Genomic Analysis of Colombian Leishmania panamensis strains with different level of virulence

Abstract

The establishment of Leishmania infection in mammalian hosts and the subsequent manifestation of clinical symptoms require internalization into macrophages, immune evasion and parasite survival and replication. Although many of the genes involved in these processes have been described, the genetic and genomic variability associated to differences in virulence is largely unknown. Here we present the genomic variation of four Leishmania (Viannia) panamensis strains exhibiting different levels of virulence in BALB/c mice and its application to predict novel genes related to virulence. De novo DNA sequencing and assembly of the most virulent strain allowed comparative genomics analysis with sequenced L. (Viannia) panamensis and L. (Viannia) braziliensis strains, and showed important variations at intra and interspecific levels. Moreover, the mutation detection and a CNV search revealed both base and structural genomic variation within the species. Interestingly, we found differences in the copy number and protein diversity of some genes previously related to virulence. Several machine-learning approaches were applied to combine previous knowledge with features derived from genomic variation and predict a curated set of 66 novel genes related to virulence. These genes can be prioritized for validation experiments and could potentially become promising drug and immune targets for the development of novel prophylactic and therapeutic interventions.

Figure 1

Virulence of four L. panamensis strains in BALB/c mice. BALB/c mice were infected as described in material and methods and the size of the lesion was determined weekly (A) or at the 8^th week post-infection and reported in individually (B). The severity score was also monitored weekly (C) or 8 weeks post-infection in individual mice (D). Representative photographs of the infected ears at the end of the experiment (8^th week post-infection) are presented for each experimental group (E). Parasitic loads records (F). Graphs show the mean+/−SEM (A,B,C), the median (D) or the geometric mean +/− 95% CI (F).

Figure 2

Functional annotation and synteny with L. braziliensis and L. panamensis genomes. (A) Distribution of ontologies of genes annotated in the UA946 assembly. (B) Structural comparison between the assemblies of the L. panamensis strains PSC-1 and UA946. (C) Structural comparison between the assembly of the L. braziliensis strain M2904 and the L. panamensis strain UA946.

Figure 3

Comparison of ploidy between the four strains of L. panamensis with different virulence and reference genomes previously reported for the subgenus Viannia. (A) Normalized median read depth coverage using UA946 genome as reference. (B) Distribution of relative allele dosage among the six strains predicted from the relative allele counts at each site in the reference genome having base calls for at least two observed base pairs. *Reference genome of L. braziliensis reported. ⁺Reference genome of PSC-1 reported.

Figure 4

Analyses of duplications and copy number variation within L. panamensis genomes. (A) Percentage of each chromosome within each strain covered by predicted CNVs in which the copy number value (CN) is greater than 2 (e.g. duplications). (B) Distribution of CNVs by number of different observed CN values over the 5 strains for the complete CNV dataset and for the CNVs in non repetitive (NR) regions. (C) Distribution of predicted CN values on each strain for the complete CNVs dataset and of CNVs in non-repetitive (NR) regions. CNVs are classified by number of different CN values as fixed (only one value observed across the 5 samples) or non-fixed. (D) Differences in CN values among the 4 strains of L. panamensis evaluated with different virulence. *Genes involved in virulence or up-regulated in the amastigote state reported by different authors (Supplementary Table S1). The colours of the heatmap and the dendrogram were included with the purpose of highlighting the differences and similarity in CNVs among the strains analyzed. Reads of PSC-1 strain were included for comparison purposes.

Figure 5

(A) Neighbor Joining dendrogram of L. panamensis strains based in Kimura 2-parameter distances (scale at bottom) expressed in units of number of base substitutions per site. The tree is drawn to scale with branch lengths, which is in the same units as those of the evolutionary distances inferred. (B) Distribution of Ka/Ks values estimated from SNPs in protein coding regions. Blue bars represent Ka/Ks values predicted from substitutions between L. braziliensis and L. panamensis. Red bars are Ka/Ks values predicted from polymorphisms within L. panamensis. (C) Gene ontology analysis at level 3 of the 81 hypothetical proteins possibly implicated in virulence according to machine learning techniques. Green: cellular component; red: molecular function; and blue, biological process. (D) Proteins identified in the annotation of the 81 hypothetical proteins that are part of 230 possible new genes associated to virulence (Blast2GO program).