Aspergillus fumigatus strain ATCC 13073 (American Type Culture Collection, Manassas, VA, United States), originally isolated from a human pulmonary lesion, was used throughout this study as the wild type (WT) or parental strain. Unless stated otherwise, fungi were grown at 37°C on yeast extract-agar-glucose (YAG, 0.5% yeast extract, 2% glucose, 0.125% MgSO₄∙7H₂O, 0.1% Hutner’s trace elements, 0.1% Vitamins, 1.5% agar) or Kafer’s minimal medium (Hill and Kafer, 2001) supplemented with Hutner’s trace elements and vitamins. All strains were stored on silica gel at -80°C.

Mature conidia were harvested by flooding an agar plate with phosphate-buffered saline, pH 7.4 supplemented with 0.05% Tween 20 (PBST). The surface was gently agitated with a sterile cotton swab and the suspension was filtered through sterile Miracloth (Calbiochem) to remove hyphal fragments. Conidia were then washed three times in sterile PBS. Spores were counted with a haemocytometer.

Kafers minimal media with a 1% Hutner’s vitamin mixture was used to compare the growth of wild type and Δkdnase A. fumigatus strains. The carbon source was changed as indicated. Equivalent numbers of conidia were inoculated into the center of each plate and plates were incubated for 72 h at 37°C. Hygromycin (100 μg/mL) was the positive selection marker. To test the effect of sorbitol on growth, YAG solid media plates (with or without 1 M sorbitol) were inoculated with 10⁵ conidia, in a volume of 5 μL in the center of each plate. Slide cultures were also prepared for each strain on YAG supplemented with 750 μg/mL Congo Red dye and incubated for 20 h at 37°C. Images were obtained on a Zeiss inverted microscope at 100× magnification.

Growth assays were prepared and monitored following the EUCAST protocol (EUCAST E.DEF 9.3 December 2015) modified by addition of resazurin to increase the sensitivity of growth measurement (Yamaguchi et al., 2002). Assays were prepared in 96-well microdilution plates with RPMI 1640 media buffered with 3-(N-morpholino)propanesulfonic acid (MOPS) at a final concentration of 0.165 M, pH 7.0, and 2 × 10⁵ conidia/mL. Antifungal compounds were prepared in dimethyl sulfoxide (DMSO) to a maximum of 2.5% v/v. Control wells contained DMSO alone. Resazurin stock solution (Sigma–Aldrich) was prepared in PBS and filter sterilized, and media were prepared with a final concentration of 0.02% resazurin. Plates were incubated in the dark at 37°C. Growth was quantified in a SpectraMax plate reader by measuring resorufin fluorescence after 36–48 h of incubation (excitation wavelength 560 nm; emission wavelength 590 nm). Control plates were done in parallel that contained all components except conidia confirmed that antibiotics did not affect resorufin fluorescence.

The kdnase knockout was created by homologous recombination using a disruption construct containing 1000 bp sequences of DNA encoding the upstream and downstream regions of the kdnase gene, flanking a positive selectable marker for hygromycin resistance (Supplementary Figure S1). The disruption construct was generated using fusion PCR and introduced into the wild type strain using Agrobacterium-mediated transformation as follows. The hygromycin resistance gene cassette containing the Escherichia coli (hygromycin phosphotransferase, hph) under the control of the constitutive trpC promoter and terminator from Aspergillus nidulans (Hamert and Timberlaket, 1987) was amplified with the primers 5′- AAACATTCCCACTATCGCGA TCGTCGACGTTAACTGATATTGA and 5′-TAATGCTAGTCCAGCAACGACGTCGAC GTTAACTGGTTCC. The upstream and downstream 1000 base pairs of DNA that flank the kdnase gene in were amplified by PCR with the following primer pairs, upstream: 5′- GCGCTCCTATCCAGTCAGTC; 5′-TCGCGATAGTGGGAATGTTT and downstream: 5′- TCGTTGCTGGACAGCATTAT; 5′-TGCTTCATGTCATGCCTAGC. The hph cassette and kdnase flanking regions were amplified by PCR using iProof^TM High-fidelity DNA polymerase (Bio-Rad, Mississauga, ON, Canada). Fusion PCR was performed with high-fidelity AccuPrime Taq DNA polymerase (Thermo Fisher Scientific, Mississauga, ON, Canada) using the primers 5′-GGGATCCACATACCATTCTCGCCGAAC and 5′-GGGATCCAATGCTA CGGGAACACTTGG that added a BamH1 restriction site to facilitate insertion into the pCambia plasmid. The fusion PCR fragment was gel purified and cloned into pCambia 0380 plasmid (Cambia) yielding pCambia-KdnaseKO. Competent Agrobacterium tumefaciens (AGL-1) were transformed using heat shock. Transformation of A. fumigatus followed the protocol of Sánchez and Aguirre (1996) with modifications. Putative fungal transformants were harvested using a punch transfer tool and placed on fresh YAG agar plates containing 100 μg/mL hygromycin. Plates were incubated at 37°C for 72 h before mature spores were harvested. Conidial suspensions were then diluted and re-plated with an estimated 10 spores per plate to generate plates with colonies grown from single spores.

XhoI-digested DNA from the WT and mutant strains was analyzed by Southern blotting to ensure that only a single copy of the kdnase gene was integrated into the genome and that the integration was at the kdnase locus (Supplementary Figure S2). The knockout strain was also sequenced at the kdnase locus to ensure that up- and downstream sequences were not affected. Successful homologous integration of the knockout construct in the putative mutants was also confirmed by PCR (Supplementary Figure S3).

To construct Δkdnase^R, the rescued strain, iProof polymerase (Bio-Rad) was used to amplify 2470 base pairs of the wild type kdnase gene plus 928 bp of upstream sequence [including TATA and CAAT boxes (Warwas et al., 2010)] and 322 bp of downstream sequence.

The rescue construct was amplified with the primers 5′-TGCAAGCTTCAGA ACTGCCCTTTCCCTCT-3′ and 5′-GTTAAGCTTACTGTTTTGCTGCCCTCTTC-3′ containing HindIII restriction sites to facilitate insertion into pBCPhleo (Silar, 1995). pBC-phleo-Kdnase was used to transform E. coli DH5α and plasmid DNA was linearized with SpeI (New England Biolabs, Whitby, ON, Canada). Linearized plasmid DNA (10 μg) was introduced into 50 μL of swollen conidia by electroporation. Briefly, ice-cold swollen conidia were pelleted by centrifugation and washed 3 times with 4°C sterile water after which they were re-suspended in 4 mL YED media (1% yeast extract, 1% glucose, 20 mM HEPES, pH 8) and incubated for 1 h at 30°C with gentle shaking. Following incubation, conidia were centrifuged and re-suspended in 150 μl of electroporation buffer (10 mM Tris–HCL pH7.5, 270 mM sucrose, and 1 mM lithium acetate). The conidial suspension was then divided into 50 μl aliquots for use in transformation. The restriction enzyme SpeI (New England Biolabs) was used to linearize pBC-phleo-Kdnase. Linearized plasmid DNA (10 μg) was added to 50 μl of prepared conidia and the mixture was incubated for 15 min on ice before being transferred to an electroporation cuvette with a 2 mm gap. Conidia were exposed to an electric field pulse of 1 kV, 400 ohms, and 25 μF. Immediately following the pulse, 1 mL of ice-cold YED media was added to the conidia and they were cooled on ice for 15 min before being allowed to recover for 2 h at 37°C with gentle shaking. Transformants were selected on YAG media containing 100 μg/mL phleomycin. Putative transformants were subcultured onto selective media to obtain pure cultures. Ectopic integration of the rescue construct was confirmed using PCR (Supplementary Figure S3).

Petri dishes containing YAG agar or YAG + 1M sorbitol agar were prepared and overlaid with a sterile disk of cellophane. Conidia of the appropriate fungal strain (1 × 10⁶) were spread evenly over the surface of the cellophane. Each plate was incubated at 30°C for 20 h than at 37°C for 5 h and flooded with a fixative solution containing 2.5% EM-grade glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.4. Plates were incubated at room temperature for 1 h, washed with sodium cacodylate buffer and a 1-cm² piece of cellophane was transferred to a microcentrifuge tube filled with 0.1 sodium cacodylate buffer. Samples were dehydrated in an ethanol series, followed by 1:1 solution of ethanol/propylene oxide (1:1) then pure propylene oxide. The samples were soaked in a series of propylene oxide/Spurr resin mixtures (3:1, 1:1, 1:3, 0:1). Each sample was incubated in 100% resin for 2 days with resin replaced twice each day. Resin was polymerized at 70°C for 8 h. Samples were examined on a Hitachi H-7600 transmission electron microscope. From the TEM images of hyphae, cell wall thickness was quantified with ImageJ using the measure function. A random number generator produced 5 sets of x,y co-ordinates for each image. ImageJ was used to locate coordinates and measure the thickness of the cell wall directly across using the measurement tool (calibrated to each scale bar). If the chosen co-ordinate was not on the cell wall, the line tool was used to find the shortest straight line to the cell wall, and the wall measurement was made at that point where the angle of intersection with each side of the cell wall as close to 90 degrees as possible. Measurements were performed blinded on coded images. The number of separate images analyzed for each condition was as follows: WT (17), Δkdnase (17), WT + 1M sorbitol (12), Δkdnase + 1 M sorbitol (24).

Conidia (5 × 10⁶) from each strain were incubated in RPMI 1640 overnight at 37°C on sterile round coverslips (25 mm). RPMI was replaced with 3 mL of a solution containing 0.075% Alcian Blue, 75 mM lysine, 3% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.2 and further incubated at 4°C overnight. Addition of Alcian Blue and lysine have been shown to enhance visualization of the polyanionic exopolysaccharides (Fischer et al., 2005). Samples were washed 3 times each with 0.1 M sodium cacodylate buffer, pH 7.2, and distilled water. After dehydration in ethanol (25, 50, 75, 90, 100, and 100% for 10 min each), samples were critical point dried, coated with iridium (5 ± 1 nm) and examined on a Nova NanoSEM (FEI).

All experiments were carried out in accordance with the Canadian Council on Animal Care (CCAC) guidelines. This study was approved by the University Animal Care Committee at Simon Fraser University (protocol 1193B-11). The mouse model of pulmonary IA was modified from (Dixon et al., 1989). Briefly, female BALB/cJ mice (∼3 months of age) from Jackson Laboratory (Bar Harbor, ME, United States) were acclimatized for 2 weeks prior to experimentation. Mice were immunocompromised with intraperitoneal injections of 0.025 g/ml cyclophosphamide (Toronto Research Chemicals, North York, ON, Canada) in PBS (100 mg/kg) administered on day 4 and day 0, and subcutaneous injections of 0.05 g/mL cortisone acetate (Sigma–Aldrich Canada Co., Oakville, ON, Canada) in PBST (200 mg/kg) on day 1. Experimental groups were inoculated intranasally with 1.0 × 10⁶ fungal spores in 40 μL aliquots, as follows: seven mice received wild type Aspergillus fumigatus spores, seven mice received wild type Aspergillus fumigatus spores followed by a dose of 5 mg/kg of liposomal amphotericin B (AmBisome) (AB) (Gilead Sciences Inc., Foster City, CA, United States), seven mice received Δkdnase A. fumigatus spores, seven mice received Δkdnase A. fumigatus spores and 5 mg/kg AB, seven mice received Δkdnase^R strain A. fumigatus spores, three mice received saline, and four mice received saline and 5 mg/kg AB. AB was administered 4 h after spore inoculation on day 0; the drug was resuspended in water just prior to use, and administered intravenously.

Mice were maintained in a specific pathogen-free facility, four to a cage, with alternating 12-h light schedules. Mice were housed in autoclaved cages and received irradiated chow ad libitum (equal parts mixture of PicoLab Mouse Diet 20 and Picolab Rodent Diet 20). Mice were provided with a tetracycline water (0.5 g/L) mixture sweetened with aspartame (Sigma–Aldrich Canada Co., Oakville, ON, Canada). Mice were monitored and weighed daily. At experimental end-points (>20% weight loss, swelling around the eyes, piloerection, hunched posture, limited mobility and activity, rapid breathing or gasping, and antisocial behavior), mice were humanely euthanized by CO₂ followed by removal of lungs. After euthanasia, perfusion with ice cold PBS was performed to clear blood from mouse lungs. The left lobe was fixed overnight at 4°C in paraformaldehyde, then transferred to an ethanol solution, and prepared for histology. The right lobe was split into two, one section was stored in TRIzol (Invitrogen) at 4°C for cytokine analysis via qPCR, the other section was frozen in a pre-weighed bag at -20°C for DNA analysis for fungal burden.

Fixed lungs were sectioned and stained with hematoxylin and eosin (H&E) or with Grocott’s methenamine silver (GMS) to visualize fungal elements (hyphae and conidia). Fungal elements were quantified as total fungal element area using the color threshold tool in ImageJ on GMS-stained sections, and expressed as percent fungal element coverage of total lung area.

Lung sections (4 μm thick) were de-paraffinized with 3 xylene washes and re-hydrated by successive washes in 100, 90, and 70% ethanol, and then finally in dH₂O. Heat-based antigen retrieval was performed by heating slides in citrate buffer (pH 6.0) to 120°C in a pressure cooker. Slides were then incubated with 5% donkey serum for 30 min and endogenous peroxidase was inhibited with 0.3% H₂O₂ treatment for 20 min. Slides were incubated with the primary antibody overnight, the biotinylated secondary antibody for 30 min, and horseradish peroxidase-conjugated avidin-biotin complex (Vector Laboratories, Burlingame, CA, United States) for 45 min. Staining was visualized with 3-amino-9-ethyl carbazole (AEC) substrate chromagen, which results in red staining (Vector Laboratories). Sections were then counterstained with hematoxylin. The following primary antibodies were used: Mac-3 (5 μg/mL) (BD Biosciences, Franklin Lakes, NJ, United States) to visualize macrophages and myeloperoxidase (MPO) to visualize neutrophils (5 μg/mL) (Abcam, Cambridge, MA, United States). Immune cells were manually counted on a microscope. Quantification of the staining was performed in a manner that was blinded from knowledge of specific mouse IDs. The total lung area for each section was quantified in ImageJ and Mac-3-positive or MPO-positive cells were expressed per total area (mm²).

Analysis was based on the method of Bowman et al. (2001). Frozen lung samples in pre-weighed bags were minced and homogenized in cold saline (7.5 volumes/g lung) using a roller bottle, and transferred to microcentrifuge tubes. Secondary homogenization was carried out in a TissueLyser II (3 sets of 30 s at 30 oscillations/s) and centrifuged for 5 min at 800g and 4°C. Genomic DNA was extracted using the CTAB method (Carlson et al., 1991) and fungal DNA was amplified by TaqMan quantitative PCR (Life Technologies). We targeted the internal transcribed spacer region, ITS1 and the 5.8S gene located between the 18S and 28S rRNA genes, using the following primers: Forward Primer - TCTGAAAGTATGCAGTCTGAGTTGATT; Reverse Primer - GATGCCGGAACCAAGAGATC; Probe - CGTAATCAGTTAAAACTTTC. A C_t value was determined for triplicates from each sample, based on a standard curve also created using triplicate samples with genomic DNA from A. fumigatus (0.001–1.000 ng/μL).

To determine the proportion of conidia internalized by macrophages, we labeled conidia with fluorescein isothiocyanate (FITC) (470 nm). Bound/external conidia were detected by an anti-Aspergillus antibody conjugated to a red fluorophore (590 nm) added to the cell suspension at the end of the incubation period. All strains were grown on YAG media supplemented with appropriate antibiotics (100 μg/mL hygromycin for the Δkdnase strain and 50 μg/mL phleomycin for Δkdnase^R) for 72 h at 37°C and harvested in PBST as described above. To label conidia with FITC, we followed the procedure of Sturtevant and Latge (1992). The J774A.1 cell mouse macrophage cell line, originally derived from BALB/cN mice (ATCC, Manassas, VA, United States) were grown on coverslips in 24-well plates in RPMI 1640 media containing 10% fetal bovine serum (RPMI-FBS). Macrophages (2.5 × 10⁵ cells) were incubated at 37°C + 5% CO2 for 3 h and blocked in RPMI-FBS containing 0.5% BSA for 1 h at 37°C. The medium was aspirated and FITC-labeled conidia from each strain (2 × 10⁶, suspended in 1 mL RPMI-FBS) were added to the wells. Each strain was inoculated into separate wells and no spore control and no primary antibody control wells were also included. Plates were incubated for 1 h at 37°C in 5% CO₂ then placed on ice. On ice, J774 cells were washed 3× with PBST and incubated for 1 h with a 1:75 dilution of polyclonal rabbit anti-Aspergillus cell wall antibody (Wasylnka and Moore, 2002) in PBS plus 10% goat serum (GS). After washing 3 times, Alexa Fluor 594-conjugated AffiniPure goat anti-rabbit IgG (H+L) (Jackson ImmunoResearch) was added in 1:575 dilution in PBS-GS. After incubation for 1 h, cells were washed with PBS and fixed in PBS-4% formalin for 1 h. Washed coverslips were mounted with Prolong Antifade Mounting Media, sealed and stored in the dark at 4°C. Images were obtained a Zeiss Axioplan epifluorescence microscope.

Macrophage killing of conidia was measured in vitro using the nystatin protection assay (Wasylnka and Moore, 2002), except that J774A.1 cells were incubated in RPMI-FBS containing 0, 10, or 20 μg/mL hydrocortisone dissolved in DMSO for 24 h prior to adding conidia. Control samples were treated with DMSO.

The kdnase gene on chromosome 4 was disrupted by Agrobacterium-mediated transformation of A. fumigatus using a fusion PCR construct that contained a hygromycin resistance cassette for positive selection flanked by 1000 bp kdnase up- and downstream sequences (Supplementary Figure S1). Successful generation of a Δkdnase strain was determined by PCR and Southern blotting. The Southern blot confirmed homologous insertion of the knockout construct at the kdnase locus (Supplementary Figure S2). The complemented mutant, Δkdnase^R or ‘rescued’ strain, was produced by electroporation of the knockout with the wild type (WT) kdnase gene controlled by its native promoter (928 bp of upstream sequence). Successful kdnase knockout and ectopic insertion of the WT kdnase gene into the genome of the Δkdnase strain was confirmed by PCR (Supplementary Figure S3).

A growth assay was performed to determine how the loss of the kdnase gene affected the nutritional requirements of A. fumigatus. Strains were grown for 72 h on Kafer’s minimal medium containing only the specified carbon source. When grown on minimal media supplemented with Kdn, glucose, or mannose, the growth of the Δkdnase strain was inhibited compared to WT (Figure 1). Compared to Kdn, Neu5Ac (labeled sialic acid in Figure 1) is a poor carbon source for A. fumigatus; however, the Δkdnase strain had even lower growth than WT on Neu5Ac. These data also confirm our previous finding that Kdn is strongly preferred to Neu5Ac in A. fumigatus (Telford et al., 2011).

We examined the relative growth and morphology of the WT and mutant strains in the presence of a hyperosmotic stress generated by addition of 1 M sorbitol to rich medium (YAG). Growth of the WT and Δkdnase strains was similar on solid YAG medium although the morphology of the KO strain was abnormal. However, the Δkdnase strain showed poor growth on YAG supplemented with sorbitol (Figure 2). A dose response experiment showed that growth inhibition was evident at concentrations of sorbitol of 0.4 M (data not shown). DIC images of slide cultures of the two strains showed that the morphology of the Δkdnase hyphae was abnormal even in YAG alone with hyperbranching and swollen filaments (Figure 2 and Supplementary Figure S4). Quantitative analysis in RPMI-sorbitol medium showed that growth of the KO strain was <10% of WT after 48 h though some recovery was evident after 72 h (Supplementary Figure S4).

We examined the surface structure of conidia and hyphae of the WT, Δkdnase, and Δkdnase^R strains grown in RPMI using scanning electron microscopy. No apparent structural changes in conidia were evident (data not shown); however, Δkdnase hyphae had a less dense glycocalyx, a structure that is principally composed of polyanionic exo-polysaccharides. The rescued strain had an intermediate phenotype (Supplementary Figure S6). We quantified galactosaminogalactan (GAG) content of the three strains using lectin staining [soybean agglutinin (SBA) linked to a fluorophore] but found no significant differences (data not shown). Hence, changes in surface structure in the knockout cannot be attributed to alterations in GAG content. TEM images of hyphae showed that the cell walls of the Δkdnase hyphae (panels A–D in Figure 3A) appeared thicker than those of WT (panels E–H in Figure 3A) when strains were grown in medium supplemented with sorbitol (compare panels C,D with panels G,H); this was confirmed by quantitative analysis of the images (Figure 3B). These data suggest that compensatory cell wall thickening occurred in response to the hyperosmotic stress in the Δkdnase strain but not the WT strain.

Chitin is an important structural polysaccharide of the A. fumigatus cell wall (Lee and Sheppard, 2016) and its synthesis is upregulated under conditions of cell wall stress. The dyes Congo Red (CR) and Calcofluor White (CW) were used to probe changes in the cell wall chitin content as these dyes bind structural polysaccharides with a preference for chitin (Roncero and Duran, 1985). Figure 4A and Supplementary Figure S5 show that the Δkdnase strain was able to grow in concentrations of CR that almost completely inhibited WT growth on solid media. The rescued strain had an intermediate phenotype. The results with CW were similar: After 72 h of growth on YAG supplemented with 0.25% CW, the WT strain showed no growth whereas the Δkdnase strain reached ∼25 mm of radial growth. The Δkdnase^R strain grew to only 30% of the level of the kdnase knockout strain (Figure 4B).

To directly quantify the levels of selected cell wall polysaccharides, strains were grown in slide culture on YAG medium with or without sorbitol (1 M) and immunofluorescence microscopy was used to measure the β(1,3)-glucan, α(1,3)-glucan and chitin contents of the hyphae. Figure 5 shows that although no differences were detected in β(1,3)-glucan, α(1,3)-glucan reactivity was significantly increased in Δkdnase hyphae in the presence or absence of sorbitol by approximately 30% (Figures 5A,B). Binding of the chitin probe was dramatically different in the three strains; the Δkdnase strain grown in sorbitol had a threefold higher level of fluorescence compared to all other strains and conditions (Figure 5C and Supplementary Figure S7). These data suggest that resistance to CR and CW and the observed cell wall thickening in hyperosmotic conditions may be a consequence of increased chitin deposition in the cell wall of the Δkdnase strain. Alternatively, the increased fluorescence may reflect increased access of the probe to chitin as a consequence of cell wall disorganization in the Δkdnase strain.

We next tested the relative susceptibility to several antifungal agents. Nikkomycin Z is a competitive inhibitor of chitin synthase (Hector, 1993). The Δkdnase strain was more resistant to nikkomycin than WT (Figure 6A) even though WT A. fumigatus strains already have high minimum inhibitory concentration values for nikkomycin (16–64 μg/mL) that limit its clinical use (Ganesan et al., 2004). The Δkdnase strain also showed some resistance to voriconazole, a drug that alters membrane sterol composition by inhibiting cytochrome P450 lanosterol demethylase (Odds et al., 2003), but resistance was only detected in a narrow voriconazole concentration (Figure 6B). We did not detect a difference in the susceptibility of the Δkdnase strain to itraconazole, another triazole antifungal (data not shown). In contrast, Δkdnase was more susceptible than WT to both caspofungin and amphotericin B (Figures 6C,D). Caspofungin is an echinocandin antifungal that inhibits the Fks1 subunit of β(1,3)-glucan synthase (Odds et al., 2003). It is the only anti-fungal agent that directly targets the cell wall. At caspofungin concentrations <1 μg/mL, no difference in growth was observed between the two strains (data not shown); however, at >5 μg/mL caspofungin (concentrations at which ‘paradoxical’ growth occurs), Δkdnase growth was marginally, but significantly, reduced compared to the WT strain (Figure 6). Finally, Δkdnase also showed increased sensitivity to Amphotericin B (AB) compared to WT (∼ 6-fold at 0.7–0.8 μg/mL) (Figure 6). AB is a pore-forming polyene antibiotic used as an alternative to voriconazole in primary treatment of IA. Because of the clinical importance of AB and the in vitro sensitivity of the Δkdnase knockout strain, we included AB treatment in our test of AfΔkdnase virulence.

To test the virulence of the Δkdnase knockout strain and the efficacy of AB treatment in vivo, female BALB/cJ mice were first immunosuppressed with cortisone and cyclophosphamide as described in the Section “Materials and Methods.” On day 0, mice in all treatment groups (seven per group) received 10⁶ conidia intranasally of the appropriate strain. Two groups of animals were also treated with AmBisome (AB) (5 mg/kg i.v.) 4 h after exposure to Δkdnase (KO), or Δkdnase^R (R) conidia. Two additional groups of control mice received the same immunosuppressive regime and were exposed to saline intranasally (n = 3) or saline followed by AB treatment (n = 4). Control mice (not infected with A. fumigatus) had 100% survival (data not shown). Kaplan-Meier survival curves of immunosuppressed mice exposed to wild type (WT), Δkdnase (KO), or Δkdnase^R (R) conidia with or without AmBisome treatment are shown in Figure 7. All mice were weighed and monitored by a blinded observer twice per day for up to 10 days (experimental endpoint): Mice showing signs of morbidity approaching a humane endpoint were sacrificed. Most mice reached a humane endpoint by day 4 with the exception of one infected with the Δkdnase strain (KO) (1/7 survived) and four mice infected with the Δkdnase strain that also received AmBisome treatment following spore exposure (KO+AB) (3/7 survived for 10 days). These surviving mice also gained weight and had no apparent symptoms (data not shown). Using the log rank test, a significant difference in survival was found (p < 0.0007). Comparison of KO+AB and WT+AB were significantly different at p < 0.04. Fungal burden at endpoint was assessed by two methods. Fungal DNA content was measured by qPCR and data were analyzed using a one-way ANOVA followed by a Games Howell test for differences between groups. The Δkdnase plus AmBisome (KO+AB) cohort was the only group in which fungal burden was not statistically different from the controls (CON+AB), and the median fungal burden in the KO+AB treatment group was 10-fold lower than in WT+AB (Figure 8A). We also quantified fungal elements in 3 lung sections from each animal in the WT+AB (n = 7) and KO+AB (n = 6) treatment groups and the data are summarized in Figure 8B. The percent coverage of the lung by fungal elements in GMS-stained lung sections corroborated the fungal DNA burden data: WT+AB mice had values ∼10-fold higher than the KO+AB group. Thus, the increased susceptibility of the Δkdnase strain to amphotericin B we observed in vitro (Figure 6) was replicated in vivo and resulted in a lower fungal burden and greater survival of the host in the KO+AB group.

Exposure to the Δkdnase strain resulted in a significantly lower lung fungal burden compared to wild type with or without AmBisome treatment, even in the animals that died from the disease. To determine whether the relative growth of the two strains may have contributed to this difference, we measured growth in minimal essential medium (MEM) supplemented with 15% human serum. After 18 and 25 h at 37°C, the Δkdnase strain had approximately 50% of the growth of the WT strain but by 48 h, cell density was equivalent between the two strains (data not shown). Moreover, in the antibiotic sensitivity experiments after 28–48 h, the growth of WT and Δkdnase strains were equivalent in the no antibiotic control samples (Figure 6). Finally, the data in Figure 7 show that mice exposed to the Δkdnase strain without AB treatment did not survive longer than mice that were given the WT strain. We confirmed this in a separate mouse study in which we found that the survival of BALB/c female mice was the same in the WT or Δkdnase treatment groups (data not shown). This suggests the combination of amphotericin B with Δkdnase deletion was responsible for the lower fungal burden and increased survival.

Representative sections of H&E and GMS stained lung sections from AmBisome-treated animals are shown in Figure 9. Extensive cellular infiltrates and hemorrhage were evident in WT+AB group, and clusters of fungal filaments were abundant in lung sections. In the KO+AB mice, although inflammation and fungal elements were evident, these were more restricted in size and less frequently observed (Figure 9). GMS-stained sections from mice exposed to WT, Δkdnase, or Δkdnase^R without AB treatment are shown in Supplementary Figure S8.

We examined the relative numbers of macrophages and neutrophils in lung sections from the AmBisome treatment groups at endpoint using immunohistochemistry. Neutrophil (MPO-positive cells) counts/mm² in the lung of animals that received either WT or Δkdnase conidia were elevated threefold over controls (Figure 10A); no significant difference was observed between the treatment groups. Macrophages (Mac 3-positive cells) were higher than control animals for both groups receiving A. fumigatus; however, the accumulation of macrophages detected in tissues of mice exposed to WT conidia was significantly lower than in Δkdnase-exposed mice (Figure 10B). Supplementary Figure S9 contains representative photomicrographs showing the difference in the number and staining of the macrophages in a lung section from a mouse in the KO+AB group compared to one from WT+AB.

Uptake and killing of conidia by macrophages is an important innate immune defense against A. fumigatus (Philippe et al., 2003). We compared the uptake of FITC-labeled WT, Δkdnase, and Δkdnase^R conidia, quantified with immunofluorescence microscopy in the cultured mouse macrophage cell line, J774. No difference in uptake by J774 cells was observed between the 3 strains (Figure 11) and a similar experiment that measured conidial uptake by cultured human type 2 pneumocytes (A549 cells) produced the same result (data not shown). Thus, Δkdnase conidia were internalized to the same extent as WT by both professional phagocytes and epithelial cells; however, whether there were differences in the rate of conidial killing was not determined.