Targeting Iron Acquisition Blocks Infection with the Fungal Pathogens Aspergillus fumigatus and Fusarium oxysporum

Filamentous fungi are an important cause of pulmonary and systemic morbidity and mortality, and also cause corneal blindness and visual impairment worldwide. Utilizing in vitro neutrophil killing assays and a model of fungal infection of the cornea, we demonstrated that Dectin-1 dependent IL-6 production regulates expression of iron chelators, heme and siderophore binding proteins and hepcidin in infected mice. In addition, we show that human neutrophils synthesize lipocalin-1, which sequesters fungal siderophores, and that topical lipocalin-1 or lactoferrin restricts fungal growth in vivo. Conversely, we show that exogenous iron or the xenosiderophore deferroxamine enhances fungal growth in infected mice. By examining mutant Aspergillus and Fusarium strains, we found that fungal transcriptional responses to low iron levels and extracellular siderophores are essential for fungal growth during infection. Further, we showed that targeting fungal iron acquisition or siderophore biosynthesis by topical application of iron chelators or statins reduces fungal growth in the cornea by 60% and that dual therapy with the iron chelator deferiprone and statins further restricts fungal growth by 75%. Together, these studies identify specific host iron-chelating and fungal iron-acquisition mediators that regulate fungal growth, and demonstrate that therapeutic inhibition of fungal iron acquisition can be utilized to treat topical fungal infections.


Introduction
Aspergillus and Fusarium are filamentous fungi that cause lethal infections in immune suppressed individuals [1,2]. Additionally, they infect the corneas of immunocompetent individuals, and are a major cause of blindness associated with ocular trauma [3,4]. Although less common, Curvularia, Alternaria, and Penicillium species also cause keratitis [3]. Globally, the world health organization estimates that 1.8 million people in developing nations are blinded annually from corneal ulcers; furthermore, in developing nations, up to 65% of total corneal ulcers are caused by fungal infection, with approximately 1 million cases occurring annually in Asia and Africa [5][6][7].
Treatment regimens for fungal keratitis are often ineffective, with up to 60% of fungal keratitis cases requiring corneal transplantation [3]. Given the limited treatment options, there is a pressing need to develop new treatment strategies. In this effort, we recently demonstrated that inhibitors of fungal anti-oxidative responses enhanced fungal clearance in vivo and improved disease outcome [8]. As iron is essential for the redox reactions of major fungal antioxidants, including thioredoxin-dependent peroxiredoxases [9,10], and fungal iron acquisition mutants are more susceptible to oxidative stress [11], we hypothesized that targeting fungal iron acquisition may represent a potential new avenue for treatment of fungal infections.
Fungal iron acquisition primarily involves the production of hydroxamate-type siderophores that are secreted into the environment, bind iron with high affinity, and are then captured by specific siderophore receptors on the fungal cell membrane [12]. Specifically, the A. fumigatus siderophore biosynthesis pathway originates with the sidA gene, which encodes ornithine-N 5 -oxygenase, resulting in conversion of ornithine to N 5 -hydroxyornithine [13]. Utilizing this essential precursor, the siderophore biosynthesis pathway leads to either intracellular or extracellular siderophores. The sidC gene product is required for production of the intracellular siderophores, ferricrocin (FC) and hydroxyferricrocin (HFC), whereas the sidF and sidD gene products are required for production of the extracellular siderophores, fusar-inine C (FusC) and triacetylfusarinine C (TAFC) [13]. The sidG gene product is required to generate TAFC from FusC [13], whereas both the sidH and sidI gene products are required to incorporate mevalonate into the structure of extracellular siderophores [14]. As mevalonate biosynthesis is dependent on HMG-CoA reductase and this enzyme is inhibited by statins, we hypothesized that statin-mediated inhibition of fungal HMG-CoA reductase may restrict fungal iron acquisition in vivo.
Fungal siderophores are secreted into mammalian tissues during infection where they compete with host iron sequestration defenses. Under homeostatic conditions, free iron is maintained at relatively low levels by iron-binding proteins such as transferrin and ferritin [15]. In addition, mucosal secretions contain high concentrations of lactoferrin, which binds iron, and lipocalin-1, which sequesters fungal siderophores [16,17]. However, tissue damage during infection increases extracellular iron levels by releasing intracellular labile iron, ferritin, and heme-containing proteins [18]. Infection also stimulates both local and systemic immune defenses to counter microbial iron acquisition. Resident cells can secrete iron-sequestering proteins and chemotactic cytokines, which recruit neutrophils to the site of infection. Neutrophils also release pre-formed and de novo synthesized iron sequestering proteins such as lactoferrin, Lcn-2, and the hemoglobin binding protein haptoglobin [19]. Furthermore, neutrophilmediated oxidation is likely to increase the microbial requirement for iron to fuel iron-dependent anti-oxidative defenses. Lastly, production of cytokines such as IL-6 and IL-22 can induce local and systemic synthesis of the peptide hormone hepcidin, which degrades the iron exporter ferroportin and traps iron inside host cells [20][21][22].
In the current study, we examined the role of host iron sequestration and fungal iron acquisition in a murine model of Aspergillus and Fusarium corneal infection. We show that Dectin-1 and IL-6 regulate expression of genes involved in iron sequestration and that fungal growth positively correlates with serum iron levels. Using mutant A. fumigatus and F. oxysporum strains, we also demonstrate that fungal transcriptional responses to low iron levels and mevalonic acid-dependent extracellular siderophore biosyn-thesis, but not intracellular siderophores or reductive iron assimilation, are essential for fungal growth in vitro and during infection. Lastly, using iron chelators, siderophore binding proteins, and siderophore biosynthesis inhibitors including statins we provide proof-of-concept that targeting fungal iron acquisition enhances fungal clearance from infected tissues and may represent a new avenue for treatment of fungal infections.
During fungal infection of humans and mice, neutrophils comprise .95% of the cellular infiltrate in the cornea and are likely the predominant source of gene transcripts [23,24]. Therefore, prior to examining de novo transcription of ironsequestering genes during infection we first quantified the number of neutrophils infiltrating C57BL/6 and IL-6 2/2 infected corneas by flow cytometry using the Ly6G NIMP-R14 monoclonal antibody. As shown in Figure 1D, there was no significant difference in neutrophil numbers between infected C57BL/6 and IL-6 2/2 mice. RNA was then isolated from C57BL/6 and IL-6 2/2 corneas at 24 h post-infection, and gene expression was measured by Q-PCR.
During infection, lysed cells release heme, which microbes can utilize as a source of iron [18]. To restrict microbial access to heme, mammals produce hemopexin (Hpx) that binds to heme, and haptoglobin (HptG), which binds hemoglobin [15]. Figure 1E also shows that expression of Hpx and HptG in infected C57BL/6 mice is up-regulated 200-fold compared with naïve mice, whereas expression is ,10-fold increased in infected IL-6 2/2 mice.

Author Summary
Fungal pathogens, in addition to causing life-threatening systemic disease, can also invade the cornea and cause blindness and visual impairment. In the current study, we examined the role of iron acquisition in corneal infections caused by Aspergillus and Fusarium. We first demonstrated that expression of iron chelators, heme, siderophore binding proteins and hepcidin is elevated in infected corneas. Secondly, we showed impaired in vivo growth in Aspergillus and Fusarium with mutations in the pathway leading to production of iron-binding siderophores. These mutants were also more susceptible to killing by human neutrophils. Based on these observations, we targeted these pathways using topical iron chelators and found that they blocked fungal growth in the cornea. Finally, as statins target the enzyme HMG-CoA reductase, which is required for siderophore and ergosterol biosysnthesis, we found that topical statins inhibited fungal growth and reduced infection, and showed that combined treatment with the iron chelator deferiprone and statins had an additive effect on fungal infection. Together, these studies demonstrate that therapeutic inhibition of fungal iron acquisition can be utilized to treat topical fungal infections.
Lcn-1 binds to many hydrophobic molecules including phospholipids at the air-fluid interface in tears [27]. However, Lcn-1 also binds to fungal hydroxamate-type siderophores [17]. Given that human neutrophils store the bacterial-siderophore binding protein Lcn-2 in secondary granules [19], we examined if human neutrophils also produce Lcn-1. Peripheral blood neutrophils (.95% purity) from three healthy human volunteers were incubated for 1 h in RPMI media in the presence or absence of A. fumigatus crude hyphal extract, and Lcn-1 expression was examined by Q-PCR. As shown in Figure 1F, Lcn-1 gene expression was detected in human neutrophils in the presence or absence of Aspergillus hyphal extract, indicating constitutive RNA expression. Taken together, results from this set of studies indicate that following fungal infection of the cornea, Dectin-1 dependent IL-6 production induces local and systemic host responses that limit microbial access to iron. Iron availability regulates hyphal growth and the severity of A. fumigatus infection Given that fungal infection initiates an iron sequestration response, we next examined if iron availability regulates fungal growth during infection. Mice were injected intraperitoneally with 5 mg Fe-dextran (90 mmoles iron) or deferroxamine (5 mg), which is an iron-chelating xenosiderophore that is utilized by A. fumigatus [28]. Twenty four hours after the last injection, RFP-expressing A. fumigatus (Af-dsRed) conidia were injected into the corneal stroma of C57BL/6 mice [8,23]. After 24 h, total serum iron was quantified in treatment and control groups. Figure 2A shows that serum iron levels in infected mice were reduced 2-fold compared with naïve mice, indicating systemic iron-sequestration during inflammation [15]. In contrast, mice given systemic Fe-dextran, but not deferroxamine had significantly elevated serum iron compared to vehicle-treated mice. Despite the difference in serum iron levels, Figures 2B,C show that whereas fungal mass (dsRed) increases over 48 h in vehicle-treated mice, this was significantly higher in Fe-dextran and deferroxaminetreated mice. Consistent with these data, Figure 2D shows that at 48 h post-infection, CFU were significantly higher in Fe-dextran and deferroxamine-treated mice compared with control, vehicletreated mice. As filamentous fungi grow by hyphal extension and not cell division, the dsRed measure of fungal mass increases over time, whereas CFU decreases from the initial inoculum [15]. Figures 2B, E, and F show that corneal opacification was also increased in Fe-dextran and deferroxamine-treated mice compared to vehicle-treated mice, consistent with increased fungal growth. The increased fungal growth in deferroxamine treated mice is likely due to its xenosiderophore function, which can be used by Aspergillus for iron acquisition [28].
To examine the effect of limiting iron availability during infection, C57BL/6 corneas were infected with A. fumigatus dsRed conidia as described (12), and the iron chelating protein lactoferrin was added topically (10.4 mg in 8 ml) at 0 h and 6 h post-infection. As shown in Figures 2G-I, infected corneas given topical lactoferrin had significantly less fungal mass (dsRed) and CFU per eye at 24 h post-infection compared with those given vehicle alone. Taken together with results from Fe-dextran and deferroxamine-treated mice, these studies demonstrate that fungal growth in the cornea is dependent on increased free iron or bio-available iron.

Siderophores and detection of low iron concentrations but not reductive iron assimilation is required for fungal growth during infection
Given that fungal growth during infection is enhanced by the exogenous xenosiderophore deferroxamine, we next examined the role of endogenous fungal siderophores using A. fumigatus, F. oxysporum, and Alternaria brassicicola iron acquisition mutants. The A. fumigatus DsidA mutant does not synthesize extracellular or intracellular siderophores [29], whereas the DhapX strain lacks the transcription factor HapX that is activated by low iron concentrations, and which regulates expression of genes involved in iron acquisition including siderophores and repression of irondependent pathways [30]. Figure 3A shows no significant difference in fungal growth in media alone between the parent (WT) strain and the DsidA and DhapX mutants, indicating that there is no effect of these mutations on fungal growth in the presence of an exogenous source of iron. We next examined the growth of the WT strain and the DsidA and DhapX mutants in the presence of neutrophils, which we recently showed kill hyphae by producing reactive oxygen species [8].
Conidia (spores; 12,500) were cultured 4-6 h in SDB media to allow germination and production of hyphae and incubated with 1610 5 neutrophils. In contrast to RPMI media alone in which all strains grew equally, in the presence of human neutrophils, growth of the DsidA and DhapX mutants was significantly less than the WT parental strain ( Figure 3A), indicating that adaptation to iron starvation and siderophores, which requires SidA and HapX is essential for survival in the presence of neutrophils.
Consistent with a role for these genes in virulence, we found that at 48 h post-infection, corneas infected with the A. fumigatus mutant strains DsidA or DhapX had significantly lower CFU compared with the parent strain ( Figure 3B). Figure S1 shows that corneas infected with the complemented DsidA strain sidA R or the complemented DhapX strain hapX R , show no significant difference in opacification or CFU as the WT parental strain. In contrast, although the DftrA mutant lacks a membrane-bound iron transport channel protein and is therefore deficient in cellular uptake of environmental iron [29], there was no significant difference in CFU in this mutant, indicating that this transporter protein is not essential for fungal growth in vivo. Similarly, Figures 3C, E, and F show that mice infected with the DsidA or DhapX mutants have significantly less corneal opacification than the parent A. fumigatus strain, whereas the DftrA mutant was not significantly different. Histological analysis shows a pronounced cellular infiltrate and fungal hyphae in the corneas of mice infected with the parent strain, whereas no hyphae were detected in DsidA infected corneas, indicating that these mutants did not germinate in the cornea ( Figure 3D). As with A. fumigatus, mice infected with the F. oxysporum DhapX strain also exhibit significantly lower CFU than the parent strain at 48 h post-infection ( Figure 3G), and A. brassicicola siderophore mutants have significantly less CFU than the parent strain ( Figure S2). Figure 3H-J demonstrate that mice infected with the F. oxysporum DhapX strain also exhibit significantly less corneal opacification compared to the parental strain. Figure S1 shows that corneas infected with the Fusarium complemented DhapX strain hapX R show no significant phenotypic differences from the WT parental strain. Taken together, these data indicate that siderophores have a critical role in fungal growth during infection, whereas reductive iron assimilation is not essential.

Extracellular siderophores are required for fungal infection
To determine the relative contribution of intracellular versus extracellular siderophores in fungal keratitis, corneas were infected with the DsidF and DsidC mutants. Figure 4B shows that CFU from corneas infected with the DsidF mutant that leads to extracellular siderophore production was significantly lower than those infected with WT A. fumigatus, having a similar CFU as DsidA mutants. In contrast, DsidC mutants that regulate production of intracellular siderophores were not significantly different from the parent strain. Consistent with this finding, corneas infected with DsidF had significantly lower cornea opacity area and intensity values compared with mice infected with WT A. fumigatus, whereas DsidC mutants were not significantly different ( Figure 4C-E). Together, these findings indicate that extracellular, but not intracellular siderophores are essential for fungal growth in the cornea and development of keratitis.
To determine the relative contribution of the extracellular siderophores FusC and TAFC, we infected corneas with DsidD mutants, which do not produce extracellular siderophores, or with DsidG mutants, which produce FusC but not TAFC, and compared them with the parent A. fumigatus strain that produces both FusC and TAFC [13]. Figure 4B shows that mice infected with DsidD mutants had significantly lower CFU compared with WT A. fumigatus, whereas DsidG were not significantly different. Corneal opacification scores reflected the CFU data, with DsidD but not DsidG mutants having significantly less opacification than WT A. fumigatus ( Figure 4C-E). These findings indicate that sidD-mediated synthesis of FusC rather than sidG-mediated synthesis of TAFC is necessary and sufficient to support fungal growth in vivo. Figure S1 shows that corneas infected with the complemented strains sidF R and sidD R show no phenotypic differences from the WT parental strain. Figure S2 shows that both intracellular and extracellular siderophore mutants of Alternaria brassicicola also have impaired growth during infection. Together, these data clearly demonstrate that extracellular siderophores are essential for both Aspergillus and Alternaria growth during tissue infection, and that even though TAFC is reportedly more stable [13], FusC production is sufficient to maintain fungal growth in vivo.

Mevalonate incorporation into extracellular siderophores is required for fungal infection
Fungal extracellular siderophore biosynthesis requires HMG-CoA reductase-dependent synthesis of the precursor mevalonate [14]. The Aspergillus genes sidI and sidH encode a CoA-ligase and an enoyl-CoA-hydratase, respectively, which convert mevalonic acid to anhydromevalonyl CoA and incorporate this precursor through the sidF-D-G pathway into the structure of fusarinine C and TAFC ( Figure 5A) [13,14]. To determine if this pathway is essential for fungal growth during tissue infection, C57BL/6 mice were infected with A. fumigatus mutant strains DsidH and DsidI and examined as before. Figure 5B shows that mice infected with DsidH or DsidI exhibit significantly less CFU than mice infected with the WT strain, indicating that mevalonate incorporation into extracellular siderophores is essential for fungal growth during tissue infection. Further, mice infected with either DsidH or DsidI exhibit significantly less cornea opacity at all time-points compared to mice infected with WT A. fumigatus ( Figure 5C-E). Figure S1 shows that corneas infected with the complemented DsidH strain sidH R or the complemented DsidI strain sidI R show no significant difference in opacification or CFU as the WT parental strain.''

Lipocalin-1 sequesters fungal siderophores and restricts fungal growth during infection
Humans produce two lipocalins with siderophore binding activity [26]. Lipocalin-1 (Lcn-1) binds to a wide range of bacterial and fungal hydroxamate-type siderophores [17], whereas Lcn-2 binds catechol-type bacterial siderophores but not fungal siderophores [31,32]. We therefore examined the role of Lcn-1 on A. fumigatus using the same assays as above. Figure 6B shows significantly less fungal growth incubated with 40 mg/ml or 4 mg/ ml Lcn-1 than in RPMI alone, and that growth of A. fumigatus in the presence of neutrophils and 4 mg/ml Lcn-1 was significantly less than with neutrophils alone or Lcn-1 alone. Figures 6C-E show significantly less fungal dsRed and CFU in mice given topical Lcn-1 (16 mg/8 ml) at 0 h and 6 h after infection compared with infected mice not given Lcn-1. These findings indicate that topical Lcn-1 inhibits fungal growth in vivo, presumably by sequestering fungal siderophores.

Topical simvastatin and deferiprone inhibit fungal infection
As shown in Figure 5, Aspergillus SidI and SidH proteins incorporate mevalonate into the structure of extracellular siderophores and are essential for fungal growth in the cornea. Also, HMG-CoA reductase is required for mevalonate production and can be targeted by statins to inhibit siderophore biosynthesis [14,33] (Figure 7A). To determine the effect of blocking this As shown in Figure 7B, there was significantly less growth of A. fumigatus following 16 h incubation in SDB media with simvastatin, lovastatin and deferiprone, but not deferroxamine, thereby demonstrating a direct effect of statins and deferiprone on fungal growth. Similar results were obtained with Fusarium oxysporum ( Figure 7C). Interestingly, a statin dependent dose curve was observed with A. fumigatus when exposed to simvastatin or lovastatin ( Figure 7B), however, a dose curve was only observed when Fusarium was treated with simvastatin not lovastatin ( Figure 7C). This observation likely reflects differences in the pharmacokinetics of different statins and their ability to penetrate into the fungal cytoplasm and inhibit HMG-CoA reductase of multiple fungal genera and species [33].
To ascertain if these agents enhance fungal sensitivity to killing by neutrophils, A. fumigatus conidia were cultured for 4-6 h in SDB media, washed, and incubated a further 16 h with human neutrophils in RPMI media containing simvastatin, lovastatin or deferiprone. As shown in Figure 7D, fungal growth was significantly less when incubated with neutrophils and simvastatin, lovastatin, and deferiprone compared with neutrophils alone, whereas there was no significant difference in the presence of deferroxamine. The inhibitory effect of 1 mM statins on fungal growth shown in panel B was not observed in this assay, most likely due to siderophore production during the 4-6 h growth in the absence of statins.
To ascertain if statins can restrict fungal growth during infection, mice were infected intrastromally with A. fumigatus, and given topical simvastatin, deferiprone, or deferroxamine at the time of infection and after 6 h. At 24 h post-infection, mice eyes were imaged for corneal opacity and fungal dsRed, and processed for fungal CFU. Importantly, unlike prior experiments, which examined CFU at the 48 h time-point, at 24 h CFU in infected eyes does not decrease unless treated with anti-microbial agents. Therefore, in this 24 h assay and unlike prior experiments, vehicle-treated CFU do not decrease but instead represent the maximum fungal CFU value per assay. As shown in Figure 7E and F, mice treated with simvastatin or deferiprone exhibited significantly less fungal mass compared to vehicle-treated mice, which was further decreased when given both compounds. Conversely, mice given simvastatin together with deferroxamine had significantly higher fungal mass than mice given simvastatin alone, indicating that exogenous deferroxamine counters the inhibitory activity of statins. Figure 7G shows similar responses when CFU were measured 24 h post infection, with significantly less CFU in mice given simvastatin, deferiprone or both, and partial reversal of the inhibitory effect of simvastatin when mice also received topical deferroxamine, indicating that simvastatin is targeting siderophore biosynthesis in addition to ergosterol synthesis in vivo. Mice treated with deferroxamine alone showed elevated fungal mass ( Figure 7E) compared with vehicle-treated mice, but at this time point, CFU values were not significantly different from vehicle-treated mice ( Figure 7F). Taken together, these findings clearly demonstrate that topical statins and iron chelation can block fungal growth during infection.

Discussion
Previous work in our laboratory demonstrated that fungal antioxidative responses are essential for survival during tissue infection, and that fungal growth can be inhibited in vivo by Targeting Fungal Iron Acquisition PLOS Pathogens | www.plospathogens.org targeting fungal thioredoxin [8]. As thioredoxin-regulated peroxiredoxases, catalases, and other antioxidants require iron to quench reactive oxidants [9,34], the current study examined the role of host iron sequestration and fungal iron acquisition during infection. In examining endogenous iron levels, we found a twofold reduction in serum iron levels following fungal infection which correlated with elevated expression of iron-chelating proteins, heme-binding and siderophore-sequestering proteins in infected corneas, and systemic induction of hepatic hepcidin. Expression of these iron-sequestering proteins was dependent on Dectin-1, which we and others showed is important in recognizing b-glucan on germinating conidia [23,35,36]. Expression of these proteins was also dependent on IL-6, which has been shown to induce liver hepcidin and reduce systemic iron levels [15].
In this iron-restricted environment, we show that fungal siderophores are essential for microbial growth and survival. This notion is supported by our findings that siderophore mutants are unable to grow in the cornea, and that exogenous iron chelators or inhibitors of fungal siderophore biosynthesis impair fungal growth both in vitro when incubated with human neutrophils, and in vivo in a murine model of fungal infection. Given that the siderophore biosynthesis pathway is highly conserved in filamentous fungi [12], our findings are very likely relevant to fungal infections of other tissues in addition to other fungal pathogens. Figure 8 illustrates the fungal siderophore biosynthesis pathway and highlights the key findings in this study.
Firstly, we identified that both Aspergillus and Fusarium require the transcription factor HapX to sense and respond transcription-ally to low iron levels and survive during infection. Further, we show that A. fumigatus strains DsidF, DsidD, DsidH, or DsidI, that do not express the extracellular siderophores are attenuated during infection of the cornea, whereas based on the absence of a phenotype with DsidC mutants, there is no apparent role for intracellular siderophores. We show that production of extracellular siderophores is also required for infection with Alternaria. These findings are consistent with an essential role for HapX and extracellular siderophores during experimental A. fumigatus lung infection [11,14,30], the requirement of HapX for Fusarium infection of tomato plants and immunosuppressed mice [37], and the requirement of siderophores for Alternaria infection in maize [38]. Interestingly, iron starvation also activates the transcription factor AcuM, which increases HapX expression and down-regulates the iron-repressing transcription factor SreA, resulting in fungal iron acquisition [39]. The role of AcuM and SreA during corneal infection has yet to be determined.
In the present study, we also examined the potential for statins to inhibit fungal HMG-CoA reductase, which is required for mevalonic acid production and extracellular siderophore biosynthesis [14]. Consistent with a report on pulmonary aspergillosis [14], we used mutants to demonstrate that SidI and SidHdependent mevalonic acid incorporation into extracellular siderophores is essential for infection of the cornea. Further, we show that simvastatin and lovastatin inhibit fungal growth in vitro, and also enhance growth inhibition by human neutrophils. Consistent with the difference in statin activity in treating hypercholesterolemia [40], simvastatin exhibited 10-fold greater inhibitory activity than lovastatin in restricting fungal growth in vitro. During infection, topical application of simvastatin inhibited fungal growth that was only partially reversed by exogenous siderophores, indicating that statins function in vivo by inhibiting both fungal siderophore and ergosterol biosynthesis. The fungicidal activity of statins has been reported for Aspergillus, Fusarium, Mucorales, and Candida [41][42][43]. However, to our knowledge, this is the first study to clearly demonstrate a therapeutic effect of statins in an experimental fungal infection. In contrast to systemic statin treatment, topical application is likely to have minimal risk of side effects [33]. Future studies will examine if statins can restrict fungal growth at later stages of infection.
In addition to showing inhibition of siderophore biosynthesis, the current study demonstrates that exogenous lipocalin-1 impairs fungal growth in the presence of human neutrophils and during infection. This finding is consistent with the reported role for Lcn-1 in sequestering fungal siderophores, including TAFC [17]. The affinity of Lcn-1 for fungal siderophores is similar to that of fungal siderophore receptors [17], and therefore at high concentrations Lcn-1 could sequester siderophores from fungal receptors. Further, it is likely that Lcn-1-siderophore complexes are internalized through the lipocalin-interacting membrane receptor (LIMR) resulting in siderophore degradation [44,45]. Endogenous Lcn-1 is abundant in human tears (3 mg/ml), nasal mucosa, and tracheal secretions where it can function prophylactically to prevent mucosal fungal infections [17,46]. However, as we now show that human neutrophils express Lcn-1, it is possible that Lcn-1 also has a protective role during active infection.
In addition to targeting siderophores, we showed that reducing local tissue iron concentrations by topical application of the iron chelating protein lactoferrin restricts fungal growth in vivo. This finding is consistent with the reported role for lactoferrin in blocking conidia germination in vitro by human neutrophils [47], and suggests that in vivo, neutrophil-derived lactoferrin restricts the growth of conidia and hyphae by binding free iron. Given that siderophores exhibit a higher affinity for iron than lactoferrin [12], it is likely that fungal siderophores can acquire iron from lactoferrin during infection; however, the rate of siderophore iron acquisition in a lactoferrin-rich environment is likely slower than in the absence of lactoferrin given the scarcity of free iron or ironbound to lower affinity biomolecules.
Similarly, we showed that the iron chelator deferiprone sensitizes fungi to human neutrophils and blocks fungal infection. However, deferiprone is also a very small molecule (MW = 140 g/mole), approximately 600-fold smaller than lactoferrin, and is therefore released from tissues more readily than lactoferrin, resulting in both iron sequestration and depletion from infected tissues [48]. Deferiprone has been used effectively and safely to lower iron levels in patients with hemochromatosis [48], and although widely utilized in Europe, it is not currently licensed in the USA. Importantly, deferiprone, unlike deferroxamine, is not a xenosiderophore [28] and is not associated with an increased risk of fungal and bacterial infections [49][50][51][52][53][54][55][56][57].
Iron chelators have been used to treat Aspergillus and Rhizopus infections in mice [57][58][59][60], and a clinical trial examined the potential of the iron chelator deferasirox to enhance the efficacy of liposomal amphotericin B to treat mucormycosis (the DEFEAT Mucor study). However, the trial was unsuccessful due to an unexpected increased risk of death in patients receiving deferasirox adjunct therapy [61]. The DEFEAT Mucor study exhibited a limited sample size and imbalanced stratification of the sickest patients into the deferasirox treatment group and may not accurately reflect the potential of deferasirox to treat mucormycosis. However, systemic deferasirox treatment does cause side effects that include agranulocytosis and nephrotoxicity [62]. In the current study, we demonstrated that local (topical) application of iron chelators is both effective in inhibiting fungal growth and preventing corneal disease. As local administration is highly unlikely to cause systemic side effects, clinical studies using topical iron chelators and fungal iron acquisition inhibitors are unlikely to cause adverse reactions, especially if combination therapies targeting iron acquisition can use low drug concentrations. This approach could be used to treat fungal infections not only in the cornea, but also in other tissues that could be treated topically such as the tongue, skin, and nails.
In conclusion, we have identified host iron sequestration and fungal siderophore biosynthesis as essential mediators of fungal growth during infection. One approach to exploiting these findings is to chelate local iron at the infectious site utilizing deferiprone or lactoferrin. A second approach is to inhibit the ability of fungi to acquire iron utilizing the siderophore-binding protein Lcn-1 or the siderophore biosynthesis inhibitor simvastatin. However, the most efficacious strategy would likely involve the combination of iron chelation and inhibition of siderophore biosynthesis. In this study, we provide proof-of-concept that dual treatment with deferiprone and simvastatin further restricts fungal growth during infection. As both deferiprone and simvastatin, have a long history of safe use in patients, it is possible that these agents can be successfully utilized to treat a broad range of fungal infections.

Use and source of animals
All animals were treated in accordance with the guidelines provided in the Association for Research in Vision and Ophthalmology ARVO statement for the Use of Animals in Ophthalmic and Vision Research, and were approved by Case Western Reserve University IACUC. C57BL/6 mice (6-12 wk old) and IL-6 2/2 mice on a C57BL/6 background were from the Jackson Laboratory (Bar Harbor, ME), Dectin-1 2/2 mice were kindly provided by Dr. Yoichiro Iwakura (University of Tokyo; Tokyo, Japan). Table 1 lists the genotype and phenotype of all strains utilized in this study. Aspergillus fumigatus was cultured on Vogel's minimal media (VMM) +2% agar and Fusarium oxysporum lycopersici was cultured on potato dextrose agar (PDA). Alternaria brassicicola was cultured in complete media as described previously [38]. All solid media used in this study were supplemented with 10 mM FeSO 4 to enhance conidia production by siderophore mutants. For neutrophil-fungus incubation assays, all fungi were grown in RPMI media w/o FeSO 4 supplementation. The Alternaria brassicicola Dnps2, Dnps6, and Dnps2/6 strains were kindly provided by Dr. B. Gillian Turgeon (Cornell University, Ithaca, NY).

Mouse model of Aspergillus and Fusarium keratitis
Aspergillus and Fusarium strains were cultured as described above for 2-3 days, and fresh conidia were disrupted with a bacterial Lloop, harvested in 5 ml PBS, and filtered through sterile PBSsoaked cotton gauze in a 10 ml syringe to obtain pure conidial suspensions. Conidia were quantified using a hemocytometer and adjusted in PBS to a final stock solution of 15-20,000 conidia/ml. Mice were anaesthetized with 1.25% 2, 2, 2-tri-bromoethanol in PBS. The corneal epithelium was abraded using a 30-gauge needle, through which a 2 ml injection containing conidia was released into the corneal stroma using a 33-gauge Hamilton syringe. Mice were examined daily under a stereomicroscope for corneal opacification, and quantified by image analysis using Metamorph software as described [8,23]. At each time point, animals were euthanized by CO 2 asphyxiation, and eyes were either placed in 10% formalin and embedded in paraffin, sectioned at 5 mm intervals and stained with periodic acid Schiff and Hematoxylin (PASH), or were placed in 1 ml of sterile PBS, homogenized and colony forming units (CFU) were quantified by manual count.

Topical and systemic drug delivery
Compounds were suspended in a commercial eye drop formulation (Alcon laboratories) or in PBS, and 8 ml was applied topically at 0 h and 6 h post-infection. Lactoferrin (1.25 mg/ml), deferiprone (10 mM), deferroxamine (10 mM), and simvastatin (4 mM) were purchased from Sigma Aldrich (St.Louis, MO). E.coli-expressed recombinant human lipocalin-1 was purified as described previously [17] and applied topically at 2 mg/ml. Iron-dextran and deferroxamine were purchased from Sigma and administered systemically to mice by daily intraperitoneal injections of 5 mg starting at day -2 until mice were euthanized. All animals were bred under specific pathogen-free conditions and maintained according to institutional guidelines.

Quantification of Aspergillus fungal mass and colony forming units (CFUs)
Growth of the RFP expressing A. fumigatus strain in the cornea was detected by fluorescent microscopy and quantified by Metamorph image analysis [8,23]. For assessment of fungal viability, whole eyes were homogenized under sterile conditions in 1 ml PBS, using the Mixer Mill MM300 (Retsch) at 33 Hz for 4 min. Subsequently, 100 ml aliquots were plated onto bacteriologic-grade Sabouraud dextrose agar plates, incubated for 24 h at 37uC (Aspergillus) or at 30uC (Fusarium and Alternaria), and the number of CFU/eye was determined by direct counting. The weight of the whole eye is consistent from one mouse to the next regardless of infection, and as we homogenize the entire eye and not just the cornea, we calculate CFU from the entire eye not just a representative sample. Fungal dsRed and CFU analysis do not have a linear correlation as hyphae of varying lengths show differences in dsRed fluorescence, but are still counted as a single CFU. Also, as homogenization can potentially damage branched hyphae, which may be more abundant in wild type compared with mutants, we may be underestimating the difference in CFUs between strains. All CFU graphs show pooled data from at least three repeat experiments.

Quantification of IL-6 protein in mouse corneas and serum
Corneas were homogenized in 150 ml reagent diluent (R & D Systems, Minneapolis, MN) using the Retsch MM 300 ball miller at 33 Hz for 4 min (Qiagen). Mouse serum was obtained as described below and assayed directly. IL-6 protein was quantified using a mouse IL-6 ELISA kit as per manufacturer's instructions (R & D Systems, Minneapolis, MN).

Quantification of neutrophils in mouse corneas
Corneas were dissected, cut into 8 small fragments, and incubated in 80 units of collaganese (Sigma-Aldrich) for 1-2 h. The cell suspensions were filtered, centrifuged at 300*g for 5 min at 4uC and washed in FACS buffer (PBS+1% FBS+0.5% Na azide). Cells were then incubated with anti-mouse CD16/32 antibody (Fc block, clone 93, eBioscience) for 10 min followed immediately by incubation with biotinylated rat anti-mouse NIMP-R14 or isotype-control for 45 min. Cells were washed and incubated with streptavidin-PE-Cy7 for 30 min in the dark. Cell suspensions were then analyzed utilizing a C6 Accuri flow cytometer with gates set based on isotype controls.
Quantitative PCR of infected corneas C57BL/6 mice and IL-6 2/2 mice were infected with A. fumigatus strain Af-dsRed as described above. At 24 h mice were sacrificed, corneas were excised, suspended in tissue lysis buffer (Qiagen, Valencia, CA) and homogenized using the Mixer Mill MM300 (Retsch) at 33 Hz for 2 min. Subsequently, RNA was extracted from samples using RNeasy mini kit according to the manufacturer's directions (Qiagen, Valencia, CA). Samples with a 260/280 (RNA:protein) ratio of 2.0 were used to generate cDNA using the superscript first strand synthesis system (Life technologies, Grand Island, NY) using standard methods. Real Time PCR was performed on the cDNA samples using the SYBR green system (Applied Biosystems, Carlsbad, CA). All primers used in this study are listed in Table 2 and were synthesized by Integrated DNA technologies (San Diego, CA). Fold change with respect to naïve uninfected corneas was calculated using the 2 2DDct method. Data are therefore presented as fold increases of relative gene expression (log (RQ)). RT-PCR samples were also analyzed by 2% agarose gel electrophoresis.

Quantification of iron content in mouse serum
Whole blood was obtained from mice by retro-orbital bleeding, and serum was recovered following blood coagulation. An iron assay kit (ABCAM, Cambridge, MA) was subsequently used to quantify Fe 2+ and Fe 3+ in the serum using manufacturer's instructions. Briefly, 25 ml of serum was added to 75 ml iron assay buffer and 5 ml iron reducer, which reduces Fe 3+ to Fe 2+ . Next, 100 ml of the iron-probe solution was added yielding a Fe 2+ -ferene S complex that absorbs light at 593 nm. Spectrophotometry was used to detect absorbance at this wavelength.
In vitro human neutrophil:hyphae growth inhibition assay Human neutrophils were isolated from normal, healthy donors using Ficoll-Paque Plus (GE) density centrifugation as described [8]. Isolated conidia from each A. fumigatus mutant were cultured in 200 ml SDA media (12,500/well) in black-wall 96 well plates with an optically clear bottom (CoStar 3720) until early germ tubes were observed (4-6 h). Wells were washed twice with sterile ddH 2 O and incubated 16 h with either RPMI media (+ Control), PBS (2 Control), or human peripheral blood neutrophils suspended in RPMI at 0.5-1*10 5 /well, which we know does not inhibit fungal growth [8]. After 16 h incubation, plates were washed and stained with 50 ml calcofluor white stain, which binds chitin (Fluka 18909) for 5 min in the dark. Subsequently, plates were washed three times with ddH 2 O and quantified by fluorometry (360/440 nm; Synergy HT; Biotek). In assays where no neutrophils were added, fungi were incubated in SDA media alone with or without inhibitors. Fungi cultured in 96well plates exhibit maximal growth by 16 h; therefore, this assay measures only relative decreases in fungal growth.

Statistical analysis
Statistical analysis was performed for each experiment using one way ANOVA with a Tukey post-hoc analysis using Prism software (GraphPad Software Inc, La Jolla, CA). A p value,0.05 was considered significant.

Ethics statement
All animals were treated in accordance with the guidelines provided in the Association for Research in Vision and Ophthal-