Comparison of two molecular barcodes for the study of equine strongylid communities with amplicon sequencing

Abstract

Basic knowledge on the biology and epidemiology of equine strongylid species still needs to be improved to contribute to the design of better parasite control strategies. Nemabiome metabarcoding is a convenient tool to quantify and identify species in bulk samples that could overcome the hurdle that cyathostomin morphological identification represents. To date, this approach has relied on the internal transcribed spacer 2 (ITS-2) of the ribosomal RNA gene, with a limited investigation of its predictive performance for cyathostomin communities. Using DNA pools of single cyathostomin worms, this study aimed to provide the first elements to compare performances of the ITS-2 and a cytochrome c oxidase subunit I (COI) barcode newly developed in this study. Barcode predictive abilities were compared across various mock community compositions of two, five and 11 individuals from distinct species. The amplification bias of each barcode was estimated. Results were also compared between various types of biological samples, i.e., eggs, infective larvae or adults. Bioinformatic parameters were chosen to yield the closest representation of the cyathostomin community for each barcode, underscoring the need for communities of known composition for metabarcoding purposes. Overall, the proposed COI barcode was suboptimal relative to the ITS-2 rDNA region, because of PCR amplification biases, reduced sensitivity and higher divergence from the expected community composition. Metabarcoding yielded consistent community composition across the three sample types. However, imperfect correlations were found between relative abundances from infective larvae and other life-stages for Cylicostephanus species using the ITS-2 barcode. While the results remain limited by the considered biological material, they suggest that additional improvements are needed for both the ITS-2 and COI barcodes.

Keywords: Cyathostomin; Cytochrome c oxidase subunit I; Horse; Internal transcribed spacer 2; Mitochondrial; Nematode; Parasite; Ribosomal; Strongylid.

Figure 1. Comparison of the predictive abilities of cyathostomin community structure for the mitochondrial COI and ITS-2 rDNA barcodes.

Considered coefficient values are represented across three mock community sizes for the mitochondrial COI (blue) and ITS-2 rDNA (yellow) barcodes. F1-score corresponds to the trade-off between identifying true positives while minimizing the false discovery rate (A). Divergence was computed as the Bray-Curtis (species relative abundances); (B) between the expected and observed mock community composition. Differences between observed and expected alpha diversity (Shannon’s index) are given in (C). (D) Depicts the fraction of reads with no taxonomy assigned (note that because of the stringency of the mapping procedure for the COI barcode, the rate of taxonomy assignment is inflated).

Figure 2. Species-specific amplification efficiencies of the COI and ITS-2 barcodes for 11 cyathostomin species.

The amplification efficiency (in %) derived from qPCR is plotted for each species of interest. Each dot represents a single worm. Worms from the first experiment corresponds to the same batch as those used for metabarcoding sequencing, while the second experiment corresponds to a validation set. Dot shape indicates the worm sex.

Figure 3. Impact of the considered life-stage on predicted equine strongylid community diversity.

(A) The relative abundances measured in the strongylid communities from six horses using the COI or ITS-2 barcodes applied to either eggs, infective larvae or adult worms. (B and C) The first two axes of a non-linear multidimensional scaling analysis based on Bray-Curtis dissimilarity for the COI and ITS-2 rDNA region respectively. Unclassified sequences were not visualized for the ITS-2 barcode but accounted for 2.01% of read counts on average (range between 0% and 4.28%). Because of the outlying community structure found for ITS-2 on the larval samples of the W734 and W748 ponies, the NMDS was applied on only four individuals.