A fragment of the cytochrome c oxidase subunit I (COI) gene has been used increasingly for species identification and discovery in eukaryotes. However, amplifying COI has proven difficult, or even impossible, in some taxa due to non-homology between the universal primers and the target DNA region. Among the most problematic animal groups is Serpulidae (Annelida). These sedentary marine animals live in self-secreted calcareous tubes and many of them, especially of the genus Hydroides, are economically important reef-builders, foulers, and biological invaders. We developed novel taxon-specific primers for amplifying COI from Hydroides, and for the first time generated 460-bp COI sequences from 11 of 14 species attempted. Average Kimura-2-parameter interspecific sequence distance (26.2%) was >60 times greater than the average intraspecific distance (0.43%), indicating that the COI gene is effective for species delimitation in Hydroides. Although applicability of the new primers for a wide range of serpulids needs to be tested, barcoding of Hydroides is now on its way from impossible to difficult. We anticipate that COI barcoding will provide a modern species identification tool and, combined with other molecular markers, yield important insights in phylogeny and evolutionary ecology of this large and important genus.

Introduction	Go To >> Top Introduction Materials and methods Results Discussion Acknowledgements References

DNA barcoding is a tool of species identification and discovery using a short and standardised gene region (Hebert et al. 2003). The best barcode candidate for animal groups, a 648-base-pair locus of the mitochondrial cytochrome c oxidase subunit I (COI) gene, has several desirable properties: it is present in all animal taxa, short enough to be sequenced easily and long enough to provide sufficient interspecific variations, and sequenceable using ‘universal’ PCR primers. This DNA barcode has been proven effective in diverse groups of animals at the species level, with a success rate of >98% (Park et al. 2011) for as many as 116 369 animal species registered with a barcode according to BOLD3 (http://www.barcodinglife.com, accessed on 14 August 2012). DNA barcoding has many applications, such as determination of species status and description of new species (Halt et al. 2009), revealing cryptic species (Hebert et al. 2004b; Smith et al. 2006), linking adult with juvenile (Thomas et al. 2005) or male with female of the same species (Willassen 2005).

One important prerequisite for applying DNA barcoding is successful PCR amplification of the gene fragment of interest. Previous research has revealed that the standard primer set LCO1490/HCO2198 (Folmer et al. 1994) cannot be used for amplification of the COI gene from some sponges (Erpenbeck et al. 2006), insects (Kondo et al. 2008), annelids (Halanych and Janosik 2006; Halt et al. 2009; Carr et al. 2011) and fish (Sevilla et al. 2007). This problem has led some researchers to propose alternative genes for barcoding animals, including some evolving at a lower rate than COI (Sevilla et al. 2007; Smith et al. 2007). However, this deviates from the original idea of barcoding using one standardised gene region for species-level identification in all animals, making it more difficult to catalogue and compare the data. Therefore, efforts have been directed to develop new primers in order to recover the COI barcoding region for those taxa that are not responsive to the standard primer set (Carr et al. 2011; Park et al. 2011).

Polychaeta, the largest class of Annelida, with 81 families and over 12 000 described species, is often the dominant group of macrobenthos in marine ecosystems in terms of both species numbers and abundance (Hutchings 1998). Although COI has been demonstrated to be a good barcode to distinguish species and reveal cryptic species of polychaetes, it is difficult to amplify in some members of Cirratulidae, Nephtyidae, Spionidae, Sabellidae, and especially Serpulidae (Pleijel et al. 2009; Barroso et al. 2010; Nygren and Pleijel 2011; Carr et al. 2011). Serpulidae is a monophyletic clade of polychaetes (Kupriyanova et al. 2006; Lehrke et al. 2007) living in self-secreted calcareous tubes and using a beautiful branchial crown for both feeding and respiration. From a barcoding point of view, Serpulidae is by far the most problematic polychaete family, appropriately termed a ‘nightmare group’ by Dirk Steinke (pers. comm.). Most attempts to amplify and sequence COI in serpulids using universal primers (e.g. Hydroides: Toril Loennechen Moen, pers. comm.; Marifugia: Valerija Zakšek, pers. comm.; Spirobranchus: Craig Starger and Paulo Paiva, pers. comm.) or to develop taxon-specific primers (Galeolaria: Halt et al. 2009) known to us have not been successful. The only successful amplifications were reported by Carr et al. (2011), who obtained sequences using primers designed specifically for polychaetes.

Because of their hard calcareous tubes and long-living planktotrophic larvae, some serpulids, especially in the genera Hydroides, Ficopomatus, Serpula and Spirobranchus, are economically important reef-builders, foulers, and bio-invaders (Schwindt et al. 2001; Nishi and Kato 2004). Hydroides Gunnerus, 1768, the largest genus of Serpulidae with ~100 species (ten Hove and Kupriyanova 2009), includes several species of bio-invaders and foulers (Qiu and Qian 1997; Lewis 2006; Otani and Yamanishi 2010; Tovar-Hernández et al. 2009). These worms can form dense aggregates on underwater structures such as aquaculture nets, seawater intake pipes and ship hulls and buoys (Qiu and Qian 1997), and therefore are important nuisances to marine aquaculture, navigation, shipping industries and power plants. Recently, Hydroides elegans was reported to be the major fouler of submarine seawater piping systems on boats based in Sydney, and was also problematic for the new submarines built in Adelaide (John Lewis, pers. comm.). Millions of dollars are spent each year to prevent the fouling of marine organisms, especially Hydroides, on man-made structures (Dürr and Watson 2010). Foulers can modify ecosystem dynamics and species assemblages through competition for space and food. For example, outbreaks of introduced Hydroides elegans caused serious damage to cultured oyster crops in Japan (Arakawa 1971). Of the 18 polychaete species on the list of 100 ‘worst invasives’ in the Mediterranean (Streftaris and Zenetos 2006), 12 species were serpulids, including eight species of Hydroides.

Identification of the species in question is the first and foremost step towards an understanding of the risk of potential bioinvasion. Revision based on morphological characters (Bastida-Zavala and ten Hove 2002, 2003) has shown that many species of this genus can be distinguished easily by the structures of the operculum, a modified radiole of the tentacle crown serving as a plug to close the opening of the tube. Some species with identical opercula, such as H. elegans and H. norvegicus, can still be distinguished by the morphology of the chaetae (ten Hove 1974). However, in many other species, such as H. albiceps and H. trivesiculosus, H. bispinosus and H. crucigera, H. elegans and H. longispinosus, the opercula are very similar and their intraspecific variations in morphological characters largely overlap with interspecific variations (ten Hove and Jansen-Jacobs 1984; Fiege and Sun 1999; Bastida-Zavala and ten Hove 2002). Clearly, an efficient molecular approach for species identification such as barcoding is urgently needed for serpulids, especially Hydroides.

The goal of this study was to develop taxon-specific primers that can be used to reliably amplify and sequence the barcoding fragment of COI in Hydroides. We reassessed the potential application of the universal COI primers (Folmer et al. 1994) and the polychaete-specific primers (Carr et al. 2011) for use in Hydroides, designed novel Hydroides-specific primers, and tested the effectiveness of the new primers for barcoding Hydroides and a non-Hydroides serpulid available to us.

Materials and methods	Go To >> Top Introduction Materials and methods Results Discussion Acknowledgements References

A total of 21 serpulid specimens belonging to 14 species of Hydroides and one species of Serpula were used in this study. The details of taxon names, collection localities, voucher numbers, and GenBank accession numbers are given in Table 1. Freshly collected individuals were fixed in 90% ethanol and preserved in 95% ethanol. Specimens were initially identified to species on the basis of descriptions in Straughan (1967), Imajima (1976), Wu and Chen (1980), Fiege and Sun (1999), Bastida-Zavala and ten Hove (2002, 2003), and Pillai (2009). Vouchers are deposited in the Australian Museum (AM) in Sydney.

Table 1.  Species included in the present study Collection locations, voucher numbers and accession numbers are shown (for those successfully amplified by PCR)

For each individual, the tentacle crown was separated from the body for DNA extraction to reduce the chance of contamination by exogenous DNA due to the presence of algae or bacteria in the gut. The crown was immersed in 0.5 mL TE buffer for 3 h with a change in buffer every 1 h to remove ethanol in the tissue. Qiagen DNeasy Blood and Tissue Kit was used to extract the genomic DNA according to the manufacturer’s protocol. Initial attempts to amplify COI were performed with the universal primer set (LCO1490/HCO2198: Folmer et al. 1994), and the primer set for polychaetes (polyLCO/polyHCO: Carr et al. 2011). PCR reaction mixture (total volume: 20 μL) contained 10× buffer (2 μL), 50 mm MgCl₂ (0.6 μL), 10 mm dNTPs (0.4 μL), 10 μm primer mix (1 μL), ddH₂O (15.42 μL), Invitrogen Platinum Taq Polymerase (0.08 μL) and template (0.5 μL). All amplifications were carried out in an Eppendorf Mastercycler Gradient thermocycler using the following PCR protocol: 94°C for 5 min, 5 cycles at 94°C for 30 s, 47°C for 30 s and 72°C for 40 s, 30 cycles at 94°C for 30 s, 51°C for 30 s and 72°C for 40 s, final extension at 72°C for 7 min. PCR products were separated by electrophoresis using 1.0% agarose gel. The gel was stained with ethidium bromide and visualised using an UV transilluminator equipped with a digital camera.

Out of the 21 samples, amplifications were successful in three samples by using LCO1490/HCO2198 and six by using polyLCO/polyHCO (Fig. 1). PCR products were purified with QIAquick Gel Extraction Kit (Qiagen) following the manufacturer’s protocol. Because the PCR products from a single amplification with the universal primers LCO1490/HCO2198 were insufficient for sequencing, they were diluted 1 : 10, and a 0.5-μL aliquot was used as template for a secondary amplification under identical PCR protocol. Nevertheless, the second PCR failed in all reactions. The products from the first PCR were then cloned directly into the pMD™ 18-T Vector (Takara) to sequence these amplicons. Sequencing reactions were performed using an AB SOLiDTM 4.0 automatic sequencer with the universal primer M13F for vector sequencing. PCR products amplified with polyLCO/polyHCO were sequenced directly without cloning.

Fig. 1.  Gel images of PCR amplicons for 20 Hydroides specimens and one Serpula specimen using LCO1490/HCO2198, polyLCO/polyHCO and Hydro-COIF/Hydro-COIR, respectively. ‘M’ denotes molecular size ladder in base pairs. The specimen labels are also shown in Table 1.

Sequences obtained with universal and polychaete-specific primers were compared with those in GenBank (www.ncbi.nlm.nkh.gov/Genbank/index.html) by BLAST search to check whether the correct gene fragments had been amplified. All sequences were not obvious pseudogenes (i.e. no frameshift). However, the six sequences obtained by using polyLCO/polyHCO showed >90% similarity with the COI sequences of bacteria such as Vesicomya spp. and thus were likely contaminants of prokaryotes. These sequences were not used for further analysis. The three sequences obtained using LCO1490/HCO2198 were aligned with ClustalX using default settings (15-gap opening penalty and 6.66-gap extension penalty), and subsequently checked by eye using BioEdit to identify the homogenous regions for design of new primers. The new primers were tested against the same 21 individuals under the same PCR protocol as for the other two sets of primers. The PCR amplicons after gel purification provided sufficient products for direct sequencing with new primers.

A matrix of sequence data obtained by using the newly developed primers was generated using ClustalX under default settings. Genetic distances among sequences were calculated using MEGA4 (Tamura et al. 2007). The Kimura-2-parameter (K2P) model was chosen as it is appropriate for comparison of taxa with low genetic distances (Nei and Kumar 2000) and has previously been used in studies of mtDNA barcoding (e.g. Hebert et al. 2004a).

To infer relationships between Hydroides and other polychaetes, the COI sequences of all polychaete species available in GenBank were downloaded and combined with the sequences of Hydroides and Serpula obtained in this study. The Neighbour-joining (NJ) analysis was applied as it is suitable for analysis of large assemblages of taxa (Kumar and Gadagkar 2000). The construction of NJ tree was performed with the K2P model using MEGA4 to provide a graphical representation of the result. Bootstrap analyses were performed with 1000 replicates using MEGA4.

Results	Go To >> Top Introduction Materials and methods Results Discussion Acknowledgements References

Amplifications with universal primers LCO1490/HCO2198 resulted in amplicons from three samples of three species: Hydroides diramphus, H. sanctaecrucis and H. elegans (Fig. 1). The read length of these sequences, after removing primers and vector sequences, was 658 bp. Alignment of the sequences revealed two distinct regions with high homogeneity (Fig. 2), which were used to develop a set of primers for Hydroides:

Fig. 2.  Alignment of the three COI sequences showing the conserved regions of the COI locus used to design degenerate oligonucleotide primers Hydro-COIF/Hydro-COIR. Asterisks indicate conserved positions.

Hydro-COIF: 5′-TCWRTWRTKACDGTKCATGCTA-3′ and

Hydro-COIR: 5′-CMRYAGGWTSAAARAACCTAGTA-3′,

where degenerate positions are represented by the following ambiguity codes: D = A|G|T; K = G|T; R = A|G; S = C|G; W = A|T; Y = C|T (Fig. 2).

We then tested the new primers against the 20 individuals of 14 Hydroides species and one individual of Serpula cf. granulosa. Thirteen amplicons belonging to 10 species were successfully obtained from the 20 Hydroides samples (79% success based on the number of species used; 52% success based on the number of samples used); one amplicon from Serpula cf. granulosa was also obtained (Fig. 1). The PCR reaction provided sufficient products for direct sequencing. The read length was 460 bp in all reactions. No stop codons were present in any of the sequences.

Interspecific sequence divergence ranged from 10.4 to 36.9%, with a mean of 26.2%. Intraspecific sequence divergence was much smaller, ranging from 0 to 0.9% (mean: 0.43%). There was no intraspecific genetic divergence between Hydroides elegans collected from Hong Kong and that from Australia. The intraspecific genetic divergence between H. sanctaecrucis collected from Hong Kong and Mexico was only 0.41%. The genetic divergence between H. operculatus and H. cf. operculatus, and between H. branchyacanthus and H. cf. branchyacanthus was 0.4% and 0.9%, respectively.

In every species of Hydroides for which more than one individual was used, the species was monophyletic in the NJ tree (Fig. 3). The clade of Hydroides was not clustered with the clade of other serpulids sequenced in an earlier study (Carr et al. 2011). The NJ tree indicated a base position of the Hydroides clade in relation to other polychaetes except for certain genera of Opheliidae (Travisia), Spionidae (Spio) and Serpulidae (Crucigera, Protula, Protolaeospira and Spirorbis) (Fig. 3). Compared with the clade of serpulids comprising Crucigera, Protula, Protolaeospira and Spirorbis, Hydroides had a closer relationship with other groups of polychaetes.

Fig. 3.  Neighbour-joining tree for polychaete species. The number next to each node indicates a bootstrap value (>75%) as a percentage of 1000 replicates. Only clades with serpulids are shown. The species-level classification is shown in Supplementary Material 1. The labels of samples are also shown in Table 1. To the right of the Hydroides species names are drawings of the opercula. Sequences cited from Carr et al. (2011) are downloaded from NCBI with the accession number after each species name. BLAST search results indicate that sequences of Travisia pupa, Spio setosa, Travisia forbesii, Crucigera sp., Crucigera zygophora, Spirorbis sp., Protolaeospira eximia, Protula pacifica, Spirorbis sp. are not of polychaete origin.

Discussion	Go To >> Top Introduction Materials and methods Results Discussion Acknowledgements References

This study represented the first successful attempt to amplify the COI sequences from Hydroides, the most speciose genus in Serpulidae, with ~100 species distributed worldwide (ten Hove and Kupriyanova 2009; Pillai 2009). The failure to amplify the COI barcoding region with the universal primer set in previous studies of polychaetes is thought to be due to the non-homology between the primers and the target DNA region (Halanych and Janosik 2006). Efforts to overcome the problem included designing primer sets/cocktail according to the published sequences of other species that are phylogenetically close to the target species (Halt et al. 2009; Carr et al. 2011), or searching for primer sites on the tRNA gene, which usually lies within 200 bp upstream of the COI gene (Park et al. 2011). When the sequences amplified using LCO1490/HCO2198 were compared with the primer sequences, two substitutions were identified in the 5′ end of the target barcoding region (G → T/A, G → T) and one in the 3′ end (A → C/T) (Fig. 4). These nucleotide substitutions could have led to the failure of primer annealing, which eventually resulted in the failure of the PCR reaction. PolyLCO/polyHCO contains modifications to LCO1490/HCO2198 in order to recover the three substitutions on the primer regions. Specifically, W (a position with the possibility of being A or T) and T were used to recover the double Gs in the 5′ end; and R (a position with the possibility of being C or T), was used to recover A in the 3′ end, respectively. However, using this modified primer set for Hydroides samples, we only generated amplicons of prokaryotic genes. The fact that COI primers originally designed for invertebrates might effectively amplify the COI locus of certain bacteria has been reported in studies of shrimp and mollusks, which is due to the high homogeneity between the primers and the corresponding region of the bacterial COI gene (Siddall et al. 2009). Our new primers were designed within the COI region to avoid using the region that is homogenous to the prokaryotic gene. Using the newly designed primer set, we successfully amplified 79% of the tested Hydroides species and one Serpula species. This result indicates that barcoding Hydroides using a fragment of the COI gene is now not an impossible goal, although it might still not be easy.

Fig. 4.  Alignment of the three COI sequences obtained by using LCO1490/HCO2198, showing three substitutions on the primer regions of Hydroides COI sequences, compared with the sequences of LCO1490/HCO2198. Asterisks indicate conserved positions.

The NJ tree indicated a deep divergence between the Hydroides clade obtained in this study and the other serpulids (Crucigera, Protula, Protolaeospira and Spirorbis) sequenced by Carr et al. (2011). This result is suspicious for several reasons. All previous phylogenetic studies based on morphological and molecular data indicated that Serpulidae is a monophyletic group (e.g. Kupriyanova et al. 2006, 2008; Lehrke et al. 2007; Kupriyanova and Rouse 2008) and Crucigera is most closely related to Serpula (Kupriyanova and Rouse 2008; Kupriyanova et al. 2008). Moreover, a BLAST search showed a high similarity between sequences from Carr et al. (2011) and sequences from prokaryotes such as Pseudoalteromonas sp. (98%), Colwellia psychrerythraea (94%), Galeomma turtoni (83%), Vesicomya sp. (80%) and Methyloversatilis universalis (76%). In contrast, the positive hits of our Hydroides sequences were all invertebrates, with 60–70% sequence similarity with polychaetes (Eulalia levicornuta, Eumida merope, Eumida sanguinea, Nereiphylla castanea, Pseudomystides limbata), arthropods (Holopedium glacialis, Triops longicaudatus), mollusks (Gulella usambarica, Schizobrachium polycotylum, Odostomia plicata) and jellyfishes (Crambionella orsini, Aurelia aurita). Besides, serpulid sequences from Carr et al. (2011) have 2–3 indels in the reading frame and the location of the indels is inconsistent, so these sequences are not currently ruled out as being from bacteria. Although polyLCO/polyHCO is useful for several families of polychaetes, as demonstrated by Carr et al. (2011), our results suggest that this primer set is not suitable for barcoding serpulids.

Levels of interspecific sequence divergence reported in the present study of Hydroides are high (10.4–36.9%), but are consistent with data from other studies of polychaetes, which also reported high values, i.e. 16% in Harmothoe (Hardy et al. 2011); mean of 16.5% for 333 species (Carr et al. 2011). In the polychaete Pectinaria koreni, the intraspecific sequence divergence could reach 16.4% (Jolly et al. 2005), but this might indicate the presence of cryptic species and the data might actually reflect interspecific sequence divergence. The mean COI K2P sequence divergence between the Hydroides species was 60 times higher than within the species. Application of the average within-species variation threshold (i.e. 10× variation: see Hebert et al. 2004a; Witt et al. 2006) indicated the presence of 10 Hydroides species among our samples, which was consistent with the number of identified species. Entries for the same species were tightly clustered and distinct from the other species (Fig. 5). The large gap between the intraspecific and interspecific COI divergence indicated that the COI fragment amplified by our primer set is a good candidate for distinguishing species of Hydroides and can be an effective barcoding tool for this genus. It is especially useful for species identification when expertise in morphology-based identification is not available or morphology-based methods do not work. For example, H. branchyacanthus and H. cf. branchyacanthus were identified on the basis of the presence of geniculate spines with a well developed knob protruding above a subapical incurving tip in both species, nevertheless they differ in the spine size: in H. branchyacanthus the dorsal spine is larger than the other spines, whereas in H. cf. branchyacanthus all spines are similar in size and shape. Since the two taxa were placed in one cluster with low divergence (0.9%), we consider both as H. branchyacanthus. Similarly, H. operculatus and H. cf. operculatus, identical with respect to several morphological characters (i.e. 9 verticil spines, spines of identical shape, and the presence of a basal internal spinule) but differing in the number of large spines (one in H. operculatus versus three in H. cf. operculatus), showed only 0.4% sequence divergence from each other but at least 22.5% divergence from other clades; therefore both should be considered as H. operculatus.

Fig. 5.  COI (K2P) distances for the barcode region within species and genera.

In the present study, individuals of the same species collected from widely different geographical locations show low genetic divergence. This was the case for Hydroides elegans collected from China and Australia, Hydroides sanctaecrucis collected from Mexico and China, and Hydroides ezoensis from China and Russia. The great distance between collection sites of the same species but lack of genetic divergence is consistent with the observation that these species are common fouling organisms that are easily transported by ship (ten Hove and Kupriyanova 2009) and their wide distribution could be the result of recent introduction. In fact, biofouling was proposed as a major mode of dispersal for Hydroides elegans as inferred from microsatellite loci (Pettengill et al. 2007). Therefore, the COI fragment amplified by our newly designed primer set can be useful for confirmation of the specific status of invasive Hydroides.

In summary, our newly designed primer set has opened the door for barcoding of Hydroides, the most speciose and economically important genus of Serpulidae. In addition, it can amplify the COI fragment from Serpula cf. granulosa and thus, may be useful for other non-Hydroides serpulids. As already noted, the new primer set failed to amplify the barcode region of three tested Hydroides species. But the successfully amplified Hydroides sequences can serve as reference sequences for the design of novel primers and confirmation of the feasibility of creating a barcode library for Hydroides when Hydro-COIF/Hydro-COIR does not work. With the increasing application of COI barcoding and expansion of the DNA reference library, we anticipate a clearer distinction between Hydroides species, especially for those species whose identity cannot be resolved solely on the basis of morphological characters. When data from the COI barcode and other mitochondrial gene sequences become available for most species of Hydroides, analyses can then be conducted to better understand the speciation and evolutionary history of Hydroides.

Acknowledgements	Go To >> Top Introduction Materials and methods Results Discussion Acknowledgements References

We thank Emma Johnston, Ana-Maria Tovar, and Vasily Radashevsky for collecting specimens from Australia, Mexico and Russia, respectively, and Paulo Paiva, Craig Starger, Dirk Steinke, Toril Loennechen Moen, and Valerija Zakšek for sharing their experience in barcoding serpulids, and John Lewis for providing information on fouling caused by Hydroides. This project was supported by a grant from Environment and Conservation Fund, Hong Kong (ECF 7/2007).

References	Go To >>\ Top Introduction Materials and methods Results Discussion Acknowledgements References