Substrate chemistry influences the morphology and biological function of adsorbed extracellular matrix assemblies

ARTICLE IN PRESS

Biomaterials 26 (2005) 7192–7206 www.elsevier.com/locate/biomaterials

Substrate chemistry influences the morphology and biological function of adsorbed extracellular matrix assemblies Michael J. Sherratta,, Daniel V. Baxb, Shazia S. Chaudhryc, Nigel Hodsonc, Jian R. Lud, Priya Saravanapavand, Cay M. Kieltyb,c a

Division of Laboratory and Regenerative Medicine, The Medical School, Stopford Building, The University of Manchester, Oxford Road, Manchester, M13 9PT, UK b United Kingdom Centre for Tissue Engineering, The University of Manchester, Michael Smith Building, Oxford Road, Manchester, M13 9PT, UK c Wellcome Trust Centre for Cell-Matrix Research, Faculty of Life Sciences, The University of Manchester, Michael Smith Building, Oxford Road, Manchester, M13 9PT, UK d Biological Physics Group, School of Physics and Astronomy, The University of Manchester, Sackville Street Building, Sackville Street, Manchester, M60 1QD, UK Received 21 January 2005; accepted 6 May 2005 Available online 20 June 2005

Abstract In addition to mediating cell signalling events, native extracellular matrix (ECM) assemblies interact with other ECM components, act as reservoirs for soluble signalling molecules and perform structural roles. The potential of native ECM assemblies in the manufacture of biomimetic materials has not been fully exploited due, in part, to the effects of substrate interactions on their morphology. We have previously demonstrated that the ECM components, fibrillin and type VI collagen microfibrils, exhibit substrate dependent morphologies on chemically and topographically variable heterogeneous surfaces. Using both cleaning and coating approaches on silicon wafers and glass coverslips we have produced chemically homogeneous, topographically similar substrates which cover a large amphiphilic range. Extremes of substrate amphiphilicity induced morphological changes in periodicity, curvature and lateral spreading which may mask binding sites or disrupt domain structure. Biological functionality, as assayed by the ability to support cell spreading, was significantly reduced for fibrillin microfibrils adsorbed on highly hydrophilic substrates (contact angle 20.71) compared with less hydrophilic (contact angle 38.31) and hydrophobic (contact angle 92.81) substrates. With an appropriate choice of surface chemistry, multifunctional ECM assemblies retain their native morphology and biological functionality. r 2005 Elsevier Ltd. All rights reserved. Keywords: Extracellular matrix (ECM); Atomic force microscopy (AFM); Surface energy; Surface roughness; Protein adsorption; Cell spreading

1. Introduction Many synthetic polymers used in tissue engineering applications are non-charged or weakly charged. UV irradiation to increase the surface energy of materials such as expanded polytetrafluoroethylene (ePTFE) and polyurethane in an NH3 atmosphere has been employed Corresponding author. Tel.: +44 161 275 7054; fax: +44 161 275 5082. E-mail address: [email protected] (M.J. Sherratt).

0142-9612/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2005.05.010

to increase cellular adhesion [1]. Biodegradable polyesters such as poly(lactic-co-glycolic acid) (PLGA), which degrade to non-toxic lactic and glycolic acids, have also been widely used as tissue engineering scaffolds [2,3]. As with ePTFE, the cell adhesion properties of unmodified hydrophobic PLGA and related polylactide surfaces are poor and the physicochemical introduction of hydrophilic groups and the incorporation of cell adhesion peptides have been used to enhance cell attachment [4,5]. An alternative approach to enhancing cellular adhesion to synthetic substrates involves the design and manufacture of biomimetic materials in which adsorbed

ARTICLE IN PRESS M.J. Sherratt et al. / Biomaterials 26 (2005) 7192–7206

native extracellular matrix (ECM) proteins or ECM derived peptides influence cellular adhesion and phenotype [6]. Cellular adhesion to fibronectin and other ECM proteins, such as the macro-molecular assemblies formed by fibrillin and type VI collagen, is mediated by integrins which bind short ECM peptide sequences such as RGD (Arg-Gly-Asp) regulating cell adhesion, migration proliferation and secretory responses [7]. It has been proposed that the enhanced cellular adhesion observed on physicochemically modified hydrophilic PLGA surfaces may be mediated by increased adsorption of cell-binding ECM serum proteins such as fibronectin and vitronectin [4]. The ease of synthesis of short cellular recognition site peptide sequences has encouraged the study of synthetic peptides as potential biomimetic biomaterials [6]. Short peptide sequences however lack the diverse functionality of native ECM components which in vivo act as a reservoirs for soluble signalling molecules, bind other ECM proteins and perform important structural roles [8]. A major barrier to the use of native ECM proteins in the construction of biomimetic materials is substrate induced random folding which may sterically hinder binding events [6]. Regulation of cell phenotype in vitro is critically dependent on the establishment of appropriate cell to adsorbed ECM protein interactions. In turn these interactions require the preservation of native protein morphology which is dependent on the substrate [9–11]. Due to their amphiphilic nature proteins adsorb to most surfaces over a wide range of solution conditions [12]. Structural re-arrangements, which play an important role in the adsorption process, are known to influence the biological function of native ECM proteins, such as fibronectin and synthetic cell-adhesion peptides [6,13–15]. Steered molecular dynamic studies suggest that the substrate specific integrin binding of fibronectin may be a consequence of the relative movement of domains and the physical separation of the RGD (Arg-Gly-Asp) cellular recognition site and the adjacent synergy site [16]. In addition to changes in intra and inter-domain spacing (alterations in quaternary and tertiary structure), atomic force microscopy (AFM) force spectroscopy and circular dichroism (CD) studies suggest that the forces generated during protein/ substrate interactions may be of sufficient magnitude to disrupt protein secondary structure [17,18]. Fibrillin and type VI collagen are modular ECM proteins which form structurally complex supra-molecular assemblies known as microfibrils. These naturally occurring multifunctional ECM components have the potential to produce highly effective biomimetic materials supporting cell attachment/spreading and linking cells to other ECM components [19–22]. We have previously demonstrated by AFM that fibrillin and type VI collagen microfibrils exhibit substrate dependent morphologies on surfaces which vary in chemistry,

7193

topography and the degree of chemical heterogeneity [11]. The chemical modification of substrates with thin polymer films or self-assembled monolayers provides a way of manufacturing topographically similar substrates which differ only in surface chemistry. Variations in surface chemistry and topography may influence adsorbed protein morphology and are known to affect contact angle estimates of substrate wettability [23]. Coating silicon oxide (SO) wafers with PLGA or octadecyltrimethoxysilane (OTS) produces chemically distinct homogeneous substrates with a wide amphiphilic range [2,24]. Similar substrate amphiphilicities may be induced on glass coverslips using combinations of alcohol/acid washes and vapour silanisation [25,26]. We have employed atomic force microscopy to investigate the effects of surface chemistry on fibrillin and type VI collagen microfibrils adsorbed on chemically homogeneous, topographically similar silicon wafer and glass substrates. The effect of substrate induced conformational changes on biologically important interactions was determined by cell spreading assays on fibrillin microfibril ligands adsorbed to chemically modified transparent glass coverslips.

2. Experimental section 2.1. Materials and reagents Second trimester foetal calves and adult bovine eyes were obtained from the local abattoir within 1 h of death. Bacterial collagenase (type 1A), bovine serum albumin (BSA), phenylmethylsulfonyl fluoride (PMSF), N-ethylmaleimide (NEM), 10 nm gold colloids, poly-L-lysine (PLL), dichloromethane, hexadecane and Sigmacotes (chlorinated organopolysiloxane in heptane) were obtained from Sigma-Aldrich (Poole, Dorset, UK). Poly(DL-lactide-co-glycolide) (PLGA 50:50) with an average molecular weight of 50,000–75,000 and glass transition point of 45–50 1C and octadecyltrimethoxysilane (OTS) were also purchased from Sigma-Aldrich and used as supplied. Sepharose CL-2B was supplied by Amersham Biosciences UK Ltd. (Little Chalfont, Buckinghamshire, UK). Primary human dermal fibroblasts (HDF) were obtained from Cascade Biologics Inc. (Mansfield, Nottinghamshire, UK). DMEM/ HEPES cell culture medium was obtained from Invitrogen Ltd. (Paisley, Renfrewshire, UK). Neutral Decon was obtained from Decon laboratories (Hove, East Sussex, UK). All other reagents were of analytical grade. Metal support stubs and Olympus high aspect ratio etched silicon probes (spring constant of 42 N m
1) were obtained from Veeco Instruments (Santa Barbara, California, USA). Non-doped native silicon oxide wafers with /1 1 1S orientation were obtained from Compart Technology Ltd (Peterborough, Cambridgeshire, UK) and were used as supplied. Borosilicate glass coverslips (13 mm diameter) were obtained from Scientific Laboratory Supplies (Nottingham, Nottinghamshire, UK). Muscovite mica was obtained from Agar Scientific

ARTICLE IN PRESS 7194

M.J. Sherratt et al. / Biomaterials 26 (2005) 7192–7206

(Stansted, Essex, UK). Microtiter plates were obtained from Costar Group (Bethesda, Maryland, USA).

measurements were carried out at ambient laboratory relative humidity (3075%), in triplicate.

2.2. Substrate preparation

2.4. Microfibril isolation

Silicon oxide wafers were cut into 1  2 cm pieces and rinsed with 3–5% Neutral Decon and ultrapure water prior to PLGA or OTS coating. Thin PLGA films were coated using a dipcoating rig as previously described [27]. The dip-coating rig was placed on top of an anti-vibration table and the entire system was enclosed in a Perspex box to avoid any interference caused by air flow. PLGA was dissolved in dichloromethane to make a 0.15 wt% solution. Wafer cuts were lifted out of the solution at a speed of 2 mm s
1. The coated films were annealed at 100 1C for 2 h to remove residual solvent and release any strains. OTS was dissolved in a mixed solution of dichloromethane and hexadecane in a volume ratio of 1:1 to make up a 1 mM solution [28]. Dry wafer cuts were immersed in the solution for 1 h prior to rinsing with dichloromethane, ethanol and distilled water. Coated films were stored in either a desiccator or a sealed plastic. Glass coverslips (G) were cleaned by two methods. G/MHS (glass/methanol, HCl, H2SO4) coverslips were prepared by a modification of method 2 reported by Cras et al. [25]. Following initial washing in a 1:1 solution of 100% methanol:concentrated HCl for 30 min, coverslips were briefly washed in ultrapure water before being placed in concentrated H2SO4 for a further 30 min. After this time the coverslips were rigorously washed in ultrapure water and dried at 75 1C. G/E (glass/ethanol) coverslips were placed in 96% ethanol for 30 min and then removed and left to air-dry. Silanised glass coverslips (G/ES; glass/ethanol, Sigmacotes) were cleaned by the ethanol method prior to vapour phase deposition of a silanising agent [26]. Following the ethanol wash, coverslips were dried at 75 1C before being placed on a raised surface in a glass jar containing 5 ml of Sigmacotes. After 15 min exposure, the coverslips were washed in 96% ethanol before being left to air dry at room temperature. All glass coverslips were used for microfibril and cell spreading assays within 1–2 h of preparation.

The vasculature is a rich source of both fibrillin and type VI collagen microfibrils but the ciliary zonules of the eye contain fibrillin microfibrils only [19,22]. Microfibrillar species were isolated from foetal bovine aorta and adult bovine ciliary zonules in native non-denaturing conditions by bacterial collagenase digestion and size exclusion chromatography using modifications of a previously described methodology [11,29]. Briefly, tissue was incubated with 0.5 mg ml
1 collagenase, freshly prepared protease inhibitors (2 mM PMSF, 5 mM NEM) and 10 mM CaCl2 in column buffer (400 mM NaCl, 50 mM Tris–HCl, pH 7.4) for 18 h at 4 1C. Samples were subsequently centrifuged at 5000g for 5 min, and the supernatant was size fractionated on a Sepharose CL-2B column in column buffer. The excluded volume (Vo) of aorta digests contained abundant fibrillin and type VI collagen microfibrils and the Vo of ciliary zonule digests contained abundant fibrillin microfibrils.

2.3. Substrate characterisation Root mean square (RMS) roughness of silicon wafers (SO, SO/PLGA and SO/OTS) and glass coverslips (G/MHS, G/E and G/ES) was determined as described previously [11]. Briefly, intermittent contact mode AFM height images (2 mm) were captured, first order flattened and exported as ASCII files. RMS roughness values (Rq) were calculated for 300 109  109 nm boxes (28  28 pixels) on each 2 mm scan using routines written in Microsoft Visual Basic 6.0. Advancing and receding contact angles were determined for glass coverslips (G/MHS, G/E and G/ES) as described previously [11]. Briefly, advancing contact angles were quantified by the incremental addition of 5 ml drops of ultrapure water onto each substrate to a final volume of 40 ml. Receding contact angles were quantified by the sequential removal of 5 ml drops. Images horizontal to the surface were captured at each increment and the contact angle calculated from the drop height and width. All contact angle

2.5. AFM sample preparation and imaging Trimmed silicon wafers (SO, SO/PLGA and SO/OTS) and glass coverslips (G/MHS, G/E and G/ES) were adhered to 15 mm diameter metal sample stubs using clear nail varnish. Aliquots of microfibril-containing Vo were diluted 1:8 with column buffer. A sufficient volume of the diluted sample was pipetted onto the substrate and incubated for 1 min prior to three consecutive washes with 300 ml distilled H2O. Excess liquid was removed by capillary action and the prepared samples were allowed to air-dry overnight in a dust-free environment prior to microscopical analysis. Samples were imaged by intermittent contact mode in air using a Multimode AFM with a Nanoscope IIIa controller and an E scanner. Cantilever oscillation frequencies and drive amplitudes were determined by the Nanoscope software. Height images were captured at a scan rate of 1.49 Hz and scan sizes of 5 or 2 mm. The setpoint was adjusted to just below the point at which tipsample interaction was lost. All height images were first order flattened using the Nanoscope software. The instrument was calibrated periodically using a grating with 180 nm deep, 10  10 mm depressions. All images were captured at a relative humidity of 3075%. 2.6. Microfibril adsorption The relative affinity of fibrillin and type VI collagen microfibrils for each substrate following a 1 min incubation was estimated from AFM height images. Fifteen 5 mm scans, captured at random positions on each substrate, were subdivided into 25 1 mm regions. The number of microfibril repeats in each region was counted manually for each image. The number of microfibril repeats per mm2 was calculated for a total area of 375 mm2. The binding affinities determined for fibrillin microfibrils after a 1 min incubation on glass substrates may be different to the relative microfibril coverage after a 1 h incubation in preparation for cell spreading assays.

ARTICLE IN PRESS M.J. Sherratt et al. / Biomaterials 26 (2005) 7192–7206

2.7. Periodicity and flexion angle determination

7195

measuring the flexion angle defined by each set of three adjacent beads (Fig. 1a) [30]. Type VI collagen microfibril periodicities were calculated from axial height profiles extracted along the axial contour (Fig. 1a). The double bead centre was defined as the height minimum between two bead height maxima. Periodicities were determined for 400 repeats from at least 20 microfibrils. Microfibril curvature was determined by measuring the flexion angle between three adjacent double-bead centres, repeat curvature, measured between double bead centres and the centre of the interbead, quantified bending within a single repeat due to deformations of the beads and/or interbead (Fig. 1a).

Fibrillin microfibril periodicities were calculated from 1 pixel wide (3.9 nm on 2 mm scans) bead centre to bead centre axial height profiles generated by WSxM 4.0 develop 4.0 (Nanotec Electronica, available on the internet at http:// www.nanotec.es/) (Fig. 1a). Height maxima, corresponding to bead centres were located on axial height profiles using a program developed in Microsoft Visual Basic 6.0 [21]. Periodicities were determined for a total of 1000 repeats from at least 20 microfibrils adsorbed on each substrate. Microfibril curvature was determined for the same population by

Type VI collagen

Transverse height profile

Repeat curvature

Microfibril curvature

Periodicity

Fibrillin

B

IB

B

IB

(a) 12 Gaussian

R H

Height (nm)

10

Height profile

8 6 4 2

(b)

Wt

0

Wm

-2 0 (c)

20 40 60 80 Transverse distance (nm)

100

Fig. 1. Microfibril periodicity/flexion angle and tip deconvolution. (a) Periodicities of fibrillin microfibrils were measured between adjacent bead centres. Periodicities of type VI collagen microfibrils were measured between double bead centres along the axial contour. Microfibril curvature was quantified from the angle formed between three adjacent beads (fibrillin) or double bead (type VI collagen) centres. Type VI collagen repeat curvature was quantified from the angle formed between adjacent double bead centres and the centre of the interbead. Mean transverse height profiles (THP) were calculated for fibrillin and type VI collagen beads (B) and interbeads (IB). (b) Simple geometric model illustrating the relationship between the true width (W t ) and the measured width at half maximum height (Wm) of a spherical particle of height H scanned by a tip with a radius of curvature (R) [33,34]. (c) Mean, fast scan direction, height profile of 10 nm colloidal gold particles imaged with one tip. Width at half maximum height (Wm) was calculated from a unimodal Gaussian fitted to the profile data. The radius of curvature (R) of each tip was calculated from Wm and Wt (where Wt was equivalent to the height of the gold particle) using Eq. (2). Error bars ¼ SEM.

ARTICLE IN PRESS M.J. Sherratt et al. / Biomaterials 26 (2005) 7192–7206

7196

2.8. Height maps and transverse height profiles Flattened, 2 mm height images were exported as ASCII height data files. For AFM images of isolated, disperse, structures adsorbed on a flat surface such as mica, the height of the surface (S) is generally nonzero. The value of S, determined from the mode of the height distribution was subtracted from points in the scan using routines written in Microsoft Visual Basic 6.0 [31]. Surface height corrected scans were exported in raw format for import into the public domain image analysis program ImageJ (National Institutes of Health, available on the internet at http://rsb.info.nih.gov/ij/). Curved microfibril regions were straightened using ImageJ and the straighten plugin which fits a non-uniform cubic spline to user supplied points [11,32]. Mean height maps were calculated for 50 or 20 repeats (fibrillin and type VI collagen respectively) with a transverse distance of 160.2 nm and an axial extent equal to the mode of the periodicity on each substrate. Bead and interbead transverse height profiles (THP) were extracted from the height maps (Fig. 1a). 2.9. Width Lateral dimensions determined by AFM are overestimated due to the finite geometry of the tip [33]. The true width ðW t Þ of a sphere or cylindrical polymer imaged by a spherical AFM tip may be calculated from the measured width using Eq. (1) [34] (Fig. 1b). sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi W2 Wm ¼ 2  R  Wt þ t (1) 4 where Wm is the measured width at half maximum height and R is the radius of curvature of the tip. Previous studies have calculated Wt from measurements of Wm for biological polymers using estimates of R [35]. Conversely R may be calculated for a particular tip when W t is known and Wm is measured from an AFM height image (Eq. (2)). R¼

ðW m =2Þ2
W 2t =4 Wt

(2)

Small DNA fragments and amyloid-b 1-40 fibrils and protofibrils have been used to estimate R for commercial Si3N4 cantilevers and carbon nanotube modified Si cantilevers respectively [33,36]. The use of biological polymers to characterise AFM tip geometry may be limited by substrate induced polymer deformations and tip/polymer interactions [33]. In contrast colloidal gold particles are non-deformable and available in a range of sizes appropriate for the characterisation of R [37]. In this study, 10 nm diameter gold colloids were adsorbed to mica coated PLL [11]. Intermittent contact mode AFM height images (2 mm scans) were captured for three separate tips. Central height profiles (in both the fast and slow scan directions) were measured for 10 particles for each tip. Fig. 1c shows the mean fast scan height profile for one tip through 10 particles. Superimposition of fast and slow scan height profiles suggested that the tips were symmetrical (data not shown). For microfibrils adsorbed on SO/PLGA and SO/ OTS the THP closely approximated a Gaussian distribution and the measured width at half maximum height (Wm) was

calculated from unimodal Gaussians fitted to mean THP (SigmaPlot 8.0 [SPSS Science, Chicago, IL, USA]). For spherical colloidal gold particles of maximum height (H), W t  H and R may be calculated from Eq. (2). The mean radius of curvature for the three tips was 7.94 nm (SEM 1.06 nm) which was comparable to the manufacturers quoted value of o10 nm. Measured microfibril bead and interbead widths at half maximum height (Wm) were calculated from unimodal Gaussians fitted to the THP in SigmaPlot 8.0. Assuming that adsorbed microfibrils are approximately cylindrical in cross-section; Eq. (1) was solved numerically for W t using the calculated values for R and Wm. Fitting a Gaussian to the THP of microfibrils adsorbed on SO was not possible due to the extensive lateral spreading. Compared to the calculated dimensions of the AFM tips used in this study (7.94 nm) the plateau region was low in height (0.5–0.6 nm) with an edge located 45 nm from the microfibril axis. It was unlikely therefore that the measured dimensions of the plateau region were due to tip/sample convolution effects. Microfibril widths were calculated as the region of the microfibril in the THP which was X0.2 nm (Rq value for these substrates) in height for all substrates and corrected widths (W t ) were calculated from mean bead THP of microfibrils adsorbed on SO/PLGA and SO/OTS substrates. 2.10. Cell spreading Cell spreading on fibrillin microfibrils adsorbed to chemically distinct substrates was quantified as previously described [20]. Briefly, purified fibrillin microfibrils were diluted in column buffer and allowed to adsorb to freshly prepared glass coverslips (G/MHS, G/E and G/ES) placed in the wells of 24 well microtiter plates. Following a 1 h incubation at room temperature, the buffer and unbound ligand was aspirated from the wells and non-specific binding was blocked by a 1 h incubation with heat denatured BSA. As negative controls, coverslips were also blocked with BSA in the absence of fibrillin microfibrils. Trypsinized HDF, adjusted to 2  105 cells ml
1 with warm DMEM/HEPES buffer, were added to the microfibril coated or BSA controls and incubated for 40 min at 37 1C in a CO2 incubator. Following fixation with 37% formaldehyde, cell spreading was quantified by phase contrast microscopy. Phase dark cells, and those with distinct phase dark projections were counted as spread, and phase bright cells were counted as un-spread. Three fields of view were counted for each data point.

3. Results 3.1. Substrate characterisation: SO, SO/PLGA and SO/ OTS Similar RMS roughness (Rq) values were determined from AFM height images of SO wafers and SO wafers dip-coated with PLGA or OTS (Figs. 2a and b). Uncoated SO wafers had an Rq value of 0.201 nm (SD ¼ 0.054 nm), SO/PLGA (Rq ¼ 0.215 nm, SD ¼ 0.032 nm) and SO/OTS (Rq ¼ 0.150 nm, SD ¼ 0.050 nm)

ARTICLE IN PRESS M.J. Sherratt et al. / Biomaterials 26 (2005) 7192–7206

Height (nm)

SO

1

1

0

0

0

-1

-1

-1

200 400 600 800 1000

0

(a)

200 400 600 800 1000 Axial distance (nm)

200 400 600 800 1000

120

Contact angle (
°)

0.30 0.20 0.10

100 80 60 40

Substrate

(c)

SO/OTS

SO/PLGA

SO

glass

mica-PLL

mica

0

SO/OTS

SO/PLGA

SO

glass

mica-PLL

20 0.00

mica

RMS roughness (Rq)

0

140

0.40

(b)

SO/OTS

SO/PLGA

1

0

7197

Substrate

Fig. 2. Substrate characterisation: SO, SO/PLGA and SO/OTS. (a) AFM height images and 1 mm height profiles of SO, SO/PLGA and SO/OTS substrates; 1 mm extracted regions from 2 mm scans, Z-scale 5 nm. (b) RMS roughness (Rq) values measured from AFM height scans of mica, micaPLL and glass substrates [11] and SO substrates. (c) Contact angle measurements. Sessile drop column buffer contact angle measurements for mica, mica-PLL and glass [11]. Advancing water contact angles for SO, SO/PLGA and SO/OTS substrates [3,24]. Error bars ¼ SD.

(Fig. 2b). There were significant but small differences between the Rq values for each substrate (Mann–Whitney U-test; po0:001). Previous studies have demonstrated that uncoated SO wafers are hydrophilic with advancing water contact angles of ðyA Þ of 181 (721) (Fig. 2c) [24]. PLGA coated substrates are less hydrophilic; Croll and co-workers reported values for yA of 74.91 (70.61) on PLGA coated glass coverslips [3]. To our knowledge, there are no reported yA values for PLGA coated SO, however substrate amphiphilicity had no effect on yA measurements following PLGA coating of hydrophilic ðyA ¼ 81:2 2:01Þ or hydrophobic ðyA ¼ 81:4 1:81Þ glass coverslips [2]. Silylated surfaces are hydrophobic and do not show significant water adsorption with reported yA values on SO coated with an OTS monolayer in the range 111–1151 (731) [24,38,39]. 3.2. Microfibril adsorption: SO, SO/PLGA and SO/ OTS Abundant fibrillin and type VI collagen microfibrils were identified by their unique beaded morphologies on

all three substrates (Fig. 3). The relative affinity of fibrillin and type VI collagen microfibrils for each substrate was determined by AFM (Table 1). There was no significant difference (Mann–Whitney U-test) in the amount of fibrillin microfibrils adsorbed on SO, SO/ PLGA and SO/OTS substrates. With increasing substrate hydrophobicity there was a small but significant decrease in the amount of adsorbed type VI collagen microfibrils. Significantly fewer microfibrils adsorbed to SO/OTS compared to SO (Mann–Whitney U-test; po0:01) and SO/PLGA (Mann–Whitney U-test; po0:05). 3.3. Periodicity and flexion angle: SO, SO/PLGA and SO/OTS Mean fibrillin microfibril periodicity was significantly reduced on SO substrates (46.3 nm) compared with SO/ PLGA (58.4 nm) and SO/OTS (58.3 nm) (Student’s t-test; po0:001) (Table 1). There was no significant difference between fibrillin microfibril periodicity on SO/PLGA and SO/OTS substrates. Mean fibrillin microfibril flexion angle was also significantly reduced

ARTICLE IN PRESS 7198

M.J. Sherratt et al. / Biomaterials 26 (2005) 7192–7206

between fibrillin microfibril flexion angles on SO/PLGA and SO/OTS substrates. Although mean periodicity and flexion angles were reduced for the whole population of microfibrils adsorbed on SO, within individual microfibrils distinct regions of further reduced periodicity and flexion angle were often co-localised at the microfibril termini (Figs. 4a–c). Within the terminal regions of the individual microfibril depicted in Fig. 4a, mean periodicity and flexion angle were 35.7 nm and 139.61 respectively. Within the central region mean periodicity was 50.9 nm and flexion angle 160.01. Mean type VI collagen microfibril periodicity was significantly reduced on SO substrates (102.4 nm) compared with SO/PLGA (107.5 nm) and SO/OTS (105.6 nm) (Student’s t-test; po0:001) (Table 1). There was no significant difference between type VI collagen microfibril periodicity on SO/PLGA and SO/OTS substrates. Microfibril curvature was unaffected by substrate chemistry, but there were significant differences in repeat curvature (Mann–Whitney U-test) (Table 1). Mean repeat curvature decreased significantly as substrate hydrophilicity decreased; SO (165.71), SO/PLGA (161.31) and SO/OTS (156.71) (Table 1) (Figs. 4d and e).

3.4. Width: SO, SO/PLGA and SO/OTS

Fig. 3. Microfibril adsorption: SO, SO/PLGA and SO/OTS. AFM height images of foetal bovine aorta fibrillin (F) and type VI collagen (VI) microfibrils adsorbed on SO, SO/PLGA and SO/OTS substrates. Scan size 2 mm, Z-scale 15 nm.

on SO substrates (151.11) compared with SO/PLGA (163.51) and SO/OTS (161.01) (Mann–Whitney U-test; po0:001) (Table 1). There was no significant difference

Microfibrils adsorbed on hydrophilic SO had a diffuse appearance compared to microfibrils adsorbed on less hydrophilic SO/PLGA and hydrophobic SO/OTS substrates (Figs. 5a and b). These observations, although more pronounced, are in agreement with previous observations of substrate amphiphilicity-dependent morphologies on mica, mica coated with poly-L-lysine and borosilicate glass [11]. Fibrillin and type VI collagen microfibril width was substrate dependent and decreased significantly (Mann–Whitney U-test; p40:001) with increasing substrate hydrophobicity (Figs. 6a–c). Mean bead fibrillin microfibril width was 91.3 nm (SEM 2.9 nm) on SO, 81.2 nm (SEM 2.9 nm) on SO/PLGA and 49.2 nm (SEM 2.1 nm) on SO/OTS (Fig. 6c). Correcting for tip radius, W t was 16.7 nm on SO/PLGA and 10.6 nm on SO/OTS (Fig. 6c). Laterally spread material from fibrillin microfibrils adsorbed on SO formed a plateau region of uniform height (0.5–0.6 nm) which extended along the whole repeat (Figs. 6a and d). Mean bead type VI collagen microfibril width was 60.7 nm (SEM 3.7 nm) on SO, 49.6 nm (SEM 2.6 nm) on SO/PLGA and 36.3 nm (SEM 1.5 nm) on SO/OTS (Fig. 6c). Correcting for tip radius W t was 12.0 nm on SO/PLGA and 6.9 nm on SO/OTS (Fig. 6c). Laterally spread material from type VI collagen microfibrils adsorbed on SO was less than 0.3 nm in height at the periphery (Fig. 6b).

ARTICLE IN PRESS M.J. Sherratt et al. / Biomaterials 26 (2005) 7192–7206

7199

Table 1 Mean microfibril periodicity and flexion angle: SO, SO/PLGA and SO/OTS. SEM denoted by brackets. SO

SO/PLGA

SO/OTS

Fibrillin Repeats/mm2 Periodicity (nm) Microfibril curvature (1)

2.4 (0.4) 46.3 (0.3) 151.1 (0.9)

3.1 (0.4) 58.4 (0.3) 163.5 (0.6)

2.1 (0.3) 58.3 (0.4) 161.0 (0.6)

Type VI collagen Repeats/mm2 Periodicity (nm) Microfibril curvature (1) Repeats curvature (1)

5.9 102.4 156.6 165.7

5.2 107.5 155.4 161.3

4.3 105.3 156.7 156.7

(0.4) (0.7) (0.9) (0.6)

(0.4) (0.7) (0.9) (0.6)

(0.4) (0.6) (0.8) (0.6)

69

100

200

80

160

60 40 20 0

(b)

1

Flexion angle (°)

Periodicity (nm)

(a)

0

20

40 Repeat

60

120 80 40 0

(c)

0

20

40 Repeat

60

133˚

(d)

(e)

Fig. 4. Periodicity and flexion angle: SO, SO/PLGA and SO/OTS. (a) AFM height image of a fibrillin microfibril adsorbed on SO; 1  0.5 mm extracted region from a 2 mm scan, Z-scale ¼ 15 nm. The terminal repeats, beads 1 and 69, are indicated on the image. Regions of reduced periodicity and flexion angle are enclosed by a dotted line. (b) Periodicity plots of the microfibril shown in a; the periodicity range from 50 to 60 nm is shaded and regions of reduced periodicity are indicated by braces. (c) Flexion angle plot of the microfibril shown in a; regions of reduced flexion angle corresponding to reduced periodicity are indicated by braces. (d) AFM height images, 2 mm scans, of a type VI collagen microfibril adsorbed on SO/ OTS. Bending of the collagen triple helical region is localised to one end of the microfibril (dotted box). (e) Magnified region from d, the angle formed between the centre of the interbead and adjacent double bead centres is 1331 for the labelled repeat. Scale bar ¼ 100 nm, Z-scale ¼ 10 nm.

ARTICLE IN PRESS M.J. Sherratt et al. / Biomaterials 26 (2005) 7192–7206

7200

3.5. Biological function: cell spreading on fibrillin microfibrils adsorbed to G/MHS, G/E and G/ES substrates

Height (nm)

Fig. 5. Microfibril morphology: SO, SO/PLGA and SO/OTS. Height contour maps of straightened fibrillin (a) and type VI collagen (b) microfibrils (1000  160 nm). Discontinuous contours, colour coded in 1 nm steps (1–10 nm) and 0.5 nm steps (0–1 nm).

8 7 6 5 4 3 2 1 0

SO

SO/ PLGA

Untreated glass coverslips (G) had a patchy appearance which was reduced after both cleaning and silanisation treatments (Fig. 7a). Similar RMS roughness (Rq) values were determined from AFM height images of methanol/acid washed glass (G/MHS), ethanol washed glass (G/E) and silanised glass (G/ES) (Figs. 7a and b). G/MHS glass coverslips had an Rq value of 0.303 nm (SD ¼ 0.095 nm), G/E (Rq ¼ 0.293 nm, SD ¼ 0.117 nm) and G/ES (Rq ¼ 0.251 nm, SD ¼ 0.116 nm) (Fig. 7b). There were significant but small differences between the Rq values for each substrate (Mann–Whitney U-test; po0:001). Following methanol and acid washing (G/MHS), glass coverslips were highly hydrophilic (y ¼ 20:71, SD ¼ 2.41) (Fig. 7c). Ethanol washing (G/E) and ethanol washing followed by silanisation (G/ES) produced hydrophilic (y ¼ 38:31, SD ¼ 1.61) and hydrophobic (y ¼ 92:81, SD ¼ 1.11) substrates respectively (Fig. 7c). All washing and silanisation treatments reduced contact angle hysteresis compared with untreated glass (data not shown). Fibrillin microfibrils adsorbed to all glass substrates (G/MHS, G/E and G/ES) (Fig. 8a). Significantly more

SO/ OTS

SO

(a)

10

Width (nm)

60 40

SO

6 4 2

20 SO

SO/ SO/ PLGA OTS Fibrillin

SO SO/ SO/ PLGA OTS Type VI collagen

bead interbead

8 Height (nm)

Width Wt

80

(c)

SO/ OTS

(b)

100

0

SO/ PLGA

plateau

plateau

0 -80 -60 -40 -20 0 20 40 60 80 (d) Transverse distance (nm)

Fig. 6. Microfibril width: SO, SO/PLGA and SO/OTS. (a and b) Height contour maps of mean microfibril repeats. Discontinuous contours colour coded in 0.5 nm steps (1–8.5 nm) and 0.25 nm steps (0–1 nm). (a) Fibrillin microfibrils, n ¼ 50; SO 51  160 nm, SO/PLGA and SO/OTS 58  160 nm. (b) Type VI collagen microfibrils, n ¼ 20; 105  160 nm. (c) Mean bead THP width and corrected width (W t ) of fibrillin and type VI collagen microfibrils adsorbed on SO, SO/PLGA and SO/OTS. (d) Bead and interbead THP through a single repeat extracted from a fibrillin microfibril adsorbed on SO. Laterally spread material formed a plateau of 0.5–0.6 nm height. Error bar ¼ SEM.

ARTICLE IN PRESS M.J. Sherratt et al. / Biomaterials 26 (2005) 7192–7206

G

G/MHS

1

1

0

0

-1

-1 0

0

200 400 600 800 1000 Axial distance (nm)

200 400 600 800 1000 Axial distance (nm)

G/E

G/S

1

1

0

0

-1

-1 0

(a)

0

200 400 600 800 1000 Axial distance (nm)

Contact angle (
°)

RMS roughness (Rq) (b)

200 400 600 800 1000 Axial distance (nm)

140

0.50 0.40 0.30 0.20 0.10 0.00

7201

G/MHS

G/E Substrate

120 100 80 60 40 20 0

G/ES (c)

G/MHS

G/E Substrate

G/ES

Fig. 7. Glass substrate characterisation. (a) AFM height images and 1 mm height profiles of G, G/MHS, G/E and G/ES substrates; 1 mm extracted regions from 2 mm scans; Z-scale 5 nm. (b) RMS roughness (Rq) values measured from AFM height scans. (c) Sessile drop water contact angle measurements for G/MHS, G/E and G/ES substrates. Error bars ¼ SD.

microfibrils adsorbed to G/E (7.51 repeats mm
2, SEM ¼ 0.79 repeats mm
2) than to G/MHS (1.98 repeats mm
2, SEM ¼ 0.28 repeats mm
2) or G/ES substrates (1.27 repeats mm
2, SEM ¼ 0.17 repeats mm
2), Mann–Whitney U-test; po0:001. However, there was no significant difference between microfibril adsorption on G/MHS and G/ES substrates (Mann–Whitney Utest). Fibrillin microfibrils adsorbed to G/MHS substrates were morphologically similar to fibrillin microfibrils adsorbed to SO substrates (Fig. 8a). Periodicity reduction/increased flexibility and a diffuse appearance

were commonly observed within whole microfibrils and distinct regions of the same microfibril. Microfibrils adsorbed to G/ES were highly compacted and morphologically similar to SO/OTS adsorbed assemblies. G/E adsorbed microfibrils were morphologically similar to SO/PLGA microfibrils. HDF cell spreading was significantly increased on all glass substrates by the prior adsorption of fibrillin microfibrils compared to BSA alone (w2 -test; po0:001) (Fig. 8c). There was no significant difference in cell spreading on fibrillin microfibril functionalised G/E

ARTICLE IN PRESS 7202

M.J. Sherratt et al. / Biomaterials 26 (2005) 7192–7206

(Rq values of 0.037, 0.230 and 0.183 nm respectively; unpublished data and [11]). The SO and coated SO substrates used in this study present chemically homogeneous surfaces which exhibit less variation in RMS roughness than mica, chemically modified mica and glass and a greater variation in surface energy from reported contact angle measurements (Figs. 2b and c). 4.2. Protein adsorption: SO, SO/PLGA and SO/OTS

Fig. 8. Microfibril and cellular morphology on glass substrates. (a) AFM height images of adult bovine ciliary zonule fibrillin microfibrils adsorbed on G/MHS, G/E and G/ES substrates. Diffuse (D) and reduced periodicity (RP) microfibrils were commonly observed on G/ MHS substrates. 1 mm extracted regions from 2 mm scans; Z-scale 15 nm (G/MHS and G/E), 20 nm (G/ES). (b) Phase contrast microscopy images of HDF cell spreading on fibrillin microfibril functionalised and BSA blocked G/MHS, G/E and G/ES substrates; scale bar ¼ 10 mm. (c) HDF cell spreading on G/MHS, G/E and G/ES substrates. Each substrate was exposed to a 10 mg ml
1 fibrillin microfibril suspension followed by a BSA blocking step or to BSA alone.

(48.9%) and G/ES (49.8%) (w2 -test). Significantly fewer cells spread on G/MHS (20.0%) compared with G/E and G/ES fibrillin microfibril functionalised substrates (w2 -test; po0:001). Similar numbers of fibrillin microfibrils adsorbed to G/MHS and G/ES substrates after a 1 min exposure but G/ES adsorbed microfibrils supported significantly more cell spreading.

4. Discussion 4.1. Substrate characterisation: SO, SO/PLGA and SO/ OTS In our previous study we compared the morphology of microfibrils adsorbed on smooth, hydrophilic, chemically homogeneous mica; rougher, less hydrophilic, chemically heterogeneous mica-PLL and rougher, hydrophobic, chemically heterogeneous borosilicate glass

Previous studies on isolated proteins, including proteolytically processed type I collagen, have suggested that in general protein adsorption increases with increasing substrate hydrophobicity [40–42]. Extracted type I collagen is not however amphiphilic, consisting exclusively of an uninterrupted hydrophilic triple helical region. Both type VI collagen and fibrillin are amphiphilic, as is human fibronectin which adsorbs in equal amounts on bacterial and more hydrophilic tissue grade polystyrene [11,14]. The amphiphilic nature of these ECM macromolecular assemblies may allow them to adsorb to substrates via both hydrophobic and electrostatic interactions, reducing the importance of substrate amphiphilicity in mediating the amount of adsorbed protein, but inducing highly substrate specific structural rearrangements [12,42]. 4.3. Microfibril morphology—highly hydrophilic SO Fibrillin microfibrils adsorbed on hydrophilic SO were highly diffuse with a significantly reduced periodicity compared to tissue microfibrils and microfibrils adsorbed on mica, poly-L-lysine coated mica and carbon films [21,30,43]. Periodicity may be reduced in vitro by the chelation of bound Ca2+ and previously reported Ca2+ deficient microfibrils are indistinguishable in appearance, periodicity and flexion angle from SO adsorbed microfibrils [30,43]. The retention, however, of a 56 nm periodicity for microfibrils adsorbed to SO/ PLGA and SO/OTS precluded the loss of Ca2+ or the effects of drying as an explanation for altered morphology. Therefore the reduced periodicity/flexion angle appeared to be due solely to the interaction of hydrated microfibrils with the hydrophilic SO substrate. This observation also raises the possibility that reported alterations in the periodicity/curvature of Ca2+ depleted microfibrils may not occur in suspension but are induced during microfibril/substrate interaction. The reduction in periodicity and increased curvature of fibrillin microfibrils adsorbed on hydrophilic SO was associated with an increased width (Fig. 6c). We propose that lateral attraction of microfibrillar material to the hydrophilic SO substrate pulls adjacent beads together during adsorption. Adsorption to hydrophilic mica had no effect on periodicity and did not induce major lateral spreading but mica has a lower charge

ARTICLE IN PRESS M.J. Sherratt et al. / Biomaterials 26 (2005) 7192–7206

density than SO [11,44,45]. We have previously suggested that the disruption of microfibril domain organisation on mica is driven by an attraction between negatively charged carbohydrate side chains and the positively charged divalent cation counter ion layer [11]. Microfibril association with the more highly charged SO appears to be strong enough to disrupt not only domain/ domain spacing but also secondary protein structure with potentially major consequences for biological function [17,18]. Meadows et al. [17] demonstrated that characteristic domain unfolding events were absent for isolated fibronectin monomers on a negatively charged surface. Domain unfolding was only observed when fibronectin was aggregated and therefore partially shielded from the substrate. Similar shielding effects were observed for type VI collagen adsorbed on mica where the double beaded structure was retained only within microfibrillar bundles [11]. Norde and Favier [18] employed CD to demonstrate that the a-helix content of both bovine serum albumin and hen’s egg lysozyme decreased upon adsorption to hydrophilic silica particles. The susceptibility to hydrophilic substrate induced unfolding appears to vary between proteins [18,46]. Fibrillin monomers are comprised primarily of cbEGF domains with cross-sectional diameters of 2.0 and 2.2 nm respectively [47,48]. Lateral alignment of eight structurally intact monomers on a substrate would produce a feature of width 16.0–17.6 nm and height 2.0–2.2 nm, in comparison with the 90 nm wide and 0.5 nm high feature observed in the plateau region of fibrillin microfibrils adsorbed on SO [49]. The height and uniformity of the plateau region suggests that secondary structure is substantially disrupted on SO substrates. The small but significant reduction in type VI collagen microfibril periodicity on SO was also associated with lateral spreading of material. Most of the material in the bead region of type VI collagen microfibrils adsorbed on SO was less than 1 nm in height (Fig. 6a). The majority of the globular domains in type VI collagen are homologous to von Willebrand factor (vWF) A domains which occupy a volume of 33.5 nm3 [50]. Assuming A-domains are approximately spherical each domain would have a diameter of 4 nm. As with the cbEGF domains of fibrillin microfibrils the globular domains of type VI collagen are less than one quarter of the expected height and are also suggestive of a disruption of secondary structure. 4.4. Microfibril morphology—hydrophobic SO/OTS Adsorption to hydrophobic SO/OTS had no effect on fibrillin microfibril periodicity or flexion angle but induced a highly compact morphology. It has been generally accepted that globular proteins undergo more conformational changes on adsorption to hydrophobic than hydrophilic surfaces as the protein structure alters

7203

to maximise contact between non-polar groups [51]. Previous AFM studies have observed highly compact structures of plasma von Willebrand factor and fibronectin on hydrophobic substrates [9,10]. Fibronectin in solution is thought to have a compact structure with the RGD sites unavailable to cellular integrins and adsorption to hydrophilic but not hydrophobic substrates induces an extended conformation which may expose these sites [10,14]. Similar hydrophobic driven mechanisms may alter the biological function of fibrillin microfibrils through the masking of binding regions. Type VI collagen microfibril periodicity was not altered by adsorption to SO/OTS but curved repeats were often observed (Figs. 4d and e) [22,50]. Similar regions were observed with less frequency on SO/PLGA and never observed on SO. The globular regions which form the type VI collagen double beads and the four super-coiled triple helical regions which form the interbead are predominantly hydrophilic [11]. We propose that repulsion of flexible hydrophilic beads from the hydrophobic substrate pushes the beads into close contact inducing a bend in the triple helical region. The amount of flexibility exhibited by isolated collagen monomers under physiological conditions remains uncertain and the concept of the collagen monomer as a rigid, rod-shaped molecule has recently been challenged by single-molecule force spectroscopy studies [52,53]. Cross-linked collagen triple helices are known however to perform load-bearing function in vivo and the interaction of type VI collagen microfibrils with a hydrophobic substrate generates forces of a sufficient magnitude to reproducibly deform four potentially super-coiled and cross-linked collagen triple helices with possible consequences for biological function. 4.5. Microfibril morphology—slightly hydrophilic SO/ PLGA SO/PLGA is intermediate in surface energy between SO and SO/PLGA and the periodicity, flexion angle and lateral organisation of both microfibrillar species appeared to be largely unaffected by association with this slightly hydrophilic substrate. Although drying of proteins will have some effect on their morphology, all the major morphological features of hydrated fibrillin microfibrils (observed by cryo-TEM) are preserved following dehydration [21]. Bergkvist et al. [10] also noted that substrate induced morphological changes in hydrated fibronectin were preserved in dried samples. In the case of fibrillin microfibrils, the corrected W t (16.7 nm) was very close to the hydrated diameter (16.3 nm) determined by cryo-TEM [21]. It appears therefore, that intermediate surface energy substrates such as SO/PLGA and G/E (see Section 4.6) present an optimum chemical environment for the preservation of native protein morphology and retention of biological

ARTICLE IN PRESS 7204

M.J. Sherratt et al. / Biomaterials 26 (2005) 7192–7206

functionality. No data is available on the hydrated diameter of type VI collagen microfibrils. 4.6. Biological function: cell spreading on fibrillin microfibrils adsorbed to G/MHS, G/E and G/ES substrates Alcohol/acid washed and silanised glass coverslips were of comparable roughness to coated and un-coated silicon wafers (Figs. 2b and 7b) with a similar range of contact angles (Figs. 2c and 7c). The low contact angle of G/MHS washed glass suggests that the cleaning process acted through two mechanisms; removal of surface contaminants and activation of negatively charged hydroxyl groups [25]. The higher contact angle of G/E glass may be due to inefficient cleaning of contaminants or failure of the ethanol to activate the surface [25]. Silanisation of ethanol washed glass (G/ES) produced a hydrophobic surface. Where surface roughness iso100 nm chemical heterogeneity is thought to be the most important factor in generating contact angle hysteresis [54,55]. The low contact angle hysteresis and closed hysteresis loop of the silanised G/ES substrate suggests that the organopolysiloxane forms a uniform coating on the glass surface. Fibrillin microfibrils coated onto G/E or G/ES surfaces and microfibrils coated onto microtiter plates supported similar levels of HDF spreading (Fig. 8c, data not shown and [20]). Microfibrils coated onto highly hydrophilic G/MHS supported significantly reduced cell spreading in comparison to the G/ES substrate which adsorbed similar numbers of microfibrils after a 1 min incubation. We therefore attribute the differences in cell spreading to substrate chemistry induced morphological changes in adsorbed and hydrated microfibrils. HDF cell attachment is mediated by integrins a5 b1 and av b3 which interact with the eight predicted RGD cell attachment motifs within each fibrillin microfibril repeat [20,49]. Therefore changes in the ability of adsorbed microfibrils to support cell spreading are likely to involve masking or structural disruption of the RGD site and as a consequence reduced binding of either or both a5 b1 and av b3 integrins. Fibronectin, which also binds integrins via an RGD site [7], is believed to have a compact conformation in solution which is increasingly disrupted in adsorbed proteins as substrate hydrophilicity increases [10]. The transition from a compact to an open form of fibronectin is associated with increased RGD-integrin binding dependent cell spreading, proliferation and differentiation [14,56]. In contrast, the lateral compaction of fibrillin microfibrils on hydrophobic substrates does not appear to alter the availability of the RGD site to cells compared with the more hydrophilic ethanol washed substrates. The reduced cell spreading on hydrophilic (G/MHS) substrates may due to masking of the RGD site in axially compacted

(reduced periodicity) microfibrils or to disruption of domain structure in the laterally spread regions. The ability of G/MHS adsorbed microfibrils to support cell spreading above BSA levels suggests that only one of these two morphological changes influences integrin/ RGD site interactions.

5. Conclusion The biochemically distinct native macromolecular assemblies, fibrillin and type VI collagen microfibrils adsorbed readily to substrates with low (SO/OTS, G/ES), high (SO, G/MHS) and intermediate (SO/ PLGA, G/E) surface energies. Interaction with high surface energy substrates promoted extensive lateral spreading and influenced both periodicity (fibrillin and type VI collagen) and curvature (fibrillin). These morphological changes also inhibited integrin mediated cell spreading. In general, microfibrils adsorbed to lower energy substrates experienced fewer morphological changes and in the case of fibrillin microfibrils supported similar levels of cell spreading. In contrast cell spreading is inhibited when fibronectin is adsorbed to a low energy surface. We suggest therefore that intermediate surface energy substrates may provide the optimum chemical environment for the manufacture of biomimetic materials which utilise the multi-functionality of multiple native ECM assemblies.

Acknowledgements The authors thank Miss Amanda Morgan for expert technical assistance. M.J.S. would like to thank Research into Ageing for financial support. C.M.K. and D.V.B. would like to acknowledge the support of the BBSRC, MRC (UK), EPSRC and Royal Society (C.M.K.). S.S.C. and N.H. would like to acknowledge the support of the MRC (UK). P.S. and J.R.L. would like to acknowledge the support of the BBSRC.

References [1] Heitz J, Svorcik V, Bacakova L, Rockova K, Ratajova E, Gumpenberger T, Bauerle D, Dvorankova B, Kahr H, Graz I, Romanin C. Cell adhesion on polytetrafluoroethylene modified by UV-irradiation in an ammonia atmosphere. J Biomed Mater Res A 2003;67A:130–7. [2] Kiss E, Vargha-Bulter EI. Novel method to characterize the hydrolytic decomposition of biopolymer surfaces. Colloid Surf B 1999;15:181–93. [3] Croll TI, O’Connor AJ, Stevens GW, Cooper-White JJ. Controllable surface modification of poly(lactic-co-glycolic acid) (PLGA) by hydrolysis or aminolysis I: Physical, chemical, and theoretical aspects. Biomacromolecules 2004;5:463–73.

ARTICLE IN PRESS M.J. Sherratt et al. / Biomaterials 26 (2005) 7192–7206 [4] Khang G, Jong-Hwa C, Rhee JM, Lee HB. Interaction of different types of cells on physicochemically treated poly(Llactide-co-glycolide) surfaces. J Appl Polym Sci 2002;85:1253–62. [5] Quirk RA, Chan WC, Davies MC, Tendler SJB, Shakesheff KM. Poly(L-lysine)-GRGDS as a biomimetic surface modifier for poly(lactic acid). Biomaterials 2001;22:865–72. [6] Shin H, Seongbong J, Mikos AG. Biomimetic materials for tissue engineering. Biomaterials 2003;24:4353–64. [7] Newham P, Humphries MJ. Integrin adhesion receptors: Structure, function and implications for biomedicine. Mol Med Today 1996;2:304–13. [8] Rosso F, Giordano A, Barbarisi M, Barbarisi A. From cell-ECM interactions to tissue engineering. J Cell Physiol 2004;199:174–80. [9] Raghavachari M, Tsai HM, Kottke-Marchant K, Marchant RE. Surface dependent structures of von Willebrand factor observed by AFM under aqueous conditions. Colloids Surf B 2000;19: 315–24. [10] Bergkvist M, Carlsson J, Oscarsson S. Surface-dependent conformations of human plasma fibronectin adsorbed to silica, mica, and hydrophobic surfaces, studied with use of atomic force microscopy. J Biomed Mater Res 2003;64A:349–56. [11] Sherratt MJ, Holmes DF, Shuttleworth CA, Kielty CM. Substrate-dependent morphology of supramolecular assemblies: fibrillin and type-VI collagen microfibrils. Biophys J 2004;86: 3211–22. [12] Vandulm P, Norde W. The adsorption of human plasma albumin on solid surfaces, with special attention to the kinetic aspects. J Colloid Interf Sci 1983;91:248–55. [13] Haynes CA, Norde W. Structures and stabilities of adsorbed proteins. J Colloid Interf Sci 1995;169:313–28. [14] Garcia AJ, Vega MD, Boettiger D. Modulation of cell proliferation and differentiation through substrate-dependent changes in fibronectin conformation. Mol Biol Cell 1999;10:785–98. [15] Keselowsky BG, Collard DM, Garcia AJ. Surface chemistry modulates fibronectin conformation and directs integrin binding and specificity to control cell adhesion. J Biomed Mater Res 2003;66A:247–59. [16] Krammer A, Craig D, Thomas WE, Schulten K, Vogel V. A structural model for force regulated integrin binding to fibronectin’s RGD-synergy site. Matrix Biol 2002;21:139–47. [17] Meadows PY, Bemis JE, Walker GC. Single-molecule force spectroscopy of isolated and aggregated fibronectin proteins on negatively charged surfaces in aqueous liquids. Langmuir 2003;19:9566–72. [18] Norde W, Favier JP. Structure of adsorbed and desorbed proteins. Colloid Surf 1992;64:87–93. [19] Kielty CM, Sherratt MJ, Shuttleworth CA. Elastic fibres. J Cell Sci 2002;115:2817–28. [20] Bax DV, Bernard SE, Lomas A, Morgan A, Humphries J, Shuttleworth CA, Humphries MJ, Kielty CM. Cell adhesion to fibrillin-1 molecules and microfibrils is mediated by a5b1 and aVb3 integrins. J Biol Chem 2003;278:34605–16. [21] Sherratt MJ, Baldock C, Haston JL, Holmes DF, Jones CJP, Shuttleworth CA, Wess TJ, Kielty CM. Fibrillin microfibrils are stiff reinforcing fibres in compliant tissues. J Mol Biol 2003;332: 183–93. [22] Kielty CM, Shuttleworth CA. Microfibrillar elements of the dermal matrix. Micros Res Techniq 1997;38:413–27. [23] de Gennes PG. Wetting: statics and dynamics. Rev Mod Phys 1985;57:827–63. [24] Wang Y, Lieberman M. Growth of ultrasmooth octadecyltrichlorosilane self-assembled monolayers on SiO2. Langmuir 2003;19:1159–67. [25] Cras JJ, Rowe-Tait CA, Nivens DA, Ligler FS. Comparison of chemical cleaning methods of glass in preparation for silanization. Biosens Bioelectron 1999;14:683–8.

7205

[26] Bezanilla M, Manne S, Laney DE, Lyubchenko YL, Hansma HG. Adsorption of DNA to mica, silylated mica, and minerals: characterization by atomic force microscopy. Langmuir 1995;11: 655–9. [27] Murphy E, Kiddie JL, Lu JR, Brewer J, Russell J. The reduced adsorption of lysozyme at the phosphorylcholine incorporated polymer/aqueous solution interface studied by spectroscopic ellipsometry. Biomaterials 1999;20:1501–11. [28] Lu JR, Su TJ, Thirtle PN, Thomas RK, Rennie AR, Cubitt R. The denaturation of lysozyme layers adsorbed at the hydrophobic solid/liquid surface studied by neutron reflection. J Colloid Interf Sci 1998;206:212–23. [29] Kielty CM, Cummings C, Whittaker SP, Shuttleworth CA, Grant ME. Isolation and ultrastructural analysis of microfibrillar structures from foetal bovine elastic tissues. Relative abundance and supramolecular architecture of type VI collagen assemblies and fibrillin. J Cell Sci 1991;99:797–807. [30] Cardy CM, Handford PA. Metal ion dependency of microfibrils supports a rod-like conformation for fibrillin-1 calcium binding epidermal growth factor-like domains. J Mol Biol 1998;276: 855–60. [31] Ratcliff GC, Erie DA. A novel single-molecule study to determine protein-protein association constants. J Am Chem Soc 2001;123: 5632–5. [32] Kocsis E, Trus BL, Steer CJ, Bisher ME, Steven AC. Image averaging of flexible fibrous macromolecules—the clathrin triskeleton has an elastic proximal segment. J Struct Biol 1991;107: 6–14. [33] Bustamante C, Vesenka J, Tang CL, Rees W, Guthold M, Keller R. Circular DNA molecules imaged in air by scanning force microscopy. Biochemistry 1992;31:22–36. [34] Engel A, Schoenenberger C-A, Muller DJ. High resolution imaging of native biological sample surfaces using scanning probe microscopy. Curr Opin Struct Biol 1997;7:279–84. [35] Balnois E, Wilkinson KJ. Sample preparation techniques for the observation of environmental biopolymers by atomic force microscopy. Colloid Surf A 2002;207:229–42. [36] Wong SS, Harper JD, Lansbury PT, Lieber CM. Carbon nanotube tips: high-resolution probes for imaging biological systems. J Am Chem Soc 1998;120:603–4. [37] Carlson JW, Godfrey BJ, Sligar SG. Metal binding colloidal gold particles: a versatile scanning force microscope tip calibrator for fluid imaging. Langmuir 1999;15:3086–90. [38] Zhao XS, Lu GQ. Modification of MCM-41 by surface silylation with trimethylchlorosilane and adsorption study. J Phys Chem B 1998;102:1556–61. [39] Angst DL, Simmons GW. Moisture adsorption characteristics of organosiloxane self-assembled monolayers. Langmuir 1991;7: 2236–42. [40] Penners G, Priel Z, Silberberg A. Irreversible adsorption of triplehelical soluble collagen monomers from solution to glass and other surfaces. J Colloid Intfer Sci 1981;80:437–44. [41] Pamula E, De Cupere V, Dufrene YF, Rouxhet PG. Nanoscale organization of adsorbed collagen: influence of substrate hydrophobicity and adsorption time. J Colloid Intfer Sci 2004;271: 80–91. [42] Nath N, Hyun J, Ma H, Chilkoti A. Surface engineering strategies for control of protein and cell interactions. Surf Sci 2004;570: 98–110. [43] Wess TJ, Purslow PP, Sherratt MJ, Ashworth J, Shuttleworth CA, Kielty CM. Calcium determines the supramolecular organization of fibrillin-rich microfibrils. J Cell Biol 1998;141: 829–37. [44] Le Grange JD, Markham JL, Kurkjian CR. Effects of surface modification on the deposition of silane monolayers on silica. Langmuir 1993;9:1749–53.

ARTICLE IN PRESS 7206

M.J. Sherratt et al. / Biomaterials 26 (2005) 7192–7206

[45] Pashley RM. DLVO and hydration forces between mica surfaces in Li+, Na+, K+ and Cs+ electrolyte solutions, a correlation of double layer and hydration forces with surface cation exchange properties. J Colloid Interf Sci 1981;83:531–46. [46] McMillin CR, Walton AG. Circular-dichroism technique for study of adsorbed protein structure. J Colloid Interf Sci 1974;48:345–9. [47] Sakai LY, Keene DR, Glanville RW, Bachinger HP. Purification and characterisation of fibrillin, a cysteine-rich structural component of connective tissue microfibrils. J Biol Chem 1991;266: 14763–70. [48] Downing AK, Knott V, Werner JM, Cardy CM, Campbell ID, Handford PA. Solution structure of a pair of calcium binding epidermal growth factor-like domains: implications for the Marfan syndrome and other genetic disorders. Cell 1996;85: 597–605. [49] Baldock C, Koster AJ, Ziese U, Rock MJ, Sherratt MJ, Kadler KE, Shuttleworth CA, Kielty CM. The supramolecular organization of fibrillin-rich microfibrils. J Cell Biol 2001;152: 1045–56.

[50] Baldock C, Sherratt MJ, Shuttleworth CA, Kielty CM. The supramolecular organization of collagen VI microfibrils. J Mol Biol 2003;330:297–307. [51] Marsh RJ, Jones RAL, Sferrazza M. Adsorption and displacement of a globular protein on hydrophilic and hydrophobic surfaces. Colloid Surf B 2002;23:31–42. [52] Obrink B. Non-aggregated tropocollagen at physiological pH and ionic strength. A chemical and physico-chemical characterization of tropocollagen isolated from the skin of lathyritic rats. Eur J Biochem 1972;25:563–72. [53] Sun YL, Luo ZP, Fertala A, An KN. Direct quantification of the flexibility of type I collagen monomer. Biochem Bioph Res Co 2002;295:382–6. [54] Neumann AW, Good RJ. Thermodynamics of contact angles: I. Heterogeneous solid surfaces. J Colloid Interf Sci 1972;38:341–58. [55] Extrand CW, Kumagai Y. An experimental study of contact angle hysteresis. J Colloid Interf Sci 1997;191:378–83. [56] Grinnell F, Feld MK. Adsorption characteristics of plasma fibronectin in relationship to biological activity. J Biomed Mater Res 1981;15:363–81.