Cryptosporidiosis represents a major public health problem. This infection, caused by protozoa of the genus Cryptosporidium, has been reported worldwide as a frequent cause of diarrhoea, and its prevalence varies according to different regions 1. In developed countries, massive Cryptosporidium foodborne and waterborne outbreaks have been reported. In developing countries, Cryptosporidium affects mostly children under five 2. Furthermore, cryptosporidiosis remains a clinically significant opportunistic infection in immunocompromised patients, causing potentially life-threatening diarrhoea, especially in those HIV-infected without access to highly active antiretroviral therapy (HAART) 3. Additionally, these parasites not only infect humans, but also cause morbidity in farm animals, leading to economic losses 4. Most Cryptosporidium species infect the epithelium of the gut but in severe infections, dissemination can occur to extra-intestinal sites 5. Infection of the intestinal cells can result in blunting of the intestinal villi, crypt hyperplasia and inflammation. Epithelial cell apoptosis due to this parasite has also been described 67.

Molecular techniques have been developed to differentiate this parasite at the species and genotype levels, showing that there are at least 16 different species 1, and several methods have been used to study and characterize different parasite strains. However, the understanding about invasion, transmission, pathogenesis and epidemiology is limited, and no effective drug against this infection is available. Previous studies based on multilocus characterisation of Cryptosporidium isolates identified different sub-groups within some Cryptosporidium species 891011. These groups exhibited different population genetic structures (epidemic clonality and panmixia) that could be correlated to defined phenotypes.

To contribute to the comprehension of the dynamics of infection and to investigate biological divergence between different populations of Cryptosporidium, tissue culture and animal models have been used. It has been difficult to find suitable models for cryptosporidiosis, as most mammalians are susceptible to infection only as newborns 12. However, some Cryptosporidium species can be propagated in either chemically or genetically immunosuppressed mice. Particularly, adult mice with congenital mutations such as nu/nu, scid, bg/nu, xid or mice with steroid induced immunosuppression are susceptible to infection 121314151617. Efforts are still needed to develop reproducible animal models allowing both the routine passage of different species and approaching unclear aspects of Cryptosporidium infection, especially in the pathophysiology field.

We have developed a model to compare chronic muris and parvum cryptosporidiosis using SCID mice treated with corticoids in order to evaluate clinical signs, location of the infection, pathogenicity, oocyst shedding and histopathological changes in the gastrointestinal tract. This model was chosen for the following reasons: i) Previous studies have shown that host cell immunity during cryptosporidiosis is mediated by both Th1 and Th2 response 1819, thus SCID mice are more susceptible to the infection and to develop a chronic disease due to their defect in T and B lymphocytes 20. ii) It has been found that during cryptosporidiosis there is an IFNγ mucosal response with increased levels of Il15 21. IFNγ-induced enterocyte resistance against C. parvum has been reported 22. Furthermore, glucocorticoids are known to have an effect on the priming of the innate immune response 23, and could suppress IFNγ-regulated gene expression 24. Consequently, dexamethasone, a synthetic glucocorticosteroid, could be used to alter this innate immune response. iii) This model of SCID mice treated with steroids has been proven to be successful for the development of Pneumocystis, another opportunistic agent 25.

The pathological damage due to C. muris or C. parvum infection was studied in SCID mice treated or not with Dex. Main data related to infected mice included in the study are shown in Table 1.

Dex induced a significant body weight loss in both Cryptosporidium infected and uninfected SCID mice (P < 0.001). In contrast, no significant body weight change was associated with parasite infection. Once infected with C. muris, mice from groups M and MDex became chronic shedders and produced high numbers of oocysts without marked differences between the groups. On the other hand, in animals inoculated with C. parvum, the oocyst excretion was sporadic and limited in P mice, and high and chronic in PDex mice, with a statistically significant difference (P = 0.002). Figure 1 shows the average of oocyst excretion in different groups of mice. Analysis of variance showed that the administration of Dex, Cryptosporidium species, and the interaction of these factors significantly influenced oocyst excretion (all the P < 0.001). The level of oocyst excretion was higher in mice infected with C. muris than in those infected with C. parvum.

Pre-patent period ranged from 6 to 11 days (9.7 days in average) in M mice and it was 11 days for all MDex mice. Geometric means of oocyst excretion before sacrifice were 1839 and 1159 oocysts/mg faeces respectively (Table 1). At histological level, no difference between mice of groups M or MDex was observed. C. muris localization was restricted to the stomach, with no extra-gastric dissemination. Different stages of the parasite life cycle were present. Stomachs of all mice euthanatized at different dates, from day 20 to day 84 post-infection, were heavily infected mainly in fundic mucosae (Figure 2). Fundic gastric glands were dilated and covered by a flattened epithelium. Oxyntic cells were less numerous than in control mice and preferentially visible in the deeper area. Fundic mucosa was twice thinner than in control mice while pyloric mucosa was similar to that in uninfected control mice. A moderate inflammatory infiltrate, made of mononuclear cells and neutrophile polynuclears, was observed in some cases.

For C. parvum-infected SCID mice from group P, geometric mean of oocyst excretion before euthanasia was 1 oocyst/mg faeces. SCID mice from PDex group had a prepatent period ranging from 4 to 11 days (6.2 days in average) and developed a chronic infection. Geometric mean of oocyst excretion before sacrifice was 195 oocysts/mg of faeces for mice from group PDex. Two out of six PDex SCID mice were severely ill and required euthanasia. In mice infected with C. parvum, histopathological differences were revealed between Dex-treated and untreated animals. Untreated SCID mice infected with C. parvum have neither detectable parasites nor lesions at the histological level (Figure 3) at any time during the course of the study.

In Dex-treated SCID mice infected with C. parvum, parasite localization was restricted to the gastrointestinal tract, mainly at the caecal region (Figure 3). Three out of five (60%) histologically examined PDex SCID mice presented in the ileocaecal region polypoid, sessile, adenomatous masses, measuring approximately 2.5 mm in diameter. They appeared as closely packed, branching sometimes dilated tubular structures, separated by normal or inflammatory lamina propria. Focal cystic dilation was observed. Noticeably some tubules were covered by a low grade or high grade dysplastic epithelium, which showed mucin depletion and nuclear stratification. In some areas, architectural distortion was associated with marked cellular atypias (Figure 4). Epithelial cells presented a loss of their normal polarity. In addition, abnormal nuclear changes consisting of prominent nucleoli and irregularly scattered chromatin were observed (Figure 4). These mucosal changes were suggestive of intraepithelial neoplasia of low or high grade. In some areas, major cellular atypias, with foci of merged glands, typical of intramucosal adenocarcinoma, invasive into lamina propria, were found (Figures 4A and 4B).

These major histological changes were earliest observed in one mouse euthanatized at day 46 post-infection and were found in all PDex mice necropsied after (Table 1). They were always associated with the presence of C. parvum organisms (Figures 4 and 5). In contrast, no parasite was detected in the stomach, duodenum, jejunum, hepatic biliary system or pancreas. The possibility that biotic (e.g. virus) or abiotic factors present in the inoculum could be responsible for these lesions was discarded by administering Dex-treated SCID mice with a filtered inoculum (i.e. without Cryptosporidium oocysts) (group Cdex2). Neither parasites in the faeces or tissues nor lesions were detected in these mice at any time of the experience.

We developed a new animal model that allows a good propagation of two different species of Cryptosporidium. In this study, adult SCID mice treated with Dex became chronically infected with 10⁵ C. parvum or C. muris oocysts, and had a significant oocyst shedding during all the course of the experiments. Furthermore, this model was useful to compare C. muris and C. parvum infections at clinical, histopathological and parasitological levels.

In this study, variations in the expression of the disease in terms of either Cryptosporidium species or Dex administration were shown. Amounts of oocysts discharged seemed slightly lower in MDex than in M mice but this difference was not statistically significant. However, it appeared that the aggravation of the immunosuppression status did not lead to an increase in the severity of C. muris infection. Miller et al. previously reported that immunosuppressed mice were as or less susceptible to C. muris than immunocompetent mice 26. However, this conclusion can hardly be extrapolated to all Cryptosporidium species, as long as in our experiments, the oocyst shedding was markedly and significantly higher in PDex than in P group. Further studies are required to determine minimal infectious dose, ID₅₀ (infectious dose to 50 percent of exposed individuals) and other data for each species and according to the immunosuppression degree. Interestingly, data from very recent experiments with C. parvum confirmed the results of the present work (data not shown). The Dex-treated SCID mouse model revealed therefore to be reproducible.

C. muris infection caused damage to the gastric mucosa of both mice from M and MDex groups. But though M mice were less ill than MDex mice, there was no marked histological difference between them. Dilated, hypertrophied and highly parasitized gastric glands without an extra gastric dissemination of parasites were the main histopathological changes. Additionally, a little inflammatory response was observed. Other authors have reported similar findings in relation with C. muris gastric infection 122728. However, to our knowledge, this is the first report of a clear preferential localization of the parasite colonization at the fundic level of the stomach, where acid secreting glands are numerous 29. This could suggest a favourable influence of lower gastric pH on the growth of C. muris.

On the other hand, several studies using animal models have described histological changes in the intestinal epithelium due to C. parvum infection, such as villous atrophy and crypt hyperplasia in the lower small intestine 13 or in the caecum and the colon 12, cryptic hyperplasia with abscesation of crypts in the large intestine 13, stunting and fusion of villi, replacement of enterocytes by immature cells and eosinophilia of lamina propria 30, small and large intestine mucosa severely damaged with villous contraction and little or absent epithelial layer 31. Another report described an association between Cryptosporidium sp. and aural-pharyngeal polyps in iguanas. These polyps were pedunculated masses composed of glandular cystic structures lined by hyperplastic cuboidal to columnar epithelium, containing numerous parasites along the apical surfaces of the epithelial cells 32. Nevertheless, none of these studies have described the presence of carcinoma lesions associated to cryptosporidiosis. These lesions were first observed unexpectedly after 46 days post-infection, when the infection had become chronic, and were found only in C. parvum infected SCID mice treated with Dex. These findings were recorded using an inoculum relatively low (10⁵ oocysts) in comparison with the higher infectious doses, between 10⁶ and 10⁷, used by others 1233.

Several observations in our study suggest that combination of C. parvum with Dex administration is involved in the generation of these significant histological changes. Indeed, Dex seemed to be a critical factor to the development of intraepithelial adenocarcinoma in C. parvum-infected SCID mice. Dex potentially alters the innate immunity in C. parvum infected animals 2324. The higher oocyst shedding found in mice presenting these neoplasic lesions seems to confirm this assumption. Additionally, this kind of lesion was found neither in Dex-untreated C parvum-infected animals with a low oocyst excretion during the course of the study, nor in the control Dex-treated non-infected group. A possible contamination of the inoculum with a virus or other agent potentially responsible for these neoplasic lesions was also discarded. Interestingly, C. muris was not able to induce this type of epithelial changes and the reasons of this different expression of the disease are unclear. It has been reported previously a variability in pathogenicity for different hosts between different Cryptosporidium species and types, suggesting the existence of specific virulence factors among species and isolates 34. Further studies should be done to clarify this difference in pathogenicity. It is important to mention that these colonic neoplasic lesions are not listed among the typical background diseases due to the SCID mice genetic defect 35.

To our knowledge, this is the first time that C. parvum is associated with cancerogenesis. There is one case report that described cryptosporidiosis of the biliary tract clinically mimicking a pancreatic cancer in an AIDS patient 36. However, biopsy of the gallbladder revealed cryptosporidiosis causing inflammation of the biliary tract and ruled out the diagnosis of neoplasia 36. Furthermore, a recent epidemiological study in Poland reported a high frequency of cryptosporidiosis in patients with colorectal cancer 37. These data strengthen the interest of our experimental observations.

Several infectious agents, including parasites, have been linked to oncogenesis in humans. Some of these associations are strong. Particularly, Schistosoma hematobium has been described as a definitive cause of urinary bladder cancer by the International Agency for Research in Cancer. Also, a proportion of cholangiocarcinoma of the liver worldwide is attributable to Opisthorchiidae liver flukes 38. Other associations still speculative with some protozoa were suggested on the basis of epidemiological observations, such as Trichomona vaginalis and cervical cancer, or Toxoplasma gondii and tumors 39. As well, a microbial pathogen such as Helicobacter pylori has been classified as oncogenic for humans due to its strong epidemiological association with carcinoma of stomach and gastric lymphoma 3840.

Because C. parvum is an opportunistic agent that causes significant morbidity and mortality in immunocompromised patients, it is possible that individuals infected with this parasite may have a higher risk of developing colorectal malignancies, especially when immunosuppression is more severe. In a retrospective study that shows the incidence and clinical course of colorectal malignancies in HIV/AIDS patients, adenocarcinoma was the type of cancer most frequently found among them. These patients did not have familial antecedents of intestinal malignancies or other risk factors, and developed tumors at earlier ages in comparison to immunocompetent persons 41. Unfortunately, no data about gastro-intestinal parasites in these patients were given in the study 41. More studies have to be done in humans to evaluate cryptosporidiosis as a possible risk factor of colorectal cancer.

Epidemiological studies have shown that environmental factors may play a role in colon cancer susceptibility 42. The association between chronic inflammation and cancer is also well known 43. Particularly, colon cancer has also been associated to inflammatory bowel disease. In the latter case, cancer develops after sustained gut inflammation. However, in the present work C. parvum infection was associated with only mild inflammation of the intestine as reported also by others 12.

Taking into consideration that the observations described herein were replicated in a second experiment (unpublished data), it can be concluded that C. parvum is able to induce the formation of polyps and carcinogenic lesions in the gut of Dex-treated SCID mice. However, more work has to be done to elucidate this interesting association and to further understand cryptosporidiosis pathogenesis.

In summary, for the first time C. parvum was associated with the generation of polyps and in situ adenocarcinoma in the gut of Dex-treated SCID mice. Additionally, we have developed a model to compare chronic muris and parvum cryptosporidiosis using SCID mice treated with corticoids. This reproducible model has allowed the evaluation of clinical impact, oocyst shedding, location of the infections, pathogenic power, and histopathological changes indicating divergent effects of Dex according to Cryptosporidium species.

The author(s) declare that they have no competing interests.

GC & TN have equally contributed to this work. They participated in the conception and design of the study, carried out the experiments and drafted the manuscript. KG participated in the design of the experiments. NG, TC, AM participated in the performance of animal experiments. LF prepared the histological cuts. AP carried out the statistical analysis. JCC participated in the design of the study. ED participated in the design and coordination of the study and helped to draft the manuscript. CC carried out the pathological study and helped to draft the manuscript. All authors read and approved the final manuscript.