Metatranscriptomics yields new genomic resources and sensitive detection of infections for diverse blood parasites

Abstract

Metatranscriptomics is a powerful method for studying the composition and function of complex microbial communities. The application of metatranscriptomics to multispecies parasite infections is of particular interest, as research on parasite evolution and diversification has been hampered by technical challenges to genome-scale DNA sequencing. In particular, blood parasites of vertebrates are abundant and diverse although they often occur at low infection intensities and exist as multispecies infections, rendering the isolation of genomic sequence data challenging. Here, we use birds and their diverse haemosporidian parasites to illustrate the potential for metatranscriptome sequencing to generate large quantities of genome-wide sequence data from multiple blood parasite species simultaneously. We used RNA-sequencing of 24 blood samples from songbirds in North America to show that metatranscriptomes can yield large proportions of haemosporidian protein-coding gene repertoires even when infections are of low intensity (<0.1% red blood cells infected) and consist of multiple parasite taxa. By bioinformatically separating host and parasite transcripts and assigning them to the haemosporidian genus of origin, we found that transcriptomes detected ~23% more total parasite infections across all samples than were identified using microscopy and DNA barcoding. For single-species infections, we obtained data for >1,300 loci from samples with as low as 0.03% parasitaemia, with the number of loci increasing with infection intensity. In total, we provide data for 1,502 single-copy orthologous loci from a phylogenetically diverse set of 33 haemosporidian mitochondrial lineages. The metatranscriptomic approach described here has the potential to accelerate ecological and evolutionary research on haemosporidians and other diverse parasites.

Keywords: Leucocytozoon; Parahaemoproteus; Plasmodium; RNA-seq; co-infection; malaria parasite.

Figure 1.

Genus assignment based on a quartet resampling strategy. (A) For each transcriptome contig of unknown origin (green), one reference sequence from each genus (Plasmodium: orange, Parahaemoproteus: purple, Leucocytozoon: blue) is drawn from the same orthogroup alignment. (B) Sequence quartets with less than 150 bp of overlap among all four sequences are discarded. (C) If the contig has best hits (arrows) against different reference sequences in the amino acid and the nucleotide alignments, the quartet is considered ambiguous and no genus assignment is made. (D) If the nucleotide and the amino acid data result in a best hit against the same reference sequence and if at least one best hit is bi-directional (double headed arrow), the quartet is counted as positive for that genus. This procedure is executed for all possible quartet combinations within an orthogroup. Contigs with unambiguous hits against the same genus in the majority of quartets are assigned to that genus. Genus assignment was only conducted using orthogroup alignments with a minimum of two sequences for each reference genus.

Figure 2.

The phylogenetic diversity of haemosporidian parasites detected in host samples selected for transcriptome sequencing. Shown are cytb phylogenies for the genera (A) Leucocytozoon, (B) Parahaemoproteus, and (C) Plasmodium, generated from all cytb haplotypes available on the avian haemosporidian database MalAvi as of April 2019 (Bensch et al. 2009). Depicted on each phylogeny are the names of the cytb lineages (determined by cytb barcoding) that were included for metatranscriptome sequencing or were used as references. Names in red indicate lineages that were used as references for the genus assignment procedure that were obtained from previously published research, and names in blue indicate lineages that were included in the reference database and were sequenced for this study. Names in black represent lineages that were sequenced in this study and were found in mixed-genus samples, and so were used as input for genus assignment.

Figure 3.

Samples that contained different parasite genera and infection intensities differed in the composition of the orthogroups for which data was generated. A) Multiple correspondence analysis (MCA) showing that samples with Parahaemoproteus infections of >0.5% parasitemia cluster together to the exclusion of samples with low-intensity (<0.5%) Parahaemoproteus or Leucocytozoon infections. B) Venn diagrams demonstrating overlap in the number of orthogroups for which data was generated for sample groups defined by parasite genus and infection intensity.

Figure 4.

Relationship between parasitemia and number of haemosporidian transcripts. Samples that generated a higher number of haemosporidian transcripts tended to have more intense infections, contain infections from a greater number of parasite species, and had higher sequencing depth. The number of haemosporidian transcripts shown here refers to the number of transcripts for each sample in the ‘all-orthologs’ dataset.