Colloids and Surfaces B: Biointerfaces 124 (2014) 87–96

Contents lists available at ScienceDirect

Colloids and Surfaces B: Biointerfaces journal homepage: www.elsevier.com/locate/colsurfb

Understanding and controlling type I collagen adsorption and assembly at interfaces, and application to cell engineering Christine C. Dupont-Gillain ∗ Université catholique de Louvain, Institute of Condensed Matter and Nanosciences, Croix du Sud 1 (L7.04.01), 1348 Louvain-la-Neuve, Belgium

a r t i c l e

i n f o

Article history: Received 3 June 2014 Received in revised form 21 August 2014 Accepted 22 August 2014 Available online 6 September 2014 Keywords: Collagen Atomic force microscopy Protein adsorption Self-assembly Biointerfaces

a b s t r a c t Collagen is a large anisotropic and self-assembling extracellular matrix protein. Understanding and controlling its adsorption and assembly at interfaces is expected to increase our general knowledge of protein adsorption as well as to open the way to the development of biointerfaces of interest for biomaterials science and tissue engineering. The work related to type I collagen adsorption performed in our laboratory over the past twenty years is reviewed. Substrate chemical nature and adsorption conditions (collagen concentration, adsorption duration) were shown to affect collagen adsorbed amount and supramolecular organization. Collagen assemblies were formed starting from the interface, and assembly was favored by hydrophobic substrates and high adsorbed amount. Substrates were designed to better control collagen adsorption and assembly. The spatial control of adsorption was ensured by chemically heterogeneous substrates, which also affected collagen assembly when domains with a dimension smaller than the length of the collagen molecule (i.e. 300 nm) were prepared. Mixed polymer brushes were used to achieve a temporal control of adsorption: adsorption and desorption were reversibly triggered by changes of pH and ionic strength. Layer-by-layer assembly of collagen in a nanoporous template was used to elaborate collagen-based nanotubes, which were further deposited on ITO glass substrates by electrophoretic deposition. Finally, the evaluation of cell behavior on the created biointerfaces showed that the control of collagen organization can be successfully used to alter cell behavior. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Collagen, the most abundant protein in animal kingdom, is a major component of extracellular matrices (ECM) [1], where it forms insoluble fibrils and provides recognition sequences for cell integrins and adhesion proteins such as fibronectin [2]. At least ten different types of collagen molecules are found in vertebrates. Type I collagen is predominant in skin, tendons and bones. It is made of three polypeptides entwined in a superhelix, forming a 300 nmlong molecule, with a diameter of 1.5 nm and a molar mass of 300 kDa [1]. Type I collagen self-assembles into fibrils which can have a diameter up to the micrometer range [3]. Owing to their amphiphilic and polyampholyte character, as well as to their low entropy structure, proteins are very surfaceactive. As a general rule, all proteins tend to adsorb to all substrates [4]. The driving forces for protein adsorption include mainly: (i) electrostatic interactions, which strongly depend on pH and ionic strength (I) of the surrounding medium, (ii) dispersion forces,

∗ Tel.: +32 10473584. E-mail address: [email protected] http://dx.doi.org/10.1016/j.colsurfb.2014.08.029 0927-7765/© 2014 Elsevier B.V. All rights reserved.

which are however relatively weak for small protein molecules, (iii) hydrophobic forces, arising from substrate and protein molecule dehydration upon adsorption, and (iv) the gain of entropy due to structural relaxation at the interface [5]. Type I collagen adsorbs readily to most substrates [6–8]. The helical structure of the molecule is not easily perturbed by adsorption, and electrostatic interactions were shown not to control collagen adsorption [9], although this is not so easy to evidence because collagen aggregation state strongly depends on pH and I [1]. The collagen molecule is however very large and anisotropic, and attractive van der Waals forces may then dominate the adsorption process, which is also largely irreversible since it is very difficult to desorb a molecule interacting through many molecular segments with the substrate [5]. On hydrophobic substrates, hydrophobic interactions may also contribute to the adsorption mechanism, and collagen affinity for polystyrene (PS), a hydrophobic substrate, was indeed higher than for oxidized polystyrene (PSox), a more hydrophilic substrate [8]. But what are the driving forces that made us study collagen adsorption for almost twenty years? Compared to many globular proteins whose adsorption has been investigated in details, collagen presents very different properties in reason of its large size,

88

C.C. Dupont-Gillain / Colloids and Surfaces B: Biointerfaces 124 (2014) 87–96

its high anisotropy, and its self-assembling properties. This is in itself interesting, with a view to expand our general understanding of protein adsorption. Then, it opens perspectives for the design of controlled architectures at interfaces. Manipulating molecules at interfaces to build patterns with the desired geometry is a common challenge in nanotechnology, and collagen is a good candidate to reach this goal. Last but not least, collagen, the main ECM protein, offers recognition signals and suitable mechanical properties to cells. It is thus particularly useful to control cell–material interactions. Creating biointerfaces based on collagen to mimic the ECM and control cell behavior offers perspectives for applications in biomaterials science and tissue engineering. Collagen thus constitutes the ideal building block to elaborate biointerfaces with promising functions. The adequate and controlled design of such biointerfaces depends on a strong knowledge of collagen behavior at interfaces. The present paper reviews the work related to type I collagen adsorption and assembly at interfaces performed in our laboratory over the last twenty years. The factors influencing collagen adsorption and supramolecular organization were examined, and the mechanism of assembly of collagen from the interface was investigated. Substrates were designed to better control collagen adsorption, either spatially or temporally. Collagen-based nanoobjects were also synthesized to create multifunctional interfaces. Finally, the behavior of cells grown on the designed biointerfaces was studied. 2. Factors influencing adsorbed collagen supramolecular organization The role of substrate surface properties and of adsorption conditions (collagen concentration, duration of adsorption, drying step) on adsorbed collagen layers organization was thoroughly investigated [10–16]. This was mainly done using atomic force microscopy, whose resolution allows the supramolecular organization of the adsorbed layers to be unraveled. The collagen adsorbed amount was quantified based on radioassays, using collagen which was labeled by reductive methylation of lysine residues with 14 C formaldehyde. The adsorbed amount and collagen layer

organization were also approached based on the N/C ratio determined by X-ray photoelectron spectroscopy. Adsorption was always performed in static conditions at 37 ◦ C, from type I collagen solutions in phosphate buffered saline (PBS, pH 7.2, ionic strength = 160 mM). After the desired incubation time, samples (typically 1 cm × 1 cm squares) were thoroughly rinsed with ultrapure water then either analyzed as such or dried. Fig. 1 illustrates the effect of substrate surface chemical composition, collagen concentration, adsorption duration and mode of drying on collagen adsorbed amount and on the morphology of adsorbed layers. Using radioassays, collagen adsorption kinetics was monitored on polystyrene (PS), a hydrophobic substrate (water contact angle ≈90◦ ), at two different collagen concentrations (7 and 40 ␮g/ml), as well as on plasma-oxidized polystyrene (PSox), a more hydrophilic substrate (water contact angle ≈50◦ ) (Fig. 1a) [13,14]. In all cases, adsorbed amount increases sharply at short adsorption times, to reach a plateau value at longer times. The plateau value is reached faster at a concentration of 40 ␮g/ml compared to 7 ␮g/ml, and the corresponding adsorbed amount is higher on PS at a concentration of 40 ␮g/ml compared to PSox at the same concentration and to PS at 7 ␮g/ml. The adsorption isotherms for collagen adsorption on PS and PSox were also compared (see Ref. [8], not shown here). The plateau value was reached at a collagen concentration of about 20 ␮g/ml on PS, while it was only reached at a concentration of about 80 ␮g/ml on PSox. The organization of the adsorbed layers was strongly affected by the adsorption duration, as revealed by AFM images (Fig. 1b) [14]. On PS, at a collagen concentration of 7 ␮g/ml, a discontinuous layer made of individual molecules was found after 1 min of adsorption. When the adsorption time was increased to 30 min, a smooth layer was observed, which is attributed to the adsorption of a monolayer of collagen molecules. From 2 to 24 h of adsorption, the surface was then progressively covered with small filamentous structures (height 3–6 nm, length 250–500 nm), corresponding to assemblies of a few collagen molecules. A high density of these assemblies was obtained after 24 h. The presence and density of adsorbed supramolecular collagen assemblies can also be modulated by varying collagen concentration and substrate hydrophilicity (Fig. 1c)

Fig. 1. Factors influencing collagen adsorption and organization. (a) Collagen adsorbed amount as a function of time on PS (7 and 40 ␮g/ml) and PSox (40 ␮g/ml), measured using radioassays. (b) AFM images (2.5 ␮m × 5 ␮m, z-range = 10 nm) acquired in air on collagen adsorbed on PS from a 7 ␮g/ml solution, for different durations as indicated [adapted from Ref. [14] with permission from Elsevier]. (c) AFM images (5 ␮m × 2.5 ␮m, z-range = 5 nm) acquired in air on collagen layers obtained after 30 min of adsorption in the indicated conditions (SD = slow drying, which leads to dewetting).

C.C. Dupont-Gillain / Colloids and Surfaces B: Biointerfaces 124 (2014) 87–96

[10,13,14]. After 30 min of adsorption, no assemblies were found on PS using a collagen concentration of 7 ␮g/ml, while a dense layer of fibrils was found at 40 ␮g/ml. In the latter conditions, almost no assemblies were however formed on the more hydrophilic PSox substrate. Other pairs of materials, including a more hydrophobic and a more hydrophilic one, were investigated as well (not shown). Fibrillar collagen structures were found on hydrophobic CH3 -terminated self-assembled monolayers (SAMs) of alkanethiols on gold, while in the same adsorption conditions, smooth collagen layers were formed on hydrophilic OH-terminated SAMs [12]. Similarly, collagen assemblies were formed at the surface of poly(ethylene terephthalate) (PET), but not on plasma-oxidized PET [11]. A general trend can be extracted from these results: the tendency of collagen to assemble is stronger on more hydrophobic substrates. Although this may to some extent be related to a higher adsorbed amount on hydrophobic substrates, it was actually observed in situations where the adsorbed amount was similar, and it thus reflects differences in collagen–substrate interactions, as will be discussed further in the next section. The small collagen assemblies found at the interface upon spontaneous adsorption are the result of collagen self-assembly. They are not formed upon drying, and can easily be observed under liquid on wet samples, although with a lower resolution [12,13,17,18]. Collagen supramolecular structures with different geometries may also be obtained through dewetting. When the drying rate is sufficiently low, the water meniscus formed upon rupture of the water film will displace collagen molecules, forming dewetting patterns as illustrated in Fig. 1c [14,19]. This is only observed: (i) on hydrophobic substrates having a water contact angle that is high enough to provoke the rupture of the liquid film upon drying, and (ii) if the adsorbed amount is not too high, because otherwise the increased intermolecular interactions prevent collagen displacement along the surface plane. The patterns obtained by collagen dewetting were further used as a template to design nanostructured polymer surfaces. This was achieved by spin-coating a polymer solution on top of the dewetted collagen layers. Depending on the conditions (choice of solvent, polymer concentration), patterns with different geometries could be obtained [20]. To summarize, collagen layers formed by spontaneous adsorption present different supramolecular organizations depending on the substrate surface properties and on the adsorption conditions. Smooth layers corresponding to monolayers of isolated molecules are obtained on more hydrophilic substrates, and on hydrophobic substrates at low collagen concentration or short adsorption time. The formation of collagen assemblies at the interface is favored by substrate hydrophobicity, and by increasing the adsorbed amount (i.e.at higher collagen concentration or longer adsorption time). Dewetting patterns can also be produced by drying adsorbed collagen monolayers on hydrophobic substrates.

3. Mechanism of formation of collagen assemblies at interfaces Type I collagen is a self-assembling molecule, giving structural properties to the extracellular matrices through the formation of micrometer-size fibrils. Collagen assemblies are also formed in vitro under given conditions of pH and temperature [1]. While essentially monomeric solutions are found in acidic type I collagen solutions kept at 4 ◦ C, aggregates appear upon increasing the temperature and neutralizing the solution. Collagen assemblies found in adsorbed phases could thus either be the result of deposition of assemblies already formed in solution, or be directly formed at the interface. The role of substrate hydrophilicity (see Section 2) in fibril formation at interfaces indicates that assemblies are rather built from the interface. A more direct evidence of such a mechanism was

89

provided by comparing adsorbed layers obtained from solutions with different states of aggregation. Collagen solutions at a given concentration were prepared at pH 5.8 then aged for different times at 37 ◦ C. It was shown, using UV–vis spectrophotometry, that aggregation in solution increased with aging time [17]. Adsorption was then performed on PS (collagen concentration = 40 ␮g/ml; adsorption duration = 30 min). AFM images, presented in Fig. 2a, show a strong effect of collagen aggregation in solution on the presence of assemblies at the interface: when more aggregates are present in solution, less assemblies are formed in the adsorbed phase. This demonstrates that the collagen fibrils found at the interface are formed in situ and not previously in solution. Collagen monomers would adsorb first, because they diffuse faster to the interface, and would then prevent further adsorption of collagen aggregates. The decreased aggregation at the interface when aggregation in solution is increased is then attributed to a lower concentration in collagen monomers. The collagen adsorbed layer obtained from a strongly aggregated solution is thus similar to the one observed at low collagen concentration. The build-up of collagen fibrils from the interface is related to the intrinsic self-assembling property of native collagen molecules, as evidenced by the absence of assemblies in the adsorbed phase when denatured collagen is used (Fig. 2b) [18]. High resolution AFM images were acquired to elucidate the structure of the interfacial collagen aggregates, as presented in Fig. 2b. On images obtained under water, some aggregates are terminated by a rootlike structure [18]. This appears even more clearly when imaging is performed on dried samples [21]. This observation points to the formation of fibrils starting from partly adsorbed collagen molecules, whose free segments would assemble. Since assembly is strongly favored by substrate hydrophobicity (see Section 2), it seems that the collagen–substrate interaction may influence the availability of collagen segments for assembly. Further experiments were carried out to compare the collagen–substrate interaction on PS vs PSox. In Fig. 2c, AFM measurements performed on collagen adsorbed on PS and PSox are presented. Force–distance curves were recorded, and showed multiple adhesion events between the AFM tip (made of silicon nitride) and the collagen layer on PS, while this was not the case on PSox [13]. This suggests the presence of free molecular segments on PS, which could bridge the tip. Such free segments would be absent on PSox, where collagen molecules would rather adsorb in closer contact with the substrate. The AFM tip was moreover used as a “molecular broom” [22]: a 5 ␮m × 5 ␮m area was scanned three times at a given applied loading force (7 nN), then a wider area (10 ␮m × 10 ␮m) was imaged to evaluate the damages brought to the adsorbed layer. It clearly appears that the adsorbed layer formed on PS is very sensitive to the action of the AFM tip, while it is unaffected in the same conditions on PSox [10]. These results indicate that the collagen–substrate interaction is stronger on PSox compared to PS, in line with collagen molecules in closer contact with the substrate on PSox. This was further investigated by monitoring the desorption of radiolabeled collagen adsorbed on PS and PSox, either placed in contact with pure buffer or with a non labeled collagen solution. The results are presented in Fig. 2d. Collagen adsorption is essentially irreversible on both substrate, with a desorption in buffer of about 20%. However, exchange of adsorbed molecules by molecules from the solution is observed on PS, where desorption in the presence of collagen is of about 45%, but not on PSox [21]. This further confirms that the collagen–substrate interaction is stronger on PSox compare to PS. On PS, collagen molecules would only be adsorbed through a molecular segment. This leaves the remaining part of the molecule available for assembly with neighboring molecules. This also favors displacement by molecules from the solution. On PSox, collagen molecules would rather lie along the substrate plane, increasing

90

C.C. Dupont-Gillain / Colloids and Surfaces B: Biointerfaces 124 (2014) 87–96

Fig. 2. Understanding the formation of collagen assemblies at interfaces. (a) AFM images (2 ␮m × 1 ␮m, z-range = 10 nm) acquired in air on collagen layers obtained by adsorption on PS from a 40 ␮g/ml collagen solution with increasing degree of aggregation, from top to bottom [adapted from Ref. [17] with permission from Elsevier]. (b) High resolution AFM images (z-range = 15 nm), recorded in tapping mode under water, of native (top, 1 ␮m × 1 ␮m) and denatured (bottom, 1 ␮m × 0.4 ␮m) collagen adsorbed for 30 min on PS, from a 40 ␮g/ml solution. Inset: AFM image (z-range = 15 nm) acquired in tapping mode in air after 24 h of collagen adsorption on PS from a 100 ␮g/ml solution [adapted from Ref. [21] with permission from Langmuir. Copyright 2005 American Chemical Society]. (c) Force–distance curves recorded between the AFM tip and collagen adsorbed for 30 min from a 40 ␮g/ml solution on PS (top) and PSox (bottom), and damages brought by scanning with the AFM tip in collagen layers (2 h, 60 ␮g/ml) on PS (top) and PSox (bottom) (see text for details) [adapted from Ref. [10] with permission from Langmuir. Copyright 2001 American Chemical Society]. (d) Desorption (90 min) of adsorbed radiolabeled collagen (30 min, 40 ␮g/ml) by pure buffer and non labeled collagen in solution, on PS (black bars) and PSox (white bars). (e) Model of adsorbed collagen organization on PS (conditions as indicated) and PSox.

the interaction with the substrate, and decreasing the mobility of molecules, which cannot assemble in such conditions. QCM-D results (not shown here) [21] obtained for collagen adsorption on PS and PSox revealed unusually high f (≈600 Hz on PS; ≈300 Hz on PSox) and D (≈150 × 10−6 on PS; 100 × 10−6 on PSox) values, especially on PS. This is attributed to the length of the collagen molecule (≈300 nm), which can extend in solution and trap water in the adsorbed layer. The more pronounced effect observed on PS is in agreement with the presence of more collagen free segments on that substrate. Fig.2e summarizes the spatial organization of adsorbed collagen on PS and PSox, taking into account the information presented here above. On PS, at low concentration or after a short adsorption time, isolated molecules are partly adsorbed, leaving free segments in solution. Assembly is prevented by the low density of adsorbed molecules. On PS, at higher concentration or after a longer adsorption time, the

density of partly adsorbed molecules is high enough to provoke assembly. On PSox, adsorbed collagen molecules interact more strongly with the substrate, and are in lying position, which restrict their mobility, thereby preventing assembly. It is interesting to point out that the behavior of collagen is in contrast with the one of most globular proteins. It is indeed usually reported that protein adsorption is more reversible on hydrophilic surfaces [4], while the contrary is observed for collagen. This difference can be attributed to the stability of collagen conformation, owing to its triple helical structure, which prevents relaxation at the interface. Soft globular proteins tend to relax on hydrophobic surfaces, leading to irreversible adsorption. Moreover, the higher affinity of collagen for hydrophobic surfaces (see the adsorption kinetics on PS vs PSox in Fig. 1) leads to a higher density of adsorbed molecules in a short time frame, thereby decreasing the time available to interact with the substrate surface.

C.C. Dupont-Gillain / Colloids and Surfaces B: Biointerfaces 124 (2014) 87–96

4. Heterogeneous substrates to spatially direct collagen adsorption and assembly In vivo, cells are in contact with the ECM, which is highly organized. Patterning ECM proteins is a way to mimick this complex organization. Two main approaches can be used to pattern proteins at interfaces. On the one hand, proteins can be selectively deposited on defined areas of a homogeneous substrate. This can be achieved by means of physical contact between this substrate and an elastomeric stamp [23] or the tip of an atomic force microscope [24,25], or alternatively by non-contact printing techniques [26]. On the other hand, protein patterns can be obtained by spontaneous protein adsorption on substrates presenting chemical heterogeneities, with defined areas favoring adsorption while the remaining areas are designed to repel proteins [27]. A variety of methods are nowadays available to produce such chemically heterogeneous substrates, as reviewed elsewhere [28]. For example, heterogeneous substrates consisting in PS inclusions embedded in a protein-repellent matrix of poly(vinylpyrrolidone) were elaborated by dewetting of this latter polymer on a PS substrate, and collagen selectively adsorbed in the PS domains [29]. In many cases, poly(ethylene oxide) (PEO) is used to functionalize the domains preventing protein adsorption. Photolithography and plasma treatment in oxygen were used to selectively oxidize PS substrates in defined areas. The obtained PS/PSox substrates were submitted to collagen adsorption from a solution also containing Pluronic F68, a PEO–poly(propylene oxide)–PEO triblock copolymer. The outcome of the competitive adsorption between collagen and Pluronic

91

was influenced by substrate hydrophilicity: Pluronic adsorption was dominating on PS domains, hence forming a protein-repellent brush of PEO and preventing collagen adsorption, while collagen adsorption was important on PSox domains. In this way, collagen was confined in the micrometer-size oxidized domains designed by photolithography [8,30]. In these examples, the size of the domains favoring adsorption is large compared to the dimensions of the collagen molecule. The supramolecular organization of collagen in the adsorbed phase is thus not influenced by the size of these domains. Domains with smaller dimensions were further used to modulate and direct collagen assembly at interfaces, as presented in Fig. 3. Two different approaches were followed to create submicrometer-scale patterns, and the chemical composition of the involved domains was chosen based on the behavior of collagen on corresponding homogeneous substrates, as presented in Fig. 3a. Collagen was adsorbed on: (i) a monolayer of PEOterminated alkylsilanes assembled on a silicon wafer (hereafter named PEO) [31], (ii) a monolayer of methyl-terminated alkylsilanes assembled on a silicon wafer (hereafter named CH3 ) [31], (iii) a thin spin-coated layer of poly(methyl methacrylate) (PMMA) [32], and (iv) a thin spin-coated layer of PS [32]. On PEO, a sparse network of individual collagen molecules is observed. On PMMA, a relatively smooth collagen layer is found, with small aggregates whose dimensions (height ≈4 nm; length ≈300–600 nm) roughly correspond to the ones of pentamers reported to be formed at the onset of collagen aggregation [1]. On CH3 , a dense layer of short but a bit thicker collagen fibrils (height ≈4–8 nm; length

Fig. 3. Heterogeneous substrates to spatially direct collagen adsorption and assembly. (a) AFM images (5 ␮m × 2.5 ␮m; insets: 750 nm × 375 nm; z-range = 10 nm for PEO and CH3 and 30 nm for PMMA and PS) of collagen adsorbed on homogeneous PEO, CH3 , PMMA and PS substrates. (b) Left: scheme of the patterns (CH3 tracks – dark color – in a PEO matrix – light color) produced by combination of electron beam lithography and silanation, right: AFM images (2 ␮m × 1 ␮m; z-range = 10 nm) of collagen adsorbed on the corresponding patterns (width of tracks = 30 nm (top) and 95 nm (bottom)) [adapted from Ref. [31] with permission from Small. Copyright 2005 Wiley]. (c) Left: scheme of the patterns (PS – dark color – inclusions in a PMMA – light color – matrix and conversely) produced by polymer demixing in thin films, right: AFM images (5 ␮m × 2.5 ␮m; z-range = 30 nm) of collagen adsorbed on the corresponding patterns [adapted from Ref. [32] with permission from Langmuir. Copyright 2012 American Chemical Society].

92

C.C. Dupont-Gillain / Colloids and Surfaces B: Biointerfaces 124 (2014) 87–96

≈300–600 nm) is formed. Finally, on PS, larger collagen structures are formed as described in Section 2, showing several arms of ≈300 nm long and a height of ≈10 nm, associated together via an anchoring knot. X-ray photoelectron spectroscopy (XPS) was also used to determine the surface chemical composition of the same samples. The nitrogen content can be related to the amount of collagen detected at the interface since this element is not present in the substrates. The nitrogen content is the lowest on PEO (4%), in agreement with the expected protein repellency of this substrate, and with the observation of isolated molecules at the interface. Note that the same PEO layer totally prevented albumin adsorption, but collagen is a very large molecule and it is more difficult to avoid its adsorption. An intermediate value (9%) is found on PMMA, pointing to the formation of a layer with a thickness lower than the probed depth (i.e. ≈10 nm). The nitrogen content measured on CH3 and PS (≈16%) is not very far from the theoretical nitrogen content of collagen (≈19% if hydrogen, an element not detected by XPS, is excluded, as calculated from the amino acid sequence of collagen). This indicates that the obtained collagen layers almost fully cover the substrate and reach a thickness at least similar to the probed depth. Starting from the knowledge of the organization of adsorbed collagen on these homogeneous substrates, two types of heterogeneous substrates were created to trigger collagen aggregation in defined domains at the interface. A first strategy (Fig. 3b) consisted in drawing CH3 tracks, with a width in the nanometer range, in a PEO matrix. This was achieved using a combination of electron beam lithography and silane monolayer grafting [31]. A second strategy (Fig. 3c) was based on the phase separation of PS and PMMA in thin films obtained by spin-coating a mixture of these two polymers in different proportions [32]. Inclusions of PS in a PMMA matrix and, conversely, inclusions of PMMA in a PS matrix were obtained. These latter patterned substrates are less ordered than the ones obtained using electron beam lithography, but they offer the advantage of a more easy preparation process, which can be applied over large surface areas. The quality of the obtained substrates was examined using a combination of AFM (shape, dimensions and distribution of patterns) and XPS (surface chemical composition) (results not shown here). AFM used in the lateral force mode was actually used to evidence differences in friction, pointing to the different chemical composition of CH3 and PEO domains [31], while images based on phase contrast obtained in tapping mode AFM were useful to identify PS and PMMA domains [32]. Collagen was adsorbed on these heterogeneous substrates. On CH3 /PEO substrates (Fig. 3b), while almost no contrast was found on AFM height images between the CH3 tracks and the PEO matrix in the absence of collagen (height ≈0.2 nm, not shown here), a marked contrast was observed after collagen adsorption, demonstrating the accumulation of collagen in the CH3 areas. The height of the obtained collagen bundles was 6.5 ± 0.6 nm and was independent of track width. Collagen molecules, which are not much flexible, must be aligned along the tracks when the track width is small (30 nm) compared to the length of the molecule (≈300 nm). When track width increases to 95 nm, it can be observed that part of this alignment is lost, and some molecules escape from the tracks. Hence, the chemical contrast between tracks and matrix, and the anisotropy and small width of the CH3 domains are at the origin of collagen assembly in defined nanotracks [31]. On PS/PMMA substrates, aggregates were clearly formed starting from PS domains. When PS domains were under the form of inclusions in the PMMA matrix, each inclusion served as an anchoring point for a long and large collagen fibril, while only small aggregates were formed on the PMMA matrix, in agreement with the behavior observed on the homogeneous polymers. A similar behavior was found when the matrix was made of PS: collagen fibrils were formed in this matrix, leaving the PMMA inclusions only covered with individual

molecules or small collagen aggregates. Note that the dimensions of the obtained collagen fibrils were similar to the ones found on pure PS when PS formed the matrix of the system, while much longer fibrils were formed when PS formed inclusions. This is attributed to the small diameter of inclusions, which is in the range of the length of collagen molecules. The nitrogen content measured by XPS on these heterogeneous PS/PMMA substrates after collagen adsorption corresponds to the average, weighted by the surface coverage of each polymer, of the nitrogen content found in similar conditions on the pure polymers. Again, the chemical nature and the geometry of the PS and PMMA domains govern collagen organization at the interface [32]. To summarize, chemically heterogeneous substrates are useful to confine collagen in given domains at interfaces. Moreover, when the size of the domains favoring adsorption is in the range of the length of the collagen molecules or lower, such heterogeneous substrates also influence collagen assembly at interfaces.

5. Temporal control of collagen adsorption using mixed polymer brushes While the spatial control of collagen adsorption may lead to the development of biointerfaces mimicking the complex geometry of ECM, it may also be useful to trigger collagen adsorption or desorption at a given time point, enabling the control of cell processes over time. This can be achieved using so-called smart or stimuli-responsive coatings, whose properties may be altered upon application of an appropriate stimulus. The development of substrates able to reversibly adsorb and desorb proteins may be based on the switchable exposure of moieties favoring and preventing adsorption, respectively. Mixed polymer brushes are well suited to design this kind of smart coatings. We have synthesized mixed brushes of PEO and poly(acrylic acid) (PAA) using chemisorption on gold of polymers functionalized with sulfur-containing groups (thiol-terminated PEO; PAA with a disulfide bond in its center). An intimate mixing of the two polymers was obtained at the interface, and the polymer density was high enough to ensure the formation of a brush [33,34]. While PEO is known for its protein-repellent properties, PAA is a polyelectrolyte whose conformation depends on pH and ionic strength (I) of the surrounding medium. Depending on the conditions, the PAA chains may be shrunk or swollen, respectively exposing or hiding adjacent PEO molecules from the outmost surface. Hence, changes of pH and I may be used to trigger protein adsorption or desorption. The useful values of pH and I must be selected taking into account their influence on PAA conformation, but also the protein characteristics [34]. In the case of collagen, this is quite complex since the state of aggregation of collagen in solution strongly depends on pH and I. Preliminary experiments were thus carried out to identify the best conditions for adsorption and desorption [35]. Fig. 4 illustrates the behavior of mixed PEO/PAA brushes toward collagen adsorption and desorption. Adsorption was performed at pH 9 and I = 10−1 M, while the conditions used for desorption were pH 3 and I = 10−3 M. Quartz crystal microbalance was used to monitor in situ two successive cycles of adsorption and desorption, and the protein adsorbed amount was evaluated by XPS (through the nitrogen content) after adsorption solely and after successive steps of adsorption and desorption. On pure PEO brushes, almost no collagen adsorption was observed, as expected. A strong and irreversible adsorption of collagen was recorded on pure PAA brushes. Mixed PEO/PAA brushes showed the desired behavior, i.e. collagen adsorption was almost as high as on pure PAA brushes, but desorption was very significant, and increased with the PEO content in the brush. Cycles of adsorption/desorption were successfully repeated in QCM experiments [35].

C.C. Dupont-Gillain / Colloids and Surfaces B: Biointerfaces 124 (2014) 87–96

Fig. 4. Mixed polymer brushes to switch collagen adsorption on and off. Evolution of the f measured using QCM (a) and of the nitrogen mole fraction (%N) measured by XPS (b) upon adsorption (pH 9, I = 10−1 M – dark green bars) and desorption (pH 3, I = 10−3 M – light green bars) of collagen (25 ␮g/ml) on a range of PEO/PAA brushes (fraction of repeating units of PEO and PAA in the mixed brushes as indicated). Two successive cycles of adsorption and desorption were monitored by QCM. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

To sum up, mixed PEO/PAA brushes were successfully synthesized and used to reversibly and repeatedly adsorb and desorb collagen, opening the way for the temporal control of collagen-dependent cell processes at interfaces. 6. Collagen-based nanotubes To create biomimetic environments for cell and tissue engineering, it is necessary to develop methods allowing collagen-based fibers with controlled dimensions to be produced. Spontaneous self-assembly of collagen may of course be used to obtain fibers with different characteristics, but more sophisticated micro- or nano-objects could also be synthesized, bringing additional functionality. Electrospinning is the most widely used technique to produce fibers with polymers or proteins. It was successfully applied to the production of collagen [36] or collagen-based fibers [37]. An alternative approach consists in depositing a collagenbased coating within the pores of a membrane used as a template, and then to dissolve the membrane, thereby obtaining nanotubes. This gives the opportunity to precisely control the dimensions of the designed structures (length of the tubes = thickness of the membrane; diameter of the tubes = diameter of the pores). Moreover, the inner void of the nanotubes could be used further to encapsulate active molecules (growth or differentiation factors; drugs) which could be delivered to cells. Collagen-based coatings were synthesized using the layerby-layer approach, in which a polycation and a polyanion are alternately adsorbed. Here, collagen was used as the polycation and poly(styrene sulfonate) (PSS) as the polyanion. The success of assembly was first demonstrated on plane substrates [38]. Assembly was then performed within the pores of polycarbonate track-etched membranes to build nanotubes, as depicted in Fig. 5a. The successful assembly of (PSS/collagen)6 in templates with pores of a diameter of 200 and 500 nm was first demonstrated by gas flow porometry [39], and a multilayer thickness of 30 nm and 80 nm was determined. This is by far thicker than multilayers obtained in the same conditions on a flat substrate, as previously observed for (PAH/PSS) assemblies [40]. It is believed that macromolecules can

93

be interconnected across the pores of the template, forming a gel which fills the pores when hydrated. The thinner layer obtained in 200 nm pores may be attributed to the more limited diffusion of 300 nm-long collagen molecules in that case. The polycarbonate template was then dissolved in dichloromethane or dimethylformamide (DMF) to free the nanotubes (Fig. 5a). To demonstrate the synthesis of the nanotubes, the dissolution was performed on top of a silver membrane which was subsequently imaged by scanning electron microscopy (SEM), as shown in Fig. 5b for nanotubes obtained in a 21 ␮mthick template with 200 nm-diameter pores. The dimensions of the observed nanotubes were in agreement with the template geometry. Transmission electron microscopy images also revealed the hollow structures of the formed nanotubes [38]. Electrophoretic deposition (EPD) was used to immobilize these collagen-based nanotubes at the surface of indium-tin oxide (ITO)coated glass, as illustrated in Fig. 5a [41]. Briefly, the nanotube suspension (≈6 × 106 nanotubes/ml in DMF) was filled in a cell formed by two ITO glass slides, serving as the working and the counter electrodes, separated by a 5 mm-thick silicone spacer. A platinum wire was used as the pseudo-reference electrode. The EPD procedure was carried out for 1000 s at different applied voltages, as illustrated in Fig. 5c. The nanotubes could be observed using fluorescent microscopy owing to the introduction of a fluorescently labeled polyelectrolyte (Flu-PAH) in the core of the nanotubes, which were obtained using (collagen/PSS)3 /(Flu-PAH/PSS)3 assembly. A homogeneous distribution of mostly individual nanotubes was obtained using EPD. A clear correlation between applied voltage and density of immobilized nanotubes was found, giving the opportunity to tune the surface coverage by collagen-based nanotubes. The presence of collagen at the outermost surface of these nanotubes is required to provide signaling cues to cells. Time-offlight secondary ion mass spectrometry (ToF-SIMS) imaging was used to examine the chemical composition of the extreme surface of nanotubes deposited on ITO glass, as illustrated in Fig. 5d. The signal of indium (from ITO glass), which is screened by the deposited nanotubes, can be used to localize them at the interface (Fig. 5d, left). The signal from secondary ions attributed to amino acids was used to build an image, which demonstrates the presence of collagen in the outer layer of the nanotubes (Fig. 5d, right). To summarize, layer-by-layer assembly in the nanopores of a template was used to synthesize collagen-based nanotubes with controlled dimensions, which were immobilized at the surface of ITO glass using electrophoretic deposition. The designed biointerfaces are decorated with a tunable density of nanotubes, which present collagen molecules at their outer surface, and can be loaded with bioactive compounds.

7. Cell behavior on collagen-based biointerfaces As detailed here above, a good control over collagen organization at interfaces can be achieved, owing to a better understanding of its mechanisms of adsorption and assembly, as well as to the development of substrates which spatially or temporally direct collagen adsorption, and to the design of collagen-based nanoobjects. The tailored biointerfaces may be used to interfere with cell behavior. Fig. 6 summarizes our achievements in terms of control of cell–material interactions using collagen-based biointerfaces. In a first approach, PSox tracks of supracellular width, embedded in a PS matrix, were designed to selectively adsorb collagen in the presence of Pluronic (see Section 4 and Fig. 6a). Rat hepatocytes (primary culture) were seeded on the conditioned substrate, and were shown to exclusively adhere in the collagen-modified tracks, where adhesion signals are available for specific binding with cell membrane receptors (integrins). The same behavior was observed with other cell types as well [30]. Since the shape of cells

94

C.C. Dupont-Gillain / Colloids and Surfaces B: Biointerfaces 124 (2014) 87–96

Fig. 5. Collagen-based nanotubes. (a) Scheme of the process used for nanotube synthesis and collection. (b) SEM image of collagen-based nanotubes [reproduced from Ref. [39] with permission from The Royal Society of Chemistry]. (c) Fluorescence microscopy images of nanotubes collected by EPD at different voltages as indicated. (d) ToF-SIMS images of collagen-based nanotubes deposited on ITO glass by EPD. Left: indium signal, right: sum of signals from amino acids [adapted from Ref. [41] with permission from Biomacromolecules. Copyright 2011 American Chemical Society].

may modify their fate [42], this approach could be useful to keep cells in or bring cells to a given state of differentiation. Other similar approaches, based on the modification of defined areas of supracellular size with extracellular matrix proteins, have been reported in the literature [29,42–44]. Then, the effect of the supramolecular organization of collagen on the adhesion and spreading of endothelial cells (HUVEC – human umbilical vein endothelial cells) was examined (Fig. 6 b and c) [45]. Collagen layers with an increasing density of assemblies were prepared (Fig. 6b). The sparse collagen layer obtained after 1 min of adsorption with 7 ␮g/ml of collagen did not support cell adhesion, and only a few cells with round shape were found to adhere after 4 h of incubation with the substrate. The more continuous collagen layers, showing a low density of fibrils, obtained by adsorption during 24 h of a 7 ␮g/ml collagen solution, did on the contrary favor cell adhesion, which is attributed to the increased number of available collagen molecules, providing adhesion sequences to the cells. Cells adopted a very spread shape, with many actin stress fibers oriented along the cell edges. After collagen adsorption for 24 h from a 100 ␮g/ml solution, cells adhered as well to the substrate, but could however not spread, keeping a round shape. This can be related to the presence of a high density of collagen assemblies at the interface. These assemblies may either lead to a lower exposure of recognition sequences or modify their spatial distribution. The effect of collagen supramolecular organization, independently of the adsorbed amount, was also highlighted by comparing the behavior of endothelial cells on collagen layers which were wet,

dried, or dewetted (see Section 2). Fig. 6c shows that the wet collagen layer does not support cell adhesion. This can be attributed to the weak mechanical properties of the collagen layer. On a continuous collagen layer obtained by drying this wet layer, cell adhesion is good, and cell spreading is moderate. When the same wet collagen layer is dried at slow rate and undergoes dewetting, the obtained discontinuous collagen pattern provokes a very significant increase of cell spreading. This again evidences the influence of the spatial distribution of adhesion sequences on cell behavior. To further understand the role of collagen assemblies on cell behavior, MC3T3 preosteoblasts were seeded on a range of heterogeneous PS/PMMA substrates obtained by demixing of these polymers [46]. These substrates were shown to modulate the distribution and size of collagen aggregates at the interface (see Section 4). Fig. 6d shows the evolution of cell density after 4 h and 6 days in culture, on PS/PMMA substrates with 0–100% of PS in the surface layer, and either submitted or not to collagen adsorption. After 4 h in culture, almost no cells adhere on naked substrates, and a much better adhesion is observed when collagen is present at the interface and provides cells with adhesion sites. Heterogeneous surfaces do moreover better support cell adhesion compared to pure polymers, which can be related to the organization of collagen assemblies. After 6 days in culture, the difference between naked and collagen-treated substrates is less clear. This is attributed to the fact that cells had time to secrete their own extracellular matrix compounds, and to remodel the previously adsorbed layer. Substrate heterogeneity does however still play a role, with the

C.C. Dupont-Gillain / Colloids and Surfaces B: Biointerfaces 124 (2014) 87–96

95

Fig. 6. Collagen-based tailored biointerfaces influence cell behavior. (a) Optical microscope image of hepatocytes adhering selectively to PSox tracks of a PS/PSox heterogeneous substrate treated with collagen and Pluronic. The principle of the method is illustrated [adapted from Ref. [30] with permission from Elsevier]. (b) Fluorescence (red = actin) and (c) optical microscopy images of HUVECS cells after 4 h in culture on PS with adsorbed collagen (see text for details; insets: AFM images of the corresponding collagen layers). The determined cell density is given [adapted from Ref. [45] with permission from Journal of Biomedical Materials Research Part A. Copyright 2005 Wiley]. (d) Preosteoblast cells density on a range of PS/PMMA substrates, after 4 h and 6 days in culture, and in the absence or presence of adsorbed collagen. A scheme of the PS/PMMA surfaces and corresponding AFM images after collagen adsorption are given below the graph. (e) Fluorescence microscopy image of preosteoblast cells grown on a biointerface decorated with collagen-based nanotubes (blue = nuclei; red = actin; green = nanotubes) [adapted from Ref. [41] with permission from Biomacromolecules. Copyright 2011 American Chemical Society]. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

better cell adhesion found for substrates with PS inclusions in a PMMA matrix. On these substrates, the density of collagen assemblies is lower compared to pure PS, and these assemblies are longer. The distribution of extracellular compounds secreted by the cells and adsorbed on the substrate may similarly be influenced by substrate heterogeneity. Cell spreading was shown as well to depend on PS fraction at the interface, and was dependent on the presence of preadsorbed collagen after 4 h but much less after 6 days (not shown here). Finally, the interaction between MC3T3 preosteoblasts and biointerfaces decorated with collagen-based nanotubes (see Section 6) was examined. Some cells were observed to directly interact

with the nanotubes through filopodia, as shown in Fig. 6e (see white circle). Changes in cell morphology were linked to the presence of the nanotubes, while cell adhesion, attachment and growth were not modified on ITO glass functionalized with nanotubes compared to virgin ITO glass. The nanotubes thus interact with cells and do not bring any obvious cytotoxicity, and these biointerfaces may then be used to trigger desired cell behaviors by filling the nanotubes with signaling molecules which will be released locally to the cells [47]. In summary, collagen-based biointerfaces can be used to control cell–material interactions. Cell behavior may be influenced by the supracellular distribution of collagen, confining cells in defined areas, as well as by the subcellular organization of collagen, since the state of

96

C.C. Dupont-Gillain / Colloids and Surfaces B: Biointerfaces 124 (2014) 87–96

aggregation of collagen at interfaces is affecting the distribution and exposure of adhesion sequences. Collagen-based nanotubes are finally shown to positively interact with cells, opening the way to the design of multifunctional interfaces for the control of cell behavior. 8. Perspectives Our knowledge of collagen adsorption and assembly at interfaces has considerably increased over the last twenty years. This can be mainly related to the advent of atomic force microscopy and related techniques, which have revolutionized our ability to observe adsorbed proteins at the (supra)molecular scale and in different media. There is however still progress to be made, especially to better understand collagen adsorption and assembly at the molecular level. In situ methods such as surface-sensitive infrared spectroscopy, QCM and ellipsometry may help advancing in that direction. Regarding collagen-based biointerfaces, we should increase their complexity, associating other ECM components to collagen, and combining them with drug delivery approaches. This would lead to the development of multifunctional platforms for tissue engineering applications. Moreover, such tailored interfaces mimicking ECM could also be used for cancer research [48] or drug development studies (for example, screening of anti-matrix metalloproteinases). The knowledge and control of 2D organization of ECM compounds should be extended to 3D-systems, which are more representative of cell environments [49]. The cell–biointerface interaction could be directly probed using force spectroscopy with cell-modified probes [50]. Finally, instead of creating biointerfaces mimicking the ECM, substrates could be developed that direct adsorption of ECM compounds secreted by the cells, and their remodeling. Our knowledge of collagen adsorption and assembly mechanisms could serve as a guide for the design of such substrates. Acknowledgements I wish to express here my profound gratitude to Professor Paul Rouxhet, who supervised my Ph.D. thesis then always encouraged me and helped me in my quest for new knowledge. The work presented here started under his impulsion, and today, it is still a great pleasure to discuss new results or ideas with him. Many researchers have contributed to this work throughout the years. I thank all of them for the quality of their results and the stimulating discussions. This work has benefited from the financial support of Belgian National Foundation for Scientific Research (FNRS), and from BELSPO through interuniversity attraction pole programs. References [1] K.A. Piez, in: J.I. Kroschwitz (Ed.), Encyclopedia of Polymer Science and Engineering, vol. 3, Wiley, New York, 1985, pp. 699–727. [2] G.A. Di Lullo, S.M. Sweeney, J. Körkkö, L. Ala-Kokko, J.D. San Antonio, J. Biol. Chem. 277 (2002) 4223. [3] K.E. Kadler, D.F. Holmes, J.A. Trotter, J.A. Chapman, Biochem. J. 316 (1996) 1. [4] J.L. Brash, T.A. Horbett, Proteins at interfaces II, in: J.L. Brash, T.A. Horbett (Eds.), ACS Symposium Series, vol. 602, American Chemical Society, Washington, DC, 1995, pp. 1–23. [5] W. Norde, J.L. Brash, T.A. Horbett, Proteins at interfaces III: state of the art, in: T.A. Horbett, J.L. Brash, W. Norde (Eds.), ACS Symposium Series, vol. 1120, American Chemical Society, Washington, DC, 2012, pp. 1–34. [6] M. Mertig, U. Thiele, J. Bradt, G. Leibiger, W. Pompe, H. Wendrock, Surf. Interface Anal. 25 (1997) 514.

[7] S. Rössler, D. Scharnweber, C. Wolf, H. Worch, J. Adhes. Sci. Technol. 14 (2000) 453. [8] J.-L. Dewez, V. Berger, Y.-J. Schneider, P.G. Rouxhet, J. Colloid Interface Sci. 191 (1997) 1. [9] G. Penners, Z. Priel, A. Silberberg, J. Colloid Interface Sci. 80 (1981) 437. [10] C.C. Dupont-Gillain, P.G. Rouxhet, Langmuir 17 (2001) 7261. [11] V.M. De Cupere, P.G. Rouxhet, Surf. Sci. 491 (2001) 395. [12] F.A. Denis, P. Hanarp, D.S. Sutherland, J. Gold, C. Mustin, P.G. Rouxhet, Y.F. Dufrene, Langmuir 18 (2002) 819. [13] E. Pamula, V. De Cupere, Y.F. Dufrene, P.G. Rouxhet, J. Colloid Interface Sci. 271 (2004) 80. [14] I. Jacquemart, E. Pamula, V.M. De Cupere, P.G. Rouxhet, C.C. Dupont-Gillain, J. Colloid Interface Sci. 278 (2004) 63. [15] C.C. Dupont-Gillain, E. Pamula, F.A. Denis, V.M. De Cupere, Y.F. Dufrêne, P.G. Rouxhet, J. Mater. Sci.: Mater. Med. 15 (2004) 347. [16] C.C. Dupont-Gillain, E. Pamula, F.A. Denis, P.G. Rouxhet, Progr. Colloid Polym. Sci. 128 (2004) 98. [17] C.C. Dupont-Gillain, I. Jacquemart, P.G. Rouxhet, Colloid Surf. B 43 (2005) 179. [18] E. Gurdak, J. Booth, C. Roberts, P.G. Rouxhet, C.C. Dupont-Gillain, J. Colloid Interface Sci. 302 (2006) 475. [19] C.C. Dupont-Gillain, I. Jacquemart, Surf. Sci. 539 (2003) 145. [20] C.C. Dupont-Gillain, P.G. Rouxhet, Nano Lett. 1 (2001) 245. [21] E. Gurdak, C.C. Dupont-Gillain, J. Booth, C.J. Roberts, P.G. Rouxhet, Langmuir 21 (2005) 10684. [22] A.S. Lea, A. Pungor, V. Hlady, J.D. Andrade, J.N. Herron, E.W. Voss, Langmuir 8 (1992) 68. [23] H.W. Li, B.V.O. Muir, G. Fichet, W.T.S. Huck, Langmuir 19 (2003) 1963. [24] D.L. Wilson, R. Martin, S. Hong, M. Cronin-Golomb, C.A. Mirkin, D.L. Kaplan, Proc. Natl. Acad. Sci. U.S.A. 98 (2001) 13660. [25] K.B. Lee, J.H. Lim, C.A. Mirkin, J. Am. Chem. Soc. 125 (2003) 5588. [26] L. Ceriotti, L. Buzanska, H. Rauscher, I. Mannelli, L. Sirghi, D. Gilliland, M. Hasiwa, F. Bretagnol, M. Zychowicz, A. Ruiz, S. Bremer, S. Coecke, P. Colpo, F. Rossi, Soft Matter 5 (2009) 1406. [27] D. Falconnet, D. Pasqui, S. Park, R. Eckert, H. Schift, J. Gobrecht, R. Barbucci, M. Textor, A novel approach to produce protein nanopatterns by combining nanoimprint lithography and molecular self-assembly, Nano Lett. 4 (2004) 1909. [28] D. Falconnet, G. Csucs, M. Grandin, M. Textor, Surface engineering approaches to micropattern surfaces for cell-based assays, Biomaterials 27 (2006) 3044. [29] S.C. Thickett, J. Moses, J.R. Gamble, C. Neto, Soft Matter 8 (2012) 9996. [30] J.L. Dewez, J.B. Lhoest, E. Detrait, V. Berger, C.C. Dupont-Gillain, L.M. Vincent, Y.J. Schneider, P. Bertrand, P.G. Rouxhet, Biomaterials 19 (1998) 1441. [31] F.A. Denis, A. Pallandre, B. Nysten, A.M. Jonas, C.C. Dupont-Gillain, Small 1 (2005) 984. [32] E. Zuyderhoff, C.C. Dupont-Gillain, Langmuir 28 (2012) 2007. [33] M. Delcroix, G. Huet, T. Conard, S. Demoustier, F.E. Du Prez, J. Landoulsi, C.C. Dupont-Gillain, Biomacromolecules 14 (2013) 215. [34] M.F. Delcroix, S. Demoustier-Champagne, C.C. Dupont-Gillain, Langmuir 30 (2014) 268. [35] M.F. Delcroix, S. Laurent, G.L. Huet, C.C. Dupont-Gillain, Acta Biomater. (in press). [36] C.R. Carlisle, C. Coulais, M. Guthold, Acta Biomater. 6 (2010) 2997. [37] C. Huang, R. Chen, Q. Ke, Y. Morsi, K. Zhang, X. Mo, Colloids Surf. B 82 (2011) 307. [38] J. Landoulsi, C.J. Roy, C.C. Dupont-Gillain, S. Demoustier-Champagne, Biomacromolecules 10 (2009) 1021. [39] J. Landoulsi, S. Demoustier-Champagne, C. Dupont-Gillain, Soft Matter 7 (2011) 3337. [40] C.J. Roy, C. Dupont-Gillain, S. Demoustier-Champagne, A.M. Jonas, J. Landoulsi, Langmuir 26 (2010) 3350. [41] D. Kalaskar, C. Poleunis, C.C. Dupont-Gillain, S. Demoustier-Champagne, Biomacromolecules 12 (2011) 4104. [42] R. Singhvi, A. Kumar, G.P. Lopez, G.N. Stephanopoulos, D.I.C. Wang, G.M. Whitesides, D.E. Ingber, Science 264 (1994) 696. [43] M. Charnley, F. Anderegg, R. Holtackers, M. Textor, P. Meraldi, PLoS ONE 8 (2013) e66918. [44] R. Lovchik, C. Von Arx, A. Viviani, E. Delamarche, Anal. Bioanal. Chem. 390 (2008) 801. [45] Z. Keresztes, P.G. Rouxhet, C. Remacle, C.C. Dupont-Gillain, J. Biomed. Mater. Res. 76A (2006) 223. [46] E. Zuyderhoff, Ph.D. thesis, Université catholique de Louvain, 2012. [47] D. Kalaskar, S. Demoustier-Champagne, C.C. Dupont-Gillain, Colloids Surf. B 111 (2013) 134. [48] D. Docheva, D. Padula, M. Schieker, H. Clausen-Schaumann, Biochem. Biophys. Res. Commun. 402 (2010) 361. [49] M.S. Hall, R. Long, X. Feng, Y. Huang, C.Y. Hui, M. Wu, Exp. Cell Res. 319 (2013) 2396. [50] J. Friedrichs, J. Helenius, D.J. Muller, Nat. Protoc. 5 (2010) 1353.

Understanding and controlling type I collagen adsorption and assembly at interfaces, and application to cell engineering.

Collagen is a large anisotropic and self-assembling extracellular matrix protein. Understanding and controlling its adsorption and assembly at interfa...
3MB Sizes 0 Downloads 3 Views