Characterisation of the mantle transcriptome and biomineralisation genes in the blunt-gaper clam, Mya truncata

Members of the Myidae family are ecologically and economically important, but there is currently very little molecular data on these species. The present study sequenced and assembled the mantle transcriptome of Mya truncata from the North West coast of Scotland and identi ﬁ ed candidate biomineralisation genes. RNA-Seq reads were assembled to create 20,106 contigs in a de novo transciptome, 18.81% of which were assigned putative functions using BLAST sequence similarity searching (cuttoff E-value 1E − 10). The most highly expressed geneswerecomparedtotheAntarcticclam( Laternulaelliptica )andshowedthatmanyofthedominantbiological functions(muscle contraction, energyproduction, biomineralisation) in the mantle were conserved. There were however, differences in the constitutive expression of heat shock proteins, which were possibly due to the M.truncata sampling location being ata relatively low latitude, and hencerelatively warm,intermsof theglobal distribution of the species. Phylogenetic analyses of the Tyrosinase proteins from M. truncata showed a gene expansion whichwas absent in L.elliptica . The tissue distribution expression patternsof putative biomineralisation genes were investigated using quantitative PCR, all genes showed a mantle speci ﬁ c expression pattern supporting their hypothesised role in shell secretion. The present study provides some preliminary insights into how clams from different environments – temperate versus polar – build their shells. In addition, the transcriptome data provides a valuable resource for future comparative studies investigating biomineralisation. © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
Biomineralisation is an essential biological process for many living organisms. From microalgae to shellfish, multiple taxain at least 30 phylasecrete calcium carbonate crystals onto a protein matrix (Addadi and Weiner, 1997). The Mollusca is one such phylum whose success since the base of the Cambrian (545 million years ago) is partly attributed to the possession of a hard shell Marie et al., 2013;Vermeij, 2005). The mollusc shell has been the subject of scientific interest for centuries and continues to be researched from increasingly multi-disciplinary perspectives (Marin and Luquet, 2004;Zhang et al., 2012). Shells contain approximately 95-99% calcium carbonate (CaCO 3 ) and 1-5% organic matrix; the organic component is a protein matrix which applies synergetic forces that either nucleate or inhibit crystal growth Meenakshi et al., 1971;Weiner and Hood, 1975). The protein matrix is secreted by the mantle and recently, particularly since the 'omics era began, much research has focussed on understanding the molecular mechanisms of shell growthspecifically the genetic control of shell matrix protein secretion (Clark et al., 2010;Werner et al., 2013;Zhang et al., 2012).
Mya truncata is a marine bivalve which is part of the Myidae family of soft-shelled clams. These clams live buried in sediments, both intertidally and subtidally, and are important in many ecosystem functions such as long-term sediment stabilisation, bioturbation and benthopelagic coupling (Queirós et al., 2013). Members of the Myidae family are edible and important economically as a food source and others, such as Mya arenaria are invasive (Powers et al., 2006;Sousa et al., 2009). Large infaunal clam species are morphologically similar, and often inhabit similar ecological niches. Clam species from different geographic and environmental locations can therefore provide valuable models for comparing important biological processes over physical gradientssuch as temperature (Morley et al., 2007).
Despite being named "soft-shelled" clams, M. truncata have a hard shell which is composed of four structurally distinct layers: (i) the outer periostracum (approximately 3 μm and relatively thin); (ii) an outer shell layer of aragonitic granular prisms; (iii) a middle layer of aragonitic crossed lamellar; and (iv) an aragonitic inner layer of complex crossed lamellar (Fig. 1). In order to better understand the molecular mechanisms which produce M. truncata's shell, transcriptomic sequence data was generated and mined for biomineralisation candidates.
The objectives of the present study were: 1.) Develop a molecular resource to aid the study of biomineralisation in M. truncata by sequencing, assembling and putatively annotating its mantle transcriptome. 2.) Identify and further characterise candidate biomineralisation genes in the newly-assembled M. truncata mantle transcriptome. 3.) Further characterise possible biomineralisation mechanisms in M. truncata by preliminary comparison with another bivalve species living at lower temperatures, the Antarctic clam, Laternula elliptica.

Method
2.1. M. truncata mantle transcriptome 2.1.1. Animal collection and RNA extraction M. truncata (n = 9, mean shell length = 64.5 mm) were collected by the NERC Facility for Scientific Diving from Dunstaffnage Bay, North West Scotland in August 2011. Mantle tissue was dissected from each animal and RNA was extracted using TRI reagent (TRIsure) according to manufacturer's instructions (Bioline, UK) and purified on columns (Qiagen, UK).

Sequencing and bioinformatics
The total mantle RNA from nine individuals was pooled prior to sequencing. Pooled RNA was subject to 454 GS FLX Titanium sequencing at Cambridge University Department of Biochemistry Sequencing Centre. 454 reads were assembled into a de novo transcriptome with GS Data Analysis Software (454.com/products/analysis-software/) on default genomic-style parameters, resulting in a total of 20,106 contigs with an average read length of 675 bp. All contigs were compared to the NCBI non-redundant (nr) database (downloaded for in-house use January 2015) using the Basic Local Alignment Search Tool (BLAST) to search for sequence similarity and putative gene annotation (Altschul et al., 1990). The most highly expressed annotated contigs were identified to highlight dominant processes in the mantle at the transcript level. The mantle transcriptomes of M. truncata and L. elliptica were compared using tBLASTx with default parameters.
Putative biomineralisation genes were identified using keyword searches for candidates which have previously been shown to be associated with shell production and calcification (Table 1). In addition, contigs which were present in the mantle and shell proteome with similarity to biomineralisation domains (Arivalagan et al., 2016in this issue) were also included.

Tyrosinase bivalve phylogeny
Fifteen of the M. truncata mantle contigs showed high sequence similarity to Tyrosinase. To understand the evolution of Tyrosinase proteins in M. truncata, phylogenetic analyses were carried out. The fifteen Tyrosinase transcripts were mapped to a reference Tyrosinase domain (PF00264) and preliminarily aligned using Clustal-W with default parameters (Larkin et al., 2007). The alignment indicated that most contigs represented gene fragments that mapped to different, non-overlapping regions. To provide the most accurate phylogenetic analysis, only those fragments which overlapped were selected for further analysis. From the initial alignment of fifteen contigs, five were identified as potential paralogues, the remaining eight contigs could still be paralogues, but could not be included in analysis.
The derived amino acid sequences of the five selected contigs were added to a previously published bivalve Tyrosinase alignment (provided by Aguilera et al., 2014) and the phylogeny was determined using the method of Aguilera et al. (2014). Briefly, alignments were created using the MAFFT algorithm (Katoh et al., 2005), refined using the RASCAL webserver (Thompson et al., 2003) and analysed with Gblocks 9.1b (Castresana, 2000) to select conserved regions. The final alignment was used to run three phylogenetic models: Neighbor-Joining (NJ) reconstructions were performed using MEGA 5.2.2 (Tamura et al., 2011), Maximum-likelihood (ML) trees were constructed using RAxMLGUI v. 1.3 (Silvestro and Michalak, 2012) and Bayesian inferences (BIs) were performed using MrBayes v. 3.2 (Ronquist et al., 2012). Data from the three models were manually combined to produce a consensus tree (Fig. 2).
2.3. Tissue distribution of candidate biomineralisation gene expression 2.3.1. Experimental design Animals (same collection as Section 2.1, n = 5, mean shell length = 60 mm) were dissected into six different tissues: mantle, siphon, gill, foot, digestive gland and gonad. Dissected tissue samples were snap frozen in liquid nitrogen and stored at −80°C prior to RNA extraction.

RNA extraction and quantitative-PCR
Total RNA was extracted from each tissue of each animal on ice using Tri-Reagent (Bioline, UK) according to manufacturer's instructions, and purified using DNase and RNeasy columns (Qiagen, UK). All RNA samples were analysed for concentration and quality by spectrophotometer (NanoDrop, ND-1000) and tape station analyses (Agilent 2200 TapeStation). All samples were diluted to 30 ng μl − 1 total RNA, and 30 ng was used to synthesize cDNA following the manufacturer's protocol (Qiagen, QuantiTect Reverse Transcription Kit). cDNA was stored at −20°C until gene expression analysis.
A total of seven candidate genes were selected for tissue distribution gene expression analysis (Table 1). The Ribosomal 18s gene was selected as a housekeeping reference as recommended by previous work on M. truncata's sister species -Mya arenaria (Siah et al., 2008). Genespecific primers were designed for unique regions in each candidate using NCBI/Primer-BLAST (Ye et al., 2012) to produce single amplicons with a size of approximately 100-250 bp, an annealing temperature of 62-64°C and a GC content between 55 and 60%. PCR amplicons were sequenced to confirm identity.
For quantitative PCR (Q-PCR), lyophilised primers (Invitrogen) were reconstituted to 100 μmol with RNase-free water and mixed with Brilliant II SYBR® Green (Agilent, UK) following manufacturer's guidelines. Fluorescence was detected (Stratagene, Mx3000P) over 40 cycles with cycling conditions of 95°C for denaturing, primer-specific annealing 62-64°C, and extension at 72°C. All samples and standards were run in triplicate and each plate included triplicate H 2 O and no template controls. Standard curves of each gene were generated on each Q-PCR plate using four point, 2-fold serial dilutions of cDNA (from pooled cDNA). The efficiencies of the Q-PCR reactions were 90-110%, as determined using the slope of the standard curve (Efficiency (%) = [(10(slope / − 1)) − 1]).
Quantification of gene expression was conducted using the comparative CT method that normalises the gene expression of each sample in relation to an internal housekeeping gene (Ribosomal 18s). Evaluation of CT values for Ribosomal 18s across samples indicated it was an appropriate housekeeping gene, ie there was no significant difference in expression across the different tissue types. Normalized CT values were obtained by subtracting the CT value of the internal housekeeping gene from that of the candidate gene in the same sample (ΔCT). Differences between the average ΔCT and ΔCT of each sample were expressed as ΔΔCT. The fold changes (2 -ΔΔCT ) of candidate gene expression were compared across tissues.

Statistics
Data were non-normally distributed (due to the number of zeros or very low values) and could not be transformed to reach normality. Given the non-normal distribution and unbalanced design (mantle n = 5, siphon n = 4, gill n = 5, foot n = 3, digestive gland n = 5, gonad n = 5), data were compared using 95% confidence intervals around the mean average fold change and compared to zero, ie if the confidence interval overlapped with zero (indicating the fold change was equal to zero) or not. For additional stringency, mantle gene expression data were tested against a set median of zero using the nonparametric Wilcoxon Signed Rank Test.

M. truncata mantle transcriptome
RNA-Seq reads from the mantle tissue of nine animals were assembled to create 20,106 contigs in the final de novo transcriptome (reads Table 1 Candidate biomineralisation genes selected for tissue distribution analysis, putative annotations and primers used in Q-PCR. "top 50"=in the top 50 most highly expresssed, annotated mantle transcripts, "shell"=present in the shell proteome, "mantle"=present in the mantle proteome, "both"=present in both the shell and mantle proteome. available from NCBI SRA accession number: SRP064949, assembled contigs available at: http://bit.ly/1QcFiVH). 18.81% of contigs were assigned putative functions using BLAST sequence similarity searching (below an E-value of 1E−10). The top 50 most highly expressed, annotated, transcripts included many putatively involved in muscle contraction (40%) such as: Myosin, Paramyosin, Tubulin, Tropomyosin and Actin. Energy production was also a dominant process, with annotation in twelve transcripts (24%), eg NADH dehydrogenase and Cytochrome c oxidase. Other notable transcripts included three encoding putative biomineralisation genes -Calponin-3, Calponin-2 and Tyrosinase; and two chaperone genes -Heat shock protein 90-alpha 1 and Heat shock protein 70 (Table 2).
When the mantle transcriptomes of M. truncata and L. elliptica were compared using tBLASTx, 17.37% of the M. truncata contigs showed similarity to an L. elliptica contig (below an E-value of 1E−10; Supplementary Table 1). The top 50 most similar contigs included one notable  Table 2). All of the candidate biomineralisation genes selected for Q-PCR also showed strong sequence similarity (below an E-value of 1E − 15) to a L. elliptica contig (Supplementary Table 2).

Tyrosinase bivalve phylogeny
At least five putative Tyrosinase paralogues were identified in the M. truncata mantle transcriptome. The derived amino acid sequences were added to a previously published molluscan Tyrosinase phylogenetic analysis (Aguilera et al., 2014). Many of the nodes had low support values (Fig. 2, Supplementary Figs. 1, 2 and 3), however some reoccurring patterns were observed across the three phylogenetic models used. Two major clades (A & B) were resolved (although it should be noted that there is some confusion in the literature with regards to nomenclature which we believe explains clade discrepancies) in addition to two large, independent expansions in the taxa Crassostrea and Pinctada. In general the M. truncata transcripts clustered in clade A. One of the M. truncata transcripts (contig00553) clustered loosely with the TyrA3 genes from Crossostrea gigas and Pinctada fucata, whilst all other copies clustered with the L. elliptica gene, TyrB, and showed evidence of early expansion in the M. truncata genome.

Tissue distribution of biomineralisation gene expression
Seven candidate biomineralisation genes were identified for further analysis based on sequence similarity to known biomineralisation genes, some of which were also present in the top 50 most highly expressed annotated transcripts (Table 2), the shell or mantle proteome (Arivalagan et al., 2016in this issue), or any combination of the three (Table 1).
A mantle/siphon-specific signal was detected for all candidates (Fig.  3). None of the mantle 95% confidence intervals overlapped zero and a non-parametric Wilcoxon Signed Rank Test showed that all of the mantle gene expression values were above zero (P b 0.05). In contrast, all of

M. truncata mantle transcriptome
Presented here is the first substantial molecular resource for M. truncata, which is a valuable resource for biomineralisation and comparative studies. The M. truncata mantle transcriptome was similar in size (~20,000 contigs), composition of most highly expressed transcripts and percentage annotation to other previously characterised bivalve mantle transcriptomes (Clark et al., 2010;Freer et al., 2014;Joubert et al., 2010;Niu et al., 2013;Shi et al., 2013); in particular it shares some consistent features with that of the Antarctic clam L. elliptica. When the M. truncata and L. elliptica mantle transcriptomes were compared using tBlastx, 17.35% of contigs were shared (below an E-value of 1E − 10) representing a highly conserved core set of genes (Clark et al., 2010;Sleight et al., 2015). Both the M. truncata and L. elliptica mantle transcriptomes were heavily dominated by muscle related genes (Table 2; Myosin, Actin etc.), reflecting the contractile nature of this organ. In addition, putative mitochondrial respiratory chain genes (NADH dehygrogenase, Cytochrome c, Arginine kinase etc) were highly expressed in both, demonstrating that the mantle is a metabolically and transcriptionally active tissue. To date, our research has concentrated on L. elliptica and M. truncata has been chosen as a northern hemisphere, temperate comparison. M. truncata is often reported as an Arctic bivalve (Camus et al., 2002;Gillis and Ballantyne, 1999), however animals in the present study were sampled from a more southerly latitude, and hence much warmer region of their distribution on the North West coast of Scotland. M. truncata and L. elliptica are ecologically and morphologically very similar, but their physical environments (Arctic to temperate versus Antarctic), geographical extent (ranging from Arctic through subboreal to temperate versus Southern Ocean Fig. 3. Tissue distribution expression patterns of candidate biomineralisation genes determined via Q-PCR (mean average fold change ± 95% confidence intervals). Fold change calculated as 2 -ΔΔCT using Ribosomal 18s as an internal housekeeping gene. exclusively) and evolutionary history (phylogenetically distant relatives) differ significantly. In addition, the two species have different shell microstructures. As a result of their independent evolutionary trajectories in dissimilar physical conditions, with diverse selection pressures, M. truncata which inhabit temperate regions have a higher metabolic rate and shorter lifespan than L. elliptica (Camus et al., 2003(Camus et al., , 2005Peck et al., 2002;Philipp and Abele, 2010).
As well as sharing a core set of highly conserved genes, the M. truncata and L. elliptica most highly expressed mantle transcripts share more specific similarities at the individual gene level. Here we will focus on the genes likely to be involved in biomineralisation. Both transcriptomes have a single Tyrosinase gene in the most highly expressed set of transcripts, L. elliptica has Tyrosinase B and M. truncata has Tyrosinase A3. Tyrosinase is a biomineralisation protein involved in the formation of the shell matrix and periostracum (Huning et al., 2013;Sánchez-Ferrer et al., 1995;Zhang et al., 2006), its extreme high expression in the mantle organ of both clam species provides further evidence for its important role in shell deposition. A less well characterised biomineralisation candidate is Calponin, both species have two Calponin genes in the mostly highly expressed transcripts. Calponin proteins are typically involved in muscle contraction and interact closely with other muscle action proteins such as Actin and Myosin (Matthew et al., 2000). Its role in muscle contraction however, is thought to be primarily crosslinking and stabilisation of the muscle fibres (Jensen et al., 2014), and we hypothesise it could play a similar role in the stabilisation of the shell protein matrix. Several pieces of evidence support this idea: 1.) It is highly expressed at the transcript level in the mantle of the two clam species and a species of pearl oyster, Pinctada martensii (Shi et al., 2013). 2.) It has been demonstrated to be involved in the biomineralisation of bone in humans (Ueda et al., 2002). 3.) The shell matrix protein, PFMG8, has been shown (in silico) to contain a Calponin domain which has a calcium binding site (Evans, 2012). 4.) It has a mantle specific expression pattern (Fig. 3) and finally, 5.) Calponin proteins were found in both the shell and mantle proteome of M. truncata where it may interact with Myosin and contribute to shell elasticity (Arivalagan et al., 2016in this issue).
One noticeable difference between the M. truncata and L. elliptica mantle transcriptomes concerns the constitutive expression of heat shock proteins (HSPs). HSPs are involved in protein-folding and chaperoning and are either constitutively expressed, or induced in response to stress (Hartl, 1996). Both Heat shock protein 70 (Hsp70) and Heat shock protein 90-alpha (Hsp90) were highly expressed in the M. truncata mantle, where as in L. elliptica there was no high expression of any HSP family members (above 300 readswhich was the cut-off used by Clark et al., 2010). Hsp70 is classically regarded as inducible, rather than constitutive, and forms part of a classic "stress" response in many organisms (Clark et al., 2008;Clark and Peck, 2009). One possible explanation for the high background expression of inducible Hsp70 in M. truncata is that the animals in the present study were sampled from a southerly latitude and warm region of their global distribution (summer sea surface temperatures in Dunstaffnage Bay have been recorded N14°C) where they are close to their upper thermal tolerance (Amaro et al., 2005). Work by Amaro et al. (2005) demonstrated that M. truncata at its southern distribution limit (the Frisian Front in the North Sea experiencing water temperatures N14°C) have low numbers of ripe oocytes and frequent years of poor recruitment, indicating the southern populations could be in a constant state of low-level thermal stress. Hsp90 on the other hand, is thought of as a constitutively expressed protein, and its high expression in the M. truncata mantle transcriptome is likely to represent normal cellular processes in the mantle, similar to other invertebrate species (Huang et al., 2013).

Tyrosinase bivalve phylogeny
Tyrosinase is a multifunctional well-characterised, shell-associated protein which has been shown to have a functional role in the cross-linking of the soluble periostracum precursor (the periostracin) to form an insoluble periostracum (Waite et al., 1979); to be localised in the prismatic layer of shell (Nagai et al., 2007) and to be expressed in the pallial mantle and hence involved in the nacreous layer of the shell (Takgi and Miyashita, 2014). In addition, Tyrosinase is involved in numerous other biological processes such as innate immunity, pigmentation and wound healing (Aguilera et al., 2014;Zhang et al., 2006). We have identified at least five gene copies of Tyrosinase in the M. truncata mantle transcriptome which are likely to be a result of several gene duplication events followed by sub-functionalisation (Force et al., 1999). It is possible that an expansion and subsequent subfunctionalisation has produced Tyrosinase paralogues which have completely new functions besides shell formation, for example in the immune system, as suggested by Wang et al. (2009) andEsposito et al. (2012). Given the well-characterised and clearly important nature of Tyrosinase in molluscan shell formation, it is important to investigate its evolution in order to better understand how mollusc shell is produced between species. M. truncata Tyrosinase amino acid sequences were therefore investigated within a phylogenetic context. Similar to Aguilera et al. (2014) and other previous Tyrosinase phylogenies ( Aguilera et al., 2013), many of the nodes had low support values (Fig.  2, Supplementary Figs. 1, 2 and 3). Adding M. truncata sequence data to three independent phylogenetic analyses did not alter the overall tree topology and the major patterns described by Aguilera and colleagues were largely still resolvedfurther validating their work.
Four of the five M. truncata Tyrosinase amino acid sequences formed a well-supported cluster with L. elliptica Tyrosinase B. It is possible this cluster represents an expansion within the M. truncata genome. The fifth M. truncata sequence grouped with Tyrosinase A3 sequences from C. gigas and P. fucata. Given that one of the L. elliptica sequences group with the M. truncata sequences it is likely they have evolved under at least some similar selection pressures with regard to shell growth. M. truncata however, show evidence for a Tyrosinase expansion, which is absent in L. elliptica. Antarctic marine invertebrates are largely stenothermal due to evolution at constant cold temperatures for long periods of geological time (Rogers, 2007). We find that, for the Tyrosinase gene family, Antarctic clams are less diverse than their temperate counterparts M. truncata, and other marine shelledmolluscs. An explanation for higher Tyrosinase diversity in M. truncata than L. elliptica could be due to M. truncata's wider geographical distribution and spread over environmental gradients. As a species, M. truncata could require more diverse molecular machinery to cope with the diverse environments they inhabit; supporting the hypothesis that diversity is positively correlated to environmental heterogeneity and stress (Nevo, 2001;Van Valen, 1965).

Tissue distribution of biomineralisation gene expression
As expected, all of the putative biomineralisation candidate genes we selected for tissue distribution analysis (Table 1) showed a mantle/ siphon-specific expression pattern (Fig. 3). Tyrosinase and Pif are well characterised shell proteins (Nagai et al., 2007;Suzuki et al., 2009); they have recently been shown to respond to shell damage in L. elliptica (Sleight et al., 2015), and the mantle/siphon specific expression patterns found in M. truncata provide further evidence to support their hypothesised functional role in shell deposition in the two clam species. Calponin is likely to be involved in biomineralisation (as discussed at length in Section 4.1) however, like many biomineralisation proteins, it is multi-functional and also involved in muscle contraction where it interacts with Myosin and hence it also showed variable expression in the foot (a muscle). Cartilage matrix proteins are involved in calcium phosphate biomineralisation in vertebrates where they bind to calcium phosphate crystals and form part of an extracellular matrix (Acharya et al., 2014). Arivalagan et al. (2016in this issue) identified Cartilage matrix protein in the M. truncata shell proteome, and taken together with the mantle-specific gene expression pattern found in the present study, it is likely to play a similar matrixlike role in mollusc biomineralisation. Chitinase is an enzyme hypothesised to be involved in mollusc shell matrix construction (Sleight et al., 2015), as well as immunity (Badariotti et al., 2007). The function of Chitinase and Chitinase-like proteins have been investigated in arthropods with chitinous exoskeletons; they are typically involved in moult-cycles, wound healing and tissue repair (Bonneh-Barkay et al., 2010;Chen et al., 2004), however their exact function in the mollusc shell matrix is still unclear. Previous work shows Chitinase expression is up regulated in response to injury after 21 days in young, but not old, L. elliptica (Husmann et al., 2014); and Sleight et al. (2015) also found variable expression in L. elliptica over time in response to shell damage. Very high expression was found in young damaged animals after one week, whereas no expression was detected in control or damaged adult animals. The mantle specific expression pattern found in the present study, as well as Chitinase presence in the M. truncata shell proteome (Arivalagan et al., 2016in this issue), provides further support for its active involvement in shell deposition. More research is required to understand its hugely variable expression in young versus old animals and its exact function during matrix formation and shell secretion.
Complement control proteins are multifunctional and involved in both the immune system (Ferreira et al., 2010) and possibly biomineralisation (Arivalagan et al., 2016in this issue). Bivalve shell and mantle combined act as a barrier to the external environment, and as such are likely to be entwined with immune processes. Disentangling immune and biomineralisation mechanisms represents a significant challenge for researchers trying to understand how molluscs build their shells. The challenge is partly due to the dual role of haemocytes both as immune cells and hypothesised calcium carbonate chaperones (Mount et al., 2004). In addition to the observed mantle specific expression of Complement control protein in the present study and previous reports of immune genes such as Mytilin responding to damage in L. elliptica (Sleight et al., 2015), Arivalagan et al. (2016in this issue) found immune proteins in the M. truncata shell proteome (and verified their presence was not due to contamination). The expression of immune genes in the M. truncata mantle, as well as the immune proteins found in its shell, could be explained in several ways: 1.) general haemocyte circulation in the mantle could result in coincidental incorporation into the shell as an accidental bi-product of their immune function (as per the mantle's role as a barrier); 2.) whilst haemocytes actively deposit calcium carbonate to the shell secretion site they could be coincidentally trapped in the shell matrix space, as an accidental bi-product of their role in biomineralisation, as proposed by Arivalagan et al. (2016in this issue); or 3.) immune proteins could serve a genuine dual-functional role, both in aiding biomineralisaion during calcium carbonate secretion from haemocytes, and possibly also at a structural matrix-level, as well as fighting infection as an anti-microbial peptide in the shell, and in the circulating haemocytes in the mantle.

Conclusion
We present the first substantial molecular resource for M. truncata. The mantle transcriptome was 454-sequenced, de novo-assembled and BLAST sequence similarity-annotated to produce a total of 20,106 contigs, of which approximately 19% were assigned putative functions. The mantle transcriptomes of M. truncata and the Antarctic clam (L. elliptica) were compared using tBLASTx and overall, shared a core complement (17%) of highly conserved transcripts. Looking at the most highly expressed genes in the two species showed that many of the dominant biological functions (contraction, energy production, biomineralisation) were conserved. The Tyrosinase proteins from M. truncata were analysed phylogenetically and showed a small expansion which was closely related to L. elliptica, however M. truncata had more diversity in Tyrosinase proteins. The tissue distribution expression pattern of candidate biomineralisation genes was investigated using Q-PCR, all genes showed a mantle specific expression pattern supporting their hypothesised role in shell secretion. We provide very preliminary insights on how clams in different conditions (temperate versus polar) build their shellsa topic which we will continue to pursue in our research group. In addition, we provide a valuable molecular resource for future comparative studies investigating biomineralisation.