Introduction

Animal models serve to link in vitro work and clinical studies. Although results from clinical trials are the gold standard for guiding clinical decisions, they are very time-consuming and expensive. In vitro studies, while convenient and useful for adding valuable information, cannot serve as the sole basis for decisions impacting patient care. Animal models can be used in a variety of ways, from proof-of-concept studies and evaluations of virulence factors and disease pathogenesis, to preclinical work designed to elucidate the toxicology, pharmacokinetics, and efficacy of investigational drugs [1•, 2]. Animal models of invasive fungal infections have been widely used for these purposes. Many of the models that are used in medical mycology mimic the clinical pathogenesis of the invasive mycoses, and various questions can be examined rapidly while controlling for certain variables [1•]. However, no animal model is able to answer all questions, and each has limitations. Some may not accurately reflect the pathogenesis observed in humans, while others may be associated with significant variability [1•, 3]. Additionally, there may be substantial differences in pharmacokinetic parameters observed in certain animals as compared to humans. The purpose of this review is to discuss various animal models of invasive mycoses and highlight new developments that have refined these useful tools and improved our understanding of these infections.

Invasive Aspergillosis

Invasive aspergillosis is a leading cause of morbidity and mortality in immunocompromised patients. The incidence of this infection has increased in critically ill patients not traditionally considered at high risk, including those with chronic obstructive pulmonary disease and those receiving corticosteroids. Studies have shown that there is a clear dose response relationship between immunosuppression and susceptibility to invasive aspergillosis [4, 5]. Since many patients with this opportunistic mycosis have been made neutropenic due to cytotoxic chemotherapy and exposed to corticosteroids, many animal models have included both forms of immunosuppression in order to establish disease following Aspergillus inoculation. Combining these two different forms of immune suppression appears to augment each, as either alone may not result in consistently lethal infections [6]. While cytotoxic chemotherapy affects predominantly granulocytes, the effects of corticosteroids are pleiotrophic, affecting both T and B cells, macrophages, granulocytes, and monocytes [7, 8]. In addition, these agents may also influence fungal growth rate [9].

Within the last decade, changes in the epidemiology of invasive aspergillosis have been noted, with more non-neutropenic patients at risk for this invasive infection [4, 5]. In these patients, corticosteroid use is a major risk factor. Studies have demonstrated that the pathogenesis of invasive aspergillosis differs between neutropenic and non-neutropenic hosts where the primary form of immunosuppression is corticosteroids. In neutropenic murine model fungal proliferation, angioinvasion, and necrosis were abundant following intratracheal inoculation with A. fumigatus [10••]. In contrast, diffuse pneumonia associated with neutrophil influx and limited fungal burden and tissue invasion was observed in mice immunosuppressed with corticosteroids only, suggesting that the pathogenesis was due to an adverse host inflammatory response with minimal pathogen induced damage. This may also have implications for antifungal therapy, as amphotericin B was ineffective at prolonging survival in corticosteroid-immunosuppressed mice.

The form of immunosuppression may also influence the effects of virulence factors. Spikes et al. evaluated the role of gliotoxin, a mycotoxin produced by Aspergillus species, in the virulence of invasive pulmonary aspergillosis in both neutropenic and non-neutropenic murine models [11•]. In this study, gliotoxin was shown to be a virulence factor in non-neutropenic hosts, as virulence was clearly attenuated in mice administered high dose corticosteroids and infected with an A. fumigatus strain that did not produce gliotoxin secondary to deletion of gliP, a gene involved in the synthesis of this mycotoxin. In contrast, gliotoxin did not have a significant effect in neutropenic animals, as the survival rates were unchanged between groups infected with the wild-type strain and those infected with the gliP mutant. Thus, the mode of immune suppression in animal models of invasive aspergillosis may influence the virulence of the infecting organism, the pathogenesis of disease, and possibly responses to antifungal therapy.

In animal models of invasive aspergillosis, several routes of inoculation have been utilized, including intravenous injection, intranasal and intratracheal administration, and aerosol inhalation. Although intravenous injection has been extensively used and yields reproducible infection, this route of inoculation leads to limited pulmonary disease. However, it may be useful for early preclinical evaluations of antifungal agents [1216]. Intranasal and intratracheal challenge mimics the natural route of human exposure to Aspergillus species, and numerous investigators have employed these methods in different animal species [17, 18]. However, these modes of inoculation may be more labor intensive and complex than intravenous injection.

Inhalation of aerosolized conidia is a challenge method that recapitulates human exposure to Aspergillus, and various strategies have been utilized. One method utilizes a small acrylic chamber developed for aerosol delivery of Aspergillus conidia to immunosuppressed animals [19••]. This system utilizes a standard respiratory therapy nebulizer and compressed air to generate an aerosol within a chamber that is housed inside a biosafety cabinet. This standardized system has been shown to result in reproducible pulmonary infections in mice, rats, and guinea pigs [20, 21], and has been utilized by numerous investigators to evaluate various questions pertaining to invasive pulmonary aspergillosis, including virulence factors, new diagnostic assays, and novel therapeutic strategies [2224]. Although this method is economically feasible and highly reproducible, one limitation is the inability to accurately adjust the inoculum the animals are exposed to within the aerosol that is generated. One model that overcomes this limitation utilizes neutropenic rabbits that are intratracheally inoculated with a precise inoculum. This model has been has been extensively used to evaluate investigational antifungal agents, as well as changes in radiographic and surrogate markers of infection upon exposure to drug therapy [2530].

Other Moulds

Animal models have also been used to study other invasive mould infections, including fusariosis and scedosporiosis. These infections are often resistant to antifungal therapy and are associated with a poor prognosis [31, 32]. Murine models of disseminated disease have primarily been used to evaluate antifungal therapy against these mycoses [3340]. While key methods such as immunosuppression methodology and inoculum concentration varied between these studies, only modest improvements in outcomes were observed, which mimics the suboptimal results observed clinically. Studies in mice have been used to elucidate virulence differences among strains within the Scedosporium/Pseudallescheria complex [39, 41]. Harun et al. utilized an immunocompetent murine model to compare differences in virulence between S. aurantiacum and S. prolificans [41]. Distinctive DNA fingerprints were found among the S. aurantiacum strains, and virulence patterns that were independent of clinical, environmental or geographic derivation.

Animal models have also been used to study invasive mucormycosis, a highly aggressive infection that causes disease in immunocompromised and immunocompetent individuals. Robust models of disseminated mucormycosis in neutropenic mice and those with diabetic ketoacidosis have been used to evaluate disease pathogenesis, the effects of iron overload and chelation, as well as antifungal efficacy [4247]. Pulmonary disease has been established in mice using intranasal inoculation [4850]. Animal models of mucormycosis have demonstrated efficacy with combination therapy that includes an echinocandin plus amphotericin B, which has also been reported clinically [43, 51]. Interestingly, differences in outcomes have been observed between different lipid amphotericin B formulations in mice with diabetic ketoacidosis [52].

Disseminated Candidiasis

One of the most commonly used models of invasive fungal infections is that of disseminated candidiasis following intravenous inoculation. In this model, mice are typically inoculated with Candida via the lateral tail vein, and the infecting organisms are usually detected in multiple organs, including the kidneys, liver, spleen, lungs, and brains [53, 54•]. Candida species can also be isolated from the central nervous system, and a recent study demonstrated that the ability of this species to traffic to and infect the brain is due to binding to the heat shock protein gp96, a receptor expressed on brain endothelial cells, through surface expression of the fungal invasin Als3 [55, 56]. A rabbit model of hematogenous Candida meningoencephalitis has been utilized to evaluate antifungal efficacy within the central nervous system [57, 58]. However, infection seems to progress primarily in the kidneys, and is controlled in other organs, including the liver and spleen [53]. As disease progresses within the kidneys and because a high concentration of the infecting organisms can be found within this tissue shortly after challenge, this organ is primarily used by many investigators, and it was thought that mice succumbed to infection due to renal failure. However, in infections caused by C. albicans, mice succumb due to progressive sepsis. Spellberg et al. demonstrated that mice inoculated with C. albicans SC5314, a strain commonly used in animal models and for which the genome has been sequenced, became increasingly hypotensive and tachycardic during the course of infection [59]. In addition, systolic BP progressively declined, and other markers, including low bicarbonate levels, low pH, and decreasing glucose levels, were consistent with metabolic acidosis from severe sepsis. Fungal burden within the kidneys also correlated with serum creatinine levels and inversely correlated with serum pH.

The reason for the difference in the degree of disease progression between different tissues has recently been elucidated. Lionakis et al. investigated the temporal and spatial accumulation of leukocytes in different tissues following intravenous challenge of mice with C. albicans [54•]. Candida infection induced a robust neutrophilia in mice, and more leukocytes accumulated in the liver and spleen during the first 24 hours post-challenge. However, at later time points, further accumulation of neutrophils was only observed within the kidneys. Large abscess formation occurred only in the kidneys, while infection was controlled within the liver and spleens without the development of immune-related tissue damage. This late renal immune-pathology has subsequently been shown to be amplified by the chemokine receptor Ccr1, which is widely expressed on leukocytes, and mediates excessive recruitment of neutrophils from the blood to the kidneys [60].

Different Candida species can be used to establish infection. However, successful infection by species other than C. albicans may be dependent upon the host immune status, as other species, including C. glabrata, C. parapsilosis, and C. krusei are less virulent in these models and require aggressive immunosuppression in order to establish infection and achieve lethality [6165]. Immune status has also been shown to affect echinocandin efficacy. Response rates for caspofungin and anidulafungin, as measured by survival and changes in fungal burden within the kidneys, are in neutropenic mice compared to immunocompetent animals infected with C. albicans [66, 67]. Higher exposures to the investigational agent isavuconazole were also reported to be required to maintain efficacy in neutropenic mice [68]. Reduced efficacy has also been reported clinically in immunosuppressed patients treated with caspofungin [69].

In addition to studying pathogenesis and immune-mediated responses to infection, animal models of disseminated candidiasis have been extensively used in the evaluation of new diagnostic targets and assays, and therapeutic strategies. This includes studies defining the pharmacokinetic/pharmacodynamic (PK/PD) parameter associated with antifungal efficacy. Each of the clinically available antifungal agents has been evaluated in these types of studies, which have also utilized different Candida species. For the azoles and the echinocandins, the area under the concentration curve to minimum inhibitory concentration (AUC/MIC) ratio has been the PK/PD parameter most closely associated with efficacy [70, 71]. A rabbit model of hematogenous Candida meningoencephalitis has been utilized to evaluate echinocandin efficacy within the central nervous system [57, 58]. Based on the concentration dependent activity of the echinocandins and the clinical safety profile of the echinocandins [72], dosage escalation of these agents has been evaluated as a strategy to overcome microbiologic resistance. Mixed results have been reported from murine models of both C. albicans and C. glabrata, and the success of this strategy may be strain specific [64, 73, 74]. Reduced in vivo fitness and virulence have been reported in animals models for C. albicans strains with homozygous fks1 mutations compared to wild-type isolates [75]. These effects were associated with increased cellular chitin content and cell wall thickness. However, others have reported mixed results in animal models and that the reduced virulence may be strain specific [74].

While disseminated models of candidiasis following intravenous challenge are useful, it should be remembered that the majority of disease associated with these pathogens may be associated with biofilms. These organized cellular communities have reduced susceptibility to clinically available antifungal agents as well as the host immune response [76, 77]. Although numerous in vitro models have been developed to study biofilms, they cannot replicate the complex environment and changing conditions found within the host. To overcome these limitations, animal models of C. albicans biofilms have been developed. One model that has been used extensively is an in vivo central venous catheter biofilm model [78]. In this model, a subcutaneously tunneled catheter is placed into the internal jugular vein of anesthetized rats. After a 24-hour period of conditioning, a volume of a C. albicans suspension is instilled into the catheter and allowed to dwell for 4 hours, after which the inoculum is withdrawn and the catheter is flushed and locked with heparinized saline. A progressive increase in the number of viable cells composing the biofilm and within the rat kidneys has been demonstrated over a 72-hour period. Microscopically, the formed biofilm consists of a bilayered structure with yeast cells densely embedded in an extracellular matrix adjacent to the catheter surface, while the outermost layer consisted of yeast, hyphae, as well as host cells. This model has been used to evaluate the role of different adhesins, expression of different genes, and transcriptional networks in the development of C. albicans biofilms, as well as potential treatment strategies against biofilms [7881]. A similar rabbit central venous catheter biofilm model has been developed and has been used to evaluate different treatment strategies, including antimicrobial locks, against Candida biofilms [82, 83].

Mucosal Candidiasis

Mucocutaneous Candida infections, including oropharyngeal and vulvovaginal candidiasis, are common forms of candidiasis. In addition, Candida may be a component of the normal flora within the gastrointestinal tract. Various animal models of mucocutaneous candidiasis are available for the study of these infections and potential treatment strategies. However, C. albicans is not a natural colonizer of mucosal surfaces in rodents. Instead, C. pintolopesii may be found [84]. Thus, in order to establish colonization and infection with Candida species that are found within humans, various manipulations have been used. These include the use of immunosuppressive agents in order to dampen the host response to challenge, antibiotic treatment in order to alter the normal flora, and estrogen treatment in order to establish vulvovaginal candidiasis [3].

In one model of oropharyngeal candidiasis that has been extensively used, mice are immune suppressed with cortisone acetate [85]. Infection within the oral cavity is established in anesthetized animals by placing a cotton wool ball saturated with C. albicans blastospores sublingually where it is allowed to stay for 2 hours. Histopathology results have demonstrated that this model mimics the pseudomembranous oropharyngeal candidiasis observed clinically. This model has been used in the evaluations of drug and vaccine therapy, the host immune response to oral infection, biofilm regulators within C. albicans, and virulence factors [8688]. As fungal biofilms also play an important role in denture stomatitis in normally health denture wearers, other investigators have adapted oral candidiasis and denture stomatitis models in rats. One of these models utilizes acrylic denture material applied over the palate of rats rendered immunosuppressed with cortisone acetate, and the space between the hard palate and the acrylic material is inoculated with C. albicans [89]. It was reported that without antibiotic prophylaxis, a polymicrobial biofilm was formed, containing yeast, both Gram-positive and Gram-negative bacteria, and anaerobes, with microbial burden increasing over the first 48 hours. Antibiotic treatment significantly reduced the bacterial counts and resulted in a subsequent increase in fungal burden. Histopathology also showed hyphal invasion of the palate and tongue. Others have developed a less acute model of denture stomatitis, in which immunocompetent rats were fitted with a denture system that consisted of a fixed component and another that could be removed [90•]. This design allowed for the serial sampling of the Candida biofilm on the prosthetic material as well as the palate. Biofilm formation, as evident by an extracellular matrix upon confocal and scanning electron microscopy, was noted on the dentures at week 4 of the 8-week model and on the palate by week 6. A progressively increasing inflammatory response was also reported on histological analysis throughout the time course of the infection.

Rodent vulvovaginal candidiasis models have also been developed. These models are highly dependent on a prolonged state of pseudo-estrus, which can be induced through the administration of 17-β-estradiol [3]. Oophorectomy must also be done in rats in order to obtain the pseudo-estrus state. Estrogens are required, as they transform the columnar epithelium into thicker stratified squamous epithelium and increase the glycogen content, pH and growth substrates within this environment. The innate or adaptive defenses may also be dampened by estrogen, thus allowing for the establishment of infections [91]. These models have demonstrated that host defense against mucosal infections is multifaceted, and that protective mechanisms within the vagina may differ from those in the mouth or gastrointestinal tract [3, 92].

Candida albicans can transition from a commensal to opportunistic pathogen by dissemination from the gastrointestinal tract in critically ill patients and those that are highly immunosuppressed. Various models of gastrointestinal colonization with C. albicans have been used to evaluate factors important in this transition. These have primarily included murine models that have included immunosuppressed and antibiotic treated adult mice, or infant mice inoculated by oral gavage [1•]. However, achieving sustained colonization and consistent dissemination to visceral organs has been challenging [3]. One novel approach uses a purified diet that reduces the number of lactobacilli and organic acids present within the stomach, which may inhibit colonization with C. albicans. Yamaguchi et al. demonstrated that in normal adult mice fed a purified synthetic diet, fecal cultures were positive for Candida throughout the 5-week course of the model following oral inoculation [93]. In contrast, fecal cultures were negative for Candida one week after inoculation in mice fed a commercial diet. In addition, histologic evidence demonstrated mucosal invasion by C. albicans within the forestomach of mice fed the purified diet, and dissemination to visceral organs was also achieved in this group that was subsequently immunosuppressed.

Cryptococcosis

Various animal models of cryptococcosis are also available, and have been extensively used to evaluate the pathogenesis of disease, virulence factors, the host immune response, as well as therapeutic and preventative strategies [9499]. One of the primarily manifestations of disseminated disease in immunocompromised hosts is cryptococcal meningoencephalitis, as these organisms are highly neurotropic [100]. Because of this, meningoencephalitis has been extensively studied in models that use direct intracerebral or intracisternal inoculation [98, 99, 101]. Similar to other invasive mycosis pneumonia models, cryptococcosis can be inoculated into the lungs by intranasal instillation or via direct intratracheal injection. Pulmonary cryptococcosis has the advantage of mimicking the natural exposure that occurs in humans, while intravenous injection of Cryptococcus cells mimics hematogenously disseminated disease [102105]. Different animals species have been used for these models, including guinea pigs, mice, and rabbits. While invasive disease can be established in immunocompetent mice, guinea pigs and rabbits generally require immunosuppression for the establishment of infection [102]. Immunosuppression is especially important in rabbits, as these animals are naturally resistant to cryptococcosis due to their relatively high body temperature, which inhibits fungal replication.

Animal models have recently been used to evaluate differences in the pathogenesis of disease between different Cryptococcus species. Using a systemic model of infection in mice, Capilla reported that inoculation with C. neoformans var grubii lead to a temporal increase in fungal burden within the brain and liver, while fungi were not recovered from the brains of mice infected with C. gattii [106]. However, C. gattii is known to cause central nervous system disease in humans [107]. A recently adapted murine model of meningoencephalitis caused by C. gattii reported that this Cryptococcus species was more virulent than that of a C. neoformans isolate [108], a finding that is supported by clinical evidence [109, 110].

Invertebrate Models of Fungal Infections

When choosing an animal model, the benefits and limitations of the model must be considered (Table 1). For vertebrate models, the benefits include the ability to deliver a precise inoculum through various challenge routes, including those which mimic human exposure to the infecting organism. However, these models also have limitations, such as cost, time, and ethical constraints. Invertebrate models have become more widely used in the study of fungal infections. Invertebrate hosts that have been widely used in mycology include the fruit fly Drosophila melanogaster, the nematode Caenorhabditis elegans, larvae of the wax moth Galleria mellonella, and the amoebae Acanthomeba castellanii. Invertebrate models have the advantages of being less costly and are often less time-consuming than vertebrate models. The host and environment can be precisely controlled, and there are highly conserved innate immune mechanisms between invertebrates and higher animals that allow for the evaluation of innate host responses [111•]. Limitations of these models may include potential variability in the infecting inoculum, overly simplistic physiologies compared with humans, and some host mechanisms may not be highly conserved with higher animals. In addition, while antifungal activity observed in these hosts is often similar to in vitro assays and these models have been used to evaluate novel therapeutic combinations [112114], assessment of pharmacokinetic parameters such as bioavailability, distribution, and metabolism is problematic.

Table 1 General benefits and limitations of vertebrate and invertebrate models of fungal infections [1•, 3, 111•]

One invertebrate host that has been increasingly used in models of fungal infections is Drosophila melanogaster. Drosophila contain various signaling cascades of innate immunity. The Toll pathway is important for systemic immune responses, while epithelial responses to microbial exposure are mediated by the dual oxidase, JAK-STAT, and immune deficiency (Imd) pathways [115117]. Although wild-type flies are relatively resistant to fungal pathogens, lines that contain a mutation in various immune pathways, such as the Toll pathway, make these invertebrates susceptible of infection by fungi and bacteria. Modes of inoculation include injection of the organism through a puncture made on the dorsal side of the thorax, rolling of anesthetized flies over lawns of organisms, and ingestion. Drosophila have been used to evaluate several mycoses, including aspergillosis, candidiasis, fusariosis, scedosporiosis, cryptococcosis, mucormycosis, and Pneumocystis infections [111•, 113, 118, 119]. These have included studies of fungal virulence, innate immune response, as well as antifungal pharmacology. In addition, the Drosophila genome has been sequenced and annotated [120], and molecular tools, such as RNA interference, are becoming commercially available for genetic studies [121].

Another invertebrate model that has gained interest is that of Galleria mellonella. Larvae of this species have been utilized and injections can be made into the hemocoel via the prolegs. This route of delivery allows for administration of a standardized inoculum of infecting organism or a specific dose of antimicrobial agents. This model has been used to evaluate immune responses and virulence of different fungal pathogens as well as antifungals, including combination therapy with novel agents [111•, 122]. In studies of cryptococcosis, attenuated virulence of various mutant strains, including those affecting the capsule, different signaling pathways, and the production of melanin, have been reported, which are consistent with results from vertebrate models [123]. Although the Galleria genome has not been fully sequenced, recent work characterizing the genes responsible for immune responses has led to the development of gene microarrays for use in this invertebrate [124].

The transparent round worm Caenorhabditis elegans is another invertebrate host that has been used to study fungi. This model has the advantage of having a sequenced genome and the availability of various genetic tools [111•]. C. elegans is also easily propagated on a lawn of E. coli and inoculated by transferring the worms to a lawn of fungi or other bacteria. This model is adaptable to high-throughput screening for virulence genes as well as antimicrobial compounds [112]. C. elegans has been widely used to study cryptococcosis as all C. neoformans serotypes accumulate in the intestines and kill the nematodes [125]. Worms infected with less virulent mutants produce more progeny, which allows for rapid screening of virulence factors. While this model has been useful in evaluating cryptococcosis, studies with other pathogenic fungi are lacking.

Conclusions

Animal models are powerful tools that allow in-depth investigations into disease pathogenesis, host-pathogen relationships, pharmacodynamics, toxicity and efficacy of investigational therapeutic agents, and are instrumental in the development and refinement of diagnostic methodologies. In this review, we have provided an overview of commonly used animal models of invasive mycoses. We have also illustrated recent advancements in these models that have enhanced our understanding of these infections. The animal models presented range from classical, long used models to the more recently developed invertebrate models of mycotic disease. As with all models, each has distinct advantages and disadvantages. However, these models are robust tools that allow us to gain insights into the complex relationships between host and pathogen, as well as to evaluate new diagnostic and treatment strategies against invasive fungal infections.