Protein Microarrays Based on Polymer Brushes Prepared via Surface

Biomacromolecules 2010, 11, 3467–3479

3467

Protein Microarrays Based on Polymer Brushes Prepared via Surface-Initiated Atom Transfer Radical Polymerization Raphael Barbey,† Ekkehard Kauffmann,‡ Markus Ehrat,‡ and Harm-Anton Klok*,† Ećole Polytechnique Fe´de´rale de Lausanne (EPFL), Institut des Mate´riaux and Institut des Sciences et Ingeńierie Chimiques, Laboratoire des Polyme`res, Baˆtiment MXD, Station 12, CH-1015 Lausanne, Switzerland, and Zeptosens - A Division of Bayer (Schweiz) AG, Benkenstrasse 254, CH-4108 Witterswil, Switzerland Received August 24, 2010

Polymer brushes represent an interesting platform for the development of high-capacity protein binding surfaces. Whereas the protein binding properties of polymer brushes have been investigated before, this manuscript evaluates the feasibility of poly(glycidyl methacrylate) (PGMA) and PGMA-co-poly(2-(diethylamino)ethyl methacrylate) (PGMA-co-PDEAEMA) (co)polymer brushes grown via surface-initiated atom transfer radical polymerization (SI-ATRP) as protein reactive substrates in a commercially available microarray system using tantalum-pentoxidecoated optical waveguide-based chips. The performance of the polymer-brush-based protein microarray chips is assessed using commercially available dodecylphosphate (DDP)-modified chips as the benchmark. In contrast to the 2D planar, DDP-coated chips, the polymer-brush-covered chips represent a 3D sampling volume. This was reflected in the results of protein immobilization studies, which indicated that the polymer-brush-based coatings had a higher protein binding capacity as compared to the reference substrates. The protein binding capacity of the polymer-brush-based coatings was found to increase with increasing brush thickness and could also be enhanced by copolymerization of 2-(diethylamino)ethyl methacrylate (DEAEMA), which catalyzes epoxide ring-opening of the glycidyl methacrylate (GMA) units. The performance of the polymer-brush-based microarray chips was evaluated in two proof-of-concept microarray experiments, which involved the detection of biotin-streptavidin binding as well as a model TNFR reverse assay. These experiments revealed that the use of polymer-brushmodified microarray chips resulted not only in the highest absolute fluorescence readouts, reflecting the 3D nature and enhanced sampling volume provided by the brush coating, but also in significantly enhanced signal-to-noise ratios. These characteristics make the proposed polymer brushes an attractive alternative to commercially available, 2D microarray surface coatings.

Introduction Protein microarrays are powerful devices that allow highly multiplexed measurements of protein-protein interactions of protein abundances and of post-translational modifications.1-7 Because gene expression levels do not directly reflect protein abundance and because the cellular activities are strongly affected by post-translational modifications, protein microarray analysis can complement information obtained from DNA microarrays. Furthermore, proteins, and not DNA, are the major targets of pharmaceutical compounds for disease treatments. Whereas minute quantities of sample DNA can be amplified using polymerase chain reaction (PCR), the absence of a similar protein amplification method requires optimized assay procedures such as sample handling, high-affinity recognition elements, as well as highest sensitivity detection systems. To optimize signal-to-noise ratios, the surfaces of protein microarrays must provide a high binding capacity and a good accessibility to the immobilized proteins.1,4,7,8 Currently, two types of protein microarray chip surfaces can be distinguished.9-12 The first class consists of a 2D, planar protein binding interface, which can be, for example, a dodecylphosphate (DDP),13 amine,14 epoxide,14 aldehyde,15 or poly-L-lysine16 modified glass or plastic surface. A second group of protein microarrays is composed of chips covered with a porous poly(vinylidene * Corresponding author. Email: [email protected]. Fax: + 41 21 693 5650. Tel: + 41 21 693 4866. † ´ Ecole Polytechnique Fe´de´rale de Lausanne (EPFL). ‡ Zeptosens - A Division of Bayer (Schweiz) AG.

fluoride) (PVDF)17 or nitrocellulose18 membrane or a thin acrylamide19,20 or agarose21 hydrogel coating. This second class of surfaces is of particular interest because it provides a higher binding capacity than 2D surfaces. Whereas the increased binding capacity of the 3D versus the 2D surfaces is a potential advantage, accessibility, reproducibility of the surface structure, and enhanced background fluorescence are potentially limiting factors of the current 3D surfaces. Polymer brushes, viz. thin polymer coatings in which the individual polymer chains are tethered with one of their chain ends to the underlying substrate, represent a potentially interesting alternative to the 3D protein microarray surfaces discussed above, especially when they are prepared in a bottom-up (“grafting from”) fashion via surface-initiated polymerization. In contrast to, for example, thin, spin-coated polymer films or brushes obtained by grafting appropriate end-functionalized polymers onto a substrate, surface-initiated polymerizations allow access to polymer brushes with relatively high chain densities. If the brushes are prepared from monomers with protein-reactive side-chain functional groups, then thin coatings can be prepared that present very high surface concentrations of protein binding groups.22-24 Because they allow accurate control over brush density, thickness, composition, and architecture, controlled/“living” radical polymerizations are particularly attractive for surface-initiated polymerization.22 The ability of polymer brushes to act as high-capacity protein binding surfaces has been demonstrated in a number of publications.25-39 Most of these reports have focused on the protein binding

10.1021/bm101297w  2010 American Chemical Society Published on Web 11/19/2010

3468

Biomacromolecules, Vol. 11, No. 12, 2010

Barbey et al.

protein-protein interactions and which allow the comparison of the assay performance of the (co)polymer-brush-based microarray chips with that of the commercial DDP-coated chip.

Experimental Section

Figure 1. Schematic representation of the proprietary, waveguide technology-based fluorescence reader (ZeptoREADER F3000) used in this contribution.

capacity of the polymer brush, the immobilization kinetics, or both. Systematic studies that investigate the feasibility of polymer brushes to bind proteins covalently and then evaluate their performance in model assays in a commercial protein microarray format, however, have not been reported so far to the best of our knowledge. This contribution discusses the feasibility of polymer brushes prepared via surface-initiated atom transfer radical polymerization (SI-ATRP) to act as a 3D protein binding coating in a commercial protein microarray format. The brushes utilized in this report are either poly(glycidyl methacrylate) (PGMA) homopolymer or poly(glycidyl methacrylate)-co-poly(2-(diethylamino)ethyl methacrylate) (PGMA-co-PDEAEMA) random copolymer brushes. These brushes contain epoxide groups that can react with, for example, amine groups present in proteins. The random copolymer brushes are of interest because it was recently shown that the incorporation of DEAEMA both accelerates the immobilization reaction and leads to increased protein binding in comparison to a PGMA homopolymer brush.39 The microarray system utilized in this report consists of a planar tantalum pentoxide (Ta2O5) optical waveguide chip in combination with a readout system based on evanescent field fluorescence excitation (Figure 1).13 In this commercial microarray system, the waveguide is modified with a 2D homogeneous, hydrophobic DDP monolayer onto which the proteins of interest are subsequently spotted. Because the evanescent field intensity decays exponentially with increasing distance from the surface and given that the height of the surface-confined field is generally on the order of 100 to 300 nm (depending on the measurement technique and design of the sensing device), control over coating thickness and homogeneity is of particular importance for evanescent field-based measurement techniques. Therefore, because of their specific characteristics, surfaceinitiated controlled polymerizations are particularly suitable for the preparation of 3D adsorptive coatings for this class of optical waveguide-based devices. In this contribution, the performance of Ta2O5 waveguide chips modified with PGMA and PGMAco-PDEAEMA (co)polymer brush coatings will be investigated and compared to that of DDP-modified chips. This contribution consists of two parts. The first part focuses on protein immobilization and studies protein binding on the (co)polymer brushes using a commercial spotter and standard microarray protocols as a function of different parameters, including brush composition and thickness as well as protein concentration and pH of the spotting solution. The second part of this report presents two proof-of-concept experiments, which were performed to evaluate the feasibility of the (co)polymer-brushcoated microarray chips to detect protein-small molecule and

Materials. Copper(I) chloride (CuICl, > 99%, Sigma-Aldrich), copper(II) bromide (CuIIBr2, 99.999%, Aldrich), 2,2′-bipyridyl (bpy, > 99%, Sigma-Aldrich), glycidyl methacrylate (GMA, >97%, Fluka), 2-(diethylamino)ethyl methacrylate (DEAEMA, 99%, Aldrich), bovine serum albumin (BSA, g98%, lyophilized powder, Sigma-Aldrich), ovalbumin (OVA, >90%, Sigma), fucosylamide-biotin labeled BSA (BSA-biotin, Sigma), recombinant human TNFR (Thermo Scientific), antihuman TNFR purified from rabbit serum (Pierce Endogen), Alexa Fluor 647 conjugates of BSA (BSA-AF647, Invitrogen) as well as ovalbumin (OVA-AF647, Invitrogen), streptavidin (SA-AF647, Invitrogen) and goat Fab fragments that specifically bind to the Fc portion of rabbit IgG antibodies (Fab-AF647, Zenon, Invitrogen) were used as received. Neutral (pH 7.4), basic (pH 8.2), and acidic (pH 6.5) spotting buffers (ZeptoMARK native spotting buffers, Zeptosens) as well as assay buffer (ZeptoMARK CAB1) were obtained from Zeptosens. Tween 20 (Sigma-Aldrich) was used for the preparation of 0.1 wt % Tween 20 detergent solution. Unless otherwise stated, all solvents were of puriss p.a. grade. Ultrahigh quality water with a resistance of 18.2 MΩ · cm (at 25 °C) was obtained from a Millipore Milli-Q gradient machine fitted with a 0.22 µm filter. The inhibitor in glycidyl methacrylate (4-methoxyphenol) was removed by passing the monomer through a column of activated, basic aluminum oxide, whereas DEAEMA was freed from its inhibitor (phenothiazine) via distillation under reduced pressure. SI-ATRP was performed from silicon wafers cut in pieces of 2 × 0.8 cm2, from silicon oxide-covered quartz crystal microbalance (QCM) sensors (International Crystal Manufacturing) or from Ta2O5 planar waveguides (ZeptoMARK chips, Zeptosens) with dimensions of 5.7 × 1.4 cm2. The ATRP initiator, 6-(chloro(dimethyl)silyl)hexyl 2-bromo-2-methylpropanoate, has been synthesized according to a published protocol.39 Analytical Methods. Brush thicknesses were determined on silicon substrates by means of a Philips Plasmos SD 2300 ellipsometer working with a He-Ne laser (λ ) 632.8 nm) at an angle of incidence of 70°. The calculation method was based on a three-layer silicon/polymer brush/ambient model, assuming the polymer brush to be isotropic and homogeneous. A fixed refractive index of 1.45 was used for the polymer layer. All reported ellipsometric film thicknesses represent an average over five data points taken from the same substrate and are corrected for the approximately 2-nm-thick native oxide layer on the silicon substrates. Atomic force microscopy (AFM) was performed in tapping mode on a Veeco multimode Nanoscope IIIa instrument mounted with a NSC14/no Al (MikroMasch) cantilever. XPS data were recorded on an Axis Ultra instrument from Kratos Analytical. These measurements were carried out with a conventional hemispheric analyzer. The X-ray source employed was a monochromatic Al KR (1486.6 eV) source operated at 150 W and 10-9 mbar. The analysis area was 700 × 350 µm2 at an angle of 90° relative to the substrate surface (takeoff angle). The pass energies were 80 and 20 eV for survey scans and highresolution elemental scans, respectively. The operating software used was Vision 2, which corrects for the transmission function. Charge compensation was performed with a self-compensating device (Kratos patent) using field-emitted low-energy electrons (0.1 eV). All XPS spectra were calibrated on the aliphatic carbon signal at 285.0 eV. The mass of bound proteins was determined using a QCM with dissipation (QCM-D) Q-Sense E4 system (Q-Sense) recording the resonance frequencies of the third, fifth, and seventh harmonic. Microarrays were fabricated by noncontact spotting using a GeSiM Nano-Plotter NP 2.1, which deposits ∼400 pL volume for each spot. A ZeptoFOG (Zeptosens) nebulizer was used to block the substrates with BSA. A planar waveguide system (ZeptoREADER F3000, Zeptosens) or a confocal fluorescence scanner (GMS 418, Affymetrix) was used to acquire microarray images.

Protein Microarrays Based on Polymer Brushes ATRP Initiator Functionalized Substrates. Silicon and tantalum pentoxide substrates as well as QCM-D sensors were functionalized with the ATRP initiator according to a published procedure.39 In brief, silicon- and Ta2O5-coated substrates were sonicated in acetone, ethanol, and Milli-Q water for 5 min each, and the QCM-D sensors were stirred in acetone for 10 min. Then, the substrates were rinsed with ethanol and dried under vacuum. Subsequently, to remove any organic residues from the surfaces and create active silanol groups, the QCM-D sensors were placed in a microwave-induced oxygen plasma system (200 W, Diener electronic GmbH) for 15 min, whereas the silicon and Ta2O5 surfaces were immersed in piranha solution (H2SO4/H2O2 ) 7:3 v/v) for 1 h at 140 °C. (Caution! This solution reacts Violently with organic materials.) The slides were extensively rinsed with deionized water and ethanol and dried under vacuum. Next, the substrates were exposed to a 2 mM solution of the ATRP initiator in toluene for 2 h at room temperature. Thereafter, they were removed from the reaction mixture and rinsed extensively with toluene and acetone, sonicated for 15 s in acetone (except QCM-D sensor), rinsed with ethanol, sonicated for 15 s in ethanol (except QCM-D sensor), then rinsed with Milli-Q water and ethanol. Finally, the initiator-functionalized surfaces were dried under vacuum and stored until needed for polymerization. Surface Patterning. For AFM thickness determination, the ATRP initiator-functionalized silicon and tantalum pentoxide surfaces were UV-irradiated at a distance of ∼5 cm for 5 min with a 200 W Hg-Xe UV spot light source (Lightningcure LC6, Hamamatsu) through 400 Mesh TEM grids (with 10 µm bar width, Athene Grids), which were used as photomasks. Preparation of Poly(glycidyl methacrylate) and Poly(glycidyl methacrylate)-co-poly(2-(diethylamino)ethyl methacrylate) Brushes. The (co)polymer brushes were synthesized following a published protocol.39 GMA and DEAEMA were mixed in molar ratios equivalent to 100:0 or 75:25 (GMA/DEAEMA) and then added, under stirring, to a methanol/water solution (monomer(s)/methanol/water 5:4:1 v/v/v). The mixture was subjected to two freeze-pump-thaw cycles before 2,2′bipyridyl (bpy), CuICl, and CuIIBr2 were added in quantities such as to obtain a final molar ratio of monomer(s)/CuICl/CuIIBr2/bpy of 2000/ 20/1/50. After two additional freeze-pump-thaw cycles, the reaction mixture was canula-transferred into nitrogen-purged reaction vessels containing the initiator-functionalized substrates. After a predefined polymerization time, the slides were removed from the ATRP solution, thoroughly rinsed with methanol, and then successively washed in methanol for 1 h, dichloromethane, and acetone for 30 min each. At the end of the process, the (co)polymer-brush-coated substrates were rinsed with ethanol and dried under vacuum. The (co)polymer-coated silicon wafers were used to determine the brush thickness by means of ellipsometry, whereas the (co)polymer-modified Ta2O5 surfaces were used for the microarray experiments. Protein Immobilization Studies. A GeSiM Nano-Plotter NP 2.1 was used to spot ∼400 pL of protein solutions on the DDP- and (co)polymer-brush-modified Ta2O5 substrates. Spotting solutions were composed of fluorescently labeled proteins (OVA-AF647, BSA-AF647, and Fab-AF647) in various concentrations (i.e., 6, 2, 0.6, and 0.2 nM) diluted in a BSA matrix (3, 1, 0.3, and 0.1 mg · mL-1, respectively) in acidic (pH 6.5), neutral (pH 7.4), and basic (pH 8.2) spotting buffers. After spotting, the slides were incubated for 24 h in a humidity chamber (RH ) 55%) and then extensively washed with 0.1 wt % Tween 20 for another 24 h to remove any physisorbed proteins. In a second series of experiments, surface areas of about 6 × 8 mm2 were incubated in 2 nM labeled protein (BSA-AF647 and OVA-AF647, in a 1 mg · mL-1 diluting BSA matrix) solutions in basic (pH 8.2) spotting buffer for 24 h, whereas BSA and pure basic spotting buffer were used as controls. After the predefined incubation time, the slides were extensively rinsed with basic spotting buffer. Quantification of Immobilized Proteins. The amount of immobilized proteins, BSA and OVA, has been determined according to a published procedure,39 which consists of measuring the shift in resonance frequency of the (co)polymer-coated QCM-D sensors after


3469

incubation in protein-containing solutions for 6, 24, and 48 h at 25 °C. In brief, the (co)polymer-coated sensors were first measured in their dry state, then taken out from the measurement cell and incubated in 50 mg · mL-1 protein (BSA or OVA) solutions in basic spotting buffer. After the predefined incubation times, the substrates were extensively rinsed with water and ethanol and finally dried under vacuum. Then, the resonance frequency of the sensors was measured, and the amount of immobilized protein was determined using the Sauerbrey relation, assuming the film to be thin and rigid in its dry state. Biotin-Streptavidin Assay. ∼400 pL spots of BSA-biotin labeled proteins with concentrations of 30, 10, 3, and 1 nM (in an unlabeled BSA matrix of 3, 1, 0.3, and 0.1 mg · mL-1, respectively) in basic spotting buffer as well as pure Milli-Q water and basic spotting buffer control solutions were deposited in duplicate onto the different substrates. After 24 h at a relative humidity (RH) of 55%, the slides were blocked with BSA for 30 min (using the ZeptoFOG instrument) and subsequently thoroughly washed with Milli-Q water, dried, and stored at 4 °C in the dark. For the assay, the slides were first equilibrated in assay buffer (ZeptoMARK CAB1); then, the array was incubated in 500 times diluted 2 mg · mL-1 SA-AF647 solutions at room temperature in the dark. After incubation for 2.5 h, the slides were rinsed three times with assay buffer (ZeptoMARK CAB1), and the fluorescence intensities were determined by means of the planar waveguide ZeptoREADER F3000 system. Model Reverse TNFr Assay. DDP- and (co)polymer-coated substrates were spotted in duplicate with 30, 10, 3, and 1 nM recombinant human TNFR (diluted in a BSA matrix of 3, 1, 0.3, and 0.1 mg · mL-1, respectively) solutions in basic spotting buffer, and they were let to react for 24 h in a humidity chamber (RH ) 55%). After the slides were blocked with BSA for 30 min (ZeptoFOG instrument), they were washed with Milli-Q water, dried, and placed at 4 °C in the dark for storage. On the same substrates, basic spotting buffer and Milli-Q water were also spotted as controls. The assay was carried out by first equilibrating the slides in buffer and then incubating the chips in a 500 times diluted 0.5 mg · mL-1 antihuman TNFR (purified from rabbit serum) solution. After incubation for 2.5 h at room temperature in the dark, the slides were extensively rinsed with CAB1 buffer; subsequently, the array was incubated for an additional 2 h at room temperature and in the dark in a 500 times diluted 0.2 mg · mL-1 antirabbit Fab-AF647 solution (which specifically binds to the Fc portion of rabbit IgG). Finally, the slides were rinsed three times with buffer (ZeptoMARK CAB1), and the fluorescence intensities of the spots were measured with a ZeptoREADER F3000 instrument. Fluorescence Measurements. Fluorescence intensities were measured using either a confocal fluorescence scanner (GMS 418, Affymetrix) or a ZeptoREADER F3000 instrument (a planar waveguide system developed by Zeptosens).13,40 The acquired fluorescence images were analyzed with the software ZeptoVIEW 3 (Zeptosens). The reported fluorescence intensities represent an average over the whole spot area and are corrected for the surrounding background intensity. Coefficients of variation (CV ) standard deviation/mean) and signal-to-noise ratios were also determined by means of the software ZeptoVIEW 3.

Results and Discussion Brush Synthesis and Characterization. Whereas procedures to graft polymer brushes from silicon oxide surfaces via surfaceinitiated controlled radical polymerization are well-established, no reports have been published to date that describe the synthesis of polymer brushes from Ta2O5, which is the waveguiding coating of the microarray chips studied in this contribution. The strategy that was explored for the synthesis of PGMA and PGMA-co-PDEAEMA (co)polymer brushes on the Ta2O5coated chips is presented in Scheme 1. The procedure outlined in Scheme 1 was previously used to synthesize PGMA and PGMA-co-PDEAEMA brushes from silicon oxide substrates and involves the immobilization of an

3470


Barbey et al.

Scheme 1. Preparation of PGMA and PGMA-co-PDEAEMA Brushes via SI-ATRP from ATRP Initiator-Functionalized Tantalum Pentoxide Substrates

ATRP initiator (6-(chloro(dimethyl)silyl)hexyl 2-bromo-2-methylpropanoate), followed by room temperature SI-ATRP of the appropriate monomer mixture in aqueous methanol.39 A number of experiments was carried out to explore the feasibility of this strategy to graft polymer brushes from Ta2O5. Figure 2 illustrates the evolution of brush thickness with polymerization time for the SI-ATRP of GMA. Because of the transparency of the Ta2O5-coated glass slides, film thicknesses could not be easily assessed by means of ellipsometry but were estimated by analyzing AFM height profiles obtained from micropatterned polymer brushes. Figure 2 compares the growth kinetics of PGMA brushes grafted from Ta2O5 with that of PGMA brushes prepared from silicon oxide substrates that were subjected to SI-ATRP in the same reaction vessel. The similarity in growth profiles and absolute thicknesses indicates that the thickness of PGMA brushes grown from Ta2O5 can be indirectly estimated by analyzing the thickness of brushes that are grown from silicon oxide substrates in the same reaction vessel. The composition of the polymer brushes grown from Ta2O5 substrates was confirmed by XPS analysis. Figure 3 shows the XPS survey and high-resolution C1s and O1s scans of a 20-nmthick PGMA brush prepared from an ATRP-initiator-modified Ta2O5 substrate. As expected based on the chemical composition of PGMA, the survey spectrum reveals only the presence of C1s and O1s signals. In agreement with literature data,39,41 the high-resolution C1s signal can be fitted with the expected peak area ratios using five model Gaussian/Lorentzian curves, which correspond to the aliphatic carbon atoms of the polymer

backbone (C-C/C-H, 285.0 eV), the carbon atoms adjacent to the ester groups (C-CdO, 285.6 eV), the C-O moiety (286.5 eV), the carbon atoms of the oxirane rings (C-O-C, 287.0 eV), and the carbons of the ester groups (O-CdO, 289.1 eV). The O1s signal of PGMA can be fitted with three curves corresponding to the oxygen atoms from the O-CdO (533.9 eV), C-O-C (533.2 eV), and O-CdO (532.3 eV) groups. In a final series of experiments, the stability of the polymer brushes upon exposure to basic spotting buffer, which is used as one of the buffers to prepare the spotting solutions for the protein immobilization experiments and is the chosen buffer for the two model assays (vide infra), was investigated. To this end, PGMA-co-PDEAEMA copolymer brushes, which were grown from patterned silicon and Ta2O5 substrates, were exposed to basic spotting buffer (pH 8.2) for 48 h, and the patterns were observed via optical microscopy. The optical micrographs in Figure 4 do not reveal any wrinkles or other signs of detachment, which is a good indication of the stability of the brushes under the conditions used for the protein microarray assays. Protein Immobilization. Protein immobilization on PGMAand PGMA-co-PDEAEMA-coated Ta2O5 microarray chips was investigated by spotting the substrates with different fluorescently labeled proteins and analyzing the chips with a proprietary fluorescence reader (ZeptoREADER F3000 instrument). Figure 5 shows the results, which were obtained by spotting 50-nmthick PGMA- and PGMA-co-PDEAEMA-coated Ta2O5 substrates with solutions containing 6, 2, 0.6, and 0.2 nM labeled protein (BSA-AF647, OVA-AF647, and Fab-AF647) in a

Figure 2. (A) Evolution of the thickness of PGMA brushes grown from silicon oxide (2) or tantalum pentoxide (9) surfaces as a function of polymerization time. (B) Comparison of the thickness of PGMA brushes grown from silicon substrates with those grafted from Ta2O5 surfaces. Thicknesses were determined from micropatterned brushes using AFM. The dotted line is a guide to the eye representing equivalent thicknesses on both substrates.

Protein Microarrays Based on Polymer Brushes


3471

Figure 3. XPS survey and high-resolution C1s and O1s scans of a 20-nm-thick PGMA brush grown from a Ta2O5 substrate.

Figure 4. Optical micrographs of 130-nm-thick, micropatterned PGMA-co-PDEAEMA copolymer brushes grown from (A,B) silicon wafers or (C,D) Ta2O5-coated glass slides. The two top images (A,C) represent the pristine samples, whereas the bottom images (B,D) are micrographs of the samples after 48 h of exposure to basic (pH 8.2) spotting buffer.

diluting matrix of 3, 1, 0.3, and 0.1 mg · mL-1 of unlabeled BSA (i.e., approximately 45, 15, 4.5, and 1.5 µM), respectively. Spotting solutions were prepared with acidic (pH 6.5), neutral (pH 7.4), or basic (pH 8.2) spotting buffers. As control experiments, protein-free buffers were also spotted on the same slides. One reason to dilute the labeled proteins in a BSA matrix (to a molar ratio of approximately 1:8000 in this case) was to simulate more realistic conditions, i.e., when it is necessary to analyze low-abundant proteins that are present in a complex mixture containing large excess of other proteins (for example, blood serum). After spotting, the substrates were kept in a

humidity chamber for 24 h and then washed with 0.1 wt % Tween 20 for 24 h to remove any physisorbed proteins. The residual fluorescence that can be observed in Figure 5 thus reflects the proteins that are covalently immobilized on the polymer brush coating. Figure 6 is a quantitative representation of the fluorescence intensities observed in Figure 5. Figure 6 only represents those data in Figure 5 that were obtained without the use of dimethyl sulfoxide (DMSO) as additive in the spotting solution. On the basis of these two Figures, a number of observations can be made. First of all, on both investigated surfaces and for all proteins used, spotting resulted in well-

3472


Barbey et al.

Figure 5. Fluorescence images of PGMA (50 nm) and PGMA-co-PDEAEMA (50 nm) brushes grown from Ta2O5 substrates and spotted with fluorescently labeled proteins (BSA-AF647, OVA-AF647, or Fab-AF647) under different conditions (pH 6.5, 7.4, or 8.2; labeled protein concentration ) 6, 2, 0.6, or 0.2 nM; with or without DMSO as additive) and incubated for 24 h. Acquisition settings: planar waveguide readout system with (i) illumination color: red; (ii) exposure time: 1 s; and (iii) display range: 20 000.

Figure 6. Mean fluorescence intensity as a function of the pH of the spotting buffer and protein concentration for PGMA brushes spotted with (A) BSA-AF647, (B) OVA-AF647, and (C) Fab-AF647 and for PGMA-co-PDEAEMA brushes spotted with (D) BSA-AF647, (E) OVA-AF647, and (F) Fab-AF647.

defined, round spots that do not reveal any signs of smearing (Figure 5), which could have been the case if the brush substrate would have been too hydrophilic. Second, independent of the spotting conditions (protein, protein concentration, and pH), the measured fluorescence intensities are at least two times (but up to 10 times) higher on PGMA-co-PDEAEMA-coated substrates as compared to the PGMA-modified chips, which reflects the higher protein binding capacity of the former brush. Third, in the investigated range of concentrations (0.2-6 nM), the fluorescence intensity increases as the concentration of labeled proteins increases, indicating that none of the (co)polymer

brushes reaches saturation. Finally, the PGMA-co-PDEAEMAcoated slides reveal an enhanced and more spread out background fluorescence as compared to the PGMA-modified microarray chips. This is attributed to and a drawback of using Tween 20 to wash away physisorbed proteins, which spreads the labeled proteins all over the surface and may be responsible for the global increase in background fluorescence intensity as observed in Figure 5. In the following sections, the effects of the type and concentration of protein, the pH of the spotting buffer, the use of additives, as well as the influence of brush



3473

Figure 7. Mean fluorescence intensity of (A) PGMA and (B) PGMA-co-PDEAEMA brushes as a function of the protein (labeled protein concentration kept constant at 2 nM) for different pH values.

Figure 8. (A,B) Area-averaged mass of proteins immobilized on a 70-nm-thick PGMA-co-PDEAEMA brush upon exposure to 50 mg · mL-1 solutions of BSA and OVA (in basic spotting buffer) for 6, 24, and 48 h as determined from the resonance frequency shifts measured by QCM-D (average over the third, fifth, and seventh harmonic).

composition and thickness on the amount of protein that can be immobilized and the kinetics of the immobilization process will be discussed. Effect of Spotted Proteins. As a representative example, Figure 7 shows the fluorescence intensities measured on the two brushes after spotting with a 2 nM solution of different proteins in the presence of 1 mg · mL-1 unlabeled BSA in the three different buffers. Data for other concentrations are included in the Supporting Information (Figure S1-S3). Figure 7 indicates that in each buffer the measured fluorescence intensities are higher on the copolymer brush as compared to the PGMA homopolymer brush, which reflects the higher protein binding capacity of the former. Furthermore, for any given buffer, the measured fluorescence intensities decrease from OVA-AF647 to BSA-AF647 to Fab-AF647. Whereas the specific reasons for this observation are difficult to pinpoint, they are likely to be due to differences in protein size (OVA: ∼43 kDa; Fab: ∼50 kDa; BSA: ∼66 kDa) in combination with variations in the number of surface-accessible nucleophilic groups, the number of fluorophores per protein, or both. Because the differences in fluorescence intensities in Figure 7 do not necessarily reflect differences in protein binding capacity but, as mentioned above, may also be influenced by, for example, the number of fluorophores and surface-accessible groups per protein, a subsequent series of QCM-D experiments was carried out to quantify the protein binding capacity of the PGMA-co-PDEAEMA brushes. To this end, a 70-nm-thick

PGMA-co-PDEAEMA brush was incubated in 50 mg · mL-1 solutions of BSA and OVA in basic spotting buffer. Protein binding was assessed by comparing the resonance frequencies before and after exposure to the proteins. (See also Table S1 in the Supporting Information.) The results of these experiments are summarized in Figure 8 and clearly illustrate that the protein binding capacity is protein-dependent. Except for short incubation times (6 h), the amounts of OVA that can be immobilized are larger than BSA. After 48 h, ∼6 µg · cm-2 OVA can be bound as compared to ∼3.5 µg · cm-2 BSA. As also noticed for Figure 7, these differences are most likely attributed to differences in protein sizes (OVA: ∼43 kDa; BSA: ∼66 kDa) or variations in the number of nucleophilic groups that are present at the surface of proteins. Effect of pH. As illustrated in Figure 9 for labeled proteins spotted at a concentration of 6 nM, the pH of the spotting buffer influences the amount of protein that is immobilized. (For other concentrations, see Figure S4-S6 in the Supporting Information). The results reveal that in general both basic and acidic conditions enhance the amount of protein that can be immobilized on the substrates compared with neutral pH. This can be explained by the higher reactivity of oxirane rings under both acidic and basic conditions.42 In many cases, spotting proteins from basic (pH 8.2) spotting buffer resulted in the highest fluorescence intensities. Effect of Adding DMSO. As illustrated in Figure 5, spotting experiments were carried out not only in pure buffers but also

3474


Barbey et al.

Figure 9. Mean fluorescence intensity of (A) PGMA and (B) PGMA-co-PDEAEMA brushes as a function of pH for different labeled proteins (concentration kept constant at 6 nM).

Figure 10. Effect of the addition of 10 vol % DMSO to the spotting solution on the immobilization capacities of (A) PGMA or (B) PGMA-coPDEAEMA brushes at different pH values (for OVA-AF647 at a concentration of 2 nM).

using 10 vol % DMSO as additive. The rationale for exploring DMSO as additive was due to its capacity to avoid the spots to dry out too quickly (as glycerol or sugar do in other spotting buffers) as well as for its excellent solvating power (DMSO is miscible in water and a good solvent for PGMA), which might facilitate proteins to reach the whole sampling volume of the brush.43-45 As an example, Figure 10 plots the mean fluorescence intensity measured on (co)polymer-brush-modified substrates when spotted with 2 nM solutions of OVA-AF647 (in a matrix of 1 mg · mL-1 of BSA) with and without 10 vol % DMSO as additive. The data in Figure 10 indicate that the addition of DMSO is often beneficial and can enhance the protein binding capacity of the brushes but only to a relatively limited extend. At low concentrations of protein in the spotting solution, however, the improvements often are only relatively small. (See Figures S7-S17 of the Supporting Information.) On the basis of these results, the conditions that gave the highest fluorescence intensities (i.e., OVA-AF647 and BSAAF647, basic spotting buffer (pH 8.2) and without using 10 vol % DMSO as additive) were chosen for the experiments described in the remainder of this manuscript. Effect of Brush Composition and Thickness. The experiments presented in Figure 5 were useful and important to evaluate the effects of the type and concentration of protein and the pH of the spotting buffer on the protein immobilization. The results that have been discussed so far, however, have provided only limited insight into the effects of the thickness and composition of the brush coating. To evaluate further the influence of brush thickness and composition on the protein

binding capacity of the modified microarray chips, we analyzed substrates that were exposed to solutions of fluorescently labeled proteins using a confocal fluorescence scanner. In contrast to the system that was used in Figure 5, a conventional confocal laser scanner is not limited to the ∼100 nm penetration depth of the evanescent field and therefore also allows analysis of thicker brush coatings. For these experiments, Ta2O5-modified microarray chips covered with DDP (deposited as a hydrophobic self-assembled monolayer) or a PGMA (50 nm) or PGMA-coPDEAEMA (50 and 130 nm) brush coating were incubated in 2 nM OVA-AF647 and BSA-AF647 solutions (diluted in a matrix of 1 mg · mL-1 BSA) in basic spotting buffer for 24 h. In addition, control experiments were performed in which the substrates were exposed to 1 mg · mL-1 BSA (in basic spotting buffer) as well as pure basic spotting buffer only. After 24 h, the substrates were extensively rinsed with excess basic spotting buffer and subsequently analyzed with a confocal fluorescence scanner. The image in Figure 11A and the corresponding fluorescence intensities, which are plotted in Figure 11B, reveal a number of interesting characteristics of the PGMA and PGMA-co-PDEAEMA brushes with regards to the feasibility of these coatings for protein microarray applications. First of all, the control experiments that were carried out with 1 mg · mL-1 BSA or pure basic spotting buffer reveal a low intrinsic fluorescence of the polymer brush coatings, which is an important prerequisite for the development of practically useful protein microarray surface coatings (Figure 11A). Figure 11B is a quantitative representation of the fluorescence intensities that were measured after exposing the DDP- and (co)polymer-



3475

Figure 11. (A) Fluorescence images of DDP-, PGMA- (50 nm), and PGMA-co-PDEAEMA- (50 and 130 nm) coated microarray chips after 24 h of incubation in different labeled proteins. BSA and basic spotting buffer were used as controls. Acquisition settings: Confocal fluorescence scanner with (i) illumination color: red; (ii) laser power: 100%; (iii) gain: 90%; and (iv) display range: 65535. (B) Effect of brush composition and thickness on the number of immobilized proteins after 24 h of incubation time at a pH of 8.2 and a labeled protein concentration of 2 nM. Mean fluorescence intensities were taken from the micrograph in Figure 11A.

Figure 12. (A,B) Area-averaged mass of bound protein on PGMA (∼25 and ∼100 nm) and PGMA-co-PDEAEMA (∼15 and ∼70 nm) brushes after incubation in 50 mg · mL-1 OVA solutions in basic spotting buffer for 6, 24, and 48 h. The area-averaged mass was determined by QCM-D from the average of ∆fn/n for n ) 3, 5, and 7 using the Sauerbrey equation.

brush-modified substrates to the different protein and control solutions. The results clearly indicate the enhanced protein binding capacity of the 3D (co)polymer-brush-modified substrates as compared to the 2D, hydrophobic DDP-coated chip. Furthermore, comparison of the results for PGMA and PGMAco-PDEAEMA brushes of similar thickness reveals that the incorporation of DEAEMA comonomer units enhances the protein binding capacity, which confirms the results summarized in Figure 5. Finally, the data in Figure 11 demonstrate that the protein binding capacity of the copolymer-brush-based coatings depends on and increases with increasing brush thickness. This is illustrated by comparison of the data obtained with 50- and 130-nm-thick PGMA-co-PDEAEMA brush coatings and reflects the 3D nature of the polymer brush coatings. The effects of film thickness and composition on the protein binding capacity of the (co)polymer-brush-modified chips were further quantitatively investigated by QCM-D experiments. Figure 12 summarizes the results that were obtained upon exposing QCM-D sensors modified with PGMA-co-PDEAEMA brushes of two thicknesses to 50 mg · mL-1 OVA solutions in basic spotting buffer for different periods of time. For comparison, Figure 12 also includes the results of QCM-D experiments that were obtained under identical conditions using two PGMA homopolymer brushes with thicknesses of 25 and 100 nm.39 Additional details are provided in Tables S1 and S2 in the Supporting Information. Figure 12 clearly illustrates that the protein-binding capacity increases with increasing brushes

thickness. Whereas the thinnest (co)polymer brushes are able to immobilize up to ∼1.5 µg · cm-2 after 48 h of incubation, the thickest brushes reach values that are three to four times higher. Figure 12 also reconfirms the enhanced protein binding capacity of the copolymer brushes as compared to the PGMA homopolymer brushes. For example, after exposure for 48 h to an OVA solution, ∼6 µg · cm-2 OVA was bound on a 70-nmthick PGMA-co-PDEAEMA versus ∼5 µg · cm-2 on a 100-nmthick PGMA brush, despite the slightly higher film thickness for the later substrate. Figure 12B plots the evolution of the area-averaged mass of surface-immobilized OVA as a function of incubation time for the different substrates. Whereas the amount of bound protein increases more-or-less linearly with time for the thickest (co)polymer brushes, the area-averaged mass measured on the other substrates seems to level off with increasing incubation time, which could suggest a saturation of these brush coatings. Protein Microarray Experiments. After investigating the immobilization of proteins on PGMA and PGMA-co-PDEAEMA brushes and elaborating protocols to optimize the protein binding capacity of these polymer brush coatings, two proofof-concept microarray studies were carried out to evaluate the feasibility of the (co)polymer-brush-coated microarray chips to detect different protein-protein interactions (Scheme 2). In a first series of experiments, the ability of the different (co)polymer brush coatings to investigate binding of labeled streptavidin to surface-immobilized BSA-biotin was evaluated.

3476


Barbey et al.

Scheme 2. Schematic Illustration of the Two Proof-of-Concept Microarray Experiments Carried out in This Studya

a

(A) Biotin-streptavidin assay (one-step recognition) and (B) model TNFR reverse assay (two-step recognition).

Figure 13. (A) Fluorescence images obtained after incubation in SA-AF647 of different coated microarray substrates that were previously spotted with a concentration series of BSA-biotin (30, 10, 3, and 1 nM in a BSA diluting matrix of 3, 1, 0.3, and 0.1 mg · mL-1, respectively). Basic spotting buffer and water were spotted as controls. Acquisition settings were: (i) illumination color: red; (ii) exposure time: 1 s; (iii) gray filter: 0.5; (iv) display range: 30 000. (B) Mean fluorescence intensities determined from the micrograph in Figure 13A.

To this end, DDP-, PGMA- (50 nm), and PGMA-co-PDEAEMA- (50 nm) coated microarray slides were spotted with different concentrations of BSA-biotin, incubated for 24 h, blocked with unlabeled BSA to quench any unreacted epoxide groups, and subsequently incubated in a solution of SA-AF647. (See the Experimental Section for more details.) The results of these experiments are summarized in Figures 13 and 14. It is important to note here that the measured fluorescence intensities reflect binding of labeled streptavidin by surface-immobilized biotin. This is different from the data shown in Figures 5 and 11, where the measured fluorescence is due to the covalently immobilized protein. Qualitatively, the results in Figure 13 demonstrate that labeled streptavidin molecules are able to access the polymer brush and bind to immobilized BSA-biotin. Moreover, the absence of fluorescence from the control spots (i.e., pure basic spotting buffer and water) illustrates the efficiency of the BSA blocking step and demonstrates that the fluorescence that is observed on the polymer-brush-modified substrates is due to the specific binding of the labeled streptavidin to the surface-immobilized BSA-biotin. As shown in Figure 13B, apart from the lowest BSA-biotin concentrations, the fluorescence intensities measured on the PGMA-co-PDEAEMA brushes were always higher than those measured on

the DDP-coated and PGMA-brush-modified microarray chips. The enhanced fluorescence intensities measured on the PGMAco-PDEAEMA brush as compared to the DDP-covered microarray chip illustrate the increased binding capacity that is obtained by presenting a 3D sampling volume instead of a 2D planar active layer. For all investigated BSA-biotin concentrations, the fluorescence intensities measured on the PGMA-coPDEAEMA brushes were higher than those of the PGMA homopolymer brush, which is in agreement with the higher binding capacity of the former (vide supra). Interestingly, not only are the measured absolute fluorescence intensities on the polymer-brush-coated microarray chips higher compared with the DDP-modified surfaces but also the use of the polymer brush coating results in a significant improvement of the signal-tonoise (S/N) ratio (Figure 14B) and coefficient of variation (CV, Figure S18A in the Supporting Information). As an example, the S/N ratio on PGMA-co-PDEAEMA was up to four times better as compared to a DDP-coated chip, which clearly proves the increased sensitivity that can be obtained with the polymerbrush-based microarray coatings. Furthermore, the CV values are about 15% for the copolymer brushes, 20% for the PGMA brushes, and 40% for the DDP substrates, which indicates that better quality (less-dispersed) results are obtained using the



3477

Figure 14. (A) Mean fluorescence intensities and (B) signal-to-noise ratios determined from substrates that were spotted with a concentration series of BSA-biotin protein after incubation in SA-AF647. Acquisition settings were: (i) illumination color: red; (ii) exposure time: 1 s; (iii) gray filter: 0.5.

Figure 15. (A) Fluorescence images of microarray chips modified with a DDP, PGMA, or PGMA-co-PDEAEMA coating after spotting with recombinant human TNFR and a subsequent two-step assay process with antihuman TNFR and Fab-AF647. Basic spotting buffer and water were spotted as controls. Acquisition settings were: (i) illumination color: red; (ii) exposure time: 10 s; (iii) gray filter: 1.0; and (iv) display range: 20 000. (B) Mean fluorescence intensities as determined from these substrates.

(co)polymer brushes and further underlines the potential of the polymer-brush-modified substrates as an alternative for the commercially available DDP chips. Figure 14A shows that the mean fluorescence intensity levels off for the highest concentration (30 nM BSA-biotin in 3 mg · mL-1 BSA). For the PGMA and PGMA-co-PDEAEMA brushes, the intensity of the measured fluorescence increases linearly up to 10 nM BSA-biotin (in a diluting matrix of 1 mg · mL-1 BSA); then, the signal levels off. This could indicate saturation of the polymer brushes because of the high BSA concentration of the diluting matrix (3 mg · mL-1) (i.e., not enough free reactive sites to bind BSAbiotin), or this could be due to mass transfer/diffusion (of SAAF647) issues within the polymer layer. In addition to the biotin-streptavidin binding assay, a second, more challenging microarray experiment was carried out, which involved spotting and immobilization of recombinant human TNFR, followed by a two-step labeling process by successive incubation in solutions containing antihuman TNFR and FabAF647. (See the Experimental Section for more details.) The results of these experiments are shown in Figures 15 and 16. As illustrated in Figure 15B, for all investigated human recombinant TNFR concentrations, the fluorescence intensities measured on the PGMA-co-PDEAEMA brushes are higher than those on the DDP reference, which reflects the higher protein binding capacity of the 3D brush coating versus the 2D DDP substrate. As for the biotin-streptavidin assay, the fluorescence intensities on the PGMA-coated microarray chips are lower than those on the PGMA-co-PDEAEMA substrate, which illustrates

the enhancement of the protein binding capacity due to the incorporation of the DEAEMA comonomer units. Figure 16A shows that the fluorescence intensities increase linearly with the concentration of human TNFR up to a concentration of 10 nM (in 1 mg · mL-1 BSA), which is similar to what was observed for the biotin-streptavidin assay (Figure 14A). The deviation from linearity at higher TNFR concentrations could reflect saturation of the (co)polymer brushes due to the high concentration of BSA in the spotting solution or could be the result of mass transfer/diffusion limitations. The assay conditions that were used here, however, are the standard conditions developed for the DDP-coated chips and may be further optimized specifically for the polymer-brush-modified substrates. It is important to note that the assay, the results of which are summarized in Figures 15 and 16, is a two-step process that requires subsequent binding of antihuman TNFR and Fab-AF647 with surface-immobilized recombinant human TNFR and, respectively, recombinant human TNFR-antihuman TNFR dimer. The fluorescence observed in the micrograph in Figure 15A qualitatively demonstrates that antihuman TNFR and Fab-AF647 are able to penetrate the protein-loaded brush and access their binding partners. Figure 15A, especially for the PGMA-coPDEAEMA-modified chips, however, also reveals background fluorescence, which is particularly obvious from the control spots (pure basic spotting buffer and water). The background fluorescence could be due to: (i) an inefficient BSA blocking step, which leaves reactive sites in the brushes that can react with Fab-AF647 or antihuman TNFR, (ii) inefficient washing after

3478


Barbey et al.

Figure 16. (A) Mean fluorescence intensities and (B) signal-to-noise ratios determined from DDP, PGMA, and PGMA-co-PDEAEMA coated substrates as a function of the concentration of TNFR in the spotting solution.

antihuman TNFR incubation, which leaves physisorbed antihuman TNFR that is subsequently labeled with Fab-AF647, or (iii) a combination of these effects. Despite the increased background fluorescence, however, S/N ratios (Figure 16B) and coefficients of variation (Figure S18B, Supporting Information) determined for the PGMA-co-PDEAEMA modified chips are still improved compared with the DDP-coated reference chips (and the PGMA-brush-covered substrate), which indicates that application of this copolymer brush coating not only enhances the protein binding capacity but also results in an increase in sensitivity.

Conclusions This article has investigated the feasibility of PGMA and PGMA-co-PDEAEMA (co)polymer brushes to act as the protein-binding coating in a commercial protein microarray device. The (co)polymer brush coatings were prepared via surface-initiated atom transfer radical (co)polymerization of GMA and DEAEMA from tantalum-pentoxide-modified microarray chips. Protein immobilization studies, which were carried out using a commercial microarray spotter, revealed that the 3D (co)polymer-brush-modified chips had a higher protein binding capacity as compared to the 2D, commercially available DDP-covered reference chip. The protein binding capacity of the (co)polymer-brush-modified chips was found to depend not only on the nature of the spotted protein, the protein concentration, and the pH of the spotting solution but also on the composition and thickness of the brush coating. Specifically, copolymerization of DEAEMA and increasing brush thickness resulted in enhanced protein immobilization. Finally, the performance of the polymer-brush-based microarray chips was compared to that of a commercially available DDP chip. To this end, two model, proof-of-concept microarray studies were carried out, which involved the detection of biotin-streptavidin binding as well as a model TNFR reverse assay. These proofof-concept assay experiments, accomplished under real protein microarray conditions (i.e., via standard industrial protocols, using an automated nanospotter and a fluorescence readout system), were of particular significance for the evaluation of (co)polymer brushes as substrates for microarray applications. For both model experiments, the copolymer-brush-modified chips not only revealed the highest fluorescence intensities but also significantly enhanced signal-to-noise ratios compared with the DDP reference, which indicates that the copolymer brush coatings might represent an attractive alternative to commercially available protein microarray substrates. Considering the fact that the results presented in this contribution were

obtained using instrumentation and protocols that were optimized for 2D, DDP-modified microarray chips, there is still much room to improve further the performance of the polymer brush-based microarray chips presented here. Acknowledgment. This research was supported by CTI/KTI. Supporting Information Available. Effects on the protein binding capacity of the type of protein, protein concentration, pH of the spotting buffer, and of using DMSO as additive; QCM-D data used to determine the area-averaged mass of proteins bound to the (co)polymer brushes; and coefficients of variation values for the two protein microarray assays. This material is available free of charge via the Internet at http:// pubs.acs.org.

References and Notes (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22)

Mitchell, P. Nat. Biotechnol. 2002, 20, 225. MacBeath, G. Nat. Genet. 2002, 32, 526. Xu, Q.; Lam, K. S. J. Biomed. Biotechnol. 2003, 2003, 257. Wilson, D. S.; Nock, S. Angew. Chem., Int. Ed. 2003, 42, 494. Templin, M. F.; Stoll, D.; Schwenk, J. M.; Po¨tz, O.; Kramer, S.; Joos, T. O. Proteomics 2003, 3, 2155. LaBaer, J.; Ramachandran, N. Curr. Opin. Chem. Biol. 2005, 9, 14. Hall, D. A.; Ptacek, J.; Snyder, M. Mech. Ageing DeV. 2007, 128, 161. Liotta, L. A.; Espina, V.; Mehta, A. I.; Calvert, V.; Rosenblatt, K.; Geho, D.; Munson, P. J.; Young, L.; Wulfkuhle, J.; Petricoin, E. F., III. Cancer Cell 2003, 3, 317. Angenendt, P. Drug DiscoVery Today 2005, 10, 503. Kusnezow, W.; Jacob, A.; Walijew, A.; Diehl, F.; Hoheisel, J. D. Proteomics 2003, 3, 254. Angenendt, P.; Glo¨kler, J.; Murphy, D.; Lehrach, H.; Cahill, D. J. Anal. Biochem. 2002, 309, 253. Angenendt, P.; Glo¨kler, J.; Sobek, J.; Lehrach, H.; Cahill, D. J. J. Chromatogr., A 2003, 1009, 97. Pawlak, M.; Schick, E.; Bopp, M. A.; Schneider, M. J.; Oroszlan, P.; Ehrat, M. Proteomics 2002, 2, 383. Martin, K.; Steinberg, T. H.; Cooley, L. A.; Gee, K. R.; Beechem, J. M.; Patton, W. F. Proteomics 2003, 3, 1244. MacBeath, G.; Schreiber, S. L. Science 2000, 289, 1760. Haab, B. B.; Dunham, M. J.; Brown, P. O. Genome Biol. 2001, 2, 0004.1. Lueking, A.; Horn, M.; Eickhoff, H.; Bu¨ssow, K.; Lehrach, H.; Walter, G. Anal. Biochem. 1999, 270, 103. Stillman, B. A.; Tonkinson, J. L. BioTechniques 2000, 29, 630. Arenkov, P.; Kukhtin, A.; Gemmell, A.; Voloshchuk, S.; Chupeeva, V.; Mirzabekov, A. Anal. Biochem. 2000, 278, 123. Charles, P. T.; Goldman, E. R.; Rangasammy, J. G.; Schauer, C. L.; Chen, M.-S.; Taitt, C. R. Biosens. Bioelectron. 2004, 20, 753. Afanassiev, V.; Hanemann, V.; Wo¨lfl, S. Nucleic Acids Res. 2000, 28, E66. Barbey, R.; Lavanant, L.; Paripovic, D.; Schu¨wer, N.; Sugnaux, C.; Tugulu, S.; Klok, H.-A. Chem. ReV. 2009, 109, 5437.

Protein Microarrays Based on Polymer Brushes (23) Senaratne, W.; Andruzzi, L.; Ober, C. K. Biomacromolecules 2005, 6, 2427. (24) Rusmini, F.; Zhong, Z.; Feijen, J. Biomacromolecules 2007, 8, 1775. (25) Kawai, T.; Saito, K.; Lee, W. J. Chromatogr., B 2003, 790, 131. (26) Bruening, M. L.; Dotzauer, D. M.; Jain, P.; Ouyang, L.; Baker, G. L. Langmuir 2008, 24, 7663. (27) Dai, J.; Bao, Z.; Sun, L.; Hong, S. U.; Baker, G. L.; Bruening, M. L. Langmuir 2006, 22, 4274. (28) Zhang, Z.; Chen, S. F.; Jiang, S. Y. Biomacromolecules 2006, 7, 3311. (29) Xu, F. J.; Cai, Q. J.; Li, Y. L.; Kang, E. T.; Neoh, K. G. Biomacromolecules 2005, 6, 1012. (30) Kusumo, A.; Bombalski, L.; Lin, Q.; Matyjaszewski, K.; Schneider, J. W.; Tilton, R. D. Langmuir 2007, 23, 4448. (31) Dong, R.; Krishnan, S.; Baird, B. A.; Lindau, M.; Ober, C. K. Biomacromolecules 2007, 8, 3082. (32) Jain, P.; Sun, L.; Dai, J. H.; Baker, G. L.; Bruening, M. L. Biomacromolecules 2007, 8, 3102. (33) Huang, J. S.; Li, X. T.; Zheng, Y. H.; Zhang, Y.; Zhao, R. Y.; Gao, X. C.; Yan, H. S. Macromol. Biosci. 2008, 8, 508. (34) Alarcoń, C. D. H.; Farhan, T.; Osborne, V. L.; Huck, W. T. S.; Alexander, C. J. Mater. Chem. 2005, 15, 2089.


3479

(35) Dunn, J. D.; Igrisan, E. A.; Palumbo, A. M.; Reid, G. E.; Bruening, M. L. Anal. Chem. 2008, 80, 5727. (36) Singh, N.; Wang, J.; Ulbricht, M.; Wickramasinghe, S. R.; Husson, S. M. J. Membr. Sci. 2008, 309, 64. (37) Xu, F. J.; Li, H. Z.; Li, J.; Eric Teo, Y. H.; Zhu, C. X.; Kang, E. T.; Neoh, K. G. Biosens. Bioelectron. 2008, 24, 773. (38) Jain, P.; Dai, J. H.; Baker, G. L.; Bruening, M. L. Macromolecules 2008, 41, 8413. (39) Barbey, R.; Klok, H.-A. Langmuir, DOI: 10.1021/la102400z. (40) Duveneck, G. L.; Abel, A. P.; Bopp, M. A.; Kresbach, G. M.; Ehrat, M. Anal. Chim. Acta 2002, 469, 49. (41) Beamson, G.; Briggs, D. High Resolution XPS of Organic Polymers: The Scienta ESCA300 Database; John Wiley & Sons: Chichester, U.K., 1992. (42) Parker, R. E.; Isaacs, N. S. Chem. ReV. 1959, 59, 737. (43) McQuain, M. K.; Seale, K.; Peek, J.; Levy, S.; Haselton, F. R. Anal. Biochem. 2003, 320, 281. (44) Preininger, C.; Sauer, U.; Dayteg, J.; Pichler, R. Bioelectrochemistry 2005, 67, 155. (45) Mace, C. R.; Yadav, A. R.; Miller, B. L. Langmuir 2008, 24, 12754.

BM101297W

Protein Microarrays Based on Polymer Brushes Prepared via Surface

Recommend Documents