Around two hundred limpets Cellana toreuma, with shell length 3-4 cm, were collected around July in 2018 from the rocks in the intertidal zone at Zhou Shan beach, Zhejiang Province, China. The radulae were extracted from limpets, placed on petri dishes to maintain a flat shape, and immediately transferred to a -80°C freezer until further use. For the analyses reported here, freshly thawed radulae were thoroughly washed to remove all organic debris: (i) washed in water, (ii) immersed in 5% NaOH for 24h and (iii) again washed in water.

Total RNA was extracted from radulae tissue with the standard protocol of TRIzol reagent (Life Technology, Thermo, USA) for general cloning of the genes. The quality of RNA was determined by measuring the OD_260/280, OD_260/230 and the quantity of RNA was evaluated by OD₂₆₀ with a NanoDrop Lite spectrophotometer (Thermo Scientific, USA). The extracted RNA was used to synthesize single-stranded cDNA for RACE experiments using the SMARTer RACE cDNA Amplification Kit (Clontech, Japan). 3’RACE-F1, F2 and 5’RACE-F1, F2 primers were used with the primers supplied with the 3’- and 5’-RACE kits, respectively. The full-length sequence of CtCBP-1 was confirmed by the primers CtCBP-1-F and CtCBP-1-R. The deduced amino acid sequence was determined bioinformatically using the ExPASy Translate tool and the signal peptide was predicted by SignalP4.1.

The coding region of gene CtCBP-1 without signal peptide was amplified by the primers CtCBP-1-pMAL-F and CtCBP-1-pMAL-R ( Table 1 ). In addition, the CtCBP-1-pMAL-R primer contained a His6-tag, and then the amplified fragment was inserted into the pMAL-c5X vector (New England BioLabs Inc., NEB; USA) downstream of the E. coli malE gene encoding maltose binding protein (MBP) to promote the expression of the “soluble” fusion protein MBP tagged CtCBP-1. CtCBP-1 was purified for further functional analyses.

Purified recombinant plasmid pMAL-CtCBP-1 was transformed into Escherichia coli strain Transsetta (DE3) (Transgene, China) for expression. The recombinant cell was cultured in LB(Transgene, China) at 37 °C, and add 0.5 mM isopropyl 1thio-β-D-galactopyranoside (IPTG; SIGMA, USA) when the cell density reached 0.6-0.8, then cultured at 16 16°C overnight. Cells were then collected by centrifugation at 8000 g and 4°C for 5 min. The supernatant was removed and the cell pellet was resuspended with lysis buffer (50 mM Tris, 300 mM NaCl, 20 mM imidazole, pH=8.0). The cells were disrupted by ultrasonic homogeniserc. Then cells were disrupted via ultra-sonic dismembrator (Sonics & Materials Inc., USA) at 30% power with 4 s pulse on and 6 s pulse off repetition cycle on ice. After centrifugation 12000 g for 40 min at 4 Inc., USA) at 30% poweraliquoted into a fresh tube and incubated for 1h at 4°C with gentle rocking in the presence of Nickel-NTA agarose slurry (NEB, USA) that had been pre-equilibrated in lysis buffer. The agarose was washed with washing buffer (50 mM Tris, 300 mM NaCl, 40 mM imidazole, pH=8.0). Finally, the protein rCtCBP-1 was eluted using elution buffer (50 mM Tris, 300 mM NaCl, 400 mM imidazole, pH=8.0) then desalted using Hitrap desalting column (GE Healthcare, USA) to a storage buffer (50 mM Tris, 500 mM NaCl, pH=8.0) and stored at 4°C for future experiments. The fractions were analyzed through SDS-PAGE (12% (w/v) acrylamide) followed by staining with Coomassie blue. The purification of MBP was performed according to the instruction of pMAL-c5x (NEB) (GE Healthcare, USA).

Synthesis of goethite was according to the literature with modifications (Mohapatra et al., 2005). The steps are as follows: Stock solutions containing 1M ferric nitrate (initial pH 0.92 at room temperature) and 10 M sodium hydroxide were prepared using double distilled water. The pH was then adjusted with 10 M sodium hydroxide until a final value of 11.8 was obtained, and the total volume of slurry was increased to 500 mL. The precipitated slurries were aged for 24 hours at 333 K followed by filtration and washing with distilled water until the filtrates were nitrate-free. The prepared products goethite were dried for 24 hours at 373 K. 10 mg goethite (laboratory preparation) or chitin (Biodee, China) was washed and supplemented with 1 mL storage buffer (50 mM Tris, 500 mM NaCl, pH=8.0) respectively. 60 μg rCtCBP-1/MBP were mixed with prepared goethite or chitin respectively, and the mixtures were incubated at 4 °C for 1 h with gentle rolled. After incubation, the pellets were washed with 1 ml Milli-Q deionized water and storage buffer 3 times each and the liquid flowing through was collected to detect the binding abilities of different proteins. Then the washed materials were also collected separately and treated with loading buffer. SDS-PAGE was performed to analyze the binding properties of CtCBP-1 and MBP.

The morphologies of the cleaned three radular were examined by scanning electron microscope (SEM) (FEI Quanta 200, USA) after being sputter-coated with a thin layer of gold nanoparticles.

Synthesis of goethite was according to the literature with modifications (Mohapatra et al., 2005). In the control group, a quantity of 1 M sodium hydroxide was added to 500 μL of 1M ferric nitrate solution until the pH reached 12.0. The precipitated slurries were aged at 60°C for 24 h, then centrifuged at 3000 rpm for 5 min and washed with distilled water until the filtrates were free of nitrate and sulphate. The prepared products were dried at 100°C for 24 h. In the experiment group, proteins CtCBP-1 (35 µg/ml, 70 µg/ml,140 µg/ml) or MBP (35 µg/ml,70 µg/ml,140 µg/ml) were added to the ferric nitrate solution respectively.

We used Raman spectroscopy to examine samples (negative control, CtCBP-1 (35 µg/ml, 70 µg/ml,140 µg/ml) and MBP (35 µg/ml,70 µg/ml,140 µg/ml)) from in vitro crystallization experiments. The Raman spectra were scanned three times for 20 s in the range of 100 to 1400 cm^-1 using a LabRAM HR Evolution spectrometer (HORIBA JobinYvon, Japan).

We used XRD to examine samples (negative control, CtCBP-1 (35 µg/ml, 70 µg/ml,140 µg/ml) and MBP (35 µg/ml,70 µg/ml,140 µg/ml)) from in vitro crystallization experiments. XRD (D8 ADVANCE, Bruker, Germany) was applied in the range 10-90 (2 theta degree).

The recombinant proteins CtCBP-1 and MBP were labeled with Cy5, respectively. The purified recombinant proteins CtCBP-1 and MBP were desalted into labeling buffer (0.1 M phosphate, 500 mM NaCl, pH=8.3) respectively and the concentration was estimated using A280. The dye solution was prepared by dissolving 1 mg Cy5 NHS ester dye (AAT Bioquest, USA) in 100 μL DMSO and then diluting to 0.1 μg/μL with DMSO. The amount (mg) of the dye used was determined by an empirical formula (0.01 × 855 × protein weight (mg)/25,000). 1mg of the protein and the corresponding volume of the dye were uniformly mixed and then reacted in the dark for 30 minutes at room temperature respectively. The labeled proteins were then desalted in an AKTA system(GE, USA) by HiTrap TM Desalting(GE,USA) into storage buffer(20 mM Tris-HCl, 500 mM NaCl, pH = 7.8). Labeled CtCBP-1 and MBP were further used for in vitro distribution experiments. 100 µL of 100 μg/ml labeled CtCBP-1 and MBP were further incubated with NaOH-treated radula for 30 minutes in the dark. The teeth were then washed three times with water and a high salt ion concentration buffer to wash away non-specifically bound proteins. And samples were immersed in buffer and then observed by super-resolution Ti NSIM system (Structured illumination microscopy, Nikon, Japan).

We previously identified six chitin-binding proteins that may play an important role in tooth mineralization through proteomic studies of limpet teeth. Based on the abundance of expression at the transcriptome level, we selected the most abundant CtCBP-1 protein for more detailed studies. We applied RACE procedures to clone the 5’ and 3’ flanking sequences to obtain a complete transcript for CtCBP-1 (GeneBankTM accession number ON263363). The CtCBP-1 gene has a full length of 1236 bp ( Figure 1A ), and the open reading frame encodes 291 amino acids. CtCBP-1 has a theoretical molecular weight of 29.4 kDa and a theoretical isoelectric point of 4.57, which is an acidic protein. Its amino acid content distribution showed that the three amino acids with the highest abundance were Gly (27.15%), Asn (17.18%), Ser (6.87%), and the cysteine content was also high, with a content of 4.47%. Glycine was found to be dominant in the amino acid analysis of mature teeth (Ukmar-Godec et al., 2015), while cysteine was found to be able to directly promote the formation of goethite at room temperature in vitro ( Cornell and Schneider, 1989).

The BLASTp software was used to align the homologous sequence of the CtCBP-1 protein in the GenBank database. As a result, we only identified a homologous 46.26% protein sequence in limpet Lottia gigantea. It suggested that CtCBP-1 may be a protein sequence unique to limpets. The homologous sequences of CtCBP-1 and hypothetical protein LOTGIDRAFT_238408 were mainly located in a conserved functional region and were chitin-binding domains. In addition, there was no functional study of proteins, so the CtCBP-1 protein was a completely new unknown protein.

Using the SignalP 4.1 program (Nielsen, 2017), we found the presence of an N-terminal signal peptide in CtCBP-1, implying that it is a secreted protein. By the SMART software, we found that CtCBP-1 had two chitin-binding domains ( Figure 1B ). So it was possible that the protein was secreted outside the cell after expression and bound to the chitin framework to exercise functions during mineralization. In addition, through RADAR software, it was found that this protein had a repetitive sequence of about 50 AA long NG at the C-terminus ( Figure 1B ). These tandem repeats are common features in bio-mineralization proteins (Miyamoto et al., 1996), bioadhesive proteins (Liu et al., 2015) as well as chitin-binding proteins in the beak (YerPeng et al., 2015).

The recombinant CtCBP-1 was expressed with vector pMAL-c5X and sufficiently purified for in vitro functional studies ( Figure 1C ). The molecular mass of CtCBP-1 (81.4 kD) is almost identical to calculated value (CtCBP-1 with His-tag 12 kD plus MBP 40 kD). What’s more, mass spectrum was performed to confirm the protein sequence of CtCBP-1. We also expressed and purified MBP as a negative control ( Figure 1D ) for further functional analyses. In this research, the recombinant protein MBP-CtCBP-1 with a His6-tag on N-terminal, CtCBP-1, was used for further functional analyses.

Since protein attachment may play a key role in regulating crystal characteristics, it is important to perform goethite and chitin-binding assays to explore the affinity and interaction between CtCBP-1 and different crystal polymorphs. CtCBP-1 was incubated with chitin and goethite, respectively, and then eluted with deionized water, 0.5 M NaCl to wash away the nonspecifically absorbent protein before analyzing the sample by SDS-PAGE. We found that in the MBP control group, deionized water and 0.5M NaCl solution can elute MBP from chitin and goethite. However, in the CtCBP-1 experimental group, CtCBP-1 was separated from chitin and goethite only when the strong denaturing agent existed ( Figure 1E ). The above results demonstrated that CtCBP-1 expressed in vitro can effectively bind to chitin and goethite, which provided a basis for possible regulation of crystal growth.

To verify the role of protein CtCBP-1 in the formation of goethite, in vitro crystallization experiment was performed. Protein CtCBP-1 was added into the reaction at final concentrations of 35, 70 and 140 μg/ml, respectively, and established the same final concentration (35, 70 and 140 μg/ml) of MBP protein as a control to rule out of the effect of the tag. Compared to the typical needle-shape goethite crystals in the negative control group ( Figure 2A ), the crystals in the MBP group showed no obvious morphology alteration ( Figures 2C, E, G ), which seemed similar to the goethite needles in the leading parts of the limpet teeth (Wang et al., 2018), indicating that MBP tag had no effect on the crystal polymorphs. On the contrary, the crystals in the CtCBP-1 group showed significant changes in crystal morphologies. The morphology of the crystal changes from a typical needle-like goethite morphology to dense precipitates with much smaller crystals and the degree of the morphology alteration increased with elevated concentrations of CtCBP-1 from 35 to 140 μg/mL ( Figures 2B, D, F ), suggesting that CtCBP-1 affected the crystal morphology. Moreover, it is interesting to note that the altered crystal morphology is similar to the crystal morphology reported in the previous study which was changed by the addition of the chitin framework (Wang et al., 2018). Crystals were analyzed by Raman spectroscopy to confirm the crystal form and indicated that most of the particles in each group were goethite with specific peaks at around 240, 296, 396, 476, 547, and 680 cm-1, indicating that CtCBP-1 and MBP had no effect on the crystal polymorphs ( Figure 3 ).

In the presence of CtCBP-1 protein, the morphology of goethite formed in vitro was dissimilar to that of needle-like crystals located at the leading edge of the cusp ( Figure 4 ), indicating that CtCBP-1 may be involved in regulating the crystal morphology during the formation of goethite in vivo. In contrast, in the absence of CtCBP-1, the morphology of goethite synthesized in vitro was similar to that of needle-like crystals at the trailing edge of the cusp ( Figure 4 ), except that it was less oriented. This may be due to the in vivo chitin fibers assisting in the orientation of the crystals.

By XRD, it is also shown that in the presence of 140 μg/mL CtCBP-1, the crystals become amorphous to a large extent as the peak at 25 degree is not sharp ( Figure 5D ). While the other samples and control groups ( Figure 5A ) showed goethite crystal type ( Figures 5B, C–G ). This indicates that a high concentration of CtCBP-1 can also affect the crystallization of goethite by forming amorphous particles. The inconsistency between Raman and XRD may be due to the fact that Raman spectroscopy is a local detection method with a detection range of only a few microns and its small laser penetration depth (~5.0 μm), which is not representative of the overall structure. However, another analysis, such as TEM, is needed to clarify the results.

The previous study found that CtCBP-1 protein affected the morphology of needle goethite formation and that there is a concentration effect on this effect. To gain more insight into how CtCBP-1 affects goethite crystallization, we labeled CtCBP-1 with the fluorescent dye Cy5-NHS ester and examined the localization of CtCBP-1 on the cusp. Super-resolution images were harvested at a Z-axis resolution of 0.1 μm and reconstructed as three-dimensional images by Nikon structural illumination microscopy (NSIM) at an excitation wavelength of 645 nm ( Figure 6 ). It can be noted that the fluorescence signal shows a non-uniform distribution from the anterior to the posterior part of the cusp ( Figures 6A, B ). The CtCBP-1 protein labelled with Cy5 was found to bind specifically to the anterior margin of the cusp, concentrating primarily on the marginal side, whereas the MBP protein from the control ( Figures 6C, D ), which was uniformly distributed across the cusp, has no spatial specificity. In the mollusk chiton Acanthopleura hirtosa, which also has a radulae structure, the anterior and posterior portions of its teeth have different densities and arrangements of chitin fibers. The chitin fibers in the anterior part of its tooth cusps gradually become sparse and lack order. In contrast, the posterior part of the tooth cusps become more organized (K. et al., 1989). We speculate that this binding result of CtCBP-1 protein is due to the fact that it contains chitin-binding sites, resulting in a higher distribution of CtCBP-1 protein in the anterior edge of the cusp where chitin fibrils are densely arranged, while chitin fibrils are sparsely arranged and have fewer protein-binding sites at the posterior edge of the cusp. Combined with previous crystallization experiments, we hypothesized that CtCBP-1, which is specifically bound at the anterior margin of the cusp, could help goethite needles to align more closely at this site or bind to the chitin fiber frame as a specific nucleation site.