INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Infections caused by Cryptosporidium parvum lead to chronic diarrhea and wasting in people with AIDS, those with malignancies, and malnourished children (4, 10). Therapeutic agents currently available for treatment of such infections include paromomycin (20) and nitazoxanide (5, 6), but they are only partially effective. Many chemotherapeutic and immunotherapeutic agents have been evaluated in cell culture and/or various animal models (1, 12, 19, 21). The most commonly used rodent models include the dexamethasone-treated rat or mouse (2, 13, 22) and the congenitally immunodeficient SCID mouse (11, 14, 17). Although these animals can be infected with C. parvum, the sites of infection are generally limited to the pylorus, segments in the small intestine, the cecum, and the colon. These sites of infection do not closely resemble the pattern found in immunodeficient humans, in whom the entire small intestine can be colonized. Infected SCID mice normally remain healthy for several weeks and then develop chronic infections involving the hepatobiliary (HB) tract. The infectious dose required to induce consistent infections in the rodent models is 10⁶ to 10⁷ oocysts, which is manifold higher than the ~100 oocysts needed to infect human volunteers (7). In addition, these animals do not develop symptoms that are directly related to their luminal infection, as do humans.

Mice bearing a targeted disruption of the gamma interferon (IFN-) gene (IFN- knockout [GKO] mice) are remarkably susceptible to infection with C. parvum (15). Moreover, the infection profile in GKO mice closely mimics that seen in humans and animals with severe clinical cryptosporidiosis. Because GKO mice are far more susceptible to cryptosporidiosis than any other rodent (15), we designed the present study to examine the suitability of the GKO mouse as a potential model for the evaluation of therapeutic drugs.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Three sets of GKO mouse experiments were conducted. The first set identified parameters useful in establishing the model for drug evaluation. It describes the course and clinical outcome of small-inoculum infection with C. parvum, including the parasite distribution in the gut and the extent of associated mucosal lesions at two time points. Our prior studies did not quantify the extent of the disease (15). The second set of experiments was designed to explore the minimum parasite dose required to consistently induce infection in all animals. In the third set of experiments, paromomycin was used to validate the model for drug evaluation. Paromomycin, though noncurative, is a consistently partially effective agent in both humans and animal models. Its use also allowed a comparison of the GKO mouse model with other currently used animal models. The oocysts used in these experiments were from the GCH1 isolate, which we have maintained via serial passage in calves for more than 6 years. It has been described in detail previously (17) and is infectious to mice, calves, and humans.

Infection of GKO mice with C. parvum and its pathogenesis. Twenty-one 8-week-old male GKO mice (C57BL/6 background), purchased from Jackson Laboratories, were housed in microisolators, and each was challenged with 5,000 C. parvum oocysts (3, 19). Seven male non-GKO C57BL/6 mice were also challenged as a control group. All mice were monitored for oocyst shedding three times per week by counting acid-fast-staining oocysts in 2 µl of feces. Body mass was measured weekly, and overall clinical appearance was assessed daily. Seven mice were euthanized 15 days after challenge for histologic analysis. All surviving mice were euthanized 32 days after challenge. Mucosal scores were determined at eight gastrointestinal sites: the stomach, the duodenum, three equally spaced mid-small intestinal sites, the terminal ileum, the cecum, and the colon. The scores of the eight gut sections were combined to calculate the mucosal score for each mouse, reflecting the extent of C. parvum infection in the luminal portion of the intestine. Each site was scored as follows: 0, no infection; 1, very-difficult-to-find parasite forms; 2, sparse but easy-to-find parasite forms; 3, abundantly present but focally distributed parasite forms; 4, extensive presence of parasite forms, covering most mucosal surfaces; or 5, extensive presence of parasite forms, covering the entire mucosal surfaces. The highest possible score was 40 (8 × 5). The HB tracts of all mice were also examined histologically.

Oocyst dose-response. Thirty-five 6-week-old male GKO mice were randomized into five groups of seven mice each and maintained inside microisolators. In one experiment, members of groups 1 to 4 were challenged orally with 5,000, 1,000, 500, and 100 oocysts, respectively. Group 5 served as an uninfected control group. A second, confirmatory experiment was performed, with members of groups 1 to 4 receiving 5,000, 1,000, 100, and 10 oocysts, respectively. Infected mice were monitored as described above.

Dose-response of infected mice to treatment with paromomycin. Twenty-six 4-week-old male GKO mice were randomized and challenged with 5,000 oocysts at 7 weeks of age. Four days later, these mice began to excrete oocysts. Beginning on day 5 of infection, members of groups 1 (six mice), 2 (six mice), and 3 (seven mice) were respectively treated with 2,000, 1,000 and 500 mg of paromomycin/kg of body weight/mouse/day, given in two divided doses for 10 days. A fourth infected group of seven mice received a placebo (phosphate-buffered saline). Mice were euthanized 15 days after challenge, and mucosal scores were determined. This experiment was repeated with 100 oocysts as the challenge dose.

Statistical analysis. Data were analyzed with the Statistical Package for Social Sciences (Software Products SPSS Inc., Chicago, Ill.). Analysis of variance was conducted for all comparisons of multiple groups. If groups were significantly different by analysis of variance at the P < 0.05 level, then subgroup analysis was done. When parametric tests were inappropriate, standard nonparametric tests (detailed in Results) were used. Differences for which the two-tailed P value was 0.05 are reported as significant. We also report regression analyses using r² values, which are more conservative and robust than r values. To normalize oocyst shedding data, the numbers of oocysts detected were log transformed. To avoid taking the log of zero when no oocysts were detected, we used the common statistical maneuver of adding 0.5 to all oocyst shedding scores before transformation.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Clinical symptoms and mucosal scores. All mice challenged with 5,000 oocysts excreted oocysts within 4 days. The number of oocysts peaked on or about day 10 and declined until day 32 postchallenge (Fig. 1). Of the 21 infected mice, 4 died on day 12, 1 died on day 14, and 1 expired on day 16 after challenge. The body weights of infected mice declined from day 7 after challenge until the end of the experiment. On day 23, the C57 mice were significantly heavier than the GKO mice (means ± standard deviations, 28.90 ± 1.06 and 22.74 ± 0.68 g, respectively; P < 0.001), as they were on day 33 (31.75 ± 0.35 and 21.78 ± 0.52 g, respectively; P < 0.001). By day 10, the mice were hunched, emaciated, and reluctant to move, huddled in one corner of their cage, had scruffy and stained coats, and showed markedly decreased food consumption. They clinically deteriorated until the end of the experiment.

View larger version (19K): [in this window] [in a new window]   FIG. 1.   GKO mice (C57 background) () and control C57 mice () were each infected with 5,000 oocysts 28 days after observation was begun. At all times after infection, the log-transformed oocyst shedding score was significantly greater in GKO mice than in control C57 mice (P < 0.001 at all time points). The overall oocyst shedding score (log mean ± standard deviation) for the GKO group was 1.38 ± 0.08, while that for the C57 control group was 0.26 ± 0.02 (P < 0.001). CI, confidence interval.

Oocyst dose-response. There were no differences in the shedding of oocytes by mice challenged with either 5,000 or 1,000 oocysts; the values had peaked on day 9 or 10 after challenge in both experiments. Oocyst shedding peaked 5 days later (on day 14 after challenge) and 10 days later (on day 19 after challenge) in mice inoculated with 100 oocysts and 10 oocysts, respectively. Regardless of the inoculum size, all mice eventually reached the same level of shedding (Fig. 2). The timing of death was related to the inoculum size, with deaths in the small-inoculum groups being delayed in a dose-dependent fashion compared to those in the higher-inoculum groups. By day 19, three mice from each group of seven receiving 5,000, 1,000, or 100 oocysts had either died or were euthanized to prevent further suffering. The original inoculum was highly predictive of the mean weight, in grams, of the surviving mice (r² = 0.350, P = 0.001). Members of the group receiving 10 oocysts began to show symptoms of wasting on day 20. Both experiments showed similar patterns of shedding and weight loss.

View larger version (23K): [in this window] [in a new window]   FIG. 2.   Groups of seven GKO mice were infected with 0 (), 10 (), 100 (), 1,000 (), or 5,000 () oocysts. The time to peak patency was related to the inoculum, and before mice began to die of the infection, oocyst excretion was temporally related to the initial inoculum. Each order-of-magnitude fall in the inoculum led to a 2- to 3-day delay to the time to peak patency. In the displayed as well as a confirmatory experiment, only 6 of 14 mice infected with 5,000 oocysts were alive on day 14, whereas 11 of 14 mice infected with 100 oocysts were alive on day 14 (2 = 3.743, P = 0.053). All mice given 10 oocysts became infected. On day 12, mouse no. 3 and 4 were positive; on day 14, mouse no. 2 and 4 (but not no. 3) were positive; on day 16, mouse no. 1, 2, 5, 6, and 7 (but not no. 3 or 4) were positive; and on day 19, all mice were positive. Thus, when a 10-oocyst inoculum is used, shedding may initially be at the limits of detection, yet it still rises to the same level as in mice infected with a larger inoculum. CI, confidence interval.

Responses of infected mice to different doses of paromomycin. Mice challenged with 5,000 oocysts excreted high levels of oocysts by 5 days after challenge. Mice in the paromomycin-treated groups were protected from the clinical depression, weight loss, and mortality suffered by the untreated placebo group. All mice in the three paromomycin-treated groups, except for one mouse that died in a laboratory accident, survived the 15-day experiment. Mice treated with the lowest dose of paromomycin (500 mg/kg/day) still shed considerable numbers of oocysts (Fig. 3); curiously, however, they (as well as the other paromomycin-treated mice) showed no apparent clinical symptoms of infection. The mean weight ± the standard deviation of the mice treated with any dose of paromomycin fell from 18.52 ± 0.27 g on day 5 to 17.27 ± 0.32 g on day 15 (1.25 g), whereas the weight of the untreated mice fell from 18.67 ± 0.44 g on day 5 to 14.80 ± 0.38 g on day 15 (3.87 g), an over threefold difference. This difference in weight was very highly significant on each of the days after therapy was begun (P = 0.002, 0.001, and 0.002 on days 9, 12, and 15, respectively). In the experiment in which 5,000 oocysts were used as the challenge dose, two of seven placebo-treated mice died 9 days after infection (4 days into treatment) and a third mouse died on day 12. These three mice were the smallest within the group, suggesting that a higher body mass was protective against early death.

View larger version (22K): [in this window] [in a new window]   FIG. 3.   GKO mice were infected with 5,000 oocysts, and treatment with 0 (), 500 (), 1,000 (), or 2,000 () mg of paromomycin/kg/day began on day 5 of infection. There was a highly significant inverse relationship between fecal shedding of oocysts and the treatment dose administered to the mice (P < 0.0001). On day 9, 2,000 mg of paromomycin/kg/day decreased the fecal shedding of oocysts by ~2.5 log units. Oocyst shedding was very highly significantly decreased by paromomycin in a dose-dependent fashion (r2 = 0.5142, 0.7215, 0.7752, and 0.6703 on days 7, 9, 12, and 15, respectively; P < 0.0001). In multiple regression analysis, the dose of paromomycin was very highly significant (P < 0.0001) whereas the day of the experiment was not significant (P = 0.2060). CI, confidence interval.

View larger version (23K): [in this window] [in a new window]   FIG. 4.   The effect of paromomycin treatment  (0 [], 500 [], 1,000 [], or 2,000 [] mg/kg/day) on oocyst shedding was assessed in mice that had been infected with 100 oocysts. Treatment was begun on day 8 of infection. By day 11 and thereafter, there was a dose-related decrease in oocyst shedding in all groups receiving paromycin treatment. This relationship was highly significant on all days  (F = 235.35, P < 0.001, regression analysis). On day 18, there were no mice in the untreated group because all  (7 of 7) had died, whereas in the treated groups only 1 of 21 had died  (by Fisher's exact test, P < 0.001 for differences in survival). CI, confidence interval.

View larger version (25K): [in this window] [in a new window]   FIG. 5.   The weights of uninfected GKO mice  () and of GKO mice infected with 100 oocysts on day 0 and treated with paromomycin  (0 [], 500 [], 1,000 [], or 2,000 [] mg/kg/day) on day 8 and thereafter are displayed. Four days after infection, there were no differences in weight among the treatment groups. In contrast, by 11 days after inoculation and thereafter, there was a highly significant difference in the weights of the uninfected control mice and the untreated  (placebo) control group  (means ± standard deviations, 21.31 ± 0.38 g and 14.25 ± 0.57 g, respectively; P < 0.001). The infected, untreated mice weighed only 67% as much as the uninfected mice. Paromomycin treatment very significantly blunted this decrease in weight in a dose-dependent fashion  (F = 25.881, P < 0.001). CI, confidence interval.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

GKO mice are profoundly susceptible to infection, weight loss, and death after infection with 10⁷ GCH1 C. parvum oocysts (15). We have found that an infectious dose of as few as 10 oocysts establishes a lethal infection in these mice. These results indicate that of the animal models studied to date, the GKO mouse is by far the most susceptible to cryptosporidiosis. Infection reproducibly leads to high levels of luminal infection, with the symptoms and signs of infection being temporally related to the size of the infectious inoculum. We used the drug paromomycin to optimize this model and found that it led to consistent decreases in the infectious burden and prevented death. These characteristics strongly suggest that this model will be useful in the evaluation of anti-Cryptosporidium drugs.

In the search for effective therapy against persistent C. parvum infection, the use of a suitable immunodeficient model is highly desirable. The animal model should  (i) allow a rapid and consistent establishment of a persistent infection in the adult, (ii) require a low infectious dose, and (iii) exhibit clinical symptoms such as a profound watery diarrhea, dehydration, malabsorption, weight loss, and emaciation, as well as death. The parasite should ideally colonize a major portion of the upper gastrointestinal tract and have the ability to spread to the HB tract. Of the rodent models that are being used to screen drugs for effectiveness against cryptosporidiosis, none meets all of the above criteria. These include the normal neonatal mouse (9, 16), the neonatal or adult SCID mouse (11, 14, 17), the dexamethasone-suppressed adult mouse (22), the immunosuppressed rat (2, 13), and the anti-IFN--conditioned weaned SCID (18) models. Each model has its advantages and limitations.

The GKO mouse offers a number of significant advantages over the existing rodent models. The most significant advantage is the rapid development of clinical symptoms of wasting and weight loss. No other model has this characteristic. A second major advantage is the infection of the entire small intestine (unlike the patchy infection seen in the other models), which closely resembles the parasite distribution in severely infected or immunocompromised humans. We believe that it is this extensive infection of the small intestine that accounts for the rapid intestinal dysfunction seen in GKO mice. The ability of paromomycin to decrease the small intestinal parasite burden, as well as to prevent death, reinforces this conclusion. A third major advantage is the very low dose of oocysts (10) needed to infect GKO mice, compared with the 10⁶ to 10⁷ required in other models. This dose is even lower than the minimum calculated to be required to infect humans in one volunteer trial (7). These characteristics bring the GKO model closer to the situation in humans than even the calf and the piglet models of diarrhea, which require higher doses to induce clinical disease (8, 17). It is of note that the use of an inoculum of 10 oocysts (instead of 5,000 or 10⁷) did not reduce the eventual severity of sickness; rather, it only increased the prepatent time. A fourth advantage of the GKO model is the number of clinical parameters that can be used for analyzing the impact of drug therapy on acute infection. In addition to oocyst shedding and mucosal scores, the two key parameters used in other rodent models, weight loss, physical appearance, behavior and well-being, gross gut appearance, and death are also available for evaluation in the GKO mouse model. Fifth, there is no need to use fragile neonatal mice, which are difficult to handle, given the exquisite sensitivity of adult GKO mice to infection. Sixth, the GKO model is rapid, requiring 3 weeks for completion of a drug assay, and requires no further conditioning, as do the immunosuppressed rodent and the IFN--conditioned SCID mouse models.

One potential drawback to this model is the absence of HB disease. HB tract involvement in SCID or anti-IFN--conditioned mice occurs only with chronic infection. The acute nature of cryptosporidiosis in GKO mice does not allow for chronic disease. A second drawback is the potential loss of mice in the placebo group, as was seen in the last experiment. This can be overcome by using older mice, which appear to survive longer, and increasing the group size. Third, despite the profound wasting manifested by weight loss, C. parvum-infected GKO mice do not develop diarrhea, a characteristic shared with all other rodent models. They do, however, display other symptoms consistent with gut dysfunction, as do immunocompromised humans.

Overall, these results strongly support the utility of the GKO mouse as a model superior to others currently used for drug discovery and evaluation. Treatment with paromomycin had a significant impact on the outcome of C. parvum infection, since it improved considerably the clinical symptoms, ameliorated parasite-induced weight loss, improved (and altered the distribution of) mucosal scores, and prevented death during the period of observation. These protective effects were dose related and demonstrated the ability of this model to show the efficacy of even drugs that are not curative. In the quest for curative drugs for this infection, this model will also likely provide a strong challenge for any putatively curative agent, since remnant subclinical infection during treatment will likely evolve into symptomatic infection and death once treatment ceases. We believe the GKO mouse is a powerful tool for the evaluation of, among others, anti-Cryptosporidium therapy, with a number of significant incremental advantages over prior models.