Hsp90 governs dispersion and drug resistance of fungal biofilms

Abstract

Fungal biofilms are a major cause of human mortality and are recalcitrant to most treatments due to intrinsic drug resistance. These complex communities of multiple cell types form on indwelling medical devices and their eradication often requires surgical removal of infected devices. Here we implicate the molecular chaperone Hsp90 as a key regulator of biofilm dispersion and drug resistance. We previously established that in the leading human fungal pathogen, Candida albicans, Hsp90 enables the emergence and maintenance of drug resistance in planktonic conditions by stabilizing the protein phosphatase calcineurin and MAPK Mkc1. Hsp90 also regulates temperature-dependent C. albicans morphogenesis through repression of cAMP-PKA signalling. Here we demonstrate that genetic depletion of Hsp90 reduced C. albicans biofilm growth and maturation in vitro and impaired dispersal of biofilm cells. Further, compromising Hsp90 function in vitro abrogated resistance of C. albicans biofilms to the most widely deployed class of antifungal drugs, the azoles. Depletion of Hsp90 led to reduction of calcineurin and Mkc1 in planktonic but not biofilm conditions, suggesting that Hsp90 regulates drug resistance through different mechanisms in these distinct cellular states. Reduction of Hsp90 levels led to a marked decrease in matrix glucan levels, providing a compelling mechanism through which Hsp90 might regulate biofilm azole resistance. Impairment of Hsp90 function genetically or pharmacologically transformed fluconazole from ineffectual to highly effective in eradicating biofilms in a rat venous catheter infection model. Finally, inhibition of Hsp90 reduced resistance of biofilms of the most lethal mould, Aspergillus fumigatus, to the newest class of antifungals to reach the clinic, the echinocandins. Thus, we establish a novel mechanism regulating biofilm drug resistance and dispersion and that targeting Hsp90 provides a much-needed strategy for improving clinical outcome in the treatment of biofilm infections.

Figure 1. Compromise of Hsp90 function does not block C. albicans biofilm development in vitro.

(A) Biofilms were grown in 96-well microtiter plates in RPMI at 37°C. After 24 hours wells were washed with PBS to remove non-adherent cells and fresh RPMI medium was added containing various concentrations of the Hsp90 inhibitor geldanamycin (GdA). Biofilms were grown for an additional 24 hours at 37°C. Metabolic activity was measured using an XTT reduction assay and quantified by measuring absorbance at 490 nm. Error bars represent standard deviations of five technical replicates. Biofilm growth in the presence of GdA was not significantly different from the untreated control (P>0.05, ANOVA, Bonferroni's Multiple Comparison Test). (B) Strains of C. albicans were grown in 96-well microtiter plates in RPMI at 37°C for 24 hours with or without 20 µg/mL doxycycline (DOX). Metabolic activity was measured as in Figure 1A. Doxycycline-mediated transcriptional repression of HSP90 in the tetO-HSP90/hsp90Δ strain yielded a small reduction in biofilm growth (P<0.01). Asterisk indicates P<0.01 compared to all other conditions. Error bars represent standard deviations from five technical replicates. (C) Hsp90 levels are dramatically reduced in a C. albicans biofilm upon treatment of the tetO-HSP90/hsp90Δ strain with 20 µg/mL doyxcycline in RPMI at 37°C. Total protein was resolved by SDS-PAGE and blots were hybridized with α-Hsp90 and α-tubulin as a loading control.

Figure 2. Genetic depletion of Hsp90 does not block C. albicans biofilm formation in vivo.

The tetO-HSP90/hsp90Δ strain was inoculated in rat venous catheters in the presence or absence of 20 µg/mL doxycycline (DOX). Biofilms were examined by scanning electron microscopy imaging at 24 hours. The top row represents 50 X magnification while the bottom row represents 1,000 X magnification. Biofilm thickness and structure were similar in the presence or absence of doxycycline.

Figure 3. Pharmacological inhibition of Hsp90 alters C. albicans biofilms architecture.

C. albicans cells were grown on silicon elastomer squares in RPMI at 37°C for 24 hours. C. albicans wild-type biofilms were left untreated (A), or treated with 10 µg/mL geldanamycin (GdA) for 48 hours (B). Biofilms were stained with concanavalin A conjugate for confocal scanning laser microscopy visualization, and image reconstructions were created to provide side views (top panel). Representative images are shown. Confocal scanning laser microscopy depth views were artificially coloured (middle panel) with blue representing within 10 µm from the silicon, orange representing approximately 300 µm from the silicon, and red representing over 400 µm from the silicon. Scanning electron microscopy images are shown in bottom panel. Biofilms treated with GdA show a thinner lower layer of yeast than the untreated control.

Figure 4. Depletion of Hsp90 reduces biofilm dispersion and viability of the dispersed cell population.

C. albicans biofilms from the tetO-HSP90/hsp90Δ strain were cultured in the presence or absence of 20 µg/mL doxycycline (DOX). (A) The number of dispersed cells released from biofilms was monitored over a 24 hour period. (B) The viability of dispersed cells was determined by plating on YPD agar.

Figure 5. Inhibition of Hsp90 function dramatically enhances the efficacy of fluconazole against C. albicans biofilms in vitro.

(A) Strains of C. albicans were grown in 96-well microtiter plates in RPMI at 37°C. After 24 hours cells were washed with PBS to remove non-adherent cells and fresh medium was added with varying concentrations of the azole fluconazole (FL) and either the calcineurin inhibitor FK506 or the Hsp90 inhibitor geldanamycin (GdA) in a checkerboard format. Metabolic activity was measured as in Figure 1A. The FIC index was calculated as indicated in Table 1. Bright green represents growth above the MIC₅₀, dull green represents growth at the MIC₅₀, and black represents growth below the MIC₅₀. Data was quantitatively displayed with colour using the program Java TreeView 1.1.3 (http://jtreeview.sourceforge.net). Inhibiting calcineurin or Hsp90 function has synergistic activity with fluconazole. (B) Strains of C. albicans were grown in 96-well microtiter plates in RPMI at 37°C. When indicated, 20 µg/mL doxycycline (DOX) was added to the medium. After 24 hours cells were washed with PBS to remove non-adherent cells and fresh medium was added with varying concentrations of fluconazole. Metabolic activity was measured as in Figure 1A. Genetic depletion of Hsp90 reduces the MIC₅₀ of fluconazole to a greater extent than deletion of its client proteins calcineurin or Mkc1.

Figure 6. The Hsp90 client proteins Cna1 and Mkc1 exhibit reduced dependence on Hsp90 for stability under biofilm compared to planktonic conditions.

(A) Genetic depletion of Hsp90 does not reduce calcineurin levels in biofilm conditions. The tetO-HSP90/hsp90Δ strain with one allele of CNA1 C-terminally 6xHis-FLAG tagged was grown in planktonic or biofilm conditions with or without doxycycline (DOX, 20 µg/mL) for 48 hours. Total protein was resolved by SDS-PAGE and blots were hybridized with α-Hsp90, α-FLAG to monitor calcineurin levels, and α-tubulin as a loading control (left panel). Cna1 levels from two independent Western blots were quantified using ImageJ software (http://rsb.info.nih.gov/ij/index.html). The density of bands obtained for Cna1 was normalized relative to the density of bands for the corresponding tubulin loading control. Levels were subsequently normalized to the untreated control for the planktonic or biofilm state (right panel). (B) Depletion of Hsp90 does not deplete Mkc1 in biofilm conditions. The tetO-HSP90/hsp90Δ strain with one allele of MKC1 C-terminally 6xHis-FLAG tagged was grown in planktonic or biofilm conditions with or without DOX for 48 hours. Total protein was resolved by SDS-PAGE and blots were hybridized with α-Hsp90, α-His₆ to monitor Mkc1 levels, α-phospho-p44/42 to monitor dually phosphorylated Mkc1, and α-tubulin as a loading control (left panel). Mkc1 levels from two independent Western blots were quantified using ImageJ software. The density of bands for Mkc1 was normalized relative to the density of bands for the tubulin loading control. Levels were subsequently normalized to the untreated control for the planktonic or biofilm state (right panel).

Figure 7. Depletion of Hsp90 reduces biofilm matrix glucan.

Strains of C. albicans were cultured in 6-well polystyrene dishes for 48 hours with or without 20 µg/mL doxycycline (DOX). Matrix samples were collected and matrix β-1,3 glucan levels were meausured using a limulus lysate based assay. Asterisk indicates P<0.01 (ANOVA, Bonferroni's Multiple Comparison Test) compared to all other conditions.

Figure 8. Compromise of Hsp90 function genetically or pharmacologically enhances the efficacy of fluconazole in vivo.

(A) The tetO-HSP90/hsp90Δ strain was inoculated in rat venous catheters for 24 hours with or without 20 µg/mL doxycycline (DOX) followed by intraluminal azole treatment for an additional 24 hours. Following drug exposure, catheters were removed for visualization by scanning electron microscopy. The first column represents treatment with 250 µg/mL fluconazole (FL), followed by treatment with both 20 µg/mL DOX and 250 µg/mL FL. The top row represents 50 X magnification and the bottom row represents 1,000 X magnification. The combination of FL and DOX abrogates biofilms. (B) Biofilms were cultured as in A with 250 µg/mL FL, 100 µg/mL 17-AAG, or the combination of drugs. Serial dilutions of the catheter fluid were plated for viable fungal colony counts. Results are expressed as the mean colony forming unit (CFU) per catheter. The combination of FL and 17-AAG reduces fungal burden in the catheter compared to individual drug treatments (Asterisk indicates P<0.001, ANOVA, Bonferroni's Multiple Comparison Test).

Figure 9. Pharmacological inhibition of Hsp90 enhances the efficacy of echinocandins and azoles against A. fumigatus biofilms and affects biofilm morphology.

(A) A. fumigatus was grown in 96-well dishes in RPMI at 37°C. After 24 hours cells were washed with PBS to remove non-adherent cells and fresh medium was added with varying concentrations of the echinocandin caspofungin (CF), the azole voriconazole (VL), and the Hsp90 inhibitor geldanamycin (GdA) in a checkerboard format. Drug treatment was left on for 24 hours. Metabolic activity was measured as in Figure 1A. The FIC index was calculated as indicated in Table 2. Bright green represents growth above the MIC₅₀, dull green represents growth at the MIC₅₀, and black represents growth below the MIC₅₀. (B) A. fumigatus cells were left untreated, or treated with 32 µg/mL CF or 256 µg/mL VL in the absence and presence of 50 µg/mL GdA for 24 hours. Following drug exposure, biofilms were fixed and imaged by scanning electron microscopy. Biofilms treated with antifungal show increased cellular damage in the presence of GdA. The white arrows indicate burst and broken hyphae in the biofilms treated with CF and GdA.