ABSTRACT
Antifungal therapy can fail in a remarkable number of patients with invasive fungal disease, resulting in significant morbidity worldwide. A major contributor to this failure is that while these drugs have high potency in vitro, we do not fully understand how they work inside infected hosts. Here, we used a transparent larval zebrafish model of Aspergillus fumigatus infection amenable to real-time imaging of invasive disease as an in vivo intermediate vertebrate model to investigate the efficacy and mechanism of the antifungal drug voriconazole. We found that the ability of voriconazole to protect against A. fumigatus infection depends on host innate immune cells and, specifically, on the presence of macrophages. While voriconazole inhibits fungal spore germination and growth in vitro, it does not do so in larval zebrafish. Instead, live imaging of whole, intact larvae over a multiday course of infection revealed that macrophages slow down initial fungal growth, allowing voriconazole time to target and kill A. fumigatus hyphae postgermination. These findings shed light on how antifungal drugs such as voriconazole may synergize with the immune response in living hosts.
INTRODUCTION
More than 2 million life-threatening invasive fungal infections are estimated to occur each year, primarily in immunosuppressed patients (1). Among these infections is invasive aspergillosis, caused mainly by Aspergillus fumigatus and estimated to occur in >200,000 people per year (1). Three main classes of antifungal drugs are used in the clinic, polyenes, azoles, and echinocandins, with the first discovered in the 1950s and others developed recently (2). While there are a range of MICs reported for these drugs, depending on both the species and the strain, these drugs have excellent activity against A. fumigatus in in vitro testing, with MICs of <10 μg/ml (3). Yet patients with invasive aspergillosis who are treated with frontline drugs have only an ∼50% survival rate (4). One of the reasons for this treatment failure is growing antifungal resistance (5), but even patients infected with susceptible fungal strains can succumb despite treatment (6). Additionally, while these drugs are tested for their ability to inhibit spore germination (7), most patients present with already germinated, invasive hyphae (8). Antifungal therapy can also cause severe side effects in patients, including hepatotoxicity, nephrotoxicity, hormone dysregulation, fever, chills, and vomiting (9, 10). New drugs are being developed or repurposed for antifungal therapy (11), but the current situation suggests that we first need to understand better how these antifungal drugs work in live, infected animals. All of these drugs are primarily tested and studied in vitro, but the presence of immune cells and the complexity of whole tissues and organs has major effects on drug delivery and efficacy in infected patients (12, 13).
The larval zebrafish is an ideal intermediate animal model in which to study antifungal drug efficacy in vivo. Zebrafish have an immune system that is largely conserved with humans (14), and the pathogenesis of invasive fungal infections in larval zebrafish recapitulates that of human patients (15), including a model of invasive aspergillosis (16, 17). These studies utilize the many established immunodeficient models in larval zebrafish, including deficiencies in neutrophils and macrophages, the two major immune cell types present in larval zebrafish. Drugs are commonly tested and screened in zebrafish larvae and can easily be delivered to the animal through addition to larval water and absorption through the skin (18). Finally, the transparency of larval zebrafish allows for direct observation of fungal growth and host cell responses through light microscopy in a living, intact host over the course of multiple days (17).
In this study, we took advantage of this live imaging ability to observe the temporal effects of an antifungal drug on fungal growth in real time. We found that voriconazole treatment can fully protect wild-type larval zebrafish from A. fumigatus infection. Unexpectedly, this protection is partially dependent on the presence of host macrophages, while neutrophil function is dispensable. From live imaging experiments following a multiday course of infection, our data suggest that by inhibiting germination of spores, macrophages slow down the course of infection, allowing the drug more time to act. Previous in vitro experiments suggested that voriconazole has both fungistatic and fungicidal activity, inhibiting A. fumigatus germination and killing hyphae postgermination (19, 20), but we find that only the latter occurs in infected, voriconazole-treated larval zebrafish. These data demonstrate that intermediate models such as larval zebrafish can provide insight into how antimicrobial therapies work inside live, intact hosts.
RESULTS
Voriconazole efficacy in larval zebrafish depends on the presence of host immune cells.To first test whether larval zebrafish are a good host model for antifungal drug studies, we infected larval zebrafish with A. fumigatus spores using a now well-established model of microinjection into the hindbrain ventricle (16, 17). We then added voriconazole to the larval water and monitored survival for 7 days. In both a neutrophil-defective host model (mpx:rac2D57N) and a matched control (mpx:rac2wt) (64), 1 μg/ml or 10 μg/ml of voriconazole in the larval water fully rescued survival of these larvae (Fig. 1A). For comparison, voriconazole is estimated to have an MIC in the range of 0.25 to 1 μg/ml in vitro against a variety of A. fumigatus isolates, including the strain used here, Af293 (21–23). Similar results were seen in a different neutrophil-defective host (see Fig. S1A in the supplemental material), confirming that voriconazole can protect larval zebrafish from infection with A. fumigatus.
Voriconazole efficacy in larval zebrafish depends on host immune function. Larvae were injected with A. fumigatus spores, voriconazole was added to the larval water, and host survival was monitored for 7 days. All data are from 3 pooled independent replicates. (A) Neutrophil-defective larvae (mpx:rac2D57N) or control larvae (mpx:rac2wt) were injected with an Af293-derived strain. (B) Larvae lacking both neutrophils and macrophages (pu.1 morphant) were injected with an Af293-derived strain. (C) Larvae lacking macrophages (irf8−/−) were injected with an Af293-derived strain. (D) Neutrophil-defective larvae (mpx:rac2D57N) were injected with a CEA10-derived strain. (E) Neutrophil-defective larvae (mpx:rac2D57N) were injected with patient-derived isolates of A. fumigatus that are in vitro azole-resistant (strain F11628 or F13747) or azole-susceptible (AF41) or an Af293-derived control.
However, when we expanded these studies to other immunosuppressed host models, voriconazole treatment was not as effective. In pu.1 morphant larvae, which do not have neutrophils or macrophages, 70% of larvae treated with 1 μg/ml voriconazole still succumbed to infection (Fig. 1B). In irf8 mutant larvae, which lack macrophages and have an expanded number of neutrophils, voriconazole treatment again did not fully rescue survival (Fig. 1C). Significantly more irf8−/− larvae succumb to infection after treatment with 1 μg/ml voriconazole than irf8+/+ or irf8+/– controls (Fig. S1C; P = 0.03 and 0.003, respectively). As protection was observed in the respective controls (Fig. S1B and C), the failure of voriconazole treatment in pu.1 morphant and irf8−/− hosts likely is due to specific differences in the host immune cells present. Similar results were obtained with a different A. fumigatus strain, demonstrating that in this model, fungal strain differences are not a major determinant of voriconazole efficacy (Fig. 1D, Fig. S1D to F).
Together, these results suggest that how voriconazole controls fungal growth in vitro is likely not the same as in vivo and that voriconazole requires host innate immune cells, specifically macrophages, to control fungal infection in larval zebrafish. To test these hypotheses in a different way, we infected a neutrophil-defective host with patient-isolated azole-resistant A. fumigatus strains (24). Larvae infected with strains F11628 or F13747 were still somewhat protected by 1 μg/ml voriconazole treatment (Fig. 1E), even though voriconazole has an MIC of >4 μg/ml against these strains in vitro (24). These data provide further evidence that the efficacy of antifungal drugs in vitro cannot fully predict how these drugs work in a whole-animal model.
Voriconazole does not kill A. fumigatus spores in larval zebrafish.We therefore wanted to investigate the mechanism by which voriconazole protects larvae from A. fumigatus infection, focusing again on Af293-derived strains and the lowest dose of voriconazole tested which can be protective, 1 μg/ml, to avoid any toxicity associated with higher doses. We began by looking at control wild-type hosts, in which we consistently found that voriconazole treatment rescues larval survival (Fig. 1A, Fig. S1B and C). From previous studies, we knew that in wild-type hosts, the germination rate of spores is low, but these spores can persist inside macrophages for more than 7 days of infection (16, 17). We hypothesized that the combination of a macrophage response and voriconazole might lead to better killing of these spores. To test this, we first performed fungal CFU counts from infected larvae through a 7-day infection. However, the percentage of CFU present in wild-type larvae over time was virtually identical in control and voriconazole-treated larvae, suggesting that voriconazole treatment does not increase spore killing in larval zebrafish (Fig. 2A). To confirm these results, we also quantified spore killing using a live-dead staining method and confocal microscopy. In these experiments, the cell wall of yellow fluorescent protein (YFP)-expressing spores was labeled with an Alexa Fluor tracer molecule to allow visualization of both live (YFP+ Alexa Fluor+) and dead (YFP– Alexa Fluor+) spores, and the spores were injected into a larval host with labeled macrophages (mfap4:tomato-CAAX). The percentage of spores killed in control and voriconazole-treated larvae was not significantly different at 2 days postinjection (dpi) (Fig. 2B and C). Similar results were obtained when we focused our analysis only on spores that were clearly inside macrophages (Fig. S2), supporting the conclusion that voriconazole treatment does not increase killing of A. fumigatus spores in vivo.
Voriconazole treatment in larval zebrafish does not kill A. fumigatus spores. (A) Wild-type larvae were injected with A. fumigatus spores and treated with either 1 μg/ml voriconazole or DMSO vehicle control. Individual larvae were homogenized and plated to enumerate CFU on indicated days. All data are from 3 pooled independent replicates, with an average of n = 8 larvae per time point per replicate. (B, C) Transgenic larvae with labeled macrophages (mfap4:tomato-CAAX) were injected with a YFP-expressing strain of A. fumigatus labeled with an AlexaFluor633 tracer molecule, treated with either 1 μg/ml voriconazole or DMSO vehicle control, and imaged at 2 dpi. (B) Imaging area in the larvae and representative images are shown. Arrowheads mark killed spores (YFP–), and asterisks mark live spores (YFP+). Note: the live spore shown in voriconazole condition is germinating. (C) The percentage of spores alive per larvae was quantified over 3 independent experiments and pooled. Across 3 replicates, a total of 30 and 28 larvae and 1,500 and 1,269 spores were imaged in DMSO and voriconazole conditions, respectively. Individual data points represent single larvae and are color-coded by replicate; lines and error bars represent lsmeans and SEM.
Voriconazole kills A. fumigatus postgermination.If voriconazole does not kill spores, we hypothesized instead that it is fungistatic, preventing spore germination and hyphal development as it does in vitro against A. fumigatus (19). To test this, we used the neutrophil-defective mpx:rac2D57N zebrafish line in which we can observe hyphal growth (16) but which is protected by voriconazole treatment (Fig. 1A). We crossed this line with a macrophage-labeled line (mpeg1:EGFP), infected the resulting larvae with red fluorescent protein (RFP)-expressing A. fumigatus, and monitored fungal growth and macrophage response daily through a 5-day infection. Spores in voriconazole-treated larvae still germinated, and while the rate of appearance of germinated spores was somewhat lower than in control larvae, the difference was not statistically significant (Fig. 2A), suggesting that voriconazole acts even later in infection.
We next looked specifically at larvae that had germinated spores in them early in infection (2 dpi) to determine what happens to the fungus after it germinates. In control larvae, A. fumigatus can keep growing and eventually kill the larval host (Fig. 2B). In voriconazole-treated larvae, on the other hand, we saw hyphae being degraded over time such that most of the hyphal fragments were completely destroyed by 3 days postgermination (5 dpi) (Fig. 2B). In both treatment groups, macrophages are present and surround the growing fungus. Quantification of fungal burden in all of the larvae that had germinated spores inside them at 2 dpi showed that while 4/10 control larvae could not control fungal growth and succumbed to infection, all 7 voriconazole-treated larvae were able to degrade the germinated fungus and survive (Fig. 3C).
Voriconazole acts postgermination in neutrophil-defective larval zebrafish. Neutrophil-defective (mpx:rac2D57N) larvae with labeled macrophages (mpeg1:EGFP) were injected with RFP-expressing A. fumigatus spores and treated with either 1 μg/ml voriconazole or DMSO vehicle control. Larvae were imaged daily for 5 days. All data are from 3 pooled independent replicates. (A) Individual larvae were scored for the presence of germinated spores inside them, and the cumulative incidence of germination is plotted. (B) Example images of single DMSO-treated and voriconazole-treated larvae are shown for multiple days after germinated spores were observed at 2 dpi. (C) In all larvae that had germinated spores at 2 dpi, fungal burden was quantified from images (RFP+ area) and is displayed as a heatmap.
To better visualize the dynamics of fungal burden and macrophage activity, we performed time-lapse imaging of larvae every minute for ∼40 hours after germination was observed using light sheet microscopy. Again, we observed that in control larvae, fungal growth can proceed unabated and kill the larvae (Movie S1), while in voriconazole-treated larvae, fungal growth is contained (Movie S2). Altogether, these experiments demonstrate that voriconazole protects infected larval zebrafish by killing A. fumigatus hyphae after they begin to develop.
Macrophages are required to inhibit spore germination and give the drug more time to work in larval zebrafish.A remaining question is why the efficacy of voriconazole is different in different immunodeficient hosts, and specifically, why voriconazole works better when macrophages are present. Previous work from our labs demonstrated that a major role of macrophages in a normal immune response to A. fumigatus in larval zebrafish is to inhibit germination (17). Indeed, when we monitored germination in irf8 mutant larvae, we observed a very high incidence of germination, with ∼100% of larvae containing hyphae at 2 dpi (Fig. 4A). For comparison, in neutrophil-defective and wild-type hosts, we had previously observed only ∼60% of larvae with germinated spores at 3 dpi (Fig. 3A) (17). Similar to voriconazole treatment in neutrophil-defective hosts, treatment of infected irf8 mutant larvae with voriconazole did not inhibit germination significantly (Fig. 4A). These data further support the conclusion that voriconazole does not prevent A. fumigatus spore germination in larval zebrafish.
Voriconazole requires multiple days to act in larval zebrafish. (A to C) Macrophage-deficient (irf8−/−) larvae with labeled neutrophils (mpx:mCherry) were injected with YFP-expressing A. fumigatus spores and treated with either 1 μg/ml voriconazole or DMSO vehicle control. Larvae were imaged daily for 5 days. All data are from 3 pooled independent replicates. (A) Individual larvae were scored for the presence of germinated spores inside them, and the cumulative incidence of germination is plotted. (B) Example images of single DMSO-treated and voriconazole-treated larvae are shown for multiple days after germinated spores were observed at 1 dpi. (C) In all larvae that had germinated spores at 1 dpi, the fungal burden was quantified from images (RFP+ area) and is displayed as a heatmap. (D) Neutrophil-defective larvae (mpx:rac2D57N) or control larvae (mpx:rac2wt) were injected with A. fumigatus spores. Voriconazole was added to the larval water either immediately or at 2 dpi, and host survival was monitored for 7 days. All data are from 3 pooled independent replicates.
We therefore looked specifically at what happens to these irf8 mutant larvae after spores germinate inside them. Focusing on larvae that had germinated spores at 1 dpi, we again found that control larvae often could not control fungal growth, with hyphae overtaking the entire hindbrain region and killing the larvae (Fig. 4B). While many neutrophils respond to the fungus after germination, in general, fungal burden remained high in the control larvae, and 9/21 larvae succumbed by 4 days postgermination (5 dpi) (Fig. 4C). In larvae treated with voriconazole, the fungal burden was lower overall than in control larvae, but hyphal growth persisted for more than 4 days after germination, and we never observed the same clearance of hyphae that we did in larvae with functional macrophages (Fig. 4B and C).
Voriconazole requires multiple days to act in vivo.Our experiments in neutrophil-defective versus macrophage-deficient larvae highlight a striking difference in the rate of spore germination in these hosts (Fig. 3A and 4A). Almost 100% of macrophage-deficient larvae had germinated spores inside them at 2 dpi, while only ∼60% of neutrophil-defective larvae did at 3 dpi. We therefore hypothesized that macrophages promote voriconazole efficacy in larval zebrafish simply by inhibiting spore germination and giving the drug more time to target and kill the fungus before hyphal growth becomes too invasive and tissue-damaging (Fig. 5). If this hypothesis is correct, we would expect that in a host that is protected by voriconazole treatment (such as neutrophil-defective larvae), the addition of voriconazole later in infection would not be as effective. Indeed, in neutrophil-defective mpx:rac2D57N larvae, treatment with voriconazole starting at 2 dpi did not fully rescue host survival (Fig. 4D), similar to what we observed in macrophage-deficient hosts. On the other hand, in control hosts with a full and active innate immune system that can both inhibit the germination of A. fumigatus and target the fungus postgermination, adding voriconazole 2 dpi still fully rescued larval survival (Fig. 4D), underlining the importance of the host immune activity in anti-fungal drug efficacy in larval zebrafish.
Model of voriconazole activity in larval zebrafish. Voriconazole efficacy builds up throughout the course of treatment, reaching full efficacy after ∼3 days. Macrophages delay the process of A. fumigatus spore germination such that the occurrence of invasive fungal growth corresponds with this peak of efficacy, allowing the drug to fully clear any developed hyphae. In larvae without functional macrophages, germination happens much faster, and hyphae appear before the drug is fully effective, allowing for uncontrollable fungal growth and host death.
DISCUSSION
The molecular targets of all of the commonly used antifungal drugs are known, but we still do not fully understand the range of their phenotypic effects both on the fungus directly or on the pathogenesis of infection in humans or animal models. Triazole antifungal drugs, including voriconazole, inhibit ergosterol biosynthesis by targeting the cytochrome P450-dependent enzyme 14α-lanosterol demethylase (25). Ergosterol is a fungus-specific sterol that is an important component of the cell membrane, similar to cholesterol in mammals and vertebrates.
It is thought that inhibiting ergosterol synthesis prevents new membrane production and therefore new growth (26). There is also evidence that buildup of methylated sterols stresses and disrupts existing membranes in fungi (27). By affecting cell membranes, these drugs can also affect the function of intracellular vacuoles, including impairing vacuolar acidification (28). As a result, many studies have found that voriconazole can be fungicidal, not just fungistatic (20, 29–32). However, whether this fungistatic activity in vitro is against conidia, hyphae, or both is disputed (20, 29–33). Here, we report that in larval zebrafish, voriconazole primarily acts against the hyphal stage of A. fumigatus and does not significantly inhibit conidial germination. In vitro studies finding that this fungicidal activity is time- and concentration-dependent are consistent with our findings in larval zebrafish that several days of voriconazole treatment are required before the fungus is destroyed (31, 32). These questions have been hard to tackle in animal models in the past, as many are not amenable to long-term imaging, and therefore the effect of the drug on different stages of infection cannot be measured. However, in mammalian models of infection, voriconazole is typically given to the animal 24 hours after infection, after the infection is already established, and the drug is still effective at high doses at increasing animal survival and decreasing the fungal burden (34–39).
Variables that may differ between the larval zebrafish host and mammals include host temperature difference (28.5°C versus 37°C) and drug metabolism. In humans, voriconazole is metabolized in the liver by cytochrome P450 (CYP) enzymes via N-oxidation into a metabolite that no longer has antifungal activity, but sufficient dosing can still create steady-state levels of the drug in patients (40). We do not currently know if or how this drug is metabolized in the larval zebrafish model, and while CYP-mediated drug metabolism varies among species (41), there are similarities between CYP enzymes in zebrafish and humans (42). Higher concentrations of voriconazole can also cause host toxicity, including off-target effects on host sterol synthesis and homeostasis (43, 44), and can have adverse effects in patients, such as cutaneous malignancy, photosensitivity, and periostitis (45). Some triazole drugs are toxic to zebrafish, especially during development (46–48), and there may be some toxicity associated with higher doses of voriconazole (10 μg/ml) in immunosuppressed zebrafish. However, we do not find any toxicity of voriconazole at the dose at which we performed the bulk of our experiments (1 μg/ml).
We unexpectedly found that the full efficacy of this drug depends on host immune cells, specifically macrophages. Macrophages phagocytose A. fumigatus spores and prevent their germination in both larval zebrafish and mice (16, 17, 49), and our data suggest that it is this activity that slows down fungal growth and allows the drug more time to act. However, in some larval zebrafish infection contexts, macrophages are not required for host survival. Irf8−/− larvae that lack macrophages but have an increased number of neutrophils are still resistant to certain strains of A. fumigatus because their neutrophil response is sufficient to clear the fungus (17). Similarly, different studies using mouse models disagree on the importance of macrophages in the immune response to A. fumigatus, with some suggesting that they are dispensable (50) and others demonstrating an increased fungal burden in their absence (51). In human patients as well, macrophage deficiency is not considered a major risk factor for invasive aspergillosis, although many immunosuppressive treatments that cause susceptibility, such as corticosteroid treatment or immune-modulatory treatments such as anti-tumor necrosis factor (anti-TNF), and genetic deficiencies, such as chronic granulomatous disease, affect the function of many immune cell types, including macrophages (52–54). Our data suggest that in infection contexts where macrophages are important, such as in larval zebrafish infected with the Af293 strain of A. fumigatus, these cells are also important for voriconazole efficacy.
Macrophages may act in other ways to increase the efficacy of voriconazole in larval zebrafish that have yet to be explored. Macrophages could increase uptake of the drug from the larval water or concentrate the drug at the site of fungal growth. Both epithelial cells and murine macrophages in vitro can take up and concentrate a different azole drug, posaconazole, and increase drug targeting of A. fumigatus (55, 56). While similar results were not observed with voriconazole, it could be different in an in vivo situation. Antifungal drugs can also directly modulate host immune function in cell culture studies. Monocytes synergize with voriconazole to lead to more A. fumigatus hyphal damage (20, 29, 30). One mechanism for this synergy is proposed to be an increase in the expression of NF-κB-regulated chemokines and cytokines, including TNF-α and interferon-γ (IFN-γ), which can be stimulated in monocytes by either voriconazole alone or voriconazole combined with A. fumigatus hyphae (20, 30). A different study showed that when monocytes were stimulated with voriconazole plus A. fumigatus spores, the drug actually reduced NF-κB activation and TNF-α production (57), which could also partially explain why voriconazole was more effective against hyphae than spores in our experiments. Another line of evidence supporting the hypothesis that voriconazole treatment can act independently of direct targeting of the fungus is our observation that voriconazole provides protection against azole-resistant strains in larval zebrafish, which was similarly observed in murine models (35). The effect of this drug on immune cell behavior and gene expression in animal models such as larval zebrafish will be an important area of future study.
Altogether, we report that the efficacy of antifungal drugs inside infected larval zebrafish depends on the presence of macrophages. This finding suggests that more studies should aim to understand how existing antifungal drugs work in different immunosuppressed conditions with different proximal causes of immunodeficiency, such as patients who have had hematopoietic stem cell transplants versus cancer patients undergoing chemotherapy. To our knowledge, this has not been studied in higher vertebrate models. Some studies have compared the efficacy of different antifungal drugs in patients with allergic bronchopulmonary aspergillosis (ABPA) and cystic fibrosis (58, 59), but very little is known about how underlying risk factors for invasive aspergillosis affect patient outcome after treatment with different antifungal drugs. While the experiments we present here were performed in an intermediate vertebrate larval zebrafish model and need to be expanded into higher vertebrate models, they highlight the possible utility of the larval zebrafish model in drug development and testing.
MATERIALS AND METHODS
Zebrafish lines and maintenance.All zebrafish were maintained as previously described (60). Animal care and use protocol M005405-A02 was approved by the University of Wisconsin-Madison College of Agricultural and Life Sciences (CALS) Animal Care and Use Committee. Prior to any experimental manipulations, larvae were anesthetized in E3 medium containing 0.2 mg/ml tricaine (ethyl 3-amino-benzoate; Sigma). All zebrafish lines used in this study are listed in Table 1 and were maintained on the wild-type AB background.
Zebrafish lines used in this study
The mpx:rac2D57N transgenic line expresses a mutated (amino acid 57, D to N), dominant-negative copy of the rac2 gene specifically in neutrophils, preventing neutrophil migration to sites of inflammation. Its respective control, mpx:racwt, expresses a wild-type copy and has neutrophils that can migrate normally (64). The irf8−/− line was maintained as adult heterozygotes by out-crossing. For larval survival experiments, homozygous mutants (and their +/+ siblings) were grown to adulthood and in-crossed. For imaging experiments, larvae were generated from an irf8+/− mpx:mCherry in-cross. To enrich for the irf8−/− genotype, larvae with high numbers of mCherry+ cells were selected for experiments, as the irf8 mutation also increases neutrophil numbers. The actual genotype of each larva was then confirmed by PCR and restriction fragment length polymorphism (RFLP) at the conclusion of the experiment as previously described (61).
pu.1 morpholino injection.A previously published and established splice-blocking morpholino oligonucleotide (MO; GeneTools) against pu.1 (ZFIN MO1:GATATACTGATACTCCATTGGTGGT) (62) was kept at a stock concentration of 1 mM in water. For injection, the MO was diluted in water with 0.5× CutSmart buffer (NEB) and 0.1% phenol red to a final concentration of 0.5 mM. We injected 3 nl into the yolk of single-cell embryos. A standard control MO (GeneTools) was used at a matching concentration.
Preparation of A. fumigatus spores and microinjection.All A. fumigatus strains used in this study are listed in Table 2. TBK1.1, TBK5.1, and TFYL49.1 had previously been used in larval zebrafish and were shown to have similar characteristics to their parental strains in larval zebrafish (17). All experiments were done with Af293-derived strains unless otherwise noted. Experiments with an Af293-derived strain used TBK1.1, except for experiments in Fig. 3 and Movies S1 and S2, which used TBK5.1. Experiments with a CEA10-derived strain used TFYL49.1. Patient-derived isolates were a gift from David W. Denning.
A. fumigatus strains used in this study
Spores were grown at 37°C on solid glucose minimal media (GMM), isolated for injection in phosphate-buffered saline (PBS), and microinjected into the hindbrain ventricle of 2-day postfertilization (2-dpf) larvae as previously described (16). We aimed for an infectious dose of ∼50 spores per larva. Unless otherwise stated, larvae were kept in 96-well plates after injection.
Voriconazole treatment.Voriconazole (Sigma) was resuspended to stock concentrations of 10 mg/ml and 1 mg/ml in dimethyl sulfoxide (DMSO) and kept in aliquots at 4°C. Stock was then added to E3 with a 1:1,000 dilution to result in final concentrations of 10 μg/ml or 1 μg/ml, respectively. A DMSO control was also added at 1:1,000 to E3. Unless otherwise noted, larvae were treated within 2 hours after injection, and voriconazole was left on larvae for the entire experiment and not refreshed.
CFU counts.Single larvae were placed in 1.5-ml microcentrifuge tubes and first rinsed with 500 ml of fresh E3 for 30 min to wash out voriconazole and/or DMSO. CFU counts then were performed as described previously (17). Briefly, individual larvae were homogenized and spread on solid GMM plates, and CFU were counted 2 to 3 days later. All CFU data were normalized to the average initial injection dose for each replicate and condition.
Larval imaging.To prevent pigment formation, 0.2 mM N-phenylthiourea (PTU; Sigma-Aldrich) was added to E3 water starting at 12 to 24 hours postinjection (hpf). Larvae with fluorescent transgenes were prescreened on a zoomscope (EMS3/SyCoP3; Zeiss; Plan-NeoFluarZ objective) prior to the experiment. Live-dead spore labeling was performed as described previously (17). To image these larvae at 2 dpi, larvae were mounted in a glass-bottom dish with 1% low-melting-point agarose, and images were acquired with a laser-scanning confocal microscope (FluoView FV1000; Olympus) with a numerical-aperture (NA) 0.75/20× objective and FV10-ASW software (Olympus). For daily imaging, after spore injection and voriconazole treatment, larvae were individually maintained in wells of a 48-well plate. Each day, larvae were removed one at a time, anesthetized in E3 with tricaine, and placed into zWEDGI chambers (63) to correctly orient their hindbrains toward the bottom of the dish. Images were acquired on a spinning disk confocal microscope (CSU-X; Yokogawa) with a confocal scan head on a Zeiss Observer Z.1 inverted microscope, Plan-Apochromat NA 0.85/20× or EC Plan-Neofluar NA 0.3/10× objectives, and a Photometrics Evolve EMCCD camera. Images were acquired with ZEN software (Zeiss). After imaging was complete, larvae were rinsed in E3 with PTU and placed back into the same plate well, and liquid was replaced with fresh E3 containing PTU and voriconazole.
Imaging quantification and display.To quantify killing of Alexa Fluor-labeled spores (Fig. 2B and C), the analysis was performed blinded. Live and killed spores were manually counted using Fiji, through all slices of the z-stack, using the Cell Counter plugin. Displayed images of live-dead staining were processed using Fiji with bilinear interpolation to increase the pixel density 2-fold and with a Gaussian blur (sigma = 1.5). Images represent single z-slices. In larvae imaged daily (Fig. 3 and 4), the presence of germination was scored manually. The fungal area was quantified from maximum intensity projections of z-stacks inside a region of interest (ROI) drawn around the hindbrain by auto-thresholding with the Otsu algorithm in Fiji. Heatmaps were generated with MultiExperiment Viewer (MeV). Displayed images of daily larval imaging were processed in Fiji with bilinear interpolation to increase the pixel density 2-fold. Images represent maximum intensity projections. In all displayed images, brightness and contrast were manually adjusted in Fiji.
Light sheet microscopy.At 2 dpi, larvae were screened for the presence of germinated spores on a spinning disk confocal microscope as described above. Positive larvae were then mounted in a 0.8-mm or 1.6-mm inner diameter fluorinated ethylene propylene (FEP) tube in 0.4% low-melting-point agarose with 0.3 mg/ml Tricaine and an agarose plug. Where applicable, 1 μg/ml voriconazole was also added to the low-melt agarose. The entire head region of the larvae was then imaged with a custom-built multidirectional Selective Plane Illumination Microscope (mSPIM) (71). 3D image stacks were taken every minute with z-plane spacing of 2 μm. Different channels were taken sequentially. For visualization, maximum intensity projections were created postprocessing with Fiji.
Statistical analyses.For all statistical analyses, three independent experiments were performed and pooled. Experimental numbers are noted in figures and/or legends. For larval survival data, endpoint survival was compared using Fisher’s exact test in R, and adjusted P values (padj) are reported. For live-dead staining and CFU quantification, data were compared between experimental conditions using analysis of variance. Results are summarized in terms of least-squared adjusted means (lsmeans) and standard error (SEM). Graphs may show both calculated lsmeans ± SEM and individual data points, color-coded by replicate. For cumulative appearance of germination, replicates were pooled and curves were compared using Cox proportional hazard regression analysis, with the experimental replicate included as a group variable, as previously described (16, 17). All calculations were done in R, and graphs were generated in Microsoft Excel 2016.
ACKNOWLEDGMENTS
We thank members of the Huttenlocher and Keller labs for helpful scientific discussions.
This work was supported by NIH grant R35GM118027-03 to A.H. and NIH grant R01AI065728-10 to N.P.K. E.E.R. was supported by an NIH individual fellowship (F32AI113956).
E.E.R. was responsible for the conceptualization, methodology, and formal analysis. E.E.R. and J. He performed the investigation. J. Huisken contributed resources. E.E.R. wrote the original draft of the article, and E.E.R., N.P.K., and A.H. contributed to the review and editing of the article. N.P.K. and A.H. provided funding for this work.
FOOTNOTES
- Received 3 May 2019.
- Returned for modification 7 September 2019.
- Accepted 10 November 2019.
- Accepted manuscript posted online 18 November 2019.
Supplemental material is available online only.
- Copyright © 2020 American Society for Microbiology.