Whole genome profiling of spontaneous and chemically induced mutations in Toxoplasma gondii

Abstract

Background: Next generation sequencing is helping to overcome limitations in organisms less accessible to classical or reverse genetic methods by facilitating whole genome mutational analysis studies. One traditionally intractable group, the Apicomplexa, contains several important pathogenic protozoan parasites, including the Plasmodium species that cause malaria.Here we apply whole genome analysis methods to the relatively accessible model apicomplexan, Toxoplasma gondii, to optimize forward genetic methods for chemical mutagenesis using N-ethyl-N-nitrosourea (ENU) and ethylmethane sulfonate (EMS) at varying dosages.

Results: By comparing three different lab-strains we show that spontaneously generated mutations reflect genome composition, without nucleotide bias. However, the single nucleotide variations (SNVs) are not distributed randomly over the genome; most of these mutations reside either in non-coding sequence or are silent with respect to protein coding. This is in contrast to the random genomic distribution of mutations induced by chemical mutagenesis. Additionally, we report a genome wide transition vs transversion ratio (ti/tv) of 0.91 for spontaneous mutations in Toxoplasma, with a slightly higher rate of 1.20 and 1.06 for variants induced by ENU and EMS respectively. We also show that in the Toxoplasma system, surprisingly, both ENU and EMS have a proclivity for inducing mutations at A/T base pairs (78.6% and 69.6%, respectively).

Conclusions: The number of SNVs between related laboratory strains is relatively low and managed by purifying selection away from changes to amino acid sequence. From an experimental mutagenesis point of view, both ENU (24.7%) and EMS (29.1%) are more likely to generate variation within exons than would naturally accumulate over time in culture (19.1%), demonstrating the utility of these approaches for yielding proportionally greater changes to the amino acid sequence. These results will not only direct the methods of future chemical mutagenesis in Toxoplasma, but also aid in designing forward genetic approaches in less accessible pathogenic protozoa as well.

Figure 1

Lineage overview of the parasite strains used in this study. All are of the Type I genotype, with GT1 the sequenced reference genome available on ToxoDB.org [48]. The GT1 and RH strains are most likely descendants from the same clone [49]. All sequenced strains and mutants are outlined by red boxes; all are derived from the RH strain. Drug selections to generate transgenic lines are given between brackets in the lower box of each line. Mutagen dosage is expressed as the percentages of parasites being killed by mutagen treatment (measured by plaque assays). Phle: phleomycin; D-HR: double homologous recombination; 6-TX: 6-thioxanthine; Chl: chloramphenicol; FUDR: 5-fluoro-2'-deoxyuridine; HXGPRT: hypoxanthine-xanthine-guanine-phosphoribosyl transferase.

Figure 2

Genomic coverage and quality of sequencing reads. Average fold genomic coverage for the various strains is plotted as indicated. Bar color reflects the Illumina read length used as indicated in the legend. In the Table in the lower half the percentage of the genome covered at least 10-fold is shown as percentage of the complete GT1 reference genome (in all cases over 99.5%). Read quality is reported as the mismatch rate between reads and the reference genome. n-F-P2 [n = new] is a re-sequenced sample of the F-P2 mutant with longer read length [7], named o-F-P2 [o = old].

Figure 3

Comparison of non-mutagenized parent strains: 2F-1-YFP2, Blader RH-ΔHGXPRT (B-RH) and Gubbels RH-ΔHGXPRT (G-RH). (A) Venn diagram of shared and unique SNVs between the three strains and the GT1 reference genome. (B) The incidence of various mutations causing changes in amino acid coding. For 2F-1-YFP2, B-RH, and G-RH the unique SNVs vs GT1 are shown (these correspond to the 54, 66, and 19 SNVs in panel A, respectively). “RH-srd.” refers to all SNVs shared between B-RH and G-RH and “all-srd.” refers to the SNVs shared between all three lines (these correspond to the and 85 and 984 SNVs in panel A, respectively) Syn.: synonymous; non-syn: non-synonymous. * 50 non-coding SNVs map to the apicoplast genome. (C-E) The incidence of the various base pair changes (SNVs) in the mutants and groups as indicated. Note that we used the n-F-P2 reads instead of 2F-1-YFP2 reads in this analysis since the quality is much higher (Figure  2); ENU specific n-F-P2 mutations were removed from the comparative analysis.

Figure 4

ENU generated mutants. (A) The incidence of various SNVs in protein coding regions versus those outside annotated open reading frames. Syn.: synonymous; non-syn: non-synonymous. The number between brackets is the percentage of mutations in each category. (B) The incidence of SNVs across all seven ENU mutants in panel A. Mutant F-P2 was generated using a 55% killing dose; for all others a dose inducing 70% killing was used. Mutants F-P2, AX-H9, CF-B19, and FE-N3 were generated by Gubbels; mutants SBR1-3 were generated by the Blader lab.

Figure 5

EMS generated mutants. (A) The incidence of various SNVs in protein coding regions versus those outside annotated open reading frames. Syn.: synonymous; non-syn: non-synonymous. The number between brackets is the percentage of mutations in each category. (B) The incidence of SNVs across the EMS mutants screened for resistance to pharmacologically induced egress using DTT (E2D2 and E4D5) or invasion enhancing compound 2 (E3E2) [42, 43]. These three mutants were generated using an EMS dosage inducing 70% killing. (C) The incidence of SNVs across the EMS mutants screened for resistance against 20 μM FUDR [15]. EMS7.5 and EMS10 were generated using 7.5 and 10 mM EMS, inducing 80% and 90% killing, respectively.

Figure 6

FUDR resistant mutants. (A) The resistance of Toxoplasma strains to varying concentrations of FUDR. Survival was determined by plaque formation relative to an untreated control. Error bars represent SD from 6 independent experiments. Both strains were initially selected for survival in the presence of 20 μM FUDR. (B) Proposed etiological mutation in UPRT identified in mutant EMS7.5. Genbank accession numbers for HsUPRT and DmUPRT are NP_659489 and CG5537, respectively.

Figure 7

Genome wide distribution of mutations. (A) Circos plot of all mutants and gene density. The Toxoplasma GT1 reference genome is composed of the 14 chromosomes displayed in the outermost circle. The SNVs shared between all 3 RH parent lines and the GT1 reference are represented by green dots. Strain and mutant specific variations are represented by black dots and are grouped as either parent lines (yellow) or by mutant genealogy and mutagen as shown in Figure  1. Gene density was mapped using a sliding window approach with a window size of 100 kb. (B-D) QQ-plot of SNV distances vs. exponential distribution. To test the assumption that mutations are distributed randomly, the inter-SNV distances (taken from Additional file 2: Table S2) are plotted against the exponential distribution, with λ = 1/mean, using a Quantile-Quantile plot; B = distances between ENU induced SNVs (N = 368, mean = 159582), C = distances between EMS induced SNVs (N = 144, mean = 359992.3), D = distances between shared SNVs between the parent strains (N = 968, mean = 61631.86). The solid red line in each plot represents the null assumption y = x.