Ps19, a novel chitin binding protein from Pteria sterna capable to mineralize aragonite plates in vitro

Mollusk shell is composed of two CaCO3 polymorphs (calcite and aragonite) and an organic matrix that consists of acetic acid- or ethylenediaminetetraacetic acid (EDTA)-soluble and insoluble proteins and other biomolecules (polysaccharides, β-chitin). However, the shell matrix proteins involved in nacre formation are not fully known. Thus, the aim of this study was to identify and characterize a novel protein from the acetic acid-insoluble fraction from the shell of Pteria sterna, named in this study as Ps19, to have a better understanding of the biomineralization process. Ps19 biochemical characterization showed that it is a glycoprotein that exhibits calcium- and chitin-binding capabilities. Additionally, it is capable of inducing aragonite plate crystallization in vitro. Ps19 partial peptide sequence showed similarity with other known shell matrix proteins, but it displayed similarity with proteins from Crassostrea gigas, Mizuhopecten yessoensis, Biomphalaria glabrata, Alpysia californica, Lottia gigantea and Elysia chlorotica. The results obtained indicated that Ps19 might play an important role in nacre growth of mollusk shells.


Introduction
The biological process by which an organism crystalizes minerals is known as biomineralization. Calcium carbonate is a mineral that can be biomineralized into three polymorphs: vaterite, aragonite, and calcite; the last two are commonly found in mollusk shell, the second largest group of metazoans [1,2]. In bivalve shell, calcite forms an outer prismatic layer while aragonite forms flat tablets that assemble in a "brick-wall-like" structure or nacre sheet, forming an inner nacreous layer [2][3][4][5]. Nacre has outstanding mechanical properties due to the combination of stiffness, strength, and toughness given by its unique microstructure arrangement [6].
Shell formation in mollusks is controlled by epithelial cells from the mantle, the organ responsible for secreting the required components for calcium carbonate crystal biomineralization, such as calcium, carbonate ions and organic macromolecules (proteins, β-chitin, glycoproteins, polysaccharides) that form the organic shell matrix [7,8] matrix are located either in the inorganic phase (within the calcium carbonate crystal) or between crystals forming the organic framework [9]. The organic matrix guarantees mineral and macromolecule interactions as it has a main role in crystal nucleation and growth [10]. The mechanism of nacre formation has been studied through extraction and functional studies of various shell matrix proteins (SMPs); nevertheless, nacre biomineralization is a very complex process; thus, soluble and insoluble proteins in ethylenediaminetetraacetic acid (EDTA) or 10% acetic acid have been studied because of their biochemical and crystallization nacre properties [11,12]. Previous studies have reported that peptide sequences from insoluble and soluble matrixes share conserved regions or domains; however, shell matrix protein sequences are generally dominated by Asp, Glu, Ser, Ala, Gly, Pro and Cys residues, which are often concentrated in short or long repetitive domains [5,[13][14][15]. Gly and Asn residues usually create specific domains that have been associated with nucleation [16] and inhibition [2,17] functions in the biomineralization process while Gly/Tyr residues or GYS motif have been associated to polysaccharide-binding abilities [2,[16][17][18][19][20]. Conservatively, in the insoluble fraction, a combination of chitin-protein complexes exhibit hydrophobic biochemical properties (enrichment in aliphatic amino acids, such as glycine and alanine) while the soluble shell fraction is found to be polyanionic, and particularly aspartic acid-rich [15].
Nacre proteins containing chitin-binding domain (CBD) includes the blue mussel shell protein (BMSP), pearlin/N16 family, hichin, Pif-like, among other proteins. This domain plays an important role in organic scaffold construction following crystal deposition [2,29,30]. Another domain found in insoluble SMPs is the von Willebrand A (vWA) domain, involved in cell adhesion and often associated with CBD, found in proteins as BMSP, Pif family, and others [27,30]. The calcium binding domain is also of vital importance because these proteins are associated with nucleation of aragonite and calcite and found in proteins, such as N16, N14, MSI60 and Aspein [24,[31][32][33].
Besides the intrinsic characteristics of the protein sequences, some SMPs possess posttranslational modifications, such as phosphorylation, glycosylation, and sulfation that determine their structure and their interactions with other molecules; moreover, enzymatic regulations allow them to determine which calcium carbonate polymorph forms by interacting with the mineral [1,17]. For example, pearlin is the only protein of its family, which by itself binds to calcium, and in vitro it had the ability to induce aragonite crystals in the presence of Mg 2+ while N14 mostly combines with N66 (a carbonic anhydrase) to induce aragonite plates; Prismalin-14 binds to Ca 2+ and displays inhibitory activity on calcium carbonate crystallization in in vitro assays [19]. These interactions suggest that posttranslational modifications are crucial for some SMPs to display their role in shell biomineralization [17].
In recent decades, much attention has been drawn to describing shell matrix protein interactions with calcium ions and other biomolecules, as well as their role in nacre formation to understand their contribution to its unique mechanical and biological properties. Many shell matrix proteins remain unknown despite the continuous efforts to isolate, describe and characterize them all. Some of these isolated proteins have not been completely characterized yet, and their role in biomineralization still remains unknown. In this sense, it is important to pursue the efforts to understand the shell biomineralization process. This research aimed to identify and characterize a novel protein called Ps19, the most abundant protein located in the acetic acid-insoluble fraction of the Pteria sterna shell, a local pearl oyster of economic importance, to understand its role in shell biomineralization. Surprisingly, this Ps19 exhibited calcium-and chitin-binding capabilities involved in the biomineralization process. These facts revealed new insights in the shell biomineralization mechanisms that were analyzed by crystallization of aragonite plates in in vitro studies.

Biological material
Three shells from adult oysters were provided by Perlas del Cortez S. de R.L. MI. located at Bahia de La Paz, B.C.S. Shells were transported to the Molecular Genetics Laboratory facilities at CIBNOR.

Shell matrix proteins extraction
The organic matrix of the shell of Pteria sterna (20 g of pulverized shell) was extracted by decalcification with cold acetic acid (4˚C, 10% v/v) as previously described by Montagnani [2]. After decalcification, the solution was centrifuged at 4,500 ×g for 30 min, and the supernatant, Acetic Soluble Matrix (ASM) was collected; the pellet or Acetic Insoluble Matrix (AIM) was collected in a separate tube. The ASM was dialyzed (12-14 kDa cutoff Spectra/Por 1 membranes, Repligen, No.132680, California, USA) against distilled water overnight at 4˚C under constant low stirring. Then, water exchange was made with distilled water and dialyzed for another 8 h. Afterward, the dialyzed extract (ASM) was concentrated by cold per-evaporation (4˚C, 36 h). The AIM was rinsed with distilled water and water was removed; both fractions (ASM and AIM) were stored at −20˚C for further analysis.

Sodium dodecyl sulphate-polyacrylamide gel electrophoresis
Sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) was performed according to Laemmli [34]. Samples ASM (30 μg of protein) and AIM (64 μg of protein) of the shell from P. sterna were mixed with sample buffer 4× (0.5 M Tris-HCl pH 6.8, 20% glycerol, 10% SDS, 10% β-mercaptoethanol and 0.05% bromophenol blue) and boiled for 10 min, then loaded into a 16% polyacrylamide gel. Broad range molecular weight standard (Bio-Rad 1610317, California, USA) was loaded into the gel. Electrophoresis was conducted at 90-V at room temperature, using a vertical electrophoresis unit (Bio-Rad Protean II, California, USA). After electrophoresis, the gel was stained with Coomassie Brilliant Blue R250 (CBB) for 2 h, washed out and analyzed for protein, using a gel imager (Chemi Doc XRS, Bio-Rad, California, USA). Also, proteins were stained with silver nitrate [35,36]. The same procedure was followed for purified protein from P. sterna (Ps19), loading 6.8 μg of protein.

Protein quantification
Quantification of the most abundant protein present in the AIM (Ps19) was performed by pixel densitometry by separating the protein sample in a 16% SDS-PAGE gel and stained with Coomassie Brilliant Blue (R250). First, a standard curve was made with ovalbumin protein (0.25-8.0 μg�μL -1 ); the image was scanned with a Chemi Doc XRS (Bio-Rad, California, USA).
Then, the density of the pixels from each band was calculated using the Image Lab 5.1 software, and the linear equation was obtained. The protein amount of Ps19 from the shell of P. sterna was calculated through the obtained equations of ovalbumin standard curve (y = 5 × 10 −6 x − 0.94).

Protein purification by preparative SDS-PAGE
The AIM of P. sterna was fractionated on a discontinuous preparative polyacrylamide gel electrophoresis following instructions in the Mini-Prep Cell Manual (Bio-Rad, model 491 Prep Cell, USA, California). Briefly, a 25-mg sample obtained from the AIM was combined with 250 μL electrophoresis sample buffer 4× (0.5 M Tris-HCl pH 6.8, 20% glycerol, 10% SDS, 10% β-mercaptoethanol and 0.05% bromophenol blue) and loaded into a 12% polyacrylamide rod gel, 7 cm high. Electrophoresis was performed under 200-V at room temperature. When the tracking dye front reached the bottom of the gel rod, 100 fractions of 150 μL were collected. Random fractions were loaded into a 16% polyacrylamide SDS-PAGE to locate the protein of interest in the fractions and later delimitate all of those that contained the protein. All fractions containing the protein were pooled, concentrated, and at the same time, electrode buffer eliminated by centrifugal filter 10 000 MW cut-off (Amicon Ultra-4, EMD Millipore, Germany, Darmstadt) at 7,500 ×g, at 4˚C for 10 min, using 30 mM Tris-HCl buffer pH 8.8. Pure protein (100 μL) was sampled in 10 μL aliquots and stored at −20˚C for further analysis. Purified Ps19 concentration was calculated by densitometry as described before.

Protein MS analysis
Ps19 protein was separated in a 16% SDS-PAGE. After electrophoresis, the gel was stained with Coomassie Blue R-250 and the band was manually excised. The sample was processed at Laboratorio Universitario de Proteómica, UNAM, where it was digested by trypsin (Promega Sequencing Grade Modified Trypsin; Madison, WI, USA) and analyzed by Liquid Chromatograph-Mass Spectrometry (LC-MS). The obtained data were compared with the database at PepBank of Massachusetts General Hospital, GenBank and non-redundant proteins from NCBI.

Chitin-binding assay
Chitin-binding assay was performed as described by Montagnani [2]. Briefly, 13.6 μg of purified protein Ps19 or BSA (used as a negative control) were mixed with 100 μL of distilled water and incubated with 1 mg of shrimp shell chitin (Sigma-Aldrich, C9752, St. Louis, MO, USA) under constant stirring for 2 h at 25˚C. The sample was centrifuged for 5 min at 13,000 ×g at room temperature; the supernatant was recovered and stored at 4˚C. The pellet was rinsed thrice with distilled water (200 μL) before washing with 0.2 M NaCl (100 μL), and finally centrifuged for five min at 13,000 ×g; the supernatant was recovered and stored at 4˚C. The pellet and recovered supernatants were mixed with electrophoresis sample buffer 4× (0.5 M Tris-HCl pH 6.8, 20% glycerol, 10% SDS, 10% β-mercaptoethanol and 0.05% bromophenol blue), boiled for 10 min, and loaded into a 16% SDS-PAGE gel. Afterward, proteins were revealed with silver nitrate staining [35] and visualized in a Chemi Doc XRS (Bio-Rad, California, USA).

In vitro CaCO 3 crystallization in the presence of purified protein
To evaluate the crystal formation of Ps19, it was incubated with three different saturated solutions according to Weiss [39] and Hillner [40]:

Raman
Raman spectroscopy was performed to the in vitro obtained crystals by using an InVia micro Raman spectrometer (Renishaw, Nuevo Leon, Mexico) with an excitation line of 532 nm provided by a YAG laser of 100 mW with a spot size of 2 μm x 2 μm. The crystals were scanned for 90 seconds from 100 to 1900 cm −1 by triplicate for the specific identification. For all measurements, the slits were set at 200 μm and a 100× objective was used. The analysis was carried out at Laboratorio Nacional de Investigaciones en Nanociencias y Nanotecnología (LINAN)-IPICYT.

Isolation and characterization of the Ps19 from the shell of P. sterna
The organic matrix from P. sterna shell was extracted with cold acetic acid. The acid-soluble (ASM) and acid-insoluble (AIM) fractions were separated by electrophoresis and stained with CBB, which showed that AIM had more protein bands than the ASM (Fig 1). The AIM showed an abundant protein band with a relative molecular mass determined from a comparison of relative migration on SDS-PAGE of 19 kDa. The 19kDa protein band, named Ps19, from the AIM was purified in a single step using a preparative electrophoresis. The Ps19 was eluted in the fractions from 58 to 83 (S1 and S2 Figs). This purification procedure yielded a total of 68 μg of Ps19 purified protein from 890 μg of unpurified protein contained in the crude extract of the AIM from P. sterna (S3 Fig) shell, which had an 18% yield of protein with 85% purity (Table 1). Ps19 is a glycoprotein with acid amino acid residues with potential calcium-binding properties according to PAS and Stain All/silver staining (Fig 2). The chitin-binding capability of Ps19 was analyzed by incubation of the isolated protein with chitin and successive washing solutions. Ps19 was clearly present in the pellet treated with Laemmli buffer indicating that Ps19 is capable of binding α-chitin unlike BSA (negative control) that is not present in the pellet (containing α-chitin), but in the aqueous fraction of the washes (Fig 3).

Protein sequencing
The Ps19 was sequenced de novo by Liquid Chromatography-Mass Spectrometry, its amino acid composition indicated the presence of a high proportion of Leu (17%), Asp (11%), Ser (8%) and Ala (8%); thus, it is potentially an acidic protein with a 17% of acidic residues (Asp/

PLOS ONE
Ps19, a novel chitin binding protein from Pteria sterna Glu). Using the Basic Local Alignment Search Tool for proteins (BLASTp) against non-redundant and Mollusca proteins database with a cutoff of 0.3 for the E-value and using as query sequence the peptides obtained by Mass Spectrometry, the Ps19 match with two uncharacterized proteins from Crassostrea gigas with 89 and 78% of identity and 2.1 and 1.0 E-values respectively, with two proteins from Mizuhopecten yessoensis with also 89 and 78% of identity and 3.0 and 0.5 E-value respectively. Other peptides showed 67% identity with a hypothetical protein from Lottia gigantea and 100% identity with a BTB/POZ domain, ankyrin repeat from Aplysia californica with 100% identity and 0.5 E-value. Finally, the last peptide had 80% similarity with 0.4 E-value to a serine beta-lactamase-like protein from Biomphalaria glabrata and 64% similarity with a hypothetical protein from Elysia clorotica and 2.5 E-value (Table 2).

PLOS ONE
Ps19, a novel chitin binding protein from Pteria sterna the typical geometry of calcite, with smooth faces and sharp angles (Fig 4a and 4c). Solution B (40 mM MgCl 2 pH 8.2, 100 mM NaHCO 3 ), displayed aragonite crystals, with a needle-like morphology and variable sizing (Fig 4b). When Ps19 was added to solution A and solution B, aragonite plates were appreciated (Fig 4d and 4e), the solution containing CaCl 2 (solution A) formed 30 μm sizing plates (Fig 4d), while solution B (containing MgCl 2 ) formed plates of 100 μm size (Fig 4e). Both crystal structures formed well defined octagonal shaped plates. However, no crystals were formed when solution C (100 mM CaCO 3 ) was incubated with Ps19 (Fig 4f).

Raman analysis
Raman spectroscopy was used to confirm the identity of calcite and aragonite crystal formation. The spectra for the samples corresponding to Fig 4a and 4e are shown in Fig 5. Usually, calcite crystals contain two CaCO 3 units, for a total of ten atoms; its nucleation and crystal growth have been extensively studied by Raman spectroscopy and theoretical calculations [42]. The presence of the active vibrational modes for Raman frequencies at 282.47 cm -1 (translatory oscillations of CO 3 groups), 712.48 cm -1 (v4, asymmetric bending), and 1087.27 cm -1 (v1, symmetric stretching of CO 3 groups), confirms the correct formation of the rhombohedral primitive cell of calcite with space group D 6 3d ðR � 3CÞ and parameters of a = 5.03 and c = 17.325 Å, respectively [43]. On the other hand, the Raman spectrum for the aragonite shows similar vibrations modes than calcite, but with a slight shift to lower energies for all modes, according to its orthorhombic primitive cell. The crystal contains four CaCO 3 units, for a total of twenty atoms; the presence of the Raman-active modes at 206.1 cm -1 , 702.595 cm -1 , and 1085.16 cm -1 , has been reported for aragonite crystalline powder and confirms the correct formation of the orthorhombic lattice with space group D 16 2h ðPnmaÞ and cell parameters of 5.008, 8.029, and 5.861 Å for the a, b, and c axes, respectively [44].

Discussion
A novel shell matrix protein from the acetic acid-insoluble matrix from P. sterna shell was isolated and characterized. Previous studies have revealed that acetic acid insoluble shell matrix

PLOS ONE
Ps19, a novel chitin binding protein from Pteria sterna proteins have Asp and Glu residues in high proportion turning them into acidic proteins, which make their isolation a challenging process. Ps19 showed a high proportion of these residues (17%); therefore it may have been an acidic protein and its isolation was difficult, having low yield (18%) extraction but high purity (85%). Acetic acid-insoluble proteins from other species have been extracted and an attempt to describe and characterize them all has been done, and although some have been described, many still remain to be characterized; their role in shell biomineralization is still unknown [12,28,45].
Biochemical assays for Ps19 have shown it has chitin-binding abilities, which is a vital characteristic for interactions between shell matrix proteins and organic biomolecules to create a frame for calcium carbonate mineralization [5,26,46,47]. β-chitin is one of the main organic components of interlamellar nacre plates, and it is also considered as a possible nucleation site for aragonite sheets due to its interaction with other matrix proteins [48][49][50]. According to this evidence, Ps19 may not be only involved in aragonite plate mineralization but also in its nucleation. Proteins like Pif-like and hichin (AIM proteins) are known for having chitin-binding conserved domains rich in Gly and Cys residues [27,30], but according to BLAST analysis, neither of these known domains were found in the peptide sequence of Ps19; however, it poses this capability in vitro, suggesting other unknown chitin-binding domains might exist. The Pearlin/ N16 family does not have a defined chitin-binding domain; nonetheless, the biochemical analysis showed they pose chitin-binding ability [2,32] coinciding with Ps19 results.
Previous reports have shown that proteins, such as N14, N16, MSI60 have specific GXN domains [31][32][33] or D-rich domains as Aspein [24], which are involved in calcium-binding in Pinctada fucata [51]. Ps19 showed calcium-binding capability, but its amino acid sequence did not suggest the presence of any conserved domain according to the BLAST analysis with a 0.1 cutoff E-value; however, when this value was broadened to 3.0, it showed two hits with 0.5 Evalue for DGDVLMWALR, one with 100% identity for a BTB/POZ domain of Aplysia californica which is a domain found in different proteins related to cytoskeletal arrengment, regulation of animal transcription, and other functions involving protein-protein interactions [52], and the second to a Mizuhopecten yessoensis uncharacterized protein with 78% similarity. The similarity with the BTB/POZ domain may indicate Ps19 is a protein involved in protein-protein interactions, perhaps to help in the aragonite plate formation, what kind of protein-

PLOS ONE
Ps19, a novel chitin binding protein from Pteria sterna protein interactions it is involved it is yet to be studied. Another hit, with 0.4 E-value for FPGSPYVELYKWR peptide, showed 80% similarity with a serine beta-lactamase-like protein from Biomphalaria glabrata. Beta-lactamase is a mitochondrial protein related to antibiotic resistance with enzymatic activity [53,54], so further analysis of Ps19 is needed to prove if it has enzymatic activity or at least share some characteristics with this family of proteins. The other hits showed similarity with uncharacterized proteins or with hypothetical proteins, so their function is not well described and more information is required to make an statement.
Glycosylation is an important feature for SMPs because it imparts an acidic pH to the protein [55]. It has been proposed that acidic sugars, especially sulfated, may concentrate calcium ions at the vicinity of the acidic proteins, inducing crystal nucleation [56]. They may also play an important role in mineral surface recognition and in polymorph selection [55][56][57]. Ps19 showed glycosylation, which may be an acidic protein; therefore, its glycosylation may be responsible for its capacity to crystalize aragonite plates in vitro despite the salt it is incubated with.
Saturated solutions have demonstrated to contain -HCO 3 , Na 2+ or Ca 2+ that crystalize calcite crystals while incubating with Mg +2 will form aragonite crystals; when shell matrix proteins are aggregated to this solutions, they might induce calcite or aragonite crystals respectively [40,41]. Solutions forming calcite or aragonite crystals were consistent with the reported results in this research; however, Ps19 is capable of inducing aragonite crystals despite the salt it is incubated with (solution A or B), indicating it selectively induces aragonite plates similar to nacre plates in vitro. No crystal formation was observed when Ps19 was incubated with solution C (100 mM CaCO 3 ), which might be associated to protein concentration as previously reported by other authors. This result suggests that high protein concentration may inhibit calcium carbonate crystal formation, acting as a nucleating regulator [39,58,59].
Calcium carbonate polymorph selection does not only depend on the salt composition from the incubation solution; under experimental conditions, aragonite crystals are easier to form on confined spaces (> 25 nm diameter tubes), but in nature such conditions do not exist. Mollusk nacre plates are greater than 25 nm diameter due to matrix shell protein interactions between minerals and fibroin-like framework molecules [60]. In this study, Ps19 was capable of binding chitin and Ca 2+ ions although according to the partial peptide sequence, it did not show any known domain associated to these characteristics, such as GN-repeat or GYS motifs, which have been observed in most matrix shell proteins [1,2,27,31,32]. As other SMPs, Ps19 binds polysaccharides and calcium ions, which are necessary for mineralizing calcium carbonate; therefore, Ps19 acts as an intermediary between calcium carbonate crystals and the organic matrix; moreover, it might function as an aragonite nucleation template according to in vitro crystallization assays although further studies are needed to corroborate this role in mineralization. These characteristics have demonstrate the importance of Ps19 in shell biomineralization.
These evidences highlight that despite Ps19 did not show any chitin-binding conserved domain, it can bind chitin and interact with Ca 2+ to nucleate aragonite plates in vitro. Therefore, other chitin-binding sequences must exist, as well as other calcium-binding domains. Similarity with a BTB/POZ domain may suggest Ps19 has an specific domain for interacting with other proteins involved in biomineralization, also, similarity with a serine beta-lactamase-like protein suggest it may have enzymatic activity, but this is yet to be tested.

Conclusions
This study described the biochemical characteristics of Ps19 from the shell of bivalve Pteria sterna. Ps19 is a glycoprotein with calcium and chitin-binding capability, which are important characteristics for aragonite nucleation. Ps19 is capable of cristallizing calcium carbonate in vitro by forming aragonite plates that are necessary to structure nacre. Further research must be performed to obtain the full protein sequence of Ps19 to characterize its structure and understand its potential interaction with calcium and chitin molecules to form nacre in mollusk shells. Writing -original draft: Raquel G. Arroyo-Loranca.