Biomacromolecules 2009, 10, 1287–1294
1287
Low Biofouling Chitosan-Hyaluronic Acid Multilayers with Ultra-Low Friction Coefficients Jeroen H. H. Bongaerts,†,‡ Justin J. Cooper-White,*,§ and Jason R. Stokes‡,⊥ Unilever Corporate Research, Colworth Science Park, Sharnbrook, Bedfordshire, United Kingdom, and Tissue Engineering and Microfluidics Laboratory, Australian Institute for Bioengineering and Nanotechnology, The University of Queensland, St. Lucia, Brisbane, Australia Received September 24, 2008; Revised Manuscript Received March 13, 2009
Resistance to biofouling is an advantageous material property in a variety of biomedical and biofluid processing applications. Protein-resisting surface coatings must also be resistant to wear and degradation and in certain applications good aqueous lubricating properties are required. We show that cross-linked polyelectrolyte multilayers, consisting of chitosan and hyaluronan on polydimethylsiloxane (PDMS) surfaces, form a highly lubricating film that is resistant to wear and protein adsorption. The multilayer film shows much stronger resistance to protein adsorption from human whole saliva than both hydrophobic and hydrophilic PDMS surfaces; the latter two showed identical adsorbed salivary film thicknesses. The boundary friction coefficient under aqueous conditions was extremely low (µ ∼ 0.01) between multilayer-coated PDMS substrates and the film is robust against dry rubbing and many hours of tribological experiments in a range of aqueous lubricants. The origins of the assembly’s low friction coefficients and robustness are discussed. In addition, we found that the addition of negative phosphate ions to water lowers the boundary lubricating properties of negatively charged hydrophilic PDMS surfaces by 1 order of magnitude to µ ∼ 0.01. We consider this to arise from the large hydration sheaths and resulting “ballbearing” properties of the hydrated phosphate ions, which form a lubricating barrier against asperity contact. These findings offer new insights toward biolubrication processes and suggest that chitosan-hyaluronan polyelectrolyte multilayer films have the potential to be used in (bio-) applications requiring low friction as well as resistance to biofouling and wear.
1. Introduction Protein resistance is an advantageous property of material surfaces utilized within medical prostheses and implants, such as contact lenses, catheter tubes, and artificial joints. Nonspecific protein adsorption otherwise can initiate a chain of events that ultimately leads to inflammation and rejection of the device or implant. In addition to being protein resistant, some applications require surfaces to be highly lubricating and, most importantly, robust enough to resist both wear in rubbing contacts and degradation upon exposure to biological fluids (for example, artificial joints). Having a nonfouling surface with low surface friction is also useful in many other bioengineering processes that require the prevention of the formation of biofilms via the adhesion of biological components to device surfaces. One example where this is particularly important is in microfluidic devices targeted for diagnostics and biofluid analysis and manipulation. Surface coatings that are biocompatible, resistant to protein adsorption, and lubricating under physiological conditions thus constitute a highly relevant technological challenge. In particular, aqueous bio(mimetic)-lubrication has become an increasingly active area of research.1,2 It is well-known that hydrophobic surfaces are especially prone to nonspecific protein adsorption due to hydrophobichydrophobic interactions. It is for this reason that polydimeth* To whom correspondence should be addressed. Tel.: +61 7 3346 3858. E-mail:
[email protected]. † Current address: SKF Engineering and Research Centre, Kelvinbaan 16, 3439 MT, Nieuwegein, The Netherlands. ‡ Unilever Corporate Research. § The University of Queensland. ⊥ Current address: The University of Queensland, Division of Chemical Engineering, St. Lucia, Brisbane, Australia. E-mail:
[email protected].
ylsiloxane (PDMS), a common material in implants and in biomedical devices, is often rendered hydrophilic by plasma treatment.3 This converts methyl groups to hydroxyl and carboxyl groups at the exposed surface and also dramatically improves the aqueous lubricating properties of PDMS.4,5 However, the plasma-treated surfaces are unstable and revert back to being hydrophobic with time.6-8 Alternatively, adsorption and grafting of certain (co)polymers, most notably polyethylene glycol (PEG) or polyethylene oxide (PEO) can be used to hydrophilise surfaces more permanently.9 Grafting polymer entities such as amphiphilic block copolymers10,11 and mucin12 onto bare or plasma-treated PDMS substrates provides effective aqueous lubrication through a hydrated polymeric film, that also provides a steric barrier for asperity contact. Lee et al.10 showed that poly(L-lysine) (PLL)g-PEG bound to negatively charged oxide surfaces acts as a polymer “brush” and that the boundary friction coefficient between PDMS surfaces was around ∼0.03 under solvent conditions where the polymer was extended and highly solvated. This surface was shown to display some resistance to protein adsorption and wear in rubbing contacts. The creation of such a biocompatible lubricating surface coating that prevents adsorption of biological components may also be a very suitable starting point for designing surfaces that selectively bind particular proteins; selective binding can be essential for mediating cell adherence and integration of an object under physiological conditions. Mucin coatings, for example, have been shown to provide a protective film as well as simultaneously promoting specific interactions on a range of implant surfaces.13-15
10.1021/bm801079a CCC: $40.75 2009 American Chemical Society Published on Web 04/07/2009
1288
Biomacromolecules, Vol. 10, No. 5, 2009
Croll et al.16 have recently demonstrated that a chemically grafted polyelectrolyte multilayer consisting of the natural biopolymers chitosan (CHI), which is positively charged, and negatively charged hyaluronic acid (hyaluronan, HA), is resistant to a range of proteins when the multilayer is terminated by HA. Such surfaces are believed to resist nonspecific protein adsorption due to the highly hydrated state of the uppermost HA layer, which does not allow interactions of the substrate with the hydrophobic parts of the proteins. The use of CHI-HA multilayers on prosthetics has also been investigated for oral applications, including coating of dentures and as a means by which to control drug release.17,18 These in vitro and in vivo studies have shown that cross-linking the multilayer film provided greater resistance to wear and degradation from enzymes in saliva. However, quantitative measurements on the tribological properties of these CHI-HA films have never been previously assessed. In this paper, we investigate both the lubricating and protein-resistant properties of the cross-linked polyelectrolyte multilayer system consisting of chitosan and an outermost HA layer in vitro. These layers are deposited on a deformable PDMS substrate, which is selected as our biomimetic substrate material since most bodily surfaces are compliant and hydrophobic. The robustness and aqueous lubricating properties of the multilayer films are assessed using a tribometer by measuring the friction forces between a PDMS ball and disk coated with the polyelectrolyte multilayers. In a tribological contact, the resulting contact pressures for such compliant substrates are in the MPa regime, rather than the GPa regime that is the norm for more traditional tribological steel-steel contacts. When operating at these lower pressures we are in the iso-viscous elasto-hydrodynamic lubrication regime, where viscosity is not influenced by the contact pressure.19-21 The layer-by-layer (LBL) build-up is followed separately using the quartz-crystal microbalance with dissipation monitoring (QCMD), following similar procedures to those used by Croll et al.16 The QCM was subsequently used to investigate the ability of the CHI-HA multilayer to prevent nonspecific protein adsorption by exposing it to a diluted solution of 10% human whole saliva (HWS) and following the adsorption and desorption kinetics. Saliva was selected as the biological fluid “acid test” of protein resistance while also assessing the resistance of multilayer film to degradation by the enzymes present in saliva such as lysozyme and R-amylase. Saliva is a readily obtainable source of a large collection of highly surface-active proteins, with varying molecular weight, with positive, neutral, and negative net charge. Subsequently, salivary proteins are extremely efficient in adsorbing onto a wide range of surfaces, both hydrophobic (internal oral surfaces) and hydrophilic (teeth enamel).22-24 We aim to show that the cross-linked biomolecular CHI-HA multilayers form a highly lubricating robust surface film that is resistant to wear and significantly reduces the adsorption of salivary proteins, highlighting its potential use in applications where biocompatibility, protein resistance, and biolubrication are required.
2. Materials and Methods 2.1. Materials. The polyelectrolyte polymers that were used for the multilayer build-up are identical to those used by Croll et al.16 Dimethylformamide (DMF) and toluene were all AR grade and obtained from various sources. Dimethyl carbonate (DMC), 1-ethyl-3-(3dimethylaminopropyl) carbodiimide (EDAC), N-hydroxysuccinimide (NHS), morpholino(ethanesulfonic acid) (MES), glycerol, hydrochloric acid, and sodium chloride were obtained from Sigma-Aldrich and used as received. Chitosan 500, MW 900 kDa (Wako Fine Chemicals), was
Bongaerts et al. kindly donated by Professor Yoshinari Baba of Miyazaki University. Hyaluronic acid, MW 1.6 MDa, was obtained from Lifecore Biomedical. Water was demineralized and filtered (0.2 µm) using a Barnsted Nanopure Diamond Analystical ultrapure water system (resistivity of 18.2 MΩ-cm). Phosphate buffer saline (PBS) solutions were prepared using tablets (Biochemika, Fluka), producing 137 mM NaCl, 2.7 mM KCl, and 10 mM phosphate buffer solution (pH 7.4 at 25 °C). In one experiment, we prepared a 10 mM sodium phosphate buffer at pH 7.4, without salt present, to check for the influence of NaCl on tribological measurements. Human whole mouth saliva was stimulated by mechanical chewing action on a silicone rubber tube (2 cm length, 6 mm diameter), following the procedure described by Bongaerts et al.25 2.2. Quartz Crystal Microbalance. The LBL build-up and protein resistance of the coated surfaces was followed using an E4 quartzcrystal microbalance with dissipation monitoring (Q-sense, Sweden).26 The crystals are placed in the analysis chamber and measurements are taken continuously at the fundamental frequency, as well as the third, fifth, seventh, and ninth overtones. The desired solution is pumped through the measurement chamber at a flow rate of 0.1 mL/minute at a temperature of 22 °C. When switching from one solution to the next, flow is stopped and the inlet tube is moved between solutions. The fundamental frequency is disregarded for further analysis, as this is often significantly affected by interactions with the crystal mount. In this paper, only the fifth overtones are shown, for simplicity, which 5th has a resonance frequency of Fres ) 24.79 ( 0.02 MHz. Gold QCM crystals were coated by a thin layer of the same crosslinked PDMS as was used for the tribological surfaces. This was achieved by spin-coating of a toluene solution containing 5% PDMS (1:10 ratio of cross-linker (Dow Sylgard 184 silicone elastomer curing agent) and PDMS base) and subsequent curing at 70 °C overnight. This produces a hydrophobic surface that is chemically identical to those used in the friction measurements. The PDMS-coated gold crystals that are used for LBL deposition are rendered hydrophilic by exposing it to an oxygen plasma treatment (Cressington Carbon Coater, 208 Carbon), which renders the surface partially negative due to the creation of hydroxyl and carboxyl groups. A current of 15 mA in a vapor pressure of 0.2 mbar was applied for 99 s. The hydrophilic crystals are placed in the QCM flow cell immediately, well before the surface reverts back to being hydrophobic. The QCM data are presented as shifts in the resonance frequencies of the fifth overtone, which for rigid films is a measure of the adsorbed mass per unit area following the Sauerbrey equation. The film’s viscoelasticity or “softness” is indicated by the change in the relative width of the resonance peak, that is, the dissipation of the crystal’s oscillation. Further analysis is given where appropriate. 2.3. Tribology Experiments. The friction measurements are performed at room temperature (T ) 22 ( 2 °C) using a Mini Traction Machine (MTM, PCS Instruments Ltd., U.K.). The rubbing contact consists of a PDMS (Sylgard 184, Dow Corning) ball with a radius of 0.95 cm and a disk with a radius of 23 mm and a thickness of 4 mm. The root-mean-square (r.m.s.) roughness of the ball is 26 nm and that of the disk is 9 nm. The Youngs’ modulus of the PDMS used is E ) 2.4 MPa. More details of the preparation of the balls and disks are presented elsewhere.4 A normal load, L, is applied to the contact by pressing the sphere against the disk and is continuously monitored by a force sensor. The sphere and disk are rotated at different speeds such that there is a relative motion between the surface of the ball and the disk. This results in a slide-to-roll ratio (SRR) defined as SRR ) (Vball - Vdisk)/U, where U is the entrainment speed defined as U ) (Vball + Vdisk)/2, and Vball and Vdisk are the surface velocities of the ball and disk surface, respectively. The lateral friction force Ff exerted on the ball is measured through a force transducer, yielding the apparent friction coefficient µ ) Ff/L. Details of the setup and the procedure are reported elsewhere.4 The j average pressure in the contact at a typical load of L ) 1 N equals P ) 0.21 MPa. The slide-to-roll ratio is set to SRR ) 50% for all
Low-Friction Low-Fouling CHI-HA Multilayers
Biomacromolecules, Vol. 10, No. 5, 2009
1289
Table 1. Details of the Dipping Solutions Used in the LBL Deposition Procedure name
description
MES solution NHS solution EDAC solution HA dipping solution (50 µg/mL)
5 mM MES in water, pH ) 4.3 70 mg/mL NHS is dissolved in DMF 50 mg/mL EDAC is dissolved in DMF HA is dissolved first in 5 mM MES to a concentration of 2 mg/mL and subsequently filtered using a 0.2 µm filter. This solution is diluted in 5 mM MES to a final HA concentration of 50 µg/mL. The HA solution is functionalized for cross-linking by first adding 1 part in 200 of the EDAC solution; after 30 s, add 1 part in 200 of the NHS solution CHI is dissolved first in 5 mM MES and 50 mM HCL buffer to a concentration of 2 mg/mL and subsequently filtered using a 0.2 µm filter with positive surface charge (µposidyne acrosdisc from Pall Life Science). This solution is diluted in 5 mM MES to a final CHI concentration of 50 µg/mL
CHI dipping solution (50 µg/mL)
experiments, following our previous work.4,25 For clarity of the graphs, no error bars are shown; the scatter in the data is a good indication of their accuracy. 2.4. Multilayer Deposition. The LBL deposition procedure was taken and slightly adapted from Croll et al.16 to enable formation of a CHI-HA multilayer on the PDMS substrates for use in the tribometer. It was found that very robust layers are formed if the multilayers are chemically grafted onto the PDMS substrate. This is achieved using EDAC and NHS to functionalize the PDMS. First the PDMS surface is rendered hydrophilic by exposing it to the same oxygen plasma treatment as applied to the QCM crystals. The hydrophilic PDMS ball and disk are immediately exposed to sequential dip coating in alternating HA and CHI solutions (for 15 min in each solution), with rinses in MES in between each layer (for 30 s), as per Croll et al.16 The only difference to the procedure used by Croll et al. was that instead of a NaOH-buffered 5 mM MES solution at a pH of 5.5, a simple 5 mM MES solution was used with a pH of about 4.3. The details of the dipping solutions and rinses are listed in Table 1. The coated PDMS substrates were stored in MES buffer until they were used. The LBL deposition onto PDMS-coated QCM crystals was performed and monitored in the QCM disk in situ. LBL deposition of polyelectrolytes is sensitive to the exact flow conditions under which the polymers are deposited, which makes it difficult to replicate perfectly the LBL deposition conditions used for the macroscopic PDMS ball and disk. Preliminary experiments showed, for example, that a 15 min rinse step using the MES solution in the QCM resulted in much lower adsorbed amounts than in the absence of rinsing. In particular, the chitosan layer was very thin and rigid following extensively rinsing in MES. Therefore, to more closely mimic the LBL deposition procedure on the ball and disk, a consistent rinse step of 30 s was performed in between the CHI and HA deposition steps (which each lasted for 15 min). The rinsing step also provides a physical separation between the CHI and HA solutions in the Teflon tubes leading to the QCM crystals, which can otherwise result in the formation of CHI-HA aggregates in solution, blocking the tubes of the QCM.
all cases with limited drift in the measured frequencies. Material is deposited onto the surfaces at each deposition step to produce a thicker film that becomes increasingly more visco-elastic (less rigid) due to increased dissipation within the layers. The total frequency and dissipation shifts after each deposition step are shown in Figure 2, where the untreated hydrophilic PMDS in MES solution (prior to EDAC/NHS addition) was taken as the reference starting point. The frequency shift increases with each deposition step. Interestingly, after the addition of EDAC/NHS to functionalize the PDMS, there was a significant decrease in frequency response, some of which was reversed upon the first deposition of chitosan, suggesting some rearrangement of the PDMS surface and possible absorption of water or ions into this surface during functionalization. The addition of chitosan resulted in a reduction in the dissipation, suggesting that the incorporation of chitosan into the PDMS surface displaced previously bound water or ions. Thereafter, the additional deposition steps were similar in magnitude for each biopolymer. Our observed total frequency shift for each bilayer is significantly lower than the one observed previously by Croll et al.,16 where the frequency shift per bilayer was in the order of 120 Hz. This is likely to be caused by the fact that, unlike Croll et al., our solution was not buffered, which resulted in a different pH of 4.3 in our case, instead of pH ) 5.5, which will have reduced the overall charge on the HA in comparison to the chitosan. Shifts in frequency after changing the fluid flowing over the crystals from MES to PBS buffer were also recorded (data not
3. Results and Discussion 3.1. Multilayer Build Up. The deposition process of chitosan and HA on a plasma-treated PDMS film can be followed in Figure 1 in terms of the frequency shift and dissipation of the fifth overtone of the QCM crystal. A larger negative frequency shift corresponds to more deposited material, to a first approximation. Both two bilayers and five bilayers were deposited and then left in 5 mM MES; this allowed for the investigation of the importance of the number of bilayers in rendering the surfaces protein resistant. The 30 s MES rinsing steps in between the deposition steps can be identified as the small upturns (downturns) in the frequency (dissipation) curves. The LBL deposition in the QCM produces almost identical coatings in
Figure 1. QCM results of the LBL assembly of 2 and 5 CHI-HA multilayers on PDMS-coated gold crystals. Only the 5th overtone (Fresonance ) 24.79 ( 0.02 MHz) is shown. The dashed gray lines and the corresponding numbers indicate where the deposition of 1, 2, 3, 4, or 5 bilayers has been completed.
1290
Biomacromolecules, Vol. 10, No. 5, 2009
Bongaerts et al.
Figure 2. Frequency and dissipation shifts of the 5th overtones (Fresonance ) 24.79 ( 0.02 MHz) after each deposition step. Even steps correspond to a layer ending with HA (indicated by the vertical dashed lines), odd steps to CHI. Step 0 corresponds to the untreated hydrophilic PMDS in MES solution, prior to EDAC/NHS addition.
Figure 3. Friction force in the boundary lubrication regime plotted versus applied load. Entrainment speed U ) 5 mm/s and SSR ) 50%. (O) Bare PDMS + water, (gray box with black star) bare PDMS + HWS, (9) two bilayers CHI-HA on PDMS + MES, (gray circle) two bilayers CHI-HA on PDMS + PBS, (4) two bilayers CHI-HA on PDMS + H2O.
shown). The electrolyte concentration and pH is different for these fluids (5 mM and pH 4.3 for MES; 150 mM and pH 7.4 for PBS) and one would expect this to influence the highly charged multilayer assembly. This was indeed observed for all samples as a decrease in the resonance frequency (inferring an increase in mass) and an increase in the dissipation. Such behavior is expected for layers that swell and incorporate more water, hence increasing their effective mass and rendering the layer more visco-elastic and less rigid. It is interesting to note that the shifts were, within experimental accuracy, the same for both two and five bilayers. We believe that this indicates that only the surface layer is affected by the change in salt conditions. 3.2. Lubricating Properties. The lubricating properties of the PDMS ball and disk, coated with two bilayers of CHI-HA, were measured at room temperature (22 °C) using water, 5 mM MES solution, and a PBS buffer (from tablet, pH 7.4 with 150 mM NaCl) as the lubricant. The load-dependent friction forces are presented in Figure 3 at an entrainment speed U ) 5 mm/s and SRR ) 50%; these conditions correspond to the boundary lubrication regime for water between hydrophobic PDMS (HB PDMS). Figure 3 clearly demonstrates that the multilayer assembly reduces the friction forces significantly, when compared to bare PDMS and it does so by almost 2 orders of
Figure 4. Speed-dependent friction curves at L ) 1 N, SRR ) 50%. Symbols: (O) bare HB PDMS + water, (gray box with black cross) bare HB PDMS + PBS buffer, (black star) bare HL PDMS + water, (9) two bilayers CHI-HA on PDMS + MES, (gray circle) two bilayers CHI-HA on PDMS + PBS, (4) two bilayers CHI-HA on PDMS + H2O. The thick dashed line corresponds to the HB master curve, the thin dashed line to the HL master curve.
magnitude. There is little difference between using demineralized water, 5 mM MES, or PBS buffer. The exceptionally low friction coefficient observed for polyelectrolyte bilayers is comparable to that obtained for mechanically stimulated HWS physically adsorbed onto hydrophobic PDMS ball and disk.25 The speed-dependent friction coefficients (Stribeck curve) at a load L ) 1 N are shown in Figure 4 for PDMS surfaces, oxygen plasma-treated PDMS surface (HL PDMS), and multilayer-coated PDMS surfaces. For bare HB PDMS lubricated by water and other aqueous (nonabsorbing) fluids, there is a high friction-coefficient plateau at speeds up to ∼80 mm/s, indicating the boundary lubrication regime, where surfaces are in direct contact. At high speeds (U > 1000 mm/s), elastohydrodynamic lubrication is observed where the friction coefficient increases linearly with increasing speed and the surfaces are separated by a full lubricating fluid film. Between the boundary and hydrodynamic regimes is the mixed regime where both asperity-asperity contacts and hydrodynamics play a role. Note that the PBS buffer in conjunction with HB PDMS slightly reduces the boundary friction coefficient and shifts the Stribeck curve to lower speeds when compared to water. The low-speed
Low-Friction Low-Fouling CHI-HA Multilayers
friction coefficient, corresponding to the boundary lubrication regime for HB PDMS surfaces, is reduced by 1 order of magnitude when the PDMS surface has been rendered hydrophilic (Figure 4), as observed previously.4,5 At the lower speeds, the multilayer-coated PDMS tribopair results in the lowest friction coefficients, lubricated by either water, MES or PBS; they are reduced by more than 2 orders of magnitude when compared to HB PDMS. Note that full-film hydrodynamic lubrication is the same regardless of surface treatment, although it is entered into at lower speeds for the better lubricating surfaces than for the HB PDMS tribopair. The polyelectrolyte CHI-HA multilayer-coated PDMS surfaces show extremely low boundary friction coefficients of µ e 0.01. However, the low boundary friction we have observed is in contrast to previous investigations on cross-linked HA and chitosan on their own. Benz et al.27 compared the friction of physisorbed, covalently bound, and cross-linked HA in PBS buffer on the surface force apparatus. None of these systems were highly lubricating and only the cross-linked HA showed a load-bearing capacity. It was hypothesized that HA was not directly responsible for the efficient biolubrication of cartilage as previously thought, but that it did take part in the formation of a gel layer on cartilage surfaces by combining with other lubricating species. Kampf et al.28 found that while chitosan itself was lubricating, the friction coefficient between crosslinked chitosan films was quite high when measured using the surface force apparatus. The poor lubrication properties of crosslinked chitosan were hypothesized to be due to an adhesive interaction mediated by the cross-linking agent and subsequent enhanced segmental friction between gel layers. In our multilayer, each bilayer constitutes a layer of chitosan cross-linked to the adjacent layer of HA. That is, each chitosan molecule is confined and both covalently and electro-statically bound to the HA layers above and below it. In contrast to the work mentioned above,27 we observe extremely efficient boundary lubrication for our multilayer, HA-terminated surface. While the robustness of our multilayer film is associated with its cross-linked nature, we believe the low friction coefficient in the boundary regime arises from the highly charged nature of the outermost HA layer; this leads to both double-layer repulsion, as well as extensive hydration of the surface and, hence, the presence of a solvent layer, adjacent to and within the uppermost HA gel layer. A similar low boundary friction coefficient and explanation has been found by Gong et al.29 between two polyelectrolyte bulk gels, and by Raviv et al.30 who highlight that hydration layers surrounding ions can act as “molecular ball bearings” to provide efficient lubrication between charged surfaces. To investigate the origins of the effective lubrication by the CHI-HA multilayer further and especially the role of charge in water-lubricated tribological contacts, we measured the friction between the HL PDMS ball and disk (no multilayers) lubricated by water solutions containing various electrolytes. As shown in Figure 5, the friction coefficients for HL PDMS lubricated by 150 mM NaCl solution and 5 mM MES are, within experimental error, the same as those for pure water. However, 10 mM phosphate buffer at pH 7.4, both in the absence and presence of 150 mM NaCl, displays friction coefficients that are significantly lower than those obtained for hydrophilic surfaces in salt-free water. Moreover, they are actually equal to those of the multilayer-coated ball and disk. In addition, we found that a 10 mM PBS buffer at pH 5 (without NaCl) gave a similarly low friction coefficient. This suggests that, for PDMS, efficient boundary lubrication (µ < 0.01) can be obtained by rendering the surface hydrophilic using oxygen plasma
Biomacromolecules, Vol. 10, No. 5, 2009
1291
Figure 5. Friction coefficients obtained for a HL PDMS tribopair and water and a low salt and high-salt PBS buffer. Symbols: (black star) HL PDMS + water, (-) HL PDMS + 150 mM NaCl, (gray box) HL PDMS + 5 mM MES, (0) HL PDMS + 10 mM phosphate buffer, no salt, (4) HL PDMS, 10 mM phosphate buffer + 150 mM NaCl, (gray circle) two bilayers CHI-HA on PDMS + PBS. The thin dashed line corresponds to the HL master curve.
treatment and adding phosphate ions to the aqueous lubricant. Rendering the PDMS surface hydrophilic effectively turns the surface into a hydrated cross-linked polyelectrolyte layer, much like the investigated HA layer. Plasma treatment exposes charged hydroxyl and carboxyl groups on the surface, thus allowing a layer of water molecules to associate with the surfaces. Our results suggest that phosphate ions confer an additional barrier against asperity contact and facilitate better lubrication for HL PDMS in comparison to water and NaCl solutions. The phosphate ions at pH ) 7.4 are mainly present as (H2PO4)-1 and (HPO4)-2 at a concentration of 3.92 and 6.08 mM, respectively. Sodium counterions are likely to screen the negative surface charges, thus facilitating the entrainment of negatively charged phosphate ions into the contact. We propose that the comparatively large size and nature of the hydrated phosphate ion sheaths act both as surface separators and as “molecular ball-bearings” following the proposed mechanism for confined hydrated ions of Raviv and Klein.31 To our knowledge, all previous reports of lubrication by “ionic ballbearings” concerned cases where the ions were either attracted to the surface by coulomb interaction or were part of polyelectrolyte brushes; this is not the case here for the phosphate ions. We also found that the entrainment of hydrated phosphate ions enhanced aqueous lubrication between hydrophobic surfaces as shown in Figure 4. We note that the shift in the PDMS Stribeck curve to lower speeds, the enhanced entrainment, and slightly lowering of the boundary friction coefficient is very reminiscent of the effect nonionic surfactants have above the critical-micelle concentration, see Graca et al.33 We have found that the cross-linked multilayers are extremely robust. For example, two bilayers of CHI-HA are able to withstand rubbing under dry conditions and still retain their lubricating properties upon rehydration (Figure 6). Under dry conditions, the friction coefficient is relatively high at µ ≈ 1, but decreases back to 0.02 on addition of water, indicating the importance of the hydration of the layers for efficient boundary lubrication. The deposited CHI-HA multilayer also remained intact following many hours of use during repeat tests in air, water, and buffer solutions. This robustness of the polyelectrolyte multilayers requires the layers to be cross-linked by EDAC/ NHS in the HA solution, as noncross-linked multilayers are
1292
Biomacromolecules, Vol. 10, No. 5, 2009
Figure 6. Friction coefficient for a PDMS ball and disk, coated by two bilayers of CHI-HA, lubricated by water at U ) 5 mm/s, and L ) 1 N. At t ) 280 s, indicated by the vertical line (gray solid line), the sample area is drained and the ball and disk surfaces are dried using a continuous flow of dry air. The other vertical lines indicate times at which water was added again to the dried contact: (gray dashed line) add three drops of water, (black dashed line) add three more drops of water, and (black solid line) fill sample pot with water.
rubbed off within a matter of seconds leading to an immediate increase in friction by at least an order of magnitude (not shown for brevity). 3.3. Resistance to Saliva Adsorption. The suggested blankslate properties of the CHI-HA multilayers are investigated by determining the response of the polyelectrolyte multilayers to exposure to human whole mouth saliva in both the QCM and tribometer. For the QCM studies, mechanically stimulated HWS was diluted to a concentration of 10% saliva in PBS buffer. Diluted saliva was used for practical reasons as saliva volumes of tens of mLs were required in the QCM experiments. Vassilakos et al.22 have shown that a similar amount of material is absorbed from 10% diluted samples compared to whole saliva. The diluted solution was centrifuged at 3000 rpm for 9 min to remove any solids and precipitates that were present already or were formed upon dilution. The supernatant was decanted and used in the QCM experiments. The adsorption of components from saliva in the QCM is shown in Figure 7 for HB and HL PDMS, as well as for the two and five CHI-HA bilayers on HL PDMS. The baseline for T < 0 corresponds to the situation for each surface in PBS buffer prior to addition of saliva, which is added at T ) 0. The degree of adsorption on both HB and HL PDMS is very similar with a final frequency shift of ∼93 Hz. Saliva seems to adsorb slower onto the HL crystals and to have not reached equilibrium in the HL case completely even after 1 h. The dissipation for both HB and HL PDMS reaches D ≈ 14 × 10-6, which indicates a relatively viscoelastic (“soft”) adsorbed layer. Note that the salivary components were allowed to adsorb until they reached almost their equilibrium adsorbed amount, which was reached after approximately 1 h. The results for HB and HL PDMS show that overall negative charge or hydrophilicity is not sufficient to prevent protein adsorption from saliva, as the adsorbed amount onto HL PDMS is almost identical as that onto HB PDMS. The adsorbed amount of material from saliva, which is in first order linear to the observed frequency shift, is in the case of both two and five CHI-HA bilayers a factor ∼4 lower (at ∼22.5 and 25.6 Hz for the two samples with two bilayers; at 24.4 Hz for the sample with five bilayers) than the final adsorbed amount for both bare HL and bare HB PDMS surfaces.
Bongaerts et al.
Figure 7. Fifth overtones of the QCM experiment upon saliva adsorption at T ) 0 and rinsing using PBS buffer after T ∼ 3600 s. Frequency shifts are on left axis, upper curves: (9) five bilayers, (gray circle) two bilayers, (thick gray line) HL PDMS, (thick black line) HB PDMS; dissipation is on right axis, lower curves: (0) five bilayers, (O) two bilayers, (thin gray line) HL PDMS, (thin black line) HB PDMS.
Figure 8. Friction coefficients for PDMS ball and disk coated by two bilayers of CHI-HA, lubricated by (in this order) water, HWS and 0.1% SDS. Symbols: (gray circle) two bilayers in H2O, ([) two bilayers with 1.5 mL HWS + 0.5 mL H2O, (black box with white star) two bilayers in 0.1% SDS, (black star) HL PDMS + water. The thin dashed line corresponds to the HL master curve.
Furthermore, the change in dissipation is minimal (D e 10-6) for the CHI-HA multilayered substrates, indicating that a very solid-like layer gets adsorbed and that the hydration of the underlying polyelectrolyte multilayers is not affected. Similar frequency shifts after reaching equilibrium adsorption are found for five CHI-HA layers as for two CHI-HA layers. However, upon rinsing with PBS buffer greater desorption is observed from five bilayers than for two bilayers, for which desorption is small; the final (remaining) frequency shift after rinse for two and five bilayers is 19.9 and 8.5 Hz, respectively. This indicates that even though adsorption from saliva is similar for two and five bilayers, the adsorbed components are more loosely bound to five bilayers than to two bilayers, rendering it easier to remove during rinsing. Figure 8 shows that the friction coefficient is slightly greater for saliva (1.5 mL mechanically stimulated saliva, 0.5 mL water) on the multilayer-coated surface compared to water alone, which is also similar to saliva adsorbed onto a naked PDMS surface. In an effort to remove any adsorbed saliva, the ball and disk were rinsed in 0.1% sodium dodecyl sulfate (SDS) and the friction coefficients measured. Following the rinse in SDS, the
Low-Friction Low-Fouling CHI-HA Multilayers
Figure 9. Saliva adsorption and desorption curves (using PBS buffer) for a QCM crystal coated by five CHI-HA bilayers. Frequency shifts (left axis): (thick gray dashed line) first saliva adsorption and subsequent desorption curve (from rinsing with PBS), (thick solid black line) second saliva adsorption and desorption curve (from rinsing with PBS) after 2 h of desorption and soaking overnight in PBS. Dissipation curves (right axis): (thin gray dashed line) first saliva adsorption and subsequent desorption curve, (thin solid black line) second saliva adsorption and desorption.
friction coefficient using the SDS solution is the same as that in water alone. It is likely that the SDS has removed the component that was adsorbed from saliva since SDS is known to denature proteins. This also suggests that SDS, a highly surface active anionic surfactant, has not negatively affected the multilayer assembly, confirming that the CHI-HA multilayer is extremely robust against external influences. The CHI-HA multilayers do not act as a perfect “blank-slate” for saliva, as has been observed by Croll et al. to be the case when tested with whole serum and serum proteins, but they do significantly reduce the adsorption of salivary proteins to the surfaces. Only a very rigid layer gets adsorbed on the multilayers, resulting in a very different structure of the adsorbed salivary film when compared to that of the film adsorbed on PDMS. At this point in time it is unknown exactly which salivary constituents are responsible for this adsorbed layer. Given that the HA layer is highly negatively charged, the most likely candidates are positively charged species that are known to exist in saliva such as cations, positively charged peptides, or small positively charged proteins. Further work is ongoing. Salivary proteins are extremely surface active and one could envisage any initial adsorption of saliva’s constituents to form an anchor for the other proteins to latch onto, thereby destroying the protein-resistant nature of the surface in the long run. This was found not to be the case in a further experiment. Figure 9 shows two adsorption and desorption curves of saliva onto the PDMS crystal coated by five CHI-HA bilayers. The desorption of proteins from this surface was initiated by continuous rinsing with PBS. The first curve is the first (or “maiden”) adsorption (and subsequent desorption) curve of saliva onto the multilayers, already shown in Figure 7. The second curve is obtained after rinsing this adsorbed saliva layer with PBS for several hours, subsequently leaving it overnight in PBS buffer and performing a second adsorption experiment with a fresh diluted (10%) saliva solution. The adsorbed amount from saliva in the second curve is significantly lower than that of the first curve, indicating that the ability of the multilayers to prevent adsorption of most of the salivary proteins is preserved, if not improved. The frequency shift after the second rinse apparently returns to zero, which would indicate that all the adsorbed species desorb during the rinse. Note, however, that at this stage, a positive drift occurred
Biomacromolecules, Vol. 10, No. 5, 2009
1293
in the QCM system and we expect that some of the material remains after the second rinse. The protective nature of the multilayers with preadsorbed saliva could, speculatively, have been caused by a slight screening of the negative charges on the HA layer by the material that was left after rinsing. If true, this material is able to do so without destroying the hydration of the outermost HA layer, which is suggested to be the mechanism for preventing protein adsorption. 3.4. Discussion. The highly lubricating properties of the cross-linked CHI-HA multilayer under aqueous and physiological conditions, where the boundary friction coefficient was µb ∼ 0.01, is similar to other efficient boundary lubricants that have been reported in the literature between compliant (“soft”) surfaces (e.g., PDMS) such as saliva25 and grafted polylectrolyte brushes.10 Similarly, the origin of low boundary friction under aqueous conditions for the polyelectrolyte cross-linked CHIHA multilayer-coated PDMS surfaces is likely to arise due to the high charged density and hydration of the gel-layer. This may provide some electrostatic repulsion between surfaces, but more importantly, it enables the presence of a solvent water layer associated with the ions on and within the cross-linked multilayer coating. This film is load bearing and, when chemically grafted onto the substrate, resists wear due to the visco-elastic gel-like nature of the hydrated multilayer. While similar lubricity to the CHI-HA multilayer film can be achieved by plasma treating PDMS to render it negatively charged and hydrophilic, particularly in the presence of phosphate ions, this does not render the surfaces protein-resistant. We observed a similar amount of adsorbed salivary components on both HB and HL surfaces. This is possibly due to the fact that the whole surface is not rendered hydrophilic homogeneously, and hydrophobic patches remain onto which the proteins adsorb. Another similar explanation could be that the surface charge density of the HL PDMS chains is lower than that of HA, yielding a lower hydration state of the surface layer and resulting in the surface being less resistant to protein adsorption. However, by coating the surface by a CHI-HA multilayer, the equilibrium adsorbed amount is reduced by a factor 4 when compared to PDMS (HL or HB). The layer adsorbed on the multilayer assembly is also very different than that on bare HB and HL PDMS surfaces. The observed dissipation in the QCM experiment for the multilayer assembly is minimal (D ≈ 10-6), which indicates a very rigid layer consisting of relatively small, poorly hydrated proteins or biomolecules. The amount of material that remains adsorbed from saliva, following a rinse by PBS buffer, is significantly lower for PDMS coated by five bilayers of CHI-HA than for both bare PDMS (HB and HL) and PDMS coated by only two bilayers. This indicates that, for five bilayers, any adsorbed material is bound less strongly than for two bilayers. Future work will include an investigation into whether the protein resistance of these surfaces can be improved by the addition of further bilayers, by changing the molecular weight of the biopolymers, the conditions during multilayer assembly and so on. In addition, future work will aim to identify the composition of the salivary components absorbed to the multilayer surface, to enable design of a surface that is resistant to further adsorption of oral salivary proteins and potentially other biological active molecules, while still providing a hydrated lubricating surface.
4. Conclusion We have demonstrated that assembling a cross-linked CHIHA polyelectrolyte multilayer, terminated by an outermost HA
1294
Biomacromolecules, Vol. 10, No. 5, 2009
layer, onto a PDMS substrate creates a biocompatible surface that inhibits protein adsorption, forms a highly lubricating film under physiological conditions, and one that is load bearing and robust enough to be resistant to wear and degradation over time. On exposure to a real biofluid, that is, human whole mouth saliva, which consists of hundreds of different proteinaceous components, only a small amount of adsorption occurs that has a negligible effect on the dynamics of the adsorbed polyelectrolyte multilayer or its lubrication properties. In comparison, saliva components readily adsorbed onto both bare (hydrophobic) and plasma-treated (hydrophilic) PDMS. In addition, the resulting multilayer, when exposed to human saliva becomes coated by a thin solid-like layer that shows improved resistance to further adsorption from salivary proteins and is stable under the time frames and conditions investigated. For the first time, we also observe that phosphate ions significantly improve the lubrication of both hydrophilic negatively charged PDMS contacts and hydrophobic PDMS contacts. We suggest that the highly hydrated phosphate ions act as “molecular ball-bearings”30-32 in the rubbing contact. The role of hydrated ions in macroscopic rubbing contacts will be investigated further using a range of ions with varying hydrated sheaths and ionic strengths, since it appears that this may be an important requirement in biolubrication processes. In conclusion, the coating of surfaces using cross-linked CHIHA multilayers is potentially feasible for biomedical and other applications that require lubrication and protein resistance. We recommend that further tests using other relevant biomaterials coated with this polyelectrolyte multilayer should be performed in vitro using a tribological device to test their susceptibility to wear and degradation under physiological and rubbing conditions, prior to use under in vivo conditions. Acknowledgment. The authors would like to thank Lubica Macakova for her help with the QCM experiments and in preparing the PDMS-coated QCM crystals, as well as for helpful discussions. We would also like to thank Thomas Branfield for performing the tribological experiments on uncoated hydrophilic PDMS. The authors thank Gleb Yakubov for helpful discussions. J.J.C.-W. wishes to thank Unilever for supporting his secondment to Unilever Corporate Research laboratories where all experimental work was performed. This work was also performed as part of the Australian Research Council Linkage Project LP0776963.
References and Notes (1) Lee, S.; Spencer, N. D. Science 2008, 319 (5863), 575–576. (2) Dowson, D.; Neville, A. Proc. Inst. Mech. Eng., Part J 2006, 220 (J3), 109–123. (3) Everaert, E. P. J. M.; van de Belt-Gritter, B.; van der Mei, H. C.; Busscher, H. J.; Verkerke, G. J.; Dijk, F.; Mahieu, H. F.; Reitsma, A. J. Mater. Sci.: Mater. Med. 1998, 9 (3), 147–157.
Bongaerts et al. (4) Bongaerts, J. H. H.; Stokes, J. R.; Fourtouni, K. Tribol. Int. 2007, 40, 1531–1542. (5) Lee, S.; Spencer, N. D. Tribol. Int. 2005, 38 (11-12), 922–930. (6) Everaert, E. P.; vanderMei, H. C.; Busscher, H. J. J. Adhes. Sci. Technol. 1996, 10 (4), 351–359. (7) Everaert, E. P.; vanderMei, H. C.; Devries, J.; Busscher, H. J. J. Adhes. Sci. Technol. 1995, 9 (9), 1263–1278. (8) Lawton, R. A.; Price, C. R.; Runge, A. F.; Doherty, W. J.; Saavedra, S. S. Colloids Surf., A 2005, 253 (1-3), 213–215. (9) Murthy, R.; Cox, C. D.; Hahn, M. S.; Grunlan, M. A. Biomacromolecules 2007, 8 (10), 3244–3252. (10) Lee, S.; Muller, M.; Ratoi-Salagean, M.; Voros, J.; Pasche, S.; De Paul, S. M.; Spikes, H. A.; Textor, M.; Spencer, N. D. Tribol. Lett. 2003, 15 (3), 231–239. (11) Lee, S.; Iten, R.; Mu¨ller, M.; Spencer, N. D. Macromolecules 2004, 37 (22), 8349–8356. (12) Lee, S.; Mu¨ller, M.; Rezwan, K.; Spencer, N. D. Langmuir 2005, 21 (18), 8344–8353. (13) Feiler, A. A.; Sahlholm, A.; Sandberg, T.; Caldwell, K. D. J. Colloid Interface Sci. 2007, 315 (2), 475–481. (14) Shi, L.; Ardehali, R.; Valint, P.; Caldwell, K. D. Biotechnol. Lett. 2001, 23 (6), 437–441. (15) Shi, L.; Ardehali, R.; Caldwell, K. D.; Valint, P. Colloids Surf., B 2000, 17 (4), 229–239. (16) Croll, T. I.; O’Connor, A. J.; Stevens, G. W.; Cooper-White, J. J. Biomacromolecules 2006, 7 (5), 1610–1622. (17) Etienne, O.; Schneider, A.; Taddei, C.; Richert, L.; Schaaf, P.; Voegel, J. C.; Egles, C.; Picart, C. Biomacromolecules 2005, 6 (2), 726–733. (18) Picart, C.; Schneider, A.; Etienne, O.; Mutterer, J.; Schaaf, P.; Egles, C.; Jessel, N.; Voegel, J. C. AdV. Funct. Mater. 2005, 15 (11), 1771– 1780. (19) de Vicente, J.; Stokes, J. R.; Spikes, H. A. Tribol. Lett. 2005, 20 (34), 273–286. (20) Lee, S.; Spencer, N. D. Tribol. Int. 2005, 38 (11-12), 922–930. (21) Dowson, D. Proc. Inst. Mech. Eng. 1966, 181 (3), 45–54. (22) Vassilakos, N.; Arnebrant, T.; Glantz, P. O. Scand. J. Dent. Res. 1992, 100 (6), 346–353. (23) Lindh, L.; Glantz, P. O.; Svensater, G.; Arnebrant, T. J. Dent. Res. 2002, 81, A282. (24) Lindh, L.; Arnebrant, T.; Isberg, P. E.; Glantz, P. O. Biofouling 1999, 14 (3), 189–196. (25) Bongaerts, J. H. H.; Rossetti, D.; Stokes, J. R. Tribol. Lett. 2007, 27, 277–287. (26) Marx, K. A. Biomacromolecules 2003, 4 (5), 1099–1120. (27) Benz, M.; Chen, N. H.; Israelachvili, J. J. Biomed. Mater. Res., Part A 2004, 71A (1), 6–15. (28) Kampf, N.; Raviv, U.; Klein, J. Macromolecules 2004, 37 (3), 1134– 1142. (29) Gong, J. P.; Kagata, G.; Osada, Y. Macromol. Symp. 2003, 195, 209– 216. (30) Raviv, U.; Giasson, S.; Kampf, N.; Gohy, J. F.; Jerome, R.; Klein, J. Nature (London) 2003, 425 (6954), 163–165. (31) Raviv, U.; Klein, J. Science 2002, 297 (5586), 1540–1543. (32) Klein, J.; Raviv, U.; Perkin, S.; Kampf, N.; Chai, L.; Giasson, S. J. Phys.: Condens. Matter 2004, 16 (45), S5437-S5448. (33) Graca, M.; Bongaerts, J. H. H.; Stokes, J. R.; Granick, S. J. Colloid Interface Sci. 2007, 315 (2), 662–670.
BM801079A