Murine Model for Fusarium oxysporum Invasive Fusariosis Reveals Organ-Specific Structures for Dissemination and Long-Term Persistence

The soil-borne plant pathogen Fusarium oxysporum causes life-threatening invasive fusariosis in immunocompromised individuals. The mechanism of infection in mammalian hosts is largely unknown. In the present study we show that the symptoms of disseminated fusariosis caused by F. oxysporum in immunosuppressed mice are remarkably similar to those reported in humans. Distinct fungal structures were observed inside the host, depending on the infected organ. Invasive hyphae developed in the heart and kidney, causing massive colonization of the organs. By contrast, chlamydospore-like survival structures were found in lung, spleen and liver. Systemically infected mice also developed skin and eye infections, as well as thrombosis and necrosis in the tail. We further show that F. oxysporum can disseminate and persist in the organs of immunocompetent animals, and that these latent infections can lead to lethal systemic fusariosis if the host is later subjected to immunosuppressive treatment.


Introduction
Fungi of the genus Fusarium are important plant pathogens commonly found in soil, water and decaying organic matter [1]. In addition, Fusaria can cause a broad spectrum of diseases in humans, ranging from superficial or localized infections in healthy hosts to lethal disseminated fusarioses in immunocompromised patients [2]. Today, Fusarium is the second major cause of mould infections in immunocompromised patients after aspergillosis, and the incidence is increasing [3]. Fusarium species are among the most drug resistant fungal pathogens [4,5,6] and thus associated with high morbidity and mortality rates [7,8]. Rapid diagnosis is essential for successful antifungal therapy and survival of the patient. In many cases the entry sites of disseminated Fusarium infections remain unclear, but reported portals of entry include onychomycosis, the respiratory tract (particularly paranasal sinuses), the gastrointestinal tract, and the central venous lines. However, few model systems for evaluating virulence and pathogenesis of this group of fungi have been described. Here we developed a murine infection model and use it to demonstrate that the infection symptoms caused by disseminated Fusarium infection in immunosuppressed mice are remarkably similar to those reported in humans [2,3,9]. One of the most frequent symptoms of infection by Fusarium species is the development of skin lesions, which are most commonly found on the extremities. They are described as painful subcutaneous nodules that can result in thrombosis and tissue necrosis [10]. Skin lesions are often the only source of diagnostic material, because blood cultures remain negative in many cases, in spite of blood dissemination [11]. Therefore, it is of interest to characterize the clinicopathological features of skin lesions and their role in Fusarium infection and to establish them in the diagnosis and management protocols of fusariosis. Previous studies suggest that Fusarium can cause skin lesions even in immunocompetent individuals, and that upon immunosuppression these foci can lead to the development of invasive and disseminated fusariosis [12]. It has been recommended that patients to undertake severe immunosuppressive therapy should undergo a thorough skin evaluation before initiation of therapy [9].
Soil-borne Fusarium oxysporum is the causal agent of vascular wilt, a devastating disease affecting a large variety of economically important crops worldwide [13]. F. oxysporum also causes invasive infections in immunosuppressed individuals, being the second most frequent species of the genus after Fusarium solani, [3,14]. The incidence of invasive fungal infections leading to significant morbidity and mortality is rising because of the increase of the pool of immunocompromised patients caused by a dramatically increase number of solid organ transplants and the use of newer and more potent chemotherapeutic agents [15]. This is coupled with an increase in the number of reports of severe cases of invasive fusariosis often with lethal outcomes, due to the broad resistance to antifungal drugs [12,16]. F. oxysporum infections are often misdiagnosed as a result of the histopathological similarities with Aspergillus or non-Aspergillus hyalohyphomycoses [2,12,17,18].
Previous work established that a tomato pathogenic isolate of F. oxysporum f.sp. lycopersici can cause disseminated infection in immunocompromised mice [19]. The ability to cause disease in both plants and mammals makes F. oxysporum a unique multihost pathogen for studying fungal infection across different host kingdoms. To date several knockout mutants have been tested for virulence in immunosuppressed mice [19,20,21,22,23].
In the present study we used the F. oxysporum-mouse model to investigate infectious growth of F. oxysporum in mammals. We found that F. oxysporum develops distinct invasive structures, including hyphae, microconidia and chlamydospores, depending on the infected organ. We further show that F. oxysporum can cause thrombosis and necrosis in the tails of immunosuppressed mice, which can serve as model to study fungal skin infection in mammals. Finally we present strong evidence for dissemination of F. oxysporum in immunocompetent animals. Our results suggest that latent organ infections without macroscopic disease symptoms can initiate lethal systemic fusariosis, if the host is later subjected to immunosuppressive treatment.

F. oxysporum Isolate and Culture Conditions
F. oxysporum f.sp. lycopersici wild type strain 4287 was originally obtained from J. Tello, University of Almeria, Spain and stored as a glycerol microconidial suspension at 280 C. The pathogenicity of this wild type isolate on tomato plants is routinely confirmed by plant infection assays [24]. Fusarium conidia were cultured in potato dextrose broth at 28uC and 150 rpm for 4 days. For preparation of challenge inocula, microconidia were isolated by filtration as described previously [24], collected by centrifugation, washed, and resuspended in sterile physiological saline. The conidial concentration was adjusted with a hemocytometer to the desired density. The actual inoculum level was confirmed by plating serial dilutions on potato dextrose agar plates (PDA; Sigma) and incubating for 24 h at 28uC.

Animal Infections
All animal experimentation was done in accordance with UK Home Office regulations and was approved by both the UK Home Office and the University of Aberdeen ethical review committee.
Female BALB/c mice (Harlan, UK; 6-8 weeks old) were maintained in groups of up to 6 animals per cage. All mice were fed sterilized laboratory chow and water ad libitum. Each animal was individually marked and was weighed daily. Immunosuppression was performed by intraperitoneal injection of 150 mg cyclophosphamide (Sigma) per kg body weight, which was repeated every 3 days thereafter. Depending on the experiment, the immunosuppressive treatment initiated either on day 3 prior to infection (day 23), on the same day of infection (day 0), or 3 or 7 days after infection (day 3 or 7, respectively).
Mice were infected by intravenous injection of 0.2 ml of a conidia suspension into the lateral tail vein. Mice were observed for up to 28 days post-challenge and were humanely terminated when they showed signs of severe illness and/or their body weight reduced by more than 20% of their initial body weight. Data were used to construct Kaplan-Meier survival curves, with differences determined by log rank statistics using GraphPad Prism 5.

Tissue Burden and Histopathology
When mice were culled the hearts, lungs, kidneys, spleens and livers were aseptically removed, and one half of each organ was weighed and homogenized in 0.5 ml sterile saline. Ten fold serial dilutions of this homogenate were spread onto PDA. Plates were incubated at 28uC, colonies were counted after 48 h and CFU per gram for each organ were calculated. Fungal colony counts were converted to Log 10 values and compared using the analysis of variance test. Data were analysed with the software GraphPad Prism 5.
The remaining halves of the organs were embedded in Cryo-Mbed (Bright Instruments, UK) and flash frozen. Sections (8 mm) were stained with Periodic acid-Schiff (PAS)-hematoxylin and examined by light microscopy. In addition, samples of spleen tissue

Effect of Inoculum Size and Timing of Immunosuppressive Treatment on the Severity of F. oxysporum Systemic Infection
The effect of inoculum size on systemic infection by F. oxysporum in immunosuppressed mice was investigated. Intravenous inoculation with 2610 7 microconidia led to rapid development of systemic infection, with all mice becoming severely ill within 24 h (Fig. 1A). An inoculum of 2610 6 microconidia also caused 100% mortality, however disease progression was significantly slower (p = 0.0016), with all mice dying within 5 days. Finally, inoculation with 10 6 microconidia caused a significant drop in mortality, with only two of the six infected mice succumbing to infection within the 20 d observation period (p = 0.0024).
We next tested the effect of timing of immunosuppression on disease severity. When the immunosuppressive treatment was initiated 3 d prior to infection (day 23) and repeated every 3 days thereafter, all mice succumbed to infection ( Fig. 1A & B). By contrast, a delay in the start of the immunosuppression (day 0) led to a significant reduction in mortality (p = 0.0009) (Fig. 1B). Fungal burdens in the heart, lung and liver declined by 1-2 orders of magnitude over 18 days following inoculation, but remained at the same level in the kidneys and declined only slightly in the spleen (Fig. 1C).

F. oxysporum Displays Distinct Invasion Strategies in Different Organs of the Host
Variations in the fungal burdens observed (Fig. 1C) prompted us to conduct a histopathological analysis of the invasive growth of F. oxysporum in mice subjected to immunosuppressive treatment on day 0, and every 3 days thereafter. PAS-hematoxylin-staining of tissue sections revealed striking differences in fungal development between different organs. At 24 h post inoculation, a significant number of microconidia had germinated in the heart and kidney ( Fig. 2A & D), whereas only ungerminated microconidia were observed in lung, spleen and liver (data not shown). After 4 d postinoculation (dpi), branched fungal hyphae had developed in the heart and kidney (Fig. 2B & E), leading to the production of large mycelial colonies and the complete invasion of the infected tissue area at 16 dpi (Fig. 2C & F). Invasive growth was particularly aggressive in the kidneys (Fig. 3), where the fungal biomass was detectable macroscopically on the surface of the organ (Fig. 3A &  C). By contrast, invasive growth was not detected in the lung, spleen and liver. Instead, F. oxysporum developed chlamydosporelike structures, which were detected at 4 dpi and remained visible at 16 dpi (Fig. 2G, H & I).

F. oxysporum Causes Thrombosis and Necrosis in Tails and Infections in Eyes of Immunosuppressed Mice
We noted that the immunosuppressed mice infected with F. oxysporum developed bulges and swellings which appeared at multiple sites along the tail, not restricted to the site of inoculation (Fig. 4B). Some of these swellings subsequently turned into open lesions (Fig. 4C & E). In addition, some mice displayed macroscopic symptoms of necrosis at the tail. The necrotic tissue was initially visible as a black region at the tip (Fig. 4D & E) which spread upwards along the tail, in some cases leading to loss of the tip (Fig. 4F). Calcofluor white (CFW) staining of tissue samples obtained from the open wounds revealed the presence of chlamydospores and hyphae of F. oxysporum (Fig. 4G & H).
Another observation associated with the presence of F. oxysporum in the immunosuppressed mice was the development of infection symptoms around and within the eye (Fig. 4I-M). The presence of fungal structures in the infected eye tissue was confirmed microscopically (Fig. 4N).

F. oxysporum Disseminates and Persists in Immunocompetent Mice
Inoculation F. oxysporum typically leads to 70-100% mortality in immunosuppressed mice, whereas immunocompetent animals fail to display detectable signs of illness ( [19] and this work, data not shown). Thus, we investigated whether F. oxysporum was able to disseminate and persist in an immunocompetent host. Strikingly, fungal burdens at 4 dpi in immunocompetent mice inoculated with 2610 7 microconidia were similar or only slightly lower than in animals subjected to immunosuppressive treatment on day 0, and every 3 days thereafter (Fig. 5A). F. oxysporum microconidia and germlings were observed in the heart, lung, kidney, spleen and liver of immunocompetent animals (Fig. 5C-F). At 11 dpi the fungal burdens in immunocompetent animals had declined by approximately two orders of magnitude in most of the organs, but remained relatively high (,10 5 CFU per g tissue) in the spleen (Fig. 5B). Indeed, PAS-hematoxylin and CFW staining confirmed the presence of chlamydospores, as well as germinated and ungerminated microconidia in different organs of infected animals at 4 and 11 dpi (Fig. 4G). The presence of fungal propagules in the organs of immunocompetent mice even at .10 dpi suggests that F. oxysporum can develop resting structures such as chlamydospores, but also actively growing mycelium to persist within the immunocompetent host. Such a host might subsequently become susceptible to invasive fusariosis if the immune system is disrupted. When immunocompetent mice inoculated with F. oxysporum were subjected to immunosuppressive treatments starting at either 3 or 7 dpi and repeated every 3 days thereafter, one mouse in each group died at 16 and 15 dpi, respectively (Fig. 6A). Fungal burden was determined in different organs from one of the killed mice, revealing high counts of fungal propagules (Fig. 5B). Tissue sections confirmed the presence of invasive mycelial growth in multiple organs, including heart, kidney spleen and liver, indicative of systemic fusariosis (Fig. 5C-F).

Discussion
Fungi of the genus Fusarium are common soil inhabitants and notorious plant pathogens, and have long been recognized as etiologic agents of focal infections of the skin, nails, and cornea of humans [25]. Moreover, invasive Fusarium infections in immunosuppressed patients are associated with high mortality rates [7,8]. Rapid antifungal therapy, achieved by a prompt diagnosis, is essential for survival of these patients. A critical step is the identification of Fusarium, which is often made difficult by some histopathological similarities with Aspergillus. Infections caused by both Fusarium and Aspergillus are characterized by hyaline branching septate hyphae at acute angles, which invade the blood vessels causing thrombosis and tissue infarction [18,26]. Therefore, a detailed characterization of the invasive behaviour of Fusarium within the host, including growth and development of fungal structures formed by the pathogen and associated with disease, is required and will have important consequences for the diagnosis and management of fusariosis. Here we used the mouse model to investigate Fusarium infection in mammals. The usefulness of the animal model is highlighted by the finding that the disease symptoms observed in mice are remarkably similar to those reported in human fusariosis [2,3,9].
A key result of this study is that F. oxysporum displays distinct invasion strategies in different organs of mice. While the fungus initiated hyphal growth 24 h after inoculation in the kidney and the heart, no such growth was detected in the spleen, liver or lung. Instead, chlamydospore-like structures were formed in these organs. This novel finding represents an important advance in our understanding of Fusarium infections in mammals. Chlamydospores are thick-walled cells generally developed through the modification of hyphal and conidial cells. Their formation is induced by aging or unfavorable environmental conditions such as low temperatures or carbon starvation. Chlamydospores represent the principal structure for long-time survival during unfavorable periods in the soil, and play an important role as primary inoculum for plant root infection [27,28,29,30,31]. Our findings highlight the importance of detailed histopathological analysis of infection structures, in addition to routine determination of fungal burden. For example, kidneys contained reduced fungal burdens compared to the spleen, but microscopic analysis revealed massive mycelial invasive growth in this organ, which in some instances was even visible macroscopically on the organ surface. Thus, although fungal burden is a useful parameter for assessing fungal dissemination in the host, it fails to provide detailed information on the impact of filamentous pathogens on the infected organs. It was reported previously that during F. solani or A. fumigatus infection, quantitative culture led to underestimation of absolute fungal burden as compared to non-culture-based methods such as quantitative PCR or determination of galactomannan levels by enzyme immunoassays (EIAs) [32,33,34,35]. In an inhalational rat model of invasive pulmonary aspergillosis (IPA), both real-time nucleic acid sequence-based amplification (NASBA) and qPCR showed a progressive increase of fungal biomass in lung tissue, whereas CFU counts were stable over time [36].
A frequent symptom associated with Fusarium infections is the development of superficial skin lesions, described as subcutaneous nodules that can result in thrombosis and tissue necrosis [10]. In more than 60% of disseminated Fusarium infections skin biopsies of affected patients revealed necrosis of skin papules [37] and microvessel thrombosis associated with fungal hyphae [10]. Here we found that infection of immunosuppressed mice by F. oxysporum led to macroscopic symptoms in the tail, including necrosis of the skin, swellings and wounds and even loss of the tail tip. Microscopic analysis of the skin lesions confirmed that the nodules contained both chlamydospores and hyphae of F. oxysporum. CFW staining proved to be highly useful for detection of progressive systemic fusariosis in the mouse system, and could therefore be used as a rapid diagnostic tool in immunocompromised patients. Because skin lesions often appear during early stages of Fusarium infection before the disease spreads to the trunk and the extremities, they are crucial for the diagnosis and management of fusariosis, allowing rapidly initiation of a specific treatment in order to prevent progression of skin lesions and further necrosis.
F. oxysporum keratitis is one of the most important causes of corneal ulcers, ocular morbidity and visual loss in developing nations [38]. It is also a common type of infection caused by F. solani in immunocompetent individuals worldwide [39,40,41,42,43,44]. A previous study on disseminated F. solani infection via inoculation of microconidia in the lateral tail vein of immunocompetent mice reported disease symptoms both in intraocular structures and in neighbouring muscles [45] Here we provide evidence for the presence of fungal biomass around and within the eyes of immunosuppressed mice, associated with disseminated F. oxysporum infection. This result supports the view that fungal keratitis should be established as part of the diagnostic protocol for disseminated Fusarium infections in order to prevent a delay in diagnosis or inadequate treatment, which may lead to loss of the affected eye. Further, we observed that mice affected by disseminated F. oxysporum infection showed a ''twister'' phenotype indicating a possible infection of the brain. Supporting this idea, analysis of fungal burden in the brain of an animal displaying a twister phenotype showed the presence of fungal biomass (data not shown), suggesting that F. oxysporum is able to enter the mouse brain.
Initiation of immunosuppressive treatment at day 23 led to death of all infected animals, while mortality was significantly lower when immunosuppression was started in parallel to infection. Thus, the immune status of the host has a major effect on the severity of infection by F. oxysporum. Importantly, we show here for the first time that F. oxysporum can also disseminate and persist in immunocompetent individuals. Unexpectedly, fungal burdens in immunocompetent mice at 4 dpi were only slightly lower than in immunosuppressed animals, with microconidia and germlings observed in the heart, lung, kidney, spleen and liver of immunocompetent animals even after more than 10 dpi. The finding that F. oxysporum can persist in an immunocompetent mammalian host is highly relevant, because these fungal foci could lead to subsequent systemic infection upon immunosuppressive treatment. Indeed, one mouse in each group later succumbed to fungal infection. We conclude that invasive fusariosis in these animals was caused by chlamydospore-like structures which had persisted in the organs, suggesting that latent fungal survival structures have the potential to initiate invasive mycelial growth once the immune system is no longer effective.