Development of primers for the mitochondrial cytochrome c oxidase I gene in digenetic trematodes (Platyhelminthes) illustrates the challenge of barcoding parasitic helminths



    1. Department of Integrative Biology, University of Guelph, Guelph, ON, Canada N1G 2W1,
    2. Department of Neurosciences, University of Toledo, Toledo, OH 43614-2598, USA,
    Search for more papers by this author

    1. Department of Biology, Concordia University, 7141 Sherbrooke Street West, Montreal, QC, Canada H4B 1R6,
    Search for more papers by this author

    1. Department of Biology, Concordia University, 7141 Sherbrooke Street West, Montreal, QC, Canada H4B 1R6,
    Search for more papers by this author

    1. Fluvial Ecosystem Research Section, Aquatic Ecosystem Protection Research Division, Water Science and Technology Directorate, Science and Technology Branch, Environment Canada, St. Lawrence Centre, 105 McGill, 7th Floor, Montreal, QC, Canada H2Y 2E7
    Search for more papers by this author

    1. Department of Integrative Biology, University of Guelph, Guelph, ON, Canada N1G 2W1,
    Search for more papers by this author

Teresa J. Crease, Fax: 519-767-1656; E-mail:


The phylum Platyhelminthes is a diverse group of flatworms that includes parasites with serious impacts on human health, animal husbandry, aquaculture and wildlife management. Here we present degenerate primers for the barcode region of the mitochondrial cytochrome c oxidase I (COI) gene in flatworms. Although amplicons were obtained from a wide taxonomic range in the Cestoda and Trematoda, COI fragments from many taxa in these classes did not amplify. Primers specific to trematodes in the family Diplostomidae were also developed. Amplification success was much higher with diplostomid-specific primers and sequences were obtained from 504 of 585 specimens of Diplostomum and Tylodelphys. Sequences from the barcode region resolved all specimens to the species level, with mean divergence between congeners of 19% (3.9–25%). Because many of our specimens were small, we initially amplified part of the nuclear small subunit ribosomal (r) RNA gene to evaluate the quality and quantity of DNA in our specimens. Short sequences (~380 nt) of this gene were recovered from most specimens and can be used to distinguish specimens at the family level and often the generic level. We suggest that rRNA genes could be used to screen samples of completely unknown taxonomy, after which specific COI primers could be used to obtain species-level identifications.


The digenetic trematodes (Platyhelminthes) comprise an estimated 24 000 species, many of which have yet to be described (Poulin & Morand 2004). Adults parasitize vertebrates and larval stages typically require a mollusk, usually a snail, as the first intermediate host. Most species also require a second intermediate host, which may be an invertebrate or a vertebrate, depending upon the species.

Some digeneans, such as Schistosoma spp. and Clonorchis sinensis, are important pathogens in humans, while many others have serious impacts on animal husbandry, aquaculture and wildlife management (Roberts & Janovy 2000). Accurate identification of these parasites is important for the diagnosis, treatment and control of pathogenic infections. It is also essential in broader studies relating to digenean diversity, distribution and ecology.

Species-level identification of most digeneans is based exclusively on adult morphology. Difficulties arise because they are small, soft-bodied, have few stable morphological characters and are subject to host-induced phenotypic variation (Graczyk 1991). These problems are particularly acute for larval stages, which differ morphologically from the adults and, with fewer morphological features than adults, are virtually impossible to identify to species. Except for a few well-studied digeneans, larval stages can only be identified to the species level by experimental completion of the life cycle and subsequent study of the adult specimens. This is seldom a practical option and in most cases the identity of digenean larval stages can, at best, only be resolved to the generic level. In this context, molecular markers offer powerful and much-needed tools that have the potential to distinguish between morphologically similar species at any stage in their life cycle.

Many of the difficulties associated with identification of digenean species also apply to other parasitic helminths. Sampling adult helminths usually requires postmortem examination of the host. However, it is also possible to obtain eggs, larvae, or segments (cestodes) in host excreta and many parasites (digeneans, nematodes) have free-living larval stages that can be collected directly from the environment. In these instances, it can be challenging to determine even higher order classification, particularly for the nonexpert. Here again, molecular identification systems hold much promise.

There is ample evidence that sequences from the 5′ end of the cytochrome c oxidase I gene (COI), that is, the DNA barcode, can be used to identify species across a broad taxonomic range (Hebert et al. 2003, 2004; Smith et al. 2005, 2007; Ward et al. 2005; Cywinska et al. 2006; Hajibabaei et al. 2006; Saunders 2008). To date, most studies employing molecular markers to distinguish digenean species have used the internal transcribed spacer (ITS) regions of ribosomal (r) DNA (Nolan & Cribb 2005; Olson & Tkach 2005). However, the few studies that did use COI sequences have shown that they distinguish congeneric digeneans more clearly than does the ITS region (e.g. Bowles et al. 1995; Morgan & Blair 1998; Vilas et al. 2005). Most of the digenean COI sequences used to date lie downstream of the barcode region (e.g. Bowles et al. 1995; Morgan & Blair 1998; Morgan et al. 2005). However, it seems likely that sequences from the upstream barcode fragment may also provide a useful method for interspecific differentiation in this group.

A principal advantage of the barcoding approach is that use of a standardized marker, a ~600-nt fragment at the 5′ end of COI, ensures that sequence data are comparable across studies. A prerequisite to acquiring these data is the development of primers that amplify this region in the broadest possible range of taxa, thus allowing samples of unknown taxonomic affinity to be identified to species. The most widely applicable primers used in barcoding, those of Folmer et al. (1994), are very divergent from many of the published platyhelminth COI sequences (T. Crease, personal observation). Herein, we present the preliminary results of efforts to design primers that will recover barcode sequences in diverse platyhelminths, with particular focus on the Diplostomoidea. In addition, we assess the usefulness of sequences from a small region of the nuclear 18S rRNA gene as a preliminary screening tool in the barcoding of platyhelminth parasites.

Materials and methods

Primer development

Our initial aim was to design primers that would amplify a fragment corresponding to the barcode region at the 5′ end of the COI gene in the broadest possible range of platyhelminth taxa. We aligned the barcode region of COI from complete platyhelminth mitochondrial genome sequences available from GenBank in 2006 to identify regions suitable for primer development. Seventeen sequences were aligned from representatives of two cestode families, three digenean families and one turbellarian family (AF216697, M93388, AF216697, AF540958, AF216698, DQ157223, DQ157222, AF445798, AB107234, AY195858, AB107242, DQ089663, AF216699, AF297617, AF346403, AB049114, AF314223). Based on these sequences, we designed two degenerate primers for the barcode region of COI, MplatCOX1dF and MplatCOX1dR (Table 1) and attached 5′ M13 tails so that amplicons could be sequenced with M13 primers. The degenerate forward primer ends 8 nt upstream of the Folmer A primer, and the degenerate reverse primer ends 22 nt downstream of the Folmer B primer, so these primers amplify an additional 30 nucleotides compared to the Folmer primers.

Table 1.  PCR primers used to amplify mitochondrial COI and nuclear 18S rRNA gene fragments from samples in the Platyhelminthes. The M13 tails [M13(–21)F and M13(–27)R] at the 5′ ends of the degenerate COI primers (MplatCOX1d) are underlined. The Plat-diploCOX1 primers were designed to amplify members of the family Diplostomidae
Primer namePrimer sequence 5′–3′Approximate product size (nt)Region amplified

Using sequences obtained with the degenerate primers, we designed a set of primers specific to the family Diplostomatidae, Plat-diploCOX1F and Plat-diploCOX1R (Table 1). The forward primer ends 92 nt downstream of the degenerate forward primer, and the reverse primer ends 66 nt upstream of the degenerate reverse primer, so these primers amplify 158 fewer nucleotides than our degenerate primers. Diplostomid-specific primers were developed for two reasons: first, the degenerate primers did not yield sequences in the majority of these specimens, and second, most specimens examined in this study belong to this family.

Specimen collection, polymerase chain reaction and sequencing

The majority of specimens used in this study came from various avian, amphibian and piscine hosts collected in the Saint Lawrence River basin in Quebec, Canada. Most samples were larval digeneans from fish hosts caught and frozen in 2006. Specimens were identified using morphological characters to the lowest possible taxonomic level, which was generally to genus in the case of larval specimens, and then stored in 95% ethanol. DNA was extracted using either a glass-fibre extraction protocol (Ivanova et al. 2006), a QIAGEN DNeasy Extraction Kit, or a chloroform-isoamyl DNA extraction protocol (modified from Sambrook & Russell 2001).

We attempted to amplify COI from 571 digenean and 20 cestode specimens using the degenerate primers, and from 613 digeneans using the diplostomid-specific primers (Table S1, Supporting information). As most of the specimens were small, we suspected low quantities of DNA might affect polymerase chain reaction (PCR) success rates (Ivanova et al. 2006). Therefore, to assess the quantity of DNA in initial samples, a fragment of the nuclear 18S rRNA gene from 90 digeneans and 20 cestodes was amplified and sequenced using both novel and previously published primers (Table 1). For comparative purposes, we also attempted to amplify and sequence the ITS1 + 5.8S + ITS2 region of nuclear rDNA in 102 digenean specimens using previously published primers (Table 1).

All PCRs had a total volume of 25 µL and included 1× PCR buffer (20 mm Tris-HCl pH 8.4, 50 mm KCl), 2.5 mm MgCl2, 1.25 pmol of each primer (see Table 1), 50 µm of each dNTP, 0.6 U of Platinum Taq Polymerase (Invitrogen) and approximately 5 ng (18S PCR) to 50 ng (COI PCR) of DNA template. PCR conditions were 94 °C for 2 min, 35 cycles of 94 °C for 30 s, 50 °C for 30 s, and 72 °C for 1 min, with a final extension at 72 °C for 10 min. Amplicons were visualized on 1% TAE agarose gels stained with ethidium bromide after which 1 µL was sequenced in a 12-µL reaction using 0.5 µL of the BigDye Terminator version 3.1 Cycle Sequencing mix (Applied Biosystems) and 10 pmol of primer. Amplicons generated with the degenerate COI primers were sequenced with the M13(–21)F and M13(–27)R primers, while those generated with the diplostomid-specific, 18S and ITS primers were sequenced with the primers used in the PCRs (Table 1). The 18S amplicons were only sequenced in one direction, with the 18S9F primer. Sequencing reactions were analysed on an ABI 3730 capillary sequencer (Applied Biosystems) in the Genomics Facility or the Canadian Centre for DNA Barcoding, both at the University of Guelph.

Sequences were assembled in Sequencher version 4.5 (Gene Codes Corporation) and manually edited. The aligned sequences were imported into mega 3.1 (Kumar et al. 2004) where pairwise sequence divergence estimates were generated using the Kimura 2-parameter model with pairwise deletion. The neighbour-joining algorithm (NJ) was then used to generate a phenogram from the resulting matrix of sequence divergence values.

Some specimens analysed in this study were adult trematodes that were identified to species using morphological characters. For example, specimens of Diplostomum baeri and Diplostomum indistinctum were identified by Galazzo et al. (2002). However, most specimens were larval trematodes that could only be identified to genus. In these cases, species were distinguished based on sequence divergence levels and NJ phenograms. Many species distinguished in this way showed narrow host specificity (data not shown), which constitutes independent, additional evidence of their correspondence to species (Roberts & Janovy 2000; Poulin & Morand 2004). The divergence of COI sequences and host specificity of larval trematode species presented here will be the topic of a separate publication.



There was no indication that the quantity or quality of DNA in our samples affected the performance of the COI primers in digeneans. Amplicons of the 18S gene were obtained in all 90 trematode samples assayed, including 38 in which COI did not amplify and 58 that failed to yield COI sequences (Table S1). In addition, an estimate of DNA quantity (specimen size) showed no relationship with the probability of sequencing success (R2 = 0.006, P = 0.824). Only 17 of 102 samples yielded ITS sequences.

The degenerate COI primers yielded amplicons of expected length in 314, and sequences in 231 of the 572 digenean samples assayed, including representatives of five genera in the Plagiorchiida, two genera in the Echinostomida, and eight genera in the Strigeida (Table S1). The diplostomid-specific COI primers yielded amplicons in eight of nine strigeidid genera tested, and 504 sequences from 610 samples assayed (Table S1).

Significantly, fewer sequences were obtained with the degenerate COI primers than with the diplostomid-specific primers (χ2 = 289, d.f. = 1, P < 0.0005). In the most intensively sampled taxa, the degenerate COI primers yielded sequences in 27 of 34 (79%) Diplostomum, 65 of 128 (51%) Posthodiplostomum, 47 of 133 (35%) Ornithodiplostomum, 42 of 67 (63%) Apatemon, and 9 of 69 (13%) Ichthyocotylurus samples. Chromatogram trace signals obtained with the degenerate primers were often unclear for the first 100 to 150 nt at the 5′ end. The diplostomid-specific primers performed best on Diplostomum and Tylodelphys, in which informative sequences were obtained in 504 of 585 (86%) specimens assayed.

Overall, potentially informative sequences (> 150 nt in length) of the barcode region of COI were obtained in 706 of the 1138 (62%) digenean specimens with one or both sets of novel primers. Here we report sequences only from single representatives of each species detected. Bi-directionally sequenced fragments range from approximately 100 to 650 nt in length; the mean length is 442 nt.

The mean pairwise divergence between congeneric sequences from the barcode region is 19% (3.9–25%, Fig. 1). Increasing COI sequence length does not affect species-level resolution. The topology of NJ trees based on ~480-nt sequences is identical to those based on fragments that are ~70 nt longer (data not shown).

Figure 1.

Neighbour-joining tree of barcode-region COI sequences for representatives of trematode taxa obtained in the present study. Sequences are from a single representative among multiple specimens analysed, with the exception of Echinostoma sp., Plagioporus sinitsini, Gorgoderina sp. and those marked with an asterisk, in which case a COI sequence was obtained from only a single specimen. Sequences have been deposited in GenBank under accession nos FJ477181–FJ477191 and FJ477193–FJ477223. †Tentative identification.

Few closely related specimens yielded both COI and ITS sequences. Barcode-region COI sequences have diverged by 15% and ITS sequences by 3.9% in two species of Diplostomum (ITS data from Galazzo et al. 2002), while two species of Ornithodiplostomum show 6.9% divergence in COI sequences and 4.2% divergence in ITS sequences (Fig. 2). Short 18S sequences (~380 nt) fail to distinguish two genera (Posthodiplostomum and Ornithodiplostomum) but resolve higher taxa into distinct clusters (Fig. 3).

Figure 2.

Comparison of taxonomic resolution provided by rDNA and COI sequences from specimens in the trematode family Diplostomidae. (a) Neighbour-joining tree of ITS1 + 5.8S + ITS2 rDNA sequences from closely related specimens barcoded in this study. (b) Neighbour-joining tree of barcode-region COI sequences from the same specimens as (a). Sequences of rDNA have been deposited in GenBank under accession nos FJ469594–FJ469596; rDNA sequences from Diplostomum baeri (AY123042) and Diplostomum indistinctum (AY123043) are from Galazzo et al. 2002.

Figure 3.

Neighbour-joining tree generated from partial 18S rRNA gene sequences (380 nt) from representatives of the class Trematoda (subclass Digenea), and the class Cestoda in the phylum Platyhelminthes. Sequences have been deposited in GenBank under accession nos FJ469581–FJ469593. †Tentative identification.


Degenerate COI primers yielded PCR amplicons of expected length in eight of 20 samples comprising representatives of four genera in the Cyclophyllidea. Sequences were obtained from only a single sample (Hymenolepididae: Cloacotaenia megalops, Barcode accession FJ477192; data not shown in figures). Amplification of COI was unsuccessful in 11 other cyclophyllidean genera and in two pseudophyllidean genera (Table S1). PCR amplicons of the 18S gene were generated in all cestode samples, indicating that the quality of DNA templates was unlikely to be the cause of unsuccessful COI amplification.


It was clear from the initial alignment of flatworm COI sequences that the high level of sequence divergence would make it difficult to design primers that would successfully amplify this gene across the entire phylum. Indeed, the noisiness of the upstream portion of sequences acquired with the degenerate primers suggests that short, nontarget amplicons were generated and sequenced along with the primary product. Nonetheless, the taxonomic range of our specimens that did yield amplicons or sequences, despite small sample sizes in many groups (Table S1), suggests that the degenerate primers may be useful for barcoding digeneans. In particular, they can be used to generate preliminary data for the design of barcode-region primers specific to lower taxonomic groups. This approach was fruitful with the Diplostomidae, where amplification and sequencing success was much higher with our family-specific primers. Moreover, as sequences from the barcode region of COI continue to be published for additional platyhelminth taxa (e.g. Park et al. 2007), designing primers specific to lower taxa will become easier.

In other studies using molecular markers to distinguish species, primers specific to groups recalcitrant to more generalist primers have also been employed (e.g. Morgan et al. 2005; Smith et al. 2005; Ward et al. 2005; Hajibabaei et al. 2006; Smith et al. 2007; Zarowiecki et al. 2007; Saunders 2008). This approach requires that samples be partially identified in order to determine which set of primers to use. If only the higher taxonomy of the sample is known, then a ‘cocktail’ of primers specific to a lower taxonomic groups can be used (Ivanova et al. 2007). Alternatively, short (~100-nt) sequences generated with ‘mini-barcode’ primers can be used to assign samples to family and, in most cases, to species (Meusnier et al. 2008). However, it remains to be seen whether mini-barcode primers work with common metazoan parasite taxa such as trematodes and cestodes. With parasite samples, even higher taxonomy can be difficult to determine. In such cases, we suggest that a two-tiered approach to DNA barcoding using short sequences from nuclear rRNA genes could be used to identify a specimen to family. The advantage of these genes is that it is possible to design primers that are more truly ‘universal’ and that work reliably on DNA samples of low quality or quantity (Frézal & Leblois 2008). The 18S primers used here anneal in highly conserved regions of the gene in flatworms and also work in a broad range of taxa including arthropods (Daphnia, Drosophila) as well as vertebrates (lizards, humans) (T. Crease, personal observation). Moreover, longer sequences from rDNA subunits can resolve specimens to even lower taxonomic levels (e.g. Mariaux 1998; Olson et al. 2003). We are not advocating the particular 18S gene region or the primers we used for the purpose of preliminary specimen identification. Rather, we suggest that short sequences from nuclear rRNA genes or some other easily recovered target could be amplified with truly universal primers and used to screen samples of completely unknown taxonomic affinity. This would provide enough information to select from multiple primer sets for species-level barcoding using the standard region of the COI gene.

Barcode-region COI sequences yielded good species-level resolution in our samples. For example, COI sequences showed better interspecific resolution than ITS sequences, although poor amplification rates with the ITS primers prevented us from comparing the two markers in many sibling taxa (Fig. 2). There may be more optimal markers than COI for identifying species in Schistosoma (Zarowiecki et al. 2007) or other flatworms (Littlewood et al. 2008), but for barcoding purposes a single marker must be adopted to ensure comparability. Our data suggest that even with imperfectly universal primers, the standard region of COI is a practical target for barcoding digeneans.

From a practical perspective, sequences obtained with these novel primers have already enabled us to better understand the diversity in a large sample of larval strigeidids from fishes in the Saint Lawrence River. Although the parasites of freshwater fishes in Canada are relatively well studied (e.g. Gibson 1996), little is known of the diversity and biology of some of the most common pathogens, such as Diplostomum spp. (Galazzo et al. 2002). Cryptic species occur in these taxa and a barcoding approach will permit resolution of questions concerning species diversity, geographical distribution, and host spectra of these problematic pathogens.


This research was supported through funding to the Canadian Barcode of Life Network from Genome Canada (through the Ontario Genomics Institute), NSERC and other sponsors listed at, as well by NSERC Discovery Grants to J.D. McLaughlin and T.J. Crease. We thank Elizabeth Holmes at the University of Guelph Genomics facility, and Natalia Ivanova, Liuqiong Lu and Janet Topan at the Canadian Centre for DNA Barcoding, for generating some of the sequences included in this study. We are grateful to Dr John Barta (University of Guelph), Dr Peter Watts (University of Toronto) and Sarah Elsasser (Laurentian University) for providing us with cestode and trematode specimens.

Conflict of interest statement

The authors have no conflict of interest to declare and note that the funders of this research had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.