Adhesion of mussel foot proteins to different substrate surfaces

Abstract

Mussel foot proteins (mfps) have been investigated as a source of inspiration for the design of underwater coatings and adhesives. Recent analysis of various mfps by a surface forces apparatus (SFA) revealed that mfp-1 functions as a coating, whereas mfp-3 and mfp-5 resemble adhesive primers on mica surfaces. To further refine and elaborate the surface properties of mfps, the force–distance profiles of the interactions between thin mfp (i.e. mfp-1, mfp-3 or mfp-5) films and four different surface chemistries, namely mica, silicon dioxide, polymethylmethacrylate and polystyrene, were measured by an SFA. The results indicate that the adhesion was exquisitely dependent on the mfp tested, the substrate surface chemistry and the contact time. Such studies are essential for understanding the adhesive versatility of mfps and related/similar adhesion proteins, and for translating this versatility into a new generation of coatings and (including in vivo) adhesive materials.

1. Introduction

Mussels survive in turbulent ocean environments by robust attachment to wave- and wind-swept substrata using a proteinaceous holdfast or byssus. Each byssus consists of specially assembled fibre, adhesive and coating proteins that cooperatively provide tenacity despite cyclic mechanical, chemical and biological stresses that arise from changes in the salinity, temperature, exposure and immersion, flow and microbial density in seawater. There are interesting parallels between the capacity of the byssus to maintain its integrity in the sea and the desirable properties for a biomaterial, such as a dental or tissue adhesive, introduced into a variety of physiological environments in the human body and submitted to flowing body fluids, degradation by endogenous enzymes and immunogenic attacks. Therefore, understanding the adhesion mechanism of mussel adhesive and coating proteins is surprisingly relevant to the design and development of biomedical adhesive and coating materials [1–6]. Mussels are also able to form underwater bonds to various substrates, such as glass, plastic and metal oxides. Indeed, the mechanism of water-resistant adhesion and coating to various substrates has attracted significant interests for potential use in biomedical applications. Understanding the interaction between individual mussel foot proteins (mfps) and different substrates will provide useful insights for the design of novel medical adhesives and coating materials.

Mfps from the Mytilus genus, comprising more than eight different proteins, are secreted from the mussel foot and responsible for adhesion and coating of the mussel byssus. All these proteins contain the post-translationally modified amino acid 3,4-dihydroxyphenyl-l-alanine (DOPA), derived from hydroxylation of tyrosine residue, and they are positively charged polyelectrolytes with high isoelectric points (pI ∼ 10). It was reported that DOPA enables mfps to interact with various substrates under water, and that oxidative conversion of DOPA to o-quinone significantly reduces adhesion of mfps [7–10] except stainless steel [11]. Mfp-1 has been speculated as a coating based on its outermost distribution in the byssus, whereas mfp-3 and mfp-5 were speculated to be adhesive primers owing to their distribution at interfaces between adhesive plaque and substratum [5,8,12].

Mfp-1 is composed of 64 tandem repeats of a decapeptide [Pro-Lys-Ile-Ser-DOPA-diHyp-Hyp-Thr-DOPA-Lys] (the red-rod schematic in figure 1 stands for the decapeptide), in which HyP, diHyP and DOPA denote trans-4-hydroxyproline, trans-2, 3, cis-3, 4-dihydroxyproline and DOPA, respectively [13]. Mfp-3 are interfacial adhesive proteins composed of more than 35 variants. Protein masses for the entire mfp-3 family range between 5 and 7.5 kDa. All variants contain 4-hydroxyarginines (HyArg) and approximately 20 mol% of DOPA as post-translationally modified amino acids [14]. Mfp-5 is also an interfacial adhesive protein containing o-phospho-serine as a unique post-translationally modified amino acid. Mfp-5 has the highest DOPA (approx. 25.5 mol%) among all plaque proteins, i.e. with around one in every four residues [14]. The detailed comparison of the three kinds of proteins used in the study is shown in table 1 [13–15].

Figure 1.

A schematic of the chemical structures of three mfps and four substrate surfaces, highlighting likely interactions between the mfps (mfp-3 as an example) and the different surface types. The red rod in mfp-1 stands for the decapeptide.

Table 1.

Comparison of the protein properties.

			mfp-1	mfp-3	mfp-5
mass (kDa)			92.0	5.3	9.5
pI			10.3	10.1	8.3
post-translational modifications	hydroxylation (Hy)	DOPA	13.0	19.0	25.5
		diHyP	5.9	0	0
		HyP	18.2	0	0
		HyArg	0	1.0	0.5
		total	37.1	20.0	26.0
	Phosphorylation		0	0	9.4
basic amino acids	Lys		19.4	15.0	19.5
	Arg, HyArg		1.0	9.5	3.1
	His		0.4	1.0	6.5
	total		20.8	25.5	29.1
amino acids favouring protein flexibility [16,17]	Gly, Thr, Arg, Ser, Gln, Asn, Asp, Pro, Glu, Lys (mol%)		50.4	70.0	61.1
hydrophobicity	aromatic (mol%, Phe, Tyr, Trp, DOPA)		19.2	26.5	26.2
	aliphatic (mol%, Val, Leu, Ile, Pro, Met)		9.4	9.0	2.9
	total		28.6	35.5	29.1

The interfacial properties of mfps were here investigated by a surface forces apparatus (SFA) [4,9,10,12,18–23], which has been widely used to measure the intermolecular and surface forces in various biological and non-biological systems with nanonewton force sensitivity and less than 0.1 nm distance resolution [4,24–29]. To date, the comparison of the molecular interaction forces between the different mfps (mfp-1, mfp-3, mfp-5, especially mfp-5) and various surface chemistries have not been systematically studied. In the present work, the interaction force–distance profiles were directly measured between mfp-1, mfp-3 or mfp-5 and various substrates—mica, silicon dioxide (SiO₂), polymethylmethacrylate (PMMA), polystyrene (PS)—using an SFA in aqueous solutions to better understand the underwater adhesion and coating mechanisms of mfps.

2. Material and experimental methods

2.1. Protein purification from mussel feet

Mussels (Mytilus californianus) were collected for the purification of mfp-1 and mfp-3 from Goleta Pier in Goleta, CA, USA. The feet of mussel were carefully dissected, depigmented by scraping with a razor blade and stored at −80°C before use. Blue mussel (Mytilus edulis L.) feet for the purification of mfp-5 were obtained in flash-frozen 500 g lots from the North East Transport of Union, Maine. Mfp-1, mfp-3 and mfp-5 were purified from frozen mussel feet according to published procedures [15,18,30]. Sample purity was assessed by acid urea poly-acrylamide gel electrophoresis, amino acid analysis and MALDI-TOF mass spectrometry. The mole % DOPA in purified mfp-1, mfp-3 and mfp-5 was approximately 13, approximately 23 and 28 mol%, respectively, determined by amino acid analysis after a 1 h hydrolysis in 6 N HCl at 158°C. Purified samples were freeze-dried and resuspended in 50 mM acetic acid (0.1 mg ml⁻¹) and thereafter divided into convenient aliquot volumes for storage in vials at −80°C prior to testing. Low pH and protection from light are necessary to reduce DOPA losses during handling and storage [4,18]. All aliquots were used within one month or otherwise discarded because protein adhesive quality falls off abruptly during long-term storage [10]. Milli-Q water (Millipore, Mississauga, ON, Canada) was used for all glassware cleaning and solution preparation.

2.2. Chemicals

The buffer for SFA measurements consisted of 0.1 M acetic acid (HAc) (Fisher Scientific, Ottawa, ON, Canada), and sodium acetate (NaAc) (Merck& Co. Limited, Montreal, QC, Canada), and 0.25 M potassium nitrate (KNO₃) (MP Biomedicals, Solon, OH, USA) at pH 5.5. Aqueous solutions were prepared in Milli-Q water (Millipore) and filtered through 0.2 µm filters (Nalgene, Rochester, NY, USA). PS (MW 10⁶, M_W/M_n 1.10) was obtained from Polysciences (Warrington, PA, USA). PMMA (MW 35 kDa) was purchased from Scientific Polymer Products, Inc (Ontario, NY, USA). PS and PMMA were dissolved in toluene (Fisher Scientific) and the solutions were filtered through 0.2 µm PTFE filters (Fisher Scientific) before use.

2.3. Preparation and characterization of substrate surfaces

Four different substrate surfaces were prepared for the SFA measurements: mica, mica-supported SiO₂, mica-supported PMMA and mica-supported PS (figure 1). On the basis of a previously reported procedure [18,26,29], two thin and back-silvered mica sheets (1–5 µm thick) were glued separately onto cylindrical silica discs (radius R = 2 cm) and the exposed mica surfaces were directly used or further coated with different chemicals (i.e. SiO₂, PMMA and PS) as follows. Thin layers of SiO₂ (approx. 15 nm) were deposited onto mica by E-beam evaporation (PVD-75, Kurt J. Lesker) at approximately 0.05 nm s⁻¹ with 1.5 × 10⁻⁵ Torr of O₂ and (2–8) × 10⁻⁶ Torr of H₂O. Thin layers of PS or PMMA were coated onto mica by spin-coating using 0.5 wt% PS or PMMA solution in toluene and vacuum dried at 23°C overnight.

The surface roughness of the different substrates was characterized by atomic force microscopy (AFM, Agilent Technologies 5500, Santa Barbara, CA, USA). The surfaces were imaged with a silicon tip (AppNANO, ACT-200, Si, N-type, tip radius < 10 nm, resonant frequency 318 kHz) operating in the tapping mode in air. Water contact angle measurements were performed using a contact angle goniometer (KRUSS DSA 10, Germany).

2.4. SFA force measurements

The force measurements between proteins and substrate surfaces were performed using an SFA (Surforce LLC, Santa Barbara, CA, USA) in a configuration reported previously [4,10,12,18,26,28]. A protein film was adsorbed onto each type of substrate surface according to a recent procedure [4]. Briefly, for each SFA measurement, 50 µl of the protein solution (10 µg ml⁻¹ in 0.1 M NaAc, 0.25 M KNO₃, pH 5.5) were placed onto the substrate and incubated for 10 min in a chamber saturated with water vapour. Then the surface was rinsed with pure buffer and mounted in the SFA chamber together with another bare substrate surface in a cross-cylinder configuration. The interaction forces F between the two surfaces in pH 5.5 buffer were measured as a function of the absolute surface separation distance D as determined using multiple beam interferometry.

During a typical SFA force measurement, the two surfaces (e.g. an mfp film and a substrate surface) were first brought towards each other to reach a ‘hard wall contact’ and kept in contact for a certain time, followed by separation. The ‘hard wall’ distance is defined as the confined distance between the two surfaces, which did not appear to significantly change on increasing the normal (compressive) load or pressure. If two surfaces attract one another, an adhesion force F_ad is measured during separation, and the surfaces jump apart from adhesive contact when the tensile load exceeds F_ad. The adhesion energy per unit area W_ad is related to the measured adhesion force (F_ad/R) by W_{ad =} F_ad/1.5πR for soft deformable surfaces [20,31]. All the experiments were carried out at room temperature (23°C).

2.5. Hydropathy and flexibility analysis

To interpret SFA data, analysis of the hydropathy and chain flexibility of the mfps were performed (see the electronic supplementary material for detailed analysis methodology for hydropathy and flexibility of mfps). For the hydropathy analysis, a Hopp and Woods hydropathy analysis was used (http://web.expasy.org/protscale/) [32], which is based on Tanford and Nozaki's hydropathy measurements with DOPA [33], reflecting a high degree of post-translational modification on mfps and overall random coil conformation of mfps. Most mfps have a random conformation in aqueous solution while mfp-1 has poly-proline II domains separated by unstructured sequences [34,35]. The B-factor, also known as the atomic displacement or temperature factor determined from X-ray crystallographic studies, reflects the degree of thermal motion and static disorder of an atom in a protein crystal structure, and has been applied for predicting protein chain flexibility. Amino acids can have two types of groups, ‘rigid’ and ‘flexible’, on the basis of the B-factor that reflects the chain flexibility of 31 proteins of known three-dimensional structure in the Protein Data Bank (Brookhaven, USA) [16,17]. The portions of flexible amino acids in each mfp were calculated here to predict protein chain flexibility.

3. Results

3.1. Properties of mfps and substrate surfaces

The three kinds of mfp proteins used (mfp-1, mfp-3, mfp-5) all have high isoelectric points (pI) and exhibit a high degree of post-translational modification particularly in the hydroxylation of tyrosine to DOPA. On the other hand, they differ in molecular weights, pI values, post-translational modifications type and ratios, basic and aromatic amino acid content, and flexibility (table 1) [13–15]. Mfp-3 is a small protein (5.3 kDa) with the highest flexible amino acid residues among mfps (70 mol%) determined by B-factor that reflects the degree of flexibility of amino acids [16,35]. Mfp-3 contains 25.5 mol% basic residues and a pI of ∼10.1, which has 35.5 mol% hydrophobic amino acid, including approximately 20 mol% DOPA [14]. Similar to mfp-3, mfp-5 is a small protein (9.5 kDa) with 61.1 mol% of flexible residues based on B-factor. Mfp-5 contains 29.1 mol% basic amino acids and has a pI ∼ 8.3. Mfp-5 also contains negative charges (9.4 mol% phosphorylation, 2.8 mol% Glu), and has 29.1 mol% hydrophobic residues, including approximately 25.5 mol% DOPA [15]. Compared with mfp-3 and mfp-5, mfp-1 is a large protein (92.0 kDa) with much less flexibility (50.4 mol% of flexible residues). Mfp-1 contains 20.8 mol% basic amino acid and has a pI of 8.3–10.3, which has 28.6 mol% hydrophobic amino acid, including approximately 19.2 mol% DOPA. Mcfp-1 is the most rigid protein among all the mfps tested based on the B-factor and is composed of tandem repeats of stiff decapeptide units [13,36].

The four different substrate surfaces studied have wide-ranging surface chemistries (e.g. chemical compositions, structures, etc., summarized in figure 1). The hydrophobicity determined by water contact angle measurements increases in the order of mica < mica-supported SiO₂ < mica-supported PMMA < mica-supported PS, with water contact angles of less than 5°, 20°, 70° and 92°, respectively (consistent with values reported previously [28,37,38]). The root mean square (r.m.s.) roughness determined by AFM for mica, mica-supported PMMA and PS were all less than 0.5 nm, while for mica-supported SiO₂ the r.m.s. roughness was about 1.0 nm (consistent with previous reports [28,39]).

3.2. Adhesion of mfps to mica, SiO₂, PMMA and PS

To understand the adhesive interaction mechanisms of mfps, the surface forces measurements were performed in an ‘asymmetrical’ configuration between three different mfps and four different opposing substrate surfaces. The adhesion results of the three mfps (mfp-1, mfp-3, mfp-5) to the four surfaces (mica, SiO₂, PMMA and PS) are presented in the following sections. The typical force–distance profiles are shown in figures 2–4, and the comparison of the interaction energies from figures 2–4 are summarized in figure 5.

Figure 2.

Mfp-1 adhesion to different substrates: (a) mica, (b) SiO₂, (c) PMMA and (d) PS with different contact times of 2 min (blue), 10 min (red) and 60 min (purple) after bringing the two surfaces into contact in buffer consisting of 0.1 M sodium acetate, 0.25 M KNO₃, pH5.5. The normalized forces, F/R, are denoted in the left ordinate, whereas the corresponding interaction energies per unit area, W (defined by W = F/1.5πR), are on the right-hand ordinate. F_ad/R is the normalized adhesion force.

Figure 4.

Mfp-5 adhesion to different substrates: (a) mica, (b) SiO₂, (c) PMMA and (d) PS with different contact times of 2 min (blue), 10 min (red) and 60 min (purple) after bringing the two surfaces into contact, in buffer consisting of 0.1 M sodium acetate, 0.25 M KNO₃, pH5.5. The normalized force, F/R, is denoted in the left-hand ordinate, whereas the corresponding interaction energy per unit area, W (defined by W = F/1.5πR), is indicated by the right-hand ordinate. F_ad/R is the normalized adhesion force.

Figure 5.

Relationship between adhesion energy (W_ad = F_ad/1.5πR) and contact time for (a) mfp-1, (b) mfp-3 and (c) mfp-5 on four different substrates: (black circles) mica, (green squares) SiO₂, (red diamonds) PMMA and (blue triangles) PS. Each point and error bar represents the mean of three force runs and its standard deviation.

The solution condition (pH 5.5 and salt concentration of 0.35 M) for the SFA measurements was chosen based on the following considerations. Seawater chemistry appears to be irrelevant to the initial deposition of adhesive proteins. Like a rubber plunger, the mussel foot positions itself snugly onto a patch of surface and imposes a new set of solution conditions in the sealed space between itself and the substratum. These conditions include an acidic pH (approx. pH 5) and low salt concentration 0.1 M [10], which are crucial adaptations, given that most mfps undergo spontaneous oxidation and are insoluble at the pH and ionic strength of seawater [10,12]. Our experiments using SFA were designed to be consistent with the known details of mussel adhesion, i.e. at a chosen pH of 5.5 with a buffer salt concentration of 0.35 M. Potassium nitrate was used in place of sodium chloride in the buffer solution to reduce the possible corrosion of the semi-reflecting silver layers under the mica substrates induced by the high concentration of chloride ions in the surface forces measurements.

3.2.1. Adhesions of mfp-1 to mica, SiO₂, PMMA and PS

Mfp-1 demonstrated weak adhesion to both hydrophobic and hydrophilic surfaces, but the adhesion clearly depends on the surface type and contact time (figure 2a–d).

As shown in figure 2a, the adhesion strengths of mfp-1 to mica were F_ad/R ∼ −2.3, −2.9, −3.7 mN m⁻¹ (or W_ad ∼ 0.5, 0.6 and 0.8 mJ m⁻²) for 2, 10 and 60 min contact times, respectively. These results are consistent with a recent report [18]. The adhesion of mfp-1 on SiO₂ is relatively weak, with adhesion forces of F_ad/R ∼ −0.5, −0.7 and −0.4 mN m⁻¹ (W_ad ∼ 0.10, 0.13 and 0.07 mJ m⁻²) for 2, 10 and 60 min contacts, respectively (figure 2b). The adhesion strengths of mfp-1 to PMMA were F_ad/R ∼ −0.05, −0.5, −0.6 mN m⁻¹ (W_ad ∼ 0.01, 0.11, 0.13 mJ m⁻²) for 2, 10 and 60 min contacts, respectively (figure 2c). The adhesion of mfp-1 on PS were F_ad/R ∼ −0.2, −0.6, −1.6 mN m⁻¹ (W_ad ∼ 0.06, 0.15, 0.33 mJ m⁻²) for 2, 10 and 60 min contacts, respectively (figure 2d). Overall, Mfp-1 can be readily deposited onto the four substrates using current deposition technique, and form protein layers of reproducible film thickness as determined by SFA (10 ± 0.5 nm), and the adhesion of mfp-1 with the four substrates increases in the order: mica > PS, SiO₂, PMMA (figures 2 and 5a).

3.2.2. Adhesion of mfp-3 to mica, SiO₂, PMMA and PS

Mfp-3 can adhere well to both hydrophilic and hydrophobic surfaces, and the adhesion strengths are again substrate-dependent, as shown by the typical interaction force–distance profiles (figure 3a–d for mica, SiO₂, PMMA and PS, respectively). Typically, the adhesion increased with increasing contact time for almost all the cases studied for mfp-3, most probably owing to the local conformational rearrangements of the protein molecules resulting in more effective adhesive ‘bonds’ to the substrate surfaces. For a more flexible protein, such conformational rearrangements can be achieved relatively easily with contact time for better adhesion. Whereas for a rigid protein, increasing contact time may not induce sufficient conformational rearrangements, and its adhesion shows relatively weak dependence on contact time, i.e. the entropic hindrance from neighbouring rigid side groups can impede the conformational arrangements of the functional groups for efficient adhesive interactions. Compared with mfp-1, mfp-3 is a smaller and more flexible protein, which makes its surface adhesion more dependent on the contact time.

Figure 3.

Mfp-3 adhesion to different substrates: (a) mica, (b) SiO₂, (c) PMMA and (d) PS with different contact times of 2 min (blue), 10 min (red) and 60 min (purple) after bringing the two surfaces in contact, in buffer consisting of 0.1 M sodium acetate, 0.25 M KNO₃, pH5.5. The normalized forces, F/R, are denoted in the left-hand ordinate, whereas the corresponding interaction energies per unit area, W (defined by W = F/1.5πR), are in the right-hand ordinate. F_ad/R is the normalized adhesion force.

The adhesion strengths of mfp-3 to mica were observed to be F_ad/R ∼ −1.4, −1.5 and −3.0 mN m⁻¹ (W_ad ∼ 0.29, 0.31 and 0.64 mJ m⁻²) for 2, 10 and 60 min contacts, respectively (figure 3a). These results are consistent with previous reports [5,12]. For mfp-3 against SiO₂, the adhesive interaction strengths were determined to be F_ad/R ∼ −1.2, −3.0 and −14.1 mN m⁻¹ (W_ad ∼ 0.26, 0.64 and 2.99 mJ m⁻²) for 2, 10 and 60 min contacts, respectively (figure 3b). The adhesion of mfp-3 to SiO₂ showed the greatest dependence on the contact time. The adhesion strengths of mfp-3 to PMMA were F_ad/R ∼ −0.9, −3.0 and −6.1 mN m⁻¹ (W_ad ∼ 0.2, 0.64 and 1.30 mJ m⁻²) for 2, 10 and 60 min contacts, respectively (figure 3c). For mfp-3 against PS, the adhesion strengths were essentially spontaneously achieved at F_ad/R ∼ −12.1, −13.6 and −12.7 mN m⁻¹ (W_ad ∼ 2.58, 2.88 and 2.69 mJ m⁻²) for 2, 10 and 60 min contacts, respectively (figure 3d). Therefore, mfp-3 protein molecules show the highest adhesion to PS among the four substrates for short contact time, and PS ≈ SiO₂ > PMMA > mica for long contact time (60 min). The thickness (i.e. hard wall distance) of mfp-3 layer on the four substrates was determined to be 5 ± 0.5, 6 ± 0.5, 25 ± 0.5 and 6 ± 0.5 nm for mica, SiO₂, PMMA and PS, respectively (figure 3), which indicates that the adsorption/deposition of mfp-3 to PMMA is greater than to mica, SiO₂ and PS. It should be noted that the proteins concentration and adsorption time during deposition were fixed, and the protein layer thickness was determined by repeated measurements for at least three times.

3.2.3. Adhesions of mfp-5 to mica, SiO₂, PMMA and PS

Similar to mfp-3, mfp-5 exhibits strong adhesion to both hydrophilic and hydrophobic surfaces (figure 4), and the adhesion strength depends both on surface type and contact time. Typical force–distance profiles are shown in figure 4a–d for mica, SiO₂, PMMA and PS. Similar to mfp-3, adhesion of mfp-5 to different substrates was found to increase with the contact time.

The adhesion of mfp-5 to mica was F_ad/R ∼ −4.6, −7.2 and −7.5 mN m⁻¹ (W_ad ∼ 0.98, 1.53 and 1.59 mJ m⁻²) for 2, 10 and 60 min contacts, respectively (figure 4a). The adhesion of mfp-5 to SiO₂ was observed to be F_ad/R ∼ −0.5, −0.5 and −11.5 mN m⁻¹ (W_ad ∼ 0.11, 0.11 and 2.44 mJ m⁻²) for 2, 10 and 60 min contacts, respectively (figure 4b). The adhesion of mfp-5 to PMMA was F_ad/R ∼ −1.0, −2.9 and 2.0 mN m⁻¹ (W_ad ∼ 0.21, 0.62 and 0.42 mJ m⁻²) for 2, 10 and 60 min contacts, respectively (figure 4c). The adhesion of mfp-5 to PS was F_ad/R ∼ −5.0, −7.4 and −11.2 mN m⁻¹ (W_ad ∼ 1.07, 1.57 and 2.37 mJ m⁻²) for 2, 10 and 60 min contacts, respectively (figure 4d). Overall, the adhesion of mfp-5 to the four different substrates follows the order PS ≈ mica > PMMA ≈ SiO₂ for short contact times (2–10 min), and PS ≈ SiO₂ > mica > PMMA for longer contacts (60 min; figure 5c). The hard wall distances for mfp-5 interaction cases of mica, SiO₂, PMMA and PS were determined to be 9 ± 0.5, 20 ± 0.5, 24 ± 0.5 and 20 ± 0.5 nm, respectively (figure 4), which indicates that the deposition of mfp-5 on SiO₂, PMMA and PS was more extensive than on mica.

4. Discussion

4.1. Effects of molecular weight and chain flexibility on the adhesion of mfps to different substrates

The molecular adhesion of mfps is generally correlated with the backbone flexibility in the proteins. It should be noted that for non-adhesive interactions, a flexible polymer in surface contact normally pays a high penalty in conformational entropy; thus its adhesion process is more entropically hindered with a more flexible backbone. For the adhesive bonding of mfps to different substrates in this study, higher chain flexibility of mfps is expected to have positive impact on the adhesion process as the local structure of the protein can better and more quickly adapt to the specific surface chemistry [40]. It should also be noted that the flexibility of surface bound polymer chains might not be the same as that in solutions, and further studies are necessary to fully resolve the relation between chain flexibility and adhesion. Chain flexibility generally decreases with increasing molecular weight in polymers [40]. Among the three mfps studied, the molecular weights are highest for mfp-1 (92.0 kDa), then mfp-5 (9.5 kDa) and lowest for mfp-3 (5.3 kDa) (table 1). In addition, mfp-1 has the least flexibility based on the flexible amino acids composition by B-factor (50.4 mol%) compared with mfp-3 (70 mol%) and mfp-5 (61.1 mol%). A recent study based on circular dichroism (CD) and sum frequency generation (SFG) vibrational spectroscopy also revealed that mfp-3 exhibits a flexible random coil conformation in solution that easily adapts to different surface chemistries, whereas mfp-1 is mainly composed of a stiff decapeptide repeats with a poly-proline type II helix separated by flexible hinges [41–43]. Therefore, the molecular flexibility of the three mfps is mfp-3 > mfp-5 > mfp-1.

The SFA results further show that the increasing adhesion with increasing contact time (independent of substrate chemistry) followed the order mfp-3 > mfp-5 > mfp-1 (figure 5), which is also consistent with the order of molecular flexibility. The effect of chain flexibility on mfps adhesion is more significant on a rough surface such as SiO₂ where tailored smooth contact can be obtained more easily for flexible macromolecules but not for rigid macromolecules [44]; thus, a significant adhesion increase was observed for mfp-3 and mfp-5 on SiO₂ for relatively long contact times (60 min; figure 5).

4.2. Proposed adhesion mechanisms of mfps to different substrates

All three mfps showed adhesive capabilities on the four substrates of different surface chemistry. Schematics of the molecular structures and possible interaction schemes are shown in figure 1. Several interaction mechanisms can be involved during the interactions of mfps to the four surface types, including electrostatic, hydrogen bonding, hydrophobic interactions, cation–π, π–π stacking and metal-complexation (figure 1).

All three mfps have basic pIs, are positively charged at the experimental conditions (pH 5.5) and thus can attract negatively charged mica or SiO₂ surfaces under the given pH [18]. The dissolution of K⁺ ions from mica and proton dissociation from mica and SiO₂ (pK_a 7.0) makes these surfaces negatively charged [45]. Under the experimental conditions (0.35 M salt), the concentrated K⁺ ions in the solution can compete with the mfp molecules for adsorption sites on the substrate surfaces and thereby largely suppress the net electrostatic interaction energies [18]. Therefore, electrostatic interactions are not likely to play a major role in the adhesion of the three mfps under the experimental conditions used.

Hydrogen bonds can form between the hydroxyl or amine groups (hydrogen donors) on mfps and the oxygen atoms (hydrogen acceptors) on mica, SiO₂ and PMMA (–O–H…O, –N–H…O), and between the oxygen or nitrogen atoms on mfps and hydroxyls on SiO₂ (–O–…H–O, –N–…H–O) [7,12,18]. However, it should be noted that water molecules in the solution are also able to form hydrogen bonds with the mfps and the substrates (so-called hydration layers).

Hydrophilicity calculations on the primary amino acid sequences of the mfps with inclusion of post-translational modifications indicate that the three mfps consist predominantly of hydrophilic domains based on the Hopp and Woods hydropathy index with nine amino acids as a window (see the electronic supplementary material, figure S1). Mfp-3 and mfp-5 show strong adhesion to PS, the most hydrophobic substrate investigated. This could be due to the interaction between exposed hydrophobic amino acid residues in the mfps and PS and/or cation–π interactions or π–π stacking, as discussed in more detail below.

Cation–π and π–π interactions play important roles in the interactions of many bio-interfaces (e.g. DNA structure, protein binding). Cation–π can be formed between the positively charged amines in mfps (i.e. Lys, Arg, over approx. 20 mol%) and aromatic groups on the substrate surface, and π–π stacking can be formed between the aromatic groups on mfps (over approx. 20 mol%) and the phenyl groups on PS [18,46–49]. In terms of the interaction energy, the cation–π interaction is comparable to hydrogen bonding while π–π stacking is weaker than hydrogen bonding [46–49]. The cation–π interaction has recently been implicated in the adhesion of green mfp (pvfp-1) [46] and mcfp-1 [18], which can probably also contribute to the adhesion of mfps to PS here.

Another possible interaction involved is the complexation between metal ions and the phosphate groups present in mfp-5 (9.4 mol% phospho-Ser) [15,50,51]. Small amounts of Al are present on the surface (exposed basal plane) of mica [37] that may associate with the phospho-Ser groups in mfp-5 or DOPA groups in all three mfps at pH 5.5.

4.3. Interactions of mfps with different substrates

The possible interaction mechanisms between the mfps and the four substrates tested can now be discussed in more detail.

4.3.1. Mica

Mica is a hydrophilic mineral and its exposed surface in water solution is polysiloxane with minor replacement of Si by Al (figure 1). DOPA bidentate hydrogen bonding to mica is believed to be the main contributor for the adhesion of mfps to mica [7,9,10,18]. In particular, the distances between adjacent O atoms on mica (0.28 nm) [52] and between OH groups in DOPA (0.29 nm) almost certainly facilitates DOPA bidentate hydrogen bonding to the mica surface [7]. Previous SFA measurements demonstrated that periodate treatment (DOPA oxidation) abolishes almost all of mfp-3 adhesion and more than 75 per cent of mfp-1 adhesion on mica [9,10,18]. Thicknesses of both mfp-1 and mfp-3 films measured by the SFA increased upon DOPA oxidation. This is in stark contrast to the previously reported contraction in periodate-treated mfp-1 films analysed by surface plasmon resonance and the quartz crystal microbalance [53,54]; these authors attributed film contraction to dehydration associated with protein cross-linking. In the present case, we attribute the periodate-treated mfp-1 and mfp-3 film expansion to tautomerization of DOPA-quinone to Δ-DOPA, because it is reversible and is known to induce significant reduction in their conformational flexibility [9]. Further studies are necessary to resolve the relationships between cross-linking and tautomerization.

For mfp-1 adhesion on mica, the significant hydroxyproline content could also contribute to the adhesion [18]. Thus, the hydrogen bonding follows the order of DOPA content in mol% (table 1) [13–15], i.e. mfp-5 > mfp-3 > mfp-1. Because of the additional contribution from hydroxyprolines in mfp-1, mfp-1 and mfp-3 may have similar hydrogen bonding strengths.

In addition to the hydrogen bonding, mica can support the formation of metal complexes via oxidized Al groups interacting with phosphoester groups in mfp-5 and DOPA groups in all mfps (figure 1). Such effects follow the order of mfp-5 > mfp-3 > mfp-1. Electrostatic effects can be neglected, considering the results of periodate treatment [9,10,18].

The above mechanisms are consistent with the experimental results from the SFA measurements reported here in that mfp-5 shows the highest adhesion, whereas mfp-3 and mfp-1 have similar adhesion to mica (figure 5).

4.3.2. SiO₂

For interactions with SiO₂, the adhesion follows the order mfp-3 > mfp-5 ≈ mfp-1 for short contact times, changing to for relatively longer contacts (60 min; figure 5). The significant dependence on contact time is attributed to molecular weight and chain flexibility as discussed in §4.1 (). Similar to mica, bidentate hydrogen bonding by DOPA is the major contributor to mfp adhesion to SiO₂ (mfp-5 > mfp-3 ≈ mfp-1). Electrostatic interactions can play a minor role, following the order of the relative proportion of basic residues in the mfps (mfp-3 > mfp-1 ≈ mfp-5).

The adhesion energy of mfp-1 to SiO₂ was lower than that to mica, whereas the adhesion of mfp-3 and mfp-5 to SiO₂ was higher than to mica after 60 min contact. The main reason for the different adhesion trends of mfps on mica and SiO₂ could be a surface roughness issue: the root mean square (r.m.s.) roughness determined by AFM was approximately 0.2 nm for mica, whereas on SiO₂ it was 1.0 nm. For example, in the case of mfp-1, the higher roughness of SiO₂ would inhibit the smooth adhesive contact [44] between mfp-1 and SiO₂ owing to local rigidity (presence of the stiff decapeptide) of mfp-1 chains, thereby reducing the adhesion energies. On the other hand, mfp-3 and mfp-5 might adapt better to the surface roughness of SiO₂ than mfp-1 owing to their higher chain flexibility and smaller molecular weights, thereby allowing for more and stronger adhesion bonds than on mica. The adhesion energies of mfp-3 and mfp-5 show stronger contact time dependence on SiO₂ surfaces (figure 5b,c) than for the other three substrates, which implies that longer contact times lead to better conformational re-arrangements of binding sites on mfp-3 and mfp-5 with the rough SiO₂ surface, as expected.

4.3.3. PMMA

For PMMA, the observed adhesion decreased as follows: mfp-3 > mfp-5 ≈ mfp-1. On the basis of the interaction mechanisms discussed in §4.2, the ability of the three mfps to provide hydrogen donors with consideration of all hydroxylation follows the order of mfp-1 > mfp-5 ≈ mfp-3, whereas the hydrophobic interactions follow the order mfp-3 > mfp-5 > mfp-1. These trends suggest that hydrophobic interactions prevail in the interactions between mfps and PMMA.

4.3.4. PS

Hydrophobic, cation–π and π–π stacking interactions can all contribute to and be important interaction mechanisms for the adhesion between mfps and PS. Considering that mfp-3 has the highest content of hydrophobic side-chains (35.5 mol%), followed by mfp-5 (29.1 mol%), and finally mfp-1 (28.6 mol%) (table 1) without adjusting for the effects of hydroxylation [13–15], the attractive hydrophobic interaction follows the order of mfp-3 > mfp-5 > mfp-1. Because the cation–π interaction strength is proportional to the amount of Lys and Arg in the mfps, which follows the order of mfp-3 (24.5 mol% Lys, Arg) > mfp-5 (22.6 mol%) > mfp-1 (20.4 mol%). π–π stacking is related to the amount of aromatic groups and roughly follows the order of mfp-3 (26.5 mol%) > mfp-5 (26.2 mol%) > mfp-1 (19.2 mol%). Therefore, after considering all the contributions above, the predicted adhesion of the mfps would follow the order of mfp-3 > mfp-5 > mfp-1, which agrees well with experimental results from SFA measurements.

It should be noted that, compared to mfp-3 and mfp-5, the relatively weak adhesion capability of mfp-1 to the different substrates is consistent with its coating as opposed to adhesive function in cuticle of byssal thread [12,18,50]. By contrast, mfp-3 is an adhesive primer for mussel adhesion and it should have underwater adhesion ability regardless of surface chemistry. Recent studies aided by SFG vibrational spectroscopy and CD strongly support the notion that mfp-3 adopts different conformations at various interfaces depending on specific chemical interactions [41–43]. Therefore, relatively stronger adhesion ability of mfp-3 to the tested substrates than mfp-1 and mfp-5 is partially due to a superior conformational adaptability of mfp-3 on the different surface chemistries.

5. Conclusions

The molecular interactions between three different kinds of Mytilus adhesive proteins (mfp-1, mfp-3, mfp-5) on four different substrates (mica, SiO₂, PMMA, PS) were directly measured in saline buffer using an SFA. The results provide important insights into the wet adhesion mechanisms, which were found to depend on both protein properties and substrate surface chemistry. All three proteins show adhesive versatility to both hydrophilic and hydrophobic substrates. Several interaction mechanisms are proposed, including electrostatic interaction, hydrogen bonding, hydrophobic interactions, cation–π, π–π stacking and metal-coordination. The extent to which these interactions contribute to adhesion depends on how well the critical attributes of each protein is matched to the surface tested. On the protein side, basic, aromatic and hydrophobic side-chains, the spacing between the two hydroxy groups of DOPA and chain flexibility influence the magnitude of measured mfp adhesion on all substrate. On the substrate surface side, roughness, charge, and the O–O distances of substrate surface functions are critical factors. Our results provide important insights into the design and development of biomimetic underwater adhesives and coating materials as well as anti-fouling materials.