High spatial resolution mapping of the mucosal proteome of the gills of Crassostrea virginica: implication in particle processing

Living organisms are unceasingly exposed to challenging physical, chemical and biological factors, so that their fitness, and often their survivorship, relies on the efficiency of the barriers they have built. In the case of the metazoa, mucosal secretions associated with epithelial layers represent the first line of defense against various attacks. Mucus is secreted by all epithelia containing living cells on their surfaces such as on the internal organs of vertebrates (e.g. gastrointestinal or tracheobronchial tracts) and the epidermis of fish, amphibians, cnidarians and mollusks. It is made of mucin-like molecules, water, electrolytes, epithelial and blood cells and a wide range of bioactive molecules produced by mucus-secreting cells (Schachter and Williams, 1982; Simkiss and Wilbur, 1977). The consistency, viscosity and elasticity of mucus are generally attributed to the concentration of polymers (e.g. mucopolysaccharides, mucins, mucin-like glycoproteins) within the gel and to the physical entanglement of these polymers with other molecules (Cone, 2009; Rose et al., 1984; Smith and Morin, 2002).

In mollusks, mucus has a central role in multiple biological functions (Davies and Hawkins, 1998), including locomotion and navigation (Denny, 1989; Prezant and Chalermwat, 1984; Smith et al., 1999), attachment (Smith, 2002), protection against freezing (Hargens and Shabica, 1973) or desiccation (Denny, 1989; Wolcott, 1973), and defense against predators (Gavagnin et al., 1994; Gustafson and Andersen, 1985). The energy allocated to mucus production in mollusks can exceed 15% of the total energy gained from food, highlighting the importance of mucus in the biology of these animals (Davies and Hawkins, 1998). One of the most important biological functions of mucus in bivalve mollusks (e.g. oysters, mussels, clams) is interaction with microbes (reviewed by Allam and Pales Espinosa, 2015). For instance, the mucus layer covering the surface of bivalve pallial organs (organs present in the shell cavity such as gills, mantle) is the first constituent encountered by waterborne microbes that attach to these organs before the establishment of mutualistic (symbionts) or parasitic associations (Allam et al., 2013; Burreson and Ford, 2004; Dahl et al., 2010; Dubilier et al., 2008). Furthermore, previous studies have shown that molecules present in mucus contribute to the establishment and the success or failure of many of these host–microbe associations (Allam et al., 2013; Dufour, 2005; Kremer et al., 2013; Pales Espinosa et al., 2014, 2013; Southward, 1986).

Mucus is also commonly used by mollusks to capture and transport particles on ciliated epithelia for cleansing and feeding (Barille and Cognie, 2000; Beninger et al., 1993; Morton, 1977; Urrutia et al., 2001). In suspension-feeding bivalves, particles are captured by the gills, embedded in mucus and transported on the feeding organs (i.e. gills, labial palps) to be either rejected as pseudofeces or directed to the mouth and ingested (Beninger et al., 1993; Urrutia et al., 2001; Ward et al., 1993). This food particle sorting mechanism has been well described for over a century and is considered to represent an important strategy allowing bivalves to optimize energy gain by ingesting nutrient-rich particles while rejecting poor quality ones in pseudofeces (Allen, 1921; Bayne et al., 1993; Cognie et al., 2001; Newell and Jordan, 1983). Although the precise mechanism of sorting in suspension-feeding bivalves remains unclear, in situ observations demonstrated that mucus covering bivalve feeding organs plays an important role in particle processing as a vehicle for particle capture, post-capture transport, ingestion and rejection steps (Beninger et al., 1993; Riisgard et al., 1996). Particles directed as pseudofeces are embedded in cohesive mucus and rejected back into the environment via a ventral tract entangling unwanted live cells, debris, and abiotic material of low nutritional value. Those directed for ingestion are transported to the mouth via a dorsal tract in a low viscosity mucus (Beninger et al., 1992; Ribelin and Collier, 1977). But mucus is not just a mere carrier for food particles: recent investigations showed that specific interactions take place between mucus and food particles, mediating particle selection. In this context, our previous work demonstrated that mucus covering the feeding organs of the eastern oyster Crassostrea virginica (Pales Espinosa et al., 2009) contains sugar-binding proteins (i.e. lectins) that differentially bind microalga cell surface carbohydrates (MCSCs), triggering particle selection (Pales Espinosa and Allam, 2018). Moreover, a series of feeding experiments showed that oysters preferentially ingest particles covered with glucose and mannose residues, as a likely result of interactions between MCSCs and mucosal lectins present on their feeding organs (Pales Espinosa et al., 2016a).

A low-resolution reference map of proteins found in pallial mucus covering the gills, labial palps and mantle of C. virginica has already been generated, highlighting the presence of a wide variety of putative adhesion/recognition molecules (Pales Espinosa et al., 2016b). Although this first analysis reveals the presence of multiple lectins, the sampling approach used generated bulk mucus from each organ without the spatial resolution needed to gather information about the specific role of mucus in particle selection. The current study was designed to generate a high spatial resolution of the proteomic composition of the mucus that covers the gills of C. virginica, with an emphasis on the ventral (i.e. associated with the rejection of particles) and dorsal tract (i.e. associated with the ingestion of particles). This was mainly accomplished using endoscopy-aided micro-sampling of mucus from each tract in vivo, followed by multidimensional protein identification technology and complemented by in situ hybridization of candidate proteins. Our working hypothesis was that the functional disparities between both gill tracts is the result of differences in the proteomic make-up of mucus present in each tract.

Adult Crassostrea virginica (Gmelin 1791) (80–100 mm in length, n=15) were obtained from a commercial source located on Long Island Sound, NY, USA in September 2014, cleaned of epibionts and maintained in a flow-through system using natural seawater pumped from Long Island Sound (∼20°C) until use. To prepare animals for the procedure, a small section of the inhalant margin of the upper and lower valves of each specimen was carefully trimmed without damaging the underlying mantle tissue. Oysters were then allowed to recover for at least 1 day before mucus collection. During the mucus collection procedure, bivalves were placed in an aerated assay chamber (∼1 liter) filled with filtered seawater at ambient temperature (∼20°C). Mucus samples were collected in vivo using a micropipette connected to a peristaltic pump. The sampling pipette was mounted on a micromanipulator and positioned with the aid of an endoscope (for complete procedure, see Ward et al., 1991, 1998). Mucus samples were collected from the ventral and dorsal tract of the gills (Fig. 1B,F) and kept on ice during the procedure. Simultaneously, a cocktail of general protease inhibitors (50 µl of 1× solution prepared following manufacturer's recommendation per 50 ml mucus, S8820, Sigma) was added into each sample during the collection to prevent protein degradation. A total of 22 samples (i.e. 11 from each tract, volume ranging from 5 to 57.5 ml, Table S1) were collected, immediately frozen and stored at −80°C until analysis that happened within a week.

Before protein analysis, samples from the same tract were randomly combined (final volume ranging from 80 to 97.5 ml) into a total of 3 pools (Table S1) and concentrated by filtration (Amicon Ultra-15 Centrifugal Filter Units with Ultracel-3 membrane, Millipore, Burlington, MA) as per the manufacturer's recommendations. A quality control step was implemented to check protein abundance. An aliquot (25 µl) was mixed with 25 µl of 2× Laemmli sample buffer (Bio-Rad, Hercules, CA) heated to 100°C for 10 min and separated on a precast 12% Tris-Glycine gel (Jule Biotechnologies, Inc., Milford, CT). After electrophoresis, gels were stained using a standard silver stain protocol. The volume of the concentrated protein solution was finally reduced to 100 µl using a Speed-Vac. Non-protein components were removed from the protein solution by deoxycholate-TCA precipitation using a modification of the method of Peterson (1977). The resultant protein pellet was dissolved in 20 µl 8 mol l⁻¹ urea, 25 mmol l⁻¹ NH₄HCO₃. Protein concentrations were determined using the Peterson modification of the Lowry assay (Peterson, 1977). The protein solution was subjected to trypsin digestion as follows: reduced in 4 mmol l⁻¹ DTT (30 min, room temperature), alkylated in 8.4 mmol l⁻¹ iodoacetamide (30 min, room temperature in the dark), before the urea concentration was reduced to 1.7 mol l⁻¹ and the solution incubated 16 h at 37°C in the presence of trypsin Gold (Mass Spectrometry Grade, Promega, Madison, WI) at >1 µg/40 µg protein. After incubation, the digest was added with 2% formic acid.

The samples were analyzed for protein content using a modification of the multidimensional protein identification technology (MUDPIT) method (Washburn et al., 2001). Samples were pressure bomb loaded through the proximal end of a ‘mudpit’ column constructed of 250 µm ID fused silica tubing (PT Polymicro Technologies, Phoenix, AZ) with Kasil frit at distal end. The column was packed with 3 cm of strong cation exchanger (SCX, 5 µm) matrix (Whatman) distally and 3 cm C₁₈ matrix (5 µm ProntoSil 120-5-C18H, Bischoff Chromatography, Leonberg, Germany) proximally. Following sample loading, the column was washed for 10 min with Buffer A [2% acetonitrile (ACN), 0.1% formic acid (FA)] at 300 nl min⁻¹. The mudpit column was connected with a microtee to a fritless electrospray interface (Gatlin et al., 1998) feed column for automated microcapillary liquid chromatography-tandem mass spectrometry. The nano electrospray feed column to the mass spectrometer consisted of a fused-silica capillary (100 µm ID) which was pulled using a P-2000 CO₂ laser puller (Sutter Instruments, Novato, CA) to a 5 µm ID tip and packed with 10 cm of 5 µm ProntoSil C18 matrix using a pressure bomb and subsequently equilibrated in Buffer A.

The dual column construct was placed in line with an Eksigent 2D NanoHPLC unit flowing at 300 nl min⁻¹. The HPLC separation was provided by a 13 step, three component gradient. Each step consisted of the following, in sequence: 5 min wash with 100% Buffer A; 5 min wash with a fixed percentage of Buffer C (0.5 mol l⁻¹ ammonium acetate, in Buffer A); 10 min wash with 100% Buffer A; 60 min gradient of 0% to 40% Buffer B (90% ACN, 0.1% FA); 30 min wash, 100% Buffer A. The 13 steps varied the fixed Buffer C from 0 to 100%. The application of a 2.2 kV distal voltage electrosprayed the eluting peptides directly into an LTQ Orbitrap XL ion trap mass spectrometer (Thermo Fisher, San Jose, CA). Full mass spectra (MS) were recorded on the peptides over a 400 to 2000 m/z range at 60,000 resolution, followed by five tandem mass (MS/MS) events sequentially generated in a data-dependent manner on the first, second, third, fourth and fifth most intense ions selected from the full MS spectrum (at 35% collision energy). Charge state dependent screening was turned on, and peptides with a charge state of +2 or higher were analyzed. Mass spectrometer scan functions and HPLC solvent gradients were controlled by the Xcalibur data system (Thermo Fisher, San Jose, CA).

MS/MS spectra were extracted from the RAW file with ReAdW.exe (http://sourceforge.net/projects/sashimi). The resulting mzXML data files were searched with The GPM X! Tandem (The GPM, thegpm.org; version CYCLONE 2013.02.01.1) against a custom proteome database (48,093 entries) built using the Crassostrea virginica open reading frames produced from the oyster transcriptome generated by McDowell et al. (2014). This approach excludes proteins potentially derived from microbes associated with pallial mucus. Fixed cysteine carbamidomethylation and optional methionine oxidation and threonine, serine and tyrosine phosphorylation were applied during the search. Only peptides with a P value of ≤0.01 were analyzed further. In addition, a Decoy database (all proteins in reverse order) was also added from this database with compass (Wenger et al., 2011). This database was searched with a fragment ion mass tolerance of 0.40 Da and a parent ion tolerance of 1.8 Da. Scaffold (v.4.4.3, Proteome Software Inc., Portland, OR) was used to validate MS/MS based peptide and protein identifications. Peptide identifications were accepted if they could be established at greater than 95.0% probability by the Scaffold Local FDR algorithm. Protein identifications were accepted if they could be established at greater than 5.0% probability to achieve an FDR less than 1.0% and contained at least 2 identified peptides. Protein probabilities were assigned by the Protein Prophet algorithm (Nesvizhskii et al., 2003). Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony. Proteins sharing significant peptide evidence were grouped into clusters. The sequences of the proteins identified in the 2 mucosal tracts were then uploaded into the Blast2GO application (Götz et al., 2008) to be annotated.

Adult C. virginica (80–100 mm in length, n=6) were obtained from a commercial source located on Long Island Sound, NY, in November 2017. Oyster gills were dissected and fixed in 10% neutral buffered formalin for 48 h before being dehydrated in an ascending ethanol series, embedded in paraffin blocks and cut in serial sections (5 µm thickness). Four consecutive sections were processed for standard Hematoxylin and Eosin staining (1 section) or for in situ hybridization (ISH, 3 sections). RNA in situ hybridization assays were performed using RNAscope^®, an RNA in situ hybridization technique previously described by Wang et al. (2012). The RNAscope^® 2.5 HD Red Reagent Kit (Advanced Cell Diagnostics, Newark, CA) was applied in accordance with manufacturer's instructions. Three probes custom-synthesized by the manufacturer were used: one against the mucosal C-type lectin (CvML3912, Pales Espinosa and Allam, 2018), chosen because of its abundance in oyster mucus and its demonstrated role in particle selection in oyster; a control probe targeting oyster 18S (Cv18S-rRNA; X60315.1) was used to assess RNA integrity and evaluate ISH reaction success; a negative control probe specific to the bacterial dapB gene (EF191515) was also used to evaluate non-specific binding and background staining. Slides were then counterstained and observed using a microscope. Thirteen plicae were selected to enumerate CvML3912-positive cells on the apex [4 apical ordinary filaments (AOFs)] and on the side [principal filament (PF) and lateral ordinary filaments (LOFs)] of each plica (Fig. 1C).

Statistical analysis compared protein composition in mucus collected from the ventral and dorsal tracts. For downstream analysis, only proteins for which at least 2 unique peptides were identified, that were present in two out of six samples, and that presented a log_e value <−9 were considered (40% of the initial proteins). Standardization to the sum of proteins identified in the corresponding sample was realized and protein abundance data were analyzed in MultiExperiment Viewer software (MeV, v.4.9). Significance Analysis of Microarray (SAM) methods were used to identify proteins differentially abundant in samples from the different tracts following the approach described by Roxas and Li (2008). A gene ontology (GO) enrichment analysis was performed in Blast2GO using the Fisher's Exact Test (P value of 0.05 as cut-off) to compare protein found in the ventral and dorsal tracts against the total proteome.

Several proteomics studies on mollusk shell (Mann and Jackson, 2014; Marie et al., 2012) and mucosal secretions (Caruana et al., 2016; Pales Espinosa et al., 2016b) have provided reference maps allowing for further exploration of suspected processes and functions attributed to mucosal proteins. In this new study, a total of 1833 proteins matching C. virginica predicted proteins were identified in samples collected from the two mucosal tracts combined (Table S2). Several stringent selection criteria (see Materials and Methods) yielded 735 selected proteins that were finally grouped into 14 categories (Fig. 2) based on their functional annotation (NCBI database) and a complementary search using Blast2go (GO terms, Enzyme Codes, IPR). Some of these proteins have an intracellular origin (e.g. tubulin or ribosomal proteins) likely because of the presence in the pallial mucus of hemocytes and exfoliated epithelial cells, but also to the transudation of plasma components into pallial mucus. In contrast, 56 of these 735 proteins match the GO terms ‘extracellular’, ‘cell-matrix adhesion’ and ‘integral to membrane’ and 155 additional ones present a signal peptide or a transmembrane domain (SignalP-TM; Table S3) suggesting that these particular proteins are secreted. Among the most abundant proteins present in the combined tracts of C. virginica gills (Table 1), the dominin (extracellular superoxidase dismutase, BAF30874; Itoh et al., 2011; Xue et al., 2019) and its isoforms represent about 60% of the total spectral counts. The major plasma protein 2 (also known as segon, AFH41574; Xue et al., 2019, 2012) is the second most abundant protein with 5.6% of the total spectral counts. This finding is in agreement with a previous study showing that the two proteins represent about 70% of the total proteins from oyster hemolymph (Xue et al., 2019). They are both suspected to be involved in shell formation, possibly explaining their presence in pallial mucus (Xue et al., 2019). The 20 most abundant proteins in both tracts also include a carbonic anhydrase 2-like (2.3% of the total spectral count, XP_011434938) as well as proteins involved in cytoskeletal filaments structure (e.g. actin, tubulin, calmodulin-like), cell matrix formation (e.g. SCO-spondin), adhesion or recognition (e.g. x-box binding, c-type mannose receptors 2), proteolysis or cytolysis (e.g. aminopeptidase-like, lysozyme 2), as well as several hypothetical proteins.

While most of the 735 proteins identified in mucus were common to both tracts, the relative abundance of 56 proteins was found to be significantly different between both sample types (i.e. the ventral tract versus dorsal tract, Table 2, Table S4). Among the 56 tract-specific proteins, the abundance of 34 was significantly higher in the dorsal tract compared to the ventral, while 22 proteins were higher in the latter (Table 2). An enrichment analysis showed that two categories (i.e. ‘carbohydrate binding, recognition’ and ‘mucus layer’) were particularly enriched compared to others (Fig. 3). These proteins were grouped based on available information and are discussed from the lens of oyster interaction with waterborne microbes (e.g. particle transport and selection process, microbial neutralization and digestion).

In the present study, results show that the mucus in the ventral tract is characterized by a high abundance of 12 SCO-spondin/mucin-like, with 3 of these proteins being significantly higher than levels detected in the dorsal tract (GO terms ‘mucus layer, glycosylation’, Table 2, Table S3). The overall standardized spectral count of mucin-like proteins reached 483 in the ventral tract versus 273 in the dorsal tract. In addition to the mucin-like proteins, mucus from the ventral tract is also characterized by the presence of a large number of proteins with adhesive property. For instance, the normalized total count of proteins with the GO term ‘cell adhesion’ was 63% higher in the ventral tract (465 spectral counts, Table S3) than in the dorsal (296 spectral counts), with the presence of numerous proteins containing von Willebrand factor (VWA) and EGF domains, including matrilin-like protein as well as a significant enrichment in galactose-binding lectin (galactoside-specific lectin) and fucose-binding lectin (fucolectin-7-like; Table 2, Table S3).

In contrast, the mucus from the dorsal tract is particularly enriched in proteins characterized by the GO term ‘carbohydrate binding’ (Table S3). The overall standardized spectral count of this group of 20 proteins in the dorsal tract (368) was more than double that in the ventral tract (168) and the abundance of 16 of these proteins was significantly higher in the dorsal tract as compared to the ventral tract. These included three C-type lectins: C-type lectin 2-like protein and two C-type mannose receptors 2 (Table 2). Among the C-type lectins found in the dorsal tract (Table S3), the three C-type mannose receptors 2 (cds.c102634_g1_i1|m.3912/CvML3912; Pales Espinosa and Allam, 2018; cds.c102634_g1_i2|m.3914; cds.c102634_g1_i3|m.3915) present highly similar protein sequences (e.g. carbohydrate recognition domain and determinant motifs for calcium binding and sugar specificity).

In order to better understand why these C-type lectins are more abundant in the dorsal tract, the precise localization of the cells that produce the CvML3912 mRNA and cognate protein, in situ hybridization (ISH) was performed on oyster gills and results are presented in Fig. 4. Positive cells were recorded along the different types of gill filaments (Fig. 1C,E) and an average number was calculated for the apical zone (apical ordinary filaments, 15.6±0.73) and the lateral zone (i.e. lateral ordinary and principal filaments, 19.5±0.79). The results indicate a higher number of positive cells on the lateral zone of the plicae as compared with the apical zone, supporting the enrichment of the fluid mucus present in the dorsal tract with CvML3912, and possibly other isoforms or closely related C-type lectins.

A main function of gills in suspension-feeding bivalves (e.g. oyster C. virginica), in addition to respiration, is the capture and transport of food particles (Atkins, 1936, 1937; Galtsoff, 1964; Ribelin and Collier, 1977). Using elaborate ciliary mechanisms, particles captured on gills are directed either to a dorsal tract (i.e. basal ciliated tract, Fig. 1) or to a ventral tract (i.e. marginal ciliated groove). More specifically, particles reaching the principal filaments of the gills are carried to the dorsal tract while particles attaining the ordinary filaments, and most specifically the apical ordinary filaments, are either directed to the ventral tract in the counter-current created by cilia beating ventrally [Fig. 1E, frontal median tract (FMT); Beninger et al., 2005; Beninger and StJean, 1997] or to the dorsal tract in the current created by cilia beating dorsally [Fig. 1E, two frontal lateral tracts (FLTs); Beninger et al., 2005; Ribelin and Collier, 1977], but in this last case, always via the principal filaments (Ward et al., 1994). Overall, most particles traveling via the ventral tract are fated for rejection in pseudofeces while particles trapped in the dorsal tract are directed to the mouth via the palps (where secondary sorting can occur) to be ingested, even though this mechanism is also dependent on particle concentration, particle size and on the satiation status of the bivalves (Beninger et al., 1992; Cognie et al., 2003; Ward et al., 1994). Regardless of the path the particles follow, their transport is enabled by the presence of mucus. In metazoans, mucus is often three-dimensionally structured with the presence of two distinct layers covering epithelial cells (Ross and Corrsin, 1974). The inner layer is in direct contact with the epithelial cells and is often made of low viscosity mucus that allows cilia beating. The outer layer is typically made of non-continuous and viscous secretions that entrap particles and is directed by cilia movements. This two-layer model has been confirmed in bivalves (Beninger et al., 1997), and is particularly relevant along the rejection pathway (i.e. ventral tract) where particles are embedded in viscous rafts of mucus floating on a low-viscosity mucus. This viscous mucus (enriched in acid mucopolysaccharides) is mainly produced by a high number of mucocytes lining the epithelium of the apical ordinary filaments (Beninger and Dufour, 1996; Beninger and StJean, 1997). Among the molecules known to affect mucus viscosity (Girod et al., 1992), the mucin/mucin-like proteins, their relative concentrations, and the degree of their glycosylation and hydration have been found to be the most critical (Cone, 2009; Lai et al., 2009; Linden et al., 2008). For example, the viscosity of a gel made of mucins from the giant West African snail (Archachatina marginata; Momoh et al., 2019) or from the coral (Montastrea faveolata; Jatkar et al., 2010) increases with the increase in mucin concentration. Our results showing a high abundance of mucin-like proteins and other proteins with adhesive properties (e.g. proteins with VWA and EGF domains) in the ventral tract are in agreement with earlier studies reporting a higher viscosity of this mucus as compared to mucus from the dorsal tract. Even though the exact roles of the VWA and EGF domains are not well known, several studies highlighted the fact that they may cross-link components of the mucus, therefore increasing mucus adhesion (Li and Graham, 2007; Smith et al., 2017).

The ventral tract is an area particularly exposed to water turbulence and currents and therefore the high viscosity of mucus present in this area helps trap particles aimed for rejection (Beninger and Dufour, 1996). It is noteworthy that the ventral tract was enriched in galactose- and fucose-binding lectins (adhesive properties), suggesting that particles having cell surfaces rich in galactose or fucose residues may preferentially bind to mucus present in this rejection tract. These results corroborate our previous findings showing that microalgae with cell surfaces rich in galactose-related carbohydrates are preferentially rejected in pseudofeces (Pales Espinosa et al., 2016a).

In contrast, the mucus present in the dorsal tract (i.e. ingestion pathway) originates from the principal filament troughs and the lateral ordinary filaments (Fig. 1C), both described as having a low density in mucocytes (Beninger and Dufour, 1996). This mucus, as discussed above, is characterized by a low abundance of mucin-like proteins and displays a lower viscosity than the mucus from the ventral tract (Beninger and Dufour, 1996; Beninger and StJean, 1997). It can easily flow toward the plicae trough (i.e. principal filament) driven by water currents that enter oyster pallial cavity in a ventrodorsal orientation and are deflected laterally towards the plical troughs (Fig. 1D). This low viscosity mucus carries a multitude of molecules produced by the filament cells into the plical troughs (above the principal filament) while avoiding the dispersal and loss of these molecules in the surrounding seawater. If the mucus from the dorsal tract is less viscous as compared to the ventral mucus, its adhesive property seems to be high and specific. These characteristics are likely due to the presence of numerous C-type lectins and most specifically several ‘C-type mannose receptors 2’, whose affinity for mannose/glucose residues was demonstrated for at least one of them (i.e. cds.c102634_g1_i1|m.3912; Pales Espinosa and Allam, 2018). Our results showed that this lectin is produced by the cells of the principal and ordinary filaments and is, on average, more abundant in the area made by the lateral ordinary filaments (Fig. 4). mRNAs of several C-type lectins have already been reported in the epithelium of the pallial organs and digestive tract of different bivalve species (Pales Espinosa et al., 2010; Yamaura et al., 2008) and further proteomic analysis of C. virginica confirmed that some of these cognate proteins were secreted into the pallial mucus (Pales Espinosa et al., 2016b). This new high spatial resolution analysis allowed the detection of additional lectins, including CvML, a C-type lectin previously suspected to be secreted into mucus (i.e. C-type lectin 1; Jing et al., 2011; Table S3). These findings support a high specificity of mucus from the dorsal tract to a certain type of ligand. In particular, the marked enrichment of mannose/glucose-binding lectins in the dorsal tract (i.e. ingestion pathway) is in line with previous results showing preferential ingestion of microalgae having mannose/glucose on their cell surface (Pales Espinosa et al., 2016a).

Altogether, the results presented in this study, in conjunction with previous information (Beninger et al., 2005; Pales Espinosa and Allam, 2018; Pales Espinosa et al., 2016a), provide a fine-scale mechanistic explanation for the particle selection process in oysters and likely other suspension-feeding bivalves (Fig. 5). Particles (e.g. microalgae, unicellular parasites, debris) not predominantly covered with mannose/glucose residues are more likely to be trapped by the thick mucus present at the apex of the plicae, possibly via adhesion proteins, and directed to the ventral tract to be rejected in pseudofeces. By contrast, particles covered with mannose and glucose residues are more likely to be bound by mannose/glucose-binding lectins (e.g. C-type mannose receptors 2) present abundantly in the fluid mucus that flows along the plical troughs and would be then directed to the dorsal tract for ingestion.

Pallial mucus in bivalves is involved in the processing of an extraordinarily large number of waterborne microbes that enter the pallial cavity. Some of these microbes will serve as food as described above but others may be harmful for the health of these animals. The role of mucus in host–microbe interactions and animal protection is now well recognized across various taxa and has gained prominence in the past few decades as a main component of the innate and acquired immune system (Allam and Pales Espinosa, 2015; Russell et al., 2015). In C. virginica, mucosal secretions are not only an excellent physical barrier but also contain host defensive cells (Lau et al., 2017) and a multitude of bioactive compounds (Pales Espinosa et al., 2016b) that act against microbe proliferation.

In this study, the analysis of both ventral and dorsal tracts revealed the presence of numerous proteins involved in host-microbe interactions, and more specifically, in defense against pathogens. For example, several proteins regrouped under ‘lysozymes, proteases and peptidase inhibitors’ were more abundant in the ventral tract (Table S3). This is for example the case of 3 peptidase inhibitors (kazal-type serine proteinase inhibitor, metalloproteinase inhibitor 1-like and pancreatic trypsin inhibitor) that were significantly more abundant in the ventral tract as compared to the dorsal tract (Table 2). Protease inhibitors regulate the activity of peptidases (Rawlings et al., 2004) and are considered to represent determinant resistance factors against infectious diseases in mollusks by preventing the harmful activity of exogenous proteases produced by invading microorganisms (La Peyre et al., 2010; Xue et al., 2006; Yu et al., 2011).

By contrast, other proteins linked to defense against pathogens were more abundant in the dorsal tract as compared to the ventral tract (Table S3). This is, for example, the case for three ‘peptidoglycan recognition proteins’ (i.e. PGRPs) and two lysozymes (cds.c101519_g1_i1|m.3174, cds.c113002_g1_i1|m.21368) known to have a bacteriolytic role in several bivalves (Maginot et al., 1989; Su et al., 2007), including oysters (Cronin et al., 2001; Itoh and Takahashi, 2008; Xue et al., 2010). Similarly, the abundance of the ‘complement c1q tumor necrosis factor-related protein 3’ (i.e. C1q-TNF, cds.c110839_g1_i1|m.15938), several ‘complement C1q-like’ (e.g. cds.c114572_g2_i1|m.25743), a ‘β-glucan-binding protein’ (i.e. BGBP, cds.c122006_g1_i2|m.52775) and two ‘peptidases’ (kyphoscoliosis peptidase and tolloid-like protein 2 harboring a trypsin domain, Table 2) were found to be significantly higher in the dorsal tract. In higher vertebrates, C1q-TNF family (e.g. CTRP3) is thought to mediate a large number of biological processes, including inflammation and glucose homeostasis (Li et al., 2011). In bivalves, this family of proteins is considered as an essential contributor to non-self recognition and immunity (Gestal et al., 2010). Similarly, the BGBP family is well known to play a significant role in invertebrate immunity (Vargas-Albores and Yepiz-Plascencia, 2000), including in bivalves (Liu et al., 2014). Furthermore, the abundance of the glycosyl hydrolase ‘β-galactosidase-1-like protein’ (cds.c122884_g1_i1|m.56982; Table 2), which is known to catalyze the hydrolysis of galactosides into monosaccharides, was also found to be significantly higher in the dorsal as compared to the ventral tract. β-galactosidases are produced by hemocytes (Moore and Gelder, 1985) as well as the secreting cells (i.e. apocrine cells) located in the epithelium of the digestive tract (e.g. esophagus, intestine) of mollusks (Martin et al., 2011). They actively participate in the intracellular digestion of microbes after phagocytosis (Moore and Gelder, 1985). This enzyme could contribute to the early digestion of microbes although its source in the dorsal tract is unknown but might be related to a possible higher abundance of specific secretory cells along the lateral ordinary filaments and/or principal filaments of the gill plicae.

A difference between the ventral and dorsal tracts was also found for the abundance of proteins whose functions in bivalve mollusks are not well defined, making the interpretation of the findings tentative (Table 2). This was the case, for example, for two proteins involved in intracellular transport (‘sodium-coupled neutral amino acid transporter partial’ and ‘sodium-dependent phosphate transport protein 2b’, also called NaPi2B; Table 2) that were significantly more abundant in the ventral tract as compared to the dorsal tract. In humans, neutral amino acid transporters (e.g. SLC1A5) are suggested to play an important role in amino acid depletion in mucus that cover lungs in order to deprive pathogenic organisms from their nutrients, limiting their propagation (Mager and Sloan, 2003). In vertebrates, NaPi2b has been suggested to play a role in the synthesis of surfactant in lung alveoli (Hashimoto et al., 2000) whose main role is to facilitate respiration. A similar role in facilitating gas exchange between gill cells and their environment could be suggested for this protein in bivalves.

As another example, two proteins involved in the ‘chitin metabolic process’ (peritrophin-1-like and uncharacterized protein loc101848577) were significantly more abundant in the ventral tract as compared to the dorsal (Table 2). Chitin is known to play an important role in shell formation in oysters (Suzuki et al., 2007; Zhang et al., 2012) even though this complex mechanism is not well understood. In addition, some peritrophins from insects possess one or several highly glycosylated mucin-like domains (Hegedus et al., 2016; Wang and Granados, 1997), which may contribute to mucus viscosity and possibly explain their presence in the mucus from the ventral tract.

Several collagen proteins (cell structure, cytoskeleton, shape and mobility) were more abundant in the dorsal (225 spectral counts; Table S3) as compared to the ventral tract (62 spectral counts) with the abundance of two of these (collagen alpha-6 chain, cds.c113381_g1_i1|m.22332 and cds.c114415_g2_i1|m.25328; Table 2) being significantly higher. In animals, collagen alpha-6 chain protein is the major structural component of the basement membrane. In bivalve gills, collagen is also present in the pair of skeletal rods that strengthen all types of filaments (Galtsoff, 1964; Le Pennec et al., 1988) and in muscle tissues (Medler and Silverman, 1998). The presence of these proteins in mucus from the dorsal tract can be explained by the large size of the skeleton rods and the position of the interlamellar septa (muscular tissue), both located beneath the principal filaments in close proximity to this tract.

The dorsal tract was also significantly enriched with ‘carbonic anhydrase 2-like’ (Table 2). Carbonic anhydrases are enzymes that catalyze the formation of hydrogen carbonate (HCO₃⁻) from carbon dioxide (CO₂) and water (Khalifah, 1971). In mollusks, these molecules were suggested to play a major role in acid–base balance (Wang et al., 2017), mediating the accumulation of calcium in mantle and gill tissue, and enabling the biomineralization process (Cudennec et al., 2006; Miyamoto et al., 1996). It has also been proposed that carbonic anhydrases are involved in ion regulation processes (osmoregulation) by generating HCO₃⁻ that can serve as counter-ions in sodium (Na⁺) and potassium (K⁺) uptake (Henry and Saintsing, 1983; Hu et al., 2011). While carbonic anhydrases have been found abundantly in the gills of bivalves (Duvail et al., 1998; Henry and Saintsing, 1983), it remains unclear why this protein is more abundant in the dorsal tract.

This fine-scale analysis of mucus revealed major proteomic differences between the dorsal and ventral tracts of the gill of C. virginica and suggests that each of these tracts upholds functional specialization, including their precise role in particle transport. Results showed that the dorsal tract (transport of particles intended for ingestion) is enriched with mannose- and glucose-binding lectins, providing a mechanistic explanation of previous experimental findings showing preferential ingestion of microalgae with cell surfaces covered with mannose and glucose residues (Pales Espinosa et al., 2016a). In parallel, the enrichment of mucin-like molecules and other adhesive proteins in the ventral tract is in line with prior studies showing a high viscosity of mucus in this tract. Overall, the results demonstrate that the molecular signature of mucus in each tract is different and can be linked to their specific function. However, the lack of information about the function of some proteins limits our ability to generate a complete picture of the functional topography of oyster gills. Additional studies should evaluate the effective ability of mucus from each tract to differentially interact with waterborne microbes.

We thank Drs E. Ward and M. Rosa (technical assistance with oyster and mucus collection), Dr D. Martin (support for proteomic analysis) and Drs E. Corre and S. Bassim (bioinformatics support).