ARTICLE pubs.acs.org/Biomac
Articular Cartilage Proteoglycans As Boundary Lubricants: Structure and Frictional Interaction of Surface-Attached Hyaluronan and HyaluronanAggrecan Complexes Jasmine Seror,† Yulia Merkher,‡ Nir Kampf,† Lisa Collinson,§ Anthony J. Day,§ Alice Maroudas,‡ and Jacob Klein*,† †
Department of Materials and Interfaces, Weizmann Institute of Science, Rehovot, 76100, Israel Department of Biomedical Engineering, Technion Institute of Technology, Haifa 32000, Israel § Wellcome Trust Center for Cell-Matrix Research, Faculty of Life Sciences, University of Manchester, Oxford Road, Manchester M13 9PT, United Kingdom ‡
bS Supporting Information ABSTRACT: Mammalian synovial joints are extremely efficient lubrication systems reaching friction coefficient μ as low as 0.001 at high pressures (up to 100 atm) and shear rates (up to 106 to 107 Hz); however, despite much previous work, the exact mechanism responsible for this behavior is still unknown. In this work, we study the molecular mechanism of synovial joint lubrication by emulating the articular cartilage superficial zone structure. Macromolecules extracted and purified from bovine hip joints using well-known biochemical techniques and characterized with atomic force microscope (AFM) have been used to reconstruct a hyaluronan (HA)aggrecan layer on the surface of molecularly smooth mica. Aggrecan forms, with the help of link protein, supramolecular complexes with the surface-attached HA similar to those at the cartilage/synovial fluid interface. Using a surface force balance (SFB), normal and shear interactions between a HAaggrecan-coated mica surface and bare mica have been examined, focusing, in particular, on the frictional forces. In each stage, control studies have been performed to ensure careful monitoring of the macromolecular surface layers. We found the aggrecanHA complex to be a much better boundary lubricant than the HA alone, an effect attributed largely to the fluid hydration sheath bound to the highly charged glycosaminoglycan (GAG) segments on the aggrecan core protein. A semiquantitative model of the osmotic pressure is used to describe the normal force profiles between the surfaces and interpret the boundary lubrication mechanism of such layers.
’ INTRODUCTION Articular cartilage is the smooth tissue that covers the sliding surfaces at the end of bones in synovial joints, such as hips and knees. Together with the fluid entrapped within the synovial cavity (synovial fluid (SF)), it is responsible for the lubrication of the joints, enabling smooth articulation at mean pressures up to an order of 100 atm and at shear rates from rest up to 106 to 107 s1.1,2 When healthy, the synovial joint is an extremely efficient tribological system, leading to friction coefficients in the range of 0.001 to 0.005 at the physiological working conditions recalled above.1,2 Such low friction at these high physiological pressures in aqueous media is almost unique in nature and unequaled by synthetic systems. Many models have been proposed to explain the origin of this very efficient lubrication. These have been based both on nonboundary-lubrication ideas, including hydrodynamic and elasto-hydrodynamic lubrication,3 interstitial fluid pressurization,4 and others, and more recently on boundarylubrication concepts that consider the nature of the molecular layer at the outer cartilage surface. Such models are recalled in the review of Klein.5 In contrast, when lesions occur as a result of r 2011 American Chemical Society
trauma or during the course of joint diseases such as osteoarthritis (OA),6 the breakdown of lubrication may eventually lead to complete degradation of the cartilage layers, causing acute pain and disability. Understanding the molecular mechanisms of the very efficient lubrication in healthy cartilage is thus a problem of considerable interest, and the synovial joint has been widely studied.7 However, although much is known about the articular cartilage bulk structure and composition,815 the detailed mechanism that provides the lubrication at the joint surfaces is still not well understood.5,16 This is the motivation of the present study. Cartilage is a porous tissue composed largely of water (6885% of total weight17); the extracellular matrix (ECM) is an entangled network of collagen fibers,5,17 elastin,18 hyaluronan (also known as hyaluronic acid (HA)),5,19,20 proteoglycans (PGs) (mostly aggrecans),5,17,2123 phospholipids, and other Received: April 11, 2011 Revised: June 20, 2011 Published: August 08, 2011 3432
dx.doi.org/10.1021/bm2004912 | Biomacromolecules 2011, 12, 3432–3443
Biomacromolecules
Figure 1. Detailed structure at the outer cartilage surface is thought to comprise charged macromolecules5 as in this schematic the major ones are hyaluronan (darker, thick curves, blue online), bottlebrush-like aggrecans (red online), and lubricins (lighter, thick curves, green online) extending from the surface to form a brush-like layer. Schematic adapted from Klein (2009);29 see Figure 3 for a more detailed aggrecan structure.
smaller proteins including superficial zone protein (SZP; also termed lubricin),5,8 cartilage link protein (LP), and fibronectin.5,17 Electron microscopy showed that the articular cartilage superficial zone is essentially acellular to a depth of order of one micrometer from its surface,5,24 significantly larger than the characteristic dimension (up to 400 nm)5 of the cartilage macromolecules that are accumulated both at the surface and in the SF.5,2527 These macromolecules are synthesized by the chondrocytes (cartilage cells) and (some of them) by the synoviocytes (the cells on the synovial membrane walls). Those synthesized within the cartilage permeate its space and eventually diffuse into the SF passing through the superficial zone (outer cartilage) surface.11,25,28 At some point in their passage through this interface, they are partially entrapped in the superficial zone and partially emanating from it into the synovial cavity in a brushlike configuration, as discussed in detail by Klein5 and as indicated schematically in Figure 1. Articulation of the cartilage layers is then mediated by these interfacially attached macromolecules (such as PGs, lubricin, and phospholipids) on each of the surfaces as they slide past each other under compression, and such macromolecular layers may play a major role in determining the sliding friction. For several years, HA was thought to be the molecule responsible for synovial joints lubrication because of its bulk viscoelastic behavior,30,31 but at the physiological shear rate in a flexing joint (105 s1 or higher5) its viscosity decreases to values comparable to the viscosity of water,5,32 losing its lubrication properties. Lubricin9,3338 (or a part of it39) and surface-active phospholipids (SAPLs)4043 have also both been proposed to be the major players in synovial joint lubrication, although measurements on lubricin layers as a boundary lubricant at a mica surface33 have shown that by itself it is not a particularly good lubricant at pressures higher than 10 atm. The question of what is the principle cause of the extremely efficient boundary lubrication in synovial joints remains open. Our overall aim is to gain insight into the role of these components in the lubrication mechanism of synovial joints by examining the effect of the different major cartilage macromolecules as well as smaller molecules in modulating interfacial
ARTICLE
friction. In the present study, we focus on aggrecan, a ubiquitous cartilage PG, which plays an important structural role, being highly ionized and thus leading to the high osmotic pressure that contributes to the special mechanical properties of cartilage (see below). Aggrecans are composed of a core protein of ∼2300 amino acids with a dense brush region of ∼100 covalently attached glycosaminoglycan (GAG) chains of chondroitin sulfate (CS) and keratan sulfate (KS) in a bottle-brush-like configuration, leading to a molecular weight of (2 to 3) 106Da (scheme in Figure 3).22,4447 CS is composed of ∼40 repeat disaccharides of D-glucuronic acid (GlcA) containing a carboxyl group and N-acetyl-D-galactosamine (GalNAc), which can be sulfated at the C-4 or C-6 position.15,22 When fully ionized, the GAG chains attribute to each aggrecan molecule a charge density of ∼9000 negative charges/aggrecan.46,48 This high density of fixed negative charges draws water into the tissue, yielding a high osmotic pressure responsible for more than 50% of the compressive modulus of the articular cartilage.17,22 Physiologically, aggrecans exist mostly as aggregates where an order of up to a hundred or even more aggrecan molecules are noncovalently attached to a single HA chain via the HA binding region, the globular domain called G1 at the N-terminus of the aggrecan core protein.22,4951 The aggrecan/HA aggregate is stabilized by cartilage link protein, which has the capacity to bind simultaneously to aggrecan and to HA.5153 Several studies showed that whereas aggrecan can interact with HA in the absence of cartilage link protein (via the HA binding region), those aggregates formed with link protein are more stable12,28,52,54,55 and form larger and denser aggregates;12,56 this is because cartilage link protein, aggrecan, and HA interact together in a cooperative manner, leading to the formation of a stable ternary complex. Raviv et al.57,58 have shown that brushes of charged synthetic polymers (polyelectrolytes) physiosorbed to opposing sliding surfaces in an aqueous medium result in extremely efficient lubrication, 57,58 although up to only ca. 0.3 MPa, or some 3 atm, much lower than in the major human joints. The origin of this behavior was attributed to the exceptional resistance to mutual interpenetration displayed by the compressed counterion-swollen brushes, together with the fluidity of the hydration layers surrounding the charged, rubbing polymer segments.57 This and other recent studies5,5961 (in some of which59 low friction was achieved up to physiologically high pressures of ∼7 MPa) emphasize the importance of hydration layers about charges in aqueous media as the basic lubrication element.5 In the light of these findings, we here investigate, using a surface force balance (SFB), how surface-attached, brush-like layers of aggrecan molecules extracted from articular cartilage attached to HA chains, emulating their physiological configuration, may modulate the friction compared with the behavior of HA chains alone. In this study, we were able for the first time to reconstruct on a mica surface a simplified version of the HAaggrecan complex at the superficial zone surface, mimicking the configuration at the interface of the most abundant PG in the cartilage. In this Article, we present the controlled protocols for immobilizing HA aggrecan complexes on a surface using avidinbiotin chemistry and examine the normal and shear interactions between such a surface and a molecularly smooth, negatively charged, mica test surface. The cartilage tissue is clearly not similar to an atomically smooth mica surface, but it has been shown5 that under physiological pressures the sliding cartilage surfaces are expected to deform to make molecularly intimate contact so that when sliding the interface is, locally, molecularly smooth. This first 3433
dx.doi.org/10.1021/bm2004912 |Biomacromolecules 2011, 12, 3432–3443
Biomacromolecules study thus establishes the baseline behavior for sliding of an HAaggrecan boundary complex against an opposing surface. (Future papers will describe interactions and sliding between two symmetric HAaggrecan-coated surfaces: whereas this is a more realistic scenario, it introduces the risk of artifacts due to avidinbiotin bridging interactions arising from the way the macromolecular complexes have been constructed, and we chose therefore to first study the simpler and thus more instructive system described here). Following the Materials and Methods section, the Results section describes systematic normal and shear force measurements at every stage of the surface-layer construction including the intermediate stages involving interactions between avidin-coated and avidin-HA coated surfaces, as well as the final avidinHAaggrecan-coated surface. The Discussion section interprets the normal profiles using a semiquantitative model to yield a picture of the aggrecanHA complexes on the surface, which is finally used to interpret the boundary lubrication mechanism of such layers.
’ MATERIALS AND METHODS Materials. Water for the SFB experiments was purified with a Barnstead water purification system (Barnstead NANOpure Diamond); it had resistivity of 18.2 MΩ and total organic content (TOC) < 1 ppb (so-called conductivity water). For the AFM imaging, Milli-Q Gradient A-10-system purified water (resistivity 18.2 MΩ, TOC e 3 ppb) was used. Ruby Muscovite mica grade 1 supplied by S & J Trading was utilized for both SFB experiments and AFM imaging. 3-Aminopropyltriethoxysilane (APTES, 440140), Avidin from egg white (A9275), and PBS (phosphate-buffered saline tablet, Tru-Measure chemical (P4417)) were supplied by Sigma Aldrich. MILLEX HV Duropore PVDF 0.45 μm membrane filters were supplied by Millipore. Guanidine hydrochloride (GuHCl), N-ethylmaleimide (NEM), and ethylenediaminetetraacetic acid disodium salt dehydrate (EDTA) were purchased from Fluka Chemical. ε-Amino-n-caproic acid (6-amino-n-hexanoic acid), benzamidine hydrochloride, hydrate, tris[hydroxymethyl]aninomethane (Trizma Base), cesium chloride (CsCl), dimethyl-methylene blue (DMMB), glycine, CS, Blue Dextran, cytochrome c from equine heart, β-amylase (M = 200 000 Da), Coomassie Brilliant Blue G 250, bromophenol blue, and sodium dodecyl sulfate (SDS) were from Sigma. Sodium acetate was from Acros Organics. Acetic acid and Tween-80 were from Riedel-de Haen. Sodium Chloride (NaCl), hydrochloric acid (HCl), precast polyacrylamide gradient gels Tris-HCL 420%, dithiothreitol (DTT), β-mercaptoethanol (2ME), and Precision Plus protein kaleidoscope standards were from Bio Lab. Bicinchoninic acid (BCA) protein assay reagent kit was from Pierce. Dialyzing tubes C.O. 12 00014 000 Da, 25 mm width were from Spectra/Por. Sephacryl S-200 high-resolution was from GE Healthcare B-Sciences AB. Bovine articular cartilage was from femoral heads of 1518 months old animals. Tissue was classified as normal according to the clinical diagnosis or was visually normal, and frozen at 20 °C until analyzed to keep its properties close to live tissue.62 Biotinylation of HA. Lightly biotinylated HA (bHA) was made essentially as described in ref 63. In brief, 5 mg of medical-grade HA (Genzyme; 0.5 to 1.5 MDa) was dissolved overnight in 0.1 M MES, pH 5.5 at a concentration of 5 mg/mL. To this was added 13 mL of 25 mg/mL EDAC (Pierce and Warriner, Chester, U.K.) in 0.1 M MES, pH 5.5 followed by 20 mL of 50 mM biotin-LC-hydrazide (Pierce and Warriner) in dimethyl sulfoxide, and the sample was mixed by rotation at room temperature overnight. The reaction mixture was dialyzed extensively against water, and particulate material was removed by centrifugation (12 000g for 1 min). The concentration of the bHA was
ARTICLE
determined using the metahydroxybiphenyl reaction64 relative to standards made from HA dried in vacuo over cobalt chloride. The bHA (in 0.02% (w/v) NaAzide) was stored at 4 °C. Extraction of Proteoglycans. Aggrecan molecules were extracted from cartilage tissue and characterized with atomic force microscope (AFM) before being introduced to the SFB. Articular cartilage was removed from the bone, sliced on a freezing microtome into 20 μm thick slices, and freeze-dried. Minced cartilage was extracted in 4 M GuHCl solution containing protease inhibitors for 48 h at 4 °C65 with the relation between the solution and tissue set to 1 g wet tissue to 10 mL of solution.66 Following the extraction, purification and fractionation of the aggrecan was carried out using well-established procedures.6570 (See the Supporting Information.) Isolation of Cartilage Link Protein. Link protein was isolated as previously described52 (Supporting Information) and characterized using sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDSPAGE). The identity of purified LP was confirmed by mass spectrometric analysis (LC-MS) of tryptic peptides (coverage 61%). (See the Supporting Information.) Atomic Force Microscopy. To characterize aggrecan molecules after the extraction from bovine cartilage, we used a DI NanoScope v6.13 apparatus following the procedure described by Ng et al.22 In brief, an atomically smooth freshly cleaved mica surface was functionalized with 0.1% aqueous APTES solution v/v to positively charge it. We deposed 50 μL of 0.3 μg/mL aggrecan aqueous solution onto the pretreated mica surface. Because aggrecan molecules are highly negatively charged because of the high concentration of GAG chains, they readily attach to a positively charged surface. After 2030 min of incubation, the samples were rinsed with Milli-Q water and allowed to dry in the laminar hood. Tapping mode imaging was performed (in air) in ambient temperature and humidity. An AC240 Olympus tip was used (force constant ∼2 N/m and resonant frequency ∼70 kHz, curvature radius K), the surfaces experience a jump into adhesive contact (due to Euler-like instabilities79) from a separation D = 80 ( 3 nm to a final distance of D = 6 ( 0.15 nm. The final D (6 ( 0.15 nm) fits well the dimensions of a single avidin molecule (4.5 nm 5 nm 5.5 nm82), indicating that a monolayer of avidin is adsorbed on the mica surface. From the best fit of the data (dashed line in Figure 4) using the full PB equation (eq 1) augmented with the vdW attraction (eq 2) for two asymmetric surfaces σ+ 6¼ | σ|,79 where | σ| is the effective bare mica surface charge density (σ = e/(30 nm2)), we extract the surface charge density of the avidin-coated surface σ+ = +e/(60 nm2). According to the fit, the counterions brought by the avidin molecules slightly increase the salt concentration to 2 105 M. As previously seen for chitosan-covered surfaces against bare mica,79 the attraction seen at large separations is many orders of magnitude larger than can result from vdW forces; therefore, it can be attributed as purely electrostatic in origin. Avidin overcompensates the charge on the mica surface, reversing it from net negative to net positive. As the surfaces approach each other, counterion pairs leave the gap between them to gain more entropy in the surrounding reservoir, leading to the long-range attraction.83 Avidin + bHA versus Bare Mica. During overnight incubation in the bHA solution, chains of HA attach to the avidin layer on the mica surface because of avidinbiotin interactions as well as electrostatic attraction arising from their opposing charges. Figure 5 shows typical F/R versus D profiles between an avidin + bHA-bearing surface and a bare mica surface. The normal profiles show no force until a separation D of ∼120 nm, where they experience a weak repulsion, followed by a rapidly increasing repulsive force to a final separation D = 9.8 ( 0.3 nm. This means that on compression the HA chains could be squeezed on the 3436
dx.doi.org/10.1021/bm2004912 |Biomacromolecules 2011, 12, 3432–3443
Biomacromolecules
Figure 4. Forcedistance profiles between two bare mica surfaces and a bare mica versus avidin-coated surface. Open symbols represent control Fn(D) profiles normalized by the radius of curvature R between two bare mica surfaces. Closed symbols represent the forces between a positively charged avidin-bearing surface and a negatively charged bare mica across water. The jump into adhesive contact from D = 80 ( 3 nm is indicated by arrows. The continuous (blue online) and the dashed lines are, respectively, fits to the PB equation (eq 1) augmented by the vdW expression (eq 2) for two bare mica surfaces (σmica = e/30 nm2, C = 105 M) and an avidin-bearing surface versus bare mica (σavidin = +e/ 60 nm2 vs σmica = e/30 nm2, C = 2 105 M). The top inset is a schematic of an SFB; in the underneath inset, the control force profiles between two bare mica surfaces are reported on a semilogarithmic scale. The inset in the bottom schematically shows a bare mica surface on top of an avidin-bearing one.
avidin layer, yielding a configuration where the overall avidin bHA layer was a few nanometers thicker than the avidin layer alone. This is consistent with results shown for other long polysaccharide chains adsorbed on mica84 and with the thickness of a single HA chain (∼0.3 nm).85 In some traces, we were able to recognize a few nanometers jump in from a distance D = 18 ( 4 nm (upper inset in Figure 5). Upon decompression, after coming into adhesive contact, the surfaces undergo a jump out to a distance of ∼1.7 ( 0.1 μm corresponding to a pull-off force Fpulloff = 0.4 ( 0.02 mN and an adhesion energy (W0 = 2Fpulloff/3πR) of ∼11 mJ/m2. The broken line in Figure 5 is the full PB equation + vdW attraction fit corresponding to two negatively charged surfaces, where the surface charge densities are, respectively, σ= e/300 nm2 for the avidin-bHA bearing surface and σ = e/30 nm2 for the bare mica surface. Taking into account the surface charge density of the bare mica covered by avidin molecules, we can deduce the charge density of the HA layer alone to be σHA = e/75 nm2. Avidin + bHA + Aggrecan + LP versus Bare Mica. Following the above Fn(D) measurements between the HA-coated surface and mica, an overnight incubation of this surface in a solution of aggrecan molecules and LP leads the formation of LP-stabilized
ARTICLE
Figure 5. Fn(D) profiles normalized by the radius of curvature R between an avidin + bHA-bearing surface and a bare mica across water. The blue continuous and the broken curves are, respectively, the PB equation fits (eq 1) augmented by the vdW expression (eq 2) for two bare mica surfaces (continuous blue line taken from Figure 4) and an avidin-bHA-bearing surface versus a bare mica (σavidin+bHA = e/ 300 nm2 versus σmica = e/30 nm2, C = 2 105 M). In the lower inset, the same profiles are reported on a semilogarithmic scale; the upper inset on an expanded scale shows profiles where a small jump into contact (arrows) is visible. In the lower inset is a schematic illustration of a bare mica surface on top of an avidin + bHA-bearing one.
aggrecanHA complexes. When a mica surface coated with aggrecan molecules (bound to HA chains with the help of the link-proteins) approaches a bare mica surface, repulsive forces start at a surface separations around 180 nm, a much higher value than is experienced with HA alone, and increase monotonically with the decreasing of D (Figure 6). On strong compression (up to ca. 18 atm), the surfaces reach a separation D = 12 ( 0.2 nm, suggesting a further 2 to 3 nm thickness of the layer due to the HA-attached aggrecan, beyond that arising from the avidin + bHA alone (9.8 ( 0.3 nm). SFB Surface Force Measurements: Shear Interactions. Avidin versus Bare Mica. Once the avidin-bearing surface and the bare mica surface jump into adhesive contact, the upper mica surface is made to move laterally back and forth by applying equal but opposite voltages to the opposing PZT sectors. Because of the strong mica + avidin versus mica adhesion arising from the electrostatic attraction, the static frictional force between the surfaces exceeds the maximal shear force (equal to KsΔx0,max) that we are able to apply, so that no sliding occurs and the surfaces are rigidly coupled during this applied lateral motion. Figure 7 shows typical traces between two such surfaces in water that illustrate this: trace A represents the back-and-forth lateral motion (ΔX0) applied by the PZT as a function of time, while trace B represents the shear force due to bending of the lateral springs: sliding is clearly not seen because of the strong adhesion. 3437
dx.doi.org/10.1021/bm2004912 |Biomacromolecules 2011, 12, 3432–3443
Biomacromolecules
Figure 6. Normal force profiles (normalized by the radius as Fn(D)/R) between an avidin + bHA + aggrecan + LP-bearing surface and mica across water. The continuous blue line is for two bare mica surfaces taken from Figure 4. The inset shows the same profiles on a semilogarithmic scale and a schematic illustration of the surface interaction configuration. Broken line: normal interactions from preliminary experiments at 0.1 M salt. (See the Discussion.)
Figure 7. Typical lateral force between an avidin-bearing surface and bare mica in conductivity water. Trace A is the shear motion applied to the top bare mica surface (Δx0 = 355 nm) at driving frequency ν0 = 0.5 Hz. Trace B is the shear response transmitted to the lateral springs after the surfaces jumped into adhesive contact. The inset is a cartoon illustrating a bare mica surface moving back and forth on top of the avidin-bearing one.
Avidin + bHA versus Bare Mica. Once HA chains are added to the bottom surface, there is little measurable shear force between them as they approach until the onset of detectable steric repulsion, at surface separations in the range D = 11 ( 2.8; that is, the HA layer acts as a good lubricant at low pressures. On stronger compression the friction increases sharply and, already at separations in the range D = 9.8 ( 0.3 nm, the surfaces are rigidly coupled (within the limit of the shear forces we can apply). Trace A in Figure 8 shows the back-and-forth motion of the upper surface, whereas the following traces show the shear force transmitted to the lateral springs at different pressures and surface separations D at a given contact point. The inset on the right shows the shear force traces as a function of frequency (derived from the fast Fourier transform (FFT) of the left-hand traces); the arrows indicate the drive frequency of our measurements (ν0 = 0.5 Hz) and thus enable detection of the onset of
ARTICLE
Figure 8. Typical lateral forces when bare mica is moved back and forth on top of an avidin + bHA-bearing surface in conductivity water (second approach). Trace A is the shear motion applied to the top mica surface (Δx0 = 380.6 nm). Traces BE are the shear responses transmitted to the lateral springs at different surface separations and different pressures. Right inset: shear traces as a function of frequency (FFT of the left-hand traces); the driving frequency ν0 = 0.5 Hz is indicated by arrows. On the bottom is a schematic illustration of a bare mica surface moving back and forth on top of the avidin + bHA bearing surface.
weak frictional forces, even when they are obscured by the noise in the direct Fs versus t traces. Avidin + bHA + Aggrecan + LP versus Bare Mica. A very different behavior is seen once aggrecan molecules together with link proteins are added to the existing HAavidin layers. In contrast with what is seen for the HA chains alone, the shear force increases only rather slowly up to pressure of P = 16 to 17 atm (traces BI in Figure 9). At this pressure, the surfaces are in the range of 12 ( 0.2 nm apart. Bearing in mind the aggrecan dimensions (∼250 nm for the molecule contour length and ∼20 nm for the GAG chains of CS, according to our AFM images (as in Figure 3)) and considering that they are on top of layers of HA and avidin, it is clear that the molecules are strongly compressed and flattened between the surfaces (and may be squeezed onto the surface in the interstitial regions between the avidin and the HA molecules) as the surfaces slide past each other. (See the Discussion.) The shear traces in Figure 9 demonstrate that if aggrecan is added to the HA layers, then the surfaces slip past each other at pressures well above those where they were rigidly coupled to HA alone (trace E in Figure 8). The friction coefficient remains as low as ca. 0.01 until a pressure of P = ca. 16 to 17 atm is reached, as shown in Figure 10, which summarizes all shear force Fs versus normal force Fn data both for the surfaces bearing avidin-HA alone and for those bearing the avidinHAaggrecan + LP layers. The slope of the two data plots gives the friction coefficient μ = (∂Fs/∂Fn) in the two different configurations. In both cases, μ is very low ( 16 atm, up to the highest pressures in this study, the friction coefficient increases to 0.06. Broken curve: based on results from preliminary experiments at 0.1 M salt. (See the Discussion.)
Negatively charged HA chains have been successfully attached to the negative mica surface with an avidinbiotin bond. In principle, some attachment of the bHA to avidin may occur via electrostatic (ionic) interactions alone, but the biotin on the HA chains ensures a much stronger bond with the avidin layer. Even though interactions between an avidin bearing surface and bare mica are not directly relevant for the understanding of synovial joint lubrication, they are an essential and clear control in the present work; long-range attractions, followed by a jump into adhesive contact (Figure 4), confirm that the system is clean and that a monolayer of avidin is physically adsorbed on one mica, allowing us to proceed with HA adsorption (Figure 5). Moreover, the fits of the force versus distance profiles to the PoissonBoltzmann equation (eq 1) enable us to evaluate the effective surface charge densities in different stages in the experiment and from that to obtain a more quantitative picture of the surface layers, as below. When avidin is attached to the mica, it overcompensates its negative charge density σmica = e/30 nm2 to σmica+avidin = +e/ 60 nm2 (Figure 4). Likewise, when bHA is added, the negative 3439
dx.doi.org/10.1021/bm2004912 |Biomacromolecules 2011, 12, 3432–3443
Biomacromolecules
ARTICLE
pressure of steric origin (i.e., excluded volume effects) exerted by the confined monomers (Πmon). Π ¼ Πc=i þ Πmon
ð4Þ
where Πc=i ≈ nKB T
ð5Þ
and (from the mean field expression for osmotic pressure of polymer solutions88) Πmon ≈ Figure 11. Normal force interactions Fn(D) normalized by the radius of curvature between one surface bearing avidin + bHA + aggrecan + LP against mica (open symbols) and the same surface bearing only avidin + bHA against bare mica (closed symbols).
charge on the chains slightly overcompensates the positive charge of the avidin underneath, leading to σmica+avidin+bHA ≈ e/300 nm2) (Figure 5). This results in the weak long-range repulsion experienced before steric repulsion appears as the HA chains contact the opposing mica surface. From this, we may try to estimate the adsorbance of the HA on the surface, as follows: for a molecular weight of the HA = MHA = 106 Da, the number of disaccharide units is nds = (MHA/mds), where mds ≈ 380 is the disaccharide molecular weight. It is a good assumption that the disaccharide units are singly charged (pKa ≈ 3)86 so that for the positive charge (arising from the avidin on the mica) to be just compensated by the HA, ignoring the slight overcompensation, the mean area occupied by a single HA molecule is AHA = (e 3 nds/ σmica+avidin) ≈ 1.6 105 nm2. This estimate of AHA will be useful later in assessing the structure of the HAaggrecan complex. In line with previous studies,87 we found (Figure 5) that HA chains have some adhesive interaction when compressed against a bare mica surface. However, the adhesion energy value we estimated from the pull-off force (∼11 mN/m2) is lower than the one reported in Benz et al.,87 probably because of the differences in the way the HA layer is constructed. We may offer two possible scenarios for the origin of such an attraction: on the one hand, it might be that at small surface separations reached under compression the van der Waals (vdW) attraction exceeds the osmotic repulsion exerted by the trapped counterions, resulting in net attraction; on the other hand, even if the overall charge of the avidin + bHA layers is very slightly negative (as indicated by the estimate of the effective charge density, Figure 5) locally, at the bond between the two molecules, there is a residual net positive charge on the avidin that results in adhesion to the negatively charged mica. In accordance with previous SFB works on HA facing bare mica,87 our results show that HA alone does not behave as a particularly good lubricant. Once aggrecan molecules are attached (stabilized by LP) to the HA layer, the repulsion between the surfaces starts at much larger separations (D around 180 nm) and increases monotonically (Figure 6). In Figure 11, the normalized force profiles Fn(D)/R versus surface separation D, in the presence and in the absence of aggrecan molecules, are compared to emphasize this increase in the repulsion (osmotic and steric). We now analyze the interactions between the aggrecan layer and the opposing surface. The pressure exerted by aggrecan molecules may be viewed as the sum of two terms: the osmotic pressure exerted by the trapped counterions (Πc/i) and the
ϕ2 KB T v
ð6Þ
where the monomer volume fraction ϕ (volume of monomers/ unit volume) is given by ϕ¼
Nm v D
Here v is the volume of one monomer and Nm is the total number of monomers per unit area. The parameter n is the number of counterions per unit volume, and it is equal to fNm/D, where f is the charged polymer fraction. For simplicity, we have approximated our aggrecan molecules as composed only of CS disaccharides each having v ≈ 100 Å3. Bearing in mind that each disaccharide may have one to three charges when fully ionized48 and that the Bjerrum length (∼0.7 nm) is similar to the length of one disaccharide (∼1 nm), we take f = 1. Integrating the total osmotic pressure between L, the separation at which (from Figure 6) steric interactions begin to appear (L ≈ 60 nm), and D, the surface separation (D e L), we obtain the interaction energy E per unit area, which, within the Derjaguin approximation (D , R), is equal to (from eqs 46, after a little algebra) Z L F ¼E¼ ΠðD0 Þ dD0 2πR D Z L ¼ nKB T 3 ð1 þ ϕðD0 ÞÞ dD0 ð7Þ D
and therefore ∂ FðDÞ ¼ 2π 3 ΠðDÞ ¼ 2π½nKB Tðϕ þ 1Þ ∂D R
ð8Þ
Extrapolating the derivative of F(D)/R from the experimental data, and knowing all other parameters in eq 8, we may estimate the total number of monomers Nm per unit area as Nm ≈ 0.3 monomers/nm2. (A more direct approach for estimating the adsorbance, for example by biochemical analysis of the adsorbed material, might also be feasible but would require adsorption on much larger mica areas to provide sufficient material for analysis.) From Figure 6, we see that steric interactions commence at distance L around 5060 nm. We may therefore consider our aggrecan molecules of length lagg ≈ 250 nm to be tilted on the surface, forming an angle of
