K. Kamino, National Institute of Technology and Evaluation, 2-5-8 Kazusa, Kisarazu 292-0818, Japan Fax: +81 438 20 5766 Tel: +81 438 20 5764 E-mail: firstname.lastname@example.org
Barnacles are a unique sessile crustacean that attach irreversibly and firmly to foreign underwater surfaces. Its biological underwater adhesive is a peculiar extracellular multi-protein complex. Here we characterize one of the two major proteins, a 52 kDa protein found in the barnacle cement complex. Cloning of the cDNA revealed that the protein has no homolog in the nonredundant database. The primary structure consists of four long sequence repeats. The process of dissolving the protein at the adhesive joint of the animal by various treatments was monitored in order to obtain insight into the molecular mechanism involved in curing of the adhesive bulk. Treatments with protein denaturant, reducing agents and/or chemical-specific proteolysis in combination with 2D diagonal PAGE indicated no involvement of the protein in intermolecular cross-linkage/polymerization, including formation of intermolecular disulfide bonds. As solubilization of the proteins required high concentrations of denaturing agents, it appears that both the conformation of the protein as building blocks and non-covalent molecular interactions between the building blocks, possibly hydrophobic interactions and hydrogen bonds, are crucial for curing of the cement. It was also suggested that the protein contributes to surface coupling by an anchoring effect to micro- to nanoscopic roughness of surfaces.
Database Sequence of Megabalanus rosa cp52k mRNA for 52 kDa cement protein has been submitted to the DNA Data Bank of Japan under accession number AB623048.
fragments derived from whole cement proteins by CNBr cleavage
barnacle cement proteins, followed by apparent molecular weight estimated by SDS/PAGE
mussel foot protein
GSF1 and GSF2
cement fractions separated according to their solubility in GdnHCl
Attachment of materials in water is a difficult task, and man-made adhesives for under-water attachment are still largely undeveloped. However, diverse biological processes have been developed in a variety of organisms to make underwater attachment possible . Attachment to abiotic surfaces is crucial for the organisms’ lifecycles and is controlled by complex physiological networks. Recent analysis of the biological adhesives in mussel (Mytilus sp.) , sandcastle worm (Phragmatopoma californica)  and barnacle (Megabalanus rosa)  indicated that each employs a complex of proteins for the purpose. The complex nature of the adhesives may suggest difficulties in firm and durable underwater attachment  even in nature.
The adhesives must maintain fluidity during the secretion process from their secretory organ  to the extracellular space, and then undergo self-assembly and irreversible curing for formation of a durable adhesive joint. Studies on mussel and tubeworm have shown that polymerization of proteins via intermolecular cross-linkages is essential for curing of the adhesive, with the multi-functional amino acid peptidyl 3,4-dihydroxyphenylalanine playing a central role . The dependence of polymerization on coordination/covalent bonding in the formation of mussel byssus may be due to the requirement for rapid formation of an adhesive joint with almost full strength due to the environment that the animal inhabits .
Barnacles are a unique sessile crustacean that attach irreversibly to various foreign materials for most of their lifecycle. This attachment is based on secretion of a biological underwater adhesive [9,10]. Parts of barnacles have a calcareous basis (Fig. 1A), and the adhesive (cement) firmly attaches to two materials in water, of which one is always their own inorganic shell. The foreign materials to which barnacles attach include metals, minerals, wood and synthetic polymers , although the diversity of these surface properties may not be evident in nature due to colonization by microbial biofilms. Peptidyl 3,4-dihydroxyphenylalanine has never been identified in barnacle cement [12,13], and there are apparently no homologies between barnacle cement proteins and other adhesive proteins, e.g. from mussel and tubeworms . Involvement of intermolecular cross-linkages or polymerization has not yet been found during self-assembly and curing of barnacle cement . The mode of attachment is different from those of mussel and tubeworm. Barnacle cement is generally secreted to an enlarged marginal area of their bases during periodic growth [15,16]. The old cement layer formed previously in the central area supports the attachment sufficiently for the time being, thus the adhesive joint that is being newly prepared in the enlarged area may not be required to have higher adhesive strength as required for mussel byssus . Further, barnacle cement is present as a thin layer to attach and support a much larger area: the ratio between the diameter and thickness of adhesive joints is 104 in barnacles but 102 in tubeworms. Thus, barnacle cement may have a unique self-assembly and curing mechanism among biological adhesives.
Limited attention has been given to the conformation of proteins as building blocks in biological adhesive proteins. A 20 kDa barnacle cement protein (cp20k) , which is a possible surface functional protein, is the only such protein whose 3D structure has been defined [18,19]. The primary structure of foot protein 2 (fp-2) in mussel byssus comprises tandem repeats of an epidermal growth factor-like motif , and thus may adopt a specific conformation, although no information on its conformation is available yet. However, foot protein 1 (fp-1) in the layer coating the mussel byssal thread is thought to have a random polymer like conformation . No further studies have been performed regarding the conformations of other proteins or their relationship to the functionality in underwater attachment.
The barnacle cement at the joint is insoluble , and is usually limited in amount because the thickness of the joint is only a few micrometers [22–24]. These difficulties have prevented direct analyses of the natural cement. However, it has been observed that intact animals that have been carefully dislodged from their substratum and maintained in seawater occasionally secrete a white opaque rubbery secretion at the outer surface of the calcareous basis that was once attached to the foreign material . This material is called the secondary cement , and has been useful, especially for identification of the cement components, due to the availability of larger sample quantities. It has been confirmed that the secondary cement is equivalent to the natural cement in terms of protein composition [13,25], although the degree of self-assembly and curing may not be the same between the natural and secondary cements . Only two methods to make the cement soluble have been developed. The proteolytic method involves cleavage of the Met–X bond by cyanogen bromide (CNBr) , and the non-proteolytic method involves treatment with an excess of reducing agents and denaturants at an elevated temperature . Combination of the two methods revealed that three cement proteins, namely cp100k, cp52k and cp68k, are major components of barnacle cement in terms of their amount , and the former two proteins are essential for the insoluble nature of the cement. The cDNA for the cp100k has been isolated , and it has been shown that this protein is unique as it has no homolog in the non-redundant GenBank CDS translations + PDB + SWISSPROT + PIR + PRF database (ftp://ftp.ncbi.nih.gov/blast/db/). However, no information is available for cp52k. Furthermore, almost no biochemical data on the natural cement have been reported.
In this study, the cDNA of cp52k was cloned, and the protein was fully characterized. The results obtained indicated that both the conformation of the protein as building blocks and non-covalent interactions of the protein are crucial for curing of the bulk cement. No evidence was found for involvement of the protein in intermolecular cross-linkage/polymerization in the cement.
Cloning of the 52 kDa cement protein cDNA and structural outline of the protein
The N-terminal amino acid sequence of the mature polypeptide of cp52k was determined to be TGSRPYFPVS. The cDNA of cp52k (Appendix S1) had a length of 1950 bp, and the coding region was determined to be 564 amino acids long (Fig. S1). Amino acid sequences of two peptide fragments, CB-1 and CB-4 , derived from the cement by CNBr cleavage of the Met–X bond, were found to be present in the deduced sequence, and were virtually in agreement with the deduced sequence except for a few single amino acid variations (W293F, I308L and I312L). An MS/MS ion search for cp52k in the natural cement using the Mascot search engine (http://www.matrixscience.com) with LC-MS/MS analysis gave a high score and matched the deduced sequence from the cDNA of cp52k (Fig. S2 and Table S1). The relative molecular mass and isoelectric point of the mature polypeptide of cp52k were predicted to be 62 319.6 and 10.5, respectively. The deduced sequence showed that hydrophobic residues constituted 49% of total residues, and hydrophilic and neutral residues constituted 29% and 22%, respectively (Table S2). The primary structure consists of four long tandem sequence repeats, each consisting of 113–129 amino acids (Fig. 2). The full sequence contains six Cys residues (1.1% of the total residues), and each repetitive sequence contains one Cys residue that is located at almost the same region in each repetitive sequence. The other two Cys residues are located in the C-terminal region of the protein.
A BLAST search of the non-redundant GenBank CDS translations + PDB + SWISSPROT + PIR + PRF database (ftp://ftp.ncbi.nih.gov/blast/db/) failed to yield any homologous proteins. No sequence similarity between cp52k and foot proteins in the mussel (fp-1–6 and precursors of collagens in the byssal thread)  or the cement proteins of tubeworm (Pc1–3)  was observed. The primary structure of cp52k also showed no homology to other cement proteins of barnacle, i.e. cp100k, cp19k  and cp20k . A Northern blotting analysis, using cp52k DNA as the probe, indicated that its mRNA is specifically expressed in the basal portion of the soft tissue in which the cement gland is located (Fig. S3). Specific staining for detection of glycosylation in cp52k gave a positive result (Fig. 3). Treatment with N-glycanase resulted in disappearance of the signal, thus the protein is N-glycosylated. Asn261 was predicted to be the N-linked glycosylation site in a search for both N- and O-glycosylation using the NetNGlyc 1.0 server (http://www.cbs.dtu.dk/services/NetNGlyc/) and the NetOGlyc 3.1 server (http://www.cbs.dtu.dk/services/NetOGlyc/), which use artificial neural networks to predict glycosylation sites in proteins on the basis of the sequence context. However, although the presence of Pro262 immediately after Asn261 may prevent glycosylation due to conformational constraints. Asn513 was also predicted to be a possible N-linked glycosylation site; however, its score was slightly lower (0.44) than the threshold value (0.5), and it is not clear whether Asn513 is indeed glycosylated.
Characterization of natural cement
In order to identify the location of cp52k, the barnacle base plate on polyethylene substratum and peripheral shell were separately recovered (Fig. 1A) and probed by Western blotting after electrophoretic separation. The results showed that cp52k is present in the protein fraction from the calcareous base, but not from the peripheral shell (Fig. 1B).
The process of dissolving the natural cement bulk was investigated by using a combination of treatments and probing the proteins by Western blotting (Fig. 4). First, solubilization of the bulk proteins using a protein denaturant and additional reducing treatment was examined (Fig. 4A). Fraction obtained by solubilization with 7 m guanidine hydrochloride (GdnHCl) solution at pH 7, 60 °C, was designated as GdnHCl-soluble fraction 1 (GSF1), and that solubilized with 7 m GdnHCl plus dithiothreitol at pH 8.5, 60 °C, was designated as GdnHCl-soluble fraction 2 (GSF2), which correspond to fractions obtained from the secondary cement . Each treatment was repeated three times, and the soluble fractions obtained each time were analyzed. The results showed that a single treatment of the natural cement with 7 m GdnHCl at pH 7, 60 °C, rendered a limited amount of cp52k soluble. A second round of the same treatment solubilized another fraction of the protein, but the third round of treatment gave no band on Western blotting. Treatment with dithiothreitol in 7 m GdnHCl at pH 8.5, 60 °C, rendered a greater amount of the protein soluble. The proteins found in both GSF1 and GSF2 of the natural cement had the same mobility on SDS/PAGE and the same N-terminal sequences. No band with a molecular mass higher than 52 kDa was detected in either GSF1 or GSF2.
To investigate the effects of GdnHCl or dithiothreitol separately, these treatments were performed independently. Dithiothreitol treatment of the natural cement in the absence of GdnHCl and heating did not render the protein soluble (Fig. 4B). Treatment for the GSF1-preparation rendered all of cp52k soluble. Thus, heating at 60 °C (Fig. 4C) and higher concentrations (7 m) of GdnHCl (Fig. 4D) were found to be indispensable to render the protein soluble. It was also confirmed that treatment with lower concentrations of GdnHCl (2 or 4 m) in the presence of dithiothreitol at pH 8.5, 60 °C, did not render the protein soluble (Fig. 4E).
Because the sample of natural cement used in this study was accompanied by calcareous base shell (see Experimental procedures), the residual material after treatment with both dithiothreitol and 7 m GdnHCl at 60 °C was further examined following decalcification of the shell portion and subsequent CNBr cleavage. Unexpectedly, the residual material produced typical CB peptides  (data not shown) derived from both cp100k and cp52k by CNBr cleavage. To evaluate the amount of cp52k that remained insoluble, each fraction was subjected to CNBr cleavage, and the peptide fragments were detected by Western blotting (Fig. 5A). The comparison indicated that full treatment with both dithiothreitol and 7 m GdnHCl at 60 °C without decalcification resulted in solubilization of approximately half of the proteins, leaving the other half insoluble (ratios varied somewhat between samples). Because the surface of the calcareous shell is microporous, it was assumed that the cement is trapped in the microporous structure of the calcareous base shell, resulting in limited accessibility of the chemicals used to render the cement soluble. Therefore, the cement was first subjected to decalcification prior to dithiothreitol/7 m GdnHCl treatment at 60 °C. The combination of decalcification and dithiothreitol/GdnHCl treatment at 60 °C solubilized all of the protein, and the processes of solubilizing the cement were the same (Fig. 5B). No band of molecular mass higher than the monomer peptide of cp52k was present on the Western blot (Fig. 5B).
The cp52k protein in GSF1 of the natural cement was subjected to investigation with regard to whether a disulfide bond(s) forms intermolecularly or not. 2D diagonal PAGE of GSF1 of the natural cement gave only one spot on the diagonal line (Fig. 6A), indicating that the protein has no intermolecular disulfide bond. Western blotting of GSF1 after CNBr cleavage gave two bands at 27 and 24 kDa under non-reducing conditions on SDS/PAGE, but only one band at 12 kDa was obtained under reducing conditions (Fig. 6Bi).
The disulfide bonding of the protein that remained insoluble after treatment to obtain the GSF1 preparation was further characterized. The residual material after GSF1 preparation was decalcified, and subjected to CNBr cleavage. The results for this fraction correspond with those obtained for GSF2 prepared under non-reducing conditions. The pattern on Western blotting under non-reducing conditions was the same as that of GSF1 (Fig. 6Bii).
A necessary condition for a protein to be a cement component is localization at the adhesive joint between the base shell and foreign material. We confirmed that cp52k is localized at the natural barnacle adhesive joint between the barnacle’s calcareous base and foreign material. Absence of the protein in the fraction from the peripheral shell indicated that it is not involved in other functions such as calcification. The results are consistent with those of other barnacle cement proteins [18,29], and the proteins are confirmed to be the cement components. Expression of the mRNA in the basal portion of the soft tissue where the cement gland is located is also consistent with the location of other barnacle cement proteins [17,26,29]. Assignment of two fragment peptides, CB-1 and CB-4, to the primary structure of cp52k indicated that all eight of the longer peptide fragments, which are detected in SDS/PAGE of the whole cement after CNBr cleavage , are derived from either cp100k  or cp52k. The insoluble nature and high amount of cp52k indicated that it is a possible bulk cement protein, and is involved in the processes of self-assembly and curing during formation of the adhesive joint of barnacle.
It is essential to understand how the bulk of the cement self-assembles and is cured in water. Monitoring of processes leading to solubilization of the bulk protein is a valid approach to assess the dominant mechanism for curing.
A primary question is whether intermolecular cross-linking occurs or not. The soluble fraction prepared by treatment with 7 m GdnHCl in the presence of dithiothreitol contained no bands of molecular mass higher than that of the monomer of the protein. CNBr cleavage of the whole natural cement also yielded only a peptide fragment with the molecular mass of the monomer under reducing conditions, consistent with the above results. These results clearly indicate that no intermolecular cross-linkages occur in the bulk protein of the natural cement, although whether intermolecular disulfide bonding is present in the protein remains to be elucidated by the results mentioned above.
Western blot analysis with various combinations of treatments confirmed that treatments involving both a protein denaturant and a reducing agent are indispensable for destruction of the adhesive bulk. The requirement for a reducing agent implies the existence of disulfide bonding in the bulk proteins, either inter-molecular or intramolecular cross-linking. The cp52k protein in GSF1 of the natural cement was investigated to determine whether intermolecular disulfide bond(s) exist in the protein. 2D diagonal PAGE of cp52k in GSF1 indicated that the protein does not contain intermolecular disulfide bonds, at least in the fraction examined. The protein that remained insoluble showed the same pattern of disulfide bonding as in GSF1, because Western blotting analysis following CNBr cleavage under non-reducing conditions indicated linkages of peptide fragments containing Cys residue(s) in the protein (see below). In conclusion, the protein in the natural cement is likely to have no intermolecular disulfide bonding. Cys residues in the protein probably form intramolecular disulfide bonds, which may contribute to stabilization of the protein conformation as building blocks as well as optimization of the protein–protein interactions required for formation of the bulk of the cement.
The absence of intermolecular covalent cross-linkages suggests the significance of molecular interaction. Another aspect that appeared crucial for solubilization of the proteins is the higher concentrations of protein denaturing agent required. Treatments with < 4 m GdnHCl were insufficient to solubilize the bulk cement despite the reducing conditions employed and heating at 60 °C. Higher concentrations of protein denaturant are generally required for complete unfolding of protein conformations. The inadequacy of lower concentrations of GdnHCl indicates that complete denaturation of the protein conformation is required to destroy intermolecular interactions of bulk proteins in the natural cement. Thus, assembly of the building blocks to form the bulk of the natural cement relies on and is optimized by the conformation of the building blocks. GdnHCl is known to break hydrogen bonds and diminish hydrophobic interactions. Given the high abundance of hydrophobic amino acid residues in all biological adhesive proteins, molecular interactions, possibly hydrophobic, are suggested to be essential to link the molecular building blocks.
This study indicated that self-assembly and curing of the cement bulk rely on molecular interactions and the protein conformation, which maximizes the interactions. This contrasts with mussel byssus and tubeworm cements, in which polymerization via covalent cross-linkages plays a crucial role. Non-proteolytic methods are generally ineffective in solubilizing proteins in mussel byssus and tubeworm cements. fp-2, which forms the bulk of the mussel byssal disk, has been shown to contain no non-covalent protein–protein interactions , thus intermolecular cross-linking of the protein must be essential for the assembly. The lower complexity of the primary structures of the proteins in tubeworm cement may facilitate formation of a random polymer-like structure that is thought to contribute to optimiza-tion of the functionality of side chains of amino acids, such as the catechol group in 3,4-dihydroxyphenylalanine, for cross-linkages (and surface coupling). These differences in the molecular systems used in the adhesives from different organisms are supported by the fact that no hydrophobic components in either mussel byssus and tubeworm cements were found. The cross-linkage-dependent curing mechanism in mussel may be due to the requirement for rapid curing . In contrast, barnacles may have time to cure newly formed cement due to its mode of attachment. This time may be essential to rearrange mutual topology and/or conformational changes of the bulk proteins to maximize interaction for tenacious adhesive joint, as observed for rearrangement of protein surface adsorption [31,32]. The properties of the proteins found in barnacle cement suggest that the molecular mechanisms for production of biological adhesive joints in water are complex and diverse.
A proportion of the barnacle natural cement was not rendered soluble without decalcification of the accompanying calcareous shell. This may suggest permeation of the bulk protein into the calcareous base, and that the protein also contributes to surface coupling through anchoring to micro- to nanoscopic roughness of surfaces in the base shell and foreign materials. Foreign materials adhered to by biological adhesives include hydrophobic ones, although much less attention has been focused on surface coupling to underwater hydrophobic surfaces by biological underwater adhesives. The present study focused on curing, and although the contribution of the protein to surface functionality in the attachment was not addressed, the involvement of cp52k in hydrophobic surface coupling cannot be excluded. Some studies have suggested that barnacle cement is a functional amyloid [25,26,33]. Amyloid fibrils are stabilized by hydrophobic and electrostatic interactions, especially through formation of a network of hydrogen bonds . Although it is unclear how the fibrous structure is involved in surface coupling and formation of the adhesive joint, barnacles appear to have considerably complex interactions and cascades of events to produce their tough adhesive joint. These need to be addressed in future studies.
There was a discrepancy between the apparent molecular mass estimated by SDS/PAGE (52 kDa) and that deduced from the cDNA sequence (62 kDa). The results of direct N-terminal amino acid sequencing and Western blotting analysis after CNBr cleavage, which produced a band with a molecular mass corresponding to that of the C-terminal region (Fig. 5A, CB52k-IX in Fig. 7, and Fig. S1), indicated that no proteolytic processing occurred in the protein. LC-MS/MS analysis of the protein coupled with a Mascot search also detected peptides from the C-terminal region. The absence of bands in at a molecular mass lower than 52 kDa on Western blotting also supported no proteolytic processing of the protein. These facts suggest that the protein has abnormal migration behavior on SDS/PAGE. A similar discrepancy between the molecular mass estimated from SDS/PAGE and that calculated from the sequence was reported for the cp100k protein . These differences may result from the high hydrophobicity of the proteins.
Glycosylation has been reported in fp-1 of zebra mussel  and green mussel , although fp-1 actually functions as a coating of the byssal thread. N-glycosylation of cp52k is the first example of a post-translational modification among barnacle cement proteins, although its significance requires further studies.
The primary structure of cp52k has no simple similarity to those of other cement proteins of barnacle. In addition to the low similarity of primary structures and the varied molecular masses of barnacle cement proteins, they are characterized by biased amino acid compositions , and can thus be classified into three types: a hydrophobic protein (cp100k), a hydrophilic protein rich in charged amino acids (cp20k), and two proteins rich in Ser, Thr, Gly, Ala, Val and Lys residues (cp19k and cp68k). Although the similarity of their sequences is limited, cp100k and cp52k are similar in several ways: they are the most common proteins in the cement , are extremely insoluble in most of the solvents/treatments , are the most hydrophobic of the cement proteins, and are similar in amino acid composition to some extent (Table S2). Thus cp52k may belong to the same group as cp100k.
The Western blotting analysis with CNBr cleavage under non-reducing conditions suggested limited intramolecular disulfide bonding of cp52k (Fig. 7). In principle, ten peptide fragments (Fig. 7) are expected to be released by CNBr treatment, as the protein contains nine Met residues (Fig. S1). Of the six Cys residues in the protein, four are in the peptide fragments CB52k-II, -VI, -VII and –IX, and two are in the C-terminal peptide fragment (CB52k-X). The antibody used binds to two peptide fragments in the C-terminal region of the protein: one containing one Cys residue (CB52k-IX, molecular mass 12 385.5), and another containing two Cys residues (CB52k-X, molecular mass 821.4). The molecular masses of the three remaining peptide fragments containing a Cys residue are 5198.8 (CB52k-II), 12 764.9 (CB52k-VI) and 15 843.8 (CB52k-VII). Two bands on the Western blot under non-reducing conditions are thought to correspond to a heterodimer containing CB52k-IX and a heterotrimer containing CB52k-X. Thus, the pattern on the Western blot appears to indicate linkage of all five peptides containing Cys residue(s) in the protein. There are three possible linking patterns: (1) linkage of CB52k-IX with CB52k-VI (combined molecular mass 25 150.4) and linkage of CB52k-X with the remaining two peptides, CB52k-II and CB52k-VII (combined molecular mass 21 864.0), (2) linkage of CB52k-IX to CB52k-VII (combined molecular mass 28 229.3) and linkage of CB52k-X to the remain-ing two peptides (combined molecular mass 18 785.1), and (3) linkage of CB52k-IX to CB52k-II (combined molecular mass 17 584.3) and linkage of CB52k-X to the remaining two peptides (combined molecular mass 29 430.1).
In summary, this study fully characterized an abundant bulk protein in barnacle underwater adhesive, and presented evidence that curing of the protein at the adhesive joint is optimized via both the conformation of the protein and non-covalent intermolecular interactions, possibly hydrophobic interactions and hydrogen bonds. Barnacles have a unique molecular mechanism, at least for curing of the adhesive. Unraveling the structural/conformational determinant for its self-assembly will be the subject of future studies.
Preparation of the natural cement
Natural barnacle cement from the adhesive joint was prepared as follows. Barnacles attached to a polyethylene buoy (Fig. 1A) were carefully dislodged by powerful vibrating the substratum, and specimens that had no damage to their calcareous base shell were used to collect natural cement. The outside of the base shell was directly attached to the polyethylene substratum via the cement, therefore the cement must be localized on the outer surface of base shell plate. Preliminary experiments indicated that almost all of the cement remained on the surface of the barnacle base shell rather than the polyethylene substratum after dislodging. Immediately after dislodging from the substratum, the soft body was carefully and completely removed from the calcareous shell, and only the base shell plate was collected. The inner surface of the base shell plate was repeatedly cleaned with a toothbrush and pure water. The natural cement accompanied by base shell plates from several adult barnacles was mixed together and crushed, then divided into samples, for which the wet weight was determined.
Characterization of the natural barnacle adhesive
The natural cement was subjected to several treatments. The non-proteolytic treatments used to render the cement soluble comprised combinations of reduction by dithiothreitol, denaturing by GdnHCl, and/or heating at 60 °C. Reduction was performed using 0.5 m dithiothreitol at pH 8.5 under a nitrogen atmosphere. Protein denaturation was performed using 7, 4 or 2 m GdnHCl at pH 7 or pH 8.5 at room temperature or 60 °C. Tris/HCl and Hepes were used for buffering at pH 8.5 and pH 7, respectively. 7 m GdnHCl solution buffered at pH 7 using 10 mm Hepes was used for GSF1 preparation, and 0.5 m dithiothreitol and 7 m GdnHCl solution buffered at pH 8.5 using 0.1 m Tris/HCl was used for GSF2 preparation. Each treatment was performed twice for 1 h each unless otherwise indicated. The supernatant for each treatment was collected after centrifugation at 15 000 g for 10 min. Sediment after each treatment was washed with pure water overnight to shift to a subsequent different treatment. Supernatants were dialysed against 1% acetic acid at 4 °C, and stored at −20 °C.
Another method used to render the cement soluble by partial proteolysis was CNBr cleavage of Met–X bonds, where X is any amino acid. Samples of the natural cement were decalcified by dialysis against 5% acetic acid at 4 °C, and were evaporated/lyophilized prior to CNBr treatment. CNBr cleavage was performed in 1% CNBr/70% formic acid at ambient temperature for 24 h in the dark. The reactant was once evaporated to remove CNBr. The resultant peptides were dissolved in 70% formic acid, and then centrifuged at 15 000 g for 10 min at 4 °C. The supernatant was evaporated, freeze-dried twice, and then dissolved in denaturing buffer of SDS/PAGE containing 8 m urea and omitting reducing agent at 60 °C for 30 min.
Natural adhesive is a thin layer, typically a few micrometers thick, thus the amount of sample is limited. To assess possible environmental contamination of the cement, sensitive and specific detection using antibodies is required. Thus, we used Western blotting in combination with a series of sequential treatments of the natural cement. Western blotting analysis was performed using a rabbit polyclonal antibody against part of the cp52k protein, raised using a bacterial recombinant of the C-terminal region (∼ 10 kDa) as the antigen. The antibody for cp52k is expected to bind to two fragment peptides in the C-terminal region derived by CNBr cleavage of the protein, with relative molecular masses of 12 393 and 822, respectively. However, only the longer one is detected on Western blotting, because of the very low molecular mass of the small peptide, which makes it difficult to separate by SDS/PAGE.
A detailed summary of the Experimental procedures, including molecular cloning of cDNAs encoding cp52k and characterization of natural barnacle adhesive, is given in Appendix S2, which also includes relevant references.
We are deeply grateful to Professor J.-R. Shen (Faculty of Science, Okayama University, Japan) for his careful correction of the manuscript and discussion on protein denaturation. Part of this work was performed as an industrial science and technology project entitled ‘Technological Development for Biomaterials Design Based on Self-organizing Proteins’ and supported by the New Energy and Industrial Technology Development Organization of Japan.