Development of an Orogastrointestinal Mucosal Model of Candidiasis with Dissemination to Visceral Organs

Abstract

Studies were done to develop a murine model that mimics the pattern of mucosal candidiasis followed by disseminated disease seen in patients given cytotoxic chemotherapy. Developmental studies showed that suppression of mice with 5-fluorouracil beginning 3 days prior to infection and given every 7 days thereafter necessitated antibacterial treatment but resulted in a reproducible model. Candida albicans given in the drinking water resulted in oral infection by day 3 that significantly increased from days 10 to 15 and mucosal infection with 4 to 7 log₁₀ Candida CFU in the esophagus, stomach, small intestine, and cecum. Dissemination to livers occurred and was 100% on days 5 to 15; fewer animals had kidney infection. The median kidney or liver CFU were 2 or 3 log₁₀ CFU, respectively, on day 15; despite this, mortality was low through 21 days of infection. As a demonstration of the utility of the model to test antifungal activity, daily treatment with 10 or 50 mg/kg itraconazole significantly reduced dissemination to the liver and kidneys and reduced tongue CFU compared to controls. Overall, these studies indicate that a nonlethal model of oral and gastrointestinal mucosal candidiasis with dissemination can be established in mice. Drug efficacy in treating localized infection and in preventing or treating disseminated infection can be studied.

Candidiasis of the oral mucosal surfaces and the intestinal tract is problematic for a variety of patient populations. Those at the highest risk include those with AIDS and those on immunosuppressive therapy (e.g., cancer chemotherapy or high-dose steroids). We previously reported a model of orogastrointestinal mucosal candidiasis that closely mimics the clinical manifestations observed in AIDS patients and demonstrated its utility for the study of therapeutics (5, 7). Although quite useful, this model does not result in the translocation of Candida albicans across the intestinal mucosa to cause disseminated disease (5, 7).

Disseminated candidiasis in cancer chemotherapy patients is thought to arise from the translocation of C. albicans across gut mucosa damaged from chemotherapy treatment. A murine model of gastrointestinal candidiasis mimicking this situation and resulting in systemic dissemination and death has been previously reported (35). However, our attempts to replicate this model resulted in minimal dissemination and little to no lethality due to Candida albicans. In addition, others have previously reported inconsistent or no dissemination from gut tissues in similar models (3, 9, 10, 34, 45, 46). The goal of the present studies was to further develop and standardize a model of disseminated disease arising from translocation from gut colonization, induce oral mucosal disease, and determine the utility of this model for the study of therapeutic intervention.

(These data were presented previously [K. V. Clemons, G. M. Gonzalez, G. Singh, J. Imai, M. Espiritu, R. Parmar, and D. A. Stevens, Abstr. 7th Am. Soc. Microbiol. Conf. Candida Candidiasis, abstr. 132C, 2004].)

MATERIALS AND METHODS

Mice.

Female CD-1 mice (Charles River Laboratories, Portage, Mich.) were used in all studies. Animals were 7 weeks of age in experiment 1 and 5 weeks of age in other experiments. Mice were caged in groups of five using sterilized bedding and were provided sterilized food and acidified water with or without antibiotics ad libitum.

Inoculum preparation.

Candida albicans strain #5, a strain that is well characterized with respect to its in vivo properties, was used in all studies (4-7, 12, 23, 40, 41). Inocula were prepared and infection of mice was done as previously described (5, 7). In brief, the organisms were grown in sterile bottles, each containing 100 ml of SAAMF (15) broth and incubated for 48 h at 35°C on a gyratory shaker. C. albicans was harvested by centrifugation, washed once with saline, and then suspended in saline. The cells were counted using a hemacytometer, and dilutions were made in sterile water. The final inoculum was >10⁸ cells/ml. Viability was determined by plating.

Infection of mice.

The water bottles were removed 8 h prior to replacement with the inoculum suspension of C. albicans (5, 7). The mice were allowed to drink from this suspension for 24 h, at which time the inoculum suspension was removed and replaced with drinking water with or without antibiotics (day 0). Antibiotic doses were calculated on the assumption that mice drink 5 ml of water per day.

Immunosuppressants.

5-Fluorouracil (5-FU) injection (ICN Pharmaceutical, Costa Mesa, CA) given intravenously, cyclophosphamide (Cytoxan; Mead Johnson, Princeton, NJ) given intraperitoneally, and triamcinolone acetonide (Kenalog; Bristol-Myers Squibb Co., Princeton, NJ) given subcutaneously were used in these studies. Dosages and regimens varied by experimental design.

Therapy studies.

In two separate studies, the effectiveness of antifungal therapy was examined. Mice received 10 or 50 mg/kg of itraconazole (ICZ) with cyclodextrin or cyclodextrin alone by gavage once daily prepared as previously described (16, 17). Calcium alginate swabs were wetted in a suspension of 1% clotrimazole in polyethylene glycol 400 (swab volumes are estimated to be 20 to 50 μl), which was applied by rolling the swab onto the oral surfaces twice daily in a third experiment. All treatments were given for 10 days beginning on day 4 postinfection.

Parameters of evaluation.

Survival rates and CFU were used to evaluate establishment of infection. CFU were determined by quantitative plating of tissue homogenates, which included defined lengths of esophagus, stomach, small intestine, cecum (5, 7), and liver and kidneys (6, 12, 23, 38). CFU present in the oral cavity were determined by swabbing the oral surfaces of euthanized mice with a calcium alginate swab, placement of the swab in phosphate-buffered saline to dissolve it and release the organism, and serial plating of dilutions. The lower limit of detection of CFU for tissue samples is approximately 10 CFU per sample (entire organ or sample of tissue) and approximately 5 CFU per sample swabbed from the oral cavity.

Statistical analyses.

Survival differences were determined by log-rank test, and differences in comparative CFU were determined by Mann-Whitney U test.

RESULTS

Optimization of immunosuppressive regimen.

In three preliminary experiments, the relationship of 5-FU dose and regimen to gut and disseminated disease was studied (summarized in Table 1).

TABLE 1.

Initial 5-FU dose finding for the establishment of gastrointestinal infection with Candida albicans

5-FU regimen (mg/kg), day(s)a	No. of mice with Candida infection (total no. of mice)
	Gutb	Disseminated sitesc
200, −3	0, 5, 3, 4 (6)	1, 3 (6)
300; −3	5, 6, 6, 6 (6)	6, 6 (6)
200; −3, +10	6, 10, 9, 10 (10)	6, 9 (10)
200; −3, +6	6, 6, 6, 6 (6)	6, 6 (6)

^aMilligrams per kilogram of 5-FU; day of challenge with C. albicans was day 0.

^bNumber of mice with infection at day 22 postinfection in esophagus, stomach, small intestine, and cecum, respectively. Numbers in parentheses are the total numbers of mice studied.

^cNumber of mice with infection at day 22 postinfection in the kidneys and liver, respectively.

These informative studies provided a basis to extrapolate a fourth, extensive study. Mice received no 5-FU or were given an intravenous injection of 200 mg of 5-FU/kg body weight 3 days prior to infection (day −3), and subsequent doses were given every 5, 7, or 13 days. Deaths were tallied through 20 days of infection (Fig. 1). The majority of deaths occurred in mice given the more frequent dosing of 5-FU (Fig. 1). Necropsies of animals that died or were euthanized throughout the study and quantitative plating of kidney homogenates indicated substantial bacterial infection in those given 5-FU every 5 or 7 days. For those that did show some candidal infection, the log₁₀ CFU were lower (i.e., 2.58 and 3.49 log₁₀ CFU) than what would be expected for death due to infection, based on previous studies of systemic candidal infection (6, 12, 23, 38). No animals given no 5-FU died. Recovery of Candida from the various tissues on day 20 is shown in Fig. 2. Although all surviving mice were infected with Candida in one or more tissues, including those not given 5-FU, dissemination from the gastrointestinal tract occurred in few animals (Fig. 2).

FIG. 1.

Optimization of immunosuppression for the establishment of gastrointestinal infection with C. albicans. A Kaplan-Meier plot of the cumulative mortality of CD-1 mice infected orally with C. albicans and given no treatment (n = 11) or groups of 10 mice treated with 5-FU every 5, 7, or 13 days (q5d, etc.) beginning on day −3 is shown. 5-FU treatments (200 mg/kg) were administered intravenously in a lateral vein of the tail.

FIG. 2.

Optimization of immunosuppression for the establishment of gastrointestinal infection with C. albicans. Shown are scattergrams of the CFU of C. albicans recovered at day 20 after infection from the esophagi, stomachs, small intestines, ceca, kidneys, and livers of surviving mice infected orally with C. albicans and treated as described in the legend for Fig. 1. Each data point represents the log₁₀ CFU per entire organ (6, 12, 23, 38) or fixed organ length (5, 7). A value of 0 indicates that the fungal burden was below the detectable levels of the assay. Animals that died before the conclusion of the study were omitted from the analysis. The bar represents the median group value.

Adjunct antibacterials and effect of other immunosuppressive regimens.

The results of the prior experiment suggested that a secondary bacterial infection was responsible for the deaths. To reduce bacterial proliferation in the gut, with the objective of reducing secondary infections, an empirically selected antibiotic regimen was added. All mice received gentamicin at 0.2 mg/ml, clindamycin at 1 mg/ml, and vancomycin at 1 mg/ml added to sterile drinking water, starting on day −3, fresh daily. Imipenem/cilastin was given daily at a dose of 5 mg/mouse subcutaneously (3 days prior to challenge) and then intraperitoneally thereafter. Four groups of 10 mice were tested. One group received no 5-FU. Three groups were given 5-FU at 200 mg/kg intravenously 2 days prior to infection. Subsequent 5-FU dosing was either 200 mg/kg every 7 days, 150 mg/kg every 7 days plus cyclophosphamide at 200 mg/kg every 5 days, or 5-FU at 200 mg/kg every 7 days plus 1 mg of triamcinolone every 7 days.

Mortality was monitored through 20 days of infection, and the results are presented in Fig. 3. Most deaths occurred in the groups given 5-FU in combination with cyclophosphamide or triamcinolone. Necropsies of animals that died during the study were done; kidneys were removed to monitor dissemination, and quantitative cultures were plated onto blood agar to determine the number of viable CFU in the organ. The cultures indicated Candida albicans dissemination in all mice but few bacteria (<2 to 3 log₁₀ CFU per organ) and no bacteria in the group given 5-FU at 200 mg/kg every 7 days. The kidney burdens of C. albicans recovered were low; i.e., in mice given 5-FU plus cyclophosphamide, the range of log₁₀ CFU was 1.20 to 4.70 log₁₀ CFU, and in mice given 5-FU plus triamcinolone, the range of log₁₀ CFU was 2.30 to 5.08 log₁₀ CFU. Thus, death may have been due to the toxicity of the immunosuppressant regimen rather than C. albicans, where we expect, from prior studies (6, 12, 23, 38), that there should be 7 log₁₀ CFU or more in animals dying from Candida infection.

FIG. 3.

Use of adjunct antibacterial regimens and other immunosuppressive regimens. Shown is a Kaplan-Meier plot of the cumulative mortality of CD-1 mice infected orally with C. albicans and given antibiotics. Mice received no immunosuppressant, 5-FU, 5-FU plus cyclophosphamide (CTX), or 5-FU plus steroid (triamcinolone). There were 10 mice in each group.

Recovery of organisms at day 20 (Fig. 4, no-immunosuppressant group) suggested that antibiotics alone resulted in the dissemination of C. albicans, presumably by initially enhancing Candida gut proliferation. Antibiotics in combination with 5-FU induced colonization of the gastrointestinal mucosa and dissemination of C. albicans to internal organs (kidneys and livers) to a significantly greater degree than did antibiotics alone (esophagus and stomach, P < 0.001; small intestine, P = 0.004; cecum, P = 0.035; kidneys, P = 0.0003; liver, P = 0.0007). In these two groups, there was minimal lethality by day 20 postinfection. These groups show that regardless of possible concerns that some deaths could have been due to bacterial dissemination in previous experiments, with the appropriate immunosuppressive regimen plus antibiotics, a satisfactory model of disseminated candidiasis from the gut could be established.

FIG. 4.

The use of adjunct antibacterial regimens and other immunosuppressive regimens. Shown are scattergrams of the CFU of C. albicans recovered from different organs at day 20 postinfection. Each data point represents the log10 CFU per entire organ (6, 12, 23, 38) or fixed organ length (5, 7). The bars represent the median group values. All mice received the antibiotic regimen. The groups are same as those described in the legend for Fig. 3. NI, no immunosuppressant.

Oral infection and dissemination from the gut.

The results of the previous experiment indicated that a 5-FU regimen given every 7 days would be useful. Furthermore, the inclusion of the antibiotic regimen to prevent secondary bacterial infections was determined to be valuable. The aim of this experiment was to examine the occurrence of colonization of the oral mucosa and to determine the rate of dissemination from the intestinal tract to the kidneys and liver.

Infection and dissemination rates were examined in one part of the study. These mice were given 200 mg/kg 5-FU every 7 days, beginning on day −2 as before; the same antibiotic regimen was used. Cohorts of five predesignated mice were sampled on days 1, 3, 5, 7, and 10 postinfection. The mice were killed, the tongue was swabbed with a sterile calcium alginate swab, and kidneys and livers were removed.

Figure 5 presents the results of the CFU determination for the oral mucosa, kidneys, and livers of mice at various times postinfection. CFU were recovered from the oral mucosa of all five mice by day 3 of infection and from three of five mice on day 1 postinfection. The level of infection remained constant from day 3 through day 10 and significantly increased from day 10 to day 15 by about 10-fold (Fig. 5). Small (ca. 1 mm or less) yellowish-white plaques could be observed on the mucosal surfaces, particularly on the ventral surface of the tongue, as early as day 5 postinfection.

FIG. 5.

Dissemination of C. albicans from the gut after oral infection with C. albicans. Recovery of C. albicans from the tongue, kidney, and liver of infected mice is shown. Mice were treated with 5-FU at 200 mg/kg every 7 days. Samples were taken at the various days postinfection shown. Untreated refers to no antifungal therapy.

The rates of dissemination from the intestinal tract of mice proved most consistent for the liver. Only three of five mice had CFU in the liver on day 1 or 3 postinfection. All mice had CFU recovered from the liver on days 5 to 15, with the number of CFU at day 15 significantly higher than that at days 1 to 7. Infection did not appear to be progressive in the liver, since counts were not different on days 7 to 15 (Fig. 5).

In the kidneys, fewer animals had recoverable CFU. Only two mice had CFU on day 3 postinfection, and none had CFU in the kidneys on day 1 or 5. At later times, some mice had no recoverable CFU: 1 of 5 mice on day 7, 3 of 5 mice on day 10, and 4 of 10 mice on day 15. Thus, the rate of dissemination for the kidneys appeared to be at most about 60%, whereas it was 100% for the liver (Fig. 5).

Therapeutic studies.

Three different therapeutic studies were done to examine the utility of the model as a tool for studying antifungal therapy for the clearance of oral colonization and gastrointestinal colonization and the prevention of dissemination from the intestinal tract to visceral organs. ICZ was selected as a representative of drugs used in clinical medicine for gut prophylaxis.

The initial antifungal therapy study was done to determine drug efficacy against oral colonization and prevention of dissemination. Mice were immunosuppressed, treated with antibiotics, and infected as described above for the prior experiment. Four groups of mice (n = 10 mice per group) were given no antifungal treatment, ICZ at 10 or 50 mg/kg, or cyclodextrin beginning on day 4 for 10 days. On day 15, all mice were euthanized with CO₂, the kidneys and liver were removed for CFU determination, and the tongue of each mouse was swabbed.

No mice in any group died during the study. Treatment of mice with 10 or 50 mg/kg of ICZ orally showed that ICZ had efficacy. Both regimens significantly lowered tongue CFU on day 15 (P < 0.02 or 0.0002), with 6 of 10 mice given 50 mg/kg of ICZ having no detectable CFU compared to 4 of 10 mice given 10 mg/kg of ICZ and 0 of 10 mice given no treatment or cyclodextrin diluent (Fig. 6).

FIG. 6.

Effect of ICZ therapy on the recovery of C. albicans from the tongue, kidney, and liver of infected mice in samples taken at day 15 postinfection. Regimens of 5-FU were the same as those described in the legend for Fig. 5. Untreat, untreated; cyclodex, cyclodextrin.

ICZ treatment also reduced dissemination to the liver and kidneys compared to the controls (Fig. 6). Eight of 10 mice given ICZ at 10 mg/kg had no detectable infection in the kidneys. In the livers, ICZ treatment was also efficacious compared with controls (P < 0.0004), and the 50-mg/kg dose cleared 5 of 10 mice of infection. There was no significant difference in efficacy between the two dosages of ICZ in either organ. Only 14% and 50% of the mice had bacterial colonies present in the kidneys and livers, respectively, at day 15, but only 11% of the organs had >3 log₁₀ bacterial CFU.

A second antifungal therapeutic study was done to examine the reproducibility of the model in therapy studies. The experimental conditions and dose groups were the same as those of the preceding experiment. During the course of this experiment, 5 of 10 mice given cyclodextrin died between days 12 and 15; no mice in the other groups died. Deaths were attributed to Candida infection, as bacterial CFU in the kidneys and livers were <3 log₁₀ CFU. Overall, the number of CFU of Candida recovered from the tongue, kidneys, and livers of the surviving animals was higher than that of the preceding experiment by 10- to 100-fold and was indicative that deaths could be due to fungal infection (Fig. 7). With respect to the efficacy of ICZ, it again reduced colonization and dissemination.

FIG. 7.

Second therapeutic study of the effect of ICZ. Shown is the recovery of C. albicans from the tongue and tissues of infected mice. The 5-FU regimens were the same as those described in the legend for Fig. 5. Samples were taken at day 15 postinfection. The bars represent the medians of the groups (not shown for the cyclodextrin group, where half of the animals died).

For the tongue, ICZ at 10 or 50 mg/kg was effective versus the controls (P = 0.0001), but the results with these doses were not different (Fig. 7). One animal from each drug regimen had no recoverable CFU from the tongue. Although fewer animals were free of tongue colonization in the second therapeutic experiment than in the initial therapeutic experiment, these results are similar and indicate the efficacy of ICZ, albeit non-dose responsive in this dose range. Similarly, ICZ (both dose regimens) reduced dissemination from the gut tissues to the kidneys (P = 0.001) and liver (P ≤ 0.0005) compared to the untreated controls; this was also the case for comparisons with the cyclodextrin control group, if death is assigned a worse outcome than survival with any residual burden (20, 39), with P values of ≤0.003 and ≤0.0004, respectively (the exception was ICZ at 50 mg/kg versus untreated controls, in the kidney). However, in contrast to the first antifungal therapeutic experiment, no animals were free of infection in either organ (Fig. 7).

The CFU remaining in the gastrointestinal tissues showed that both doses of ICZ significantly lowered the CFU of Candida by 10- to 100-fold in all four tissues compared to that found with no treatment or cyclodextrin (P ≤ 0.003 for all comparisons). The cecum, and then the stomach, appears to be the most difficult of the tissues to reduce colonization. No animals were free of Candida in any of the four gastrointestinal tissues (Fig. 7).

In a third antifungal therapeutic experiment that was used to examine the utility of the model for studies of therapeutic agents against oral mucosal disease, mice were treated with a topical agent, a suspension of 1% clotrimazole in polyethylene glycol 400, twice daily. The clotrimazole suspension significantly reduced the median CFU recovered from the tongue by 30-fold in comparison with the untreated control group (P = 0.014). It was also noted that the number of plaques appeared to be fewer, as well as smaller, in treated animals than in controls.

DISCUSSION

The treatment of patients with high-dose immunosuppressive regimens, broad-spectrum antibacterial antibiotics, or cancer chemotherapy contributes to the prevalence of candidiasis (1, 2, 8, 21, 24, 25, 28, 32). Treatments to clear colonized oral mucosa or the gastrointestinal tract can be difficult, and these patients may progress to develop disseminated disease, likely of gastrointestinal origin. Thus, the availability of a model that mimics these manifestations becomes valuable for the study of therapeutics as well as pathogenesis. The results of our studies indicate that a useful and reproducible model of oral mucosal candidiasis and intestinal colonization with subsequent dissemination to the liver and kidneys can be established in immunosuppressed adult mice.

Here, we have described nine experiments. During the course of our studies, we determined that careful control of various parameters was critical. The determination of an appropriate immunosuppressive regimen was necessary to cause reproducible dissemination rates, with CFU counts from the kidneys and liver showing some progression of disease with time. Furthermore, the immunosuppressive regimen could not be so severe as to be lethally toxic, as reflected by the results using 5-FU plus cyclophosphamide or triamcinolone, nor could it be too innocuous, as in giving 5-FU every 13 days. We found the inclusion of a broad-spectrum antibiotic regimen necessary to reduce the potential misinterpretation of deaths as being due to C. albicans, as demonstrated by early experiments, where deaths were not apparently due to C. albicans but rather were due to a secondary bacterial infection. The effects of various immunosuppressive regimens, antibiotics, and other factors on gut colonization and in some instances dissemination and course of disease have been previously reported by other investigators (10, 14, 19, 22, 26, 30, 31, 33, 34, 36, 37, 47). Although candidal infection of other critical organs not studied, such as brain or heart, could be an alternative explanation to bacterial infection or immunosuppressant toxicity as a cause of death in our studies, progressive infections in other organs are not seen after intravenous challenge with Candida (6, 12, 23, 38) or suggested by gross pathological examinations.

To determine the usefulness of our model in the study of drug efficacy in preventing or curing disseminated infection, we performed efficacy studies using ICZ. Our results demonstrated that ICZ showed efficacy in reducing gut tissue burdens of C. albicans as well as systemic efficacy in the prevention or reduction of burden in the kidneys and lungs. This gut efficacy is similar to those reported previously by others using other configurations of rodent models of gastrointestinal candidiasis, where azoles and triazoles showed efficacy (5, 7, 11, 13, 14, 27, 42-44). The parameters that we chose to establish the model and the design of our experiments were essential to the success of the efficacy studies. The careful evaluation and control of the occurrence of secondary bacterial infections using the broad-spectrum antibiotic regimen along with a nonlethally toxic immunosuppressive regimen and the ending of the experiment on day 15 prior to deaths were critical. By performing our studies in this way, one can be assured, in contrast to models whose endpoint is mortality (35), that the antifungal being assayed is affecting the fungal infection.

Our model is also useful for studies of drug efficacy against oral infection. We demonstrated consistent infection and recovery of yeasts from the oral mucosal surfaces of infected mice. Either oral ICZ therapy or topical clotrimazole therapy significantly reduced the burden of C. albicans in the oral mucosa as well as the numbers and sizes of grossly observable plaques. Others have previously reported models of orogastric disease in rodents and used those models for studies of pathogenicity and drug efficacy (18, 26, 29).

Overall, our studies have allowed us to develop an adult mouse model of orogastrointestinal candidiasis that affects the oral mucosa and all tissues of the gastrointestinal tract. The model proved to be consistent and reasonably reproducible in recovery of burdens and tissue infections, in contrast to our experience with previously reported models (35). These qualities allowed for drug efficacy to be examined. Differences in fungal burdens between experiments possibly reflect the method of initiation of the infection, as the inoculum that each animal receives is not precisely controlled and is dependent upon how much the animal drinks; these variations emphasize the importance of controls in each study. The reproducibility of severity of infection would likely be tighter from experiment to experiment if a gavage of the inoculum were given; however, this would lose an advantage we strived for, namely, the simplicity and ease of repeated experiments. The worse outcome associated with cyclodextrin in one experiment remains unexplained.

In addition, the model results in consistent dissemination to the kidneys and livers after 15 days, which occurs as early as 3 days, particularly in the liver. Thus, the model also appears to be useful for studies of dissemination as well as for studies of pathogenesis and drug therapy or immunomodulation. Lastly, the use of a different regimen of immunosuppression prior to infection may prove useful in altering this model to one that can lead to mortality due to the dissemination of C. albicans from the intestinal tract to visceral organs. The studies detailed in Fig. 3 suggested that this goal can be achieved. Additional preliminary studies (data not shown) with cyclophosphamide and triamcinolone suppression also suggest this possibility. Further studies are under way to refine such a model.