Figure 1

Figure 1. Geographic distribution of TcII, TcIII, TcV, and TcVI Trypanosoma cruzi clones, South America, 2002–2010. A total of 57 T. cruzi biologic clones were assembled for analysis. Of these, 24 were isolated...

For analysis, we assembled a panel of 57 T. cruzi biologic clones from a range of representative hosts/vectors across South America: 24 uncharacterized clones from Colombia and 33 reference clones (Figure 1; Technical Appendix 1 Table 1). From 2002–2010, we isolated the uncharacterized clones from humans; triatomine vectors (Panstrongylus geniculatus, Rhodnius prolixus, and Triatoma venosa insects); and sylvatic mammalian hosts (Dasypus spp. armadillos) in 3 T. cruzi–endemic departments in northern Colombia.

We genotyped all isolates using nuclear housekeeping genes GPX, GTP, Met-II, TcAPX, and TcMPX (6,8) (online Technical Appendix 1 Table 2); 25 microsatellite loci (Technical Appendix 1 Table 3) (9); and 10 mitochondrial gene fragments (10). Diploid multilocus sequence typing (MLST) data were analyzed by locus and concatenated according to their relative chromosomal positions in MLSTest (11); heterozygous variable sites were handled as average states. Gene haplotypes were inferred using PHASE version 2.1 (12). PCR products were cloned and sequenced to confirm ambiguous gene phases. We constructed maximum-likelihood and Bayesian phylogenies for nuclear haplotypic and concatenated mitochondrial data (13).

Figure 2

Figure 2. Phylogenetic trees showing relationships between Trypanosoma cruzi hybrids from Colombia and reference T. cruzi strains from across South America. A) Unrooted neighbor-joining tree based on pairwise distances between microsatellite loci. B)...

For microsatellite loci, we defined sample clustering using a neighbor-joining tree based on pairwise distances between multilocus genotypes (Figure 2) (13). We calculated DTU-level heterozygosity (Bonferroni-corrected) and evaluated genetic diversity using sample size–corrected allelic richness and private allele frequency per locus (Table). To examine TcV/TcVI allele inheritance, we classified genotypes at each locus as hybrid (TcII/TcIII) or nonhybrid (TcII/TcII or TcIII/TcIII) based on the presence or absence of specific parental alleles (Technical Appendix 2).

All putative hybrids from Colombia were highly heterozygous and minimally diverse. They possessed TcII and TcIII alleles at an approximate 1:1 ratio and, compared with parental DTUs, they displayed fewer private alleles or single-nucleotide polymorphisms; these strains fulfilled all the expectations of progeny from a recent Mendelian hybridization event(s) (Table). Based on nuclear MLST and microsatellite data, all hybrids from Colombia were classified as TcVI, not novel recombinants.

Examination of TcII and TcIII alleles across 5 nuclear loci showed that hybrid haplotypes from Colombia were shared among other TcVI strains from the Southern Cone region of South America and showed negligible affinities to parental alleles from Colombia (Technical Appendix 1 Figures 1, 2). Microsatellite profiles also supported this allopatric inheritance: only a minority of private parental alleles from Colombia were common to local TcVI hybrids. At mitochondrial loci, TcVI clones from Colombia were noticeably divergent from local TcIII maxicircle haplotypes and those observed in reference TcVI strains (Figure 2). Of note, 1 hybrid from Colombia (AACf2 cl11), which was unequivocally classified as TcVI by both types of nuclear loci, possessed a TcV-type mitochondria. All isolates in this study were biologic clones, ruling out mixed infections as a potential confounder.

Overall, our data support the hypothesis that 2 separate recombination events led to the formation of TcV and TcVI. These interlineage hybrids have distinct nuclear and mitochondrial MLST genotypes and related but independent microsatellite profiles, and most alleles that distinguish between hybrid DTUs (70.4% [38/54 alleles]) were also present in their corresponding parental strains. Interlineage differences (fixed at 84% [21/25 of loci]) between TcV and TcVI are not consistent with allelic sequence divergence (Meselson effect); for such divergence, a much higher frequency of private alleles, compared with parental genotypes, would be expected at rapidly evolving microsatellite loci.

TcVI clones from Colombia had more private microsatellite alleles per locus (0.86) than their southern counterparts (0.43), despite their unequivocal origin in the Southern Cone. This phenomenon could be attributable to de novo mutations or a founder effect with respect to the northerly introduction of TcVI. Support for the latter cause is evidenced by an overall reduction in genetic diversity among hybrids from Colombia compared with TcVI strains from the Southern Cone (allelic richness 1.87 vs. 2.46, respectively). However, we cannot discount some sampling bias because reference Southern Cone strains represented a much wider geographic range.

A novel observation among TcVI strains from Colombia was the presence of an anomalous TcV maxicircle. This pattern of inheritance could reflect 1) recent mitochondrial introgression from TcV into TcVI, leaving undetectable signatures of nuclear hybridization by our markers or, possibly, none at all (10,14), or 2) potential backcrossing of TcVI into TcIII. Genetic exchange has not been described among hybrid DTUs, but it might be expected to be more permissive between closely related strains (14). We also isolated hybrid AACf2 cl11 from a dog. T. cruzi hybridization has been proposed to arise within mammalian cells (14), and mixed infections in such hosts are common. Alternatively, TcV and TcVI may have evolved from the beneficiaries of different alleles during a single hybridization event between heterozygous parents with mixed TcIII-type mitochondrial complements; although, to date, reported levels of mitochondrial heteroplasmy in T. cruzi are low (10).

Our understanding of the geographic and ecologic distribution of T. cruzi DTUs is changing because of parallel improvements in sampling strategies and genotyping techniques. Human Chagas disease in Colombia is currently associated with DTUs TcI, TcII (to a lesser extent), and oral outbreaks of TcIV (5). In this study, we isolated T. cruzi hybrids from 3 domestic triatomine vectors, a peridomestic dog, and congenital infections among local patients. Given that no reservoir hosts of TcV and TcVI have been described (15), the hybrids from Colombia are more likely the result of long-range anthropogenic introduction than local sylvatic invasion, especially considering the successful establishment of these DTUs among domestic infections in the Southern Cone. Further intensive sampling efforts in northern South America are warranted to elucidate the transmission cycle ecology of TcVI and to accurately assess the epidemiologic risk of human Chagas disease associated with this low-diversity hybrid lineage.