Mar Biol (2016) 163:256 DOI 10.1007/s00227-016-3040-6

ORIGINAL PAPER

Chalky versus foliated: a discriminant immunogold labelling of shell microstructures in the edible gigas

Vincent Mouchi1,2,9 · Franck Lartaud3 · Nathalie Guichard4,5 · Françoise Immel4,6 · Marc de Rafélis2,7 · Cédric Broussard8 · Quentin G. Crowley1 · Frédéric Marin4

Received: 31 August 2016 / Accepted: 11 November 2016 © Springer-Verlag Berlin Heidelberg 2016

Abstract Mollusc shells are organic–inorganic biocompos- discontinuous pockets of ‘chalky layers’, a porous microstruc- ites, arranged in a limited number of superimposed calcified ture typical of bivalves of the ostreid family. By developing a layers that generally exhibit very different organization of polyclonal antibody (in two rats) elicited against a proteina- their crystallites. Because of their attractive mechanical and ceous shell fraction, we obtained differential staining of the crystallographic properties, these shell layers have been the two microstructures. We assert that our labelling is microstruc- focus of several physical and biochemical characterizations. ture discriminant. The difference in labelling of the two shell In particular, recent proteomic data obtained from individual microstructures suggests either that they are formed by a vari- layers suggest that their protein contents are different. How- ation of the secretory repertoire of the shell-forming cells of ever, the direct visual evidence that some macromolecular the calcifying epithelium or that the chalky layer may components are layer-specific is rather tenuous. This paper is be formed via a completely different mechanism. Our results based on a non-conventional immunogold labelling approach allow a first glimpse on the subtle regulatory mechanisms that to localize proteins in the shell of the edible oyster Crassos- drive the process of chalky and foliated layers deposition. trea gigas. The shell microstructure of this model organism is predominantly composed of foliated calcite, interspersed by Introduction

Responsible Editor: A.G. Checa. To protect their soft body, most molluscs secrete an exter- nal rigid exoskeleton, the shell. The shell is an inorganic– Reviewed by E. Harper, A. Osuna-Mascaró and an undisclosed expert. organic biocomposite, predominantly made of calcium car- bonate, with a minor fraction of occluded organics, about Electronic supplementary material The online version of this 1% of the shell weight (Marin et al. 2012). This fraction, article (doi:10.1007/s00227-016-3040-6) contains supplementary a mixture of proteins, glycoproteins and polysaccharides, material, which is available to authorized users.

* Vincent Mouchi 5 Present Address: UMR CNRS 6457 SUBATECH, La [email protected] Chantrerie, Nantes, France 6 Present Address: UMR CNRS 5200 laboratoire de Biogenèse 1 Department of Geology, School of Natural Sciences, Trinity Membranaire, Université Bordeaux Segalen, Villenave College Dublin, Dublin 2, Ireland d’Ornon, France 2 UPMC Univ Paris 06, CNRS UMR 7193, ISTeP, Sorbonne 7 Present Address: UMR CNRS 5563 Géosciences Universités, 75005 Paris, France Environnements Toulouse (GET), Université de Toulouse III 3 UPMC Univ Paris 06, CNRS, Laboratoire d’Ecogéochimie Paul Sabatier, Toulouse, France des Environnements Benthiques, Observatoire 8 UMR CNRS 8104 INSERM U1016 ‑ Institut Cochin, Océanologique de Banyuls, Sorbonne Universités, Université Paris Descartes, Paris, France 66650 Banyuls/Mer, France 9 Present Address: UPMC Univ Paris 06, CNRS UMR 7193, 4 UMR CNRS 6282 Biogéosciences, Université de Bourgogne ISTeP, Sorbonne Universités, 75005 Paris, France – Franche Comté (UB-FC), Dijon, France

1 3 256 Page 2 of 15 Mar Biol (2016) 163:256 collectively described as the shell matrix, is the main Crassostrea gigas shell allows a very precise temporal regulator of mineral deposition (Weiner and Traub 1984; calibration, independently from the shell microstructures Lowenstam and Weiner 1989; Simkiss and Wilbur 1989). (Lartaud et al. 2010b). The shell of C. gigas exhibits two During calcification, the shell matrix is secreted by the cal- main textures: the foliated and chalky microstructures. cifying mantle epithelium, together with inorganic precur- Foliated calcite can be described as ‘a laminar structure sor ions including calcium, bicarbonate and minor elements consisting of parallel calcitic laths arranged in sheet dip- such as magnesium and strontium (Marin et al. 2012). ping at the same angle and in the same general direction All of these ingredients interact together at the interface over a large portion of the depositional surface’ (Carter between the mantle tissue and the growing shell and self- and Clark 1985). The chalky microstructure is typical of assemble to form crystalline architectures that are exqui- Ostreidae and predominantly found in this bivalve family. sitely crafted (Carter 1990). It forms discontinuous, lenticular bodies that are interca- Because of their mechanical properties, marine mol- lated between the folia. Numerous studies have reported lusc shells are often taken as a model for biomineraliza- different putative mechanisms for the formation of chalky tion studies (Addadi et al. 2006) and, more generally, as an layers (Orton and Amirthalingam 1927; Korringa 1951; inexhaustible source of inspiration for generating organic– Palmer and Carriker 1979; Vermeij 2014). So far, no con- inorganic composites with tailored mechanical proper- sensus emerged. ties and shapes (Cranford and Buehler 2010). In addition In the present paper, we used an immunolabelling to these biotechnological applications, mollusc shells are approach—complementary to that developed by one of studied for their capacity to record with high reliability, us (F. L.) with manganese—to label the shell of C. gigas. the variations of physicochemical parameters of seawater To this end, we employed the following strategy: the shell (Rhoads and Lutz 1980). In recent years, this second aspect matrix was extracted, characterized on monodimensional has become more pressing in the context of concern over electrophoretic gels, and one fraction was further puri- global changes, in particular, of acidification (Orr fied by preparative SDS-PAGE before being analysed by et al. 2005). proteomics, and tested for its ability to interact with the are excellent mollusc models for such environ- in vitro precipitation of calcium carbonate. This protein mental studies. Oysters are marine and brackish bivalves of fraction was subsequently used to elicit polyclonal antibod- the pteriomorphid subclass with a number of advantages; ies, which, after accurate testing, allowed immunolabelling firstly, they are ubiquitous, occurring all over the world of the shell. This labelling is microstructure discriminant. and seas with the exception of polar regions, allow- The difference in labelling of the two shell microstructures ing comparisons at the global scale. Secondly, they with- suggests that their elaboration rests upon a variation of the stand different salinities (Bricteux-Grégoire et al. 1964) secretory repertoire of the shell-forming cells of the calci- and different water depths (van Rooij et al. 2010), and can fying mantle epithelium. consequently be used as markers in different environments. Thirdly, their shell is entirely made of low magnesium cal- cite—with the exception of the restricted myostracal layer Materials and methods that is aragonitic (Taylor et al. 1969)—and thus resists diagenetic transformations better than most aragonite and Sample preparation high-magnesium calcite shells (Ahr 2008). Oysters exhibit a large stratigraphical range that covers the last two hun- Fresh oysters were collected live in Dublin area, Ireland. dred million years of the history of Earth (Márquez-Aliaga For specimens used for matrix extraction, shells were emp- et al. 2005), allowing extensive palaeoenvironmental recon- tied, the muscle scar scrupulously cleaned, and the outer struction studies at different geological periods. Finally, surface of the shells mechanically abraded with a rotary their shell is thick, i.e. ideal from a practical viewpoint for tool (Dremel) to remove all epibionts that could act as a measuring with high-accuracy environmental proxies along source of contaminant material. Whole shells were then transects, in both recent (Lartaud et al. 2010a; Mouchi et al. chemically cleaned in dilute sodium hypochlorite (0.26% 2013) and fossil (Bougeois et al. 2014, 2016) specimens. active chlorine) for 48 h and rinsed thoroughly in deion- Our study focuses on one member of the ostreid fam- ized water several times. Shells were mechanically crushed ily, Crassostrea gigas, the edible cupped oyster, also in small fragments (4-5 mm) that were placed in sodium called the giant , and a model of economic hypochlorite (0.26% active chlorine) for 24 h. They were interest (Gosling 2003). Crassostrea gigas can withstand rinsed several times in Milli-Q water, dried at 37 °C and huge environmental variations, including alternation of powdered using a mortar grinder (Pulverisette 2, Fritsch, emersion/immersion, drastic change in salinity and rapid Idar-Oberstein, Germany). The powder was sieved to select increase/decrease in temperature (Lartaud et al. 2010a). particles with a grain size below 200 µm.

1 3 Mar Biol (2016) 163:256 Page 3 of 15 256

For immunogold localization, additional shells were (Laemmli-soluble AIM). The proteins were fractionated on taken from the collection of the Institut des Sciences de la the gel for 15 min at 100 V and then for about one hour at Terre de Paris (ISTeP, UPMC). These specimens had been 150 V. The gel was stained with silver nitrate according to previously bred for two years at Baie des Veys (Normandy, the protocol of Morrissey (1981), with the modification that France) according to a published paper (Lartaud et al. the colour development was stopped with 3 M citric acid. 2010b). The shells were emptied and chemically cleaned with hydrogen peroxide (6%) for 6 h followed by 0.15 N Protein extraction and fraction purification nitric acid for 20 min and washed in demineralized water. The hinge areas were cut from the shells, and hinge sec- Following the separation of the proteins in gel, one of tions were glued with epoxy to glass plates and sawed to a the most abundant proteins was purified on a large scale, thickness ranging from 500 µm to 1 mm, in order to visual- according to the procedure described in Marin et al. (2001) ize the chalky and foliated shell microstructures. and Marin (2003): 2 mL of Milli-Q and 2 mL of LSB (2 ) × were admixed to 124 mg of extracted AIM, and the prepa- Shell matrix extraction ration was denatured by heating at 100 °C for 10 min. Frac- tionation of the proteins of the LS-AIM was performed at The protocol used for matrix extraction was, with slight 180–200 V for the stacking and 300 V for the running in differences, used in previous papers (Osuna-Mascaró et al. a 12% acrylamide preparative gel was cast in a Bio-Rad 2014; Kanold et al. 2015): 25.08 g of shell powder was model 491 Prep Cell. The complete LS-AIM extract was suspended in 100 mL of Milli-Q water under constant stir- eluted from the gel in about 13 h and collected in 80 tubes ring, at 4 °C. The powder was slowly decalcified overnight (5 mL per tube, flow rate of elution: 0.5 mL/min). All 80 1 with cold dilute acetic acid (10% vol vol− ), with additions fractions were tested with dot blot using a Bio-Rad Bio-Dot every 5 s (100 µL) using an electronic burette (Titronic on a PVDF membrane. The membrane was subsequently Universal, Schott, Mainz, Germany). At the end of the treated with polyclonal antibodies elicited against cal- decalcification, the resulting solution (>1 L) was centri- prismin, a 37 kDa protein from the prismatic layer of the fuged for 15 min at 3900g. The supernatant containing the bivalve Pinna nobilis (Marin et al. 2005). This antibody acid-soluble matrix (ASM) was separated from the pellet presented a strong cross-reactivity with the shell matrix of of the acid-insoluble matrix (AIM). The AIM was scrupu- C. gigas in preliminary tests and recognized epitopes of rel- lously rinsed by a series of resuspensions in milli-Q water/ atively abundant protein of the AIM in Western blot (results centrifugation (5 cycles). Each time, the resulting super- not shown here). Two consecutive fractions presenting such natants were added to the ASM. The AIM was finally lyo- strong reaction were pooled, and the resulting solution dia- philized and weighed. The ASM was filtered on a Nalgene lysed (Spectra/Por tube, 4 °C) and lyophilized. The purity device with a 5-μm filter; then, its volume was reduced by of this extract, referred as F21-22, was tested on a 12% pol- ultrafiltration using a 400 mL Amicon cell with a 10 kDa yacrylamide gel, similarly to what described above. cut-off membrane. At the end of the concentration process (final volume around 15 mL), the solution was dialysed at Proteomics on the purified fraction 4 °C in a Spectra/Por tube (cut-off 1000 Da) against milli- Q water, with 5 water changes in 4 days. The solution was The identification of the protein content of F21-22 was per- lyophilized overnight and the resulting pellet, weighed. formed via a proteomic approach, according to an in-gel A second extraction was performed in similar conditions, digestion, as previously described (Kanold et al. 2015). The with 30.03 g of shell powder. fraction was denatured and run on a precast 12% acryla- mide mini-protean TGX gel (Bio-Rad). The gel was fixed Shell matrix analysis on 1D gel electrophoresis overnight (colloidal Coomassie blue) and then washed in Milli-Q water, and a band manually sliced near 30 kDa. ASM and AIM fractions were both analysed by conven- The slice was cut into cubes, which were subsequently tional monodimensional electrophoresis on 12% poly- placed in an Eppendorf tube. Then, in-gel digestion was acrylamide mini-gels (mini-protean III, Bio-Rad, Hercu- carried out with trypsin, according to a published proce- les, CA, USA) following the manufacturer’s instructions. dure with minor adjustments (Shevchenko et al. 2001): Both matrices were resuspended in Laemmli sample buffer the sample was destained twice with a mixture of 100 mM 1 (LSB) containing ß-mercaptoethanol and denatured by ammonium bicarbonate (ABC) and 50% (vol vol− ) ace- heating at 100 °C for 10 min. The solutions were cooled tonitrile (ACN) for 45 min at 22 °C and then dehydrated down on ice and then centrifuged for 2 min. A fraction of using 100% ACN for 15 min, before being reduced with the AIM was further solubilized by the Laemmli buffer, 25 mM ABC containing 10 mM DTT for 1 h at 60 °C and and this soluble fraction was referred to as the LS-AIM alkylated with 55 mM iodoacetamide in 25 mM ABC for

1 3 256 Page 4 of 15 Mar Biol (2016) 163:256

30 min in the dark at 22 °C. Gel pieces were washed twice Antibodies production, ELISA testing and Western with 25 mM ABC and dehydrated (twice, 15 min) and blots dried (10 min) with 100% ACN. Gel cubes were incubated with sequencing grade-modified trypsin (Promega, USA; The fraction was used to produce polyclonal antibodies 1 12.5 ng μl− in 40 mM ABC with 10% ACN, pH 8.0) over- (Eurogentec, Seraing, Belgium) in two rats, SER323 and night at 40 °C. After digestion, peptides were washed with SER324, following a standard immunization procedure: the 25 mM ABC, dehydrated with 100% ACN and extracted rats were injected (60 μg of antigens per injection) at day twice with a mixture of 50% ACN–5% formic acid (FA). 0 and then at days 14, 28, 56 and 132, and their blood was Extracts were dried using a vacuum centrifuge concentrator collected at day 0 (pre-immune serum, PPI), 38 (small bleed, plus (Eppendorf). PP), 66 (large bleed, GP) and 142 days (final bleed, SAB). For MS and MS/MS Orbitrap, analyses were performed The titres of the different antibody solutions were checked using an Ultimate 3000 Rapid Separation Liquid Chroma- by conventional ELISA (Clark and Adams 1977; Thresh tographic (RSLC) system (Thermo Fisher Scientific) online et al. 1977): in brief, the antigens were incubated in a Nunc with a hybrid LTQ-Orbitrap Velos mass spectrometer MaxiSorp 96-well microplate (200 ng per well, 90 min, 1 (Thermo Fisher Scientific). Briefly, peptides were dissolved 37 °C). After blocking step (0.5 wt% vol− gelatin in TBS), in 4 μL of 10% ACN-0.1% FA. Then, peptides were loaded the microplate was incubated 90 min. with the antibody solu- and washed on a C18 reverse-phase precolumn (3-µm par- tions (PPI, PP, GP, SAB, diluted 1/100 to 1/200000) and then ticle size, 100-Å pore size, 150 µm i. d., 1 cm length). The with the secondary antibody (goat anti-rat, Sigma A 8438, loading buffer contained 98% H2O, 2% ACN and 0.1% diluted 30000 times). The microplate was thoroughly rinsed TFA. Peptides were then separated on a C18 reverse-phase with TBS/Tween 20 (using a manual Nunc-Immuno Wash 12 resin (2-µm particle size, 100-Å pore size, 75 µm i. d., microplate washer) between antigen incubation and block- 15 cm length) with a 1-h gradient from 100% A (0.1% FA ing and after the first and second antibodies incubations. The and 100% H2O) to 50% B (80% ACN, 0.085% FA and 20% microplate was revealed with the substrate solution, consist- H2O). ing of p-nitrophenylphosphate (5 mg tablet in 10 mL) dis- The Linear Trap Quadrupole Orbitrap mass spectrometer solved in a water:diethanolamine solution (10:1), pH 9.8. acquired data throughout the elution process and operated After short incubation at 37 °C, it was read with a multichan- in a data-dependent scheme with full MS scans acquired nel spectrophotometer at 405 nm. We checked that the two with the Orbitrap, followed by up to 20 LTQ MS/MS CID pre-immune sera gave no reactivities and consequently used spectra on the most abundant ions detected in the MS scan. the different sera for further characterization. Mass spectrometer settings were: full MS (AGC: 1 106, Western blots (Towbin et al. 1979) were used to test × resolution: 6 104, m/z range 400–2000, maximum ion the specificity of the antibodies against the matrix of the × injection time: 500 ms) and MS/MS (AGC: 5 103, maxi- shell of C. gigas. Both LS-AIM and ASM were tested on × mum injection time: 20 ms, minimum signal threshold: 12% polyacrylamide mini-gels. After migration, the pro- 500, isolation width: 2 Da, dynamic exclusion time setting: teins from the gels were electro-transferred on a PVFD 30 s). The fragmentation was permitted for precursors with membrane (Immobilon, Millipore) for 90 min at 100 V in a a charge state of 2, 3, 4 and above. For the spectral pro- Bio-Rad Mini Trans-Blot Module. The membrane was then cessing, the software used to generate.mgf files was Pro- blocked in a TBS solution containing 1% gelatin for 30 min teome discoverer 1.3. The threshold of signal to noise for before placed in a TBS solution containing 1% gelatin, extraction values is 3. Database searches were carried out 0.05% Tween 20 and the antibodies diluted 1500 times. The using Mascot version 2.4 (Matrix Science, London, UK) membrane was incubated for 3 h at 37 °C and then rinsed on ‘other metazoa’ proteins (35,149,712 sequences) from several times in a TBS/Tween 20 solution. It was subse- the NCBInr databank containing 12,374,887,350 residues quently incubated 90 min in a TBS/Tween 20/gelatin solu- (January 2014) (www.ncbi.nlm.nih.gov/) and an in-house tion containing secondary antibodies (goat anti-rat, Sigma, shell protein databank (762 sequences containing 220,545 ref. A8438) coupled with alkaline phosphatase and diluted residues). The search parameters were as follows: carbami- 30,000 times. Finally, the membrane was rinsed thoroughly domethylation as a variable modification for cysteins and (5 10 min) in TBS/Tween 20 and incubated for five min- × oxidation as a variable modification for methionines. Up to utes in the dark in CDP-Star (Sigma, ref. C0712) solution. 1 missed tryptic cleavage was tolerated, and mass accuracy The chemoluminescent signal was recorded by mounting tolerance levels of 10 ppm for precursors and 0.45 Da for the membrane between two write-on transparency sheets in fragments were used for all tryptic mass searches. Posi- a cassette and exposing it shortly to a X-OMAT Kodak film tive identification was based on a Mascot score above the which was conventionally developed and fixed. In addition, significance level (i.e. b5%). The reported proteins were the nitrocellulose membrane was stained with NBT/BCIP always those with the highest number of peptide matches. (SIGMAFAST tablets, Sigma, ref. B5655).

1 3 Mar Biol (2016) 163:256 Page 5 of 15 256

Specificity of the antibody responses to microstructures for 1 min in TBS before being blocked in filtered TBS/gel- atin (0.5%) solution (pH adjusted to 7.5 with dilute NaOH In order to check the extent to which the antibodies could solution) for 30 min. The shell samples were subsequently differentiate the matrix of the individual shell microstruc- incubated 3 h in TBS/Triton/gelatin containing the antibod- tures of C. gigas, we performed specific ELISA with ies (GP and SAB) diluted 500 times. They were rinsed five extracts from the chalky and foliated layers as previously times in TBS/Triton for 10 min and incubated 3 h in TBS/ described (Marin et al. 1999). In brief, left valves were cut Triton/gelatin (0.5%) containing the secondary antibody in half to expose the hinge region showing the two micro- (goat anti-rat, 5 nm gold conjugate, BBI ref. EM.GTMA5, structures. The chalky and foliated microstructures were dilution 1/100). They were thoroughly washed in TBS/Tri- collected separately using a dental drill. 40 mg of each ton and then in water and slightly dried by capillarity. Frag- powder were dissolved overnight in 4 mL EDTA solu- ments were finally silver-enhanced (BBI ref. SEKL.15) for 1 tion (10 wt% vol− ). After a short centrifugation (3900g, 15 min before being rinsed in Milli-Q and dried at 37 °C. 10 min), aliquots of the EDTA extracts were directly incu- Several negative controls were performed, by using PPI, or bated in 96-well microplates (37 °C, 90 min.). A conven- by replacing the antibodies (first and/or second) by TBS/ tional ELISA test, as described above, was performed with Triton/gelatin. Observations were performed on the Hitachi each of the sera (PP, GP, SAB) obtained from both rats. The TM1000 Tabletop Microscope without carbon coating. The microplate was read at 405 nm. PPI (pre-immune serum) test was repeated three times. was used as a negative control.

In vitro crystallization test Results

Both the ASM and the purified fraction of the AIM (F21- Microstructures of the shell of Crassostrea gigas 22) were tested for their capacity to influence the growth of calcium carbonate crystals, according to a procedure As shown in Fig. 1, the shell of C. gigas is composed pre- derived from that of Albeck et al. (1993). Briefly, calcite dominantly of two microstructures: the main one is foli- crystals were grown by the interaction between vapours ated calcite, classically described by Taylor et al. (1969), of ammonium bicarbonate and CaCl2 solution (10 mM) Runnegar (1984) and Carter and Clark (1985), and consist- containing a small quantity of ASM or F21-22. The CaCl2 ing of parallel calcitic laths arranged in sheets; the chalky solution (200 μL) with different quantities of matrix (from layer, discontinuous, forming lenses in the hinge region 1 0.3125 to 20 μg mL− ) was placed in the wells of a 16-well (Fig. 1a) and intercalating thin layers in the other parts of culture slide (Lab-Tek, Nunc). The cover of the slide was the shell. From a microstructural viewpoint, chalky layers pierced to allow diffusion of ammonium bicarbonate appear far more porous and made of a framework of blade- vapours. The culture slide with its cover was sealed with shaped (Margolis and Carver 1974) crystals that develop Parafilm and then placed at 4 °C in a 5-L closed desicca- more or less perpendicularly to the mineral depositional tor containing crystals of ammonium bicarbonate for 72 h. plan. The blades are linked with each other by leaflets that

Control scenarios of only CaCl2 solution were tested in par- branch at different angles (Fig. 1b), leaving a large amount allel. After incubation, the solution was carefully removed of empty space. Between the frameworks, the space is filled from each well using a blunt-ended needle connected to a by tangled crystals. In addition to the foliated and chalky vacuum system. The glass plate of the slide was dissoci- microstructures, a thin prismatic calcitic layer is observed, ated from the well spare part and directly observed under constituting the outermost part of the shell (not shown in a Hitachi TM1000 Tabletop Microscope without carbon Fig. 1). coating. This experiment was repeated four times to ensure homogeneity of the results. Shell matrix extraction and characterization on mini‑gels Protein localization by immunogold by SEM Similar amounts of ASM and AIM were obtained from Small freshly fractured (<5 mm) fragments were placed in the different fractions. From the first batch (25.08 g shell sodium hypochlorite solution to remove superficial con- powder), we quantified 11.25 mg of ASM and 116.84 mg taminants. They were rinsed (milli-Q water), dried and then of AIM, representing 0.045 and 0.465% of dry weight of 1 slightly etched in EDTA solution (1 wt% vol− ) for 3 min to shell powder, respectively, representing together about expose antigenic determinants. For the cleaning and etch- half a per cent of organics. The AIM/ASM ratio is about ing steps, ultrasonic baths were not used, to avoid fragmen- 10. When proteins were fractionated on a monodimen- tation of the shell pieces. The samples were then washed sional gel and stained with silver, the profile (Fig. 2a)

1 3 256 Page 6 of 15 Mar Biol (2016) 163:256

a b 2 µm

c f

1mm 500µm

Fig. 1 Microstructures in the umbo region of a C. gigas shell, longi- left to bottom right. b Detailed view (white square from a). ‘c’ indi- tudinal section, crossing the middle of the hinge region (perpendicu- cates the chalky structure while ‘f’ corresponds to the foliated layers lar to the opening plan of the valves). a Growth direction is from top

ASM LS-AIM F21-22 12 Fraction detection Extraction by dot-blot 170 170 130 130 Fractions pooling 95 Preparative 95 electrophoresis and dialysis 72 72 55 55 43 Freeze-drying Fraction collection 43 34 b 34 26 12345 6789101112 26

17 13 14 15 16 17 18 19 20 23 24 17 10 21 22 a c d 10

Fig. 2 Characterization on monodimensional gels of the shell matrix resis) showing the cross-reactivities of the different fractions with of C. gigas and purification of one fraction. a Silver-stained gel elec- anti-caspartin antibody (see text). Only the 24 first fractions are illus- trophoresis of the acid-soluble (ASM, lane 1) and of the Laemmli- trated here. Note that the fractions of tubes 21 and 22 (F21-22) give soluble acid-insoluble (LS-AIM, lane 2) matrices. Markers of differ- the strongest signal. d Gel electrophoresis of the F21-22 fraction, ent molecular weight (in kDa) are indicated on the left. b Summary showing the apparent purity of this fraction, which was subsequently of the protocol used from the extraction to the freeze-drying of the used for eliciting polyclonal antibodies. Markers of different molecu- purified fraction.c Dot blot (performed after preparative electropho- lar weight (in kDa) are indicated on the right shows few proteins on each of the fractions distinct Protein purification and testing of antibodies from some smearing material. ASM (Fig. 2a lane 1) contains 3 main proteinaceous components at approxi- The whole LS-AIM was fractionated on a preparative gel mately 45, 27 and 12 kDa. The electrophoretic pattern electrophoresis, and the fractions were dot-blotted. The of LS-AIM (Fig. 2a lane 2) exhibits similarities with fraction of interest was eluted in tubes 21-22, and further that of ASM, since these 3 proteins are present, in addi- processed, including extensive dialysis and freeze-drying tion to three other diffuse proteinaceous components at (Fig. 2b). It was referred as F21-22 (Fig. 2c). After extensive approximately 60, 34 and 22 kDa. In both extracts, the dialysis and freeze-drying, 2.13 mg of purified protein was 45 kDa proteinaceous component is negatively stained. obtained from 30 g of shell powder. The fraction, when tested In the ASM, the upper part of the gel (above 130 kDa) on a mini-gel, is revealed as a thick proteinaceous component and the zone between 17 and 30 kDa present also this around 27 kDa (Fig. 2d). After production of polyclonal anti- particularity. bodies in two rats (SER323 and SER324), titres from the PPI,

1 3 Mar Biol (2016) 163:256 Page 7 of 15 256

SER323 SER324 a 2 b 2

SAB SAB 1.5 GP 1.5 GP ) ) PP PP (405nm

(405nm PPI 1 PPI 1 Ab s Abs

0.5 0.5

0 0 100 1000 104 105 106 1001000 104 105 106 Dilution Factor Dilution Factor

Fig. 3 Titres of the ELISA tests performed on the antibodies produced by SER323 (a) and SER324 (b). PPI pre-immune bleed, PP small bleed, GP large bleed, SAB final bleed. Note that the two rats behave differently to the repetitive injections

a SER323 b SER324 PPI GP SAB PPI GP SAB ASM LS-AIM ASM LS-AIM Std ASMLS-AIM F21-22 ASM LS-AIM ASMLS-AIM Std ASM LS-AIM F21-22

170 130 170 95 130 72 95 55 72 55 43 43 34 34 26 26

17 17

Fig. 4 Western blots of the ASM, LS-AIM and F21-22 fraction with (b) were tested. Std markers standard; the corresponding molecular polyclonal antibodies elicited against the F21-22 fraction. The PPI, weights (in kDa) are indicated on the right GP and SAB antisera produced in rat SER323 (a) and rat SER324

PP, GP and SAB bleeds were determined in ELISA (Fig. 3). are shown in Fig. 4. For each of them, we present the data Pre-immune bleeds from both rats show no reaction to the tar- obtained with PPI (negative control, left), second bleed geted fraction. For rat SER323, the final titre (about 1:1500) (GP, centre) and final bleed (SAB, right). For the clarity is almost reached after the first injection (PP bleeding), and of the results, we only illustrate the Western blots obtained the differences of immunological reactivity between the suc- after the chemical staining of the membrane with NBT/ cessive bleeding are minimal. For rat SER324, we observe a BCIP. Entirely superimposable results, although more progressive increase in the titre, in correlation with the suc- blurred, were obtained with the chemoluminescent CDP- cessive immunizations. In this case, titres are 1:100, 1:500 Star. None of the two PPIs react with the shell extracts, and 1:1000 for PP, GP and SAB, respectively. ASM or LS-AIM. When tested on F21-22, the antibodies from the two rats successfully recognize this fraction, giv- Western blots of shell extracts with the anti‑F21‑22 ing a high-intensity signal. Although the antibodies were antibody elicited against a discrete molecular weight fraction, their response against the whole ASM and LS-AIM encompasses The results of the Western blot of the shell extracts with the a broad range of molecular weights, from above 170 kDa to anti-F21-22 antibody from the rats SER323 and SER324 about 10 kDa. For rat SER323 (Fig. 4a), we notice that the

1 3 256 Page 8 of 15 Mar Biol (2016) 163:256

Table 1 Summary of the LC–MS/MS analysis of the purified F21-22 fraction of the shell matrix of C. gigas Accession numbera Protein identificationb MW (AA nb) Protein score MS/MS peptidesc Peptide score gi|317376184| Gigasin-6 Crassostrea gigas 34106 (302) 65 R.STIQEVYK.N 35 K.NPGVIVSVVK.D 8 K.NEIYTPLGMAK.S 21/(18)/(12) gi|762132907| Gigasin-6 isoform X1 Crassostrea gigas 62490 (552) 65 Same peptides as for Gigasin-6 gi|762132909| Gigasin-6 isoform X2 Crassostrea gigas 61546 (543) 65 Same peptides as for Gigasin-6 gi|762104436| Nacrein-like protein Crassostrea gigas 51244 (441) 63 K.TLSCLMEK.Y 12 K.KPSDYFIK.N 9 R.VEDTDNNPLK.E 42 gi|512134004| Nacrein-like protein Crassostrea gigas 48258 (413) 58 K.TLSCLMEK.Y 12 K.KPSDYFIK.E 9 R.VEDTENNPLK.E 36/(18)/(9) gi|762164175| Cell death abnormality Crassostrea gigas 28195 (255) 56 R.SDFECPR.D protein 1-like R.AAGSISGGDPAT- GTEAADTGSGM.- a Accession number of each protein hit according to NCBI database b Protein name according to NCBI database; MW: theoretical molecular weight in Daltons, calculated from the identified protein (in parenthe- ses, AA nb number of amino acid residues) = c List of peptides identified by the analysis antibody allows visualizing proteins that cannot be discrim- in the two proteins, while the third differed only by one inated on the silver-stained gel, in particular proteins of amino acid residue (D or E). Two peptides—among which high molecular weights around 72, 130 and above 170 kDa. a 23 amino acid residues long hydrophobic peptide—could These proteinaceous components are observed both for assign the cell death abnormality protein 1-like. The posi- ASM and LS-AIM from GP and SAB (Fig. 4a). Other tions of these peptides along the different protein sequences proteins do cross-react around 50 and 40 kDa in the same are visualized in supplementary Fig. 1. Additional in silico lanes. For rat SER324, we obtain a different pattern, since investigations in less stringent conditions (lower threshold, the corresponding antibody stains preferentially the smear not shown in Table 1) generated six peptides—all located than the discrete proteins. In the ASM (Fig. 4b, SAB), the in the N-terminal region of the protein—that match with a F21-22 fraction is well marked. For both rats, the staining transcription termination/anti-termination protein NusA- of LS-AIM is more pronounced than that of ASM. Signals like from the Mediterranean Fruit flyCeratitis capitata given by GP bleeds are weaker than those of SAB bleeds, (gi|498978467). The significance of these additional hits is particularly for rat SER324. This finding is congruent with not understood. the ELISA results. In vitro crystallization in the presence of ASM and Proteomics on the F21‑22 fraction of F21‑22

The proteomic investigations, as summarized in Table 1, Results of the in vitro crystallization assay with fraction yielded a series of peptides that match with three pro- F21-22 and with ASM are shown in Fig. 5. The control sce- teins or protein families of C. gigas that are, respectively: nario with no protein (Fig. 5a) produces single crystals that Gigasin-6, and two of its isoforms Gigasin-6 X1 and exhibit the typical rhombohedral morphologies of calcite. Gigasin-6 X2; two nacrein-like proteins; a cell death abnor- Effects are markedly different between the two extracts. 1 mality protein 1-like. Gigasin and its two isoforms X1 and At low concentration (0.31 µg mL− ), ASM exhibits a pro- X2 were identified by an identical set of three different nounced effect on the crystal shape, with the formation of peptides. Each of the two nacrein-like proteins were also polycrystalline aggregates (Fig. 5e). At the same concentra- identified by three peptides, two of which being identical tion (Fig. 5b), fraction F21-22 exerts almost no effect on

1 3 Mar Biol (2016) 163:256 Page 9 of 15 256

b c d

a

250µm 50µm 30µm

f g 100µm

50µm 50µm 30µm e

Fig. 5 In vitro crystallization of calcium carbonate at different concentrations of F21-22 (b–d) and of ASM (e–g). a 0 µg/mL. b, e 0.3125 µg/mL. c, f 5 µg/mL. d, g 20 µg/mL

SER323 SER 324 1.4 1.2

1.2 Chalky fraction 1 Chalky fraction Foliated fraction Foliated fraction 1 0.8

0.8 (405nm ) (405nm ) 0.6 0.6 Ab s Ab s

0.4 0.4

0.2 0.2

0 0 a PPI PP GP SAB b PPIPPGPSAB

Fig. 6 Differences in ELISA tests on chalky and foliated fractions with antibodies produced by SER323 (a) and SER324 (b) rats. Abs absorb- ance, arbitrary units, PPI pre-immune bleed, PP small bleed, GP large bleed, SAB f