Heteroresistance to Fluconazole Is a Continuously Distributed Phenotype among Candida glabrata Clinical Strains Associated with In Vivo Persistence

Abstract

Candida glabrata causes persistent infections in patients treated with fluconazole and often acquires resistance following exposure to the drug. Here we found that clinical strains of C. glabrata exhibit cell-to-cell variation in drug response (heteroresistance). We used population analysis profiling (PAP) to assess fluconazole heteroresistance (FLC(HR)) and to ask if it is a binary trait or a continuous phenotype. Thirty (57.6%) of 52 fluconazole-sensitive clinical C. glabrata isolates met accepted dichotomous criteria for FLC(HR) However, quantitative grading of FLC(HR) by using the area under the PAP curve (AUC) revealed a continuous distribution across a wide range of values, suggesting that all isolates exhibit some degree of heteroresistance. The AUC correlated with rhodamine 6G efflux and was associated with upregulation of the CDR1 and PDH1 genes, encoding ATP-binding cassette (ABC) transmembrane transporters, implying that HetR populations exhibit higher levels of drug efflux. Highly FLC(HR) C. glabrata was recovered more frequently than nonheteroresistant C. glabrata from hematogenously infected immunocompetent mice following treatment with high-dose fluconazole (45.8% versus 15%, P = 0.029). Phylogenetic analysis revealed some phenotypic clustering but also variations in FLC(HR) within clonal groups, suggesting both genetic and epigenetic determinants of heteroresistance. Collectively, these results establish heteroresistance to fluconazole as a graded phenotype associated with ABC transporter upregulation and fluconazole efflux. Heteroresistance may explain the propensity of C. glabrata for persistent infection and the emergence of breakthrough resistance to fluconazole.

Importance: Heteroresistance refers to variability in the response to a drug within a clonal cell population. This phenomenon may have crucial importance for the way we look at antimicrobial resistance, as heteroresistant strains are not detected by standard laboratory susceptibility testing and may be associated with failure of antimicrobial therapy. We describe for the first time heteroresistance to fluconazole in C. glabrata, a finding that may explain the propensity of this pathogen to acquire resistance following exposure to fluconazole and to persist despite treatment. We found that, rather than being a binary all-or-none trait, heteroresistance was a continuously distributed phenotype associated with increased expression of genes that encode energy-dependent drug efflux transporters. Moreover, we show that heteroresistance is associated with failure of fluconazole to clear infection with C. glabrata Together, these findings provide an empirical framework for determining and quantifying heteroresistance in C. glabrata.

FIG 1

Fluconazole population analysis profiles define heteroresistance phenotypes. PAP curves for C. glabrata strains with various degrees of FLC^HR are shown (A to F). Population analysis patterns were obtained by fluconazole agar dilution as described in Materials and Methods. The heterogeneity range (HR) is marked in dark blue, and the area under the PAP curve (AUC) is marked in light blue. Note that the AUCRs differ widely between strains that share the same HR values (B and C, D and E). Each data point represents the mean ± the standard error of the mean.

FIG 2

FLC^HR is distributed continuously among C. glabrata strains. The fluconazole AUCR distribution of 44 C. glabrata strains (A) shows a continuum of heteroresistance states over a wide range of values. Isolates Cg1775 and Cg1646, which were used in mouse infection studies, are represented by black squares. The correlation between the heterogeneity range and the AUCR (B) demonstrates that a dichotomous definition of FLC^HR based on an HR breakpoint of 16 correlates with an AUCR breakpoint of 0.43. However, the AUCR varies widely within each HR value category.

FIG 3

Differing PAP patterns of fluconazole, caspofungin, and anidulafungin. The PAP patterns of three representative C. glabrata strains are shown: fluconazole-susceptible nonheteroresistant (FLC^N) reference strain CBS15126, FLC^HR clinical strain Cg1462, and fluconazole-resistant clinical strain Cg1708. ANF, anidulafungin; CAS, caspofungin; FLC, fluconazole. Each data point represents the mean ± the standard error of the mean.

FIG 4

Rhodamine 6G efflux correlates with FLC^HR. (A) Rhodamine 6G efflux was measured in the presence of 8 mM glucose (+ glucose) and under glucose starvation conditions (− glucose) to differentiate ABC transmembrane transporter specific activity. White bars, FLC^N strains; dark bars, FLC^HR strains. In aggregate, FLC^HR had significantly greater ABC-type efflux activity. Yet, as with the AUCR, rhodamine 6G efflux was distributed continuously rather than bimodally between FLC^HR and FLC^N strains. RLU, relative luminescence units. (B) Correlation between rhodamine 6G efflux activity and degree of FLC^HR, expressed as FLC-AUCR. Empty circles, FLC^N strains; black squares, FLC^HR strains.

FIG 5

Visualization of intracellular rhodamine 6G accumulation. Rhodamine 6G accumulation in cells was detected by fluorescence microscopy. Panels: A to C, FLC^N isolate Cg1775; D to F, FLC^HR isolate Cg2268; G and H, FLC^R isolate Cg1708; A, D, and G, light microscopy; B, E, and H, fluorescence microscopy (rhodamine filter); C, F, and I, composite images. All images were captured at ×400 magnification by using manually fixed exposure values. DIC, differential interference contrast.

FIG 6

Expression of genes encoding efflux transporters. (A) Relative expression of efflux transporter genes CDR1, PDH1, and SNQ2 and transcription factor PDR1 in four FLC^N isolates (white bars), six FLC^HR isolates (gray bars), and two FLC^R isolates (black bars). Each dashed line marks the arbitrary 2.0-fold threshold of upregulation. *, P < 0.05; **, P < 0.01 (for the comparison of gene expression with the mean expression of four FLC^N isolates). Note that the y axis is on a different scale for each gene. (B) Induction of efflux transport-associated genes following a 3.5-h incubation in drug-free medium or medium containing fluconazole (FLC) at 8 or 64 µg/ml. *, P < 0.05; **, P < 0.01 (for relative gene expression induction following drug exposure). Expression was normalized to ACT1 and then compared to levels of expression in the same strain that was not exposed to the drug. PDR1 expression never reached the 2-fold threshold for upregulation. Each data point represents the mean ± the standard deviation of three experiments. (C) Rhodamine 6G efflux of C. glabrata strains Cg304, Cg1775 (FLC^N), and Cg2268 (FLC^HR) after incubation with fluconazole (8 µg/ml, black bars) or a saline control (gray bars). *, P < 0.05; **, P < 0.01; ***, P < 0.0001.

FIG 7

Persistence of viable C. glabrata in kidney tissue despite fluconazole treatment. Persistence of viable C. glabrata in homogenized kidneys of immunocompetent BALB/c mice was determined 7 days after tail vein infection. Diamonds, Cg1775 (FLC^N isolate); triangles, Cg1646 (FLC^HR isolate); squares, Cg1708 (FLC^R isolate); empty symbols, mice injected intraperitoneally with fluconazole 100 mg/kg/day; full symbols, mice treated with saline control. Values at the top are the numbers of mice with persistence of viable C. glabrata in the kidneys at the end of 7 days of fluconazole treatment, the total number of mice in each group, and the percentage with persistent infection.

FIG 8

Phylogenetic distances among strains with different fluconazole susceptibility phenotypes. Phylogenetic relationships were derived by comparing the sequences of the IGS region between nuclear genes CDH1 (CAGL0A00605g) and ERP6 (CAGL0A00627g) of C. glabrata strains. The cladogram was constructed by MCMC methodology. Posterior probabilities are shown at the nodes. The color gradient correlates with the degrees of FLC^HR (AUCR legend). Capital letters mark the main clusters referred to in the text. Fluconazole-resistant strains are framed.