DNA extracts from 188 C. ubiquitum–positive specimens (collected during 2002–2012) were used, including those from animals, humans, and drinking source water. Animal specimens were obtained from various species of ruminants and rodents, a horse, a raccoon, and a primate, Verreaux’s sifaka (Propithecus verreauxicoquereli). Animal specimens were collected in the United States, Peru, Brazil, the United Kingdom, Spain, the Czech Republic, the Slovak Republic, Turkey, Algeria, South Africa, China, Nepal, and Australia (Table 1). Human specimens were obtained from the United States, Canada, Peru, the United Kingdom, and Turkey (Table 1). Water samples were collected from storm water and river water in the United States. The storm water samples were collected from a drinking-source watershed in New York where most of the rodent specimens were also collected (13,17). These specimens were initially found to be positive for C. ubiquitum by DNA sequence analysis of an ≈830-bp fragment of the small-subunit rRNA gene (25).

To identify a subtyping marker for C. ubiquitum, we sequenced the genome of an isolate from a specimen (33496) from a Verreaux’s sifaka by 454 technology using a GS FLX+ System (454 Life Sciences, Branford, CT, USA). This specimen was selected for whole-genome sequencing because of the high number of oocysts present, the availability of ample fecal materials for isolation of oocysts by sucrose and cesium chloride gradient flotation and immunomagnetic separation, and minor contamination from nontarget organisms in extracted DNA. Of the 3,030 assembled contigs of 11.4 MB nucleotides generated from 1,069,468 sequence reads, 1contig (no. 0067), consisting of 45,014 bp, had a high sequence similarity to the 5′ and 3′ ends of the gp60 gene and the flanking intergenic regions. Alignment of the contig 0067 sequence with the nucleotide sequences of the C. parvum gp60 gene (AF203016 and AY048665) led to the identification of sequences conserved between C. ubiquitum and C. parvum, which were used to design a nested PCR that amplified the entire coding region of the gp60 gene, except for the 54 nt at the 3′ end. The sequences of primers used in primary and secondary PCR were 5′-TTTACCCACACATCTGTAGCGTCG-3′ (Ubi-18S-F1) and 5′-ACGGACGGAATGATGTATCTGA-3′ (Ubi-18S-R1), and 5′-ATAGGTGATAATTAGTCAGTCTTTAAT-3′ (Ubi-18S-F2) and 5′-TCCAAAAGCGGCTGAGTCAGCATC-3′ (Ubi-18S-R2), which amplified an expected PCR product of 1,044 and 948 bp, respectively.

The partial C. ubiquitum gp60 gene was amplified by nested PCR in a total volume of 50 μL, containing 1 μL of DNA (primary PCR) or 2 μL of the primary PCR product (secondary PCR), primers at a concentration of 0.25 µM (Ubi-18S-F1 and Ubi-18S-R1) or 0.5 µM (Ubi-18S-F2 and Ubi-18S-R2), 0.2 µM deoxyribonucleotide triphosphate mix (Promega, Madison, WI, USA), 3 µM MgCl₂ (Promega), 1 × GeneAmp PCR buffer (Applied Biosystems, Foster City, CA, USA), and 1.25 U of Taq DNA polymerase (Promega). The primary PCR reactions also contained 400 ng/μL nonacetylated bovine serum albumin (Sigma, St. Louis, MO, USA) to reduce PCR inhibition. PCR amplification consisted of an initial denaturation at 94°C for 5 min; 35 cycles at 94°C for 45 s, 45 s at 58°C (primary PCR) or at 55°C (secondary PCR), 1 min at 72°C; and a final extension for 7 min at 72°C. Both positive (DNA from the sifaka specimen) and negative (reagent-grade water) controls were used in each PCR run.

Products of the secondary gp60 PCR were sequenced in both directions on an ABI 3130 Genetic Analyzer (Applied Biosystems). The sequences were assembled by using ChromasPro 1.32 (www.technelysium.com.au/ChromasPro.html), edited by using BioEdit 7.04 (www.mbio.ncsu.edu/BioEdit/bioedit.html), and aligned by using ClustalX 2.1 (www.clustal.org/). To assess the genetic relatedness of various subtype families of C. ubiquitum, we constructed a neighbor-joining tree by using MEGA 5.05 (www.megasoftware.net/). The reliability of cluster formation was evaluated by the bootstrap method with 1,000 replicates. To assess potential recombination among various subtype families, we used DnaSP 5.10 (www.ub.es/dnasp/) to calculate recombination rates on the basis of segregating sites (excluding insertions and deletions). Unique nucleotide sequences derived from this work were deposited in GenBank under accession nos. JX412915–JX412926 and KC204979–KC204985.

We compared the difference in the distribution of C. ubiquitum XIIa and non-XIIa subtype families between rodents and ruminants and between humans in the United States and the United Kingdom by using a χ² test implemented in SPSS 20.0 for Windows software (IBM Corp., Armonk, NY, USA). The difference was considered significant when the p value obtained was <0.05.

The gp60 gene in contig 0067 in the whole-genome sequencing of specimen 33496 was 945 bp in length, coding for a peptide that consisted of 315 aa. Except for the 5′ and 3′ regions, the gp60 gene of C. ubiquitum had no obvious similarity to the gp60 genes of C. parvum and C. hominis at the nucleotide level. At the amino acid level, the sequences shared 43%–48% sequence identity with those of C. parvum and C. hominis (Figure 1). The gene has the same structure of gp60 in C. parvum and C. hominis, with the first 19 aa coding for a signal peptide and the last 20 aa for a transmembrane domain. As in C. parvum and C. hominis, upstream of the transmembrane domain, the C. ubiquitum gp60 gene sequence had a C-terminus hydrophobic region that is likely linked to a glycosylphosphatidylinositol anchor, 2 potential N-linked glycosylation sites, and numerous O-linked glycosylation sites (26,27). However, no TCA/TCG/TCT trinucleotide repeats were seen in the 5′ region of the gp60 gene of C. parvum, C. hominis, and related species (28), and no N-terminal amino acids, DVSVE, were found between the putative cleavage site of the signal peptide and trinucleotide repeats (27). The putative furin cleavage site RSRR or KSISKR between gp40 and gp15 of C. parvum and C. hominis (29) was not found in C. ubiquitum (Figure 1).

Of the 188 DNA preparations of C. ubiquitum, 149 yielded PCR products of the expected size, and 145 were successfully sequenced, including 86 from animals, 41 from humans, and 18 from water (Table 1). Nucleotide sequences of isolates from 31 specimens were identical to those of the C. ubiquitum reference sequence (contig 0067 from specimen 33496). The remaining sequences had nucleotide differences of 8.2%–38.2% from the reference (Table 2). The gp60 nt sequences of C. ubiquitum differed from each other in length at most by 171 bp, with all insertions/deletions in trinucleotide, thus maintaining the reading frame (Technical Appendix Figure). Most of the length differences were due to 4 sequences generated from rodents in the Slovak Republic. The remaining sequences differed from each other in length by at most 12 bp. Most nucleotide sequence polymorphism occurred in the 3′ half of the gene (Technical Appendix).

The nucleotide sequences of the gp60 gene of C. ubiquitum formed 6 subtype families in a neighbor-joining analysis (Figure 2). They were named XIIa, XIIb, XIIc, XIId, XIIe, and XIIf, in concordance with the established nomenclature of gp60 subtype families (24). Subtype families XIIe and XIIf formed a cluster highly divergent from the dominant cluster of the 4 subtype families of XIIa, XIIb, XIIc, and XIId (Figure 2). In the dominant cluster, XIIc and XIId diverged at the 3′ end, and the 3′ end of the gene of XIIc had a high identity to XIIa, an indication of possible genetic recombination at the gp60 locus (Technical Appendix). Indeed, DnaSP analysis (www.ub.edu/dnasp/) revealed a minimum of 23 potential recombination events among the 6 subtype families (XIIa–XIIf), and 7 recombination events among the 4 subtype families (XIIa–XIId) in the dominant C. ubiquitum cluster. Pairwise recombination event comparisons revealed that the genetic recombination was mostly between XIIa and other subtype families (Table 2).

Within the XIIa subtype, minor sequence differences were found among specimens, leading to the formation of 3 common subtypes (subtypes 1–3) and several rare subtypes with a single nucleotide substitution each (for example, 19716, 35147, 37827, 37828, and 37830). A few nucleotide substitutions were also seen in the subtype family XIId (Technical Appendix). These subtypes within subtype families XIIa and XIId could not be named using the subtype nomenclature based on the number of trinucleotide repeats TCA, TCG, and TGT (24) because of the lack of such repeats in the gp60 gene of C. ubiquitum (Technical Appendix).

Among 86 isolates from animal specimens characterized, all 68 isolates from Old and New World ruminants belonged to the XIIa subtype family, including 56 from ovine specimens from the United States, Peru, Brazil, Spain, the United Kingdom, Turkey, China, and Australia; 1 from a goat specimen from Algeria; 3 from yak specimens from China; 1 from an alpaca specimen from Peru; and 7 from 5 species of wild ruminants from Nepal, the Czech Republic, and South Africa. One isolate from an equine specimen from the United Kingdom also belonged to the XIIa subtype family. In contrast, XIIb, XIIc, and XIId subtypes were all seen in specimens from several species of rodents and a few other wildlife species (1 raccoon and 1Verreaux’s sifaka) in the United States, and XIIe and XIIf were seen in specimens from field mice in the Slovak Republic (Table 1, Figure 2). The difference in the distribution of XIIa and non-XIIa subtype families between ruminants and rodents was significant (χ² = ∞; p <0.001). This was also the case when the comparison was made with only specimens from the United States (χ² = 26.00; p<0.001).

Among 41 isolates from human specimens successfully subtyped, 25 from the United States belonged to the subtype families of XIIb (9), XIIc (4), and XIId (12); 13 isolates from the United Kingdom belonged to the subtype families of XIIa (8), XIIb (3), and XIId (2); and 3 isolates from Turkey, Peru, and Canada belonged to the XIIa subtype family (Table 1; Figure 3). The difference between the distribution of XIIa and non-XIIa subtype families in humans the United States and United Kingdom was significant (χ² = 19.49; p<0.001).

Among 18 water samples from the United States, all 15 storm water samples were collected from a drinking source watershed, and 2 river water samples had the XIIb subtype family. Another river water sample revealed organisms from the XIId subtype family (Table 1).