J. Phys. Chem. B 2006, 110, 14581-14589
14581
Novel Method to Prepare Morphologically Rich Polymeric Surfaces for Biomedical Applications via Phase Separation and Arrest of Microgel Particles Iseult Lynch,*,† Ian Miller,‡ William M. Gallagher,‡ and Kenneth A. Dawson† Irish Centre for Colloid Science & Biomaterials, Department of Chemistry, UniVersity College Dublin, Belfield, Dublin 4, Ireland, and Conway Institute of Biomolecular and Biomedical Research, Department of Pharmacology, UniVersity College Dublin, Belfield, Dublin 4, Ireland ReceiVed: February 23, 2006; In Final Form: May 19, 2006
We outline here a simple method to prepare polymeric surfaces of controlled surface topography on the micrometer scale, via assembly and arrest of microgel particles, for use in a range of biological applications to modify cell adhesion and spreading. In previous work by other groups, it has transpired that topography on the nanoscale is unlikely to be useful for this purpose, as roughness on this scale is often covered or coated by serum derived proteins during the early stages of cell adhesion and cells can easily bridge nanoscale roughness. Therefore, in our work, we have focused on roughness or topographic variations on the micrometer length scale. The basic idea is to modify the interactions between particles, thereby causing the microgel particles to phase separate into particle-dense and particle-dilute domains and to arrest these domains on the surface. The result is the creation of surfaces with controlled topography. By changing the particle size, it is possible to alter the size of the pores formed and their distribution in the film. Preliminary results show that the system can readily be arrested into a homologous series of such structures (formed from microgel particles of the same size and same chemical structure) with biological implications. At the extremes of this series, large phenotypic differences are observed between cells, ranging (at one end) from localization of the cells in the pores to (at the other end) cells that avoid such localization, and remain extended, growing along the ridges between the pores. This constitutes a sort of cell localization transition on a surface with identical chemical components, where only the morphology has been adjusted.
Introduction Cell-biomaterial interactions determine the viability or otherwise of cells on a particular surface and thus the success or failure of a particular material as, for example, a medical device coating or an antibacterial surface. Thus, if an understanding of the factors controlling cell-biomaterial interactions can be reached, then it will be possible to design surfaces whose properties range from ideal for cell adherence and proliferation, to surfaces which are repellent to cells and cause cell death, to surfaces that are selectively adhesive to particular cell types. The nature of a biomaterial surface is known to play a key role in the processes of cell adhesion, spreading, growth, and migration. Surface characteristics such as hydrophobicity, surface energy, texture, surface charge, and chemical composition are all known to affect the phenotypic response of interacting cells.1 It is well-known that cells attach to surfaces via the formation of focal contact points, the formation, density, and distribution of which regulates cell adhesion.2,3 Surface topography may affect many cellular responses to materials, such as the inflammatory response at an implanttissue interface, fibroblast attachment to the surface, angiogenesis, and a host of other cellular processes such as cellular differentiation, DNA/RNA transcription, cell metabolism, and protein production.4,5 There is significant evidence to suggest that the ability of a cell to form good contacts with a surface (or actually with the * To whom correspondence should be addressed. E-mail: Iseult@ fiachra.ucd.ie. Corresponding author is also at Physical Chemistry 1, Lund University, P.O. Box 124, 22100, Lund. † Department of Chemistry, University College Dublin. ‡ Department of Pharmacology, University College Dublin.
extracellular matrix (ECM) deposited on the surface) determines the cell viability and that cell foreshortening caused by the dissolution of the ECM may be the signal that initiates the celldeath program.6 This possibility is supported by the finding that endothelial cells spread and grow on large (>100 µm diameter) microcarrier beads7 whereas they rapidly die when bound to small (4.5 µm) ECM-coated beads8 due to the lack of possibility for cell extension.9 Chen et al. found that confining the spreading of human capillary endothelial cells to 10 and 20 µm circles resulted in significantly more of the cells undergoing apoptosis than cells grown on identical substrates where their spreading was not confined.10 Chen et al. also looked at the spreading of a single cell across multiple closely spaced adhesive islands of either 3 or 5 µm diameter (to approximate the size of an individual focal adhesion) and found that cell bodies spread across the intervening nonadhesive areas of the substrate, forming vinculin-containing focal adhesions only on the adhesive islands. By changing the spacing between the islands, they could increase cell spreading 10-fold without significantly altering the total cell-ECM contact area.10 Interestingly, DNA synthesis scaled with projected cell area not with cell-ECM contact area. Also, apoptosis was switched off by cell spreading, even though the cell-ECM contact area remained constant. Thus, cell shape per se appears to be one of the critical determinants in switching between life and death and between proliferation and quiescence. Lee et al. showed that fibroblasts cultured in vitro on polycarbonate membranes with well-defined surface topography (track-etched micropores, 0.2-8.0 µm in diameter) displayed gradually decreasing cell adhesion and growth with increasing
10.1021/jp061166a CCC: $33.50 © 2006 American Chemical Society Published on Web 07/12/2006
14582 J. Phys. Chem. B, Vol. 110, No. 30, 2006 micropore size on hydrophilic surfaces and almost no adhesion at all on the hydrophobic microporous surfaces.11 It seems that the fibroblasts had difficulty overriding the surface discontinuities if the micropores were large but that they could easily override small micropores. From this summary of the literature, it is clear that although there are many unknowns (and the whole arena is not clearly framed as a quantitative research topic), control of surface morphology has the power to profoundly impact the nature of intra- and intercellular processes, with all the implications of biomedical research. Thus, it has transpired that topography on the nanoscale is unlikely to be of much additional use as a control parameter in determining the cell-material interactions. Roughness on this scale is often covered or coated by serumderived proteins during the early stages of cell adhesion, and cells can easily bridge nanoscale roughness. At that scale, therefore, much of the interest revolves around the detailed chemical composition of the surface and the nature of the adsorbed-protein layer to which that leads.12-14 This is not to say that there are no consequences of nanoscale roughnesss merely that these are less dramatic than the role of protein adsorption, and are much affected by the chemical details, and protein environment. Roughness on a longer topographic length-scale regime (up to several micrometers) is, however, linked to many aspects of the intracellular machinery of a biological cell. This is, in one sense, to be expected. A typical animal cell is on the order of 5 µm, and the organelles (subcellular components) are substantial fractions of that total. Also, information from tissue engineering studies is suggestive of the likely useful role of this length scale in controlling the overall biological response in the cell-material interaction. Thus, many cells reside in a matrix with voids on the scale of many tens of nanometers, and this has led to numerous attempts to mimic such matrixes.15,16 Phenomenological information from a wide variety of studies suggests that control of the external stress experienced by cells is a key factor affecting their metabolism, signaling, and other processes. How and why different cell types react differently to surface roughness is not understood, but a different response to such stress could be one explanation. The practical implications of this are clear. There are already suggestions that the performance of implant materials can be improved using slightly roughened surfaces, which result in increased adhesion of connective tissue cells and closer apposition of tissue to the implant.1 Such observations support our view that there is a substantive new role for controlled surface morphology in medically related materials. Currently, however, the field is hampered by a lack of a quantitative understanding of the role of surface topography on cell behavior. Some have suggested that the lack of defined surfaces with uniform morphologies is a limitation.11 Cellular behavior can change dramatically with small changes in the environment, and finding the critical regime is not always easy, requiring laborious development of many different samples. We also emphasize the lack of a homologous series of materials with controlled surface morphology and identical chemical constitution, for this is essential in separating the effects from two different sources. It is difficult to find systems where morphology can be changed without altering other (surface or chemical) parameters simultaneously. Such systems could ideally be formed from a range of different materials and compositions, producing a wide range of homologous series of materials.
Lynch et al. In this paper, we frame such an approach using a concept that has recently attracted a lot of attention in the fundamental science of materials.17,18 That is, we can use control of attractions between particles to create gels, or other more complex morphologies, and then seek to “dynamically arrest” them to make new materials. For example, depletion-induced attractions can cause microgel particles to phase separate into particledense and particle-dilute domains. The kinetics of this separation process is somewhat understood and (in the unstable regime) is driven by unstable compositional fluctuations of more and less dense particles regions of quite well-defined wavelength.19,20 If the attraction is sufficiently short-ranged and strong, then these kinetically determined structures can be “arrested” or frozen in. The strength and range of the depletion-induced attraction are (respectively) determined by the concentration and molecular weight of the added (free) polymer. Subsequent (or simultaneous) drying of the sample complicates the theoretical description of this whole process, but porous matrixes can be produced nevertheless. It is interesting that control of the particle size, concentration of the added free polymer, and (to some extent) the rate of drying of the dispersion allow us to control the morphology in terms of both the size of the pores formed and their distribution in the film. It is also worth noting that previous studies of cast films of the same free polymers exhibited no porous structures under a broad range of comparable conditions.13,14,21 This and the fact that we work in a solvent where water is miscible (especially at the very low concentration present in pure ethanol) make it less likely that the structures we observe here are connected to “breath figures” (formed by precipitation of polymer at the interface of assembled water droplets in the solvent) that have been reported in the literature.22,23 We have previously shown that by changing the composition of the polymers used to make the microgel particles, we can alter the particle hydrophobicity and the surface composition.21 We have used microgel particles composed of 50% N-tertbutylacrylamide (NtBAM) and 50% N-isopropylacrylamide (NiPAM), that is, 50:50 NtBAM/NiPAM microgel particles, as the interactions between cells, and flat surfaces made from linear copolymers of this composition have been extensively characterized by us previously.13,14 Thus, any changes in the behavior of the cells on surfaces made from the microgel particles compared with those on flat surfaces made from the linear copolymers can only result from changes in the physical topography of the surface (that is, physical stress induced effects) and not from any chemical or compositional difference. We have focused exclusively on particles with a ratio of 50:50 NtBAM/ NiPAM in the present work, as this composition was determined previously to promote the highest level of cell adhesion and spreading of the linear copolymer films studied, as well as having the largest impact on the gene expression.13,14 Cross-linked microgel particles in a range of sizes (70-700 nm diameters) were prepared and mixed with a compositionally equivalent linear 50:50 poly(NtBAM-co-NiPAM) copolymer, which can be considered as a nonadsorbing polymer under the conditions used (i.e., in ethanol). In this work, we focus on a detailed characterization of the structure and morphology of the particle surfaces and present only brief results of the behavior of cells on the surfaces. Surfaces were characterized using atomic force microscopy (AFM), fluorescence microscopy, and confocal microscopy, and the effect of the microgel particle size and amount and the amount of free polymer added on the pore size and morphology of the surfaces was determined. Initial analysis of macroscopic cellular responses (cell spreading)
Morphologically Rich Polymeric Surfaces indicate that there are indeed profound differences in the behavior of cells on rough or porous surfaces compared with those on the smooth copolymer surfaces, which can only be due to stress factors introduced by the surface morphology, such as the limited or directed (i.e., along a ridge) growth induced by the morphological features of the surface. Experimental Details Materials. N-Isopropylacrylamide (NiPAM) monomer (purity >99%) from Phase Separations Ltd. (Clwyd, U.K.) and N-tertbutylacrylamide (NtBAM) from Fluka (Dorset, England) were recrystallized twice from hexane before use. N,N1-Methylenebisacrylamide (BisAM) (purity >99.5%) from Fluka (Dorset, England), ammonium peroxydisulfate (APS) (purity 99.99%) from Aldrich (Dorset, England), and N,N,N1,N1-tetramethylethylenediamine (TEMED) from Sigma (Dorset, England) were used as supplied. Microgel Particle Synthesis. Microgel particles composed of 50% NiPAM and 50% NtBAM (w/w) were synthesized by emulsion polymerization in water at 60 °C, with 10% of total monomer weight of bis-acrylamide as the cross-linker, ammonium persulfate as the initiator, and Triton X-100 as the surfactant, as described previously.24 As the structures of the two monomers are very similar (one H in NiPAM is replaced by a CH3 in NtBAM) random copolymer particles result from this synthesis method. Through the use of this procedure, microgel particles of diameter 200 nm were formed. To make particles of 50 nm, 400 nm, and 700 nm, SDS was used as the surfactant, with the amount of SDS used controlling the particle size.25 The SDS micelle size is controlled by the SDS concentration (with larger numbers of smaller micelles forming with increasing SDS concentration). As the growing polymers are hydrophobic (the reaction is carried out at 60 °C well above the collapse temperature of poly(NtBAM-co-NiPAM) copolymers 24), they diffuse into the hydrophobic micelle cores, which limits their growth and ensures good monodispersity. In all cases, the microgel particles were cleaned extensively by dialysis, freeze-dried, and their size characterized by transmission electron microscopy (TEM) where analysis of many particles showed a size distribution of less than 5%. It is important that the particles be extensively dialyzed as traces of low MW free polymer impurities in the microgel particle systems have been shown to lead to effective depletion forces and short-ranged attractions.26 Polymer Synthesis. Linear copolymers of similar composition, 50:50 poly(NtBAM-co-NiPAM), were prepared by free radical polymerization in benzene. A 10% w/v solution was used, resulting in a polymer of MW 1.5 × 104, as determined by combined gel permeation chromatography (GPC) and light scattering. Film Preparation from Particle Suspensions. Dispersions of 4 wt % of the microgel particles in ethanol either alone or with various concentrations (between 0.4 and 0.8 wt %) of linear poly(NtBAM-co-NiPAM) 50:50 were spread over glass slides (LabTec II Chamber slides) and allowed to dry overnight (however, the evaporation of ethanol occurred rapidlysin the first few minutes), resulting in “bumpy” surfaces. The amount of solution needed to ensure coverage of the slides was determined by varying the microgel particle concentration between 1 and 9 wt % with 0.4 wt % free polymer added.14 Characterization of the Mircogel Particle Films. Atomic Force Microscopy. AFM measurements were made in tapping mode using a NanoScope III AFM with Veeco NanoProbe anisotropic tips, to determine surface height variation and to
J. Phys. Chem. B, Vol. 110, No. 30, 2006 14583 obtain surface height plots. Films were prepared from the particles alone (no added free polymer) by casting onto freshly cleaved mica surfaces. Experiments were carried out on films prepared from 70, 200, and 400 nm particles. The 700 nm particles were too big to be studied effectively using AFM due to limitations in the length of the probe tip. For the same reason, we could not effectively measure the surfaces prepared using particles and free polymer, as the height variations were beyond the capability of our cantilever to effectively scan. By the use of AFM, it was possible to see the particle ordering over a 5 × 5 µm range. Fluorescence Staining of the Films. After the surfaces had been cast, and left overnight to dry, they were stained with Fitc. A drop of cold fluorescein 5-isothiocyante isomer 1 (Fit-C, Molecular Probes) made up in DMSO (10% w/v) was added to the particle films at 4 °C, and they were left in the cold room for an hour or so. The plates were then warmed to 37 °C prior to washing thoroughly with warm water. The surfaces could now be visualized using fluorescence and confocal microscopy. Fluorescence Microscopy. The distribution of the microgel particles on the surface and the surface coverage by the particles was determined by fluorescence microscopy using a Zeiss Axioplan Imaging microscope with a ×10 objective, and the images were obtained using a Zeiss Axiocam camera. Through the use of the ×10 objective, it is possible to see the bulk structure with a view of 200 × 200 µm, as well as the variations in the overall surface topography induced by adding free polymer and/or changing the microgel particle size. Confocal Microscopy. Confocal microscopy was carried out to determine the film thicknesses, the morphology throughout the film, and the pore sizes in the different films. Micrographs of the films were obtained using an LSMJ10 META confocal microscope (Cark Zeiss Ltd). Fit-c labeled surfaces were activated using a 488 nm argon laser. Micrographs were processed using a Zeiss LSM Image Browser and Adobe Photoshop software. Through the use of a ×20 objective, z-stacks were obtained through the film thickness, with the number of images taken being calculated from an initial scan of the films thickness to ensure a good overlap between successive images and thus to enable the best 3D reconstruction of the film. Each z-slice was approximately 1 µm apart. Once the overall structure of the films had been measured, individual pores in the films were measured using a ×63 oil-immersion lens. In this way, the pore dimensions (length, width, and depth) could be determined in detail and correlated to the different film preparations. Preliminary Cell Morphology Assay. HeLa (human epithelial adenocarcinoma cells) were obtained from the European Collection of Animal Cell Cultures (http://www.ecacc.org.uk). Relevant cell culture media and associated supplements were as detailed by the respective distributors. All cell types were maintained at 37 °C under 95% air/5% CO2. Chamber slides that had been precoated with the microgel particle films were seeded with 1 mL of cell suspension (2 × 104 cells/mL) and incubated under standard cell culture conditions for 24 h. Cells were fixed in a prewarmed 3.7% (v/v) solution of formaldehyde in PBS. Following fixation, cells were stained for 20 min using 0.1% (w/v) crystal violet in PBS. For fluorescence-based localization of F-actin and DNA, cells were permeabilized with 0.1% Triton X-100 (Sigma) in PBS for 10 min. Cells were then incubated in 1% (v/v) rhodaminephalloidin (Molecular Probes) in PBS for 45 min in the dark, followed by incubation in 0.03% (v/v) 4′,6-diamidino-2phenlindole (DAPI, Molecular Probes), in PBS for 30 s. The
14584 J. Phys. Chem. B, Vol. 110, No. 30, 2006
Lynch et al.
Figure 1. AFM images of a surfaces prepared from 200 nm 50:50 BAM/NIPAM particles alone. The image on the left shows the variation in the height of the surface, the middle image shows the variation in the amplitude of the surface, and the image on the right shows a surface density plot.
Figure 2. (a) Fluorescence microscopy images of surfaces prepared from 200 nm 50:50 NtBAM/NiPAM particles alone, showing the smoothness of the surfaces and the lack of topographic and morphologic features. (b) 200 nm particles with 0.4 wt % free polymer added. Magnification is ×10, and the images are stained with Fit-c.
slide was then washed with warm PBS and sealed with Vectashield (Vector Labs). Rhodamine-phalloidin labels Factin, while DAPI stains double-stranded DNA. Cells were mounted and viewed at ×20 magnification with a Zeiss Axioplan imaging microscope and images were obtained using a Zeiss AxioCam camera at a resolution of 1300 × 1030 pixels. Results Surfaces from Particles Alone. Surfaces made from the microgel particles alone (i.e., without added free polymer) were expected to be completely smooth, with just the periodic undulations due to the particulate nature of the particles, and this was indeed the case, as shown in Figure 1 via AFM. Films were cast from 4 wt % of the different size 50:50 NtBAM/ NiPAM particles in ethanol onto freshly cleaved mica and left to dry overnight. The height variation (Figure 1a) and the amplitude variation (Figure 1b) were recorded (only 200 nm particles are shown). From the amplitude measurements, it is possible to clearly see the particulate structure of the film, as well as to see how smooth the surfaces are, and how the microgel particles assume a regular close-packed arrangement. From the height measurements, it can also be seen that the surfaces are indeed extremely flat, which is confirmed by the surface plots shown in Figure 1c, where the peaks and dips due are to the particles. The height variation is only about 30 nm in total for the 200 nm and is less than 10 nm for the 70 nm particles (data not shown). The small height variation measured is an effect of the AFM technique itself, where the length of the probe tip is not sufficient to reach into the valleys between the particles and also due to the scanning nature of the measurement. For this reason, we could not effectively measure
the arrested-particle surfaces either (as contact with the surface was lost in the pores and tips broke upon reaching the ridges). The expected height variation determined is also much less than the particle size, due to the fact that the particles are “soft” and thus can spread out over the surface to some extent, which can be seen by estimating the number of particles per row (∼10 for the 200 nm particles shown in Figure 1, which is only about half that which could theoretically fit into a 5 µm window). However, due to their cross-linked nature, the particles do not spread completely, at least on the time scale of our surface characterization and cell adhesion studies. The largest particle size used, the 700 nm beads, were also studied using AFM, but here, no structural detail could be observed, due to the large size of the particles and also the length of the probe tip which limits the resolution. Looking at the surfaces on a more global scale (looking at a region of approximately 200 µm2 instead of the 1-5 µm2 areas imaged with AFM) using fluorescence microscopy with ×10 magnification, it was observed that at this distance the surfaces are also extremely flat. Figure 2a shows an optical microscopy image of a surface prepared from 200 nm particles alone at ×10 magnification. It is clear that the surfaces formed from microgel particles are essentially flat, with very little topography or morphology discernible. Similarly, feature- and morphologyfree images were obtained for the other particle sizes and for films prepared from linear copolymers (data not shown). In some of the early experiments, it turned out that the particles were not dialyzed sufficiently during the cleaning stage, and thus, some residual free polymer that had not been crosslinked into the microgel particles remained from the synthesis procedure. This residual-free polymer was sufficient to cause a
Morphologically Rich Polymeric Surfaces depletion attraction and induce particle ordering, of the type that we wanted to explore.26 However, we were interested in using a well-defined and well-characterized system, so we continued the dialysis of the particles for an extended time to ensure no free polymer remained (dialysis in ×20 excess volume of Milli Q water for 1 month, changing the water daily) for all further experiments. Surfaces from Particles and Added Free Polymer. Determination of the optimal concentrations of microgel particles and free polymer to achieve full surface coverage and morphological features is as follows: Using extremely well-cleaned particles, and controlled amounts of added free polymer, we were interested in investigating the concentration ranges that resulted in interesting surface morphologies and complete coverage of the glass surface. Thus, the optimal concentration of particles, and the particle-polymer ratio resulting in reproducible morphologies were determined using fluorescence microscopy. The addition of small concentrations of the nonadsorbing polymer 50:50 poly(NtBAM-co-NiPAM) to 4 wt % dispersions of the 200 nm 50:50 NtBAM/NiPAM particles prior to casting the films resulted in dramatic differences in the surfaces formed compared to those formed from particles alone (i.e., no free polymer added). Figure 2b shows a ×10 fluorescence microscopy image of a surface prepared with 4 wt % 200 nm particles and 0.4 wt % added free polymer. It is clear from this image that adding a small amount of polymer increases the amount of surface roughness observed compared with the particles alone (Figure 2a) and results in the formation of a porous or spongy structure with high regularity. The depletion reaction was found to occur at 0.4-0.5% w/w added free polymer. It did occur at lower free polymer concentrations, but not consistently, and not with full coverage of the surface. At 0.6% free polymer, the pores are a bit bigger in diameter than was the case with the lower percentage of free polymer, however, the coverage of the slide is lower, which was observed as large dark areas in the micrograph images, which correspond to the glass slide (data not shown). The optimal particle concentration required to give full coverage of the surface was found to be 4 wt % of the microgel particles for the 200-700 nm particles. Lower particle concentrations resulted in bare glass remaining at the center of the film (as a result of the interaction between the ethanolic solution and the plastic of the chamber slide walls). With the 70 nm particles, it was actually very difficult to see the development of the structure, due to the extremely high fluorescence resulting from the very large number of particles in places where particles were dense (i.e. forming a carpet), and where structure could be seen, it was not that well developed (see Figure 3a). The optimal polymer concentration to induce the depletion effect was found to be 0.4-0.5 wt % for the all of the particles sizes. Thus, 4 wt % particles and 0.4 wt % free polymer were used for all further studies. Effect of Microgel Particle Size. The size of the microgel particles is expected to play a significant role in the scale of the topography, as the physical size of the particle-dense regions must depend on the particle size. Thus, an assembly of several particles of diameter 70 nm must be significantly smaller than an assembly of several particles of diameter 200 nm (and even more so for the bigger particles of 400 and 700 nm). Comparison of the effect of particle size on the surface morphology is shown in Figures 3-5. Figure 3 shows the middle image from the z-stack (the scan through the depth of the film) for each of the four particle sizes
J. Phys. Chem. B, Vol. 110, No. 30, 2006 14585
Figure 3. Confocal microscopy images of a single z-slice for each of the four microgel particle sizes studied, where the films were prepared from 4 wt % of the microgel particle solutions with 0.4 wt % added free polymer: (a) 70 nm microgel particles, (b) 200 nm microgel particles, (c) 400 nm microgel particles, and (d) 700 nm microgel particles. Magnification is ×20, and excitation is at 488 nm. It is clear that full surface coverage was not achieved for the 70 nm particles, although clearly morphology was beginning to emerge.
studied (70, 200, 400, and 700 nm) prepared from 4 wt % particle solutions with 0.4 wt % of free high MW polymer added to each. Significant differences can be seen between the surfaces with increasing particle size. First, the size of the topographic features appears to decrease with increasing particle sizesthat is, the pores become smaller. Second, the thickness of the films decreased with increasing particle size, despite using the same particle concentration and spreading the same amount of solution over the slides. Thus, the thickness of the film from the 200 nm beads was 25 µm, whereas that from the 700 nm beads was just 13 µm (the film thicknesses were determined using the confocal microscope, where the first and last images are determined from where fluorescence is no longer observable). This is of course related to the number of particles, as there are many times more of the 200 nm particles than of 700 nm particles at the same weight fraction. Upon scanning through the films, it could also be observed that for the films composed of the smaller particles (70 and 200 nm), there were several layers of shallow pores one on top of the other, whereas for the larger sized particles (400 and 700 nm), the surface was composed of just a single layer of very deep pores. The pore length and breadth were also larger for the smaller particle sizes (see below). Thus, by the use of the same overall concentration of particles and added free polymer, the larger particles form films composed of a single layer of deep pores, whereas the smaller particles from films composed of several layers of shallow pores. At present, we have no explanation for why this is, and further studies will be conducted to try to understand and explain this. To get a more detailed characterization of the effect of microgel particle size on the pore structure of the films, the same series of films was studied using a ×63 oil-immersion lens, which enabled us to scan through a single pore in each film, to
14586 J. Phys. Chem. B, Vol. 110, No. 30, 2006
Lynch et al.
Figure 4. Details of the pore structures of the four films from Figure 3 above, taken using the confocal microscope at ×63 magnification (oilimmersion lens). The main images show the actual pores, while the orthogonal projections are given above and to the right of the images along the green and red lines, respectively. These projections are the fluorescence intensity along the lines, and represent the microgel particle intensity, as it is these that are fluorescent: (a) 70 nm microgel particles, (b) 200 nm microgel particles, (c) 400 nm microgel particles, and (d) 700 nm microgel particles.
TABLE 1: Effect of Microgel Particle Size on the Pore Dimensions in Films Prepared from 4 wt % Microgel Particle Solutions with 0.4 wt % Free Polymera particle size
pore depth
pore length
pore width
200 nm 400 nm 700 nm
7 ( 5 µm 14 ( 3 µm 13 ( 5 µm
65 ( 5 µm 50 ( 3 µm 36 ( 5 µm
40 ( 5 µm 22 ( 3 µm 13 ( 5 µm
a Values were determined as averages of analysis from confocal micrographs and fluorescent micrographs.
determine the depth, length, and width of individual pores in each of the films. Figure 4 shows the z-stack (a 3D reconstruction of the z-slice images) through an individual pore for each of the microgel particle sizes, 70, 200, 400, and 700 nm. In these images, the pore bottom and the pore walls are clearly visible. It is clear from these images that the pore size changes significantly with changing microgel particle size. For each of the microgel particle sizes, orthogonal projections of the fluorescence intensity are also shown. The orthogonal projections (above and to the right of the images) show the particle distribution along the green and red lines, respectively, indicating the particle-rich and particle-poor areas. For the 70 and 700 nm particle films, the images are quite grainy, and in the 70 nm case, it is clear that the pore structure is not entirely complete. The pore dimensions (depth, length, and width) for the films composed of 200 nm or bigger particles are listed in Table 1, and in all cases, clear trends were observed. The pore length and width decreased with increasing microgel particle size, whereas the pore depth increased with increasing particle size (see Table 1). The numbers given in Table 1 are averages from several pores. As
the pores in the film from the 70 nm particles were incomplete, data for these were not obtained. The different pore structures formed with the different microgel particle sizes are shown as surface plots (2D3D plots) in Figure 5 for each of the four particle sizes studied. From these surface plots, the pore walls and pore bottoms can be seen clearly. Cell Morphology Studies. Figure 6 shows preliminary images of HeLa cells grown on 200 nm microgel particle surfaces (with free polymer) and on control tissue culture polystyrene (TCP) surfaces. On the TCP surfaces, the cells have their characteristic stellate morphology, with good cell spreading and formation of focal contact points (Figure 6a). However, on the “bumpy” surfaces, it appears that the cells have two optionssthey can either grow inside the pores, where their spreading is restricted to the size of the pore as shown in Figure 6b, or they can grow along the particle-rich ridges between the pores, in which case the cells adopt an elongated structure, as shown in Figure 6c. It has not yet been clarified under which conditions the cells behave in each of the two distinct ways or the effect of the different cellular morphologies on cell proliferation or gene expression. Discussion The formation of controlled porous structures is of considerable general interest, and it would be useful to understand its origins. We are not able to completely resolve this question (this has been difficult in most drying morphology experiments). However, we interpret the observations as follows. Mesoscopic structuring in polymer films is not new. It has been observed upon phase separation of block copolymers, and
Morphologically Rich Polymeric Surfaces
J. Phys. Chem. B, Vol. 110, No. 30, 2006 14587
Figure 5. Surface plots (2D3D) of the z-stacks shown in Figure 4 for each of the four microgel particle films, showing the pore structures clearly: (a) 70 nm microgel particles, (b) 200 nm microgel particles, (c) 400 nm microgel particles, and (d) 700 nm microgel particles.
Figure 6. Fluorescence images of HeLa cells grown on (a) tissue culture polystyrene and (b) and (c) films made from 200 nm 50:50 NtBAM/ NiPAM microgel particles. In (b), the cells are confined to the pores, and in (c), the cells are growing along the ridges between the pores. Magnification is ×10. Staining with Rhodamine-phalloidin to label the F-actin network (red) and DAPI to stain double-stranded DNA (blue).
in the context of “breath figures” formed from a volatile solvent that is immiscible with water,22 and some effect on ordering has also been previously associated with surfactant added to the microparticles.27,28 Since we have worked in a solvent that is miscible with water, and no comparable effects are observed in parallel studies of free polymer rather than particles (as would be expected for breath figures), it seems likely that the phenomena we observe are somewhat different in origin. Similarly, previous observations of surfactant-induced morphology have been interpreted mainly as the surfactant being “extruded” from the drying film.27 Again, we consider it unlikely that extrusion of free added polymer is the origin of this effect (though some aspect of this cannot be excluded entirely). Rather we think it more likely that the morphologies are determined by new effective and short-ranged attractions caused by the particles, under rapid drying conditions. (It is worth noting that, at least to the qualitative characterization determined here, the morphology is relatively insensitive to drying rate, providing drying occurs normally or is accelerated, whereas slowed drying by increase of vapor pressure of ethanol leads to loss of structure). We consider therefore that the phenomenon is associated with organization of the particles, via new effective
forces induced by added free polymers, somewhat modified by and frozen in by the process of drying. A possible more detailed mechanism is as follows. Under a wide variety of conditions, colloidal particles gel or “solidify” without the formation of crystalline order, a process which has been termed “dynamical arrest”.17 Such glasslike arrested states arise due to attractive forces that have a short range compared with the size of the repulsive core, which is typical of colloidal particles whose size is rather large and whose interactions are of microscopic scale. Thus, while repulsive hard-sphere particles stop moving at 58% volume fractions (to form a “repulsive” glass) simply because they run out of space, attractive particles can also lose ergodicity because they “stick”.17 The primary control parameter determining glassification in attractive systems is therefore the range of the attraction. One way to introduce attractive interactions between particles is by adding a nonadsorbing polymer to the system, which leads to the particles experiencing an effective depletion attraction, whereby the polymer is excluded from the region between the surfaces of two nearby particles, leading to an increased osmotic pressure which pushes the particles close together.29 The range and depth of the depletion attraction are known to be controlled
14588 J. Phys. Chem. B, Vol. 110, No. 30, 2006
Lynch et al.
Figure 7. Schematic representation of the process of phase separation and arrest on surfaces of microgel particles, using an attractive interaction between the particles, such as that induced by a depletion attraction due to added free polymer: (a) the attraction between the particles themselves and the particles and the surface, (b) the phase separation and arresting this on the surface, and (c) the final surface with particle-rich and particlepoor domains.
by the polymer’s size and concentration, respectively. It is wellknown from extensive scientific studies for another purpose18 that such systems have a low-concentration, high-concentration phase equilibrium where particle-poor and particle-rich phases coexist. Under certain circumstances, this phase coexistence is “arrested” (frozen and immobilized) before the phases fully separate, leading to a large range of gels and amorphous porous structures, some of which may have applications as useful biological materials. The concept of depletion-induced structural morphology may now be working in the formation of different surfaces. By varying the initial components (particle size and concentration of both particle and polymer), it is possible to change the attraction between the microgel particles themselves and the microgel particles and the surface (Figure 7a), and thus, it is possible to tune across the phase separation (shown in Figure 7b). Controlled drying would then cause the system to enter (albeit in a somewhat poorly controlled manner) the unstable phase-separating regime. The kinetics of phase separation in unstable systems involves spinodal waves (unstable compositional waves of more and less dense particle dispersions), and one could envisage entering this regime as shown schematically in Figure 7c. However, there are elements of this picture that would have to be studied in more detail to fully clarify the mechanism. The overall outcome is that the resulting surface topography can be tuned all the way from flat (except from the intrinsic bumps of the particles) via bumpy, to a porous material at the surface. We show here that using different sizes of 50:50 NtBAM/NiPAM particles and nonadsorbing free polymer we get a range of surfaces with different pore size and depth (topography), as well differences in the number of layers of pores that form (morphology) which may be related to the large difference in the number of particles present with the different sizes of microgel particles at the same concentration. The relative sizes of the ridges and the pores can be altered in a number of ways, via the depletion interaction directly (size and concentration of added polymer) or via the particle size as shown in this work where the pore characteristics were hugely dependent on the microgel particle size as shown in Figures 4 and 5 and Table 1. It is reasonable to expect that even richer surface topography and morphology can be achieved using binary mixtures of two different particle sizes, and work to investigate this possibility is underway. At an even higher level of complexity, assembly of the microgel particles can be driven by more sophisticated self-assembly techniques utilizing complementarity of the particles themselves (such as charge comple-
mentarity, host-guest type interactions, or molecular recognition lock-and-key mechanisms), and an investigation of such systems will be our next step. As described here, the process of phase separation and arrest is an extremely simple and versatile approach to preparing surfaces with rich morphology. Additional advantages of this method of preparing surfaces of controlled topography include the fact that it is applicable to all polymer types, including particles made from polymers that have already received approval from the FDA for use in medical devices, as well as the fact that it does not require advanced surface chemical functionalization or machining or patterning of the surface but instead utilizes intrinsic forces within particle-polymer mixtures to self-assemble the surfaces which are then arrested in the phase-separated state. Our preliminary results on cell spreading and cell morphology have shown that by varying the surface topography we could change the cell morphology, although we have currently tested just one cell type and only in a very preliminary manner. Now that we have an understanding of how to control the surface morphology via phase separation and arrest of the microgel particles, we will extend this work to determine under what circumstances the two different cell morphologies become favorable. We currently have more detailed studies underway of the effect of surface topography on cellular phenotype, cell adhesion and spreading, and global gene expression of the HeLa cells, using oligonucleotide array-based gene expression profiling. We will also extend this to other cell types, such as primary human adult skin fibroblasts, human foetal lung SV-40transformed fibroblasts, and smooth aortic muscle cells, among others, to determine if particular cell types favor particular surface morphologies. Conclusions The formation of tissue culture surfaces of controlled surface roughness (topology) and morphology by phase separation and arrest of microgel particles at the surface is reported. The surface chemistry is controlled by the microgel composition, and parameters such as hydrophobicity or charge density can be varied. The phase separation is induced by the addition of low concentrations of a nonadsorbing free polymer which induces a depletion attraction and drives the particles together, forming regions that are particle rich (ridges) and regions that are particle poor (pores). The phase-separated system is then arrested at the surface resulting in the formation of a range of novel porous structures. The size of the pores, the thickness of the ridges,
Morphologically Rich Polymeric Surfaces and the number of layers of morphology are related to the microgel particle size and can be controlled to a high degree. This represents a versatile and adaptable route to complex morphologically rich surfaces for use in a range of biological applications, such as implant surfaces, tissue engineering, and others. Acknowledgment. This work was funded by grants from the Health Research Board, Ireland and Enterprise Ireland. The authors thank the reviewers for their insightful comments on this manuscript. References and Notes (1) Ito, Y. Biomaterials 1999, 20, 2333. (2) Pettit, D. K.; Horbett, T.; Hoffman, A. S. J. Biomed. Mater. Res. 1992, 26, 1259. (3) Garcia, A. J.; Ducheyne, P.; Boettiger, D. J. Biomed. Mater. Res. 1998, 40, 48. (4) Singhvi, R.; Stephanopolos, G.; Wang, D. I. C. Biotechnol. Bioeng. 1994, 43, 764. (5) Curtis, A. S. G.; Clark, P. Crit. ReV. Biocompat. 1990, 5, 343. (6) Ingber, D. E.; Madri, J. A.; Folkman, J. Endocrinology 1986, 119, 1768. (7) Ingber, D. E.; Folkman, J. M. J. Cell Biol. 1989, 109, 317. (8) Re, F.; Zanetti, A.; Sironi, M.; Polentarutti, N.; Lanfrancone, L.; Dejana, E.; Colotta, F. J. Cell Biol. 1994, 127, 537. (9) Dike, L. E.; Ingber, D. E. J. Cell Sci. 1996, 109, 2855. (10) Chen, C. S.; Mrksich, M.; Huang, S.; Whitesides, G. M.; Ingber, D. E. Science 1997, 276, 1425. (11) Lee, J. H.; Lee, S. J.; Khang, G.; Lee H. B. J. Biomater. Sci., Polym. Ed. 1999, 10, 283. (12) Lynch, I. D.; K. A.; Linse, S. Sci. STKE 2006.
J. Phys. Chem. B, Vol. 110, No. 30, 2006 14589 (13) Allen, L. T.; Tosetto, M.; Miller, I.; O’Connor, D.; Penney, S. C.; Lynch, I.; Keenan, A. K.; Pennington, S. R.; Dawson, K. A.; Gallagher, W. M. Biomaterials 2006, 27, 3096. (14) Allen, L. T.; Fox, E. J. P.; Blute, I.; Kelly, Z. D.; Rochev, Y.; Keenan, A. K.; Dawson, K. A.; Gallagher, W. M. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 6331. (15) Hubbell, J. A. Curr. Opin. Biotechnol. 2003, 14, 551. (16) Beattie, D.; Wong, K. H.; Williams, C.; Poole-Warren, L. A.; Davis, T. P.; Barner-Kowollik, C.; Stenzel, M. H. Biomacromolecules 2006, 7, 1072. (17) Dawson, K. A. Curr. Opin. Colloid Interface Sci. 2002, 7, 218. (18) Dawson, K. A.; Lawlor, A.; DeGregorio, P.; McCullagh, G. D.; Zaccarelli, E.; Foffi, G.; Tartaglia, P. Faraday Discuss. 2003, 123, 13. (19) Foffi, G.; Zaccarelli, E.; Sciortino, F.; Tartaglia, P.; Dawson, K. A. J. Stat. Phys. 2000, 100, 363. (20) Segre, P. N.; Prasad, V.; Schofield, A. B.; Weitz, D. A. Phys. ReV. Lett. 2001, 86, 6042. (21) Lynch, I.; Blute, I. A; Zhmud, B.; MacArtain, P.; Tosetto, M.; Allen, L. T.; Byrne, H. J.; Farrell, G. F.; Keenan, A. K.; Gallagher, W. M.; Dawson, K. A. Chem. Mater. 2005, 17, 3889. (22) Karthaus, O.; Maruyama, N.; Cieren, C.; Shimomura, M.; Hasegawa, H.; Hashimoto, T. Langmuir 2000, 16, 6071. (23) Lord Rayleigh. Nature (London) 1911, 86, 416. (24) Lynch, I.; Dawson, K. A. J. Phys. Chem. B 2004, 108, 10893. (25) McPhee, W.; Tam, K. C.; Pelton, R. J. Colloid Interface Sci. 1993, 156, 24. (26) Bartsch, E.; Antonietti, M.; Schupp, W.; Sillescu, H. J. Chem. Phys. 1992, 97, 3950. (27) Chevalier, Y.; Pichot, C.; Graillat, C.; Joanicot, M.; Wong, K.; Maquet, J.; Limdner, P.; Cabane, B. Colloid Polym. Sci. 1992, 270, 806. (28) Ramos, l.; Lubensky, T. C.; Dan, N.; Nelson, P.; Weitz, D. A. Science 1999, 286, 2325. (29) Pham, K. N.; Puertas, A. M.; Bergenholtz, J.; Egelhaaf, S. U.; Moussaı¨d, A.; Pusey, P. N.; Schofielf, A. B.; Cates, M. E.; Fuchs, M.; Poon, W. C. K. Science 2002, 296, 104.