Development of sensitive and specific molecular tools for the efficient detection and discrimination of potentially invasive mussel species of the genus Perna.

Marine mussels of the genus Perna include three species: P. canaliculus, P. viridis and P. perna. While P. canaliculus appears to be greatly restricted to its endemic range of New Zealand, P. perna and P.viridis introductions have been recorded outside their native ranges in several regions of the globe. Such introductions have often resulted in significant negative ecological, economic and social impacts. Perna perna and P.viridis are exotic to Australia and are listed under the Australian Government National System for the Prevention and Management of Marine Pest Incursions as high priority species. Rapid detection of marine pests such as Perna species remains fundamental to their effective containment and control. The present study reports on the development and validation of both conventional and real-time PCR assays suited to the rapid identification and discrimination of juvenile and adult specimens of P. viridis, P. canaliculus and P. perna. The development of a sensitive high-throughput real-time PCR assay offers further potential for the efficient detection of the presence of single Perna specimens in mixed populations of native mussel species, and for early detection of larval stages in ballast water and plankton samples. This assay offers considerable advantages over traditional identification methods and represents an important step in developing capacity for efficient identification and management of Perna species incursions in Australian waters.


Introduction
Marine mussels of the genus Perna are currently recognized as three species, P. canaliculus (Gmelin 1791), P. viridis (Linnaeus 1758) and P. perna (Linnaeus 1758) (Siddall 1980;Vakily 1989;Wood et al. 2007).The New Zealand green-lipped mussel P. canaliculus, an important aquaculture species, is endemic to New Zealand, extending from the far north of the North Island to as far as Stewart Island in the south (Jeffs et al. 1999;Powell 1979;Wood et al. 2007) (Figure 1).The Asian green mussel P. viridis is native to Asian waters, occurring extensively throughout the Indo-Pacific region from the Persian Gulf throughout India and South East Asia (Baker et al. 2007;Siddall 1980) (Figure 1).The brown mussel P. perna is native to most of the Atlantic coast of South America (Beauperthuy 1967;Siddall 1980), to the south coast of Portugal (Lorenço et al. 2012) through much of the African coastline (Nordsieck 1969;Siddall 1980) and to Sri Lanka (Sadacharan 1982;Wood et al. 2007) (Figure 1).In southwest India a fourth species Perna indica (Kuriakose 1980;Rao 1974), classified for many years as P. perna (Vakily 1989), was recently suggested to be a distinct species within the genus (Divya et al. 2009(Divya et al. , 2010)).The taxonomic status of this species is, however, recognised as requiring further clarification.Siddall (1980), Vakily (1989) and Wood et al. (2007), with an added extension of P. perna distribution to the south coast of Portugal (Lourenço et al. 2012), the west coast of India (Vakily 1989), (Divya et al. 2010) and Sri Lanka (Sadacharan 1982).Locations where P. perna and P. viridis are known to have subsequently become established following introduction are marked with dots: 1. Cook Islands, 2. Tahiti (French Polynesia), 3. Mexico, 4. Florida (USA), 5. Georgia (USA), 6. Jamaica, 7. Trinidad, 8. Philippines, 9. Okinawa (Japan) 10.Japan, 11.New Caledonia, 12. Fiji, 13.Tonga and 14.Western Samoa.Countries from which samples were collected are named in the map.Details regarding collection points within these countries can be found in Table1.
While the only two P. canaliculus occurrences outside New Zealand, as far as the authors are aware, have been reported from South Australia (SA) and eradicated (in the 90's, Justin McDonald personal communication, and more recently, Wiltshire et al. 2010), P. perna and P. viridis introductions have been recorded outside their native ranges in several regions.Perna perna has been established in the western Gulf of Mexico since the 90's (Hicks and McMahon 2002;Hicks andTunnel 1993, 1995) and was recently removed following its incursion in an oil rig in New Zealand waters (Hopkins et al. 2011a).A series of P. viridis introductions, resulting from shipping, fisheries and aquaculture activities, have been reported to occur in China and Japan since the 60's (Habe 1976;Hanyu and Sekiguchi 2000;Hanyu et al. 2001;Ye 1997;Yoshiyasu et al. 2004).Perna viridis introductions for aquaculture have also occurred in numerous Pacific islands (Bell et al. 1983;Coeroli et al. 1984;de Gaillande 1979;Eldredge 1994;Vereivalu 1990), the Caribbean (Agard et al. 1993;Hicks 2001) and Atlantic coasts of North and South America (Benson et al. 2001;Penchaszadeh and Velez 1996;Power et al. 2004;Rylander et al. 1996) (Figure 1).The successful establishment of P. viridis in these areas has had concerning economical and ecological impacts.The species' fast growth rate and lack of competitors has led to populations of P. viridis rapidly outcompeting native species at introduced locations (Hicks 2001;Hicks et al. 2001;Ingrao et al. 2001).Heavy fouling of man-made structures such as aquaculture cages, buoys, boats and water intake pipes of power stations has resulted in costly maintenance to shipping and other coastal industries (Rajagopal 1991;Rajagopal et al. 2006).
In Australia, P. viridis was introduced to Trinity Inlet, Cairns, Queensland in 2001 via ship hull biofouling.Successful eradication efforts were put in place following the detection of breeding individuals in this area in 2002 (Hayes et al. 2005).Perna viridis incursions (detection of individuals but not of established breeding populations) are frequent as this species is among the most commonly identified alien invasive species within the biofouling community of vessels entering Australian waters (McDonald 2012;Piola and McDonald 2012).Perna perna and P. viridis are listed under the Australian Government National System for the Prevention and Management of Marine Pest Incursions (DAFF 2010) as high priority species.The eradication of P. viridis from Australia (Hayes et al. 2005) and of P. perna from New Zealand are among the few examples of successful eradication of marine pests worldwide.Such processes are extremely complex and expensive, as well as highly dependent on the early detection of the pest species in order to avoid irreversible establishment and dispersal (Elton 1958;Hayes et al. 2005;Hopkins et al. 2011a,b;Willan et al. 2000).
The early detection of marine pest species can only be accomplished through the establishment of baseline surveys (Campbell et al. 2007;Hewitt andMartin 1996, 2001), repeated monitoring of high risk areas (Hewitt et al. 2004), and screening and identification of a wide range of taxa using methods which are typically time-consuming and expensive (Bott et al. 2010;Hayes et al. 2005).As taxonomic expertise declines globally (Hopkins and Freckleton 2002;Kim and Byrne 2006), molecular techniques like Polymerase Chain Reaction (PCR) and real-time PCR are recognised as important complementary methods in species identification, as well as very attractive alternative tools in marine pest research and monitoring programs (Blanchet 2012;Bott and Giblot-Ducray 2011;Holland 2000;Mountfort and Hayden 2007;Wood et al. 2013).The use of DNA-based techniques has proven to be of particular relevance in the identification of marine invertebrate species, whose morphologic characters are frequently plastic, are influenced by environmental factors, may be lacking or not distinctive at early life stages and may often only be recognised by highly trained taxonomists (Deagle et al. 2003;Dias et al. 2009;McBeath et al. 2006;Siddall 1980;Willis et al. 2011).
PCR, although widely used and useful in species identification, is a qualitative technique that requires a certain initial amount of tissue sample for DNA extraction and the post-processing of samples for analysis.This has motivated some authors to apply hydrolysis probe based real-time PCR to the identification and quantification of marine invertebrate species whose initial life stages consist of planktonic larvae (Dias et al. 2009;McBeath et al. 2006;Pan et al. 2008;Vadopalas et al. 2006).This one-step, fully automated methodology allows detection to occur during the reaction, reducing the time, labour and costs arising from post processing of samples in agarose gels.By using highly specific fluorescent probes, the sensitivity and specificity is also greatly improved in comparison to conventional PCR, allowing for the detection from samples containing a significantly lower initial concentration of target DNA.Because the intensity of fluorescence during the exponential amplification phase, measured when it first rises above background level or Critical Threshold (Ct), is directly correlated with initial template quantity, real-time PCR has also been used quantitatively (for a review of this technology see Vasalek and Repa 2005).
The present work was initiated with the objective of developing a potentially high-throughput realtime PCR assay capable of rapidly detecting and discriminating the mussel species P. perna, P. canaliculus and P. viridis.The adoption of this assay by the Western Australian biosecurity program represents an important step towards developing capacity for the rapid and efficient identification and management of Perna species incursions in Australian waters.

Sample origin and DNA extraction
In this study we used DNA extracts from 31 individuals of mussels P. perna, P. viridis and P. canaliculus sampled in the study of Wood et al. (2007) and four P. perna adductor muscle tissue reference samples obtained from the Marine Invasive Taxonomic Service (MITS) in New Zealand.DNA extracts from the study of Wood et al. (2007) were kept stored at -80ºC, transported to the WA Fisheries and Marine Research Laboratories in a cool package and stored at -20ºC until use.No field study was conducted and therefore no specific permits were required.Locations from where all 35 samples were obtained in previous study by Wood et al. (2007) and by MITS can be found in Table 1 and are indicated in Figure 1.From the four mussel adductor muscle samples, 5 mg of tissue was sub-sampled to an eppendorf tube, homogenized in 200 µl of lysis buffer (Fisher Biotec) using a micropestle and incubated overnight with 20 µl of proteinase K (Fisher Biotec) at 60ºC.DNA was extracted using a Fisher Biotec FavorPrep Tissue Genomic DNA Extraction Mini Kit, following the manufacturer's instructions.All DNA samples were stored at -20ºC until further use.

Primer design and PCR
P. perna, P. viridis and P. canaliculus mitochondrial DNA sequences available from the work of Blair et al. (2006) (nad4 gene spanning the IGS to the cox1 gene; GenBank DQ343568 to DQ343611), and Perna individuals sampled in the work of Wood et al. (2007) (cox1 region; DQ917582 to DQ917618) were aligned and screened for primer and probe candidate regions.Five sets of primer combinations were designed using the Primer Express version 3.0 software (Applied Biosystems).Generic primers, capable of amplifying targets from all members of the genus, were designed based on conserved regions of the sequences.These (conserved) primer regions flanked variable regions that were suited for the further design of species-specific probes.Primer specificity was tested against each Perna species through PCR amplification in 25 µl reaction mixtures containing 2 µl (10-20 ng) DNA, 1.25 mM of each dNTP, 62.5 mM MgCl 2 , 2.5 µl of 10x reaction Buffer, 2.5 µM of each primer, one unit of Taq DNA polymerase (Fisher Biotec) and PCR-grade water (Fisher Biotec).PCR conditions consisted of an initial incubation at 94ºC for 5 min, followed by 30 cycles at 94ºC for 30 s, 56ºC for 30 s, 70ºC for 90 s, and a final extension step of 72ºC for 5 min in an Applied Biosystems (ABI) 2720 thermal cycler.A negative control, with no template DNA added, was included in all PCR assays.PCR products were separated by electrophoresis using 1.5% agarose (Fisher Biotec) gels stained with ethidium bromide (Fisher Biotec) alongside a 100 bp molecular weight marker (Axygen Biosciences) and visualised under UV light.

Sequencing
The most suitable set of primers across the three species (plus a set of flanking primers) were selected and used in the sequencing of the target region across all 35 Perna individuals.Sequencing of unpurified PCR products was performed using the service provided by the Australian Genome Research Facility (AGRF) in Perth.All samples were sequenced in both directions and consensus sequences generated using Sequencher 5.0 (Gene Codes Corporation).Sequences were aligned and analysed in the BioEdit 7.1.3.0Sequence Alignment Editor (Hall 1999), using CLUSTAL W (Thompson et al. 1994) with default parameters.

Real-time PCR
Three specific hydrolysis TaqMan ® -MGB (FAM, VIC and NED) probes (one for each Perna species) were designed using the Primer Express version 3.0 software (Applied Biosystems), based on the preliminary specificity trials described above and sequences obtained for all 35 individuals.To help guarantee the specificity of the method, all the designed primers and probe sequences were subjected to a BLAST search, to check the GenBank database for any similar sequences and potential cross-reactions.Probe specificity was further analysed across all 35 individuals of the three Perna species, using both single-probe and multiplex assays.Multiplex assays were performed on an ABI Step One Plus™ real-time PCR system using a cycling profile of 50ºC for 2 min (AmpErase ® uracil Nglycosylase incubation), 95ºC for 20 s (DNA polymerase activation) followed by 45 cycles of 95ºC for 1 s (denaturation) and 60ºC for 20 s (annealing / extension).Reactions were conducted in a final volume of 20 µl containing 1 µl of DNA template, 1x TaqMan ® Fast Advanced master mix (Applied Biosystems), 900 nM of each primer and 200 nM of each TaqMan ® probe (Applied Biosystems).Efficiencies of primers and probes -Efficiency (%) = [10 (-1/slope) ]-1  100were assessed using standard curves based on both triplicate single-probe and triplicate multiplex reactions of 10-fold dilutions of DNA samples of each Perna species.

PCR
From the five sets of generic primers designed to target regions conserved across the three Perna species, the set that gave the best results was the one flanking the variable IGS region.The forward primer (Fw A) was designed at the 3' end of the nad4 gene and the reverse primer (Rev A) at the beginning of the cox1 gene (Table 2).This set not only allowed for the amplification of all three species with a single pair of primers, but also generated a species-specific size product for each species -P.perna 281 bp, P. canaliculus 249 bp and P. viridis 201 bp -due to the high variability of the IGS region (Figure 2).This assay thus has the potential to be used as a stand-alone assay to differentiate species of the Perna genus in the absence of access to real-time PCR systems.

Sequencing of the marker region
Sequencing of the region spanned by the Fw A/Rev A set of primers (primer region inclusive) was possible for 27 of the 35 individuals using either a single set of flanking primers (nad4 fw2  and COI rev5, Table 2), or a combination of four primers (Fw A and COI rev5, nad4 fw2 and Rev A) with subsequent alignment of the two fragments for the generation of a final consensus sequence.For the remaining 8 individuals for which complete sequence was not obtained (Mor2, Mor3, Mor4, Chen2, Phil1, Phil2, Thai1 and Viet1) only the sequences between the target primers (Fw A and Rev A) were included in the general alignment.No effort was made to extend these sequences due to constraints in the amount of each DNA extract available.The sequences obtained were used for the basis of primer optimization and species-specific probe development around the variable IGS region.All sequences were deposited in GenBank (see Table 1 for accession numbers).

Real-time PCR method specificity and efficiency
Results from the GenBank BLAST searches of all the designed primers and probe sequences indicated these to be specific to the targeted Perna species.One should note that the target marker region had not been previously sequenced for P. viridis in the work of Blair et al. (2006) and was not available in the GenBank database for this or any closely related species.Primers and probes were designed to be able to detect and differentiate between P. perna, P. canaliculus and P. viridis and were specific for these species.Detections from both singleprobe and multiplex assays across all 35 individuals used in this study (Cts 18-34) were obtained for the species-specific DNA in the reactions.No detections were obtained from negative controls.In order to obtain accurate, reproducible and comparable results, real-time PCR reaction efficiency should be as close to 100% (slope of -3.33) as possible (Pfaffl 2004;Valasek and Repa 2005).Standard curves based on either single or multiplex probe reactions of P. canaliculus, P. viridis and P. perna DNA extracts revealed slope values between -3.8 (Efficiency= 83%) and -3.6 (Efficiency= 90%) (Figures 3 and 4).There was high correlation between cycle number and dilution factor, R 2 = 0.996 -0.999 for all dilution series (Figures 3  and 4).Triplicate single-probe reactions yielded similar Ct values, with SD± 0.0 to 0.7 across all dilution series (standard deviations of triplicate Ct values were not included on Figures 3 and 4 as they are too small to be visualised).

Discussion
In this study, we have developed both a conventional PCR and real-time PCR assay capable of rapidly detecting and discriminating the mussel species P. perna, P. canaliculus and P. viridis.The simultaneous identification of all three species based on a single marker can save time and costs, offering considerable advantages over previous PCR identification approaches.The specificity and efficiency of the real-time PCR method indicates its potential for detection of Perna species in pooled tissue samples, and at larval and juvenile stages.This method could therefore save further time and costs of processing samples individually and prove of most value in the identification of Perna species at early life stages, when morphological characters can be particularly difficult to distinguish between species.

PCR
Previously available assays for the molecular identification of Perna species include that used by Holland et al. (1999), which relied on chromosome morphology to confirm the identification of P. perna introduction in the Gulf of Mexico.Santaclara et al. ( 2006) also described a method in which Perna species can be identified through a combined approach of either of two PCR methods (Kenchington et al 1995;Perez et al. 2004) combined with Restriction Fragment Length Polymorphism (RFLP).Blair et al. (2006) developed a PCR method consisting of a combination of a general forward and three distinct reverse primers (designed in the mtDNA nad4 and IGS regions), each giving a size-specific band for each species.Although more straight-forward than the previously described techniques, this method requires conducting three separate reactions to identify each Perna individual.These authors attempted to design a reverse primer common to all three species but were unable to sequence the IGS region of P. viridis, and suggested that this could be due to a different mitochondrial gene order in this species.In our work, we were able to generate sequence across the IGS region for all species, which was confirmed to be of different size for each species and considerably smaller in P. viridis.The use of a single set of primers capable of amplifying species-specific fragments in a single reaction (Figure 2) greatly simplifies the identification process and costs associated with it.However, when using the PCR primers developed in the present work, one should note that the species-specific fragments are less than 50 bp different in size from each other (P.canaliculus 249 bp, P. viridis 201 bp and P. perna 281 bp) and therefore the use of longer electrophoresis time and positive controls for the three species is strongly recommended.

Real-time PCR
The real-time PCR developed in the present study is able to detect and discriminate among all three Perna species.The real-time PCR efficiency obtained across a wide series of 10fold dilutions of DNA of each Perna species was high (> 80%).Although our results indicate a loss of 4% efficiency when changing from a single-probe to a multiplex reaction, this seems to be an acceptable compromise considering the time and cost advantages of the multiplex assay.
Considerable effort was made to obtain Perna species samples from as many populations as possible across the globe.Such samples allowed us to perform a comprehensive analysis of the IGS region intra and inter-species variability and better design species-specific probes.Based on such knowledge, and when applied to pest species detection from ships (fouling, ballast water) entering the waters of any nation, the real-time PCR method developed in this work is expected to be able to detect incursions of nonnative Perna species from nearly every region of the globe.Nevertheless, one should note that the specificity of the method has been checked against sequences of organisms deposited in GenBank.Cross-reaction with species not previously sequenced remains possible and therefore electrophoresis (to check for right product size) and sequencing of Perna positives is strongly recommended, especially in a biosecurity context.

Future applications to biosecurity surveillance
Future applications of this method include the detection of Perna species from plankton and bulk tissue samples.The method specificity and efficiency outlined in this study represents a first positive step towards further potential applications such as the detection and relative quantification of larval stages in water samples.The development of such an application would allow important detection and/or monitoring to take place from ballast water of vessels and of plankton from, for example, high risk areas (ports, marinas) or following incursion/eradication events of Perna species.Pooling small tissue samples from up to 30 individual mussels for a single DNA extraction and real-time PCR reaction detection has been performed confidently for the efficient detection of closely related mussel species (Mytilus spp.) in previous studies (Dias et al. 2008).Such an approach enables the analysis of 30 samples in a single real-time PCR reaction.Although sample preparation is inevitably longer, each assay can include up to 94 reactions (plus two controls) in a 96-well plate, meaning that the pooling of 30 samples per reaction would allow for the analysis of 2820 mussels in a 33 min run, allowing detection results to be generated in the same day.
The development of such rapid and specific methods could prove most valuable in assisting with the visual identification of pest species (especially at early stages) during biofouling inspections.Verifying the presence/ absence of suspected pests in a short time frame (same day) could at times avoid the unnecessary cleaning of vessels, considerably reducing the time and costs associated with these inspections.Taqman, single-probe, real-time PCR assays for the detection of some of the bivalve species of biosecurity concern to Australia, namely P. canaliculus, Corbula gibba and Musculista senhousia have been recently developed and/or optimized by Bott and Giblot-Ducray (2011) at the South Australian Research and Development Institute (SARDI).These assays have proved efficient, time and cost effective in the screening of these pest species from both tissue and plankton samples.In order to establish the molecular capacity for pest detection, the costs of acquiring the necessary equipment and the relatively lengthy process of training staff, developing, testing and validating assays is inevitable, representing the major drawbacks of such applications.However, we believe such drawbacks can be overcome by the benefits of having these tools in place, not only for the regular monitoring of pest species, but also in the face of an incursion event.

Figure 2 .
Figure 2. Agarose gel showing Perna species-specific band length.P. canaliculus 249 bp (Pc), P. viridis 201 bp (Pv) and P. perna 281 bp (Pp) species-specific PCR products are shown along a negative control and flanked by a 100 bp molecular weight marker (Axygen Biosciences).

Figure 3 .
Figure 3. Efficiency of real-time PCR single-probe reactions.Average cycle threshold values obtained through single-probe realtime PCR reactions of 10 fold dilutions of total DNA extracted from adult P. perna (Pp), P. canaliculus (Pc) and P. viridis (Pv).Slope values giving reaction efficiency of Taqman®-MGB probes for each species are shown on the graphic.Standard deviations (SD) are ±0.2,±0.7 and ±0.3 for P. perna (Pp), P. canaliculus (Pc) and P. viridis (Pv) respectively and are too low to be visualised in the figure.
Figure 4. Efficiency of real-time PCR multiplex reactions.Average cycle threshold values obtained through multiplex real-time PCR reactions of 10 fold dilutions of total DNA extracted from adult P. perna (Pp), P. canaliculus (Pc) and P. viridis (Pv).Slope values giving reaction efficiency of Taqman®-MGB probes for each species are shown on the graphic.Standard deviations (SD) are ±0.6,±0.5 and ±0.3 for P. perna (Pp), P. canaliculus (Pc) and P. viridis (Pv) respectively and are too low to be visualised in the figure.

Table 1 .
Details of mussel samples used in this study: sample collection locations, names, voucher references and GenBank accession numbers.
Note: oligos marked with an asterix ( ⃰ ) were designed in the lagging DNA strand.