Cryptosporidium is a genus of protozoan parasites infecting a wide range of hosts 1 . All groups of vertebrates are susceptible to Cryptosporidium infection worldwide. This parasite is the etiological agent of cryptosporidiosis, which is mainly characterized by diarrhea in humans and livestock. Massive outbreaks of enteritis in people such as in Milwaukee, Wisconsin (USA) have increased public awareness of this parasite 2 . In humans, the prevalence and severity of infection increase in immunodeficient individuals such as AIDS patients. In immunocompetent patients, the disease is self-limited 3 . No drug therapy is yet available and the high resistance of oocysts to environmental conditions and chemical treatment make cryptosporidiosis difficult to control 4 . Cattle have been considered to be a primary reservoir for Cryptosporidium oocysts for zoonotic C. parvum 5 . These animals could be a risk factor via environmental contamination from their manure being spread on farmland or their grazing on watersheds 6 . On farms, transmission of Cryptosporidium spp. can result from ingestion of contaminated food or water, by direct transmission from host to host, or through insect vectors 7 . In cattle, infection by Cryptosporidium spp. was first reported in 1971 8 . Since vaccines have become commercially available against Escherichia coli K99, rotavirus, and coronavirus, Cryptosporidium has emerged as the main neonatal diarrheic agent in calves 9 . In farm animals, the economic impact is related to morbidity, mortality and growth retardation 10 . Among the 24 species previously described (if the three fish species are accepted without complete genetic characterization) 1 11 12 13 , C. parvum, C. bovis, C. ryanae and C. andersoni usually infect cattle. C. parvum has zoonotic potential and is a frequent cause of human cryptosporidiosis 14 . C. bovis and C. ryanae have not been found in humans and there is only one description of C. andersoni in a patient 15 . Recent reports have described an age-related distribution of these aforementioned species in dairy cattle on the east coast of the United States 16 17 18 , India, China, Georgia 19 , Malaysia 20 , and Denmark 21 . The most prevalent species were C. parvum in preweaned calves, C. ryanae and C. bovis in postweaned calves and C. andersoni in adult cows 16 17 .

In France, previous studies on the prevalence of Cryptosporidium in cattle were based on microscopic determination 22 or coproantigen detection using ELISA 23 . These studies on dairy calves reported a within herd prevalence of Cryptosporidium without identifying species or the relation to the host's age. The present study was conducted in 15 farms in Brittany, France to determine the prevalence of Cryptosporidium in veal calves. We used genotyping and subtyping for the molecular study of Cryptosporidium isolates. Follow-up of the same animal allowed us to determine the outcome of the infection and the age distribution of Cryptosporidium species.

Calves under 1 month of age are frequently infected with Cryptosporidium sp 29 which results in economic loss 10 . In France, up to date, the prevalence of Cryptosporidium in diarrheic calves has been studied only by Elisa and microscopic strategies 22 23 30 . No data are available on a molecular basis to study Cryptosporidium species in calf herds in that country. The present study based on 18S rDNA and gp60 gene analysis is the first in France to include molecular characterization to describe the prevalence and the host age related susceptibility to different Cryptosporidium species after a follow up of the same animal.

Our results showed that all fifteen farms were contaminated with Cryptosporidium. The parasite prevalence on farms ranged from 10% to 100% of the sampled animals. This observation was in accordance with results in Michigan (USA) where this parameter ranged from 0% to 100% 31 . The prevalence of 70.4% obtained in this work tended toward the upper end of the scale compared to other investigations done in France which ranged from 15.6% in beef herds 30 to 95% in suckling calves 23 and in other European countries where prevalence ranged from 3.4% to 96% 32 33 . However, the sampling program did not allow the study of animals under 5 weeks of age. Indeed, the animals arrived in these structures at the age of 2 to 3 weeks and farmers did not allow sampling before two complete resting weeks for each animal. Therefore, our results could underestimate the real prevalence as Huetink et al. showed that the percentage of parasite excreting animal declines after the third week of age 34 and that the first peak of prevalence is at the age of 15 days 17 .

In our study, the higher prevalence of cryptosporidiosis was observed in calves 5 weeks old (47.9%) and the lowest (14.3%) in the 22-week-old animals. This observation shows that prevalence of Cryptosporidium infection decreases with increasing age of the cattle in France as in many other countries 17 19 33 34 35 36 37 38 .

Additionally, our data confirmed the presence in France of a host age-related susceptibility to the infection with different Cryptosporidium species. C. parvum was predominantly detected in 5-week-old calves (86.7%) compared to C. ryanae or C. bovis detected in 4.4% and 1.5% of the positive samples respectively. It is noteworthy that these results are very similar to data obtained in Ireland on calves under 30 days of age with 95%, 3.6%, and 1.3% of prevalence of the same species, respectively 39 and in the UK on animals over 3 weeks old with 93% C. parvum, 6% C. bovis, and 2% C. ryanae 40 . In contrast to previous studies 17 41 , C. ryanae and C. bovis were found with similar prevalence predominantly in 15 week and 22 week old calves. This association between the age of the cattle and the Cryptosporidium species identification has been supported by several studies 17 19 21 38 40 but different reports suggest that Cryptosporidium species repartition regarding the age of the host could be due to a technical artifact. Despite the fact that the methodological strategy based on PCR using genus specific primers and partial direct sequencing of the 18S rDNA is commonly used to identify Cryptosporidium species 42 , this molecular tool is limited in the case of mixed infections. Feng et al., 19 suggested that the important shedding of C. parvum in preweaned calves had probably masked the concurrent infection of these animals by C. bovis or C. ryanae. Furthermore, previous reports suggested that a dominant Cryptosporidium species in a sample can be preferentially amplified by PCR 43 44 . It is noteworthy that this situation of mixed Cryptosporidium species infection in farm animals would be more prevalent than originally believed 45 46 47 . Mixed Cryptosporidium species could also explain sequencing difficulties encountered in this work. The simultaneous presence of several species in the same sample could lead to amplification and sequencing of different genetic fragments leading to unreadable superimposition of electrophoregrams.

Consequently, in our work based on the utilization of Cryptosporidium generic primers, the amplification of a single fragment with a single sequence is not conclusive evidence that the sample contains only a single species. However, based on our results, it is possible to confirm the predominance of different species of Cryptosporidium by group of age among the calves.

Particularly, our data showed that animals can be sequentially infected with C. parvum, C. ryanae and C. bovis as well as C. parvum, C. bovis and C. ryanae. This observation provides evidence that a previous infection with C. parvum did not protect calves against an infection with other Cryptosporidium species. Fayer et al. suggested that the peak of cryptosporidiosis prevalence in young calves could reflect the immaturity of the immune status 48 . It was also suggested that the low excretion of C. parvum oocysts in older calves might be related to the development of immunity that also protected the animal against a secondary challenge 49 . It has been reported that immunity arises in the first two weeks after infection 50 . Interestingly, Fayer et al. 51 described that calves previously challenged with C. parvum were able to excrete oocysts after a second challenge with C. bovis but not with C. parvum. The authors concluded that immunity to C. parvum was not extended to C. bovis. Consistently, in our study, the presence in the same animal during sequential sampling of C. parvum, C. bovis and C. ryanae suggests that immunity against C. parvum and against C. bovis did not extend to C. ryanae. Furthermore, the observation that one animal excreted sequentially C. parvum, C. ryanae and C. bovis suggests that immunity against C. ryanae did not extend to C. bovis as well.

Finally, the risk to human health posed by Cryptosporidium infected cattle in France was assessed. The detection of C. ubiquitum (a rare infectious agent detected in humans 52 ), C. ryanae and C. bovis (which are mainly specific for cattle) led to consider that the 22-week-old calves are not likely a public health concern. However, the major detection of C. parvum, a prevalent zoonotic species, in 5-week-old calves was in agreement with the report of Atwill et al., who considered that the contribution of cattle to human cryptosporidiosis is limited to calves under 2 months of age 53 .

To determine C. parvum subtypes, the sequence analysis of a fragment of the gp60 gene was done. Our results show that in the region of Brittany, all identified C. parvum gp60 subtypes belonged to the IIa family which was previously found in both animals and humans 42 . Particularly, human infections with the IIa subtype are commonly seen in areas with intensive animal production 54 . Among the 48 gp60 subtypes formerly described in cattle 55 , only six were identified in this work, being IIaA15G2R1 the most commonly found. This subtype has been widely reported in calves and humans in different countries such as in Portugal 54 , Slovenia 56 and The Netherlands 57 . This observation confirms previous works and suggests a zoonotic transmission of the parasite also in this region.

It is noteworthy that the three predominant subtypes (IIaA15G2R1, IIaA17G1R1, and IIaA16G3R1) found in this work were also described in cattle with an equivalent distribution in The Netherlands 57 and England 40 . Thus, the subtype IIaA15G2R1 was found in 74.5% of the samples in this work, 68.9% in The Netherlands and 68.6% in England. The IIaA17G1R1 was identified in 11.7% of the samples in this report, 10.8% in The Netherlands and 13.8% in England. The IIaA16G3R1 determined in 5.9% of our samples, was characterized in 4.65% in The Netherlands and 5.8% in England. It is remarkable that subtypes, IIaA16G2R1, IIaA16G1R1 and IIaA13G2R1 were equivalently underrepresented in these three countries. This observation could suggest that the proportion of a gp60 subtype would not be randomly represented in a population.

Finally, the zoonotic transmission assessment of C. parvum in France would require a comparative investigation of variable genetic loci both in human and animal samples.

This is the first report on the molecular identification of Cryptosporidium species or genotypes in veal calves in France. According to data reported previously in many countries, a sequential distribution of species is observed in cattle according to age. C. parvum was mainly observed in the youngest calves, while C. ryanae and C. bovis became predominant in stool specimens collected in older animals. In some cases, several Cryptosporidium species were successively detected in the same calf, suggesting that the immune defense against C. parvum is not efficient against C. ryanae or C. bovis. Finally, the major identification of the IIaA15G2R1 subtype in France suggests that 5-week old calves could be a reservoir for zoonotic parasites transmissible to humans.

The authors declare that they have no competing interests.

JF and KG participated in the conception and design of the study, carried out the experiments and drafted the manuscript. HL designed the sampling strategy and collected samples on farms. JF, KG and AFD designed the protocol for molecular assay and participated in the analysis result. OHG carried out molecular assays. EDC, GC and PH participated in the coordination of the study and helped draft the manuscript. All authors read and approved the final manuscript.