|Home | About | Journals | Submit | Contact Us | Français|
The development of azole resistance in Candida albicans is most problematic in patients with AIDS who receive long courses of drug for therapy or prevention of oral candidiasis. Recently, the rapid development of resistance was noted in other immunosuppressed patients who developed disseminated candidiasis despite fluconazole prophylaxis. One of these series of C. albicans isolates became resistant, with an associated increase in mRNA specific for a CDR ATP-binding cassette transporter efflux pump (K. A. Marr, C. N. Lyons, T. R. Rustad, R. A. Bowden, and T. C. White, Antimicrob. Agents Chemother. 42:2584–2589, 1998). Here we study this series of C. albicans isolates further and examine the mechanism of azole resistance in a second series of C. albicans isolates that caused disseminated infection in a recipient of bone marrow transplantation. The susceptible isolates in both series become resistant to fluconazole after serial growth in the presence of drug, while the resistant isolates in both series become susceptible after serial transfer in the absence of drug. Population analysis of the inducible, transiently resistant isolates reveals a heterogeneous population of fluconazole-susceptible and -resistant cells. We conclude that the rapid development of azole resistance occurs by a mechanism that involves selection of a resistant clone from a heterogeneous population of cells.
The development of fluconazole resistance in Candida albicans after long exposures to the drug is well documented for patients with AIDS and recurrent oropharyngeal candidiasis (28). Phenotypically stable resistance to azole antifungals in C. albicans can result from mutations in or increased expression of genes involved in the ergosterol synthesis pathway (including the target enzyme 14-α demethylase) and increased expression of ATP-binding cassette transporter and major facilitator efflux pumps (28). Recently, several groups have noted that resistance can develop in C. albicans after only short exposures to the drug, both in vitro (2) and in vivo (10, 14). One example of this rapid development of resistance was in a strain of C. albicans that caused disseminated infection in a patient after receipt of bone marrow transplantation (BMT) (10). The susceptible isolate became resistant to azole drugs, with an associated increase in mRNA for an efflux pump in the CDR family. In these clinical isolates, phenotypic resistance to these drugs was transient, as the azole antifungal MICs for the resistant isolates decreased after serial growth in the absence of drug (9).
Rapid development of antimicrobial resistance in vivo can result from drug pressure that selects for a resistant clone or from selection for a resistant subset of cells within a heterogeneous population. Alternatively, the presence of the drug can influence transcription of resistance genes within a clonal population of cells. Heterogeneous resistance, or selection for a resistant subpopulation of cells, is a phenomenon that has been well documented for Staphylococcus species (3) and more recently described for Cryptococcus neoformans (11).
Two series of isogenic C. albicans isolates with inducible yet transient resistance to azole drugs have been identified from patients who received BMTs at our center (data presented here and in references 9 and 10). In this study, we show that the molecular mechanism of azole resistance in the second series of isolates is similar to that in the first series, as both series of isolates become resistant with associated increased expression of CDR mRNA. The phenotypic pattern of resistance in both series of isolates suggests that the inducible, transient nature of resistance is associated with heterogeneous resistance, or selection of a resistant subpopulation of cells from a phenotypically heterogeneous population.
The isolates used in this study are listed in Table Table1.1. The first series of nine isolates of C. albicans, FH1 to FH9 has been previously described (9). This series consists of colonizing and bloodstream isolates of the same strain from a patient who underwent BMT (10). The second series of isolates, TL1 to TL3, consists of two colonizing (mouthwash) isolates (TL1 and -2) and a third isolate (TL3) obtained from the blood of a patient who underwent BMT. Individual yeast colonies were picked from Sabouraud dextrose agar (Difco), identified as C. albicans by germ tube testing and use of API 20C strips, and stored frozen at −70°C in yeast extract-peptone-dextrose (YEPD) (10 g of yeast extract, 20 g of peptone, and 20 g of dextrose per liter) containing 10% glycerol. Azole-susceptible laboratory strains (SC5314 and 3153a) and azole-susceptible (2-76) and -resistant (12-99) clinical isolates described previously (16) were also used as controls.
Powder formulations of fluconazole (Roerig-Pfizer, New York, N.Y.), itraconazole (Janssen Pharmaceutica, Beerse, Belgium), and amphotericin B (Sigma, St. Louis, Mo.) were suspended in distilled water, filter sterilized, and stored at −70°C. Media utilized in these studies include yeast nitrogen base (1.7 g/liter) (Difco, Detroit, Mich.) with ammonium sulfate (5 g/liter) and dextrose (10 g/liter) (YAD), YEPD, and RPMI 1640 with 0.165 M MOPS (morpholinepropanesulfonic acid), pH 7.0 (American Bioorganics, Niagra Falls, N.Y.). Antifungal susceptibility testing was performed by the standardized microdilution method (13). E tests were performed as instructed by the manufacturer (AB Biodisk North America Inc., Piscataway, N.J.).
Genomic DNA was prepared using glass bead cell shearing (5). Restriction fragment length polymorphism analyses were performed to determine genetic relatedness of isolates TL1 to -3 and to compare susceptible and resistant isolates after serial transfer. Southern blotting was performed (7), and blots were hybridized with the C. albicans strain-specific Ca3 probe (20). The Ca3 probe was labeled with [α-32P]dATP with random priming.
The sequence of the ERG11 gene of the fluconazole-susceptible isolate TL1 was compared with that of the fluconazole-resistant isolate TL3. Genomic DNA was prepared, and PCR was performed using primers that span the length of the ERG11 gene. Products were cleaned by dilution and centrifugation in Centricon 100 concentrators (Amicon, Beverly, Mass.), according to the manufacturer's directions. Nucleotide sequences were determined using a DNA sequencer with Taq dye-primer and dye-terminator chemistries (Applied Biosystems, Foster City, Calif.).
Quantities of mRNAs for genes previously reported to be involved in azole resistance (MDR1, CDR, and ERG11), other genes in the ergosterol synthesis pathway (ERG1, ERG7, and ERG9), and the housekeeping gene TEF3 were compared in the susceptible and the resistant isolates. To quantitate mRNAs for multiple genes simultaneously, reverse Northern blotting was performed (6), followed by confirmation with standard Northern blotting. Yeast RNAs were prepared using cell-shearing methods (21). To perform the reverse Northern blotting, 10 μg of total cellular RNA was denatured in 100 μl of 0.05 M sodium carbonate at 55°C for 30 min, precipitated in ethanol, resuspended in water containing RNase inhibitor (Boehringer Mannheim, Indianapolis, Ind.), and 5′ end labeled with [γ-32P]ATP using T4 polynucleotide kinase (Promega, Madison, Wis.). Labeled RNA was cleaned using a G-25 MicroSpin Column (Amersham Pharmacia Biotech, Piscataway, N.J.). Target DNAs were PCR fragments previously cloned into PCR Script-Amp cloning vectors (Promega). Ten micrograms of each of these fragments was electrophoresed in 1% agarose gels and transferred to membranes using standard techniques for Southern blotting (7). Hybridizations were performed with labeled mRNAs from both the susceptible (TL1) and resistant (TL3) isolates. Signal intensities were quantified after exposure to a phosporimaging screen (Storm 1860; Molecular Dynamics), using the Molecular Dynamics ImageQuant program. To adjust for the quantity of labeled RNA used for hybridization, signal intensities of genes of interest were normalized with TEF3.
Northern analyses were performed for each gene that had a difference (≥2-fold) in expression between the resistant and susceptible isolates (7). To control for the amount of RNA loaded into gels, hybridized membranes were stripped and rehybridized with probes specific for the actin gene. Oligonucleotide probes (actin, CDR1, and CDR2) were labeled with [γ-32P]ATP by use of T4 polynucleotide kinase, and plasmid probes were labeled with [α-32P]dATP using random priming. Oligonucleotides specific for the CDR1 and CDR2 genes were used as probes for Northern blots.
The stability of the fluconazole-resistant phenotype was determined by serial transfer in the absence of drug. A single colony of each isolate was cultured in drug-free YEPD at 30°C and transferred with 1- to 5,000-μl dilutions every 2 or 3 days. To determine if susceptible isolates become resistant in vitro in the presence of the drug (induction), single colonies were cultured in YAD containing fluconazole at a concentration equivalent to two times the original MIC for the isolate. YAD was used instead of yeast extract-based medium (YEPD) to ensure that ergosterol was not supplemented by the medium. MICs for the transferred isolates were determined weekly, and isolates were stored at −70°C in YEPD containing 10% glycerol.
A previously described method that utilizes CHROMagar (CHROMagar Company, Paris, France) containing fluconazole to identify resistant C. albicans isolates (15) was modified for population analyses. A single colony was cultured overnight in YEPD broth at 30°C, cells were counted with a hemocytometer, and 100, 1000, and 10,000 cells were plated onto CHROMagar plates without and with fluconazole (1, 4, 16, and 64 μg/ml). Growth was quantified using an automated colony counter (Eagle Eye II; Stratagene, La Jolla, Calif.) after 2 days of incubation (30°C). The number of colonies that grew in the presence of the drug at each concentration, relative to the number that grew in the absence of the drug, was calculated and plotted.
The first two TL isolates (TL1 and TL2) are susceptible to fluconazole, and the third isolate (TL3) is resistant (Table (Table1).1). TL1 and TL2 are also susceptible to itraconazole and ketoconazole (data not shown), while TL3 is resistant (MICs, 16 and 1 μg/ml, respectively). All isolates are susceptible to amphotericin B (MICs, 1 μg/ml).
Southern blots of genomic DNAs from TL1 and TL3 digested with EcoRI and probed with Ca3 showed identical banding patterns (Fig. (Fig.1).1). Identical patterns were also obtained after digestion with HindIII (data not shown). This series thus represents the same strain, or matched isolates that have increasing resistance to azole drugs. Figure Figure1B1B shows the reverse Northern blots of the azole-susceptible isolate, TL1, and the azole-resistant isolate, TL3. Quantification of signal intensity, with normalization to TEF3 for both isolates, shows that the resistant isolate contains approximately sevenfold more mRNA for CDR. All other signal intensities differed by a factor of less than 2. Increased expression of CDR was confirmed using standard Northern blotting (not shown).
The ERG11 sequences were determined for TL1 and the resistant isolate, TL3. The TL1 sequence was compared with the published sequence. At two nucleotide positions, the TL1 sequence showed two bases at the same location (allelic differences) that result in amino acid differences compared to the published sequence (D116D/E and E266E/D, where D is Asp, E is Glu, and 116 and 266 are the amino acid positions). In addition, the TL1 sequence showed four other allelic differences and five substitutions, none of which result in a change in the protein sequence. When the sequence of the susceptible isolate was compared to that of the resistant isolate (TL3), there was a mutation (T580C) that results in a Phe-to-Leu change at position 145 (F145L). This mutation was present in both alleles, and although mutations at this site have been reported previously, it is unclear whether it is associated with azole resistance (8). In addition, all of the allelic differences, including the two allelic differences that result in amino acid differences (D/E116E and E/D266E), were resolved in the resistant isolate.
Our previous studies showed that azole drug resistance developed in a colonizing isolate of C. albicans along with an associated increase in mRNA for a member of the CDR efflux pump family. We explored the mechanism of resistance further by performing Northern analyses with probes specific for the CDR1 and CDR2 genes (6). The CDR1 mRNA levels increase approximately 2.3-fold between FH1 and FH8. CDR2 is not detectable in FH1, and the mRNA levels in FH8 are at least 24-fold above background (data not shown).
The fluconazole, ketoconazole, and itraconazole MICs of the resistant isolates in the first series (FH5 and FH8) decreased after transfer in the absence of drug (9) (Fig. (Fig.2A).2A). To determine which specific efflux pump is involved in resistance for these transferred isolates, Northern analyses were performed using probes specific for CDR1 and CDR2. Quantification of CDR1 (normalized to actin) showed a slight decrease in the susceptible transfers compared to the resistant isolates. However, mRNA for CDR2 (normalized to actin) decreased threefold for FH5 and twofold for FH8 (data not shown), suggesting that CDR2 is involved in resistance.
Serial transfer experiments were also performed with the resistant isolate from the TL series. As shown in Fig. Fig.2A,2A, the fluconazole MIC for this resistant isolate also decreased after serial growth in the absence of the drug. The experiment was repeated twice, with the same decrease in MIC. Identical restriction fragment length polymorphism banding patterns using the Ca3 probe suggest that contamination did not account for the loss of resistance (data not shown). These data suggest that the resistant isolates FH5, FH8, and TL3 are unique compared to other fluconazole-resistant isolates from a patient with AIDS (isolate number 12-99), as 12-99 is stably resistant after serial transfer in the absence of the drug (29).
To determine if the original, fluconazole-susceptible isolates in both series (FH1 and TL1) become resistant to the drug in vitro, the isolates were serially transferred in the presence of fluconazole. Figure Figure2B2B diagrams the increase in fluconazole MIC in both FH1 and TL1. The fluconazole-susceptible laboratory controls (3153a and SC5314) and the fluconazole-susceptible clinical isolate 2-76 (27) did not show increased resistance after growth in the presence of the drug (Fig. (Fig.2B).2B). The experiments were repeated twice, with similar results, and genetic similarity was confirmed with Southern analyses using the Ca3 probe (not shown).
Quantification of mRNA for the CDR efflux pump documented that the resistant transferred isolates FH1-R (for FH1, resistant) and TL1-R (for TL1, resistant) contained approximately 15- and 3-fold more CDR mRNA than their susceptible partners (FH1 and TL1), respectively. Also, TL1-R contained 2.5-fold more mRNA for ERG11 than its susceptible partner, TL1. The phenotypic resistance of the transferred isolates FH1-R and TL1-R was also unstable, as the fluconazole MICs returned to their original values after serial growth in the absence of the drug. Figure Figure33 demonstrates the inducibility and transience of ketoconazole, itraconazole, and fluconazole resistance in FH1. The susceptible isolate of the second series (TL1) demonstrated a similar inducibility and transience of resistance to the three azole antifungals (data not shown). The susceptible isolates obtained after transfer, FH1-R-T (for FH1, resistant, transfer) and TL1-R-T (for TL1, resistant transfer), have genomic DNA banding patterns identical to those of their partner strains (data not shown), but the patterns of the two strains differ from each other.
To determine if the transient, inducible azole resistance in the FH and TL isolates is associated with selection for a resistant subpopulation or increased expression of CDR in one clone, analysis of resistance within the cellular population was performed. Figure Figure44 shows the results of at least two population analyses for the susceptible, inducible strains FH1 and TL1 compared to the susceptible, noninducible strains 3153a, 2-76, and SC5314. FH1 has a majority of cells for which the MIC corresponds with the serial dilution MIC (4 μg/ml) but there is also a minority population of cells (approximately 10%) for which the apparent MIC is >64 μg/ml. TL1 has a majority of cells for which the MIC is 1 μg/ml and a minority for which the MIC is between 4 and 64 μg/ml. Alternatively, the noninducible isolates have homogeneous populations, for which the MICs correspond with the serial dilution MICs (≤1 μg/ml).
Population analyses revealed a homogeneous resistant population for the resistant clinical isolate FH8 (Fig. (Fig.5)5) and the stably resistant control 12-99 (not shown). However, after serial passage in the absence of fluconazole, the population analysis of the transiently resistant isolate FH8-T changed to a heterogeneous pattern, with a minority of cells for which the MIC is between 16 and 64 μg/ml. Also, a minority population of cells for which the MIC is 64 μg/ml was apparent in the population of FH1-R, but it disappeared in the isolate that was transferred in the absence of drug (FH-R-T) (Fig. (Fig.5).5). The same results were noted with population analyses of the resistant isolates and susceptible transfers from the TL series (data not shown).
The population analyses were performed by sampling liquid cultures of single colonies grown overnight. To determine if cells within a single colony are heterogeneous in the absence of replication in liquid media, growth on plates containing fluconazole (64 μg/ml) was quantified after direct inoculation of single colonies of FH1 and the control strain SC5314 and compared with overnight growth to saturation. After 2 days of incubation, equal numbers of directly inoculated FH1 cells (3.4%) and cells of the FH1 isolate inoculated from overnight growth in liquid (3.5%) were resistant. In contrast, no colonies of SC5314 were present on fluconazole plates. These data suggest that the heterogeneous phenotype is present within single colonies and is not a product of replication in liquid media.
After 2 days of incubation, the heterogeneous isolates show growth on medium that contains fluconazole at a concentration greater than the MICs for isolates. At high concentrations of the drug, these colonies are small but clearly present. Under the same conditions, homogeneous isolates show no growth on medium that contains fluconazole at a concentration exceeding the MICs isolates. To determine if a stable mutation accounts for cell growth in the presence of fluconazole, the serial dilution MICs for large colonies isolated from medium lacking fluconazole were compared to the MICs for small colonies isolated from plates containing fluconazole. The MICs for all colonies sampled from plates lacking fluconazole and containing 16 and 64 μg of drug per ml were equivalent to those obtained from each isolate's parent (FH1, 4 μg/ml; TL1, 1 μg/ml). Also, genomic DNA banding patterns after EcoRI digestion were identical for all (susceptible and resistant) colonies of FH1 and TL1 (data not shown). Thus, although growth on plates containing fluconazole suggests that the majority of the cells within the colonies are resistant, this resistant phenotype is not stable within the population of cells. The absence of large differences in the Ca3 banding patterns suggests that large DNA insertions or deletions near Ca3 do not account for the drug resistance observed.
Since population analyses are labor- and time-intensive, we sought an alternative method to detect the heterogeneous phenotype of inducible isolates. For a preliminary screen, we compared the E-test zones of inhibition of heterogeneous isolates (FH1 and TL1) and homogeneous isolates (2-76, SC5314, and 3153a). As demonstrated in Fig. Fig.6,6, the heterogeneous isolates tend to have increased satellite growth in the inner circle of inhibition, while the homogeneous isolates have clearly demarcated zones of inhibition.
We have documented that C. albicans can become resistant to azole antifungals through a mechanism that involves selection of a resistant isolate from a heterogeneous population of cells. These findings highlight the complexity of antimicrobial susceptibility testing and clinical interpretation for both bacteria and yeasts.
Recently, a heterogeneous phenotype in clinical isolates of Cryptococcus neoformans isolated from AIDS patients with persistent meningitis was described (11). The two series of C. albicans isolates described here also acquired clinically significant resistance, as they became resistant to fluconazole in vivo and caused disseminated infection despite prophylactic administration of large doses of the drug in BMT recipients. The prevalence of these organisms and whether they account for clinical failures of fluconazole therapy in other situations are unknown.
Inducible drug resistance associated with a heterogeneous phenotype in C. albicans is consistent with previous reports of heterogeneity in susceptibility phenotypes (22) and resistance mechanisms (4) within clonal populations of Candida spp. Cultures of C. glabrata, C. krusei, and C. albicans contain heterogeneous populations of colonies that vary in susceptibility to fluconazole and itraconazole (22). Also, the rapid development of reversible fluconazole resistance in C. albicans after serial growth in the presence of drug was reported (2), although the mechanisms associated with resistance were not determined. Other investigators found that populations of C. albicans that are experimentally induced to become resistant to fluconazole acquire resistance through multiple mechanisms, despite clonal derivation by single-cell progenitors (4). More recently, in vitro induction of resistance in C. tropicalis was documented, with associated increased expression of CDR1 and MDR1 (1). Our findings complement these observations by documenting that the variability in the susceptibility phenotype within cellular populations is associated with the clinical development of resistance and an inducible phenotype in vitro.
This phenomenon might be similar to the heterogeneous resistance that occurs in staphylococci and enterococci that become resistant to β-lactam and glycopeptide antibiotics (3, 23). In staphylococci, heterogeneous resistance appears to involve several genes and control is complex (3). Although the mechanisms remain elusive, factors in the cellular environment (temperature and medium pH, etc.) affect the phenotypic expression of resistance (3). One mechanism by which these organisms become resistant to β-lactam and glycopeptide antibiotics appears to be mediated by a change in cell wall components, such as penicillin-binding proteins and peptidoglycan, resulting in a decrease in susceptibility due to altered binding (and possibly cellular entry) of the drug (12, 24). Whether similar mechanisms are associated with heterogeneous azole drug resistance in C. albicans is currently unknown, but it is of interest that all of the antimicrobials that have been associated with heterogeneous resistance (β-lactams, glycopeptides, and azoles) are static drugs that directly or indirectly target the cell wall or membrane.
As noted previously (22), another possible cellular mechanism associated with heterogeneity in C. albicans might involve phenotypic switching between susceptibility phenotypes (26). Isolates of Candida lusitaniae undergo a reversible phenotypic switch that is associated with the development of resistance to amphotericin B (30). Since different strains of C. albicans undergo reversible phase switching, this is a potential mechanism for heterogeneity between susceptibility phenotypes (26).
The high frequency at which resistant cellular subpopulations are detected suggests that genetic mutations are not responsible for the generation of resistance, because mutations occur at frequencies approximating 10−6 to 10−8 per gene. Although we utilized a DNA fingerprinting method that has the best resolution, our conclusions are limited by the fact that no methods to fully ascertain the genetic distance between strains are available (25). Also, our molecular studies suggest that the two series of isolates both became resistant to the drug in vivo and in vitro along with associated increased mRNA levels for CDR1 and CDR2 (reference 9 and this study). Although we do not know how the cellular and molecular makeups of these isolates otherwise compare, it is possible that the heterogeneous phenotype is specifically associated with CDR expression, as increased expression of this efflux pump might cause a metabolic growth disadvantage that could explain the transient nature of resistance. Studies investigating the cellular phenotypes and molecular mechanisms of heterogeneous isolates are ongoing.
The development of resistance in these strains should be differentiated from the trailing phenotype that is observed in serial dilution susceptibility testing. Isolates that trail exhibit a low azole MIC after 24 h of growth and a high MIC after 48 h (18). The isolates described here do not trail. This is an important distinction, as trailing isolates behave as susceptible strains in vivo (18), while these heterogeneous isolates were clearly the cause of resistant disease in patients on prophylactic fluconazole. Preliminary population analyses on trailing isolates do not show a heterogeneous pattern as described in this study, but instead these isolates have a high degree of variability in colony morphology in the absence of drug (data not shown). Also, the trailing isolates do not become resistant with serial passage in the presence of drug in vitro (data not shown). These findings suggest that the heterogeneous and trailing phenotypes differ in both in vitro and in vivo behavior. While isolates that have the heterogeneous phenotype are a potential cause of clinically resistant infection, the significance of the trailing phenotype appears to be limited to being a cause of falsely elevated MICs in susceptibility testing.
The finding that isolates of C. albicans can become resistant to fluconazole rapidly, despite MICs being within the susceptible range, has important implications for the clinical microbiology laboratory. Previous studies have noted that serial dilution susceptibility testing is less successful in predicting therapeutic success than in predicting failure (19). Although it is likely that host factors are the major explanation, it is possible that isolate heterogeneity contributes to these limitations. Given the use of fluconazole as a single agent for candidemia in immunocompetent hosts (17), the laboratory should be able to differentiate homogeneous, susceptible isolates from heterogeneous, susceptible isolates that can become resistant after drug exposure. Our preliminary results suggest that E tests might be useful, but further studies are necessary to characterize large numbers of heterogeneous isolates and to document the correlation between heterogeneity and clinical failure.
In summary, we have described an important mechanism by which C. albicans can become resistant to azole antifungals in immunosuppressed patients. Clinical microbiologists and clinicians should be aware of this as a potential cause of azole treatment failure. Further studies are required to determine the prevalence of this phenotype and the cellular and molecular factors involved in conferring resistance.
This research was supported in part by NIH grant K08 A108044 to K.A.M. and NIH Adult Leukemia Research Center core grant CA 18029. T.C.W. was supported by NIH grant R01 DE11367 and a grant from the Murdock Charitable Trust and was a recipient of a Burroughs Wellcome Fund New Investigator Award in Pathogenic Mycology.