Surface Structuring Meets Orthogonal Chemical Modifications: Toward

Apr 23, 2017 - ... Orthogonal Chemical Modifications: Toward a Technology Platform for ... Center for Interactive Materials and Bioinspired Technologi...
0 downloads 0 Views 6MB Size
Article pubs.acs.org/journal/abseba

Surface Structuring Meets Orthogonal Chemical Modifications: Toward a Technology Platform for Site-Selectively Functionalized Polymer Surfaces and BioMEMS Maria Vöhringer,† Wibke Hartleb,† and Karen Lienkamp* Department of Microsystems Engineering (IMTEK) and Freiburg Center for Interactive Materials and Bioinspired Technologies (FIT), Albert-Ludwigs-Universität, Georges-Köhler-Allee 103, 79110 Freiburg, Germany S Supporting Information *

ABSTRACT: A manufacturing process for the site-selective modification of structured (bio)material surfaces with two different polymers/biomolecules is presented. In the first step, a chemical surface contrast is created (e.g., a gold-on-silicon contrast obtained by colloidal lithography), and is combined with two orthogonal surface reactions for polymer/ biomolecule immobilization. To demonstrate this, an antimicrobial SMAMP polymer and a protein-repellent polyzwitterion were site-selectively surface-immobilized on the gold−silicon structures. By varying the structure spacing and the surface architecture, structure−property relationships for the interaction of these bifunctional polymer surfaces with bacteria and proteins were obtained (studied by fluorescence microscopy, atomic force microscopy, surface plasmon resonance spectroscopy, and antimicrobial assays). At 1 μm spacing, a fully antimicrobially active bifunctional material was obtained, which also nearquantitatively reduced protein adhesion. As the process is generally applicable to polymers/biomolecules with aliphatic CHgroups, it is an interesting platform technology for site-selectively functionalized bifunctional (Bio)MEMS. KEYWORDS: BioMEMS, colloidal lithography, chemical surface modification, material−cell interactions, structured surface, polymer surface



INTRODUCTION Nanostructured surfaces and surfaces with submicrometer-sized features play a role in many emerging technologies, e.g., BioMEMS,1−4 organic electronics,5−9 sensors,10−16 and microfluidics.17−20 The amount of techniques used to fabricate such structures is continuously increasing and includes lithographic techniques combined with etching or evaporation processes, direct writing techniques, micro/nano-machining, and/or micro/nano-molding.21,22 Surface structures having a chemical contrast, i.e., materials consisting of more than one chemical entity, can be obtained by self-assembly of block copolymers or nanoparticles,22 by lithographic methods combined with evaporation processes,23 by microprinting techniques,24 or by direct writing techniques such as electron beam writing and dip-pen nanolithography.22,25 Recently, triggering of UVactivated chemical reactions through lithographic masks was used for site-selective and orthogonal surface patterning.26−30 Remarkably though, processes that can be used to selectively modify distinct parts of already existing micro- or nanostructures are rather scarce and often not generally applicable.31−33 We here report a process (based on colloidal lithography23,34) that combines surface structuring with orthogonal chemical surface modification steps. It can be used to site-specifically and orthogonally immobilize two different polymers or biomolecules on a surface with submicrometer-sized (or larger) structures having a gold-on-silicon contrast. The process has several strong points that make it versatile and generally applicable: First, it can be used on many surface combinations © XXXX American Chemical Society

where one part contains a thiol-reactive metal (gold, silver, copper etc.), and another surface part carries OH groups, or can be oxidized (e.g., by plasma oxidation, flaming) to obtain these groups. This includes silicon, many metals, and most polymers. Second, the size of the structures is not limited by the diffraction limit (as is the case in the above-mentioned combination of lithography with UV-activated reactions), but depends on the availability of a suitable mask through which either the thiol-reactive metal or the OH-group bearing surface part (e.g., SiO2) can be evaporated. (In colloidal lithography, structure spacings as small as 20 nm are accessible.) Third, almost any polymer or biomolecule can be immobilized with the two presented linker molecules as long as it carries aliphatic CH groups (see Discussion below). In this paper, we first describe the surface structuring process. Second, we demonstrate its application for the fabrication of structured bifunctional bioactive polymer surfaces. More specifically, we attached one antimicrobial and one proteinrepellent polymer site-selectively on a periodic gold−silicon contrast obtained by colloidal lithography. We then varied the spacing and surface architecture to obtain structure−property relationships for the interaction of the bifunctional polymer surfaces with bacteria and proteins. Specifically, we studied protein adhesion, which is important in the early stages of Received: March 6, 2017 Accepted: April 22, 2017 Published: April 23, 2017 A

DOI: 10.1021/acsbiomaterials.7b00140 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering bacterial biofilm formation.35 This is an important area of study because the contamination of medical device surfaces with microorganisms and the resulting biofilm formation lead to severe infections.36,37 Also, the effect of the domain sizes of intrinsically antimicrobial polymers on the overall activity of structured surface is not yet understood, but this aspect could be used to suppress biofilm formation. For example, recent studies have shown that surface micro- and nanostructures can be exploited to modulate bacterial adhesion and proliferation.38,39 PDMS microstructures with feature sizes larger than bacteria efficiently reduced bacterial proliferation,40 while smaller ones reduced their adhesion and degree of organization.38,41−44 Nanostructures with very high aspect ratios could even break bacterial membranes.45−48 These findings inspired us to combine submicrometer-sized structures with additional bioactive components. With the structuring process described here, we obtained a bifunctional structured material that reduced growth of E. coli bacteria quantitatively and at the same time had 99.4% less protein adhesion than a control surface. We also could show that the structure spacing had a profound impact on the surface properties as detailed below, which gives new insight into the adhesion competition of proteins and bacteria on surfaces.



and pressed on with moderate force by hand. Then, the tape was slowly pulled off at a 90° angle to the substrate. The substrates with gold islands on silicon were subsequently rinsed 3 times with acetone and dried under a nitrogen flow. Selective Surface Functionalization. Thiolation. The gold part of the gold-on-silicon substrates was chemically modified under ambient conditions. Benzophenone lipoic acid ester (LS-BP) was synthesized as described in the literature.50 A solution of BP-lipoic acid ester (5 mmol mL−1 in toluene) was prepared and the substrates were immersed in the solution overnight (∼15 h). Afterward, the samples were rinsed with toluene and dried under nitrogen flow. Silanization. The silicon part of the gold-on-silicon substrates was chemically modified with triethoxy benzophenone (3EBP) that had been synthesized as described in the literature.51 A solution of 3EBPsilane (20 mg mL−1 in toluene) was spin-coated on the samples (3000 rpm, 30 s). The samples were then cured for 30 min at 70 °C on a preheated hot plate. Functionalization of Structured Substrates with Polymer Monolayers (Reference Surfaces). Propyl SMAMP polymer and PSB were synthesized as described in the literature.50,52 The procedure was performed under ambient conditions. Polymeric solutions of PSB (50 000 g mol−1, 10 mg mL−1 in trifluoroethanol (TFE)) and SMAMP (100 000 g mol−1, 10 mg mL−1 in dichloromethane (DCM)) were prepared, stored at 7 °C and used within 10 days. The polymer solutions were filtered and spin-coated onto the structured substrates (3000 rpm, 30 s). Afterward, the samples were exposed to UV light at a wavelength of λ = 254 nm (total irradiation energy 3 J) to covalently attach the polymer to the functionalized part of the surface. Unreacted polymer was removed by immersion in solvent (DCM for SMAMP, TFE for PSB, respectively) for 18 h under ambient conditions. The samples were dried under nitrogen flow. SMAMP Polymer Activation on Surfaces. The samples were immersed in HCl (4 M in dioxane) overnight (∼15 h) to completely deprotect the SMAMP amine groups. Afterward, the samples were washed with ethanol and dried under nitrogen flow. Surface Characterization. Atomic Force Microscopy. The topography of the surfaces was imaged with a Dimension FastScan and Icon from (Bruker, Karlsruhe, Germany). Commercial Bruker FastScan-A cantilevers (length: 27 μm; width: 33 μm; spring constant: 18 N m−1; resonance frequency: 1400 kHz) and Bruker ScanAsyst Air cantilevers (length: 115 μm; width: 25 μm; spring constant: 0.4 N m−1; resonance frequency: 70 kHz) were used. All AFM images were recorded in tapping mode in air and ScanAsyst mode in air, respectively. The obtained images were analyzed and processed with the Bruker software ‘Nanoscope Analysis 1.5’. Ellipsometry. The layer thickness of SMAMP and PSB monolayers on nonstructured silicon substrates was determined with the autonulling imaging ellipsometer Nanofilm EP3 (Nanofilm Technology GmbH, Göttingen, Germany). A refractive index of 1.5 was assumed for all experiments. Average values were obtained from 3 positions on one sample. The EP4 model was used to fit the data. Contact Angle Measurements. The static, advancing and receding contact angles were measured at five different places of each sample, and the average was reported. The static contact angles were calculated with Laplace−Young equations, whereas the advancing and receding angles were determined via elliptical and tangent methods.52 Fluorescence Microscopy. Fluorescence microscopy images were taken on a Nikon Eclipse Ti−S inverted microscope (Nikon GmbH, Düsseldorf, Germany) using a green-fluorescent-protein filter and a DAPI filter at 60-fold magnification. The imaging time was varied between 80 ms and 1 s, and the images were processed with the software ImageJ. Brightness and image contrast were manually adjusted. Antimicrobial Activity Assay. The experiments were performed using a modification of the Japanese Industrial Standard JIS Z 2801:2000 ‘Antibacterial Products Test for Antibacterial Activity and Efficacy’, as reported previously.53 S. aureus (ATCC29523) and E. coli (ATCC25922) were cultured overnight in tryptic soy broth and diluted 1:10. Optical density was checked 3−4 h later and the bacterial culture (1.5 mL of S. aureus and 150 μL of E. coli) was mixed in a

EXPERIMENTAL SECTION

Synthesis and Surface Fabrication. General. All chemicals and solvents were obtained in p.a. quality from Carl Roth (Karlsruhe, Germany), Sigma-Aldrich (Munich, Germany), or Acros (Geel, Belgium), and used as received. Silicon substrates were 1.5 × 1.5 cm2 pieces cut from single-side polished silicon wafers (525 ± 25 μm thick standard Si (CZ) wafer with [100] orientation) from Si-Mat (Kaufering, Germany). 200 nm, 500 nm and 1 μm polystyrene beads were obtained from Micromod GmbH (Rostock, Germany). The spincoater used was a SPIN150-NPP (SPS-Europe, Netherlands). The UV irradiation unit was a BIO-LINK-Box (Vilber Lourmat GmbH, Germany), with 254 nm light source. Polymers were synthesized as described elsewhere.49 Preparation of Colloidal Monolayers as Lithographic Masks. One mL of 200 nm, 500 nm or 1 μm colloidal dispersions was centrifuged at 4000 rpm for 4 min in an Eppendorf tube. The water residue was removed with a syringe. Nine hundred μL of a 2 wt % 1-aminohexane in ethanol solution was added to the Eppendorf tube and the dispersion was vortexed until evenly distributed. The suspension was aspirated with a 1 mL syringe. A needle with a length of 40 mm and a diameter of 0.8 mm was attached to the syringe, and bent at about 90°. The syringe was put into a syringe pump and the pumping volume was set to 3 μL h−1 (for 1 μm and 500 nm colloids) and 2 μL h−1 (200 nm colloids), respectively. Clean silicon substrates were put in a Petri dish and 10 mL of Milli-Q water were added. A homemade plastic ring was put on the water surface to confine the colloids. The needle tip was placed accurately at the water/air interface within the plastic ring and the suspension was dispensed until the area limited by the ring was fully covered. In order to transfer the thus created colloidal monolayer to the substrates, the water was removed with a 10 mL syringe (needle placed outside the plastic ring) until a 1 mm thick film of liquid remained in the Petri dish. The Petri dishes were covered with lids to avoid dust contamination, and the water layer was then evaporated under ambient conditions for 3−4 days. Metal Deposition and Lift off. The colloidal monolayers were used as lithographic masks for the evaporation of 5 nm of chromium as adhesive layer, and 40 nm gold as chemical contrast. All metal deposition processes were performed in the cleanroom by the Reinraumservice-Center (RSC) of the University of Freiburg using the device UNIVEX 300 from Leybold Vakuum. The deposition rate for both Cr and Au was 0.2 nm s−1. After metal evaporation, the colloidal mask was removed using adhesive tape (Scotch Tape). The adhesive tape was carefully placed on the surface avoiding air bubbles B

DOI: 10.1021/acsbiomaterials.7b00140 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering chromatography sprayer bottle with 100 mL of sterile NaCl 0.9% solution and continuously stirred.53 The test samples (5 of each material), including positive and negative controls, were fixed at the center of sterile Petri dishes each and placed at a distance of 15 cm to the spray nozzle. Then the bacterial suspension was sprayed onto the samples using compressed air from a 50 mL syringe.52 Afterward, the Petri dishes were immediately covered and incubated for 2 h in a humid chamber at 37 °C under aerobic conditions and 5% CO2. Fifty microliters of sterile 0.9% NaCl solution was added onto the samples and left for 2 min. The thus dispersed bacteria were aspirated with a pipet. To ensure removal of all bacteria from that surface area, the solution was pumped back and repipetted twice, and then spread over Columbia blood agar plates. These were incubated overnight at 37 °C without agitation. The number of colony forming units (CFUs) were counted with the software “Quantitiy One”. Each experiment was performed at least twice. SPR Measurements. SPR measurements were performed on a RT2005 RES-TEC device in Kretschmann configuration from ResTec, Framersheim, Germany. Excitation was effected with a He−Ne− Laser with λ = 632.8 nm. SPR substrates were homemade (LaSFN9 glass from Hellma GmbH, Müllheim, Germany; coated with 1 nm Cr and 50 nm Au at the Clean Room Service-Center (RSC) of the Department of Microsystems Engineering, University of Freiburg, using the device CS 730 S, Von Ardenne, Dresden, Germany). SPR Sample Preparation. For the SPR sample preparation, colloidal monolayers were deposited on SPR substrates analogously to the procedure described for the silicon wafers. After evaporation of 5 nm of chromium, SiO2 was sputtered through that lithographic mask at the Fraunhofer IAF using the device SyrusPro 710 by Leybold Optics (plasma-assisted electron beam evaporation at room temperature, source material SiO2). The colloids were removed with tape, and the samples were washed as described for the silicon wafers. SPR Angular Scans. To study the buildup of the material, a full reflectivity curve was measured after each fabrication step. Before and after the adsorption experiments in the kinetics mode, full angular scans of the dry substrates were also measured. The average thickness of each substrate layer was calculated by simulations based on the Fresnel equations, which were performed with the software “Winspall”. The following permittivities ε′ and ε″ were used: LaSFN9 (ε′ = 3.4036; ε″ = 0); Cr (ε′ = −6.3; ε″ = 20); Au (ε′ = −12.3; ε″ = 1.29) SiO2 (ε′ = 2.13; ε″ = 0) PSB, SMAMP (ε′ = 2.341; ε″ = 0) LS-BP, 3EBP, fibrinogen (ε′ = 2.25; ε″ = 0), nitrogen (ε′ = 1; ε″ = 0). SPR Kinetics Experiments. Protein adsorption was studied in the kinetics mode. To set up the experiment, an angular scan of the substrate under HEPES flow was performed to detect the minimum. The protein adsorption experiments in the kinetic mode were then carried out at θexp = θmin − 1 and a flow rate of 50 μL h−1 of the fibrinogen solution (1 mg mL−1). To determine the thickness of adsorbed fibrinogen after the kinetics experiment, the surfaces were rinsed with Milli-Q water for 2 min to remove residual salt and dried under nitrogen flow. Afterward, a reflectivity curve was measured again. The thickness of each protein layer was calculated by simulations based on the Fresnel equations, as described above.



Figure 1. Illustration of the surface structuring process: first, a monolayer of polystyrene colloids was formed on silicon and used as a lithographic mask, through which chromium and gold were evaporated. After lift-off, the gold areas (yellow) were reacted with the LS-BP linker carrying a benzophenone and a sulfide group. This yielded selective anchor points for polymer 1 (green) on gold. The silicon patches (dark gray) were then reacted with the 3EBP linker consisting of a benzophenone group and a triethoxysilane group. This formed anchor points for polymer 2 (blue) exclusively on silicon.

gold-plated colloid layer was removed using adhesive tape (“liftoff”), a hexagonal pattern of gold islands on a silicon background was obtained (Figure 1). The gold islands were then functionalized with lipoic acid-benzophenone (LS-BP),50 a linker molecule carrying a disulfide group and a UV-reactive benzophenone group. Due to the selectivity of sulfur for gold, the gold islands were functionalized, but not the silicon background. Next, the first polymer was spin-coated onto the thus structured surface and irradiated with UV light at 254 nm. The UV irradiation triggered a C−H insertion reaction between the benzophenone keto group and neighboring C−H bonds of the first polymer, so that a covalent bond was formed. Unbound polymer was removed by washing, and a monolayer of surface-immobilized polymer covering only the gold islands was obtained. In the next step, a linker molecule with orthogonal reactivity relative to LS-BP was used. This molecule consisted of a benzophenone group attached to a triethoxysilane residue (3EBP, Figure 1)51,57 and selectively reacted with the OH groups of the native silicon oxide layer on the silicon background. The second polymer was then spincoated onto the structured surface and also UV irradiated. By the same mechanism as described for LS-BP, a monolayer of the second polymer reacted with the benzophenone group and was covalently and site-selectively attached to the silicon background by 3EBP. After washing, the desired bifunctional surface with the two site-selectively immobilized polymers was obtained.

RESULTS

Process Design and Target Material. For our target application, we designed a surface functionalization process that allows the selective and site-specific immobilization of two different polymers on a substrate with a gold−silicon contrast (Figure 1). Colloidal lithography was chosen to create the contrast because it is a straightforward method to generate submicrometer-sized surface structures, and the lateral dimensions of the obtained pattern can be easily altered by varying the colloid size.23 First, a monolayer of polystyrene colloids was assembled on a silicon wafer piece, and used as a lithographic mask.54−56 Next, 5 nm of chromium was evaporated through that mask as an adhesive layer for gold, then a 40 nm thick gold film was deposited. When the thus C

DOI: 10.1021/acsbiomaterials.7b00140 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering Our target material was a surface with sub-micrometer-sized antimicrobial and protein-repellent polymer patches next to each other, like the black and white fields of a chess board (Figure 2). This material was designed to have two mechanisms

Figure 2. Cartoon illustration of the target material. Antimicrobial SMAMP patches (green) and protein-repellent PSB patches (blue) are assembled next to each other, like the black and white fields of a chess board.

by which adhesion of bacterial pathogens and thus biofilm formation on the surface would be prevented. The proteinrepellent poly(sulfobetaine) (PSB) patches prevent adhesion of biomolecules and bacteria.58 Any bacteria that manage to attach to the surface in spite of the PSB patches are killed by the antimicrobial poly(oxonorbornene)-SMAMP (Figure 2).52 Both polymers are biomimetic: the polyzwitterionic PSB imitates the zwitterionic lipids of mammalian cell membranes,58 and the cationic, antimicrobial SMAMP mimics the facially amphiphilic structure of natural antimicrobial peptides.52,53,59 Since the antimicrobial SMAMP is protein-adhesive, the SMAMP patches had to be dimensioned as small as possible. However, they could not be too small, as insufficient size would limit their interaction with the bacterial cell membranes and thus make them inactive. To determine which patch size most effectively prevents both protein adhesion and bacterial proliferation, we varied the dimensions of the polymer patches from 200 nm to 1 μm. To visualize the polymers immobilized on the surface structures, they were fluorescently labeled: SMAMP was labeled with the green-fluorescent NBD-dye, PSB with the blue-fluorescent Coumarin dye (Figure 2). The synthesis of the two fluorescent polymers is published elsewhere.49 Surface Functionalization and Characterization. Atomic force microscopy (AFM) height and phase images were measured after each surface functionalization step and are included in Figures 3 and 4. Figure 3a−c shows atomic force microscopy (AFM) height images of the lithographic masks made from polystyrene colloids with spacings of 200 nm, 500 nm, and 1 μm. The images show that the degree of long-range order of the colloids decreased with decreasing spacing because of the increasing size polydispersity of the colloids. While

Figure 3. Atomic force microscopy (AFM) height images of polystyrene colloid monolayers with a diameter of (a) 1 μm, (b) 500 nm, and (c) 200 nm. These were used as lithographic masks, through which first chromium and then gold were evaporated.

comparative experiments with SiO2 colloids gave lithographic masks with better long-range order (not shown), we used polystyrene colloids for our process because these colloids could be washed away with organic solvents in case the colloid layer was not quantitatively removed during the lift-off-step. AFM height and phase images of the gold-on-silicon structures (= Au_Si) that were obtained by evaporation of Cr and Au through these lithographic masks, followed by lift-off, are shown in Figure 4, together with representative height profiles (= cross-sections through the structures). The line in the height images indicates the position where these profiles were taken. The Au_Si structures obtained using the 1 μm and 500 nm masks were well-defined triangles. In some cases, percolations of two or more gold islands due to mask defects were observed. For the 200 nm masks, the defect rate was higher, and there were point defects due to missing colloids in the masks, which gave up to 200 nm wide gold blobs after gold evaporation and lift-off, in addition to the desired hexagonal pattern. The mean height of the gold islands was 50 nm for the 1 μm samples, 35 nm for the 500 nm samples and 20−40 nm for the 200 nm samples (determined from the AFM height images). D

DOI: 10.1021/acsbiomaterials.7b00140 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering

Figure 4. Atomic force microscopy (AFM) height images, height profiles, and phase images of the Au_Si, SMAMP@Au_Si, and SMAMP@ Au_PSB@Si structures with spacings of 1 μm, 500 nm, and 200 nm. The line in the height images indicate the position where the height profile was taken. E

DOI: 10.1021/acsbiomaterials.7b00140 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering The mean width of the gold islands was 300 nm, 250 nm, and 90−200 nm (for 1 μm, 500 and 200 nm spacing, respectively); the mean diameter of the background silicon patches corresponded to the size of the colloids used. When SMAMP was immobilized on the gold islands (=SMAMP@Au_Si), the relative size of the surface features (distance between Si background and the peak of the Au islands) increased (to 70 nm for the 1 μm structures, to 40 nm for the 500 nm structures, and to 30−40 nm for the 200 nm structures). The unmodified gold islands of Au_Si were observed as smooth features with a distinct domain contrast. Fuzzy or less-defined domain edges were obtained for the SMAMP@Au_Si samples. This corresponds to a slight spillover of the polymer into the silicon domains, which is particularly prominent for the samples with the 500 nm spacing, and is expected when high molecular weight polymers (50,000−100,000 g mol−1) are attached near the gold island edges. After immobilization of the second polymer (in this case PSB) on the silicon background, these fuzzy edges vanished; indicating that a monolayer of PSB had formed on the previously bare silicon patches. (These samples are named SMAMP@Au_PSB@Si in the following.) These data confirm the successful immobilization of the polymers on their respective sites, and thus the fluorescence microscopy results described below. Similar results were obtained when the polymer sequence was inverted, i.e., when PSB was immobilized on the gold islands. These samples were named PSB@Au_Si and PSB@Au_SMAMP@Si, respectively. The height profiles of the latter samples also show that the distance between the island peaks and the background of the sample decrease to below 20 nm, indicating that the valleys between the islands were filled with polymer. To further demonstrate that the SMAMP and the PSB were site-selectively immobilized on surface structures as described in the process design, fluorescence microscopy was used. For these images, SMAMP was labeled with a green NBD chromophore, and PSB was labeled with a blue Coumarin dye (Figure 2). Since the blue-green contrast of these two polymers is difficult to discern in printed pictures, we illustrate the site-selectivity of the functionalization process first on a sample area with many mask defects (Figure 5a, left). Using a filter for green fluorescence (GFP), the fluorescence micrograph of such an area shows that the green-fluorescent SMAMP indeed covered only the gold patches and left round, dark voids where the colloids were. When imaging the same area with a DAPI filter for blue fluorescence, we see that these previously “empty” sites are now exclusively covered by the bluefluorescent PSB (Figure 5a, middle). The overlay of both images demonstrates the site-selective immobilization and good colocalization of the respective polymers on their target sites. A fluorescence microscopy image of an area with fewer defects has been included in Figure 5b, where this time the blue PSB was attached on the gold islands and green SMAMP was attached to the silicon background. While the green-fluorescent SMAMP patches are still visible though at the resolution limit of the method, the dimensions of the blue-fluorescent PSB patches are below that limit and thus cannot be clearly discerned. The surface functionalization of all sample types for all there spacings (gold islands on silicon (= Au_Si); polymers on the gold islands only (= SMAMP@Au_Si and PSB@Au_Si, respectively); and bifunctional surfaces (= SMAMP@ Au_PSB@Si and PSB@Au_SMAMP@Si)) was further confirmed using contact angle measurements, and compared

Figure 5. Fluorescence micrographs of nanostructured bifunctional surfaces made from green-fluorescent antimicrobial SMAMP polymer and blue-fluorescent polyzwitterion (PSB). (a) Fluorescence micrographs (green filter, blue filter, overlay) of a sample with high defect rate demonstrate the site-selective immobilization of SMAMP on gold and of PSB on the silicon background (= SMAMP@Au_PSB@Si). (b) Inverse contrast: PSB on the gold islands, SMAMP on the silicon background (= PSB@Au_SMAMP@Si, green filter and blue filter, respectively).

to the contact angles of several reference surfaces (a blank Si wafer; homogeneous, unstructured PSB monolayer; homogeneous, unstructured SMAMP monolayer, see Table 1). An unstructured gold surface has and advancing and receding contact angle of 71°, with no hysteresis,60 and is thus more hydrophobic than a blank silicon wafer, which had a receding contact angle of 41° in its native, oxide-covered state. Thus, it is not surprising that the surfaces with gold islands on silicon (Au_Si) had receding contact angles of 63−67° for each of the three spacings used, i.e., they become more hydrophobic than the parent silicon surface. There were no significant changes in the contact angles of the Au_Si surfaces that could be attributed to size effects. It was evident from the data for the unstructured reference samples that the SMAMP monolayer (θreceding = 33°) was more hydrophobic than the PSB monolayer (θreceding = 22°). However, only little differences in the receding contact angles were observed between the structures with PSB on the gold islands (= PSB@Au_Si) and those with SMAMP on the gold islands (= SMAMP@Au_Si) when the underlying surface pattern had the same spacing. Within each series, however, both surface types with 200 nm spacing had significantly smaller contact angles (θreceding) than the surfaces with larger spacing. We can explain this with the larger number of defects in the lithographic mask for the 200 nm spacing. These defects were filled with gold in the lithographic processing step, which resulted in a larger surface area ratio of gold to silicon, and consequently in a larger amount of hydrophilic polymer on the monofunctionalized surface with 200 nm spacing compared to the corresponding 500 nm and 1 μm structures. Alternatively, F

DOI: 10.1021/acsbiomaterials.7b00140 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering

attached to the gold background through LS-BP. SPR measurements (reflectivity versus angle) were performed after each surface modification step for. This is shown for SMAMP immobilized on SiO2 islands and PSB on the gold background (= SMAMP@SiO2_PSB@Au) in Figure 6a, and for the inverse

Table 1. Contact Angle Data (Static, Advancing, and Receding Contact Angles) for the Structured Surfaces and Selected Reference Surfaces contact angle (deg) sample type

θ

θadvancing

θreceding

88 ± 3 81 ± 1 73 ± 4

91 ± 3 86 ± 2 74 ± 4

65 ± 3 63 ± 1 67 ± 2

38 ± 3 53 ± 3 51 ± 3

43 ± 2 70 ± 1 62 ± 2

31 ± 2 42 ± 2 43 ± 1

56 ± 3 60 ± 2 63 ± 3

55 ± 3 69 ± 2 71 ± 4

33 ± 2 45 ± 1 47 ± 3

21 ± 3 43 ± 1 45 ± 1

29 ± 3 48 ± 1 51 ± 1

10 ± 2 33 ± 1 37 ± 1

52 ± 3 56 ± 2 53 ± 3

56 ± 3 57 ± 1 59 ± 2

21 ± 1 27 ± 2 34 ± 3

71 ± 1 34 ± 3 59 ± 3

75 ± 3 35 ± 3 61 ± 3

41 ± 3 22 ± 1 33 ± 3

static

Au_Si 200 nm 500 nm 1 μm PSB@Au_Si 200 nm 500 nm 1 μm SMAMP@Au_Si 200 nm 500 nm 1 μm PSB@Au_SMAMP@Si 200 nm 500 nm 1 μm SMAMP@Au_PSB@Si 200 nm 500 nm 1 μm reference samples blank Si wafer PSB monolayer SMAMP monolayer

this may also be a direct effect of the nanostructures themselves−according to Wenzel’s model, a hydrophilic surface will become more hydrophilic the rougher it gets.61 The same trend of decreasing θreceding with decreasing structure size was found for the surfaces that were backfilled with the second polymer (PSB on the gold islands and SMAMP on the silicon patches, PSB@Au_SMAMP@Si, and the inverse SMAMP@ Au_PSB@Si structure). Unsurprisingly, these fully polymercovered surfaces had overall lower contact angles than the monofunctionalized surfaces (PSB@Au_Si and SMAMP@ Au_Si, respectively). The receding contact angle of PSB@Au_SMAMP@Si was even lower than that of the parent PSB monolayer, which may also be attributed to the surface structure. While no difference was observed between SMAMP@Au_Si and PSB@Au_Si, provided they had the same structure spacing, PSB@ Au_SMAMP@Si surfaces were overall less hydrophobic than SMAMP@Au_PSB@Si.. Ellipsometry measurements could not be used to measure the average thickness of the structured surfaces because the surface patterns caused optical interference. Therefore, average thickness measurements after each functionalization step were obtained by surface plasmon resonance spectroscopy (SPR). Because SPR requires gold surfaces as sensors, the surface fabrication process had to be slightly modified: Instead of using gold islands on a silicon background, we switched to SiO2 islands on a gold background. In an analogous process to the one shown in Figure 1, we deposited SiO2 through the colloidal masks onto the gold surface. After lift-off, the thus obtained SiO2 islands were functionalized with 3EBP silane, and the first polymer was immobilized on the islands. The gold background was then reacted with LS-BP, and the second polymer was

Figure 6. Build-up of the bifunctional material on surfaces with a SiO2_Au contrast (1 μm spacing) studied by surface plasmon resonance spectroscopy (SPR) for (a) SMAMP@SiO2_PSB@Au and (b) PSB@SiO2_ SMAMP@Au. Reflectivity curves after each processing step (SiO2 islands, 3EBP functionalization, immobilization of polymer 1, LS-BP functionalization, immobilization of polymer 2) are shown.

material (= PSB@SiO2_ SMAMP@Au) in Figure 6b. Full scans of the surface angle vs reflectivity are shown for each sample. As expected, the reflectivity minimum shifted to higher measurement angles after each processing step. This already qualitatively indicates an increase of the amount of material present on the surface. To determine the average thickness of each surface feature (“layer”), we fitted the reflectivity curves with the Fresnel equation using the Winspall software (see Experimental Section). The results obtained are summarized in Table 2. The data confirm the expected thickness increase for each processing step. For SMAMP@SiO2_PSB@Au, the polymer layers were 16.7 nm (SMAMP) and 9.0 nm (PSB) thick, i.e., the higher-molecular-weight SMAMP formed a thicker layer than the PSB, which is plausible. The same was observed for the inverse PSB@SiO2_SMAMP@Au samples (SMAMP layer = 15.0 nm; PSB layer = 9.0 nm, respectively). Although the shift of the reflectivity minimum in the LS-BP scan compared to the SMAMP scan in Figure 6a appeared large, the fit G

DOI: 10.1021/acsbiomaterials.7b00140 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering

Table 2. Average Layer Thickness and Permittivity (ε′ = real part, ε″ = imaginary part) for the Bifunctional Materials Calculated from Fits to the Surface Plasmon Resonance (SPR) Curves Shown in Figure 6 SMAMP@SiO2_PSB@Au LaSNFN9 glass Cr Au SiO2 3EBP Polymer 1 LS-BP Polymer 2

PSB@SiO2_SMAMP@Au

layer thickness (nm)

ε′

ε″

1.0 48.0 11.5 0.9 16.7 2.0 9.0

3.404 −6.3 −12.54 2.13 2.25 2.38 2.25 1.99

0 20 1.72 0 0 0 0 0

indicated a layer thickness increase of 2.0 nm, which is consistent with the expected LS-BP monolayer formation. The large thickness increase for the 3EBP layer in Figure 6b (4.9 nm) can be attributed to network formation of the triethoxy groups due to slightly too long reaction times. Another set of SPR reflectivity curves for SiO 2 _Au, PSB@SiO 2 _Au, SMAMP@SiO2_Au, SMAMP@SiO2_PSB_Au, and PSB@ SiO2_SMAMP@Au at all three spacings is shown in Table S1 and confirms the here-presented SPR data for the 1 μm structures. Biological Characterization of Structured Surfaces. To understand the effect of the surface structure spacing on the early stages of biofilm formation, we studied the structured materials with two biological assays. First, the antimicrobial activity against the rod-shaped, Gram-negative Escherichia coli bacterium was tested; second, the protein-repellency of the materials was investigated using SPR kinetics experiments. For the antimicrobial activity test, the surfaces were sprayed with a suspension containing 1 × 106 E. coli bacteria per cm3 using a standardized procedure, as described in the Experimental Section.53,62 After spraying and incubation, an aliquot of the bacteria was removed from the surface and cultivated. The resulting number of colony forming units (CFUs) was counted (Figure 7a, including the nonfunctionalized structures (Au_Si), structures with one polymer on the gold islands (SMAMP@ Au_Si and PSB@Au_Si, respectively), and the bifunctional structures (SMAMP@Au_PSB@Si and PSB@Au_PSB@Si, respectively)). When only PSB was present on the gold islands (PSB@Au_Si), the antimicrobial activity was low: 20% surviving CFUs were found for the 200 nm spacing, 40% CFUs for the 500 nm spacing, and more than 80% CFUs for the 1 μm spacing. Thus, more bacteria grew on these surfaces when the spacing was large. We know from other experiments that the PSB itself is not intrinsically antimicrobially active; it had 97% CFUs in similar tests. Thus, we can infer that the reduction of bacteria by PSB@Au_Si is not an antimicrobial effect, but caused by a reduction of bacterial adhesion, which in turn affects the colony formation rates and the overall bacterial viability. PSB@Au_Si with the smallest spacing reduced bacterial adhesion the most, whereas the 1 μm spacing had only a weak effect on the colonies. Because E. coli is about 10 μm long and about 1 μm in diameter, it seems plausible to assume that the 200 nm spacing was dense enough to prevent attachment of the bacteria. For the 1 μm structures, on the other hand, E. coli apparently had enough surface contact to the nonfunctionalized silicon patches in the background to adhere there. SMAMP@Au_Si had a pronounced antimicrobial activity for all three spacings. This indicates that the SMAMP patches were sufficiently large to damage the bacterial cell

layer thickness (nm)

ε′

ε″

0.6 50.3 20.0 4.9 9.0 2.0 15.0

3.404 −6.3 −11.86 2.13 2.13 2.33 2.25 2.05

0 20 1.68 0 0 0 0 0

membrane for either spacing, which was also reported for nonstructured SMAMP monolayers and networks.52,53 The bifunctional materials SMAMP@Au_PSB@Si and PSB@Au_PSB@Si behaved quite similar to each other in the antimicrobial activity tests. For both materials, the 200 nm structures had the highest number of colony forming units (i.e., were the least antimicrobially active), while the 500 nm and 1 μm structures were significantly more antimicrobial. This indicates that the access to the SMAMP patches of the 200 nm structures was restricted compared to the situation in SMAMP@Au_Si. For the larger spacings, the antimicrobial patches were again within reach for the bacteria; therefore these materials were highly antimicrobial against E. coli bacteria. Interestingly, the nonfunctionalized Au_Si material also reduced the number of colony forming units tremendously. This is in line with previous reports that nanostructures made from nonintrinsically antimicrobial materials can inhibit bacterial growth,42−44,47,63 which will be discussed below. Surface plasmon resonance spectroscopy was used to study protein adhesion on the structured surface. We performed SPR kinetics experiments (time-dependent measurement of the reflectivity intensity at constant angle) to study fibrinogen adhesion on the structured surfaces. The results of representative kinetics experiments for SMAMP@SiO2_Au and SMAMP@SiO2_PSB@Au are shown in Figure 7b, c, respectively. The data for the other structured surfaces is included in the Table S2. In each kinetics experiment, the baseline reflectivity was first recorded for 10−20 min. After that, fibrinogen was injected and flowed over the surface (circles in Figure 7b), which resulted in an increase in reflectivity if the sample was protein-adhesive (e.g., for SMAMP@SiO2_Au, Figure 7b). When the material was protein-repellent, the curve stayed at about the same reflectivity value (particularly for SMAMP@SiO2_PSB@Au, Figure 7c). After about 15 min of protein exposure, the surface was washed by injecting buffer (stars in Figure 7b), so that any reversibly attached protein was removed. This typically reduced the reflectivity level slightly. The difference of the initial baseline and the reflectivity value after washing is a relative measure of the protein repellency of the material (assuming identical permittivity of all materials involved). To confirm this relative data quantitatively, wewashed each material was washed with water after exposure to the protein and then dried, and full angular SPR scans of the dry layer were taken. These were compared to full scans from before the kinetics experiment to determine any thickness increase due to protein adsorption (Table S1). From the dry layer thickness difference before and after protein exposure, the average layer thickness of adhered protein on the surface could be calculated in units of nm and H

DOI: 10.1021/acsbiomaterials.7b00140 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering

Table 3. Average Amount of Fibrinogen (in ng mm−2) Adsorbed on the Structured Materialsa adhered fibrinogen/ng mm−2 material type

200 nm

500 nm

1 μm

SiO2_Au PSB@SiO2_Au SMAMP@SiO2_Au PSB@SiO2_SMAMP@Au SMAMP@SiO2_PSB@Au

0.0 0.0 10.8 0.0 0.0

3.7 0.2 2.8 0.0 0.0

10.9 0.0 16.3 0.1 0.2

The layer thickness was obtained from fits to surface plasmon resonance spectroscopy angular scans before and after protein exposure, and converted into mass per area as described in the main text. a

SiO2_Au structures (500 nm and 1 μm spacing) adsorbed 3.7 and 10.9 ng mm−2 of protein, respectively. This was significantly less than unmodified gold, while SiO2_Au at a spacing of 200 nm was fully protein-repellent (the detection limit of the setup is