Molecular dissection of laboratory contamination between two schistosome populations

Abstract

Background: Genomic analysis has revealed extensive contamination among laboratory-maintained microbes including malaria parasites, Mycobacterium tuberculosis, and Salmonella spp. Here, we provide direct evidence for recent contamination of a laboratory schistosome parasite population, and we investigate its genomic consequences. The Brazilian Schistosoma mansoni population SmBRE has several distinctive phenotypes, showing poor infectivity, reduced sporocyst number, low levels of cercarial shedding and low virulence in the intermediate snail host, and low worm burden and low fecundity in the vertebrate rodent host. In 2021 we observed a rapid change in SmBRE parasite phenotypes, with a 10-fold increase in cercarial production and fourfold increase in worm burden.

Methods: To determine the underlying genomic cause of these changes, we sequenced pools of SmBRE adults collected during parasite maintenance between 2015 and 2023. We also sequenced another parasite population (SmLE) maintained alongside SmBRE without phenotypic changes.

Results: While SmLE allele frequencies remained stable over the 8-year period, we observed sudden changes in allele frequency across the genome in SmBRE between July 2021 and February 2023, consistent with expectations of laboratory contamination. (i) SmLE-specific alleles increased in the SmBRE population from 0 to 41-46% across the genome between September and October 2021, reflecting the timing and magnitude of the contamination event. (ii) After contamination, strong selection (s ≅0.23) drove the replacement of low-fitness SmBRE with high-fitness SmLE alleles. (iii) Allele frequency changed rapidly across the whole genome, except for a region on chromosome 4, where SmBRE alleles remained at high frequency.

Conclusions: We were able to detect contamination in this case because SmBRE shows distinctive phenotypes. However, this would likely have been missed with phenotypically similar parasites. These results provide a cautionary tale about the importance of tracking the identity of parasite populations, but also showcase a simple approach to monitor changes within populations using molecular profiling of pooled population samples to characterize single-nucleotide polymorphisms. We also show that genetic drift results in continuous change even in the absence of contamination, causing parasites maintained in different labs (or sampled from the same lab at different times) to diverge.

Keywords: Schistosoma mansoni; Contamination; Laboratory populations; Parasite; Pool-sequencing; Population genomics; SmBRE; SmLE.

Fig. 1

Phenotypic differences between SmBRE and SmLE. A Boxplots showing cercarial shedding from infected snails (N = 240), measured in 2015 (data from Le Clec’h et al. [17]) and 2023 over 4 weeks of the patent period (4–7 weeks post-infection of snails). Statistical comparisons between years were conducted using a Wilcoxon rank-sum test and adjusted for multiple comparisons (Benjamini–Hochberg). B Boxplots showing worm burden normalized by the number of cercariae used for hamster infection in 2015 (N = 15) and 2023 (N = 6). Statistical comparison between years were conducted with Student’s t-test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

Fig. 2

Differentiation between SmBRE and SmLE across time between 2016 and 2023. Dot plot showing smoothed average F_ST across the whole genome calculated in 20-kb windows. The solid lines indicate F_ST after smoothing with a local regression model as calculated by the locfit R package

Fig. 3

Differentiation in SmBRE and SmLE across time in comparison to 2016. Line plot showing average F_ST across the genome for each time point in comparison to pools sampled in 2016. Missing data was removed

Fig. 4

SmLE-specific allele frequencies in SmBRE pools. A Line plot showing mean allele frequency of SmLE-specific variants per chromosome and across time. The numbers next to the last data point represent the mean allele frequency. B Natural log of the genotype ratio plotted against sexual life cycles. The selection coefficient was estimated as the slope of the least-squares fit. The genotype ratio was calculated as the average genome-wide frequency of SmBRE alleles/average genome-wide frequency of SmLE alleles at each time point after the initial contamination event. C Selection coefficient (s) for individual SNPs across the whole genome. A local regression smooth line is shown in red

Fig. 5

Observed differentiation in SmLE parasites over time. A Line plot showing allele frequency change over time in specific variants in the SmLE population. Variants are labeled by chromosome and position. B Histogram illustrating the distribution of F_ST values from the comparison of 657,592 variants in female pools and 661,996 variants in male pools from 2016 with those from 2023. C Distribution of allele frequency change in the same variants between 2016 and 2023