Functional Fluorinated Modifications on a Polyelectrolyte Coated

Aug 27, 2010 - Contact angle measurements showed that this functionalized surface was as hydrophobic as the native PDMS with a virtually constant cont...
0 downloads 0 Views 3MB Size
Download PDF
Anal. Chem. 2010, 82, 7804–7813

Functional Fluorinated Modifications on a Polyelectrolyte Coated Polydimethylsiloxane Substrate for Fabricating Antibody Microarrays Huang-Han Chen, Wang-Chou Sung, Shih-Shin Liang, and Shu-Hui Chen* Department of Chemistry, National Cheng Kung University, Tainan 701, Taiwan Fluorinated compounds exhibit hydrophobic, nonstick, and self-cleaning properties, making them attractive for use as the coating material for biochips. In this study, we copolymerized the fluorinated compound 1H,1H,2Hperfluoro-1-decene (FD) with acrylic acid (AA) and bonded the resulting copolymer with protein G on the surface of polyelectrolyte coated polydimethylsiloxane (PDMS) to form a functional surface that captures antibodies. We demonstrated that the modified PDMS surface remained hydrophobic while becoming resistant to nonspecific protein binding. Thus, aqueous sample solutions formed the droplets (4 µL) on the surface without spreading and drying during the sample printing. Contact angle measurements showed that this functionalized surface was as hydrophobic as the native PDMS with a virtually constant contact angle over seven days of the study under dried condition at 4 °C. Spectroscopic measurements revealed that FD/AA copolymerization formed a homogeneous and highly dense multilayer (50 mg/mm2) with a fluorine coverage of 35.4%. Moreover, protein G was shown to covalently bind to AA molecules on the surface at a binding density of 0.24 µg/mm2. We demonstrated that the fluorinated coating withstood nonspecific binding with extremely low background emission, leading to bioassays that, without the need of blocking agents, exhibited six times more sensitivity than PEG coatings. The fluorinated PDMS antibody microarrays were further applied to accurately determine the absolute concentration of ERr in MCF-7 cells. In conclusion, the unique properties of fluorinated compounds, such as withstanding wetting, nonspecific binding and contamination, make them an excellent coating material for use in sensitive and simple on-chip assays. Array-based technologies are excellent analytical platforms for a broad range of applications including clinical molecular diagnostics, environmental microbial monitoring, new drug screening, and mechanistic studies for proteomics.1,2 Various substrates, including hard materials like glass and soft materials like nitrocel* To whom correspondence should be addressed. E-mail: shchen@ mail.ncku.edu.tw. (1) Zhu, H.; Snyder, M. Curr. Opin. Chem. Biol. 2003, 7, 55–63. (2) Joos, T. O.; Schrenk, M.; Hopfl, P.; Kroger, K.; Chowdhury, U.; Stoll, D.; Schorner, D.; Durr, M.; Herick, K.; Rupp, S.; Sohn, K.; Hammerle, H. Electrophoresis 2000, 21, 2641–2650.

7804

Analytical Chemistry, Vol. 82, No. 18, September 15, 2010

lulose, have been used for the fabrication of protein microarrays for many years. However, immunoblot and most other soft materials traditionally used in biochemical analyses are not compatible with protein microarrays for reasons including low protein binding density, spread of the spotted material, and low signal-to-noise ratio.1,3-6 Most people have turned to the use of glass slides coated with polar functional groups, such as polylysine, to increase the protein binding density.1 However, since small sample volumes (nanoliter) are applied to the hydrophilic plain glass surface, a wet environment is required during the printing process, and a high percentage of glycerol is needed in the sample buffer to keep proteins in their active wet forms. To reduce evaporation and minimize cross-contamination, nanowells on bare polydimethylsiloxane (PDMS) have been fabricated to analyze yeast protein kinases for their substrate specificities. In addition to being more amenable than hard substrate materials to creating nanowells by simple molding techniques,7 PDMS possesses many other advantages, such as its low cost, disposability, high optical transparency, biocompatibility, and chemical stability, which make it an attractive material.8 However, bare PDMS is hydrophobic and, thus, susceptible to high nonspecific protein binding, which greatly lowers its detection sensitivity.9 Many research groups have demonstrated various modification methods including polyelectrolyte multilayers (PEMS),10 silanization,11 radiation-induced graft polymerization,12,13 chemical vapor deposition,14 and phospholipid bilayer modification,15-17 to make (3) Haab, B. B. Curr. Opin. Drug Discovery Dev. 2001, 4, 116–123. (4) Templin, M. F.; Stoll, D.; Schrenk, M.; Traub, P. C.; Vohringer, C. F.; Joos, T. O. Trends Biotechnol. 2002, 20, 160–166. (5) Stoll, D.; Templin, M. F.; Schrenk, M.; Traub, P. C.; Vohringer, C. F.; Joos, T. O. Front. Biosci. 2002, 7, c13–c32. (6) Lueking, A.; Horn, M.; Eickhoff, H.; Bussow, K.; Lehrach, H.; Walter, G. Anal. Biochem. 1999, 270, 103–111. (7) Zhu, H.; Klemic, J. F.; Chang, S.; Bertone, P.; Casamayor, A.; Klemic, K. G.; Smith, D.; Gerstein, M.; Reed, M. A.; Snyder, M. Nat. Genet. 2000, 26, 283–289. (8) Lottersy, J. C.; Olthuis, W.; Veltink, P. H.; Bergveld, P. J. Micromech. Microeng. 1997, 7, 145–147. (9) Linder, V.; Verpoorte, E.; Thormann, W.; de Rooij, N. F.; Sigrist, H. Anal. Chem. 2001, 73, 4161–4169. (10) Liu, Y.; Fanguy, J. C.; Bledsoe, J. M.; Henry, C. S. Anal. Chem. 2000, 72, 5939–5944. (11) Papra, A.; Bernard, A.; Juncker, D.; Larsen, N. B.; Michel, B.; Delamarche, E. Langmuir 2001, 17, 4090–4095. (12) Hu, S.; Ren, X.; Backman, M.; Sims, C. E.; Li, G. P.; Allbritton, N. Anal. Chem. 2002, 74, 4117–4123. (13) Hu, S.; Ren, X.; Backman, M.; Sims, C. E.; Li, G. P.; Allbritton, N. Anal. Chem. 2004, 74, 1865–1870. (14) Lahann, J.; Balcelis, M.; Hang, L.; Rodon, T.; Jensen, K. F.; Langer, T. Anal. Chem. 2003, 75, 2117–2122. 10.1021/ac101799f  2010 American Chemical Society Published on Web 08/27/2010

PDMS surfaces hydrophilic so that they can withstand nonspecific binding. We have also demonstrated the ability of three-protein layers18 and hydrogel plugs19 functionalized with protein G to carry out enzyme-linked immunosorbent assays (ELISA), and the data indicates high detection sensitivity and long-term stability. Hydrophilic modifications with a low contact angle (about 20°), however, could lead to the spread of spotted materials on the plain surface or surfaces with shallow confinement, which would prohibit the use of the modified PDMS materials as substrates for use in high-density protein chips. Though hard to get wet, surfaces modified with fluorinated molecules can be cleaned up effortlessly. They are commonly used to reduce surface tension to achieve anti-stick performance. Fluorinated organosilanes applied by vapor deposition have recently demonstrated superior performance (compared to traditional bovine serum albumin (BSA), nonfat milk, and horse serum) as blocking reagents in reducing background noises, leading to enhanced fluorescence detection.20 The current study explores a novel covalent modification method in which fluorinated compounds are photocopolymerized with acrylic acid (AA) and protein G on PDMS substrate. The usefulness of this surface modification in fabricating antibody chips is also evaluated. EXPERIMENTAL SECTION Materials and Chemicals. The Sylgard 184 kit, containing PDMS oligomer and curing agent, was acquired from Dow Corning (Midland, MI). Hydrolyzed poly(styrene-alt-maleic anhydride) (h-PSMA) (MW 350 kDa), poly(ethyleneimine) (PEI) (MW 750 kDa), poly(acrylic acid) (PAA) (MW 100 kDa), 1-[3(dimethylamino)propyl]-3-ethyl-carbodiimide hydrochloride (EDC), N-hydroxy-succinimide (NHS), tetramethylbenzidine (TMB), Tween 20, 1H,1H,2H-perfluoro-1-decene (FD), 17β-estradiol (E2), FITClabeled BSA (FITC-BSA), sodium bicarbonate(NaHCO3),4-(2Hydroxyethyl)-piperazine-1-ethanesulfonic acid (HEPES), potassium chloride (KCl), ethylenedinitrilotetraacetic acid (EDTA), ethylene glycol-bis(2-aminoethylether)-N,N,N′,N′-tetraacetic acid (EGTA), and albumin from bovine serum (BSA) were obtained from Sigma (St. Louis, MO). 1,4-Dithiothreitol (DTT) were obtained from J. T. Baker (Canada). Acrylate-poly(ethylene glycol)-N-hydro-xysuccinimide (ACRL-PEG-NHS) (MW 5000) was obtained from NEKTAR (San Carlos, CA) and Laysan Bio (Arab, AL). The photoinitiator 2,2-Dimethoxy-2-phenylacetophenone (DPA) was obtained from Fluka (Buchs, Switzerland). Phosphate-buffered saline (PBS) was obtained from Pierce (Rockford, IL). Recombinant protein G was obtained from Invitrogen (Camarillo, CA). The rabbit anti-estrogen receptor R (anti-ERR), antimouse IgG-HRP, and mouse anti-ERR were obtained from Santa Cruz (Santa Cruz, CA, USA). The human recombinant ERR was obtained from Invitrogen (Carisbad, CA). (15) Yang, T.; Jung, S.; Hang, L.; Mao, H.; Cremer, P. S. Anal. Chem. 2001, 73, 165–169. (16) Yang, T.; Baryshnikova, O. K.; Mao, H.; Holden, M. A.; Cremer, P. S. J. Am. Chem. Soc. 2003, 125, 4779–4784. (17) Mao, H.; Yang, T.; Cremer, P. S. Anal. Chem. 2002, 74, 379–385. (18) Sung, W. C.; Chang, C. C.; Makamba, H.; Chen, S. H. Anal. Chem. 2008, 80, 1529–1535. (19) Sung, W. C.; Chen, H. H.; Makamba, H.; Chen, S. H. Anal. Chem. 2009, 81, 7967–7973. (20) Hsieh, H. Y.; Wang, P. C.; Wu, C. L.; Huang, C. W.; Chieng, C. C.; Tseng, F. G. Anal. Chem. 2009, 81, 7908–7916.

Acrylic acid (AA) was obtained from Fluka (Buchs, St. Gallen, Switzerland). Enhanced Chemiluminescent Luminol ReagentKit (ECL) was obtained from PerkinElmer Life Sciences (Boston, MA) for detection and the emission was captured by a digital imaging system (UVP Bio-Imaging Systems, CA, USA). PBST composed of 0.05% Tween 20 in PBS buffer was prepared in house. Polyacrylamide gel (NuPAGE 4-12% Novex Bis-Tris Gels) was obtained from Invitrogen (Carlsbad, CA) with 1× MOPS running buffer. Fabrication of the Fluorinated PDMS Microarrays. To form the PDMS prepolymer, the PDMS oligomer was mixed with curing agent at a weight ratio of 10:1, and the resulting mixture was degassed in a vacuum for 30 min. The degassed PDMS mixture was poured on the stainless steel plate for matrix assisted laser desorption/ionization instrument (MALDI, Waters) and then cured at 70 °C for four hours. Once peeled from the stainless steel plate, the resulting 12 × 8 array pattern (2.5 mm id for each spot in a space of 5 cm width ×4 cm length) of the PDMS substrate was used as the grid for solution printing. The bare substrate was then modified with polyelectrolyte multilayers (PEMS) following the procedures described previously.21 Briefly, the bare PDMS substrates were activated by an oxygen plasma, subsequently exposed to a solution of h-PSMA 0.25% (w/v), and then followed by sequential coatings with branched PEI (0.25% w/v in DI water) and PAA (0.5% w/v in DI water) for four repeated times with an additional PEI as the top layer (h-PSMA-(PEI-PAA)4-PEI). In between, the PEI and PAA exposures, the substrates were washed with DI water (3 × 20 mL). The polyelectrolyte layers were cross-linked by the mixture containing EDC (30 mg/mL in PBS buffer) and NHS reagents (10 mg/mL in PBS buffer) to form amide bonds between the PEI/PAA layers. Subsequently, ACRL-PEG-NHS (1000 µg/mL in PBS, pH 7.4) was added to react with the exposed amine group of PEI molecules in the top layer. The optimal percentage of FD was determined by its solubility test in ethanol; the optimal percentages of AA and protein G were determined by compromising the maxima value of the static contact angle and the chemiluminescence of Antimouse IgG-HRP captured by protein G that was bound to AA on the surface. The PEMS coated surface was then photopolymerized with an optimized mixture of FD (15% v/v in ethanol), AA (1% v/v in ethanol), and DPA photoinitiator (1% w/v in ethanol) under 365 nm radiation at ambient temperature for 40 min, washed with ethanol to remove extra reagents and then dried under a nitrogen stream. The surface was then incubated with a mixture of EDC (30 mg/mL) and NHS (10 mg/mL) solution for two hours at ambient temperature. After being washed and dried, the activated substrate was incubated with protein G solutions with an optimized concentration (20 µg/mL) at 4 °C for four hours to form covalent amide bonds with AA. The treated PDMS was subsequently washed with PBST buffer and stored at 4 °C until use. Surface Characterization. The binding density of FD, AA, and protein G was determined from the difference of their amounts in the original solution and in the wash solution (unbound) after coating on a 416 mm2 PDMS surface. The FD and AA in the original and wash solution were quantified by mass spectrom(21) Makamba, H.; Hsieh, Y. Y.; Sung, W. C.; Chen, S. H. Anal. Chem. 2005, 77, 3971–3978.

Analytical Chemistry, Vol. 82, No. 18, September 15, 2010

7805

etry (4000 QTRAP, Applied Biosystem, MDS Sciex, Toronto, Canada) equipped with a Turbo V ion spray source under Multiple Reaction Mode (MRM) using the spiked E2 solution (25 µg/mL) as the internal standard (IS). Mass spectrometer conditions were set up as follows: positive ionization mode with 5500 V ionization source voltage and 350 °C source temperature. Nitrogen was used as the drying and nebulizing gas. For MRM experiments, the collision energies were set to 25 or 35 eV. The precursor ion/product ion pair was set to be 447 Da/ 189 Da, 73 Da/55 Da, and 273 Da/145 Da for FD, AA and IS, respectively. The peak area of FD and AA was divided by the peak area of IS and the normalized peak areas, FD/IS, and AA/IS, were used for relative quantification. The amount of FD and AA incorporated into the coating was then determined by subtracting the amount that was washed out from the original amount applied to the surface. Protein G was detected by polyacrylamide gel electrophoresis (NuPAGE Novex 4-12% Bis-Tris Gel, invitrogen) stained with comassie blue which band intensity was digitized and quantified by UVP Bio-Imaging Systems (CA). The system is equipped with digital densitometry and a computerized image analyzer. The amount of protein G incorporated into the coating was determined by subtracting the amount that was washed out from the original amount applied to the surface. The binding density was then calculated from dividing the incorporated amount by the chip area. The contact angle measurements were carried out using a CCD camera optical meter (Victor, Japan) with 5 µL water droplets at ambient temperature, and each measurement was repeated three times. All PDMS plates were dried with a stream of nitrogen before contact angle measurements. ESCA measurements were carried out on a ULVAC- PHI 5000 VersaProbe (PHI, Tokyo, Japan) in Al KR mode. Atomic force microscopy (AFM) investigations were carried out on a Agilent 5500 AFM Microscope (Digital Instruments Inc., Phoenix, AZ) using tapping mode and an etched silicon probe (scan size ) 5 µm × 5 µm). Before the measurements, each plate was washed with PBS buffer and then dried with a stream of nitrogen. To compare the resistance against nonspecific binding, three modifications, PEMS, PEG (ACRL-PEG-NHS), and FD of individual PDMS surfaces were investigated. These investigations used the ELISA plates described previously18 as templates for fabricating microwells (100 µL) on PDMS to prevent the spread of solutions on plain hydrophilic surfaces (PEMS and PEG). Antimouse IgG-HRP solution (0.4 µg/mL) and TMB solution were sequentially added into each well (5 mm id and 5 mm in depth). After two hours of incubation and PBS wash, an ELISA reader (TECAN, Austria) equipped with a photomultiplier tube (PMT) was used to capture the emission from each well, and the measured intensities were digitized by Image J software, version 1.41. (http://rsb.info.nih.gov/ij/download.html). FITC-labeled BSA solution (100 µg/mL) incubated 2 h was also applied to the surface to investigate nonspecific binding following the same wash procedure used for the antimouse IgG-HRP solution. Fluorescence imaging (Ex ) 480 nm, Em ) 570 nm) was also captured by UVP, Bio-Imaging Systems (CA), which has a detection limit of 2 µg/ mL for FITC-labeled BSA solution. Cell Culture. The human breast cancer cell line, MCF-7, was grown at 37 °C in a humidified 5% CO2 atmosphere in Dulbecco’s 7806

Analytical Chemistry, Vol. 82, No. 18, September 15, 2010

Modified Eagle’s Medium (DMEM) supplemented with fetal bovine serum (10%) and NaHCO3. Penicillin (1%) and antibioticantimycotic (1%) were added to the medium to inhibit bacterial growth. The cell growth was monitored daily using a microscope until the cells reached a state of confluence of 80-90%. Cells were then lysed with lysate buffer A (HEPES 10 mM, KCl 10 mM, EDTA 0.5 mM, EGTA 0.5 mM, and DTT 1 mM) and buffer B (HEPES 20 mM, KCl 10 mM, EDTA 1 mM, EGTA 1 mM, and DTT 1 mM). The cell lysate was used immediately or kept frozen at -80 °C until use. Protein Microarray Chip Assay. PDMS microarray chips, each of which contained protein G in its top layer, were immersed in the antibody solution (rabbit anti-ERR, 1 µg/mL in PBST buffer) for two hours and then washed with PBST to remove unbounded species. For microarray printing, each standard recombinant ERR solution (at concentrations of 138, 69, 34, 17, 8, and 0 ng/mL in PBST buffer) and the MCF-7 cell lysate solution were pipetted (4 µL) onto PDMS microarray chips with transferred MALDI grids. Sandwich assays were used for antibody microarray detection. After two hours of incubation, mouse anti-ERR (0.4 µg/mL in PBST buffer) and antimouse IgG-HRP (0.4 µg/mL in PBST buffer) were sequentially added. Finally, ECL reagent was added and the chip was covered with precleaned glass to enable detection of the emitted chemiluminescence signals by a BioSpectrum imaging system (UVP, Bio-Imaging Systems, CA), same system as that used for fluorescence measurement. RESULTS AND DISCUSSION Optimization for Surface Modifications. Figure 1 depicts the schematic of the step-by-step surface modifications composed of PEMS, ACRL-PEG-NHS, FD/AA photocopolymerization, and protein G in sequence. Generally speaking, ACRL-PEG-NHS was covalently bound to PEI, the top layer of PEMS, via carbodiimide coupling, and FD/AA was subsequently bound to the surface via photocopolymerization. Photopolymerization is a free radical reaction including three steps: (1) chain initiation by photoinitiator, such as 2,2-dimethoxy-2-phenylacetophenone, (2) chain propagation that involves a series of addition reactions among monomers, and (3) chain termination that stops the reaction as two molecular radicals are paired. AA molecules which contain carboxylic acid were copolymerized with FD to create a functional group that could covalently bind to protein G by carbodiimide coupling. Thus, increasing AA percentage increases the available sites for protein G binding but decreases the surface hydrophobicity. In contrast, increasing FD percentage increases surface hydrophobicity. Therefore, the percentages of FD and AA for copolymerization need to be optimized to reach high hydrophobicity and high number of available sites for protein G binding. Since FD has low solubility in polar solvents, the maximum solubility of FD in ethanol was investigated and determined to be 15%. We also measured the contact angles from the surfaces exposed to 5, 10, and 15% of FD solutions, respectively, and the results did indicate that 15% FD lead to the highest static contact angle (data not shown). We then used 15% FD as a fixed percentage to mix with AA solutions which final percentage ranged from 0.25, 0.50, 1.00, 2.00, 5.00 to 10.0%. As shown in Figure 2(A), the static contact angle of the modified surface is constant (0.25, 0.50, and 1.00%) up to 1.00% AA beyond which, the static contact angle begins to drop as the AA percentage increases. To

Figure 1. Schematics of the step-by-step surface modifications composed of PEMS (including cross-linking between layers by EDC/NHS reaction), ACRL-PEG-NHS, FD/AA copolymerization, and protein G on the top of a PDMS substrate.

find out whether the binding density of protein G increases with the percentage of AA in the solution, we then applied protein G solution (20 µg/mL) to the surface, which was then subsequently detected by IgG-HRP chemiluminescence. As shown in Figure 2A, however, the binding density of protein G increases instantly as the AA percentage increases but becomes plateau as the AA percentage is beyond 1.00%. This phenomenon is indeed not yet understood. We postulated a mechanism depicted in Figure 2B to explain the data shown in Figure 2A. FD, AA, and ACRL-PEG-NHS molecules all contain a reactive double bond that could induce photopolymerization. Since both AA and FD were added together to react with ACRL-PEGNHS on the surface, so-called one-pot reaction, the final products should be mediated by three reactions among reactants: (1) FD and AA, (2) AA and the ACRL-PEG-NHS on the surface, and (3) FD and the ACRL-PEG-NHS on the surface as indicated in Figure 2(B). In this experiment, the concentration of FD (15%) and the ACRL-PEG-NHS surface density were both kept constant but the concentration of AA solution was systematically increased. Thus, increasing AA concentration will drive reaction 1, which forms AA-AA, FD-AA, or FD-FD, and reaction 2, which leads to surfacebound AA to their right. Below 1% AA in the solution, the binding density of FD was not much varied since reaction 3 was not influenced, but reaction 2 could increase the binding density of AA and thus the AA-captured protein G. Beyond 1% AA in the solution, however, reaction 1 became important, and reaction 2 could have reached a saturation point possibly because of the hindrance effect of long chain FD. Thus, the extra AAs could selfpolymerize or polymerize with FD to yield AA-AA or FD-AA polymers, which were washed away by the wash buffer (reaction 1). Since the FD concentration was kept constant in the solution while increasing AA percentage, the surface-bound FD molecules could be lost to form FD-AA polymer as controlled by the balance

of three reactions. Thus, the binding density of the AA-captured protein G was constant but the binding density of FD was decreased as the AA percentage was beyond 1%. Taken together, we postulated that the surface contact angle was majorly governed by the binding density of FD but not AA, which could explain the observation shown in Figure 2A based on the proposed mechanism. This postulation is also logical since the FD binding density is orders of magnitude higher than AA binding density (data shown in the following section). Moreover, the AA molecule is likely to be buried under the long chain of FD molecules and has little contribution to the surface contact angle. Although AA molecules may be buried under the long chain of FD, some protein G molecules could still penetrate into the polymer layer to bind with the buried AA molecules.19 Based on the results, we decided to use a monomer mixture composed of 15% FD and 1% AA for photocopolymerization on PDMS surface. The optimal concentration of the protein G solution used for the subsequent coating was also determined from the amount of the captured antimouse IgG-HRP. As shown in Figure 2C, the captured antimouse IgG-HRP reaches a maximum as the protein G concentration exceeds 10 µg/mL. To ensure complete coverage, a higher concentration, 20 µg/mL, was used for coating. Surface Characterization. Figure 3 shows the static contact angles measured from different modified layers. As shown, the static contact angle of a 5 µL water droplet on native PDMS is 87.26° (standard deviation SD ) 2.05), which corresponds to 112° when a 10-µL droplet was used but without gravity calibration (Figure S1 in Supporting Information).18 After plasma oxidization, the static contact angle decreases to 23.21° (SD ) 1.78) but increases to 83.49° (SD ) 1.41) after 7 days of storage. PEMS modification results in a hydrophilic surface with a contact angle of 6.55° (SD ) 1.13). This contact angle is slightly increased to 8.71° (SD ) 1.43) after 7 days of storage. PEG modification (ACRLAnalytical Chemistry, Vol. 82, No. 18, September 15, 2010

7807

Figure 2. Optimization for AA and protein G concentration in the reaction solution. (A) Static contact angles measured from the surface (PDMS/PEMS/ACRL-PEG-NHS) photopolymerized with various percentages of AA prepared in 15% FD ethanol solution (s) and the corresponding chemiluminescence ( · · · ) from antimouse IgG-HRP captured by protein G that was bound to AA by applying protein G solution (20 µg/mL) to the photopolymerized surfaces. (B) A postulated three-reaction mechanism among (1) FD and AA, (2) AA and ACRL-PEG-NHS on the surface, and (3) FD and ACRL-PEG-NHS on the surface for one-pot reaction (C) Chemiluminescence of antimouse IgG-HRP captured from the FD/AA surfaces that were modified by various concentrations of protein G solution (10-20 µg/mL). 7808

Analytical Chemistry, Vol. 82, No. 18, September 15, 2010

Figure 3. Static contact angles measured from the layer-by-layer surfaces: (A) a native PDMS plate, (B) an oxidized PDMS plate, (C) a PEMS-coated PDMS plate, (D) ACRL-PEG-NHS modification on the top of a PEMS-coated PDMS plate, (E) FD/AA copolymerization on the top of an ACRL-PEG-NHS-coated PDMS plate, and (F) anti-ERR immobilized on an FD/AA-coated PDMS plate. The black and white columns represent measurements on the first day and on the seventh day of storage, respectively.

PEG-NHS) increases the contact angle to 20.23° (SD ) 1.15) and exhibits a long-term hydrophilicity.22-24 Upon coating with FD/ AA, the surface becomes hydrophobic with a contact angle of 81.97° (SD ) 1.55) and remains hydrophobic (76.90°, SD ) 1.72) after antibody binding. Moreover, the contact angle does not change significantly after 7 days of storage. These results indicate that the fluorinated modification is stable and hydrophobic as bare PDMS. Since the PEI and PAA molecules are hydrophilic, reducing the number of multilayers will naturally reduce the binding density of PEI and PAA, leading to an increase of the surface hydrophobicity. Thus, if fewer layers of PEMS are used, we believe the hydrophobicity of the FD/AA modified surfaces will increase. The coating densities of FD and AA, as well as protein G, were determined from the difference of their amount measured in the original solution and in the wash solution (unbound). All experiments were repeated for three times. As shown in Figure 4A, about 25% and 15% of AA and FD, respectively, remain in the solution after binding. Based on the difference and the chip area, the binding density for FD and AA was determined to be 50 mg/ mm2 (0.11 mmol/mm2) and 0.19 mg/mm2 (2.6 µmol/mm2), respectively. Because of the small size of FD and AA, such high binding densities indicate that the copolymerization could occur in multilayers. In our previous study,19 we showed that protein G could polymerize with the hydrogel inside the PEMS and inside the bulk PDMS, indicating the reacting monomer, such as ACRL-PEG-NHS, and FD/AA could penetrate the PEMS layers to react with the noncrossed linked PEIs. Although the mechanism is not well understood yet, the nature of soft material like PDMS is very different from that of the solid material like glass. Diffusion can easily occur for soft material. It is likely that ACRL-PEG-NHS could bind to the PEI of the inner PEMS layers and thus allow the subsequent FD/ AA copolymerization to occur in inner multilayers. The amount of protein G incorporated into PDMS was also determined from the difference of protein G in the original solution and in the wash solution (unbound). As shown in Figure 4B, the concentration of protein G in the applied solution (20 µg/mL) was reduced by almost one-half based which the binding density of (22) Li, Z.; Yang, X.; Wu, L.; Chen, Z.; Lin, Y.; Xu, K.; Chen, G. Q. J. Biomater. Sci., Polym. Ed. 2009, 20, 1179–1202. (23) Cerruti, M.; Fissolo, S.; Carraro, C.; Ricciardi, C.; Majumdar, A.; Maboudian, R. Langmuir 2008, 24, 10646–10653. (24) Han, D. K.; Park, K. D.; Hubbell, J. A.; Kim, Y. H. J. Biomater. Sci., Polym. Ed. 1998, 9, 667–680.

protein G on the FD coated surface was calculated to be around 0.24 µg/mm2. The binding density of protein G is much lower than AA molecules, which form covalent bonds with protein G. We believe that one protein G could bind to multiple AA molecules through its lysine residues. However, there are still a large proportion of protein G molecules that could not penetrate the FD layer to bind to AA molecules. Although it is possible to increase the binding density of protein G by using acrylic acids with a longer chain or fluorinated compounds with a shorter chain than the chain length of the compounds we used in this study, the surface hydrophobicity will be substantially decreased. Furthermore, to verify that protein G was covalently bound to the surface, we fabricated the surface with and without adding EDC/NHS activating reagents for covalent bonding. The results shown in Figure S2 in the Supporting Information clearly indicate that without EDC/NHS for covalent bonding, the amount of nonspecifically adsorbed protein G was too low to be detected by antimouse IgG-HRP. In contrast, adding EDC/NHS to the surface for covalent bonding of protein G, the captured antimouse IgG-HRP could be readily detected as bright spots. These results indicate that Protein G molecules were covalently bound to the surface. Tapping mode AFM was used to inspect the surface topography of the modifications. Figure 5A shows that the native PDMS is relatively flat with an rms value of 1.10 nm. After FD/AA copolymerization, the roughness of the surface increases (6.33 nm), though the modification is uniform throughout the substrate (Figure 5B), indicating a thick and homogeneous surface coverage of the fluorinated modification. As the surface was subsequently topped with protein G and anti-ERR, the roughness further increased to an rms value of 9.86 nm, but the modification remained uniform across the substrate (Figure 5C). These results reflect a mixed consequence of the film thickness and the binding density of the modified layers. Although the multilayer composed of PEMS + PEG + FD/AA coating is expected to be 20-30 times longer or thicker than that of protein G+anti-ERR coating, the rms increment, 5.23 nm (6.33-1.10), arising from PEMS + PEG + FD/AA coating is less than two times of the increment, 3.53 nm (9.86-6.33), arising from the additional protein G + anti-ERR coating. Such phenomena may be explained by the fact that the binding density of FD/AA is orders of magnitude higher than that of protein G. Surfaces with high binding density normally have smaller roughness or lower rms value than surfaces with low binding density. Analytical Chemistry, Vol. 82, No. 18, September 15, 2010

7809

Figure 4. (A) MS chromatograms of the monomer solution detected in the original (Blue line) and in the wash (unbound) solution (Red line) after surface binding. The concentration of the original FD and AA solution was 250 mg/mL and 10 mg/mL, respectively, and the final concentration of 17β estradiol spiked into the solution as the internal standard was 25 µg/mL. (B) PAGE of protein G present in the applied solution (20 µg/mL) and in the wash solution (unbound) after surface binding. As specified by the manufacturer, protein G has a molecular weight of approximately 22 kDa, but its band appears in the 32-36 kDa region. All experiments were repeated for three times. All digitized intensities were expressed in a normalized scale (the original amount was scaled as 1.00) shown in the bottom of each panel. Experimental details are described in the Experimental section.

ESCA was used to characterize the chemical composition of the modifications. As shown in Figure 6A, the spectral lines of C 1s (284 eV), N 1s (400.5 eV), O 1s (532.5 eV), Si 2s (153.5 eV), and Si 2p (102.5 eV) are clearly evident from the PEMS-PEG coated surface that was cross-linked with amide bonds. A Gaussian multipeak fit reveals that the chemical states of C 1s are C-H, C-C, CdC, and C-O/C-O with respective energies at 284, 286, and 288 eV (Figure 6C). For the surface coated with FD/AA, a 7810

Analytical Chemistry, Vol. 82, No. 18, September 15, 2010

strong band of F 1s (690.8 eV) is shown (Figure 6B), but other atomic (C, N, O, and Si) lines are decreased (Figure 6B). The Gaussian multipeak fit further reveals that the chemical states of C 1s are C-F groups with two energies at 292 and 294 eV (Figure 6D). These results confirm the success of fluorinated modifications by covalent photopolymerization. Moreover, the percentage of F was estimated to be around (35.4%), indicating a high fluorine coverage, which is consistent with the high FD binding density and small surface roughness. Nonspecific Binding. Hydrophobic or ionic (acidic silanol groups) interactions are the major causes of nonspecific binding on native PDMS substrates. BSA protein is commonly used as a blocking reagent to reduce nonspecific binding in ELISA assays. Poly(ethylene glycol) (PEG) is also a common reagent used in various biochemical analyses to reduce nonspecific binding. Basically, because these chemicals are ionic or hydrophilic, they resist nonspecific binding through electrostatic or steric hindrance effects.25-27 The fluorine atom is known to have a large electronegativity and to form a stable covalent bond with carbon. However, the fluorine atom also has extremely high electron density because of its small atomic size, an attribute that causes it to resist intermolecular interactions. Thus, fluorinated compounds are stable, hydrophobic, and generally antisticking agents. We previously demonstrated that PEMS modification could turn a hydrophobic PDMS surface to a hydrophilic one and reduce nonspecific protein binding by a factor of 2-5 compared to BSA.19 Figure 7 indicates that a second PEG modification (on top of the original PEMS layer) could further reduce nonspecific binding by a factor of 2-3 compared to PEMS alone. Moreover, topping the PEG layer with an FD/AA modification resulted in extremely low background emission (Figure 7). Compared to the modification employing a PEG second layer, the reduction in nonspecific binding by FD/AA coating was estimated to be more than 1 order of magnitude. These observations are consistent with a recent report20 that fluorinated compounds were better blocking reagents than BSA and nonfat milk and, therefore, exhibited a stronger resistance to nonspecific binding. The major merit for fluorinated coating is to create a surface that is hydrophobic but could resist nonspecific binding. Such characteristics, however, is not possible with either the hydrophobic bare PDMS or the hydrophilic BSA and PEG coatings. Binding Efficiency Tests by ELISA. The sandwich detection method was used to evaluate immunoassays of the FD/AAmodified substrate, which were compared with another set of results using PEG as the top layer. For this comparison, the binding density of protein G on the PEG coated surface (0.29 µg/ mm2) was determined to be similar to that (0.24 µg/mm2) for FD-coated surface based on the same procedure described previously. ERR is a clinical marker of breast tumor, and the quantification of ERR has been routinely performed to predict prognosis in clinical laboratories. To identify the total amounts of ERR contained in different cell types is important due to the critical role of ERR in cell biology. For these reasons, we (25) De Gennes, P. G. Ann. Chim. 1987, 77, 389. Taunton, H. J.; Toprakcioglu, C.; Fetters, L. J.; Klein, J. Nature 1988, 332, 712–714. (26) Jeon, S. I.; Lee, J. H.; Andrade, J. D.; de Gennes, P. G. J. Colloid Interface Sci. 1991, 142, 149–158. (27) Jeon, S. I.; Andrade, J. D. Ibid. 1991, 142, 159–166.

Figure 5. AFM photos measured from (A) the native PDMS plate with an rms value of 1.10 nm, (B) the fluorinated (FD/AA copolymerized on ACRL-PEG-NHS-coated) PDMS plate with an rms value of 6.33 nm, and (C) anti-ERR immobilized on an FD/AA copolymerized PDMS plate with an rms value of 9.86 nm.

Figure 6. ESCA spectra measured from the layer-by-layer surfaces: (A) ACRL-PEG-NHS modification on the top of a PEMS-coated PDMS plate and (B) FD/AA copolymerization on the top of an ACRL-PEG-NHS-coated PDMS plate. Panels C and D are the C1s spectrum fitted from A and B.

set out to quantify the absolute concentration of ERR in the breast cancer cell line, MCF7.28-31 The pipetted solution formed well microdroplets (4 µL) on the FD/AA modified anti-ERR antibody chip for detection and quantification. The detection limit of the chip was estimated to be around 8 ng/mL for ERR solution, which is much lower than a commercial ELISA kit for ERR (>12 µg/mL) as specified by (28) Stroock, A. D.; Kane, R. S.; Weck, M.; Metallo, S. J.; Whitesides, G. M. Langmuir 2003, 19, 2466–2472. (29) Yongfeng, S.; Xiao, H.; James, D.; Mitchell, A. L.; Myles, B. Cell 2000, 103, 843–852. (30) Vijay, K.; Stephen, G.; Gary, S.; Meera, B.; Jin, J. R.; Pierre, C. Cell 1987, 51, 941–951. (31) Geoffrey, L. G.; Paul, G.; Michael, W.; Andrew, B.; Yvonne, H.; John, S. Science 1986, 31, 1150–1154.

the manufacturer.32 Figure 8A shows a printed chip with a series of standard recombinant ERR solutions ranging from 138 to 8 ng/ mL (n ) 3) and the MCF-7 cell lysate (n ) 3) without the blocking reagent. The spot intensities indicate quantitative modulation in accordance with their concentrations. Figure 8B contains the calibration curves constructed from the data shown in Figure 8A and an additional data set with the same sample but with the blocking step. The fitted linear calibration equation is Y = 0.0090X + 0.0777 (R2 ) 0.9986) with the blocking reagent (solid line in Figure 8B) and Y = 0.0091X + 0.0194 (R2 ) 0.9982), without the blocking reagent (dotted line in Figure 8B). Apparently, detection without the blocking reagent is comparable to that with (32) Active Motif company. Catalog Nos.49296 and 49796. www.activemotif.com.

Analytical Chemistry, Vol. 82, No. 18, September 15, 2010

7811

Figure 7. Nonspecific binding test (n ) 3) of a PEMS coated, PEG coated (ACRL-PEG-NHS modification on the top of PEMS), and fluorinated (FD/AA copolymerization on the top of PEG) PDMS plate by (A) antimouse IgG-HRP (0.4 µg/mL) and (B) FITC-BSA (100 µg/mL). Details are described in the Experimental Section.

Figure 8. (A) Standard ERR solutions with a concentration of 138, 69, 34, 17, 8, and 0 ng/mL and the solution of MCF-7 cell lysate spotted in sequence from left to right on the anti-ERR microarray of a fluorinated PDMS chip (2.5 mm spot size). Three repeats (n ) 3) were performed without the blocking step for each solution. (B) Calibration curves of the ERR constructed with (s) and without ( · · · ) blocking steps (data shown in A), as well as on the PEG modified substrate (- - - -) with a blocking step. In the blocking step, BSA (10 mg/mL in PBST) was added prior to sample incubation.

blocking reagent. This indicates that the blocking step can be eliminated for fluorinated surfaces, thereby reducing the processing time and labor and leading to a simplified bioassay. We have further constructed a calibration curve using a chip coated with PEG as the top layer21 and obtained the fitted linear equation as Y = 0.0015X + 0.0663 (R2 ) 0.9903) (dashed line in Figure 8B), indicating that the fluorinated coating is 6 times more sensitive than the PEG coating. These results are consistent with the lower nonspecific binding and thus higher specific binding associated with the FD/AA coated versus PEG-coated chips. Therefore, we believe some coating steps including the PEG layer can be 7812

Analytical Chemistry, Vol. 82, No. 18, September 15, 2010

eliminated to simplify the coating process as long as an activated top layer is created to bind FD/AA. We have also investigated the long-term reactivity of the FD/AA-coated antibody chips after storage in dried and cold (4 °C) conditions for 7 days. The results indicate no significant changes in the calibration curve (Y = 0.0098X + 0.099, R2 ) 0.9939, data not shown). This demonstrates that stored chips exhibit virtually the same reactivity as those that are freshly prepared. On the basis of the constructed calibration curve (Figure 8B), the concentration of ERR in MCF7 cells (Figure 8A) was determined to be 48 ± 2.2 ng/mL, which is consistent with results reported in the literature.31,33 A standard addition method was also used to validate the detected amount of ERR in MCF-7 cells by spiking standard ERR solution into the MCF-7 cell lysate (the final spiking concentration was 34 ng/mL). The spiked solution was analyzed by FD/AA coated anti-ERR antibody chips and the concentration of the spiked solution was determined to be 84.6 (±0.3) ng/mL based on the calibration curve constructed in Figure 8B. The concentration in the nonspiked MCF-7 cells was determined by subtraction to be 50.6 (±0.3) ng/mL, which agrees with the value obtained without spiking (48 ± 2.2 ng/mL). Alternatively, the spiked amount can be calculated as 36 ng/mL if the amount of ERR in the nonspiked solution (48 ± 2.2 ng/mL) is subtracted from the detected concentration. These concentration determinations demonstrate the excellent recovery rate (101 ± 0.0036)% of the FD/AA coated chip, which indicates an easy cleanup for FD/ AA coated antibody chips that could potentially lead to reduced contaminations. CONCLUSION We have fabricated PDMS-based antibody chips coated with fluorinated compounds by mask-less photocopolymerization and subsequent covalent bonding with protein G. The functional (33) David, J. B.; Gary, K. S.; Birgit, S.; Christian, A.; Jason, M. H.; Bradford, W. G.; Christopher, C. B.; Michael, A. B. J. Am. Soc. Mass Spectrom. 2008, 19, 729–740.

fluorinated surface demonstrated good stability in conserving longterm reactivity, hydrophobicity in forming good droplets during sample printing, and resistance against nonspecific binding. Moreover, without the blocking step, the fluorinated coating demonstrated greater sensitivity than the PEG coating with regard to absolute quantification of the low abundant protein (ERR) in cell lines (MCF7). In view of the extremely high binding density of the fluorinated compounds, the coating steps of PEMS may be reduced, which will further simplify the fabrication process. Finally, although the fluorinated coating is used on a PDMS substrate in this study, such coatings should be applicable to other substrates for fabricating protein chips.

ACKNOWLEDGMENT This work was supported by the National Science Council in Taiwan. SUPPORTING INFORMATION AVAILABLE Additional material as described in the text. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review July 7, 2010. Accepted August 11, 2010. AC101799F

Analytical Chemistry, Vol. 82, No. 18, September 15, 2010

7813