Use of PLL-g-PEG in Micro-Fluidic Devices for Localizing Selective

By utilizing flow-controlled PLL-g-PEG and PLL-g-PEGbiotin modification of predefined regions of a poly(dimethylsiloxane) (PDMS) micro-fluidic device,...
0 downloads 0 Views 256KB Size
Download PDF
Langmuir 2006, 22, 10103-10108

10103

Use of PLL-g-PEG in Micro-Fluidic Devices for Localizing Selective and Specific Protein Binding Rodolphe Marie,† Jason P. Beech,† Janos Vo¨ro¨s,‡ Jonas O. Tegenfeldt,*,† and Fredrik Ho¨o¨k*,† DiVision of Solid State Physics, Lund UniVersity, P.O. Box 118, SE-221 00, Lund, Sweden, and Laboratory for Surface Science and Technology and Institute for Biomedical Engineering, Department of Materials, ETH Zurich, CH-8092 Zurich, Switzerland ReceiVed January 20, 2006. In Final Form: April 20, 2006 By utilizing flow-controlled PLL-g-PEG and PLL-g-PEGbiotin modification of predefined regions of a poly(dimethylsiloxane) (PDMS) micro-fluidic device, with an intentionally chosen large (∼1 cm2) internal surface area, we report rapid (10 min), highly localized (6 × 10-6 cm2), and specific surface-based protein capture from a sample volume (100 µL) containing a low amount of protein (160 attomol in pure buffer and 400 attomol in serum). The design criteria for this surface modification were achieved using QCM-D (quartz crystal microbalance with energy dissipation monitoring) of serum protein adsorption onto PLL-g-PEG-modified oxidized PDMS. Equally good, or almost as good, results were obtained for oxidized SU-8, Topas, and poly(methyl metacrylate) (PMMA), demonstrating the generic potential of PLL-g-PEG for surface modification in various micro-fluidic applications.

Introduction Micro- and nano-fluidic devices have emerged as important components in the purification, fractionation, and analysis of biological samples. Together with passive or active devices, such as pumps and valves, these are the most critical building blocks for the fabrication of functional assemblies aiming at micro total analysis systems (µ-TAS). Originally fabricated mainly by silicon micromachining techniques, µ-TAS are now being increasingly fabricated in polymer materials. Because the processing steps and the raw materials are simpler and less expensive than in silicon-based micro-fluidic devices, polymeric structures are suitable candidates for single-use analytical devices for the purification and analysis of biological samples.1,2 Up to now, poly(dimethylsiloxane) (PDMS) is currently the polymer most widely and successfully used in both research and commercial systems, as demonstrated by fluidic cells for surface plasmon resonance sensors,3,4 fluorescence measurements5,6 and devices for the separation of DNA and proteins or cell sorting; in the latter case nearly always based on electrophoresis combined with a fluorescence read-out.7 To take full advantage of the components and processing techniques already developed, based on traditional materials such as silicon, silicon dioxide, silicon nitride, metals, and metal oxides, polymers other than PDMS, such as poly(methyl metacrylate) (PMMA), SU-8, and cyclo-olefin copolymer (COC), are becoming increasingly rele* To whom correspondence should be addressed. [email protected] (J.T.); [email protected] (F.H.). † Lund University. ‡ ETH Zurich.

E-mail:

(1) Sia, S. K.; Whitesides, G. M. Electrophoresis 2003, 24, 3563-3576. (2) . Guber, A. E.; Heckele, M.; Herrmann, D.; Muslija, A.; Saile, V.; Eichhorn, L.; Gietzelt, T.; Hoffmann, W.; Hauser, P. C.; Tanyanyiwa, J.; Gerlach, A.; Gottschlich, N.; Knebel, G. Chem. Eng. J. 2004, 101 (1-3), 447-453. (3) Sjolander, S.; Urbaniczky, C. Anal. Chem. 1991, 63, 2338-2345. (4) Wheeler, A. R.; Chah, S.; Whelan, R. J.; Zare, R. N. Sens. Actuators B: Chem. 2004, 98 (2-3), 208-214. (5) Chabinyc, M. L.; Chiu, D. T.; Mcdonald, J. C.; Stroock, A. D.; Christian, J. F.; Karger, A. M.; Whitesides, G. M. Anal. Chem. 2001, 73 (18), 4491-4498. (6) Ros, A.; Hellmich, W.; Duong, T.; Anselmetti, D. J. Biotechnol. 2004, 112, 65-72. (7) Chen, H.; Zhang, Z.; Chen, Y.; Brook, M. A.; Sheardown, H. Biomaterials 2005, 26 (15), 2391-2399; reviewed in ref 1.

vant materials for the fabrication of microsystems integrating several transducer elements (optical, fluidic, electronic, and mechanical).8-10 Common to all of these devices is the challenge of successfully modifying their internal surfaces so that they become highly inert to nonspecific biomolecule adsorption, thus enabling efficient handling of complex biological samples such as cell lysate, whole blood, or serum. Hence, finding a generic surface-modification protocol that efficiently reduces nonspecific adsorption of proteins has emerged as one of the most crucial steps in the development of miniaturized fluidic devices at both ends of the concentration spectrum. Nonspecific adsorption resulting from highly concentrated samples may cause clogging of the device, whereas in samples where the system of interest is intrinsically present at very low concentrations, the molecules may never reach the area intended for analysis. One widely used molecule for surface passivation is poly(ethylene glycol) (PEG) and specifically the tri-block copolymer polyethyleneoxy-polyoxypropylene-polyethyleneoxy (known as Pluronic F108 manufactured by BASF GmbH, Germany).11 Zare et al. recently described the successful use of n-dodecyl-β-D-maltoside for the same purpose.12 These molecules have in common that they adsorb readily to hydrophobic surfaces, thus preventing nonspecific adhesion of proteins and cells. In particular, PEG chains on a surface function as a water-binding hydrogel-like brush, with protein-resistant properties that are highly dependent on the length, flexibility, and density of the PEG chains. The dominating molecular mechanisms underlying protein resistance of grafted PEG are still not fully understood,13 but a brush-induced steric repulsion preventing contact between proteins and the underlying surface, and the (8) Nilsson, D.; Balslev, S.; Kristensen, A. J. Micromech. Microeng. 2005, 15 (2), 296-300. (9) Lu, H.; Schmidt, M. A.; Jensen, K. F. Lab Chip 2005, 5 (1), 23-29. (10) Lee, K. J.; Fosser, K. A.; Nuzzo, R. G. AdV. Funct. Mater. 2005, 15 (4), 557-566. (11) Caldwell, K. D. In Poly(Ethylene Glycol): Chemistry and Biological Applications; Harris, J. M., Zalipsky, S., Eds.; American Chemical Society: San Francisco, 1997; Chapter 25, pp 400-419. Hellmich, W.; Regtmeier, J.; Duong, T. T.; Ros, R.; Anselmetti, D.; Ros, A. Langmuir 2005, 21, 7551-7557. (12) Huang, B.; Wu, H.; Kim, S.; Zare, R. N. Lab Chip 2005, 5 (10), 1005. (13) Heuberger, M.; Drobek, T.; Spencer, N. D. Biophys. J. 2005, 88, 495504.

10.1021/la060198m CCC: $33.50 © 2006 American Chemical Society Published on Web 10/18/2006

10104 Langmuir, Vol. 22, No. 24, 2006

hydration shells around the PEG moieties, which energetically suppress adsorption of proteins, are considered as the two dominating mechanisms. With micro-fluidics being developed at smaller and smaller channel diameters, it is, however, impractical to rely on surface modifications requiring hydrophobic surfaces due to difficulties in wetting and bonding. To circumvent these problems, Delamarche et al. recently reported that oxidized PDMS can be pre-modified using poly(oxyethylene)-silane, enabling covalent attachment of poly(oxyethylene) chains.14 With the ultimate aim of combining our work and that of others on lab-on-a-chip-based separation schemes15 with, for example, label-free diffraction-limited detection16 and/or scanning nearfield microscopy17 devices, we demonstrate in this work a rational aqueous-based surface functionalization procedure making use of fluidics to control the adsorption of differently modified PLLg-PEG molecules18,19 to predefined regions of a PDMS microfluidic device. The benefit of using PLL-g-PEG for this purpose is that the surface modification can be done in aqueous environments, which is in contrast with covalent attachment methods20 and methods relying on hydrophobic surfaces,11 which are not easily compatible with bonding, or lipid bilayer formation,21 which is strongly material specific. First, this allows the surface to be modified after assembly of the fluidic device using bonding between oxidized PDMS and SiO2, thus circumventing the problems related to the bonding of pre-modified surfaces. Second, the documented protein-repellent property of PLL-g-PEG22 ensures extremely low nonspecific adsorption of analytes outside the detection spot(s), thus enabling highly efficient, site-selective protein capture from lowconcentration suspensions. In particular, high contrast between protein adsorption, achieved through specific recognition of streptavidin using biotin-modified PLL-g-PEG (PLL-gPEGbiotin), and low nonspecific protein adsorption, achieved by using pure PLL-g-PEG, is demonstrated from both pure buffer and serum suspensions using a device intentionally fabricated with a high surface-to-volume ratio (1 cm2:1 µL). To establish the surface-modification protocols of the device, quartz crystal microbalance with energy dissipation monitoring (QCM-D) was used to characterize and optimize the reduction of nonspecific (serum) protein adsorption to PLL-g-PEG adsorbed onto oxidized and nonoxidized PDMS. To demonstrate the general applicability of PLL-g-PEG to micro-fluidic devices, a number of other polymers relevant for, for example, nano-imprint (14) Delamarche, E.; Donzel, C.; Kamounah, F. S.; Wolf, H.; Geissler, M.; Stutz, R.; Schmidt-Winkel, P.; Michel, B.; Mathieu, H. J.; Schaumburg, K. Langmuir 2003, 19, 8749-8758. (15) Huang, L. R.; Cox, E. C.; Austin, R. H.; Sturm, J. C. Science 2004, 304, 987-990. Zheng, S.; Yung, R.; Tai, Y.-C.; Kasdan, H. MEMS 2005 Proc. 2005, 851-854. (16) Dahlin, A.; Zach, M.; Rindzevicius, T.; Kall, M.; Sutherland, D. S.; Hook, F. J. Am. Chem. Soc. 2005, 127 (14), 5043-5048. Stuart, D. A.; Haes, A. J.; Yonzon, C. R.; Hicks, E. M.; Van Duyne, R. P. IEE Proc.-Nanobiotechnol. 2005, 152 (1), 13-32. (17) Tegenfeldt, J. O.; Bakajin, O.; Chou, C. F.; Chan, S.; Austin, R. H.; Fann, W.; Liou, L.; Chan, E.; Duke, T.; Cox, E. C. Phys. ReV. Lett. 2001, 86 (7), 1378-1381. (18) Poly(L-lysine)-g-poly(ethylene glycol) (PLL-g-PEG) is a copolymer with two functional parts: poly(ethylene glycol) (PEG) chains used for its protein resistant properties grafted onto a poly(L-lysine) (PLL) backbone that adsorbs by electrostatic interaction on negatively charged surfaces.22 (19) Lee, S.; Vo¨ro¨s, J. Langmuir 2005, 21 (25), 11957-11962. (20) Chen, H.; Zhang, Z.; Chen, Y.; Brook, M. A.; Sheardown, H. Biomaterials 2005, 26 (15), 2391-2399. (21) Phillips, K. S.; Cheng, Q. Anal. Chem. 2005, 77, 327-334. Yang, T. L.; Jung, S. Y.; Mao, H. B.; Cremer, P. S. Anal. Chem. 2001, 73, 165-169. (22) Kenausis, G. L.; Vo¨ro¨s, J.; Elbert, D. L.; Huang, N.-P.; Hofer, R.; RuizTaylor, L.; Textor, M.; Hubbell, J. A.; Spencer, N. D. Phys. Chem. B: Condensed Phase 2000, 104 (14), 3298-3309. Huang, N.-P.; Michel, R.; Vo¨ro¨s, J.; Textor, M.; Hofer, R.; Rossi, A.; Elbert, D. L.; Hubbell, J. A.; Spencer, N. D. Langmuir 2001, 17 (2), 489-498.

Marie et al. Table 1. Parameters Used for Spin-Coating Crystals with Polymers polymer

solvent

content (%w/w)

speed (rpm)

thickness (nm)a

PDMS SU-8 PMMA topas 5013

N-hexane cyclopentanone anisole toluene

47 22 9-11 8.5

3000 3000 2000 3000

500 ( 50 550 ( 50 400 ( 50 650 ( 50

a The thickness was measured by QCM-D in air, and the variation in the thickness over up to 10 independent samples was typically (50 nm.

lithography (Topas)23 and surface micro-machining (SU-8, PMMA),24,25 are also included in this part of the study. Experimental Section Chemicals. Milli-Q water was used for preparing buffers and cleaning solutions. Sodium chloride, sodium hydroxide, sodium dodecyl sulfate (SDS), human serum, and streptavidin-Cy3 were purchased from Sigma-Aldrich (Germany). In all experiments, the buffer solution was 10 mM HEPES (4-(2-hydroxyethyl)piperazine1-ethanesulfonic acid) from Fluka (Germany), at pH 7.4 containing 0.1 M sodium chloride. A 1 M sodium chloride solution and a 1% SDS solution were successively used for cleaning the QCM-D measurement cell between each experiment and the micro fluidics between each device functionalization. PLL(20)-g[3.5]-PEG(2) (referred to as PLL-g-PEG throughout the text) is described elsewhere.19,22 The same copolymer was used in its biotinylated26 form (PLL-g-PEGbiotin) in order to form molecular layers capable of capturing streptavidin from the solution. The copolymers were dissolved at 0.1 mg/mL in buffer. Human serum from Sigma-Aldrich was diluted to 5% in buffer. Materials. PDMS was purchased from GE Bayer (RTV615, GE Bayer, USA). SU-8 (Microchem Nano, SU-8 2005), such as the PMMA (Microchem 950k A9), was purchased from Microchem (Massachusetts, USA). The COC used was Topas 5013 (Ticona, USA). Surface Modifications for QCM-D Measurements. QCM-D was performed using a Q-Sense D300 instrument (Q-Sense AB, Sweden) using an AT-cut quartz crystal with a fundamental frequency of 5 MHz. Gold-coated quartz crystals were cleaned in a bubbling piranha solution (1:4 hydrogen peroxide 30% in sulfuric acid 92%) for 10 min followed by rinsing and drying prior to spin-coating of polymers. Typically, the polymers were diluted in their respective solvent and the crystals were spun for 30 s at 500 rpm and for 2 min at the final speed in order to obtain submicrometer thickness, being a sufficiently thin coating for QCM-D to detect the adsorption of PLL-g-PEG and serum (Table 1). After coating, the PDMS was cured at 90 °C for 45 min in a convection oven. SU-8 was baked for 60 s at 95 °C, cross-linked by UV exposure for 30 s at 125 mJ/cm2, and baked for 120 s at 95 °C. PMMA was baked for 30 min at 180 °C. Unless otherwise stated, the polymer coatings were oxidized for 60 s in an oxygen plasma at 600 W created by standard 2.45 GHz microwaves and an oxygen pressure of 5 mbar (Plasma Preen, U.K.) prior to mounting of the crystals in the QCM-D measurement cell. The thickness of the polymer coating was measured by QCM-D in air, and variations in the polymer thickness were within 50 nm as given in Table 1. The changes in polymer thickness induced by the plasma treatment were within the uncertainty of the QCM-D measurements, corre(23) Nielsen, T.; Nilsson, D.; Bundgaard, F.; Shi, P.; Szabo, P.; Geschke, O.; Kristensen, A. J. Vacuum Sci. Technol. B 2004, 22 (4), 1770-1775. (24) Chuang, Y.-J.; Tseng, F.-G.; Cheng, J.-H.; Lin, W.-K. Sens. Actuators A 2003, 103, 64-9. Kudryasho, V.; Yua, X.-C.; Cheong, W.-C.; Radhakrishnan, K. Microelectron. Eng. 2003, 67-8, 306-311. (25) Yamasaki, K.; Juodkazis, S.; Matsuo, S.; Misawa, H. Appl. Phys. A 2003, 77, 371-3. Pfahler, J.; Harley, J.; Bau, H. Sens. Actuators A 1990, 21-23, 431434. (26) Huang, N. P.; Vo¨ro¨s, J.; De Paul, S. M.; Textor, M.; Spencer, N. D. Langmuir 2002, 18, 220-230.

PLL-g-PEG in Micro-Fluidic DeVices

Figure 1. (A) Schematic of the PDMS structure with one inlet (1) connected to an array of channels, the array of pillars and a second array of channels. Only one channel is connected to the main outlet (3), whereas the two other outlets (2 and 4) are connected to 19 channels each. The fluorescence was imaged at the entrance of the middle outlet channel (black square). (B) During functionalization with PLL-g-PEGbiotin, a PLL-g-PEGbiotin-containing buffer is introduced via outlet 3, whereas pure buffer is introduced into 1 and 4, which are then further connected to the outlet 2. The inset shows the solution flow from the middle channel to the side channels (see the Supporting Information). Later, PLL-g-PEG is introduced via inlet 1 into the whole device, thus covering the remaining areas. (C) After completed surface modification (40 min), outlets 2 and 4 are closed and the streptavidin sample is injected through the device from 1 to 3. The insets show the sample flows through the middle channel in the different cases. sponding to less than 20 nm. The density of surface charges after a similar oxidization of PDMS yielding bonding to glass was previously reported in the literature to be 1.3 10-2 C/m2.27 QCM-D Experiments. During a typical QCM-D measurement on polymer-coated crystals, the crystal was mounted in the instrument and stabilized in buffer. The crystal was exposed to the solution of PLL-g-PEG for 20 min and rinsed with 2 mL of buffer. Thereafter, the sensor surface was exposed to 0.5 mL of 5% human serum until the signal showed saturation, and then the system was rinsed with buffer. Micro-Fluidic Devices in PDMS. For the fabrication of the microfluidic devices, PDMS was cast onto an SU-8 template defined on a silicon wafer and coated with trichloro-(perfluorooctyl)silane in order to ease the release of the PDMS.28 Inlet and outlet holes were made through a glass substrate (Menzel-Gla¨zer, Germany) using sand blasting (Microetcher ERC, Danville Engineering Inc, CA). The substrates were cleaned with chloroform and the PDMS structures were rinsed in ethanol prior to being sealed on glass by oxygen plasma enhanced bonding,27 thus forming 40-µm-deep sealed structures. The devices were then filled with buffer immediately after bonding and mounted on a micro-fluidics connecting device made of PMMA with Teflon tubing and PEEK injection valves (Upchurch Scientific, USA). PLL-g-PEG, PLL-g-PEGbiotin, and streptavidin-Cy3 solutions were injected using 3 mL syringes (BectonDickinson, USA) and a syringe pump (Harvard Apparatus, USA) at 10 µL/min. (27) Duffy, D. C.; McDonald, J. C.; Schueller, O. J. A.; Whitesides, G. M. Anal. Chem. 1998, 70, 4974-4984. (28) Beck, M.; M. G.; Maximov, I.; Sarwe, E.-L.; Ling, T. G. I.; Keil, M.; Montelius, L. Microelectron. Eng. 2002, 61-62, 441-448.

Langmuir, Vol. 22, No. 24, 2006 10105 A schematic representation of the micro-fluidic device is shown in Figure 1A. It consists of an inlet channel (1) connected to a 10 × 4 mm2 array of cylindrical pillars with a diameter of 45 µm and height of 40 µm. There are 3 outlets (2, 3, 4). The middle outlet is connected to the array of pillars through a 45-µm-wide channel, whereas the two other outlets are connected to 19 channels of the same width. Using controlled flows of buffer through the inlet and the outlets of the array of pillars, the middle outlet channel of the device was coated with biotin using a 100 µL plug of PLL-gPEGbiotin solution (0.1 mg/mL), while buffer was passed through the rest of the chip (see figure legend). Subsequently, the device was filled with PLL-g-PEG solution and incubated for 10 min, causing the remaining uncovered surfaces of the device to be coated with PLL-g-PEG. Tubing, injection valves, as well as Eppendorf tubes and pipet tips used when preparing the dilutions of streptavidin were precoated with PLL-g-PEG by injection of and incubation in a 0.1 mg/mL solution for 10 min. The device was rinsed with buffer before the injection of the 100 µL sample of streptavidin-Cy3 at a flow rate of 10 µL/min, and was rinsed thoroughly with buffer before fluorescence imaging (Figure 1C). Fluorescence Microscopy. A Nikon Eclipse TE2000-U microscope (Nikon Co., Japan) equipped with an EMCCD camera (model DV887-BI, Andor Technology, U.K.) cooled to -50 °C was used in combination with a Nikon TRITC filter set to image the fluorescence of the streptavidin-Cy3. The fluorescence inside the outlet channel of the array of pillars was imaged in buffer through the PDMS using a 20× objective (NA ) 0.45, Plan Fluor, Nikon Co., Japan). The fluorescence intensity was quantified using ImageJ software (NIH, USA). The fluorescence intensity was measured over an area of 25 × 25 µm2 at the entrance of the outlet channel. The background signal measured in channels that were coated with PLL-g-PEG was subtracted. The streptavidin uptake at the surface of the outlet channel was derived by calibrating the fluorescence intensity in solution versus the concentration of streptavidin-Cy3, Briefly, 15-µm-deep PDMS channels sealed on glass were coated with PLL-g-PEG and filled with a dilution series of streptavidinCy3 in buffer. The channels were imaged in the same way as during the experiments. The fluorescence intensity inside the channel was plotted versus the number of fluorophores imaged per pixel. The data acquired were fitted to a linear function which was used as a calibration curve. Despite a high signal-to-noise ratio, even at the lowest intensities (see below), the calibration procedure yielded an error that varied from 12% at high coverage (>6 ng/cm2) to a factor 2 at low coverage (97%) for PMMA and Topas. Supported by the previously demonstrated ability of this copolymer to efficiently reduce protein adsorption on negatively charged oxides, such as glass, Nb2O5, SiO2, and TiO2, these results show that PLL-g-PEG appears as a generic surfacemodification candidate in µ-TAS consisting of traditionally used materials, providing that they are oxidized prior to surfacemodification. In fact, the purpose of the oxidation is 2-fold. First, it is intended to increase the surface charges of the polymer in order to achieve a better coverage by PLL-g-PEG. Indeed, the PLL-g-PEG coverage is somewhat lower on non-oxidized polymers, which, in turn, leads to higher nonspecific adsorption of serum (Figure 2). Compared with nonspecific adsorption of serum on the bare polymer, the reduction in nonspecific adsorption is in this case 50-60%. Second, the oxidation is used to reproduce the conditions that are often a prerequisite for efficient bonding during the assembly of micro-fluidic devices. It should also be noted that the slightly higher serum-protein uptake on PMMA and Topas correlates with a lower coverage of PLL-g-PEG, tentatively attributed to a lower density of negative charges induced by the plasma treatment on those polymers. It is also worthwhile to point out that equally good protein-resistant properties were obtained for all polymers when the substrates were dried after completed PLL-g-PEG modification and aged at 4 °C for at least a week. Localization of PLL-g-PEGbiotin for Streptavidin Capture. Inspired by these results, we demonstrate the potential of PLLg-PEG-based coatings for surface modification in micro-fluidic devices made of PDMS. To address, and specifically solve, the problem related to the depletion of analytes in micro-fluidic devices due to nonspecific protein adsorption, we used a device containing a dense array of high-aspect-ratio pillars as a test structure. This provides a large surface-to-volume ratio prior to the site of detection; the internal surface area is ∼0.98 cm2 and the internal volume is about 1 µL. Although the internal area without the pillars is only slightly lower (∼0.81 cm2), the presence of the pillars double the surface-to-volume ratio of the chip, taking the reduced volume due to the pillars into account. In addition, the device is also perfectly compatible with deterministic lateral particle separation,15 which constitutes one type of device in which a high surface-to-volume ratio is inevitable. To further address the issue of depletion due to nonspecific protein adsorption, the PLL-g-PEGbiotin-modified detection area inside the outlet channel was made more than 5 orders of magnitude smaller (6 × 10-6 cm2) than the total internal surface area (1 cm2). This was achieved by designing the device with two ports, one that serves as an inlet and the other as an outlet, respectively, whereas two additional ports serve to direct the flows inside the device during the coating of different regions of the device with PLL-g-PEG and PLL-g-PEGbiotin (see Figure 1A). By alternating the flow using these ports (see Figure 1B), the middle outlet (31) Ho¨o¨k, F.; Kasemo, B.; Nylander, T.; Fant, C.; Sott, K.; Elwing, H. Anal. Chem. 2001, 73, 5796-5804. (32) Glasma¨ster, K.; Larsson, C.; Hook, F.; Kasemo, B. J. Colloid Interface Sci. 2002, 246 (1), 40-47.

PLL-g-PEG in Micro-Fluidic DeVices

Langmuir, Vol. 22, No. 24, 2006 10107

Figure 4. Fluorescence intensity, derived from the fluorescence measured on a 25 × 25 µm2 portion of the outlet channel, versus concentration for four devices obtained at successively increasing concentrations and three devices tested at single concentrations. The fluorescence intensities were normalized to that measured at the highest concentration (160 pM). The solid line represents a least-squares linear fit (1 - R2 < 3 × 10-2). Error bars are displayed in cases when different devices were tested for overlapping concentrations.

Figure 3. Typical fluorescence micrograph of the array of pillars (- ∼45 µm, height 40 µm) (1), the area coated with PLL-g-PEGbiotin (2) at the entrance of the middle outlet channel (3) where the streptavidin-Cy3 is bound. The streptavidin-Cy3 concentration was 1.6 pM. The fluorescence intensities shown in Figure 4 are measured at the 25 × 25 µm frame displayed in the middle channel and the background level is measured in the side channels (4) connected to the other two outlets where the PDMS is coated with PLL-g-PEG. The inset shows the normalized fluorescence intensity (a.u.) after background subtraction vs the distance in µm inside the outlet channel. This type of data for varying concentrations are shown in Figure 5.

channel of the device was modified with PLL-g-PEGbiotin, whereas the rest of the micro-fluidic system was modified with PLL-g-PEG. This was followed by thorough rinsing in buffer and injection of 100 µL samples containing increasing concentrations of streptavidin through the device at a rate of 10 µL/min from left to right for a duration of 10 min33 (see Figure 1C). The fluorescence micrograph recorded after streptavidin injection at 1.6 pM and rinsing (Figure 3) provides clear evidence of successful localization of PLL-g-PEGbiotin to the outlet channel: streptavidin-Cy3 binds to the biotinylated surface inside the outlet channel (area 3), whereas no binding is observed on the surrounding areas coated with PLL-g-PEG (areas 1 and 4 in Figure 3 and the Supporting Information).34 To test the ability of PLL-g-PEG to reduce nonspecific protein adsorption and thus the depletion of the protein concentration, low-concentration streptavidin suspensions (∼1 pM to 160 pM) were injected and detected in a 25 × 25 µm2 area (see Figure 3) at the entrance of the channel. The fluorescence intensity originating from streptavidin binding obtained after 10 min is plotted versus concentration in Figure 4. These data clearly demonstrate that streptavidin can be readily detected with a contrast (net mean fluorescence intensity at the (33) Giving a speed of 0.1 m/s and a Reynolds number of 4 indicating laminar flow conditions. (34) The fluorescence background is measured in the side channels that are coated with PLL-PEG and were not exposed to the streptavidin sample (area 4 in Figure 3).

entrance of the channel compared to the fluorescence outside the channel cf. Figure 3) of >10 even at the lowest concentration of 1.6 pM (or 100 pg/mL). Note also that, since the measurements were made on a 100 µL sample exposed for 10 min, the total amount of streptavidin exposed to the device was, for the lowest concentration, only 160 attomol. Hence, although the device was intentionally designed to display a large surface-to-volume area prior to the site of detection, the limit of detection is only about 1 order of magnitude higher than what is, to our knowledge, the best sensitivity reported so far utilizing fluorescence as the mode of detection with a comparable sample volume (30 µL), although in that case the reaction time was more than 1 order of magnitude longer (2 h).35 Note also that a linear uptake vs concentration is observed, which is consistent with mass-transport limited binding. Since protein adsorption cannot be faster than the limit set by mass transport (diffusion), this allows for a comparison between these data and the highest possible streptavidin uptake. For the present design, the latter can be estimated using a model describing mass-transport-limited adsorption in a rectangular flow profile36

uptake )

( )

1 2 D 2F × 4 3 h2wl Γ 3

()

1/3

C×t

(1)

where C is the bulk concentration [g/m3], F the linear flow inside the channel [m3/s], D ) 6 × 10-11 m2/s the diffusion constant of streptavidin, and t ) 600 s the time during which the streptavidin is injected. w and h are the width (45 µm) and height (40 µm) of the channel, and l (25 µm) is the length of the portion of the channel imaged. Accordingly, the highest possible mass uptake at, for example, a concentration of 5 pM is 1.25 ng/cm2. Using a calibration curve of the fluorescence intensity vs the streptavidinCy3 density (see the Experimental Section), the mass uptake of streptavidin was calculated to 1.18 ( 0.78 ng/cm2 at this concentration. At higher concentrations, the error in the quantification of the streptavidin coverage decreases, whereas (35) Pawlak, M.; Schick, E.; Bopp, M. A.; Schneider, M. J.; Oroszlan, P.; Ehrat, M. Proteomics 2002, 2, 383-393. (36) Hlady, V.; Rickel, J.; Andrade, J. D. Colloids Surf. 1989, 34, 171-183.

10108 Langmuir, Vol. 22, No. 24, 2006

Marie et al.

sensing template to recognize streptavidin. Furthermore, the fact that the fluorescence intensity profiles in Figure 5 were recorded already after 10 min, i.e., at a time when only a fraction of saturated coverage is reached, means that they provide indirect information on the binding kinetics. Hence, the fact that the intensity profiles are essentially identical in serum and in pure buffer shows that the presence of serum in solution has an insignificant influence on the device performance also when applied to realistic biological samples, which is a key requirement in upcoming applications of lab on a chip systems.

Conclusions

Figure 5. Fluorescence intensity after background subtraction vs the distance in µm inside the outlet channel for increasing concentrations of streptavidin (5 pM via 20 pM and 80 pM to 160 pM). Also shown (inset) is a single injection of a serum suspension containing 4 pM streptavidin.

the good agreement with mass-transport limited binding is somewhat diminished. For example, at 10 pM, the obtained coverage of 1.3 ng/cm2 is expected (eq 1) already at a streptavidin concentration of 5 pM. If this depletion in streptavidin concentration is due to nonspecific adsorption on PLL-g-PEG inside the array of posts only (worst case scenario), this represents a nonspecific adsorption of streptavidin of