Clinical Trypanosoma cruzi isolates share a common antigen repertoire that is absent from culture adapted strains

Abstract

Background: Trypanosoma cruzi causes Chagas disease, a poorly understood and clinically heterogeneous disease. Recent work has demonstrated that parasites adapted to laboratory conditions are genomically variable, but little is known of the extent of genomic diversity from clinically isolated specimens.

Methods: In this retrospective observational genomic study, we isolated 15 T. cruzi specimens from three clinical studies of Chagas disease, representing different clinical contexts. We sequenced the genome of each strain and used single nucleotide variant (SNV) based analyses to estimate parasite genetic lineage, genomic population structure, regions of copy number plasticity, and to identify gene conversion events. In addition, we generated and annotated whole genome assemblies of each isolate. From these assemblies, we compared the repertoires of genes encoding for highly virulent and variable proteins that have been implicated in disease pathogenesis.

Findings: We identified parasites from two genetic lineages in this collection of clinical isolates. Our analysis revealed evidence of genomic instability. Diversity-generating copy number variation was statistically enriched in regions encoding the virulence-associated multigene families, while diversity-eliminating gene conversion events were enriched in regions depleted of multigene family members. We also discovered a set of multigene family members that is present in all of the clinically isolated parasite genomes and absent from all of the lab adapted strains, regardless of parasite lineage. Multigene family repertoires were more conserved among field isolated specimens of the same genetic lineage than among culture adapted strains of the same genetic type.

Interpretation: This study provides whole genome sequencing data for TcV parasites isolated from naturally infected human patients with Chagas disease for the first time. Our analysis of these genomes revealed substantial genomic instability, suggesting the parasite undergoes genomic change in response to the pressures imposed by the host environment. Moreover, we observed a set of virulence-associated genes that are present exclusively within clinical isolates and absent from lab-adapted strains, indicating a potential role for these genes in parasite survival in natural hosts. These findings highlight the limitations of genetic studies focused exclusively on lab-adapted parasite strains and provide insight into the genomic features of T. cruzi that are likely to be important for clinical infection.

Figure 1:. 15 clinical isolates from different infection contexts in Bolivia display underlying genetic variability

A) Jaccard similarity matrix of kmers across assembled genomes. Yellow is most dissimilar and blue is the most similar in kmer overlap. Hierarchical clustering of the data from the Jaccard matrix is shown in the dendrogram. B) PCA of allele distances among genes in hybrid genomes.

Figure 2:. Variable copy number across genomic regions and patterns of homozygosity suggests genomic diversification.

A) Estimated copy number variation for 5 kb windows in clinically isolated parasite genomes. Each row is a 5kb window colored by loss (red colors) or gain (green) of genomic content at that site. Copy number gains are censored at 15. B) Sum of sites in each genome with exactly two alleles in each 5kb window. White indicates regions with very few biallelic sites, indicating homozygosity, while blue then red indicates an increase of biallelic sites. Windows are censored at biallelic sites = 300. Grey boxes are masked regions that have less than two estimated copies, which would by definition have one copy of the allele. C) Histogram showing the distribution windows with a given number of biallelic sites. The heterozygous peak is largely at 100 biallelic sites per 5kb, or 2%, and therefore a minimum threshold of 1% of biallelic sites is set for calling a window heterozygous. The colors correspond to figure 2B. D) Heterozygosity concordance among TcV samples. For each window, the number of samples where that window is largely heterozygous (greater than 1% of sites in the window are biallelic) was calculated. The number of windows that are heterozygous in one out of 13 total samples TcV samples, up to 12 out of 13 TcV samples is shown. The majority of windows had perfectly concordant heterozygosity (n = 1035) or concordant homozygosity (n = 2985). The sites without perfect concordance in heterozygosity in all TcV samples are marked as possible gene conversion events, in blue. E. The odds ratio of a window in each stratum defined in figure 2D to have MGF members. Windows with 2, 6 and 9 samples sharing heterozygosity had zero observed MGFs and thus the CI reaches 0. Asterisks indicate ORs that are statistically significant after correction via Benjamini-Hochberg Procedure. ORs and CIs are available in Supplemental Table 3.

Figure 3:. Multigene family repertoires in clinically isolated strains are distinct from the repertoires of strains adapted to lab culture.

A) Multidimensional scaling plot of the Jaccard distance between samples’ MGF membership visualizing the presence or absence of a member of a gene cluster within each sample. Color of points indicates study, and shape indicates parasite genetic type. B) MGF clusters that are found in all clinical isolates, regardless of DTU. Columns are strains, and those indicated with yellow bars are lab adapted genomes. Blue cells are gene clusters that were present in the genome, grey cells were absent. C) Histogram of permuted set difference, establishing the null distribution for set difference between the clinical isolates and lab adapted TcII, TcV, and TcVI strains. The dotted red line represents the observed set difference between the sets of MGF clusters found in all clinical isolates and the set of MGF clusters found in none of the lab adapted TcII, TcV or TcVI strains.

Figure 4:. MGF repertoires are more conserved in recent field isolates compared to laboratory adapted strains, regardless of DTU.

Proportion of the MGF repertoire that is shared across DTU and isolation recency. For each MGF, the number of field isolated TcI, lab adapted TcI, field isolated hybrid, or lab adapted hybrid strains that each gene cluster is found in is graphed in a proportional bar graph.