Cryptosporidium oocysts were isolated from fecal specimens of 150 sporadic cryptosporidiosis cases. Two characteristic restriction profiles were obtained for Mse I digests of the 600-bp ITS1 products (Figure 1). The first type of restriction profile (Figure 1, lanes 4 and 15) showed five major bands at approximately 270, 160, 90, 75, and 55 bp. The bovine isolate, patient isolates from the 1996 Cranbrook and the 1998 Chilliwack outbreaks, and 29 (19%) isolates from sporadic cases had this restriction profile. Based on results from previous molecular characterization of a number of these isolates (1,2), this restriction profile was considered to be the C. parvum genotype 2 restriction pattern. The second restriction pattern (Figure 1, lanes 5, 6, 14, and 16) with six major bands around 185, 150, 100, 60, 40, and 30 bp was obtained from isolates of 108 (72%) sporadic cases and the one patient from the 1996 Kelowna outbreak. This restriction profile was considered to be the C. parvum genotype 1 profile, based on results from previous molecular analyses on the other seven genotype 1 isolates that were included as human genotype controls (2).

Restriction profiles with varying patterns (Figure 1, lanes 7, 9 to 13 and 17) were obtained from 13 (9%) other human isolates. Of these, nine (6%) isolates had identical restriction profiles (Figure 1, lanes 7 and 9 to 12) with five major bands at 150, 85, 70, 45, and 35 bp and were designated cervine genotype isolates. The other four isolates had unique restriction profiles and could be split into two groups of two isolates, based on the similarity of banding patterns. These were CS33 (Figure 1, lane 8) and MH222 (data not shown), which both had restriction fragments at 175 and 50 bp and additional variant bands at 65 and 70 bp, respectively. The remaining two isolates VF383 had bands at 315 and 105 bp (Figure 1, lane 13) and TK348 had bands at 325 and 85bp (Figure 1, lane 17), respectively.

To identify the Cryptosporidium species and genotype of isolates with variant restriction profiles, sequencing of a polymorphic locus on the 18S rRNA gene was carried out. Eleven isolates were selected based on their ITS1 restriction patterns. These included one isolate (TK386) with a characteristic human genotype 1 restriction profile, three isolates (TK324, TK303, DE340) with patterns similar to the human genotype 1 restriction profile but with one restriction fragment shifted slightly in molecular size (e.g., Figure 1, lane 14), three cervine genotype isolates (MH205, TK320, DE302; Figure 1, lanes 7, 10, and 11), and four isolates (CS33, MH222, VF383, TK348) with unique variant profiles (Table). Comparison of the 18S rRNA gene sequences of these isolates with 22 other published reference sequences, derived from a variety of human and animal Cryptosporidium isolates by using multiple sequence alignment and phylogenetic analysis (Figure 2), showed that the 11 isolates fell into four main groups. The first group consisted of four isolates (TK386, TK324, TK303, and DE340) that had restriction profiles identical or similar to the characteristic human genotype 1 pattern and all human genotype 1 reference isolates. All human isolates in the genotype 1 group had the characteristic polyT repeat sequence reported previously (10,26) between positions nt 686 and 698. The second group consisted of three human isolates (TK320, DE302, and MH205) and two cervine genotype isolates (Figure 2). The sequences of these three human isolates in the hypervariable 18S rRNA region were identical to that of a genotype 3 deer isolate described by Perz and Le Blancq (14). The third group consisted of two isolates (VF383 and TK348) and a pig genotype isolate (Figure 2). Sequences between the two human isolates were variant from the pig genotype sequence at only two different nucleotide positions between nt 686 and 698. The fourth group consisted of two human isolates (MH222 and CS33) and a C. meleagridis isolate (Figure 2).

Twenty-five sporadic isolates were also characterized by using a second locus on the COWP gene. Rsa I digests of the 550-bp PCR products (Figure 3) also showed the same dimorphism as the ITS1 locus with two predominant restriction patterns. Of these, 10 (40%) isolates had fragments at approximately 285, 125, 105, and 35 bp, which were characteristic for genotype 1 isolates (Figure 3, lanes 7 and 8). Another six (24%) isolates had the genotype 2 restriction profile with fragments at 410, 105, and 35 bp (Figure 3, Lanes 4 to 6). One isolate (CS33) had a variant restriction profile (Figure 3, Lane 10) with bands at approximately 370, 290, and 150 bp, which were similar in size to fragments reported for a C. meleagridis isolate (26). The 18S rRNA gene sequence of this isolate (CS33) was identical to that of C. meleagridis. The other isolate (MH205), had a restriction profile that was identical to those obtained from genotype 1 isolates (Figure 3, lane 9). This isolate had an identical ITS1 restriction profile with eight other sporadic human isolates and an 18S rRNA gene sequence that grouped with deer genotype 3 isolates from New York (Figure 2). The COWP gene sequences of MH205 and another cervine genotype isolate TK320 were determined to be identical and novel, sharing only 90% and 91% identity with the COWP gene sequences of the human (AF248741) and bovine (AF248743) alleles, respectively. BLAST analysis showed most alignment (92% identity) with a pig COWP gene sequence (AF266270) (17). However, the cervine allele had identical RsaI restriction sites to the human allele at nt 34, 228, 512, and 618, whereas the bovine allele lacked the site at nt 228. The RFLP patterns could not be determined for the remaining seven isolates as insufficient PCR product was obtained. Over half (51%) of the isolates in this study were derived from pediatric patients <10 years of age, which accounted for seven of the nine cervine genotype infections as well as the two C. meleagridis infections.

This study describes the discovery of the first zoonotic infections in humans with a novel cervine Cryptosporidium parvum genotype. Perz and LeBlancq (14) described this genotype recently after characterizing 111 Cryptosporidium isolates from wildlife in New York state. Those researchers did not detect human infections with this genotype in cryptosporidiosis cases in New York City. Other molecular epidemiologic studies in England (8,18) of 1,705 cases also did not identify cervine genotype infections, although rare zoonotic infections in humans with the dog genotype of C. parvum as well as other Cryptosporidium species such as C. felis and C. meleagridis were found (27). It is possible that cervine genotype infections in humans were not identified because the novel deer genotype had not been reported at the time of the study. As well, the PCR/RFLP profile of the COWP gene from the cervine genotype isolate was identical to that obtained from human genotype 1 isolates. Sequencing of the COWP gene from two cervine genotype isolates confirmed that the human and cervine alleles had identical RsaI restriction sites. Therefore, RFLP analysis using this endonuclease could not differentiate between isolates with these two genotypes.

Xiao et al. (28) have also found this novel cervine genotype in storm water samples collected from a stream in the watershed area of New York State that contributes to the New York City water supply. The transmission of cryptosporidiosis from wildlife to humans in British Columbia is not surprising as many communities are supplied with unfiltered drinking water drawn from surface sources where Cryptosporidium spp. oocysts have been detected (29). Many of these watersheds are situated in remote forested areas, where wildlife such as deer are present in abundance. Deer with cryptosporidiosis infections have been identified in these watersheds (Ong et al., unpub. data). Therefore, to have as many as 6% of sporadic cases infected with this novel deer genotype is not an unexpected finding.

The ITS1 and 18S rRNA genes are reportedly multicopy genes with four copies of the Type A and one copy of the Type B rDNA units per haploid genome (30). The sequence divergence found between Type A and Type B units in the ITS1 region was a concern to us initially, as we first characterized the isolates using this locus before this report. However, Morgan et al. (31), who conducted a similar PCR-RFLP analysis of the ITS1 region, found that the restriction profiles were specific for different C. parvum genotypes. This study also indicated that intraorganism variation caused by the difference between Type A and Type B rDNA units may not be such a problem. Using primers to amplify the Type B unit, Morgan et al. (31) found that the Type A unit was amplified preferentially for human genotype isolates. To confirm that the observed variation in the ITS1 RLFP patterns was not due to heterogeneous products amplified from different copies of rDNA, further characterization of a select number of sporadic isolates was performed with the 18S rRNA as well as the COWP genes. Results from these additional analyses showed that isolates with distinctly different ITS1 RFLP patterns had different COWP RFLP patterns as well as 18S and COWP gene sequences. Therefore, ITS1 RFLP was useful for generating characteristic fingerprints that could distinguish between different C. parvum genotypes and Cryptosporidium species.

Using this method of genotyping, we were able to detect two new genotypes of C. parvum that had not been reported previously. Nine isolates (including three, MH205, TK320, and DE302, which had been characterized at the 18S locus; and two, MH201 and TK320, which had been characterized at the COWP locus) had the cervine genotype. Two other isolates (VF383 and TK348) had novel genotypes that were most closely related to a pig genotype isolate from Switzerland (13). These results have important implications for drinking water quality strategies, especially for communities that obtain drinking water supplies from surface sources located in forested regions with deer populations.