Fungal biofilm architecture produces hypoxic microenvironments that drive antifungal resistance

Abstract

Human fungal infections may fail to respond to contemporary antifungal therapies in vivo despite in vitro fungal isolate drug susceptibility. Such a discrepancy between in vitro antimicrobial susceptibility and in vivo treatment outcomes is partially explained by microbes adopting a drug-resistant biofilm mode of growth during infection. The filamentous fungal pathogen Aspergillus fumigatus forms biofilms in vivo, and during biofilm growth it has reduced susceptibility to all three classes of contemporary antifungal drugs. Specific features of filamentous fungal biofilms that drive antifungal drug resistance remain largely unknown. In this study, we applied a fluorescence microscopy approach coupled with transcriptional bioreporters to define spatial and temporal oxygen gradients and single-cell metabolic activity within A. fumigatus biofilms. Oxygen gradients inevitably arise during A. fumigatus biofilm maturation and are both critical for, and the result of, A. fumigatus late-stage biofilm architecture. We observe that these self-induced hypoxic microenvironments not only contribute to filamentous fungal biofilm maturation but also drive resistance to antifungal treatment. Decreasing oxygen levels toward the base of A. fumigatus biofilms increases antifungal drug resistance. Our results define a previously unknown mechanistic link between filamentous fungal biofilm physiology and contemporary antifungal drug resistance. Moreover, we demonstrate that drug resistance mediated by dynamic oxygen gradients, found in many bacterial biofilms, also extends to the fungal kingdom. The conservation of hypoxic drug-resistant niches in bacterial and fungal biofilms is thus a promising target for improving antimicrobial therapy efficacy.

Keywords: Aspergillus fumigatus; antifungals; drug resistance; hypoxia; oxygen.

Fig. 1.

Oxygen gradients within submerged fungal biofilms are necessary for maturation. (A) Representative 3D renderings from n = 3 independent biological samples depicting the side view (XZ) of mature biofilms cultured at various oxygen tensions on normal and oxygen-permeable plates. (Scale bar, 200 µm.) (B) Heat map illustrating the alteration in hyphal architecture within biofilms cultured on normal plates as a result of reduced oxygen tensions, where architecture is defined as the angle hyphae deviate from a vertical axis. Each column is representative of n = 3 independent biological samples. (C) Heat map illustrating the altered distribution of fungal biomass as a function of height in biofilms at various oxygen tensions, as shown in A. Each column is representative of n = 3 independent biological samples. (D) Representative 3D renderings of 24-h biofilms from n = 3 independent biological samples have a collapsed architecture when cultured on an oxygen permeable surface and sealing of the surface with an oxygen-impermeable seal restores the normal plate architecture. (Scale bar, 200 µm.)

Fig. 2.

Hypoxia occurs during maturation of A. fumigatus biofilms. (A) Measurements of dissolved oxygen within developing fungal biofilms from n = 3 independent biological samples. Error bars depict SEM. (B) Extrapolation from data in A at 2,500 µm depth shows oxygen tensions gradually and significantly reduce throughout biofilm maturation. Error bars indicate SEM of n = 3 independent biological samples (one-way ANOVA with Tukey’s multiple comparisons test). (C) Computational modeling of oxygen zones within a fungal biofilm as a function of oxygen consumption and biomass. White zone: 5% < O₂ ≤ 21%, red/hypoxic zone: 0% < O₂ ≤ 5%, gray/anoxic zone: O₂ = 0%. (D) Representative images (n = 3 independent biological samples) of mutant strains of A. fumigatus with partial (CEA10 ∆srbB) or absolute (CEA10 ∆srbA; AF293 ∆srbA) low-oxygen growth defects form stunted 24-h biofilms in normoxia. The hypoxia-fit strain hrmA^R-EV forms a more robust 24-h biofilm compared to the control AF293 (representative of n = 3 independent biological samples). (Scale bar, 500 µm.)

Fig. 3.

The spatial and temporal hypoxic zone coincides with a reduced metabolic state. (A) Single representative 3D rendered micrographs of n = 3 independent biological samples show visible _perg25A-GFP reporter signal at 16-h, 18-h, and 24-h biofilms but absent at 12 h. (Scale bar, 100 µm.) (B) Heat map of the hypoxia-reporter signal relative to fungal biomass. Each column is representative of n = 3 independent biological samples. (C) Single representative 3D rendered micrographs of n = 3 independent biological samples show an absence of _perg25A-GFP reporter signal in 16-h biofilms cultured on oxygen-permeable surfaces. (Scale bar, 100 µm.) (D) Heat map of the hypoxia-reporter signal relative to fungal biomass for 16-h biofilms grown on normal or oxygen permeable surfaces. Each column is representative of n = 3 independent biological samples. (E) Xylose-inducible GFP expression indicates translational activity within the vertically grown hyphae of 18-h biofilms after 3 h of induction but absent from the horizontal hyphae at the base. Images are representative 3D renderings from n = 3 independent biological samples. (Scale bar, 100 µm.) (F) Heat map quantifying GFP fluorescence relative to overall biomass from images in A. Each column is representative of a single sample from n = 3 biological samples.

Fig. 4.

Hypoxia contributes to biofilm antifungal drug resistance. (A) Biofilms become increasingly resistant to exogenous stressors (voriconazole: 1 µg/mL, amphotericin B: 1 µg/mL, menadione: 10 µM) throughout maturation. Error bars indicate SEM with n = 3 independent biological samples. (B) Eighteen-hour biofilms remain largely resistant to treatment with voriconazole (1 µg/mL) or amphotericin B (1 µg/mL) with (AF293) or without (∆uge3) the production of the exopolysaccharide galactosaminogalactan. Error bars indicate SEM for n = 4 biologically independent samples. One-way ANOVAs performed with Tukey’s multiple comparison test. (ns: P > 0.05, not significant). (C) Twelve-hour biofilms are significantly more susceptible to antifungals (voriconazole: 1 µg/mL, amphotericin B: 1 µg/mL, menadione: 10 µM) than 18-h biofilms in normoxia (21% O₂). Treatment in hypoxia (0.2% O₂) significantly reduces the susceptibility of 12-h biofilms but not 18-h biofilms. Error bars indicate SEM of n = 4 independent biological samples. Each treatment group was analyzed separately with a one-way ANOVA with Tukey’s multiple comparison test. (D) Eighteen-hour biofilms on oxygen-permeable plates have significantly more oxygen at a depth of 3 mm. Error bars indicate SEM of n = 3 independent biological samples. Student’s t test performed between samples at each depth. (E) Eighteen-hour biofilms are significantly more susceptible to treatment on oxygen-permeable surfaces. Error bars indicate SEM of n = 5 or n = 6 independent biological samples. Student’s unpaired two-tailed t test performed between samples within each treatment group.

Fig. 5.

Respiration-induced hypoxic zones reoxygenate after treatment to facilitate biofilm growth and drug resistance. (A) FCCP (2.5 µM) significantly reduces dissolved oxygen in 24-h biofilms after 30 min. Error bars indicate SEM of n = 3 independent biological samples. Student’s unpaired two-tailed t test performed. (B) FCCP significantly reduces amphotericin B damage in 12-h biofilms. Error bars indicate SEM of n = 4 independent biological samples. One-way ANOVA with Sidak’s multiple comparison test performed. (C) Incubation in RMM that contains TCA intermediates significantly reduces dissolved oxygen in 24-h biofilms compared to incubation in GMM. Error bars indicate SEM of n = 3 independent biological samples. Student’s unpaired two-tailed t test performed. (D) Treatment of 12-h biofilms in RMM significantly reduces amphotericin B damage compared to treatment in GMM. Error bars indicate SEM of n = 4 independent biological samples. Student’s unpaired two-tailed t test performed. (ns = P > 0.05, not significant). (E) Antifungal treatment with voriconazole (1 µg/mL) or amphotericin B (1 µg/mL) significantly increases dissolved oxygen within the 24-h biofilms after 90 min of treatment. Error bars indicate SEM of n = 4 independent biological samples. Comparisons were made using a multiple t test approach with the Holm–Sidak method to determine statistical significance: (a) P = 0.003190, (b) P = 0.003288, (c) P = 0.000443, (d) P = 0.001435, (e) P = 0.003119, (f) P = 0.000011, (g) P = 0.000312, (h) P = 0.003331. Comparisons without a P value listed are P > 0.05 and not statistically significant. (F) Voriconazole (1 µg/mL) treatment increases the xylose-inducible signal for translation activity at the base of the biofilm. Images representative of n = 3 independent biological samples. (Scale bar, 100 µm.) (G) Heat map quantifying GFP fluorescence relative to overall biomass from images in F shows visible signal at the base of the 18-h biofilm following voriconazole treatment. Each column is representative of n = 3 independent biological samples. (H) Amphotericin B (1 µg/mL) treatment largely reduces the xylose-inducible signal for translation activity. Visible signal is localized only at the base of the 18-h biofilm. Images representative of n = 3 independent biological samples. (Scale bar, 100 µm.) Each column is representative of n = 3 independent biological samples. (I) Heat map quantifying GFP fluorescence relative to overall biomass from images in H shows visible signal at the base of the 18-h biofilm following amphotericin B treatment. Each column is representative of n = 3 independent biological samples. (J) Consecutive treatments with voriconazole. or amphotericin B in normoxia significantly increases the damage to 18-h AF293 biofilms, while consecutive treatments in hypoxia cause significantly less damage than in normoxia. Error bars indicate SEM of n = 6 independent biological samples. One-way ANOVA with Tukey’s multiple comparisons test performed. (ns: P > 0.05, not significant).