The expression and characterization of recombinant cp19k barnacle cement protein from Pollicipes pollicipes

Abstract

Adhesive proteins of barnacle cement have potential as environmentally friendly adhesives owing to their ability to adhere to various substrates in aqueous environments. By understanding the taxonomic breath of barnacles with different lifestyles, we may uncover commonalities in adhesives produced by these specialized organisms. The 19 kDa cement protein (cp19k) of the stalked barnacle Pollicipes pollicipes was expressed in Escherichia coli BL21 to investigate its adhesive properties. Initial expression of hexahistidine-tagged protein (rPpolcp19k-his) yielded low levels of insoluble protein. Co-overproduction of E. coli molecular chaperones GroEL-GroES and trigger factor (TF) increased soluble protein yields, although TF co-purified with the target protein (TF-rPpolcp19k-his). Surface coat analysis revealed high levels of adsorption of the TF-rPpolcp19k-his complex and of purified E. coli TF on both hydrophobic and hydrophilic surfaces, while low levels of adsorption were observed for rPpolcp19k-his. Tag-free rPpolcp19k protein also exhibited low adsorption compared to fibrinogen and Cell-Tak controls on hydrophobic, neutral hydrophilic and charged self-assembled monolayers under surface plasmon resonance assay conditions designed to mimic the barnacle cement gland or seawater. Because rPpolcp19k protein displays low adhesive capability, this protein is suggested to confer the ability to self-assemble into a plaque within the barnacle cement complex.

This article is part of the theme issue ‘Transdisciplinary approaches to the study of adhesion and adhesives in biological systems’.

1. Introduction

Adhesion in wet environments is a complex problem and is notoriously difficult to accomplish. However, bioadhesion on the liquid–solid boundary in wet environments is commonly achieved by aquatic invertebrates such as molluscs, tubeworms, barnacles and echinoderms, which can adhere to a diverse range of natural and man-made materials [1]. Understanding natural mechanisms of adhesion has huge potential for the design of new materials with biomedical applications [2], for cell adhesion materials used in tissue culture [3] as well as for environmentally friendly anti-fouling agents [4]. However, despite their potential biotechnological applications, this feat of engineering is relatively poorly understood in barnacles compared to organisms with catechol-containing adhesives such as mussels and tubeworms [5,6].

Barnacles are known for their adhesive abilities and are the only sessile members of the Crustacea who commit to ‘settling’ on one spot as larvae, which will become their permanent home. They use ‘cement’, comprised mainly of proteins, to permanently bind their base to an underwater substrate. This multiprotein complex is synthesized in a unicellular cement gland, secreted as an insoluble protein complex and cures into cement within hours [7–9]. This underwater adhesion involves surface functions such as displacement of the water layer and coupling of the adhesive to the surface, as well as bulk functions such as curing to provide stiffness and protection from microbial degradation [10]. Six barnacle cement proteins (cp) have been characterized to date: cp16k [10], cp19k [11], cp20k [12,13], cp52k [14], cp68k [10] and cp100k [15]. The insoluble proteins cp100k and cp52k may provide bulk properties in the barnacle cement whereas cp19k, cp20k and cp68k have been proposed to be ‘stickier’ and provide surface functions [9]. Cp16k is a minor constituent of the barnacle cement and has been suggested to be a lysozyme-like enzyme which is involved in surface preparation [9], while a recently described 43 kDa protein undergoes possible oxidase activity that may be moult-related [16]. The cp19k and cp68k proteins are characterized by heavily biased amino acid compositions [14], with serine, threonine, glycine, alanine, valine and lysine contributing up to 70% of their residues [9,11], whereas the calcite-specific cp20k contains a high abundance of cysteine (Cys) residues and charged amino acids [9,13]. While the hydrophobicity of cp100k and cp52k proteins suggests the involvement of hydrophobic interactions in their adhesion [9], no generic covalent bonding mechanism has been identified in barnacles to date and no clear functional motifs which are present in other marine adhesive proteins have been identified [14,17].

Unlike mussel adhesive proteins which are characterized by the presence of 3,4-dihydroxyphenyl-l-alanine (DOPA) [3,18], no post-translational modifications have been identified to date in barnacle adhesive proteins except for glycosylation in the 52k protein [14,15,19]. This offers the possibility of using recombinant expression platforms to accurately reproduce small quantities of proteins found in the bulk cement in order to understand structure–function relationships for each individual protein. There has been some progress in this area for the smaller barnacle cement proteins; for example, cp19k [11,20,21] and cp20k [13] have been successfully expressed in Escherichia coli. However, research to date has focused almost exclusively on acorn barnacle species with calcareous bases. Species with membranous bases and larger, stalked barnacles have been relatively under-studied even though balanomorph and stalked barnacles diverged 200–250 million years ago [22] and probably possess different properties. By understanding the taxonomic breath of barnacles with different substratum preferences, e.g. opportunistic preferences that include man-made substrates versus obligate attachment to rocks or conspecifics, it is hoped to uncover commonalities in adhesives produced by these specialized organisms.

Pollicipes pollicipes is a large stalked barnacle with a membranous base, which cements to intertidal rocks on extremely wave-exposed coasts (figure 1). The adhesive passes to the rocky substratum in liquid form through a pair of ‘principal canals’ [7] and hardens on exposure to the substratum (figure 1). This species grows to larger size than many acorn barnacles (figure 1b) and is somewhat unusual because instead of permanently sticking to one spot, P. pollicipes cyprids settle heavily on conspecifics but the juveniles move down along the peduncle, towards the primary substratum [23]. In fact, adults of this species can also move slightly [24].

Figure 1.

(a) Pollicipes pollicipes attached to rocks. Juvenile P. pollicipes (blue arrows) possess the capability to move along the adult stalk; (b) P. pollicipes in situ on an exposed rocky shore, with acorn barnacles indicated by an arrow; and (c) membranous base (insert, orange arrow) of P. pollicipes and hardened cement (insert, white arrow). (Online version in colour.)

In the present study, the cp19k protein of P. pollicipes (Ppolcp19k) was expressed in E. coli to investigate its adhesive properties and examine its role in the adhesion of this stalked barnacle species. The adhesion of the recombinant protein was examined on hydrophobic, neutral hydrophilic and charged model surfaces using surface plasmon resonance (SPR) to determine whether it displayed significant adhesive properties, or had some other role. The results are discussed in the context of several recent studies in distantly related barnacle species with various lifestyles.

2. Material and methods

(a). Materials

E. coli Top10 (F⁻ mcrA Δ(mrr-hsdRMS⁻mcrBC) Φ80lacZΔM15 ΔlacX74 deoR recA1 araD139 Δ (ara-leu) 7697 galU galK rpsL (Str^R) endA1 nupG); Invitrogen) was used for DNA manipulations and E. coli BL21 (DE3) (F^– ompT gal dcm lon hsdS_B$(r_B^{-}{m}_B^{-})$ λ(DE3 [lacI lacUV5-T7 gene 1 ind1 sam7 nin5]); Invitrogen) for expression of recombinant proteins. Ppolcp19k was expressed using plasmid pIG6 [25] and molecular chaperones were over-expressed from plasmids pG-KJE8, pGro7, pKJE7, pG-Tf2 and pTf16 (Takara Corp; [26]).

(b). Sample collection and cloning of Ppolcp19k gene

Pollicipes pollicipes adult barnacles (n = 3) were sampled on 1 January 2015 in Sines, Western Portugal. Individuals were instantly killed and frozen by immersion in liquid nitrogen. Samples were stored and shipped in RNAlater^® (Invitrogen, CA, USA). Samples were washed in phosphate-buffered saline (10 mM phosphate, 150 mM NaCl, pH 7.4) prior to dissection. Approximately 1 cm³ of the peduncle (stalk) from the region adjacent to the capitulum (shell plate/‘head’ area) was isolated and internal tissue was dissected. To synthesize cDNA, tissue samples were homogenized in 1 ml RLT lysis buffer (RNeasy Mini Kit; Qiagen, Hilden, Germany) with 10 µl β-mercaptoethanol added. Using the IKA^® ULTRA-TURRAX T18 homogenizer, total RNA was extracted with QIAGEN RNeasy Mini Kit including DNase I treatment, according to the manufacturer's instructions. RNA concentration was determined using a SimpliNano (GE Healthcare Limited, UK) and cDNA was synthesized from 1 µl of RNA (580 ng) using AMV reverse transcriptase system and Oligo(dT)15 primer in a 20 µl reaction (reverse transcription system kit; Promega, WI, USA). Negative control reactions, lacking reverse transcriptase enzyme, were prepared to confirm the absence of genomic DNA.

The gene sequence encoding the Ppolcp19k protein was identified from an expressed sequence tag (EST) database with GenBank Accession number FN244142 [27], later verified by sequencing and LC-MS/MS [23,28]. Oligonucleotides were designed to amplify the Ppolcp19k gene with an N-terminal E. coli ompA leader sequence, for secretion of the translated product to the periplasm in expressing cells, and a cleavable C-terminal hexahistidine tag, for detection and affinity purification of the recombinant Ppolcp19k (rPpolcp19k-his) protein. The rPpolcp19k-his product was cloned into the pIG6 expression vector [25] and its sequence confirmed prior to protein expression.

(c). Protein expression and purification

Escherichia coli BL21 (DE3) clones containing the rPpolcp19k-his plasmid, with or without co-transformed chaperone expression plasmids, were grown at 37°C for 8–10 h in 5 ml Luria Bertani (LB) medium containing 100 µg ml⁻¹ ampicillin with 250 rpm shaking. Chloramphenicol was included at 34 µg ml⁻¹ for cells harbouring chaperone expression plasmids. Overnight cultures were used to inoculate, to an OD₆₀₀ of 0.05, 50–200 ml LB broth containing antibiotics and 2–4 mg l⁻¹ of l-arabinose and/or 5 ng ml⁻¹ of tetracycline as required for induction of chaperone expression [24]. Cultures were incubated at 37°C until an OD₆₀₀ of 0.7–0.9 was reached, following which rPpolcp19k-his expression was induced at 25°C using 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG). Cells were harvested by centrifugation after 16 h, followed by preparation of soluble and insoluble protein fractions as described elsewhere [29]. Purification of recombinant rPpolcp19k was carried out using a two-stage immobilized metal affinity chromatography approach, as described in the electronic supplementary material, S1. Purified cp19k protein containing the C-terminal hexahistidine tag is referred to as rPpolcp19k-his; protein from which the hexahistidine tag has been cleaved is referred to as rPpolcp19k.

(d). SDS PAGE and western blot analysis

Protein samples were analysed under denaturing conditions in 12.5% SDS-PAGE. Gels were stained with InstantBlue™ Coomassie stain (Expedeon, UK) or proteins were transferred to an Amersham Protran™ 0.2 µm nitrocellulose blotting membrane (GE Healthcare, Germany) for immunodetection of rPpolcp19k-his. Detection was carried out using a monoclonal anti-polyhistidine peroxidase-conjugated antibody (Sigma Aldrich, Ireland) diluted 1 : 1000 in Tris buffered saline (10 mM Tris, pH 8.0, 150 mM NaCl). Colour was developed using 3,3′,5,5′-Tetramethylbenzidine (TMB).

(e). Evaluation of the adhesion ability of rPpolcp19k-his

The ability of rPpolcp19k-his preparations to adhere to hydrophobic tissue culture polystyrene plates (Corning, NY, USA) and hydrophilic glass slides (Thermo Fisher Scientific, MA, USA) was investigated as previously described [18,20]. Plates were used without pre-treatment while slides were soaked in 5% (v/v) HCl overnight, rinsed with water and dried in air prior to analysis. A 20 µl drop of approximately 500 µg ml⁻¹ protein was added to the surfaces and incubated at room temperature for 6–24 h, followed by washing with water and visualization of adsorbed proteins using InstantBlue™ Coomassie stain (Expedeon, UK). Cell-Tak (Thermo Fisher Scientific, MA, USA), commercially available Mytilus edulis mussel foot protein containing a mixture of foot protein-1 (fp-1) and fp-2 [3], was used as a positive control in adhesion assays, and lysozyme and bovine serum album (BSA) (Thermo Fisher Scientific, MA, USA) as negative controls.

(f). Surface plasmon resonance

Five types of self-assembled monolayers (SAMs) with surfaces exposing homogeneous methyl (CH₃), hydroxyl (OH), carboxyl (COO⁻), trimethylamine (N(CH₃)₃⁺) and oligo(ethylene glycol) (OEG) chemistries were prepared for SPR experiments, as described by Petrone et al. [30]. The preparation of the SAMs is described in the electronic supplementary material, S2 and table S1. SPR experiments were performed using a Biacore 3000 (GE Healthcare, Sweden) with a flow rate of 10 µl min⁻¹ at 25°C. To examine the effect of pH on the adhesion ability of rPpolcp19k, two different running buffers were used: 10 mM sodium acetate buffer (pH 4.0) containing 150 mM NaCl to represent the pH of the barnacle cement gland, and 10 mM sodium phosphate buffer (pH 8.0) containing 600 mM NaCl to represent seawater [21]. After running the relevant buffer through the cell for 5 min, protein samples were injected for 3 min, followed by a 2 min dissociation phase with the same running buffer. The adsorption of rPpolcp19k was compared with trigger factor (TF) as well as fibrinogen and BSA controls at pH 8.0, and Cell-Tak and BSA controls at pH 4.0. Fibrinogen and Cell-Tak were used as positive controls and BSA as a negative control. Adsorption in resonance units (ΔRU) was converted to adsorbed mass (m) via the relation [30]:

$$m=\mathrm{\Delta }\mathrm{RU}\times 6.5\times {10}^{-2}\hspace{0.17em}\mathrm{ng}\hspace{0.17em}\mathrm{c}{\mathrm{m}}^{-2}.$$

3. Results

(a). Cloning of Ppolcp19k

Amplification of the cp19k gene from the P. pollicipes cDNA library yielded a product corresponding to the expected size of 658 bp. DNA sequencing identified seven nucleotide differences from the P. pollicipes cp19k EST [27], of which five were silent mutations and two resulted in H73Q and G85D amino acid substitutions (electronic supplementary material, figure S1). The identified amino acid substitutions were also reported recently in P. pollicipes by Rocha et al. [23] and are probably owing to polymorphisms resulting from sampling different individuals from the P. pollicipes population.

(b). Expression and purification of the rPpolcp19k protein

Initial production of rPpolcp19k-his in E. coli BL21(DE3) under standard expression conditions resulted in low yields of soluble protein. In order to overcome bottlenecks in folding and improve protein solubility, a panel of E. coli molecular chaperones was co-overproduced with rPpolcp19k-his, resulting in increased amounts of total rPpolcp19k-his protein being produced with the GroEL-GroES, DnaK-DnaJ-GrpE, GroEL-GroES-TF and TF chaperone combinations (figure 2a). Analysis of the solubility of rPpolcp19k-his (figure 2b) revealed increased yields of soluble protein in the cases of over-production of GroEL-GroES, GroES-GroEL-TF and TF, with GroEL-GroES and TF chaperones therefore selected for scale-up of expression and purification of rPpolcp19k-his.

Figure 2.

Analysis of rPpolcp19k-his expression in E. coli. Western blot analysis of (a) whole cell extract and (b) soluble protein extract from E. coli cells expressing rPpolcp19k-his alone (lane 1) or with co-expression of GroEL-GroES, DnaK-DnaJ-GrpE (lane 2), GroEL-GroES (lane 3), DnaK-DnaJ-GrpE (lane 4), GroEL-GroES-TF (lane 5) or TF (lane 6). (c) SDS-PAGE analysis of purified rPpolcp19k-his co-overexpressed with TF. Lanes 1–2: proteins eluted using 300 mM imidazole. (d) SDS-PAGE analysis of two-step purification of rPpolcp19k-his co-overexpressed with GroEL-GroES. Lanes 1–2: proteins eluted with 500 mM imidazole. Arrows indicate proteins of the expected size of rPpolcp19k-his. (Online version in colour.)

Expression of rPpolcp19k-his in E. coli BL21(DE3) in the presence of overexpressed TF chaperone, followed by purification, yielded a protein product of approximately 24 kDa (figure 2c), significantly different from the predicted molecular mass of 19 kDa. An additional protein of approximately 48 kDa, corresponding to the size of TF and not detectable by immunoblotting (not shown), co-purified with the target protein. The yield of the TF-rPpolcp19k-his extract was 6 mg l⁻¹ of E. coli culture.

Yields of rPpolcp19k-his were further increased in the presence of co-overproduced GroEL and GroES by titrating the l-arabinose inducer to alter cellular GroEL-GroES levels. Highest amounts of soluble rPpolcp19k-his protein were achieved with 4 mg ml⁻¹ l-arabinose (figure 2d). Purification of rPpolcp19k-his yielded a single band at greater than 95% purity and with the predicted molecular mass of 19 kDa, though with some degradation evident (figure 2d). The yield was 1–2 mg of purified protein l⁻¹ of E. coli culture.

(c). Surface coat analysis of rPpolcp19k-his

Surface coat analysis was carried out to assess the adsorption of rPpolcp19k-his on hydrophilic (glass slide) and hydrophobic (polystyrene tissue culture plate) surfaces. The TF-rPpolcp19k-his complex demonstrated high adsorption on both the hydrophilic (figure 3a) and hydrophobic (figure 3b) surfaces compared to BSA and lysozyme negative controls. To determine the basis for adsorption of the TF-rPpolcp19k-his complex, purified E. coli TF was analysed and found to remain adsorbed to both surfaces (figure 3c,d) while rPpolcp19k-his (i.e. without TF) and BSA washed away, even after extending incubation from 6 to 24 h. The Cell-Tak positive control demonstrated high adsorption on both hydrophilic (figure 3b) and hydrophobic surfaces (figure 3d).

Figure 3.

Surface coat analysis of protein binding to (a,c) hydrophilic glass slide or (b,d) hydrophobic polystyrene plate. Lyso: lysozyme; BSA: bovine serum albumin; Cell-Tak: commercial Mytilus edulis mussel foot protein; TF, trigger factor. (Online version in colour.)

(d). Surface plasmon resonance

Analysis of surface adsorption of rPpolcp19k, from which the recombinant hexahistidine tag had been removed, under simulated seawater conditions (high ionic strength, pH 8.0) by SPR revealed that the barnacle protein exhibited minimal adsorption on all surface chemistries investigated (figure 4a; electronic supplementary material, figure S2). Conversely, E. coli TF adsorbed relatively highly to all surfaces with the exception of the very fouling-resistant OEG surface, to which no proteins were found to adsorb. Meanwhile, the fibrinogen positive control adsorbed highly to all other surfaces and the BSA negative control exhibited levels of surface adsorption very similar to rPpolcp19k.

Figure 4.

SPR analysis of protein adsorption on self-assembled monolayers (SAMs) under (a) seawater and (b) gland cell conditions. Data are in absolute changes in resonance units (ΔRU) after a 3 min injection of (a) 0.5 mg ml⁻¹ protein in 600 mM NaCl, pH 8.0, or (b) 0.4 mg ml⁻¹ protein in 150 mM NaCl, pH 4.0, at 10 µl min⁻¹ flow rate and 25°C, followed by 2 min wash with relevant buffer. Standard errors are from two replicate SPR experiments. (Online version in colour.)

Under conditions designed to mimic the environment of the barnacle gland cell (low ionic strength, pH 4.0), rPpolcp19k again exhibited the lowest adsorption, across all SAMs, of the three proteins investigated (figure 4b; electronic supplementary material, figure S3). Cell-Tak exhibited the highest adsorption on $\mathrm{N}{(\mathrm{C}{\mathrm{H}}_3)}_3^{\hspace{0.17em}\hspace{0.17em}+}$ and CH₃ surfaces, and BSA on the COO⁻ SAM, with the fouling-resistant OEG SAM again exhibiting the lowest values for all proteins.

4. Discussion

Barnacles with a membranous base, in particular P. pollicipes, are capable of detachment, post-settlement locomotion and reattachment [24]. Considering its ability to relocate, different taxonomic order from previous studies and soft membranous basis rather than hard calcite, the putative adhesive protein cp19k from P. pollicipes exhibits similar amino acid bias and isoelectric point (pI) as its counterparts in other species (electronic supplementary material, table S2). The present study was designed to investigate the adhesive properties of Ppolcp19k in order to delineate its role in barnacle adhesion, particularly compared to a series of recent studies on recombinant cp19k from two acorn barnacle species.

While E. coli is a popular and robust choice for the expression of recombinant proteins, many heterologous proteins are initially expressed in predominantly insoluble form and require significant troubleshooting in order to achieve soluble, functional expression in the bacterium [31,32]. One bottleneck in high-level production of heterologous proteins is saturation of the native molecular chaperone machinery of the host cell [33,34]. Co-overproduction in the expressing cells of Hsp60 chaperone family members GroEL and GroES, and TF led to large increases in soluble rPpolcp19k-his yields in this work. This chaperone over-production approach has been widely used to improve expression of recombinant targets in E. coli [26,35] though it is complicated somewhat by the difficulty in predicting in advance chaperone–target protein interactions [36], as evidenced by the absence of a positive effect of Hsp70 family chaperones DnaK, DnaJ and GrpE in this study.

Purification of rPpolcp19k-his from scaled-up expression in cells co-producing TF yielded 6 mg of protein l⁻¹ of E. coli culture, albeit this included a co-purifying, non-his-tagged protein, corresponding to the size of TF, that was not removed by repeated purifications. Additionally, the apparent molecular mass of the rPpolcp19k-his product on SDS-PAGE (24 kDa) differed significantly from its predicted molecular mass (19 kDa), a phenomenon which has been reported in TF co-expressions with other proteins [35,37]. This is possibly attributable to the high levels of E. coli TF in the cytoplasm leading to a bottleneck in translocation across the cytoplasm membrane, resulting in failure to cleave the periplasm-targeting N-terminal leader peptide of the target protein [35]. Meanwhile, expression of rPpolcp19k-his in the presence of overproduced GroEL-GroES yielded a single purified product of the expected molecular mass at a yield of 1–2 mg of protein l⁻¹ of E. coli culture.

Initial surface coat analysis revealed high levels of adsorption of the TF-rPpolcp19k-his complex on both hydrophobic and hydrophilic surfaces. Separation of the two proteins for further investigation, however, identified that the purified E. coli TF demonstrated adsorption on a similar level to the commercial Cell-Tak mussel foot protein, whereas no detectable adhesion to either surface was observed for the rPpolcp19k-his. As TF exposes hydrophilic and hydrophobic amino acid residues in its role of interacting with a diverse range of nascent polypeptides as they emerge from the ribosome exit tunnel in vivo [38,39], it is unsurprising that it is capable of adsorbing to both surface types. The absence of any detectable adsorption by rPpolcp19k-his was surprising, given previous reports of adhesive properties of cp19k proteins from Megabalanus rosa [11] and Balanus albicostatus [20]. In the latter study, however, the adhesion represented the adhesion ability of the thioredoxin-tagged fusion protein and not the native barnacle protein.

To investigate adhesion of rPpolcp19k at a microscale, SPR was used to evaluate its binding on hydrophobic, neutral hydrophilic, negatively and positively charged surfaces under conditions imitating either the gland where the proteins are produced or the secretory surface (seawater). A low pH range has been reported in the adhesive gland of adult Lepas anatifera and P. pollicipes stalked barnacles [7,40] as well as organisms such as mussels and sandcastle worms [41]. However, SPR results revealed poor adsorption of rPpolcp19k on all surfaces at pH 4.0, comparable with the BSA negative control. Under the seawater-like condition, rPpolcp19k again demonstrated the lowest adsorption across all SAMs.

Interestingly, purified E. coli TF exhibited high adsorption across multiple SAM chemistries at pH 8.0, which may be an important consideration when assessing adhesion of recombinantly produced proteins to which low levels of the chaperone may remain bound after purification. In a previous demonstration of the very strong adhesive properties of a recombinant B. albicostatus cp19k fusion protein, unidentified high molecular weight, co-purified E. coli proteins were evident in cp19k preparations [20]. The additional presence of multiple recombinant moieties (thioredoxin fusion, hexahistidine and S-tags, and thrombin and enterokinase cleavage sites) in that fusion protein makes it difficult to determine the precise basis for the adhesion in that case. TF exceeded even the adsorption of the positive control fibrinogen [30] on the negatively charged COO⁻ SAM at pH 8.0 in the present study, despite its lower pI (4.8 versus 5.3–6.1 for fibrinogen). The increased ionic strength of seawater-like conditions can cause protein aggregation and increase surface adsorption [42], though TF's unique role in binding diverse polypeptides as they emerge from the ribosome [38] may enable it to interact with the studied surface layers under a broader range of conditions than would be feasible for most proteins. Cell-Tak, meanwhile, demonstrated activity typical of a marine adhesive protein, with high levels of adsorption on both $\mathrm{N}{(\mathrm{C}{\mathrm{H}}_3)}_3^{\hspace{0.17em}\hspace{0.17em}+}$ and CH₃ surfaces evident at pH 4.0. The use of Cell-Tak as a positive control was limited to pH 4.0, as it could not be investigated at pH 8.0 owing to its susceptibility to oxidation [43].

While rPpolcp19k exhibited negligible adsorption on any surface investigated in this work, Urushida and co-workers [11] previously reported that recombinant cp19k from M. rosa adsorbed irreversibly to polycrystalline gold and hydrophobic alkylated gold in SPR experiments. It is worth noting that the amount of adsorbed M. rosa cp19k protein was 49 ng cm⁻² on their [11] alkylated gold surface while rPpolcp19k adsorbed at similar levels, ranging from negligible (OEG surface) up to 39 (electronic supplementary material, table S3) and 45 (electronic supplementary material, table S4) ng cm⁻² on CH₃ SAMs at pH 4 and 8, respectively, in the present study. Much higher levels of adsorption were shown in larval temporary adhesive: Petrone et al. [30] demonstrated the cyprid settlement-inducing protein complex (SIPC) showed similar adsorption to fibrinogen, which supports a specialized role in adhesion for this glycoprotein. The question therefore is: if cp19k from P. pollicipes is not specialized for adhesion relative to proteins apparently designed for this purpose (such as fibrinogen or SIPC), what is its role in barnacle adhesion?

This question is all the more intriguing considering that a six amino acid bias reported in several species of acorn barnacles [11] also accounts for 70.5% of residues in the P. pollicipes protein (electronic supplementary material, table S2). While overall sequence similarity across barnacle species is modest, in the range of approximately 35% maximal similarity [17], the universally strong bias towards these amino acids (Ser, Thr, Ala, Gly, Val and Lys) may confer particular abilities to the cp19k protein even when the precise sequence differs. In fact, the only major difference between species appears to be the unusually low pI in the M. rosa protein (pI = 5.8), when homologues from P. pollicipes and all other species possess pI values in the range 9.26–9.80. It is possible that hydroxyl groups on Ser and Thr residues may be important in displacing water before surface coupling [9,11] and Lys also functions in this manner when found in combination with catechols [44]. Of more significance may be the fact that cp19k is low in amino acid complexity and particularly rich in Gly across all species, with glycine making up 20.5% of the amino acid composition of Ppolcp19k (electronic supplementary material, table S2). This low amino acid complexity has been associated with nanofibre formation and Gly-rich regions, in particular, are linked to amyloid formation [45].

One of the first suggestions that functional amyloids could be a generic mechanism within bioadhesives was by Mostaert et al. [46], who suggested the sawtooth-like pattern formed from the unfolding of cross β-sheets confers strength to adhesive materials. Since then, proteins with this type of secondary structure have been shown to play a role in bioadhesion in several organisms [47,48]. The E. coli curlin protein forms fibrils to colonize inert surfaces, form biofilm and bind to host proteins [47] while other examples of functional amyloids include chorion proteins in the eggshell of silkworms [49] and the major silk protein spidroin in the silk fibres of spider webs [50]. Bulk barnacle cement stained positively for amyloid and this component was suggested to provide strength and stability to the cement in seawater [15,48]. Individual cement proteins with cross β-sheet structure such as M. rosa cp100k and cp52k have been suggested to be responsible for self-assembly and curing of the barnacle adhesive [15,51]. Similar reports for P. pollicipes cp52k protein [23] suggest that formation of barnacle cement could be similar to that of amyloid plaques. However, the same mechanism may be at work in smaller ‘adhesive’ proteins including Ppolcp19k, the predicted secondary structure of which includes β-sheets that encompass 26.4% of its amino acid residues [23], i.e. higher than other proteins in the bulk cement complex. Meanwhile, pH and/or ionic strength may serve as triggers for amyloid self-assembly [51]. Barnacle cement is secreted as a low viscosity liquid, which solidifies to become a rubbery substance and eventually hardens to a cement by an unknown ‘switch’ mechanism [7,8]. Triggers of amyloid formation are an area of active research, particularly in pathological cases, but also in functional amyloid. Recent advances in understanding possible triggers for amyloid formation in recombinant cp19k and cp20k barnacle proteins suggest that these may function under both ionic and pH control of self-assembly, in conferring strength to the adhesive and resistance to degradation of the protein plaque [2,21,52]. Meanwhile, a change in the salt concentration and pH may be required by the organism to prevent premature self-aggregation in vivo. Interestingly, amyloids have recently been shown to offer possibilities of exciting new biomaterials, e.g. CsgA from E. coli curli amyloid fibres combined with mussel foot protein has been used to engineer a bioinspired hybrid fibril-adhesive [53].

5. Conclusion

There was no evidence that recombinant cp19k from P. pollicipes (rPpolcp19k) was adapted for adhesion in our study, contrary to previous reports, upon testing on a variety of surfaces and in conditions mimicking the gland cell or the seawater environment. The amino acid bias evident in this protein is also present in a range of other distantly-related barnacle species and is typical of a protein which is pre-disposed to form amyloid (e.g. characterized by low complexity domains and enrichment with Gly residues). Therefore, we speculate that barnacle cp19k may be specialized for self-assembly into amyloid plaques under the appropriate environmental triggers (including salt concentration and pH), and that this may strengthen the barnacle cement as well as conferring resistance to degradation.

Data accessibility

Data are presented as the electronic supplementary material.

Authors' contributions

M.A.T. carried out the recombinant expression, Surface Plasmon Resonance experiments, data analysis, participated in the design of the study and drafted the manuscript; M.D. and S.M. carried out preliminary molecular laboratory work, sequence alignments and critically revised the manuscript; T.E. conceived and collected surface plasmon resonance data and critically revised the manuscript; J.G.W. and A.M.P. conceived of the study, designed the study, coordinated the study and helped draft the manuscript. All authors gave final approval for publication and agree to be held accountable for the work performed therein.

Competing interest

We declare we have no competing interests.

Funding

This work was supported by Science Foundation Ireland (SFI) co-funded under the European Regional Development Fund under grant no. 13/RC/2073. The work was also supported by SFI grant no. 09RFPMTR2311, the Irish Marine Institute (Beaufort Marine Research Awards) and the European Network of Bioadhesion Expertise (COST Action CA15216).