Kazuhisa Sekimizu, Laboratory of Microbiology, Graduate School of Pharmaceutical Sciences, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo, 113-0033, Japan. E-mail: firstname.lastname@example.org
Aims: To develop an in vivo system that could quantitatively evaluate the therapeutic effects of antifungal drugs using a silkworm infection model with Cryptococcus neoformans.
Methods and Results: Silkworms reared at 37°C died after an injection of viable serotype A C. neoformans fungus into the haemolymph. The serotype A C. neoformans, which is known to have higher mammal pathogenicity than the serotype D, was also more virulent against the silkworm. Furthermore, the deletion mutants of genes gpa1, pka1 and cna1, which are genes known to be necessary for the pathogenesis in mammals, showed an increase in the number of fungal cells necessary to kill half of the silkworm population (LD50 value). Antifungal drugs, amphotericin B, flucytosine, fluconazole and ketoconazole, showed therapeutic effects in silkworms infected with C. neoformans. However, amphotericin B was not therapeutically effective when injected into the silkworm intestine, comparable to the fact that amphotericin B is not absorbed by the intestine in mammals.
Conclusions: The silkworm–C. neoformans infection model is useful for evaluating the therapeutic effects of antifungal drugs.
Significance and Impact of the Study: The silkworm infection model has various advantages for screening antifungal drug candidates. We can also elucidate the cryptococcal pathogenesis and evaluate the in vivo pharmacokinetics and toxicity of each drug.
Pathogenic fungi cause several serious symptoms in human patients with impaired immune responses. For example, patients taking immunosuppressive drugs for receiving an organ transplant or patients suffering from AIDS are vulnerable to fungal infections. Candida, Aspergillus and Cryptococcus species are known as major human fungal pathogens worldwide (Morschhauser 2010).
Cryptococcus neoformans is a pathogenic fungus that causes opportunistic infection, respiratory illness and meningitis. It is one of the most frequent causes for death among persons with AIDS, especially in sub-Saharan Africa, where 13–44% of HIV/AIDS deaths are owing to Cryptococcus infection (Park et al. 2009). On top of this, C. neoformans not only shows natural resistance to echinocandin, a commonly used antifungal reagent, but also continues to cause serious clinical problems through the emergence of resistant strains (Morschhauser 2010). Cryptococcus neoformans is also known to resist host immunity by producing pathogenic factors such as a capsule, melanin and proteases. In addition, signalling pathways regulating the induction of fungal pathogenic factors have been reported (Steinbach et al. 2007; Kozubowski et al. 2009).
Several antifungal agents are in clinical use for treating fungal infections. However, the emergence of drug-tolerant pathogenic fungi is a serious problem. To address this problem, new antifungal agents must be developed (Ghannoum and Rice 1999). Generally, for the development of drugs against infectious diseases, chemical compound libraries are screened to identify those that inhibit pathogen growth. Most candidate compounds that inhibit pathogen growth in vitro, however, do not have therapeutic effects in animal infection models. A major reason for this is that most compounds in chemical libraries have pharmacokinetic problems and/or toxic effects in animals. Therefore, it is necessary to evaluate the pharmacokinetics and toxicity of the candidate compounds for the development of safe and effective anti-infectious drugs.
We are proposing the use of the silkworm to evaluate anti-infectious drugs. The silkworm has a number of advantages as an experimental animal. For example, methods for rearing silkworms have been well established during the long history of sericulture. Secondly, the whole genomic analyses of the silkworm have been completed, and a method to construct transgenic silkworms is now available (Tamura et al. 2000; Tomita et al. 2003; Mita et al. 2004; Shimomura et al. 2009). In addition, as we previously reported, the virulence of pathogenic bacteria and the therapeutic effects of antibiotics can be quantitatively evaluated using the silkworm. The silkworm dies after an injection with human pathogens such as Staphylococcus aureus, Pseudomonas aeruginosa and Candida albicans, and administration of antimicrobial or antifungal agents prevents these pathogens from killing the silkworm (Kaito et al. 2002; Hamamoto et al. 2004). Accordingly, the ED50 value, the amount of drug necessary to rescue 50% of the silkworm population, calculated from the silkworm infection models is often consistent with that in mammalian models (Hamamoto et al. 2004). In addition, we reported that the silkworm model is also effective for evaluating therapeutic effects of antiviral medicine (Orihara et al. 2008). Using our silkworm–baculovirus infection model, we identified a new antiviral compound, cynzeilanine, from Maoutou, an herbal medicine (Orihara et al. 2008).
The most important aspect of the silkworm infection model is that both the toxicity and the pharmacokinetics of candidate compounds can be evaluated simultaneously. Drugs absorbed by the silkworm intestine are metabolized by cytochrome P450, as is the first-stage reaction commonly formed in mammals. The drugs are sequentially excreted from the intestine after a series of conjugation reactions (Hamamoto et al. 2009), which is also commonly observed in mammals. In addition, chemical compounds that are stable in mammalian blood are also tend to be stable in silkworm haemolymph (Asami et al. 2010). The doses of various toxic compounds that kill 50% of the silkworms population (LD50) are similar to those in mammals (Hamamoto et al. 2009). Therefore, the pharmacokinetics and toxicity of drug candidates can be evaluated using the silkworm. Based on these findings, we propose that a silkworm infection model is useful for evaluating the therapeutic effects of antibiotics, antifungal agents and antiviral drugs. In this study, we tested whether the therapeutic effects of antifungal agents could be evaluated in the silkworm infection model.
We considered the applicability of the silkworm as an infection model animal for quantitative measurement of the therapeutic effects of antifungal agents against C. neoformans infection. Here, we established a silkworm infection model with C. neoformans.
Materials and methods
Culture conditions and fungal strains
Cryptococcus neoformans was cultured in liquid YPD (1% yeast extract, 1% polypeptone and 1% dextrose) medium at 30°C with shaking at 150–180 rev min−1. The overnight culture was diluted 100 times with a fresh YPD medium and cultured at 30°C for 18–24 h with shaking at 150–180 rev min−1 for the infection experiment. Details of the fungal strains used in this study are given in Table 1.
Table 1. Cryptococcus neoformans strains used in this study
Fertilized eggs of the silkworm, Bombyx mori (Hu·Yo × Tukuba·Ne; Ehime Sanshu, Ehime, Japan), were maintained in disposable plastic containers at 27°C. Hatched larvae were reared to the fifth instar at 27°C on an artificial diet, SilkMate 2S (Nosan Corp., Tokyo, Japan). All experiments were performed using fifth-instar larvae fasted overnight during the fourth ecdysis unless indicated otherwise. The hatched fifth-instar larvae were fed antibiotic-free artificial diet (Silkmate; Katakura Industries, Tokyo, Japan) for 1 day before performing the infection experiments.
Silkworm infection experiment
The silkworm infection experiment was performed according to the previously established method (Kaito et al. 2002, 2005; Matsumoto et al. 2007). Fungal suspensions (0·05 ml) were injected into the haemolymph of the larvae through the dorsal surface using a 27-gauge needle. The 18- to 24-h fungal cultures were diluted with saline and used for the experiment. The injected larvae were maintained without feed at 37°C unless mentioned otherwise. Twofold serially diluted samples were injected into silkworms and their survival was determined 48 h after injection. The LD50 value was defined as the activity that killed half of silkworms (50% lethality). Haemolymph was collected from the larvae through a cut on the first proleg at points 3 and 18 h after injection, to determine viable fungal cell counts in the silkworm haemolymph. After appropriate dilution, the samples were spread on YPD agar plates and incubated for 2 days, and the number of colonies was counted.
Cells (50 μl in 0·9% NaCl, 1·0–2·0 × 107 cells) of the H99 strain were injected into the haemolymph of silkworm larva (c. 2 g), followed by the injection of antifungal drugs (50 μl in 0·9% NaCl). More than five silkworms were used for each dose of the antifungal drugs. Silkworms were incubated at 37°C and their survival was determined 48 h after injection. The ED50 values were determined as the amount of drug required for 50% survival, normalized per 1 g of silkworm.
MICs were determined using a microdilution method recommended by the Clinical and Laboratory Standards Institute (CLSI) (Institute 2006). Each concentration (100 μl) of the antifungal drugs was placed in a 96-well microplate, and 100 μl of 1000-fold diluted fungal culture with YPD culture medium was added. The OD600 value was determined after incubation for 48 h at 30°C. The MIC was determined as the antifungal drug concentration that inhibited density to the extent that the OD600 value was <50% that of the control sample.
Amphotericin B and flucytocine were purchased from Sigma Aldrich (St Louis, MO, USA). Fluconazole and Ketoconazole were purchased from LKT Laboratories, Inc (St Paul, MN, USA). Micafungin was kindly provided by Astellas Pharma Inc (Tokyo, Japan). These drugs were dissolved in the following solutions: Amphotericin B; 10% DMSO in 0·9% NaCl, Flucytosine; 0·9% NaCl, Fluconazole; 10% DMSO in 0·9% NaCl, Ketoconazole; 10% DMSO in 0·9% NaCl and Micafungin; 0·9% NaCl. We have checked that solvent-only controls have no therapeutic effects in our experiments.
Data are shown as means ± standard deviations obtained from three independent experiments. Statistical significance between groups in Fig. 3, Tables 2 and 3 was evaluated using a two-tailed Student’s t test. Statistical analysis of the survival curves in Fig. 2 was performed using one-sided rank log tests (the prism software package, graphpad software). A P-value of <0·05 was considered statistically significant.
Table 2. LD50 of Cryptococcus neoformans strains for silkworm
LD50 (×106 CFU per larva)
6 ± 3
7 ± 1
8 ± 1
<0·0001 (vs H99)
<0·0001 (vs H99)
<0·0001 (vs H99)
Table 3. LD50 of Cryptococcus neoformans mutants for silkworm
LD50 (×106 CFU per larva)
6 ± 3
<0·01 (vs H99)
25 ± 8
<0·01 (vs H99)
31 ± 7
<0·01 (vs H99)
Establishment of the silkworm–Cryptococcus neoformans infection model
To quantitatively measure the therapeutic effects of drug candidates on infectious diseases, establishment of an appropriate animal–pathogen infection model is essential. In this study, we examined conditions under which the silkworm is killed by C. neoformans. First, we compared the silkworm killing ability of C. neoformans at 37°C and at 27°C. When the silkworm was incubated at 37°C after an injection of a range of fungal cells into the haemolymph, the number of fungal cells needed to kill half of the silkworm population (LD50) was 6 × 106 cells. In contrast, when the silkworms were incubated at 27°C, the fungi did not kill the silkworms at all, even at the dose of 6 × 107 cells (Fig. 1). Furthermore, autoclaved fractions of C. neoformans did not have silkworm killing ability (Fig. 2). Therefore, viable fungi and a higher temperature were necessary for the pathogenesis against silkworm. We also tested whether C. neoformans could grow in the silkworm after inoculation. The number of viable fungal cells was increased 10 times between 3 and 18 h after injection (data not shown). Inoculation of samples into the intestinal tract of the silkworm is the equivalent to oral administration in mammals (Hamamoto et al. 2004). We experimented on C. neoformans pathogenesis when infection occurred in the intestinal tract. Silkworms were not killed by injection of C. neoformans into the intestinal tract (i.m.) (data not shown). Cryptococcus neoformans strains are classified based on their serotypes, A and D, and the serotype A strains are more virulent than the serotype D strains in mouse models (Lin et al. 2008). We tested whether the pathogenicity of serotype A is also higher than that of serotype D in the silkworms. The LD50 of serotype A strains, such as H99, KN99a and KN99α, was seven times smaller than that of serotype D strains, such as KN3501a, KN3501α and B4500 (Table 2). Capsule formation is known to be associated with cryptococcal virulence (Kwon-Chung and Rhodes 1986). We tested whether C. neoformans formed capsule during infecting silkworms. Cryptococcus neoformans collected from haemolymph of silkworms reared at 37°C showed capsule formation. These cells showed thicker capsule and cell size enlarged to approximately 2-fold when compared to C. neoformans collected from silkworm reared at 27°C (Fig. 3). The gpa1, pka1 and cna1 genes contribute to the pathogenesis of C. neoformans in mammals, and their deletion mutants are known to have decreased virulence (Alspaugh et al. 1997; Odom et al. 1997; D’Souza et al. 2001; Steinbach et al. 2007; Kozubowski et al. 2009). We evaluated the pathogenicity of these mutant strains in our silkworm infection model. The LD50 of each deletion mutants of genes gpa1, pka1 and cna1 was more than four times higher than that of the wild-type, H99 of serotype A strain (Table 3). These results suggest that the silkworm infection model provides results consistent with those of mammalian infection models with regard to the pathogenicity of C. neoformans.
Quantitative evaluation of the therapeutic effects of antifungal drugs against Cryptococcus neoformans
We then tested whether C. neoformans infection in the silkworm model could be cured by administering antifungal drugs that are in clinical use. Amphotericin B, flucytosine, fluconazole and ketoconazole were chosen as representative antifungal agents (Morschhauser 2010). Amphotericin B, which is a polyene antifungal drug, compromises fungal cell membrane function through binding to ergosterol. Flucytosine is metabolized to 5-fluorouracil by the fungal cytosine deaminase, which is further converted to 5-fluorodeoxy UMP and 5-fluoro UTP in the fungal cell, inhibiting fungal DNA and RNA synthesis. Fluconazole and ketoconazole, which are azoles, inhibit the biosynthesis of ergosterol and damage fungal cells. Finally, micafungin, which is a candin, targets ß-1,3-glucan synthase and thereby inhibits fungal cell wall glucan synthesis (Morschhauser 2010). Administering these antifungal agents into the silkworm haemolymph prevented death of the silkworm caused by C. neoformans infection. We determined the amounts of each antifungal agent needed for the survival of 50% of the animals (ED50). We also determined the LD50 value of each agent to assess the toxicity of antifungal drugs. The toxicities of these antifungal agents against silkworms were especially low for those in clinical use, and the ratio of LD50 to ED50 was <0·08 for all four agents tested (Table 4). Micafungin, which is not clinically used to treat cryptococcal infections, did not protect the silkworms from death when infected with C. neoformans (Table 4). In addition, amphotericin B, which is not effective when orally administered to humans, also did not have therapeutic effects when injected into the silkworm intestine (Table 5).
Table 4. ED50, MIC, ED50 per MIC, LD50 and ED50 per LD50 of antifungal drugs in silkworm infection model with Cryptococcus neoformans
ED50 (μg g−1 of larva)
MIC (μg ml−1)
ED50 per MIC ratio
LD50 (mg g−1 of larva)
ED50 per LD50 ratio
14 ± 10
4 ± 2
6 ± 1
21 ± 7
2 ± 1
7 ± 6
19 ± 2
0·1 ± 0·1
Table 5. Intra-midgut administration of amphotericine B does not have therapeutic effects in a silkworm model
ED50 (μg of antifungal agent g−1 of larva) of drug administrated by the following route
14 ± 10
6 ± 1
9 ± 7
2 ± 1
9 ± 3
19 ± 2
14 ± 10
The findings of this study indicate that the silkworm is a suitable mammalian animal substitute for quantitatively evaluating the therapeutic effects of antifungal agents against C. neoformans infection. Experiments using mammals, such as mouse and rat, should be performed following the 3Rs, i.e., Replacement, Reduction and Refinement. The 3Rs concept is an internationally recognized principle for responsibly conducting animal experiments (Russell and Burch 1959). The use of silkworms is consistent with the idea of Replacement, one of the 3Rs.
To date, studies on the pathogenicity of C. neoformans using invertebrate animal hosts such as Caenorhabditis elegans, Drosophila melanogaster and Galleria mellonella as infection hosts have been reported (Mylonakis et al. 2002, 2005; Apidianakis et al. 2004). In the infection model utilizing G. mellonella, the host-killing ability of C. neoformans was more apparent at 37°C than at 30°C (Mylonakis et al. 2005), and heat-killed fungus did not kill G. mellonella (Mylonakis et al. 2005). Similar to this, we also observed that C. neoformans killed the silkworms at 37°C, but not at 27°C, and administration of autoclaved fungus did not kill the silkworms. Furthermore, we showed that injected C. neoformans proliferated in the silkworm haemolymph at 37°C. Thus, we consider that proliferation of the injected fungus is necessary for inducing silkworm death. Our previous report did not include C. neoformans infection (Hamamoto et al. 2004). Because previous experiments were performed at 27°C, which made it difficult to assess C. neoformans infection. In addition, intestinal challenge with C. neoformans failed to kill the silkworms. Earlier studies reported that oral challenge with C. neoformans could not readily cause disseminated cryptococcosis in immunocompetent mice (Salkowski et al. 1987). According to these researchers, no mortality was observed in C. neoformans-colonized immunocompetent mice during a 12-week observation period after oral challenge. Also, serotype A strains are known to have greater ability to kill mammalian hosts than serotype D strains. Consistently, in our study using silkworms, H99 of a serotype A strains had higher pathogenicity than KN3501a of a serotype D strains. The extracellular structure, which determines the serotype of C. neoformans, may affect pathogenicity against both mammals and silkworms.
The cna1, gpa1 and pka1 genes are essential for the pathogenicity of C. neoformans against mammals. Deletion mutants of these genes also showed lower host-killing activity in silkworms, compared with wild-type fungus. The cna1 gene encodes Cna1, the catalytic subunit of calcineurin. The Cna1 is reported to have a role in the calcineurin signalling pathway, which is essential for its pathogenicity against mammals (Odom et al. 1997; Steinbach et al. 2007). On the other hand, Gpa1 and Pka1 exert their pathogenicity through a mechanism independent of the calcineurin pathway (Alspaugh et al. 1997; Kozubowski et al. 2009). The results obtained using our C. neoformans infection model using silkworms were consistent with those of previously reported animal infection models. We propose that the capsule formation of C. neoformans is possibly important in its silkworm killing ability for the following reasons: (i) C. neoformans has a higher capsule-forming activity at 37°C compared with 30°C (Kwon-Chung and Rhodes 1986). (ii) Cryptococcus neoformans collected from the haemolymph of silkworm reared at 37°C had thicker capsule than when the silkworms were reared at 27°C. (iii) Deletion mutants of genes gpa1 or pka1, which are both genes necessary for capsule formation, had lower silkworm killing activity (Table 3). (iv) Serotype A has higher capsule-forming activity than serotype D (Hicks et al. 2004). (v) The H99 strain, which is serotype A, had stronger silkworm killing activity than the KN3501a strain, which is serotype D (Table 2). As Gpa1 and Pka1 are also involved in melanin synthesis, the capsule-forming mutants may need further evaluation. Recently, giant cell phenotypes have been observed for C. neoformans-infecting mice (Okagaki et al. 2010; Zaragoza et al. 2010). We propose that the difference in the silkworm rearing temperatures 27 and 37°C is the cause of the difference in the fungal cell size.
We tested whether we could quantitatively evaluate the therapeutic effects of antifungal drugs against C. neoformans using this silkworm infection model. Thus, we determined the ED50 values of amphotericin B, flucytosine, fluconazole and ketoconazole. The silkworm infection model can also be applied to evaluate the pharmacokinetics of antibiotics, based on the ED50 per MIC ratio (Hamamoto et al. 2004). The ED50 per MIC values of clinically used antibiotics are <10 (Hamamoto et al. 2004), and according to our experiments, the ED50 per MIC values of amphotericin B, flucytosine and fluconazole in the silkworm infection model were also <10 (Table 4). On the other hand, the ED50 per MIC value of ketoconazole, whose clinical use is limited to external application, was 190 in the silkworm model (Table 4). Both fluconazole and ketoconazole are azole-type antifungal drugs, and ketoconazole exhibited a lower MIC against C. neoformans. The ED50 value for ketoconazole was higher than that of fluconazole in the silkworm model (Table 4). These results are consistent with the clinical experience with both fluconazole and ketoconazole for the treatment of cryptococcal meningitis (Hoffman et al. 2000). Thus, based on the ED50 per MIC value in the C. neoformans infection model using silkworms, we can evaluate the therapeutic effects of drug candidates that cannot be estimated simply from MIC values. Furthermore, amphotericin B, which is not orally available for human patients, did not show therapeutic effects in silkworms when administered into the intestine. This result suggests that, as in humans, amphotericin B does not permeate the silkworm intestine. Therefore, the therapeutic effects of orally administered antifungal drugs could be evaluated in silkworms by injecting drugs into the intestine. This study is the first report using the silkworm infection model that therapeutic effects of antifungal drugs with different administration routes are changed.
Utilizing invertebrate infection model instead of mammals for the evaluation of therapeutic effects of drug candidates is advantageous for the following reasons: (i) lower maintenance cost, (ii) less space required for keeping the animals, (iii) fewer ethical problems compared with mammals, (iv) smaller body weight, requiring less drug for evaluation and (v) shorter time courses for infection experiments. Moreover, the silkworm evaluation system has the following merits compared with previously proposed invertebrate systems such as C. elegans and D. melanogaster: (i) the body size large enough to perform quantitative administration of drug solutions, (ii) the variety of available injection methods into the haemolymph and intestine, (iii) the tolerance to rearing temperatures of 37°C up to several days, which is necessary for virulence of certain pathogens like C. neoformans, (iv) availability of the data regarding pharmacokinetics and toxicity of chemicals, (v) established methods for rearing animals because of the long history of the silk industry and (vi) the low motility and lack of capacity to fly even after becoming a moth. The last point is very important when considering the biohazard risks of infection experiments. The novelty of our silkworm infection model when compared to the previously reported C. neoformans infection model using G. mellonella (Mylonakis et al. 2005) is as follows: (i) it is a quantitative model, where we can evaluate fungal virulence and the therapeutic effects of the drugs through LD50 and ED50 values. Quantitative evaluation has never been performed in the previous insect models. (ii) The virulence of C. neoformans and therapeutic effects differ between different administration routes such as intra-haemolymph and intra-midgut. Different administration routes have not been discussed in previous insect models. Based on these advantages, we propose that the silkworm is a useful model animal for evaluating the therapeutic effects of drugs against infection with C. neoformans, a pathogenic fungus that is causing serious clinical problems.
We thank Keiko Kataoka and Yoshihiro Nakatani for their technical assistance. Fungal strains used in this study were kindly provided by Kirsten Nielsen and Joseph Heitman. This work was supported in part by a Grant-in-Aid for Scientific Research (21790062) and also by the Program for Promotion of Fundamental Studies in Health Sciences of the National Institute of Biomedical Innovation (NIBIO); a grant from the Ministry of Health, Labour and Welfare (Research on Biological Resources and Animal Models for Drug Development); and Genome Pharmaceuticals Institute Co., Ltd (Tokyo, Japan). This study was partly supported by Cooperative Research Program of Medical Mycology Research Center, Chiba University (10-25 and 11-1).