from an AmB-susceptible phenotype to an AmB-resistant phenotype. In addition, the reverse switch from AmB resistance to susceptibility also occurs but at a lower frequency. The ability of susceptibility phenotypes to switch in C. lusitaniae, but not C. albicans, isolates is consistent with the description of clinical failure and in vitro resistance developing during AmB administration in C. lusitaniae-infected patients.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Candida strains. C. lusitaniae isolates were obtained from either the American Type Culture Collection (ATCC) or from hospitalized patients in Harper Hospital, Detroit, Mich., or from William Beaumont Hospital, Royal Oak, Mich. These isolates were recovered from multiple anatomic sites and at separate times over a 1-year period in individual patients. All isolates were identified by germ tube production and chlamydospore formation on cornmeal agar, the yeast API 20C method (Sherwood Medical, Plainview, N.Y.). In addition, yeast isolates were also verified by CHROMagar Candida plates (40, 42), randomly amplified polymorphic DNA (RAPD) fingerprinting (64), and electrophoretic karyotyping (68, 70). ATCC isolates were used as reference standards for the various Candida species.

Determination of in vitro susceptibility. AmB as a lyophilized cake of AmB and sodium deoxycholate was purchased from Gensia Laboratories (Irvine, Calif.). This was suspended in sterile water at 1 mg/ml and stored frozen in light-protected vials. In later experiments, AmB was purchased from Sigma (St. Louis, Mo.) and dissolved at 1 mg/ml in dimethyl sulfoxide.

UV light treatment. UV induction of switching was a modification of the method of Morrow et al. (39). One to ten 2- to 3 day-old colonies, grown on SAB agar, were resuspended in 30 ml of sterile distilled water, transferred into a 100-mm-diameter petri dish, and exposed with slow stirring to a ChromatoView model CC-20 short-UV light source (UV Products, Inc., San Gabriel, Calif.). Samples were withdrawn at intervals to determine the exposure time required to achieve 10% survival, typically 25 s. Afterwards, survivors were scored as CFU by plating serial dilutions onto SAB agar plates. the individual UV doses for each Candida species were determined separately.

Zymolyase incubations. Candida cultures were initiated from single colonies in SD broth (1 ml), grown overnight at 35°C with shaking, and the harvested in 1.5-ml centrifuge tubes, washed once with 1 ml of 1 M sorbitol-10 mM (Tris (pH 7.5), and resuspended in 400 µl of the same buffer at 2 × 10⁷ to 2 × 10⁷ cells/ml. This volume was split into duplicate tubes, 200 µl each. Zymolyase (1.5 U) was added to one of the duplicates, and both tubes were incubated at 35°C for 5 h. To determine viability, samples from both tubes were serially diluted into sterile distilled water before and after incubation. Afterwards, 5 µl of each dilution was plated on SD agar without AmB. Aliquots were also plated on SD agar with 0.25 and 0.35 µg of AmB per ml to determine the numbers of resistant cells with or without prior exposure to zymolayse.

Processing of human blood. Ten-milliliter samples of venous blood were drawn from healthy volunteers in Vacutainers with either EDTA or heparin as the anticoagulant. Blood samples were stored on ice or at 4°C and used immediately or within hours of sampling. Yeast cultures were stored at 4°C. Serum was prepared from some of the blood samples by centrifugation at 7,000 × g for 5 min.

RAPD fingerprinting. Individual colony lysates were prepared as previously described by using the toothpick method (64). Briefly, 1 µl of lysate was amplified in a Robocycler thermocyler (Stratagene, La Jolla, Calif.) with the following program: 5 cycles of 94°C for 30 s, 25°C for 2 min, and 72°C for 2 min; 45 cycles of 94°C for 30 s, 30°C for 30 s, and 72°C for 2 min; 45 cycles of 94°C for 30 s, 30°C for 30 s, and 72°C for 2 min; and 1 cycle of 72°C for 10 min. Typical reaction mixtures, in 10- to 25-µl volumes overlaid with mineral oil, consisted of 25 mM Tris-Hcl (pH 8.4), 50 mM KCl, 1.5 mM MgCl₂, 100 µM deoxynucleoside triphosphates, 50 pmol of 10-mer primer, and 2.5 U of Taq DNA polymerase, as provided by Bethesda Research Laboratories (Bethesda, Md.). Reaction products were analyzed by ethidium bromide agarose gel electrophoresis (1.5% agarose or 1% Sepharide; Bethesda Research Laboratories) photographed, and scanned (Logitech Scanman II).

Primers. The following sequences were used as primers at 50 pmol per 25-µl reaction volume: CX5, 5'ACACTGCTTC-3' (64); REP, 5'-GAGGGTGGCGGTTCT-3' (6, 53, 66); AP3, 5'-TCACGATGA-3' (73); and PST, 5'-CAGTTCTGCAG-3'.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Isolation and characterization of AmB-resistant forms of C. lusitaniae 7227. As an introduction to a description of switching of C. lusitaniae isolates from AmB susceptibility to resistance, we first describe assay systems used to define these phenotypes. A clonal derivative of clinical isolate 7227, called 7227R, was characterized as being less susceptible to AmB in one assay, the EAA (EEA concentration [minimal concentration of AmB which resulted in fewer than 5 colonies after incubation], >0.15 ug/ml), than the original clinical isolate. Isolate 7227R was recovered from 7227 after subculturing a single colony on SAB agar supplemented with 0.15 µg of AmB per ml. Subculturing of a single colony of 7227R resulted in the recovery of isolate 7227S, which was found to be susceptible to AmB (EEA concentration, <0.05 µg/ml). We used isolate 7227R6 (R = resistant; 6 = colony no. 6) and 7227S11 (S = susceptible; 11 = colony no. 11) for the majority of our experiments. Both isolates were derived from the parent C. lusitaniae 7227, a clinical isolate, as described previously.

Demonstration of switching of AmB susceptibility among cells from single colonies. The susceptible isolate 7227S11 was derived from a resistant strain, 7227R6, as previously described. Repeated attempts to recover more susceptible isolates similar to 7227S11 from approximately 100 clonal derivatives of 7227R6 failed. However, new AmB-resistant variants derived from 7227S11 were recovered at relatively high frequencies, ranging from 1 in 10 to 1 in 10⁴ cells in individual colonies. Thirty colonies derived from 7227S, designated strains 7227S1 to 7227S30, were plated at a density of approximately 10⁶ cells per colony. From these strains, at least 100 colonies were recovered per plate supplemented with 0.15 µg of AmB per ml. Approximately 1 in 30 of these colonies was classified as AmB resistant because the isolates were able to generate the same number of colonies upon subculturing on medium with or without AmB. These strains were subsequently designed 7227SR, followed by a number to indicate the clonal origin. Differential plating of these variants is depicted in Fig. 1. Four clonal derivatives of isolate 7227 (7227R6, 7227S11, 7227RS9, and 7227SR5) all grew with the same efficiency on drug-free SAB agar (Fig. 1, lower plates). However, two resistant derivatives (7227R6 and 7227SR5) were able to grow with the same efficiency on agar plates with or without AmB. In contrast, two susceptible derivatives (7227S11 and 7227RS9) formed few to no colonies on SAB agar plates supplemented with AmB. Thus, the important observation was that a small percentage of cells derived from most susceptible C. lusitaniae colonies grew to colony size on AmB-supplemented plates.

View larger version (81K): [in this window] [in a new window]   FIG. 1.   Reversible phenotypic switching of C. lusitaniae 7227. Single colonies were diluted and subcultured on SD agar without AmB (AmB) or with AmB at 0.15 µg/ml (+AmB). Representative plates inoculated with either 0.2 or 100 µl of the initial cell suspension are shown. The isolates are labeled as follows: designations that include "S" indicates isolates that are derived from a susceptible lineage; "R" indicates a resistant lineage; "SR" indicates isolates that switched from a susceptible to a resistant phenotype; and "RS" indicates isolates that switched from a resistant to a susceptible phenotype. Numbers following these labels indicate clonal lineages.

UV light induction of reversion in isolate 7227SR9. To avoid searching among 10⁴ colonies, we evaluated whether the switching frequency could be enhanced by exposure to UV light (39). Exposure of a single 7227SR5 colony to a UV light dose producing 10% survival of cells resulted in two revertant 7227RS colonies among the 60 tested colonies, a rate of 3 × 10². If we assume that the spontaneous rate of this reversion was the same as the switch from susceptible to resistant, i.e., 10⁴, then the UV treatment resulted in a 300-fold enhancement of the switching frequency. This assumption is reasonable, given that no susceptible colonies were detected among 1 × 10 to 2 × 10⁴ untreated switching colonies. A subculture of 7227RS9 cells on SD medium with 0.1 µm zinc (28), which is known to induce morphological switching of C. albicans (2, 5), had no effect on C. lusitaniae switching to an AmB-resistant phenotype.

Evaluation of the switching frequency from AmB susceptibility to resistance induced by heat shock and whole blood exposure. After demonstrating that UV light induced AmB switching, we investigated whether any other physiological signals which may be encountered during a systemic fungal infection have the capability to induce switching. As shown in Fig. 2, a 30-min heat shock increased the number of resistant colonies by about 10-fold (Fig. 2, right plates). Exposing the same cell lineage to whole human blood increased resistant daughter colonies by about 100-fold (Fig. 2, left plates). In repeat experiments, the level of induction by blood was approximately 20-fold. Prolonged heat shocks of up to 5 h were required for optimal conversion in the white-to-opaque transition of C. albicans (49). In contrast, heat shock exposures longer than 30 min were suboptimal for C. lusitaniae switching. Blood-induced switching was similarly reproducible for C. lusitaniae 7227 but was ineffective in inducing switching in C. albicans B311 or C. guilliermondii 7210. Switching induction by exposure to whole blood did not require a 37°C heat shock and was not achieved if the blood was stored at 4°C for more than 3 days or if serum was substituted for whole blood (data not shown).

View larger version (92K): [in this window] [in a new window]   FIG. 2.   Induction of C. lusitaniae 7227RS9 switching by exposure to whole blood (left plates) and heat shock (right plates). A single colony was suspended and distributed to SD or fresh whole blood, and aliquots (0.2 or 20 µl) were plated on SD with or without AmB as indicated. After 30 min of incubation at 37°C, larger aliquots were plated the same way.

AmB-resistant and -susceptible isolates were clonal derivatives. Several approaches were used to eliminate the possibility that resistant derivatives of susceptible lineages were random contaminants of a resistant species from the laboratory environment. First, rare laboratory contamination cannot explain the repeated, independent isolations of both resistant and susceptible derivatives from single, independent colonies of 7227S11 and 7227SR5. Second, all colonies were identical in colony morphology and color on CHROMagar Candida plates. Third, each colony was analyzed by colony filter hybridization, and all hybridized with a radioactive probe specific for C. lusitaniae (64a). Negative hybridization controls included colonies of 10 other Candida species. These data rule out the interpretation that the different phenotypes were due to colonies of other Candida species, notably C. krusei, which have different susceptibilities to AmB (44). Finally, all isolate 7227 derivatives were identical or nearly identical by RAPD fingerprinting. Individual colony lysates of the isolates were amplified by using different primers. Primer CX5 generates a single 2-kb product characteristic of C. lusitaniae (64). All 7227 lineage isolates generate this product, regardless of resistance phenotype. Primer REP hybridizes to a repetitive sequence element in Candida and so generates a large number of products (66). Neither REP nor any of 10 independent 10-mer primers generated different fingerprints for susceptible versus resistant lineages.

Comparison of AmB-resistant and -susceptible derivatives of isolate 7227. Resistant strain 7227SR5 and switched susceptible strain 7227RS9 were compared to their parents for susceptibility to antifungals and other growth conditions. 7227SR5 was as resistant to AmB as parent 7227R, and both of these were more resistant than 7227S or 7227RS9. All forms of isolate 7227 were equally susceptible to fluconazole, clotrimazole, itraconazole, flucytosine, and terbinafine. All of the strains had the same temperature and pH limits and the same carbohydrate assimilation capacities as measured by the yeast API 20C identification method.

View larger version (124K): [in this window] [in a new window]   FIG. 3.   Cellular morphologies of resistant and susceptible derivatives of C. lusitaniae 7227. Cells were suspended in sterile water from single, fresh colonies grown on SAB agar and visualized by differential interface microscopy (160X Nikon E600 Research Microscope) and digitally imaged with the SPOT camera (Diagnostic Instruments Inc., Sterling Heights, Mich.). (A) Resistant cells from isolate 7227R6; (B) susceptible cells from isolate 7227S11.

View larger version (14K): [in this window] [in a new window]   FIG. 4.   Effect of zymolyase on viability and resistance to AmB. (A) The indicated Candida strains were digested with zymolyase as described in Materials and Methods. The percentages of CFU were calculated as the number of colonies per milliliter after zymolyase digestion, divided by colonies per milliliter of the same strain incubated without zymolyase (open bars). Less than 1% of cells from isolates, other than 7227R6 and C. albicans, survived zymolyase treatment and therefore are not evident in the figure. The same strains were also plated onto AmB agar without preincubations in zymolyase (closed bars) to verify their levels of susceptibility to AmB. (B) C. lusitaniae 7227R6 was incubated for 5 h in zymolyase as described in Materials and Methods and then plated on SD plus the indicated concentrations of AmB. %CFU values are percentage of the original CFU before zymolyase was added. Strain abbreviations: R6, 7227R6; S11, 7227S11; RS9, 7227RS9, a susceptible isolate derived from 7227R6; RS9', a colony derived from the original 7227RS9; RS9.3, a variant derived from 7227RS9 that did not switch back from susceptibility to resistance; Ca, C. albicans B311.

AmB switching in other Candida species. We were unable to detect spontaneous AmB-resistant derivatives of susceptible Candida isolates in several Candida species, including C. glabrata, C. tropicalis, C. kefyr, and C. guilliermondii. In contrast, 30 of 30 independent clinical isolates of C. lusitaniae demonstrated the switching phenotype described for 7227S11 (Table 3). All tested C. lusitaniae isolates showed an incidence of at least 0.001 resistant cell per viable cell. From five randomly chosen isolates, five colonies from AmB plates were resuspended and replated on SD agar with and without AmB. All showed the resistant phenotype, i.e., equivalent numbers of CFU developed on both plates. These data suggest that AmB switching was a characteristic feature of many C. lusitaniae isolates which was not found among other Candida species.

Effects of high AmB concentrations on 7227R cells. Table 3 also demonstrates that a standard MIC determination for a C. lusitaniae isolate did not reflect the dramatic switching phenotype of the minority population on agar plates. Indeed, even the MICs for 7227R6 were only slightly higher in these assays. To clarify this difference and to determine whether the switch to resistance on agar may have clinical significance, we compared the survival rates of 7227R6 and 7227RS9 after short-term exposures to high concentrations of AmB in SD broth (Fig. 5). The exposure of susceptible cultures to 2 µg of AmB per ml reduced viability by 5 orders of magnitude within 5 h. During the same interval, resistant cultures retained full viability and continued to survive for well over 24 h. Survival rates for both cultures exposed to 1 µg of AmB per ml were similar to those for cultures exposed to 2 µg/ml. In the absence of AmB, both cultures grew to confluency within 24 h (data not shown). Exposure of resistant cultures to even higher concentrations of AmB (5 to 20 µg/ml) was not highly fungicidal; in independent trials performed over more than a 1-year period, survival ranged between 10 and 50% after up to 24 h of exposure.

View larger version (13K): [in this window] [in a new window]   FIG. 5.   Resistance of C. lusitaniae 7227 cells to fungicidal effects of AmB. Single colonies of 7227R6 () or 7227RS9 () were inoculated into 5 ml of buffered SD in 50-ml tubes for overnight growth at 30°C. Twenty microliters of grown culture was transferred into 1 ml of buffered SD with or without AmB (2 µg/ml) for the indicated times. CFU were determined by plating four 10-fold serial dilutions of each sample onto SAB agar plates. Percent survivors were calculated from the ratios of these CFU to the CFU before AmB was added. Cultures without AmB grew to confluency in less than 24 h.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Switching in C. albicans has pleiotropic effects, beyond the initial colony and cellular morphologies, including surface antigenicity, cell wall structure, adhesiveness, susceptibility to polymorphonuclear leukocytes, virulence, and possibly drug susceptibility (27, 56). Presumably, the switching properties vary with the cell lineage and the individual cell (59, 61). Molecular epidemiological data have previously demonstrated that recurring C. albicans infections in a single individual are typically subtypes of the initial isolate (29, 34-36, 45, 59, 62). This has been interpreted as a longitudinal selection of variant phenotypes within the individual host to enhance survival in a particular environment (host, anatomical location, and commensal microorganisms). These variant phenotypes might be generated by a switching mechanism that is consistent with studies that correlate invasiveness of Candida isolates with a high switching mode and with a recent gene disruption study that links virulence of C. albicans to the ability to switch to hyphal growth (7, 23, 27).

We have observed the development of AmB resistance that involves rapid in vitro switching from AmB-susceptible to AmB-resistant phenotypes only in C. lusitaniae. The switch was reversible and occurred at a high and variable rate of about 1 per 10 to 10,000 cells. As a result, many cells in each colony (contains approximately 3 × 10⁶ cells), switched phenotype from that of the cell that originated the colony. We have identified three different types of C. lusitaniae cells in standard colonies and have designated them types I, II, and III. Type I cells do not grow when replated on AmB-supplemented medium. Type II cells grow, but the resultant subculture of colonies on AmB-supplemented medium were still in a susceptible, switching mode. These cells generate only mixed-phenotype second-generation colonies, composed of mostly type II cells that produce an average of 1 in 50 daughter cells able to grow on AmB-supplemented agar. The ratio of type II to type I cells in a colony increases about seven- to eightfold in colonies recovered from AmB-supplemented agar versus those cells recovered from nonselective agar. Susceptible lineages derive from type I and type II cells and continue to switch at variable rates that range from 1 in 50 to 1 in 10,000 cells. Type III cells form colonies that have a stable resistant phenotype and grow with equal efficiency on media with and without AmB. These cells occur only once per 30 colonies subcultured on AmB-supplemented medium. The other 29 colonies are of the type II mixed-phenotype lineage. One might argue that there is no such thing as a purely resistant lineage. The "resistant" colony that arises once per 30 colonies from AmB plates may result from a switch to resistance early in the growth of that colony from its parent cell, especially if the frequency of switching from resistance to susceptibility was lower than the reverse. The other 29 type II colonies are explained by having switched to resistance late during the growth of each colony. However, resistant lineages retain a high RI even after many generations in the absence of selective pressure, and susceptible isolates can be recovered from these only after induction by UV. Present data cannot determine which of two models account for the resistant lineage. In one theoretical model, the persistently resistant lineage has turned off a switching mechanism that must be reactivated to restore switching. In a second model, resistant lineages start out with such a high percentage of resistant cells due to an early switch event. As long as the rate of switching from resistance to susceptibility is lower than the reverse switch, these colonies will appear to be composed of only resistant cells.

The resistance phenotype was specific for AmB, of all of the drugs tested, and was accompanied by changes in cellular morphology. Switching of this type seems commonplace among independent clinical isolates of C. lusitaniae but was not detected in other Candida species.

The resistant isolate 7227R6 was less vulnerable than susceptible derivatives of strain 7227 to zymolyase-mediated osmotic instability. However, 7227R6 was rendered susceptible to AmB after exposure to zymolyase. These data suggest that 7227R6 cells differ from susceptible lineages by having altered cell walls that were less readily digested by zymolyase and that the resistance may be conferred by a wall-mediated barrier that can be removed by zymolayse without decreasing cellular viability.

The level of AmB resistance in the switched isolates was sufficient to be clinically significant. In vivo, serum concentrations of AmB rarely exceed 2 to 3 µg/ml and decline steadily over the next 24 h (13). In vitro, 7227R6 was not affected by a 2- to 5-µg/ml concentration of AmB for at least 24 h. In addition, it is important to emphasize that this type of resistance, while easily measured by the EEA method or the MFC assay, would be completely missed by standard in vitro MIC protocols (Table 3). Standard MIC assays can only determine whether an isolate grows in the presence of the drug and so fail to differentiate among isolates that were or were not killed by the drug. This may offer one possible explanation for the frequently observed absence of documented in vitro AmB resistance in clinical C. lusitaniae isolates recovered from patients who were treated with AmB and clinically failed.

The switching of C. lusitaniae isolates from an AmB-susceptible to an AmB-resistant phenotype bears a number of similarities to colony morphology switching reported in C. albicans. Both occur at high frequencies relative to mutation rates, and both have phenotypes that switch in both directions. In addition, both increase switching rates after exposure to UV light, heat shock, or cellular components of blood (58). However, there are differences between the two switching systems. With C. lusitaniae AmB resistance phenotype switching, there was only a subtle change in colony morphology. Switching of C. lusitaniae is thus far limited to the in vitro AmB susceptibility phenotype switch, changes in cellular morphology, and alterations in fungal cell walls. In contrast, there were multiple switching changes seen in C. albicans. In C. lusitaniae, the least susceptible form was the elongate cellular form, in contrast to the situation observed in C. albicans (56).

C. albicans B311 does not switch to develop AmB resistance under the conditions described in this paper. Instead, C. albicans B311 and other Candida species may acquire a phenotypic resistance (AmB^PR) when the cells are preexposed to azoles (14, 15, 52, 65, 67, 71) or when they age (17, 18, 71). In either circumstance, they undergo alterations in cell wall structures that may be functionally related to the resistance described previously (4, 9, 19). This resistance differs in several respects from switching. First, AmB^PR C. albicans cells have only transient resistance; thus, they cannot be serially passaged on AmB-supplemented medium. Second, under optimal conditions, 10 to 99% of the cells in a C. lusitaniae population switch to resistance, which is much higher than the percentage of AmB^PR cells (71). It remains possible, however, that the switching and phenotypic resistance are mediated through changes involving the fungal cell wall. One is mediated via physiological mechanisms and the other is mediated via genetic alterations, and a subset of wall changes may affect cellular and morphological changes. For C. lusitaniae switching, it remains to be determined whether wall and morphology changes are essential to the process of AmB resistance.

Few studies have evaluated whether there is a link between morphological switching of C. albicans and antifungal drug resistance. It was previously noted in the white-to-opaque transition of C. albicans that the AmB MICs for opaque cells were slightly higher than those for white cells and that the opaque cells showed increased resistance to the fungicidal effect of AmB (20). However, most of the previously published studies are confusing for several reasons. First, they are correlative studies with uncharacterized clinical isolates. Clinical isolates of C. albicans that were recovered from patients on azole therapy had switched colony morphologies, and some of these isolates were resistant to higher concentrations of azoles or in some cases to other nonclinical drugs (59, 61, 71). However, not all colonies with a shared morphology were resistant, and in some cases the resistant isolates were not recovered in a manner consistent with drug selection. In a second study, the authors noted that long-term exposure to inhibitory concentrations of heavy metals resulted in metal-resistant colonies with altered morphologies. These altered morphologies, however, were also seen among the susceptible colonies (30). These data suggest that switching produces alterations from a list of affected traits in individual cells, some combination of which may prove successful under the new environmental challenge.

What is known about the possible mechanism of AmB resistance? It is proposed that AmB kills Candida and other fungal cells by forming ion channels in the plasma membrane (10, 12). Among several clinical isolates of Candida, resistance has been correlated with decreased levels of ergosterol (16, 32, 33). In addition, resistance to AmB can also result from mutations in the sterol -5,6-desaturase. This defect results in the accumulation of 14-methylfecosterol in the presence of fluconazole rather than 14-methylergosta-8,24(28)-dien-3,6-diol (24, 25). Speculatively, alterations in the cell wall might directly affect the accessibility of AmB to membrane ergosterol or more indirectly change membrane composition, in turn affecting the ability of AmB to form ion channels. Conversely, changes in membrane composition mediated by fluconazole might result in cell wall structure alterations that affect accessibility of AmB and change cell morphology.

What potential clinical relevance might AmB switching have, both in the area of clinical specimen testing and in the progression of infection during therapy? First, that certain Candida species are inherently resistant, either to azoles (C. krusei and C. glabrata) (26, 31, 37, 63) or to AmB (C. lusitaniae and C. guilliermondii) (44), has been well described. Screening small numbers of clinical specimens from an immunocompromised patient for drug susceptibility could easily be misleading, if such samples were taken when the yeast was in an AmB-susceptible mode. Typically, samples in the laboratory were diluted and inoculated at about 300 cells per microtiter well (3). This small sample of cells would contain only a few to no resistant revertant cells. Even if a small number of resistant cells were present, they would not have time to grow to an extent that would generate sufficient turbidity to produce a higher MIC cutoff.

Finally, C. lusitaniae is similar to other Candida species in producing systemic or localized infections in compromised individuals (21, 22, 37, 50, 51, 72). However, this species is unique by virtue of its innate ability to switch antifungal susceptibility phenotype during growth. This ability is a reflection of its history of developing in vitro (1, 22, 43, 48) as well as in vivo (21, 22, 37) AmB resistance. During antifungal therapy, cells that switch to a resistant mode would be selected and the infection would potentially progress with the more resistant variants. Furthermore, since many in vitro conditions are known to stimulate morphological switching, it is reasonable to expect that localized conditions in vivo might also enhance morphological and AmB susceptibility switching, potentially changing other virulence traits. Our observation, that switching is efficiently induced by exposure to whole blood, reinforces this notion.