A novel approach to the micropatterning of proteins using dewetting of polymer bilayers

Abstract

The ability to control protein and cell positioning on a microscopic scale is crucial in many biomedical and bioengineering applications, such as tissue engineering and the development of biosensors. We propose here a novel, simple, and versatile method for the micropatterning of proteins. Micropatterned substrates are produced by the dewetting of a metastable polymer film on top of another polymer film. Selective adsorption, or micropatterning, of proteins can be achieved on such substrates by choosing pairs of polymers which differ in protein affinity. In this study, patterns were produced in bilayers of poly(methylmethacrylate) (PMMA) and polystyrene (PS), and of PMMA and octadecyltrichlorosilane (OTS). Fluorescence microscopy and atomic force microscopy (AFM) provide evidence that model proteins adsorb preferentially on isolated bio-adhesive (PS and OTS) micropatches in a protein-resistant (PMMA) matrix. “Inverse” protein patterns, containing non-adhesive (PMMA) islands in a protein-adhesive (PS) matrix can also be produced. Such micropatterned substrates could potentially be used in the development of biosensors and bioassays, and in the study of cell growth and motility.

---

1 Introduction

Many areas of the biological sciences depend critically on the ability to arrange and study cells on specific sites of a surface. Topographical and chemical patterning of a surface has a demonstrated effect on cell development and functionality: studies addressing cell attachment, differentiation, proliferation and migration, benefit greatly from the development of substrates with microscopic features of controlled surface chemistry and engineered geometry.¹ There is consequently a strong interest in producing substrates that possess spatially defined bio-adhesive patterns, which promote protein and subsequent cell attachment, in a background that resists protein adsorption. Besides its application in fundamental cell-surface investigations, the ability to immobilise proteins and cells on a substrate is fundamental in the engineering of biomedical implants, such as artificial tissues,^2,3 and bio-technological devices, such as cell-based sensors for drug screening.^4,5

Importantly, the interaction of cells with their environment is mediated by extracellular proteins contained in cell culture media.⁶ For example, cells such as neurons only attach to surfaces that are pre-coated by proteins such as laminin.¹ In order to achieve the patterning of cells on a substrate, the non-specific adsorption of proteins on the surface (bio-fouling) needs to be limited and the attachment of selected proteins to designated regions of the surface controlled. Ultimately, the aim is to precisely arrange proteins on a substrate that is pre-patterned on the microscale.

To date, several elegant techniques have been employed to pattern proteins on a surface,^7,8 mainly soft lithography approaches,^9,10 such as microfluidic patterning^11,12 and micro-contact printing.^13,14 However, these methods are expensive and cumbersome, and require specialised expertise and materials, such as custom-synthesised reagents, which limits their widespread use in biomedical fields. It is desirable to develop a protein patterning method that is at the same time simple, inexpensive, and versatile.

In this paper we propose, for the first time, to pattern biological molecules on substrates produced by “dewetting”, the spontaneous decay process whereby unstable liquid films transform into their equilibrium state, i.e. a series of isolated droplets. In recent years, a thorough understanding of dewetting of liquid films from flat solid surfaces has evolved both for simple and complex fluids.^15–18 We are interested here in the more complex situation of a liquid film dewetting from a liquid substrate, a problem which has also been investigated before.^19–23

Unstable and metastable liquid polymer films of thickness around 100 nm supported on solid substrates dewet mainly by heterogeneous nucleation: debris particles or impurities in the film initiate the formation of randomly distributed holes (patches free of polymer, see Fig. 1).²⁴ The holes grow with time until they impinge on adjacent holes, resulting in the formation of ribbons of liquid along their contact line. Finally, the ribbons become unstable too, and eventually transform into a series of isolated droplets on the substrate. Very thin polymer films may also decay by spinodal dewetting, which induces correlated holes throughout the whole film.^25,26

Fig. 1 (a) AFM image of a hole formed by dewetting of a polystyrene (PS) film on top of a poly(methylmethacrylate) (PMMA) film (“inverse” bilayer). The base of the hole is made of PMMA and the rest of the film and rim are made of PS. Portions of the sample that are thicker are represented in lighter shades in the image. The scale bar is 5 μm and the vertical scale is 210 nm. (b) Cross section of a hole in a PMMA film on top of a PS film (“direct” bilayer) obtained from an AFM scan.

Here, we propose a novel method for the patterning of proteins for biological and bio-medical studies. Patterned substrates are produced by the dewetting of bilayers of two immiscible polymers. In this study, we employ polystyrene (PS), which readily attracts proteins, and poly(methylmethacrylate) (PMMA), which has a lower affinity for proteins. The method consists of two simple steps (see Fig. 2):

Fig. 2 Schematic illustration of the proposed method for the micropatterning of proteins: preparation of a “direct” protein pattern. A PMMA film dewets on top of a PS film, exposing small round patches of the underlying PS film. When the sample is incubated in a protein solution, the proteins (small black stars) adsorb selectively inside the PS holes. Similarly, an “inverse” protein pattern can be prepared by inverting the order of the two polymer films.

(1) A polymer bilayer of PS and PMMA is deposited on a silicon substrate. The bilayer is melted by heating it at a temperature T above the glass transition temperature (T_g) of the polymers, and, since the upper PMMA film is unstable, it dewets (see Fig. 2). Holes appear in the upper film that expose small round patches of the lower PS film. As a result, the bilayer is micropatterned by simply thermally annealing the system and letting it dewet, a spontaneous process for unstable films. The size and the distribution of the formed islands can be tuned by varying a few physical parameters, such as annealing time, film thickness, and molecular weight of the employed polymers.

(2) A buffered solution of proteins is incubated on the micropatterned substrate and proteins selectively adsorb onto bio-adhesive regions of the substrate. In Fig. 2 the proteins, small black stars, adsorb selectively inside the PS holes, and not on the PMMA matrix.

Alternatively, an “inverse” bilayer can be prepared by inverting the order of the two polymer layers: upon exposure of a dewetted inverse bilayer to a protein solution, the PS film matrix is covered with proteins, but the PMMA holes remain almost completely free of proteins.

We demonstrate by atomic force microscopy (AFM) and fluorescence microscopy that both the direct and inverse PS/PMMA bilayer systems are effective in achieving patterning of immunoglobulins, because the proteins adsorb preferentially on the PS regions of the substrate.

2 Experimental

Thin polymer films were prepared using atactic polystyrene (PS) of molecular weight M_w = 96 kg mol⁻¹ and poly(methylmethacrylate) (PMMA) of molecular weight M_w = 23 kg mol⁻¹ (both of low polydispersity M_w/M_n = 1.04; PSS, Mainz, Germany). The PS and PMMA films were prepared by spincoating dilute toluene solutions (99.5+%, Sigma-Aldrich) of the polymers onto polished oxidised silicon wafers (MEMC Electronic Materials Inc., St Peters, USA). The silicon wafers (Si) are covered with a layer of amorphous silicon oxide (SiO_x) of thickness 1.6 ± 0.1 nm. Prior to spincoating, the Si wafers were thoroughly cleaned in order to remove organic and particulate contaminants. Debris particles were removed using a CO₂-snow jet. Subsequently, the wafers were sonicated for a few minutes in ethanol (redistilled) and acetone (99.5+%, Sigma-Aldrich), and immersed in fresh piranha solution (1 : 1 mixture of concentrated sulfuric acid and 30% H₂O₂) for 30 min and then rinsed by immersion up to 10 times in hot Milli-Q water. Cleaning and spincoating were performed in a class-100 laminar flow cabinet.

Two types of polymer bilayers were prepared: in the “direct” bilayer (see Fig. 2), a PS film was first spincoated on top of the Si substrate; a PMMA film was then spincoated onto a freshly-cleaved mica surface, transferred from the mica onto a clean Milli-Q water surface (floating), and then picked up from the water with the PS-coated Si. The same process was used to prepare the “inverse” bilayers, simply inverting the order of PS and PMMA. The floating step is necessary because the second polymer film cannot be spincoated directly onto the first polymer film, since toluene is a solvent for both polymers. Throughout this paper we will refer to the film that is in contact with the Si substrate as the lower film, and to the film that is exposed to air as the upper film.

Some dewetting experiments were also performed using a hydrophobised Si wafer as the substrate: a monolayer of octadecyltrichlorosilane (OTS, 90+%, Sigma-Aldrich) was allowed to self-assemble on the Si substrate using an established method.²⁷ This method produces a hydrophobic, smooth, and stable OTS monolayer of thickness 2.6 ± 0.1 nm on the Si substrate.

The thickness of each of the layers was determined by ellipsometry (Beaglehole Instruments, New Zealand), using measurements at different angles of incidence. The prepared polymer films were of thickness between about 70 and 150 nm, and were smooth and uniform, with an RMS roughness below 0.4 nm (over an area of 5000 × 5000 nm²) as determined by AFM. The RMS roughness of a clean Si wafer over the same area was lower than 0.3 nm.

The wetting properties of the employed materials were investigated by measuring the advancing and receding contact angles of MILLI-Q water with a KSV CAM200 Contact Angle System (KSV Instruments Ltd., Helsinki, Finland).

The spreading parameters of PS on PMMA and of PMMA on PS are negative at high temperatures, which implies that a film of a polymer placed on top of a film of the other is either metastable or unstable.^20,23 The bilayer systems were thermally annealed on a heating stage at a temperature T greater than the glass transition temperature T_g of both polymers (T_g ≈ 100 °C for PS and 110 °C for PMMA) for a few minutes. Since the upper film of the bilayer is unstable, it dewets and forms holes that grow with time. The bilayers were then cooled to room temperature and the dewetting patterns froze in. Finally, the patterned substrates were incubated in a protein solution to induce the differential adsorption of proteins. All our experiments indicate that the base of the holes in the upper film of a bilayer is coincident with the top of the surface of the lower polymer film. This result is in agreement with what was observed for inverse bilayer systems of PS/PMMA by Harris et al.²⁰

PMMA forms a stable wetting film on top of the SiO/SiO_x substrate.²⁸ In contrast, PS films of the thickness employed here (ca. 100 nm) are metastable on the SiO/SiO_x substrate.¹⁸ Usually, in direct bilayers, we observe the dewetting of the upper PMMA film only, because the dewetting rate of PS on the SiO/SiO_x substrate is very slow; in cases where the PS lower film is thinner than the PMMA upper film we sometimes observe the appearance of small holes in the lower film within the holes of the upper film. This double dewetting, which was observed before in PS/PMMA bilayers,²³ can be avoided by using thicker lower films.

Whole molecule immunoglobulin G from bovine serum (IgG, >95%) and goat anti-rabbit fluorescein-conjugated immunoglobulin G (FITC-IgG) were purchased from Sigma-Aldrich and from Chemicon International, respectively. The lyophilized powder of IgG and the concentrated solution of FITC-IgG were diluted in a 10 mM phosphate buffer solution (PBS, 140 mM NaCl and 3 mM KCl, pH 7.4, Sigma-Aldrich) to varying concentrations, ranging from 1.5 to 25 μg ml⁻¹.

A droplet of protein solution (ca. 100 μl) was placed onto a substrate for about 1 min and then the substrate surface was rinsed with 2 ml of PBS solution and 2 ml of Milli-Q water. The surface was then dried gently under a stream of nitrogen gas. In the case of the fluorescence measurements, to avoid complete dehydration of the proteins and possible quenching of the fluorescence signal, samples were not dried with nitrogen but were covered with a drop of glycerol and a cover slip.

Atomic force microscopy (AFM) (Multimode, Digital Instruments/Veeco, Santa Barbara, USA, and MFP-3D, Asylum Research) was used in tapping mode to obtain high resolution images of the patterns and of the proteins. A Zeiss Axioplan optical microscope equipped with an Axiocam MR camera and a mercury lamp was used for the fluorescence measurements.

3 Results

The wetting properties of the polymer films and the OTS layer were investigated by measuring the advancing (θ_a) and receding contact angle (θ_r) of water on their surfaces. The OTS layer is the most hydrophobic surface, with θ_a = 112° and θ_r = 107°. The PS film has θ_a = 99° and r = 80°, while the PMMA film has θ_a = 75° and r = 60°.

Fig. 1 shows an AFM image of a typical hole formed upon annealing a polymer bilayer above T_g of the polymers. Upon dewetting, a hole is formed in the upper polymer film and the material removed from the hole accumulates in a rim, which is depicted in lighter shades in the AFM image. The rim of the hole is clearly visible in Fig. 1b, which shows a cross section of an AFM image. In the center of the hole the defect that initiated the dewetting of the film is visible. Close inspection of the rim profile reveals that the rim decreases to the thickness of the unperturbed film via a trough. The presence of this undulatory rim profile, which in very thin films may lead to the formation of so-called satellite holes,²⁹ was investigated in previous studies, but was believed to be present only in films of polystyrene of very low molecular weight, below the entanglement length.³⁰ Our experiments show that when the substrate is liquid the characteristic trough outside the rim is maintained even for higher molecular weight polymers.

The series of optical micrographs in Fig. 3 illustrates the temporal evolution of dewetting by heterogeneous nucleation of a direct bilayer of PS/PMMA films at T = 180 °C. As soon as T_g is exceeded, holes nucleate in the upper PMMA film (dark grey circular patches). Fig. 3a to d show approximately the same region of the sample in subsequent stages of the dewetting process: the hole diameters increase with time, from a few tens of nanometers (not shown), to ca. 20 μm (Fig. 3a), to ca. 100 μm when coalescence with neighbouring holes occurs (Fig. 3c). Initially, the patterns created upon dewetting of a direct bilayer by heterogeneous nucleation consist of a random distribution of circular holes. The average distance between the holes is expected to vary approximately as the square of the film thickness.³¹ The hole size is nearly monodisperse (within ±5%) and the growth rate of all holes approximately equal. The smallest hole diameter that can be observed while imaging in situ by AFM is around 100 nm. As the holes coalesce with neighbouring holes, their shape is deformed at the line of contact. Importantly, as the dewetting progresses the proportion of exposed areas of PS increases compared to the PMMA areas.

Fig. 3 (a)–(d) Dewetting of a PMMA film (ca. 80 nm thick) on top of a PS film (ca. 85 nm thick) (“direct” polymer bilayer). Time series of optical micrographs recorded in situ at T = 180 °C. The time at which each image was captured is shown. The scale bars are 100 μm.

Proteins were allowed to adsorb on the micropatterned substrates and the treated films were observed by fluorescence microscopy and AFM to reveal the surface coverage of adsorbed proteins. To visualise protein coverage on the substrates with fluorescence microscopy, a fluorescently-labelled immunoglobulin G, FITC-IgG, was deposited on the bilayers, as illustrated in Fig. 4. Most of the fluorescently labelled proteins adsorb on the PS domains, as shown clearly in the case of the direct bilayer system by the higher fluorescence signal measured inside the PS holes than in the rest of the PMMA film (part (b)). In the case of the inverse bilayer system (part (c) and inset), as expected, an inverse fluorescence pattern is observed, with higher fluorescence signals coming from the PS matrix than from inside the PMMA holes. This finding demonstrates that the different surface properties of the two polymers induce a differential adsorption of FITC-IgG, therefore achieving the desired effect of patterning the proteins on a microscale.

Fig. 4 Patterning of FITC-IgG proteins. (a) Optical micrograph in reflected light of a dewetted “direct” bilayer system (PMMA on top of PS) on which FITC-IgG proteins were adsorbed from a 15 μg ml⁻¹ solution. (b) Fluorescence micrograph of the same area as in (a). The exposure time was 400 ms. (c) Fluorescence image of a dewetted “inverse” bilayer system (PS on top of PMMA) on which FITC-IgG proteins were adsorbed from a 25 μg ml⁻¹ solution. The exposure time was 200 ms. (d) Fluorescence image of a bare dewetted inverse bilayer (PS on top of PMMA). The exposure time was 2000 ms. The insets in parts (c) and (d) show cross sections of the fluorescence images taken at the locations indicated by the black lines. The scale bars are 20 μm. The fluorescence contrast observed in parts (b) and (c) is due to the higher amount of fluorescent FITC-IgG adsorbed on the PS regions compared to the PMMA regions.

Both PS and PMMA emit a very weak fluorescence signal, which is much smaller than the signal emitted by the FITC-labelled proteins. A very faint fluorescence signal from a bare bilayer system could be collected by employing exposure times 10 times longer than those used for the FITC-labelled samples; even with this long exposure time, there is practically no fluorescence contrast between the PS and PMMA regions in Fig. 4d and inset. Thus we are convinced that the fluorescence contrast observed in Fig. 4b and c is due to the different amount of fluorescent FITC-IgG adsorbed on them.

AFM confirmed the fluorescence results and provided direct high-resolution images of the adsorbed proteins. Fig. 5a shows an AFM image of a hole in a direct bilayer system with adsorbed IgG molecules. The small oblate objects visible in the image are single IgG molecules and their size and shape correspond to the expected values for IgG, which has dimensions of 14 × 10 × 4 nm³.³² The adsorption density of IgG on PS is much higher than on PMMA, with most of the proteins adsorbing selectively inside the holes. The slightly curved front in the center of Fig. 5a is the contact line between the PMMA hole rim and the lower PS film. This front is the border between the protein-rich flat PS film (left side of Fig. 5a) and the curved PMMA hole rim, on which only a few adsorbed molecules are observed.

Fig. 5 Patterning of IgG proteins. (a) Tapping mode phase AFM image of a dewetted “direct” bilayer (PMMA on top of PS) in which IgG proteins were adsorbed from a 3 μg ml⁻¹ solution. The small particles visible inside the hole are IgG protein molecules adsorbed on the PS. (b) Tapping mode AFM image of a dewetted “inverse” bilayer (PS on top of PMMA) in which IgG proteins were adsorbed from a 15 μg ml⁻¹ solution. The scale bars are 0.5 μm. Clearly, most of the IgG proteins adsorb on the PS regions.

Fig. 5b shows the results of the adsorption of a protein solution on an inverse bilayer system. Here, too, the AFM image confirms the fluorescence results and clearly indicates that IgG proteins adsorb preferentially on the PS regions. In the inverse bilayer the adsorption of proteins results in an inverse pattern, with most of the proteins attached to the matrix of the film, and most of the PMMA holes free of proteins.

The average density of IgG molecules adsorbed on PS surfaces from solution is between 4 to 8 times higher than on PMMA surfaces, as measured from several AFM images after exposure to protein solutions of concentration in the range 1.5–20 μg ml⁻¹. It should be noted that these results are strongly dependent on the ionic strength of the protein solution: a less pronounced segregation of proteins was observed using a more concentrated buffer (100 mM PBS).

To test the viability of different bilayer systems for protein patterning, some experiments were conducted on single PMMA films onto hydrophobised (OTS-coated) Si wafers (data not shown). PMMA readily dewets from the OTS layer, producing patterns analogous to the direct PMMA/PS system. Exposure of the PMMA/OTS system to a solution of IgG resulted in noticeable selective adsorption of proteins on OTS regions. This suggests that proteins are more strongly attracted to hydrophobic surfaces; however, since OTS is more hydrophobic than PS, we would expect the segregation of proteins to be stronger on the PMMA/OTS system than on the PMMA/PS system. On the contrary, PS attracts slightly more proteins than OTS. We conclude that wettability is not the only factor determining protein adsorption.

The driving force for the selective adsorption of immunoglobulins on PS domains is still highly debated in the literature: hydrophobic surfaces are known to adsorb larger amounts of various proteins than hydrophilic surfaces^33–35 and it is often suggested that the hydrophobic attraction governs the interaction between proteins and solid surfaces. However, many other factors such as polymer rigidity, surface-or pH-induced conformational changes of the adsorbed proteins, and ionic strength of the solution, are believed to be of importance in determining protein-polymer affinity.³⁶ Our results indicate that surface wettability is not the only factor influencing protein adsorption, since OTS surfaces adsorb slightly smaller amounts of proteins than PS films, even though they are more hydrophobic. It is also clear that a high ionic strength of the solution reduces the selectivity of adsorption of IgG on PS domains.

4 Discussion

The method proposed here provides an attractive alternative to existing protein patterning methods. It is simple, and requires only very basic equipment: a spincoater, to prepare thin and smooth polymer films, and a standard heating stage, to melt the polymer films. All the preparation steps could be easily transferable to the large scale production of substrates to be used in the biological and biomedical fields.

The length scale of the dewetted patterns can be fine-tuned from a few tens of nanometers to several tens of micrometers by controlling annealing time and temperature (see Fig. 3). In contrast, other microfabrication techniques require different molds for the preparation of patterns of different sizes. Due to the heterogeneous nucleation mechanism, the great majority of holes are formed at the same time, so that the entire surface of the sample is patterned with holes of the same size and shape, independently of the overall size of the sample. In general, the growth rate of the holes increases with decreasing viscosity (i.e. with increasing temperature) and increases with decreasing film thickness. When neighbouring holes coalesce, the upper film is reduced to a number of droplets or cylinders of the upper polymer distributed in a polygonal pattern on the lower polymer film (see Fig. 3d). These late patterns could also be used for selective adsorption of proteins.

Kumar and Hahm observed the differential adsorption of proteins on PMMA/PS block copolymers,³³ but in phase-separated copolymers the polymer domains are very small (a few tens of nanometers) and close to each other, which does not suit applications in cell biology. With the method reported here, large (micrometric) protein-rich domains separated from each other by several tens of micrometers can be produced (see Fig. 4 and 5), which is an advantage for studies of cell functionality and motility.

This method is versatile because the dewetting of polymer bilayers can be tuned simply by controlling parameters such as film thickness and the substrate properties, allowing for the possibility of creating a great number of patterns. In this initial study, random patterns were produced. Regular, ordered features could also be obtained by exploiting other dewetting mechanisms, and anisotropic dewetting. For example, correlated concentric holes (satellite holes) could possibly be produced in PS/PMMA bilayers.²⁹ Regular patterns could be obtained by spinodal dewetting, which induces correlated holes throughout the whole film. Spinodal dewetting can be “guided”: well-aligned polymer lines with defined width can be obtained, simply by rubbing the substrate on which the bilayer is spincoated¹⁹ or the upper polymer film itself.^37,38 Alternatively, arrays of defects, deposited in an orderly fashion on the bilayers (printed, for example, as in ref. 39) could induce ordered dewetting. Finally, the location at which film breakup occurs could be controlled by imposing a mold with the desired pattern on the liquid polymer film.^40,41 The possibility of exploiting other dewetting mechanisms for micro- and nano-patterning of proteins will be pursued in future studies.

The patterns produced by dewetting of polymer bilayers present not only chemical contrast, but also topographic contrast between the features and the background. This combination is automatically achieved with dewetting, because the receding of a protein-resistant polymer from the substrate exposes patches of the substrate that are protein-attractive; at the same time, a round patch which is lower than the background and separated by a circular rim is formed (see Fig. 1). The rim could act as a topographical barrier to restrict the movement of cells, once they adsorb inside the protein-coated hole. The size of the circular holes obtained from nucleation dewetting is indeed compatible with the dimensions of most cells.^2,11,42

We can envisage that many other biological molecules, such as DNA and peptides, could also be patterned on polymeric substrates, simply by fine tuning the properties of the polymers that are employed in the bilayer. Furthermore, different proteins contained in one solution could be separated and isolated on different sites of the patterned substrates, according to their distinct affinity for each polymer, therefore achieving a novel affinity chromatography.

5 Conclusions

A new method for the micropatterning of proteins, that is easy to implement and extremely versatile, was introduced. With few very easy steps, patterned substrates that possess micro-islands of adsorbed proteins (on PS) separated by a protein-resistant matrix (PMMA) could be reproducibly prepared. The size and the distribution of these islands could be controlled simply by varying a few physical parameters, such as film thickness and annealing time. It is equally easy to prepare “direct” and “inverse” patterns, according to the application for which the patterns are required. Successive stages of the dewetting process achieve islands of increasing size, which in turn result in varying proportions of resistance and affinity to proteins.

In this way we were able to achieve the micropatterning of model proteins such as immunoglobulin G and fluorescently labelled immunoglobulin G. The combination of PMMA films with hydrophobised Si was also found to produce a functional “direct” pattern, which is, however, not as efficient and versatile as the PS/PMMA system. There is no limit to the size of the area that can be patterned, because the entire substrate dewets at the same time, and reproducible features are induced throughout the sample.

This method has the potential to contribute to bio-medical and bio-engineering studies that require spatially localised features of controlled surface chemistry, such as studies in proteomics and cell research, pharmaceutical screening processes (e.g. development of bio-sensors and bio-assays), and co-culture studies.

Acknowledgements

The author acknowledges funding from the Australian Research Council and assistance from Sally Stowe of the Electron Microscopy Unit of the ANU.

References

1. D. Falconnet, G. Csucs, H. M. Grandin and M. Textor, Biomaterials, 2006, 27, 3044–3063.
2. A. Curtis and C. Wilkinson, Biomaterials, 1997, 18, 1573–1583.
3. J. H. Fitton, B. A. Dalton, G. Beumer, G. Johnson, H. J. Griesser and J. G. Steele, J. Biomed. Mater. Res., 1998, 42, 245–257.
4. S. N. Bhatia, M. L. Yarmush and M. Toner, J. Biomed. Mater. Res., 1997, 34, 189–199.
5. J. L. Tan, W. Liu, C. M. Nelson, S. Raghavan and C. S. Chen, Tissue Eng., 2004, 10, 865–872.
6. H. Thissen, G. Johnson, P. G. Hartley, P. Kingshott and H. J. Griesser, Biomaterials, 2006, 27, 35–43.
7. J. D. Andrade and V. Hlady, Adv. Polym. Sci., 1986, 79, 1–63.
8. A. S. Blawas and W. M. Reichert, Biomaterials, 1998, 19, 595–609.
9. R. S. Kane, S. Takayama, E. Ostuni, D. E. Ingber and G. M. Whitesides, Biomaterials, 1999, 20, 2363–2376.
10. K.-B. Lee, S.-J. Park, C. A. Mirkin, J. C. Smith and M. Mrksich, Science, 2002, 295, 1702–1705.
11. D. T. Chiu, N. J. Jeon, S. Huang, R. S. Kane, C. J. Wargo, I. S. Choi, D. E. Ingber and G. M. Whitesides, Proc. Natl. Acad. Sci. U. S. A., 2000, 97, 2408–2413.
12. A. Folch and M. Toner, Biotechnol. Prog., 1998, 14, 388–392.
13. M. Mrksich, C. S. Chen, Y. Xia, L. E. Dike, D. E. Ingber and G. M. Whitesides, Proc. Natl. Acad. Sci. U. S. A., 1996, 93, 10775–10778.
14. R. Singhvi, A. Kumar, G. P. Lopez, G. N. Stephanopoulos, G. N. Wang, D. I. C. Wang, G. M. Whitesides and D. E. Ingber, Science, 1994, 264, 696–698.
15. G. Reiter, Phys. Rev. Lett., 1992, 68, 75–78.
16. K. Jacobs, R. Seemann, G. Schatz and S. Herminghaus, Langmuir, 1998, 14, 4961–4963.
17. P. Müller-Buschbaum, J. Phys.: Condens. Matter, 2003, 15, R1549–R1582.
18. R. Seemann, S. Herminghaus, C. Neto, S. Schlagowski, D. Podzimek, R. Konrad, H. Mantz and K. Jacobs, J. Phys.: Condens. Matter, 2005, 17, S267–S290.
19. A. M. Higgins and R. A. L. Jones, Nature, 2000, 404, 476–478.
20. M. Harris, G. Appel and H. Ade, Macromolecules, 2003, 36, 3307–3314.
21. P. Lambooy, K. C. Phelan, O. Haugg and G. Krausch, Phys. Rev. Lett., 1996, 76, 1110–1113.
22. S. Qu, C. J. Clarke, Y. Liu, M. Rafailovich, J. Sokolov, K. C. Phelan and G. Krausch, Macromolecules, 1997, 30, 3640–3645.
23. C. Wang, G. Krausch and M. Geoghegan, Langmuir, 2001, 17, 6269–6274.
24. C. Redon, F. Brochard-Wyart and F. Rondelez, Phys. Rev. Lett., 1991, 66, 715–719.
25. A. Vrij, Discuss. Faraday Soc., 1966, 42, 23–33.
26. R. Xie, A. Karim, J. F. Douglas, C. C. Han and R. A. Weiss, Phys. Rev. Lett., 1998, 81, 1251–1254.
27. J. Sagiv, J. Am. Chem. Soc., 1980, 102, 92–98.
28. E. K. Lin, W.-L. Wu and S. K. Satija, Macromolecules, 1997, 30, 7224–7231.
29. C. Neto, K. Jacobs, R. Seemann, R. Blossey, J. Becker and G. Grün, J. Phys.: Condens. Matter, 2003, 15, 3355–3366.
30. R. Seemann, S. Herminghaus and K. Jacobs, Phys. Rev. Lett., 2001, 87, 196101.
31. G. Reiter, Langmuir, 1993, 9, 1344–1351.
32. E. Droz, M. Taborelli, P. Descouts, T. N. C. Wells and R. C. Werlen, J. Vac. Sci. Technol., B, 1996, 14, 1422–1426.
33. N. Kumar and J. Hahm, Langmuir, 2005, 21, 6652–6655.
34. H. Elwing, S. Welin, A. Askendal, U. Nilsson and I. Lundstrom, J. Colloid Interface Sci., 1987, 119, 203–210.
35. B. Lassen and M. Malmsten, J. Colloid Interface Sci., 1997, 186, 9–16.
36. S. H. Lee and E. Ruckenstein, J. Colloid Interface Sci., 1988, 125, 365–379.
37. B. Du, F. Xie, Y. Wang, Z. Yang and O. K. C. Tsui, Langmuir, 2002, 18, 8510–8517.
38. X. Zhang, F. Xie and O. K. C. Tsui, Polymer, 2005, 46, 8416–8421.
39. T. Kawase, H. Sirringhaus, R. H. Friend and T. Shimoda, Adv. Mater., 2001, 13, 1601–1605.
40. Y. S. Kim and H. H. Lee, Adv. Mater., 2003, 15, 332–334.
41. M. Cavallini and F. Biscarini, Nano Lett., 2003, 3, 1269–1271.
42. J. Hyun, H. Ma, Z. Zhang, T. P. Beebe and A. Chilkoti, Adv. Mater., 2003, 15, 576–579.

---

Footnote

† The HTML version of this article has been enhanced with colour images.

---