- Email: [email protected]
Printed protein microarrays on unmodified plastic substrates
Analytica Chimica Acta 671 (2010) 92–98
Contents lists available at ScienceDirect
Analytica Chimica Acta journal homepage: www.elsevier.com/locate/aca
Printed protein microarrays on unmodified plastic substrates Meike Moschallski, Johannes Baader, Oswald Prucker, Jürgen Rühe ∗ Laboratory for Chemistry and Physics of Interfaces, Department of Microsystems Engineering (IMTEK), University of Freiburg, Georges-Köhler-Allee 103, 79110 Freiburg, Germany
a r t i c l e
i n f o
Article history: Received 3 February 2010 Received in revised form 3 May 2010 Accepted 5 May 2010 Available online 13 May 2010 Keywords: Protein immobilization Plastic chips Microarray Polymer networks
a b s t r a c t A key challenge for the generation of protein microarrays is the immobilization of functional capture probe proteins at the chip surfaces. Here, a new concept for a single step production of protein microarrays to unmodified plastic substrates is presented. It is based on the printing of polymer/protein mixtures and the photochemical attachment of the obtained microstructures to the plastic chip surfaces. In the photochemical process three reactions occur simultaneously: transformation of the polymer into hydrogel dots, covalent binding of the forming gel to the substrate, and covalent immobilization of the proteins to the three-dimensional hydrogel scaffold. As an example we use anti-bovine serum albumin as a protein (anti-BSA) and a water swellable polymer network based on polydimethylacrylamide as a scaffold, which is photochemically crosslinked using benzophenone as a crosslinking agent. In one series of microarray experiments the probe density of the immobilized proteins was determined by incorporating fluorescence-labeled anti-BSA in the hydrogels. In a typical experiment, the number of immobilized probes was determined to 4 × 109 protein molecules per spot. In other experiments, the microarrays were brought into contact with fluorescently labeled BSA. In such analyses signal-to-noise values of more than 200 were obtained and about 9 × 107 antigen molecules were bound per spot. This demonstrates that in a very simple way microarrays with large amount of probes per spot can be realized and that antibodies immobilized in the printed hydrogels remain accessible and retain their functionality. © 2010 Published by Elsevier B.V.
1. Introduction Protein microarrays are of increasing interest in (bio)medical analytics because they allow a fast and parallel analysis of proteins on a chip and require only small amounts of sample solution [1–3]. Protein chips can be used to discover new proteins or to analyze their binding behavior and the formation of protein–protein complexes [4–6]. Due to the fast and highly parallel analysis process they are becoming more and more appreciated for applications in medical diagnostics [7,8]. In the following, the most commonly used surface modification reactions for the fabrication of protein microarrays are discussed, grouped into physical and chemical binding reactions. A very simple way to bind a protein to a surface is to adsorb it from solution. The relatively weak interaction between the individual groups involved in this process is usually compensated through the fact that a large number of groups in the protein become
Abbreviations: GOPS, 3-glycidoxypropyltrimethoxy-silane; His, histidine; MABP, methacryloyloxybenzophenone; MAGE, methacrylic acid glycidylester; NaPi, sodium phosphate buffer; NHS, N-hydroxysuccinimide; OWS, optical waveguide spectroscopy; PDMAA, polydimethylacrylamide; PMMA, polymethylmethacrylate; SSC, sodium citrate buffer; VPA, vinyl phosphonic acid. ∗ Corresponding author. Tel.: +49 761 203 7160; fax: +49 761 203 7162. E-mail address: [email protected] (J. Rühe). 0003-2670/$ – see front matter © 2010 Published by Elsevier B.V. doi:10.1016/j.aca.2010.05.008
involved in the process. This in turn causes frequently large changes in the protein conformation. One example for such an adsorption based process is the use of thin poly-l-lysine films adsorbed on glass. It has been reported, that under specific circumstances 20% of the printed antibodies were able to capture their specific antigen [9]. Another physical binding process occurs, when nickel ions contained in a self-assembled monolayer at the surface strongly bind to histidine (His) residues of proteins. If the proteins are His-tag labeled in advance, this induces an oriented immobilization of the proteins at the surface. Over 80% of the applied histidine-modified proteins were reported to retain their functionality on the chip [5]. Some commercially available chips use nitrocellulose membranes attached to the chip surface. An example are Whatman’s FASTTM slides [10]. Proteins adsorb inside the cellulose network and are attached to nitrate groups by hydrophobic interaction [11]. The high porosity of the cellulose increases its binding capacity, which leads to higher signals in the assay [12]. Apart from physical adsorption, proteins can also be chemically bound to the surface once appropriate functional groups are present at the surface. The latter are introduced to the surface for example through binding of thiols (in case of noble metal surfaces) or through silanes (in case of oxidic surfaces such as silicon or glass substrates). Various chemical functionalities present in silane monolayers have been reported to covalently bind proteins. Aldehyde groups, for example, form imine-bonds with the primary
M. Moschallski et al. / Analytica Chimica Acta 671 (2010) 92–98
amino groups of proteins [13]. NHS-ester silanes also bind proteins via their free amino or thiol side groups [12]. Epoxide groups of 3-glycidoxypropyltrimethoxy-silane (GOPS) react with primary amino groups of the protein forming a covalent bond. As they are rather simple to make GOPS modified chips are widely used for protein microarrays [12,14]. All three reactions immobilize the proteins in a dense (nonoriented) monolayer at the silanized surface. However, as all protein molecules have to bind directly to the surface, the binding capacities are intrinsically limited. One way to exceed this limit is to move from 2D to a 3D system and to use polymer networks instead of self-assembled monolayers. To overcome the limited binding capacity of planar surfaces, water swellable 3D surface coatings are introduced [14]. These hydrogels can be obtained through a crosslinking polymerization of the appropriate monomer in small pads on a glass surface in the cavities of a quartz master. After activating the hydrogel with glutaraldehyde, proteins printed on the pads are covalently bound to the gel. The signal of captured antigens can be altered either by reducing the density of crosslinks in the polymer network or by increasing the height of the pads [15]. As it would be well beyond the scope of this article to review the various surface chemistries for protein chips, which have been thoroughly reviewed elsewhere [2,4,10,12], we focus on some specific aspects here. One drawback of many employed methods is the complex multi-step fabrication processes required for the generation of the chips [16–19]. Most of the techniques start by modifying the complete chip surface with a protein binding layer. Many surface reactive groups require additional crosslinkers to form covalent bonds between the protein and surface [12]. Finally a microarray consisting of the desired proteins is printed onto the chip and the proteins are immobilized on the modified surfaces. Furthermore, some techniques (especially silanization) require an inert atmosphere to modify the surface in a reproducible way. As a result of this complex process, some coatings show background fluorescence, that significantly diminishes the signal-to-noise ratio of the fluorescence measurement [12]. Surfaces, onto which the probe proteins become physically attached (adsorbed) exhibit rather weak binding forces between protein and surfaces. Thus, the proteins might desorb again or be displaced under certain conditions during the assay (‘bleeding’), which could lead to erroneous results of the assay. However, what is almost as problematic as the bleeding problem is that the surface density of immobilized probe proteins is unknown in most cases and an internal reference is usually employed to monitor the quality of the protein microarrays. Here we present a new concept for the simple fabrication of protein microarrays on unmodified polymer substrates based on a one-step reaction. The proteins are immobilized in surface-attached hydrogel microstructures in high concentrations. For this purpose a terpolymer based on water soluble dimethylacrylamide (DMAA) with the photo-crosslinker methacryloyloxybenzophenone (MABP) and methacrylic acid glycidylester (MAGE) is synthesized. MAGE introduces epoxide side groups to the polymer that can bind proteins through chemical reactions in addition to the photochemical process, which is important at high protein and low benzophenone contents. As an alternative vinyl phosphonic acid (VPA) was introduced into the polymer instead of MAGE because VPA is a charged monomer which allows strong swelling of the surface-attached hydrogels in water. One of these terpolymers is printed together with the protein onto the surface of a plastic substrate. During a single UV-exposure step simultaneously both the surface modification and immobilization of the protein occur [20]. We describe in the following the preparation process and compare it to commonly used GOPS chips.
93
2. Materials and methods 2.1. Materials Polymethylmethacrylate (PMMA) slides were purchased from FORM-IN (Heitersheim, Germany) and cleaned in ethanol before printing. Glass slides from Schott (Jena, Germany) were thoroughly cleaned by dipping sequentially into 1 M sodium hydroxide solution, deionized water (twice), 2% (v/v) helmanex in deionized water in an ultrasonic bath, water (four times), ethanol (twice), and finally water (three times). Then the glass slides were dried in a stream of nitrogen and then silanized by dip-coating with GOPS (ABCR, Karlsruhe, Germany) dissolved in toluene. Bovine serum albumin (BSA) and the corresponding polyclonal antibody anti-BSA from rabbit were obtained from Sigma–Aldrich (Munich, Germany). The proteins were labeled with the fluorescent dye Dy647 with NHS-ester function from Dyomics (Jena, Germany) following the protocol for Cy-5 conjugation of proteins by Amersham Biosciences (Freiburg, Germany). Unbound dye was removed by centrifugation in a Microcon centrifugal filter device with a 10 kDa cutoff obtained from Millipore (Billerica, MA, USA). Finally the average ratio of dye molecules per protein was determined by UV–VIS spectroscopy. 1 M aqueous solutions of sodium dihydrogen phosphate (NaH2 PO4 ) and disodium hydrogen phosphate (Na2 HPO4 ) both from VWR International GmbH (Darmstadt, Germany) were prepared and mixed to yield a 1 M sodium phosphate (NaPi) buffer, which was titrated with the 1 M NaH2 PO4 solution until a pH of 7 was reached. For washing and incubation of the protein chips during the immunoassay, 0.1% (v/v) Tween 20 purchased from VWR International GmbH was added to the buffer. This surfactant promotes washing of the protein arrays and minimizes unspecific adsorption especially of labeled proteins on the chip surface. The terpolymer polydimethylacrylamide with 10% methacrylic acid glycidylester and 1% methacryloyloxybenzophenone (PDMAA–10%MAGE–1%MABP) was synthesized in a free radical polymerization of the distilled monomers dissolved in dimethylformamide with AIBN as an initiator as reported before [21]. A content of 10% MAGE was chosen because it provided sufficient protein binding capacity but was still soluble in water or phosphate buffer. In analogy polydimethylacrylamide with 5% vinyl phosphonic acid (VPA) and 1% MABP (PDMAA–5%VPA–1%MABP) was synthesized in a toluene solution. For characterization the infrared, nuclear magnetic resonance and UV spectra of the polymers had been recorded showing that the polymerization reactions were conducted in proper form. 2.2. Instrumentation The swelling behavior of the surface-attached hydrogel was analyzed by optical waveguide spectroscopy (OWS). For this purpose a self-made surface plasmon resonance spectrometer equipped with a prism made from lanthanum glass (LaSFN9) in the Kretschman configuration was used. Its backside was coated with a 50 nm thin film of evaporated gold and a 20 nm thin film of polystyrene. Either an Omnigrid 100 contact printer from Geneworx AG (Oberhaching, Germany) or a contactless printer (SciFlexarrayer S5 from Scienion AG (Berlin, Germany)) were used to produce the microarrays. Fluorescence images of the obtained microarrays were taken using a Biodetect® 645 from Genescan Europe AG (Freiburg, Germany). 2.3. Procedures On the prism for OWS the polystyrene was dip-coated from an ethyl acetate solution (5 mg mL−1 ) at a withdrawal speed of
94
M. Moschallski et al. / Analytica Chimica Acta 671 (2010) 92–98
60 mm min−1 . On top of this adhesion layer the polymer of interest was dip-coated from an ethanol solution (100 mg mL−1 ) at the same withdrawal speed and then photo-crosslinked (254 nm, 500 mJ cm−2 ). The hydrogel under investigation was swollen repeatedly by filling the fluidic chamber. The thickness was determined with water or buffer, then the polymer layer was dried by flowing nitrogen through the empty chamber and measured in both states. The simulation curves were calculated using the software WinspallTM that uses the Fresnel equations to determine the reflectivity spectrum for a model consisting of the sequence of layers on the glass with their corresponding thicknesses and complex refractive indices. The latter parameters were varied to adapt the simulated spectrum to the measured data. Microarrays were printed from a mixed solution of polymer (PDMAA–10%MAGE–1%MABP) and antibody in 100 mM NaPi buffer at pH 7. Volumes of 400 pL up to 6.4 nL of the solution were dispensed per spot using the above mentioned microarrayers. A volume of 400 ± 10 pL dispensed per droplet was measured in images obtained with the help of a stroboscope camera during spotting. Higher spot volumes were achieved by printing up to 16 of these droplets on the same position. Usually the dry chips were irradiated with UV-light at a wavelength of 254 nm with an energy dose of 500 mJ cm−2 to crosslink the polymer and immobilize the proteins before the assay. Capture immunoassays were performed with fluorescently labeled antigen (BSA). For this, the chips were washed in 30 mM sodium citrate buffer (2× SSC) with 0.1% (v/v) sodium dodecylsulfate (SDS) and 0.05% (v/v) Tween 20 and rinsed with deionized water. Then they were incubated with 100 nM labeled BSA dissolved in phosphate buffered salt solution with Tween 20 for 3 h in a closed chamber in order to allow for sufficient time for the formation of the protein complex. Finally, the chips were washed again in the same way as in the first step, before they were dried in a stream of nitrogen and the fluorescence intensity was measured.
Fig. 1. Printing microarrays of mixed polymer protein solutions and immobilization during UV-crosslinking. (A) Macroscopic view: (1) solutions of polymer and protein in buffer are mixed and unmodified plastic slides are provided, (2) a microarray of the solution is printed on the chip by contact printing and (3) the chip is irradiated with UV-light. (B) Molecular point of view (symbols not to scale): (1) epoxide side groups of the polymer react with primary amino groups of the protein forming complexes in solution, (2) droplets of protein and polymer form at the surface and (3) the photoreaction of the benzophenone moieties in the polymer crosslinks the polymer, attaches it to the surface, and simultaneously immobilizes the proteins in the polymer network.
proteins and at the same time efficiently crosslink the hydrogel. For this reason epoxide groups are added to the polymer for protein immobilization. However, these groups can interact only with the proteins and do not contribute to the hydrogel crosslinking. 3.2. Hydrogel characterization
3. Results and discussions 3.1. Single step microarray production The polymer protein chips were produced on commercially available plastic slides as illustrated in Fig. 1 from a macroscopic point of view as well as seen at a molecular level. In this study almost exclusively PMMA slides were used. Other plastic substrates, i.e. polystyrene or hydrophilized Topas, can also be favorably employed as long as they exhibit low background fluorescence and can be wetted by the printing solution (data not shown). To generate this solution, first a mixture of the polymer in water and the protein in buffer was prepared. After a short time of incubation to account for the required reaction time of the epoxide side groups of the polymer with the primary amino groups of the proteins to form protein–polymer complexes [22], the thus obtained mixture was printed directly onto the unmodified polymer substrates by using a contact printer. Finally, the chips were UV-irradiated by flood exposure, which induces the radical-based photoreaction of the benzophenone moieties in the polymer. The photoprocess crosslinks the polymer to a network, covalently attaches it to the polymer substrate and simultaneously immobilizes the proteins in the surface-attached network as illustrated in Fig. 1. The covalent immobilization of the proteins inside the hydrogel both through the epoxide groups as well as through the radical reaction with the benzophenone moieties in the polymer minimizes bleeding of the proteins out of the polymer network during the immunoassays. As the benzophenone content is only 1% to allow high swelling factors of the hydrogel, it is not enough to bind high amounts of
A critical question which needs to be addressed before any protein is attached to the hydrogels, is how strongly the hydrogel is swollen under assay conditions, i.e. in the presence of an aqueous buffer. Swelling is a key for the performance of the microarray. Unfortunately many studies which use hydrogels do not report on swelling parameters. This is even more important on the background that surface-attached hydrogels exhibit a significantly different swelling as the identical free (i.e. not surface-attached) hydrogel [20]. This is especially crucial because the swelling of the hydrogel will strongly influence the accessibility of protein molecules located within the hydrogel. A method to elucidate this question is optical waveguide spectroscopy, which allows determining simultaneously the thickness and the refractive index of an ultrathin polymer layer. To this a hydrogel layer of PDMAA–10%MAGE–1%MABP was dip-coated on the prism for the OWS measurement (see Section 2.3). A dry film thickness of 1 m was determined after extraction in water and subsequent drying in a stream of nitrogen. Fig. 2 shows the reflectivity spectra of the polymer film in the dry state as well as that of the film swollen in deionized water. The simulated reflection curves, which were calculated according to the Fresnel equations, are adapted to the measured data to determine the film thickness and the refractive index of the swollen polymer. As the waveguide spectra show a series of (orthogonal) resonance signals, the system is overdetermined and both parameters can be calculated simultaneously from the same measurement. Swollen and dry thickness was measured for at least ten cycles of swelling and drying. The linear swelling factor of the polymer net-
M. Moschallski et al. / Analytica Chimica Acta 671 (2010) 92–98
95
Fig. 2. Reflectivity spectra of a polymer film measured by optical waveguide spectroscopy with polarized light on a LaSFN9 prism. (1) Dry film d0 = 1.1 m; n0 = 1.51 (refractive index); (2) film swollen in deionized water d = 3.9 m; n = 1.39; the solid lines are calculated according to the Fresnel equations.
of the degree of swelling of a free gel, which is in good agreement with the experimental data for both hydrogel compositions used here. Similar swelling factors have been determined for PDMAA–5%VPA–1%MABP [21]. In the following experiments we printed spots of the polymers mixed with antibodies onto plastic slides and showed their use for immunoassays. As the areas of the hydrogel spots had a diameter of 100–600 m with a dry film thickness between 20 and 100 nm, these hydrogel structures can be regarded as thin surface-attached hydrogel films and are expected to show the same swelling behavior as the films measured by optical waveguide spectroscopy, although the microspot structures could not be measured in a swollen state. Fig. 3. The linear swelling factor of the surface-attached hydrogel film plotted as a function of the NaPi buffer concentration.
work was calculated as the ratio of the swollen versus the dry layer thickness. It was analyzed for a series of solutions containing NaPi buffer between 1 M and 100 mM. Fig. 3 illustrates, that the swelling ratio d/d0 decreases linearly with the buffer concentration c and can be approximated by the following formula: d = 3.7 − 1.3c d0
(1)
Here the summand 3.7 describes the degree of swelling of the hydrogel in salt-free environment and the factor 1.3 represents the influence of added salt onto the swelling behavior. However, even at high salt concentrations the hydrogel is strongly swollen, i.e. even at the solubility limit of the buffer, which is about 1 mol L−1 for the NaPi buffer system, it is still swollen by a factor of 2.4, which means, that the layer still contains about 60% (v/v) of water. The slight shrinking of the polymer layer with increasing salt concentration can be explained as a “salting out effect”. That means the solvent quality of the aqueous solution is slightly reduced at high salt concentrations—an effect well known for the swelling of free hydrogels in contact with salt solutions [21]. In the following we use in all experiments salt concentrations of 200 mM NaPi which are typically employed in standard immunoassays, so that the surface-attached hydrogels are swollen by roughly a factor of 3.4. It should be noted, that due to the geometric constraints imposed by the surface-attachment, very thin surface-attached gels swell significantly different (i.e. less strongly) than nonattached but otherwise identical free gels [20,21]. Application of the Flory–Rehner theory of the swelling of gels to such a twodimensional system suggests that the degree of swelling of such a surface-attached gel is expected to be roughly the square root
3.3. Determination of protein immobilization efficiency In order to quantify the number of antibodies immobilized in the polymer spots, arrays of fluorescence-labeled antibodies were printed on GOPS silanized glass and on PMMA slides. PDMAA–5%VPA–1%MABP was used as the hydrogel to immobilize anti-BSA. Only 2% of the printed antibodies were labeled in order to prevent quenching of the fluorophores at the high protein concentrations under investigation. The fluorescence intensity integrated over each spot measured immediately after printing defined volumes of a set of protein concentrations was used to calibrate the signal of the corresponding known amounts of labeled proteins printed per spot. Since the signal increased linearly with antibody concentration, quenching was excluded (see also Supplemental Information). As a reference experiment antibodies were also printed without polymer to show, on which substrate the proteins form monolayers. The chips were stored at 6 ◦ C over night in order to assure that the proteins have enough time to bind to adjacent epoxide groups at the chip surface. After the over night reaction the protein chips were washed and the fluorescence signals were measured a second time. The density of immobilized antibodies in each single spot was determined as the quotient of fluorescence intensities after washing and before washing multiplied by the amount of antibodies dispensed per spot. Fig. 4 illustrates how antibodies are immobilized onto monolayers (a) and in the hydrogel (b) on both silanized glass and plastic slides. Proteins printed under otherwise identical conditions, but without polymer bind to GOPS silanized glass chips and form a monolayer. In contrast to this, they cannot bind to a PMMA slide and most antibodies are washed off the chip (see Fig. 4a). The hydrogel in contrast allows attaching the antibodies in a more three-dimensional way on both substrates as demonstrated in Fig. 4b.
96
M. Moschallski et al. / Analytica Chimica Acta 671 (2010) 92–98
Fig. 5. Number N of antibodies (anti-BSA) immobilized into a PDMAA-based polymer network according to fluorescence measurements after washing in 100 mM NaPi buffer with 0.1% Tween 20. The average values of four chips are shown printed with four identical spots of each probe. Error bars indicate the standard deviation of the complete set of data (n = 16).
Fig. 4. Schematic illustration and fluorescence images of antibody immobilization on two different surfaces: (1) Glass slide silanized with GOPS. (2) Plastic slide. On each surface: (a) monolayers of antibodies were printed and (b) antibodies were immobilized in hydrogel. The fluorescence images on the right corresponding to the schematics on the left are taken after washing from a series of solutions with varied concentrations of labeled antibodies (a) without polymer (monolayer) and (b) mixed with 1 mg mL−1 PDMAA–5%VPA–1%MABP hydrogel printed on both substrates.
The fluorescence images of antibody dilution series printed in a microarray format (Fig. 4b) clearly show that antibodies are immobilized in the hydrogel regardless, whether the hydrogel is attached to glass or to PMMA surfaces. As far as antibody monolayers are concerned, a reasonable extent of binding is only observed on epoxide silanized glass, but not on plastic surfaces, where hardly any fluorescence is detected (Fig. 4-2a). From all hydrogel samples well defined, round spots with a constant diameter of 400 m are obtained, when approximately 5 nL drops are printed. In contrast to this, the monolayer spots are smeared out, because the antibodies are not strongly bound in the monolayers and some of them bleed out during washing and become readsorbed in an adjacent location. This occurs preferably parallel to the direction of withdrawal of the chips from the washing solution. In one set of experiments, a series of antibody solutions was printed with concentrations ranging from 0 to 2000 mg L−1 . The fluorescence intensity increases with antibody concentration starting at a low intensity of 50 mg L−1 antibody, which can hardly be seen, up to very high intensities for 2 g L−1 . From these measured intensities the density of immobilized antibodies per spot can be determined by using the calibration deter-
mined from the measurement of labeled antibody spots before washing. The number of antibodies immobilized in the three-dimensional hydrogel scales linearly with their concentration in the print solution and does not reach a limit within the investigated range. In contrast the number of antibodies immobilized in a monolayer seems to reach asymptotically a limit at around 5 × 109 per spot, which is ten thousand antibodies per m2 or one antibody on 100 nm2 . 3.4. Functionality of immobilized antibodies To investigate the ability of the immobilized antibodies to capture antigen from solution, the samples described in Fig. 5 were exposed to labeled antigen in a form of a capture immunoassay according to the protocol presented in the experimental section. Fig. 6 illustrates how different densities of immobilized antibodies in such hydrogels can bind labeled antigens from solution. If one looks at these values, it is important to be aware of the background signal. As far as this is concerned, two different parameters are important. On the one hand it is important to know how much protein adsorbs unspecifically to the hydrogel, on the other hand it should be known, how the protein adsorbs unspecifically to the chip surface, which is not coated with hydrogel. On control spots, where no antibody was immobilized in the hydrogel, no appreciable fluorescence signal could be measured, which proves, that the labeled proteins adsorb, if at all, only very weakly unspecifically onto the hydrogel itself [23]. Between the spots also very low fluorescence intensity was measured. When the chip was washed with a washing
Fig. 6. Schematic illustration and intensity plot of the fluorescence intensity of a direct immunoassay with labeled antigens and immobilized antibodies on GOPS silanized glass slides in hydrogel. As a control also hydrogel spots were generated, which contained no protein (0 mg L−1 ). See also Fig. 7 for a quantitative analysis.
M. Moschallski et al. / Analytica Chimica Acta 671 (2010) 92–98
97
4. Conclusions
Fig. 7. Number of captured BSA antigens N(BSA) per spot determined from fluorescence measurement after a capture immunoassay with labeled BSA (100 nM) on anti-BSA immobilized both in PDMAA–5%VPA–1%MABP and in monolayers on a GOPS silanized glass slide as a function of the number of antibodies immobilized N(anti-BSA) per spot. The data labels specify the antibody concentrations in the spotted solution. The average values of four chips are shown printed with four identical spots of each probe. Error bars indicate the standard deviation of the complete set of data.
buffer, which contained the surfactant Tween 20, a signal-to-noise ratio between the spots and the surrounding background of around 220 was observed. This very high signal to background ratio especially on unmodified PMMA is reached because this surface does not exhibit protein binding groups. Therefore the simple addition of the surfactant that prevents hydrophobic interaction of the protein with the polymer surface efficiently “blocks” unspecific adsorption to this background. Fig. 7 shows the number of captured BSA antigens versus the number of immobilized antibodies per spot. The number of captured antigen increases linearly with antibody concentration in the hydrogel spots over the whole concentration range up to a captured amount of around 108 antibodies/spot. In contrast to this the amount of antigen captured by an antibody monolayer does not increase linearly and asymptotically reaches a limit at around 4 × 107 BSA/spot, so that saturation is already reached at a printed antibody concentration of 0.5 g L−1 . The linear behavior observed in the hydrogel system implies that a constant fraction of the antibodies is still active after immobilization, while the ratio of active antibodies in the monolayer decreases for antibody concentrations above 0.25 g L−1 . One question, which needs to be addressed, is why the number of BSA molecules captured is much smaller than that of the immobilized anti-BSA. It has been reported before, that the immobilization of antibodies in monolayers can result in a decrease in affinity or binding capacity [24]. This has been explained by steric hindrance of the antigen’s access to the paratopes at the antibody. In agreement with this in other work it has been shown, that with increasing antibody density on the surface, their efficiency to bind large antigens decreases significantly [25]. Especially, when the antibodies are randomly oriented, as in the case of amino groups binding to epoxide silane surfaces, some of the paratopes may be too close to the surface to be accessible for antigens [26]. The maximum antibody density measured in the monolayer is comparable to that reported by Kusnezow and coworkers with twenty thousand per m2 [27]. They assumed at least half of the antibodies inaccessible, but did not give quantitative data. When the results shown in Fig. 7 are analyzed, it has to be considered, that the number of capture proteins (anti-BSA) in the hydrogels and the monolayer are the same. The higher binding capacity in the hydrogels seem to indicate, that more paratopes are accessible to bind antigens, if the antibodies are immobilized in the flexible, highly swollen hydrogel compared to immobilization in a dense monolayer. This indicates that the immobilization reaction of the rare benzophenone moieties in the hydrogel does not affect the biofunctionality of the antibodies.
The benzophenone containing hydrogels allow a simple photochemical immobilization of antibodies on polymer substrates. Since the radical-based photoreaction of benzophenone is not specific, various polymer protein mixtures can be attached to essentially any polymer substrate provided, that are wetted by the aqueous print solution. The only additional prerequisite is that the substrate exhibits a sufficiently low background fluorescence to allow facile read-out. Hence, this process for protein chips opens the road to the use of simple polymer substrates that can be fabricated by injection molding in high numbers at extremely low cost. On top of it, our strategy enables to produce protein chips in a very simple way. The polymer and protein solutions are mixed and printed on (clean) unmodified polymer substrates followed by a short UV-exposure to induce the photoreaction. The polymer protein chips can be printed under ambient conditions. There is neither the need for an inert atmosphere nor for a specific chemical laboratory environment for chip production, which makes the process very simple and fast. Since the chosen hydrogel binds proteins via epoxide and benzophenone groups, this approach is not limited to antibody arrays, but can bind many kinds of proteins including enzymes, which is currently elucidated in other studies. The resolution limit of the process depends only on the printer used and the wettability of the chip surface. Hence, the 400 pL droplets printed here gave spots of about 150 m in diameter. If desired, even smaller hydrogel structures can be produced by photolithography if the protein–polymer mixture is UV-irradiated through a fine grid and then non-exposed areas of polymer and protein are washed off. It is a very interesting feature of this method of chip fabrication, that it permits a detailed quality control of the chip production process. As it is essentially a one-step process, the measurement of the volume of a dispensed droplet is sufficient to determine the number of immobilized probe proteins, because they are covalently bound by both benzophenone and epoxide side groups of the polymer and therefore cannot bleed out during the assay. Despite the simple production process proteins immobilized in these hydrogels show a high accessibility to capture antigens from a sample solution. We will show in a following communication, that the described method can be used to generate chips with extremely high signal-to-noise ratios and accordingly high sensitivity. Conflict of interest statement The authors declare no financial or commercial conflict of interest. Acknowledgment We thank Natalia Schatz for valuable technical support. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.aca.2010.05.008. References [1] T. Kodadek, Chemistry & Biology 8 (2001) 105. [2] M.F. Templin, D. Stoll, J.M. Schwenk, O. Pötz, S. Kramer, T.O. Joos, Proteomics 3 (2003) 2155. [3] S. Sauer, B.M.H. Lange, J. Gobom, L. Nyarsik, H. Seitz, H. Lehrach, Nature Reviews Genetics 6 (6) (2005) 465. [4] G. MacBeath, Nature Genetics 32 (2002) 526.
98
M. Moschallski et al. / Analytica Chimica Acta 671 (2010) 92–98
[5] H. Zhu, M. Bilgin, R. Bangham, D. Hall, A. Casamayor, P. Bertone, N. Lan, R. Jansen, S. Bidlingmaier, T. Houfek, T. Mitchell, P. Miller, R.A. Dean, M. Gerstein, M. Snyder, Science 293 (2001) 2101. [6] P. Bertone, M. Snyder, FEBS Journal 272 (21) (2005) 5400. [7] J. Qiu, J. Madoz-Gurpide, D.E. Misek, R. Kuick, D.E. Brenner, G. Michailidis, B.B. Haab, G.S. Omenn, S. Hanash, Journal of Proteome Research 3 (2004) 261. [8] P. Wagner, R. Kim, Current Drug Discovery 5 (2002) 23. [9] B.B. Haab, M.J. Dunham, P.O. Brown, Genome Biology 2 (2001) 13. [10] B.A. Stillman, J.L. Tonkinson, BioTechniques 29 (2000) 630. [11] S. Oehler, R. Alex, A. Barker, Analytical Biochemistry 268 (1999) 330. [12] W. Kusnezow, A. Jacob, A. Walijew, F. Diehl, J.D. Hoheisel, Proteomics 3 (2003) 254. [13] G. MacBeath, S. Schreiber, Science 289 (2000) 1760. [14] W. Kusnezow, J.D. Hoheisel, Journal of Molecular Recognition 16 (2003) 165. [15] O. Prucker, C.A. Naumann, J. Ruhe, W. Knoll, C.W. Frank, Journal of the American Chemical Society 121 (1999) 8766. [16] D.-N. Kim, W. Lee, W.-G. Koh, Analytica Chimica Acta 609 (2008) 59. [17] W. Lee, D. Choi, Y. Lee, D.-N. Kim, J. Park, W.-G. Koh, Sensors and Actuators B: Chemical 129 (2008) 841.
[18] A.L. Hook, H. Thissen, N.H. Voelcker, Biomacromolecules (2009). [19] D.G. Swan, M.J. Swanson, www.freepatentsonline.com, vol. 09/521545, SurModics, Inc., Eden Prairie, MN, United States, 2004. [20] R. Toomey, D. Freidank, J. Ruhe, Macromolecules 37 (2004) 882. [21] T. Neumann, Wiederverwendbare 3D-Polymer-DNA-Chips zur Echtzeitanalyse und kompartimentierten Paralleldiagnostik, dissertation, University of Freiburg, 2006. [22] S. Golze, Oberflächengebundene Polymermonolagen für die Herstellung von DNA-Chips, dissertation, University of Mainz, 2001. [23] B. Berchtold, Surface attached polymers for the reendothelialization of porcine heartvalves, dissertation, University of Freiburg, 2005. [24] T.M. Spitznagel, J.W. Jacobs, D.S. Clark, Enzyme and Microbial Technology 15 (1993) 919. [25] Y. Kwon, Z. Han, E. Karatan, M. Mrksich, B.K. Kay, Analytical Chemistry 76 (2004) 5713. [26] R.A. Vijayendran, D.E. Leckband, Analytical Chemistry 73 (3) (2001) 471. [27] W. Kusnezow, Y.V. Syagailo, S. Ruffer, K. Klenin, W. Sebald, J.D. Hoheisel, C. Gauer, I. Goychuk, Proteomics 6 (2006) 794.
Contents lists available at ScienceDirect
Analytica Chimica Acta journal homepage: www.elsevier.com/locate/aca
Printed protein microarrays on unmodified plastic substrates Meike Moschallski, Johannes Baader, Oswald Prucker, Jürgen Rühe ∗ Laboratory for Chemistry and Physics of Interfaces, Department of Microsystems Engineering (IMTEK), University of Freiburg, Georges-Köhler-Allee 103, 79110 Freiburg, Germany
a r t i c l e
i n f o
Article history: Received 3 February 2010 Received in revised form 3 May 2010 Accepted 5 May 2010 Available online 13 May 2010 Keywords: Protein immobilization Plastic chips Microarray Polymer networks
a b s t r a c t A key challenge for the generation of protein microarrays is the immobilization of functional capture probe proteins at the chip surfaces. Here, a new concept for a single step production of protein microarrays to unmodified plastic substrates is presented. It is based on the printing of polymer/protein mixtures and the photochemical attachment of the obtained microstructures to the plastic chip surfaces. In the photochemical process three reactions occur simultaneously: transformation of the polymer into hydrogel dots, covalent binding of the forming gel to the substrate, and covalent immobilization of the proteins to the three-dimensional hydrogel scaffold. As an example we use anti-bovine serum albumin as a protein (anti-BSA) and a water swellable polymer network based on polydimethylacrylamide as a scaffold, which is photochemically crosslinked using benzophenone as a crosslinking agent. In one series of microarray experiments the probe density of the immobilized proteins was determined by incorporating fluorescence-labeled anti-BSA in the hydrogels. In a typical experiment, the number of immobilized probes was determined to 4 × 109 protein molecules per spot. In other experiments, the microarrays were brought into contact with fluorescently labeled BSA. In such analyses signal-to-noise values of more than 200 were obtained and about 9 × 107 antigen molecules were bound per spot. This demonstrates that in a very simple way microarrays with large amount of probes per spot can be realized and that antibodies immobilized in the printed hydrogels remain accessible and retain their functionality. © 2010 Published by Elsevier B.V.
1. Introduction Protein microarrays are of increasing interest in (bio)medical analytics because they allow a fast and parallel analysis of proteins on a chip and require only small amounts of sample solution [1–3]. Protein chips can be used to discover new proteins or to analyze their binding behavior and the formation of protein–protein complexes [4–6]. Due to the fast and highly parallel analysis process they are becoming more and more appreciated for applications in medical diagnostics [7,8]. In the following, the most commonly used surface modification reactions for the fabrication of protein microarrays are discussed, grouped into physical and chemical binding reactions. A very simple way to bind a protein to a surface is to adsorb it from solution. The relatively weak interaction between the individual groups involved in this process is usually compensated through the fact that a large number of groups in the protein become
Abbreviations: GOPS, 3-glycidoxypropyltrimethoxy-silane; His, histidine; MABP, methacryloyloxybenzophenone; MAGE, methacrylic acid glycidylester; NaPi, sodium phosphate buffer; NHS, N-hydroxysuccinimide; OWS, optical waveguide spectroscopy; PDMAA, polydimethylacrylamide; PMMA, polymethylmethacrylate; SSC, sodium citrate buffer; VPA, vinyl phosphonic acid. ∗ Corresponding author. Tel.: +49 761 203 7160; fax: +49 761 203 7162. E-mail address: [email protected] (J. Rühe). 0003-2670/$ – see front matter © 2010 Published by Elsevier B.V. doi:10.1016/j.aca.2010.05.008
involved in the process. This in turn causes frequently large changes in the protein conformation. One example for such an adsorption based process is the use of thin poly-l-lysine films adsorbed on glass. It has been reported, that under specific circumstances 20% of the printed antibodies were able to capture their specific antigen [9]. Another physical binding process occurs, when nickel ions contained in a self-assembled monolayer at the surface strongly bind to histidine (His) residues of proteins. If the proteins are His-tag labeled in advance, this induces an oriented immobilization of the proteins at the surface. Over 80% of the applied histidine-modified proteins were reported to retain their functionality on the chip [5]. Some commercially available chips use nitrocellulose membranes attached to the chip surface. An example are Whatman’s FASTTM slides [10]. Proteins adsorb inside the cellulose network and are attached to nitrate groups by hydrophobic interaction [11]. The high porosity of the cellulose increases its binding capacity, which leads to higher signals in the assay [12]. Apart from physical adsorption, proteins can also be chemically bound to the surface once appropriate functional groups are present at the surface. The latter are introduced to the surface for example through binding of thiols (in case of noble metal surfaces) or through silanes (in case of oxidic surfaces such as silicon or glass substrates). Various chemical functionalities present in silane monolayers have been reported to covalently bind proteins. Aldehyde groups, for example, form imine-bonds with the primary
M. Moschallski et al. / Analytica Chimica Acta 671 (2010) 92–98
amino groups of proteins [13]. NHS-ester silanes also bind proteins via their free amino or thiol side groups [12]. Epoxide groups of 3-glycidoxypropyltrimethoxy-silane (GOPS) react with primary amino groups of the protein forming a covalent bond. As they are rather simple to make GOPS modified chips are widely used for protein microarrays [12,14]. All three reactions immobilize the proteins in a dense (nonoriented) monolayer at the silanized surface. However, as all protein molecules have to bind directly to the surface, the binding capacities are intrinsically limited. One way to exceed this limit is to move from 2D to a 3D system and to use polymer networks instead of self-assembled monolayers. To overcome the limited binding capacity of planar surfaces, water swellable 3D surface coatings are introduced [14]. These hydrogels can be obtained through a crosslinking polymerization of the appropriate monomer in small pads on a glass surface in the cavities of a quartz master. After activating the hydrogel with glutaraldehyde, proteins printed on the pads are covalently bound to the gel. The signal of captured antigens can be altered either by reducing the density of crosslinks in the polymer network or by increasing the height of the pads [15]. As it would be well beyond the scope of this article to review the various surface chemistries for protein chips, which have been thoroughly reviewed elsewhere [2,4,10,12], we focus on some specific aspects here. One drawback of many employed methods is the complex multi-step fabrication processes required for the generation of the chips [16–19]. Most of the techniques start by modifying the complete chip surface with a protein binding layer. Many surface reactive groups require additional crosslinkers to form covalent bonds between the protein and surface [12]. Finally a microarray consisting of the desired proteins is printed onto the chip and the proteins are immobilized on the modified surfaces. Furthermore, some techniques (especially silanization) require an inert atmosphere to modify the surface in a reproducible way. As a result of this complex process, some coatings show background fluorescence, that significantly diminishes the signal-to-noise ratio of the fluorescence measurement [12]. Surfaces, onto which the probe proteins become physically attached (adsorbed) exhibit rather weak binding forces between protein and surfaces. Thus, the proteins might desorb again or be displaced under certain conditions during the assay (‘bleeding’), which could lead to erroneous results of the assay. However, what is almost as problematic as the bleeding problem is that the surface density of immobilized probe proteins is unknown in most cases and an internal reference is usually employed to monitor the quality of the protein microarrays. Here we present a new concept for the simple fabrication of protein microarrays on unmodified polymer substrates based on a one-step reaction. The proteins are immobilized in surface-attached hydrogel microstructures in high concentrations. For this purpose a terpolymer based on water soluble dimethylacrylamide (DMAA) with the photo-crosslinker methacryloyloxybenzophenone (MABP) and methacrylic acid glycidylester (MAGE) is synthesized. MAGE introduces epoxide side groups to the polymer that can bind proteins through chemical reactions in addition to the photochemical process, which is important at high protein and low benzophenone contents. As an alternative vinyl phosphonic acid (VPA) was introduced into the polymer instead of MAGE because VPA is a charged monomer which allows strong swelling of the surface-attached hydrogels in water. One of these terpolymers is printed together with the protein onto the surface of a plastic substrate. During a single UV-exposure step simultaneously both the surface modification and immobilization of the protein occur [20]. We describe in the following the preparation process and compare it to commonly used GOPS chips.
93
2. Materials and methods 2.1. Materials Polymethylmethacrylate (PMMA) slides were purchased from FORM-IN (Heitersheim, Germany) and cleaned in ethanol before printing. Glass slides from Schott (Jena, Germany) were thoroughly cleaned by dipping sequentially into 1 M sodium hydroxide solution, deionized water (twice), 2% (v/v) helmanex in deionized water in an ultrasonic bath, water (four times), ethanol (twice), and finally water (three times). Then the glass slides were dried in a stream of nitrogen and then silanized by dip-coating with GOPS (ABCR, Karlsruhe, Germany) dissolved in toluene. Bovine serum albumin (BSA) and the corresponding polyclonal antibody anti-BSA from rabbit were obtained from Sigma–Aldrich (Munich, Germany). The proteins were labeled with the fluorescent dye Dy647 with NHS-ester function from Dyomics (Jena, Germany) following the protocol for Cy-5 conjugation of proteins by Amersham Biosciences (Freiburg, Germany). Unbound dye was removed by centrifugation in a Microcon centrifugal filter device with a 10 kDa cutoff obtained from Millipore (Billerica, MA, USA). Finally the average ratio of dye molecules per protein was determined by UV–VIS spectroscopy. 1 M aqueous solutions of sodium dihydrogen phosphate (NaH2 PO4 ) and disodium hydrogen phosphate (Na2 HPO4 ) both from VWR International GmbH (Darmstadt, Germany) were prepared and mixed to yield a 1 M sodium phosphate (NaPi) buffer, which was titrated with the 1 M NaH2 PO4 solution until a pH of 7 was reached. For washing and incubation of the protein chips during the immunoassay, 0.1% (v/v) Tween 20 purchased from VWR International GmbH was added to the buffer. This surfactant promotes washing of the protein arrays and minimizes unspecific adsorption especially of labeled proteins on the chip surface. The terpolymer polydimethylacrylamide with 10% methacrylic acid glycidylester and 1% methacryloyloxybenzophenone (PDMAA–10%MAGE–1%MABP) was synthesized in a free radical polymerization of the distilled monomers dissolved in dimethylformamide with AIBN as an initiator as reported before [21]. A content of 10% MAGE was chosen because it provided sufficient protein binding capacity but was still soluble in water or phosphate buffer. In analogy polydimethylacrylamide with 5% vinyl phosphonic acid (VPA) and 1% MABP (PDMAA–5%VPA–1%MABP) was synthesized in a toluene solution. For characterization the infrared, nuclear magnetic resonance and UV spectra of the polymers had been recorded showing that the polymerization reactions were conducted in proper form. 2.2. Instrumentation The swelling behavior of the surface-attached hydrogel was analyzed by optical waveguide spectroscopy (OWS). For this purpose a self-made surface plasmon resonance spectrometer equipped with a prism made from lanthanum glass (LaSFN9) in the Kretschman configuration was used. Its backside was coated with a 50 nm thin film of evaporated gold and a 20 nm thin film of polystyrene. Either an Omnigrid 100 contact printer from Geneworx AG (Oberhaching, Germany) or a contactless printer (SciFlexarrayer S5 from Scienion AG (Berlin, Germany)) were used to produce the microarrays. Fluorescence images of the obtained microarrays were taken using a Biodetect® 645 from Genescan Europe AG (Freiburg, Germany). 2.3. Procedures On the prism for OWS the polystyrene was dip-coated from an ethyl acetate solution (5 mg mL−1 ) at a withdrawal speed of
94
M. Moschallski et al. / Analytica Chimica Acta 671 (2010) 92–98
60 mm min−1 . On top of this adhesion layer the polymer of interest was dip-coated from an ethanol solution (100 mg mL−1 ) at the same withdrawal speed and then photo-crosslinked (254 nm, 500 mJ cm−2 ). The hydrogel under investigation was swollen repeatedly by filling the fluidic chamber. The thickness was determined with water or buffer, then the polymer layer was dried by flowing nitrogen through the empty chamber and measured in both states. The simulation curves were calculated using the software WinspallTM that uses the Fresnel equations to determine the reflectivity spectrum for a model consisting of the sequence of layers on the glass with their corresponding thicknesses and complex refractive indices. The latter parameters were varied to adapt the simulated spectrum to the measured data. Microarrays were printed from a mixed solution of polymer (PDMAA–10%MAGE–1%MABP) and antibody in 100 mM NaPi buffer at pH 7. Volumes of 400 pL up to 6.4 nL of the solution were dispensed per spot using the above mentioned microarrayers. A volume of 400 ± 10 pL dispensed per droplet was measured in images obtained with the help of a stroboscope camera during spotting. Higher spot volumes were achieved by printing up to 16 of these droplets on the same position. Usually the dry chips were irradiated with UV-light at a wavelength of 254 nm with an energy dose of 500 mJ cm−2 to crosslink the polymer and immobilize the proteins before the assay. Capture immunoassays were performed with fluorescently labeled antigen (BSA). For this, the chips were washed in 30 mM sodium citrate buffer (2× SSC) with 0.1% (v/v) sodium dodecylsulfate (SDS) and 0.05% (v/v) Tween 20 and rinsed with deionized water. Then they were incubated with 100 nM labeled BSA dissolved in phosphate buffered salt solution with Tween 20 for 3 h in a closed chamber in order to allow for sufficient time for the formation of the protein complex. Finally, the chips were washed again in the same way as in the first step, before they were dried in a stream of nitrogen and the fluorescence intensity was measured.
Fig. 1. Printing microarrays of mixed polymer protein solutions and immobilization during UV-crosslinking. (A) Macroscopic view: (1) solutions of polymer and protein in buffer are mixed and unmodified plastic slides are provided, (2) a microarray of the solution is printed on the chip by contact printing and (3) the chip is irradiated with UV-light. (B) Molecular point of view (symbols not to scale): (1) epoxide side groups of the polymer react with primary amino groups of the protein forming complexes in solution, (2) droplets of protein and polymer form at the surface and (3) the photoreaction of the benzophenone moieties in the polymer crosslinks the polymer, attaches it to the surface, and simultaneously immobilizes the proteins in the polymer network.
proteins and at the same time efficiently crosslink the hydrogel. For this reason epoxide groups are added to the polymer for protein immobilization. However, these groups can interact only with the proteins and do not contribute to the hydrogel crosslinking. 3.2. Hydrogel characterization
3. Results and discussions 3.1. Single step microarray production The polymer protein chips were produced on commercially available plastic slides as illustrated in Fig. 1 from a macroscopic point of view as well as seen at a molecular level. In this study almost exclusively PMMA slides were used. Other plastic substrates, i.e. polystyrene or hydrophilized Topas, can also be favorably employed as long as they exhibit low background fluorescence and can be wetted by the printing solution (data not shown). To generate this solution, first a mixture of the polymer in water and the protein in buffer was prepared. After a short time of incubation to account for the required reaction time of the epoxide side groups of the polymer with the primary amino groups of the proteins to form protein–polymer complexes [22], the thus obtained mixture was printed directly onto the unmodified polymer substrates by using a contact printer. Finally, the chips were UV-irradiated by flood exposure, which induces the radical-based photoreaction of the benzophenone moieties in the polymer. The photoprocess crosslinks the polymer to a network, covalently attaches it to the polymer substrate and simultaneously immobilizes the proteins in the surface-attached network as illustrated in Fig. 1. The covalent immobilization of the proteins inside the hydrogel both through the epoxide groups as well as through the radical reaction with the benzophenone moieties in the polymer minimizes bleeding of the proteins out of the polymer network during the immunoassays. As the benzophenone content is only 1% to allow high swelling factors of the hydrogel, it is not enough to bind high amounts of
A critical question which needs to be addressed before any protein is attached to the hydrogels, is how strongly the hydrogel is swollen under assay conditions, i.e. in the presence of an aqueous buffer. Swelling is a key for the performance of the microarray. Unfortunately many studies which use hydrogels do not report on swelling parameters. This is even more important on the background that surface-attached hydrogels exhibit a significantly different swelling as the identical free (i.e. not surface-attached) hydrogel [20]. This is especially crucial because the swelling of the hydrogel will strongly influence the accessibility of protein molecules located within the hydrogel. A method to elucidate this question is optical waveguide spectroscopy, which allows determining simultaneously the thickness and the refractive index of an ultrathin polymer layer. To this a hydrogel layer of PDMAA–10%MAGE–1%MABP was dip-coated on the prism for the OWS measurement (see Section 2.3). A dry film thickness of 1 m was determined after extraction in water and subsequent drying in a stream of nitrogen. Fig. 2 shows the reflectivity spectra of the polymer film in the dry state as well as that of the film swollen in deionized water. The simulated reflection curves, which were calculated according to the Fresnel equations, are adapted to the measured data to determine the film thickness and the refractive index of the swollen polymer. As the waveguide spectra show a series of (orthogonal) resonance signals, the system is overdetermined and both parameters can be calculated simultaneously from the same measurement. Swollen and dry thickness was measured for at least ten cycles of swelling and drying. The linear swelling factor of the polymer net-
M. Moschallski et al. / Analytica Chimica Acta 671 (2010) 92–98
95
Fig. 2. Reflectivity spectra of a polymer film measured by optical waveguide spectroscopy with polarized light on a LaSFN9 prism. (1) Dry film d0 = 1.1 m; n0 = 1.51 (refractive index); (2) film swollen in deionized water d = 3.9 m; n = 1.39; the solid lines are calculated according to the Fresnel equations.
of the degree of swelling of a free gel, which is in good agreement with the experimental data for both hydrogel compositions used here. Similar swelling factors have been determined for PDMAA–5%VPA–1%MABP [21]. In the following experiments we printed spots of the polymers mixed with antibodies onto plastic slides and showed their use for immunoassays. As the areas of the hydrogel spots had a diameter of 100–600 m with a dry film thickness between 20 and 100 nm, these hydrogel structures can be regarded as thin surface-attached hydrogel films and are expected to show the same swelling behavior as the films measured by optical waveguide spectroscopy, although the microspot structures could not be measured in a swollen state. Fig. 3. The linear swelling factor of the surface-attached hydrogel film plotted as a function of the NaPi buffer concentration.
work was calculated as the ratio of the swollen versus the dry layer thickness. It was analyzed for a series of solutions containing NaPi buffer between 1 M and 100 mM. Fig. 3 illustrates, that the swelling ratio d/d0 decreases linearly with the buffer concentration c and can be approximated by the following formula: d = 3.7 − 1.3c d0
(1)
Here the summand 3.7 describes the degree of swelling of the hydrogel in salt-free environment and the factor 1.3 represents the influence of added salt onto the swelling behavior. However, even at high salt concentrations the hydrogel is strongly swollen, i.e. even at the solubility limit of the buffer, which is about 1 mol L−1 for the NaPi buffer system, it is still swollen by a factor of 2.4, which means, that the layer still contains about 60% (v/v) of water. The slight shrinking of the polymer layer with increasing salt concentration can be explained as a “salting out effect”. That means the solvent quality of the aqueous solution is slightly reduced at high salt concentrations—an effect well known for the swelling of free hydrogels in contact with salt solutions [21]. In the following we use in all experiments salt concentrations of 200 mM NaPi which are typically employed in standard immunoassays, so that the surface-attached hydrogels are swollen by roughly a factor of 3.4. It should be noted, that due to the geometric constraints imposed by the surface-attachment, very thin surface-attached gels swell significantly different (i.e. less strongly) than nonattached but otherwise identical free gels [20,21]. Application of the Flory–Rehner theory of the swelling of gels to such a twodimensional system suggests that the degree of swelling of such a surface-attached gel is expected to be roughly the square root
3.3. Determination of protein immobilization efficiency In order to quantify the number of antibodies immobilized in the polymer spots, arrays of fluorescence-labeled antibodies were printed on GOPS silanized glass and on PMMA slides. PDMAA–5%VPA–1%MABP was used as the hydrogel to immobilize anti-BSA. Only 2% of the printed antibodies were labeled in order to prevent quenching of the fluorophores at the high protein concentrations under investigation. The fluorescence intensity integrated over each spot measured immediately after printing defined volumes of a set of protein concentrations was used to calibrate the signal of the corresponding known amounts of labeled proteins printed per spot. Since the signal increased linearly with antibody concentration, quenching was excluded (see also Supplemental Information). As a reference experiment antibodies were also printed without polymer to show, on which substrate the proteins form monolayers. The chips were stored at 6 ◦ C over night in order to assure that the proteins have enough time to bind to adjacent epoxide groups at the chip surface. After the over night reaction the protein chips were washed and the fluorescence signals were measured a second time. The density of immobilized antibodies in each single spot was determined as the quotient of fluorescence intensities after washing and before washing multiplied by the amount of antibodies dispensed per spot. Fig. 4 illustrates how antibodies are immobilized onto monolayers (a) and in the hydrogel (b) on both silanized glass and plastic slides. Proteins printed under otherwise identical conditions, but without polymer bind to GOPS silanized glass chips and form a monolayer. In contrast to this, they cannot bind to a PMMA slide and most antibodies are washed off the chip (see Fig. 4a). The hydrogel in contrast allows attaching the antibodies in a more three-dimensional way on both substrates as demonstrated in Fig. 4b.
96
M. Moschallski et al. / Analytica Chimica Acta 671 (2010) 92–98
Fig. 5. Number N of antibodies (anti-BSA) immobilized into a PDMAA-based polymer network according to fluorescence measurements after washing in 100 mM NaPi buffer with 0.1% Tween 20. The average values of four chips are shown printed with four identical spots of each probe. Error bars indicate the standard deviation of the complete set of data (n = 16).
Fig. 4. Schematic illustration and fluorescence images of antibody immobilization on two different surfaces: (1) Glass slide silanized with GOPS. (2) Plastic slide. On each surface: (a) monolayers of antibodies were printed and (b) antibodies were immobilized in hydrogel. The fluorescence images on the right corresponding to the schematics on the left are taken after washing from a series of solutions with varied concentrations of labeled antibodies (a) without polymer (monolayer) and (b) mixed with 1 mg mL−1 PDMAA–5%VPA–1%MABP hydrogel printed on both substrates.
The fluorescence images of antibody dilution series printed in a microarray format (Fig. 4b) clearly show that antibodies are immobilized in the hydrogel regardless, whether the hydrogel is attached to glass or to PMMA surfaces. As far as antibody monolayers are concerned, a reasonable extent of binding is only observed on epoxide silanized glass, but not on plastic surfaces, where hardly any fluorescence is detected (Fig. 4-2a). From all hydrogel samples well defined, round spots with a constant diameter of 400 m are obtained, when approximately 5 nL drops are printed. In contrast to this, the monolayer spots are smeared out, because the antibodies are not strongly bound in the monolayers and some of them bleed out during washing and become readsorbed in an adjacent location. This occurs preferably parallel to the direction of withdrawal of the chips from the washing solution. In one set of experiments, a series of antibody solutions was printed with concentrations ranging from 0 to 2000 mg L−1 . The fluorescence intensity increases with antibody concentration starting at a low intensity of 50 mg L−1 antibody, which can hardly be seen, up to very high intensities for 2 g L−1 . From these measured intensities the density of immobilized antibodies per spot can be determined by using the calibration deter-
mined from the measurement of labeled antibody spots before washing. The number of antibodies immobilized in the three-dimensional hydrogel scales linearly with their concentration in the print solution and does not reach a limit within the investigated range. In contrast the number of antibodies immobilized in a monolayer seems to reach asymptotically a limit at around 5 × 109 per spot, which is ten thousand antibodies per m2 or one antibody on 100 nm2 . 3.4. Functionality of immobilized antibodies To investigate the ability of the immobilized antibodies to capture antigen from solution, the samples described in Fig. 5 were exposed to labeled antigen in a form of a capture immunoassay according to the protocol presented in the experimental section. Fig. 6 illustrates how different densities of immobilized antibodies in such hydrogels can bind labeled antigens from solution. If one looks at these values, it is important to be aware of the background signal. As far as this is concerned, two different parameters are important. On the one hand it is important to know how much protein adsorbs unspecifically to the hydrogel, on the other hand it should be known, how the protein adsorbs unspecifically to the chip surface, which is not coated with hydrogel. On control spots, where no antibody was immobilized in the hydrogel, no appreciable fluorescence signal could be measured, which proves, that the labeled proteins adsorb, if at all, only very weakly unspecifically onto the hydrogel itself [23]. Between the spots also very low fluorescence intensity was measured. When the chip was washed with a washing
Fig. 6. Schematic illustration and intensity plot of the fluorescence intensity of a direct immunoassay with labeled antigens and immobilized antibodies on GOPS silanized glass slides in hydrogel. As a control also hydrogel spots were generated, which contained no protein (0 mg L−1 ). See also Fig. 7 for a quantitative analysis.
M. Moschallski et al. / Analytica Chimica Acta 671 (2010) 92–98
97
4. Conclusions
Fig. 7. Number of captured BSA antigens N(BSA) per spot determined from fluorescence measurement after a capture immunoassay with labeled BSA (100 nM) on anti-BSA immobilized both in PDMAA–5%VPA–1%MABP and in monolayers on a GOPS silanized glass slide as a function of the number of antibodies immobilized N(anti-BSA) per spot. The data labels specify the antibody concentrations in the spotted solution. The average values of four chips are shown printed with four identical spots of each probe. Error bars indicate the standard deviation of the complete set of data.
buffer, which contained the surfactant Tween 20, a signal-to-noise ratio between the spots and the surrounding background of around 220 was observed. This very high signal to background ratio especially on unmodified PMMA is reached because this surface does not exhibit protein binding groups. Therefore the simple addition of the surfactant that prevents hydrophobic interaction of the protein with the polymer surface efficiently “blocks” unspecific adsorption to this background. Fig. 7 shows the number of captured BSA antigens versus the number of immobilized antibodies per spot. The number of captured antigen increases linearly with antibody concentration in the hydrogel spots over the whole concentration range up to a captured amount of around 108 antibodies/spot. In contrast to this the amount of antigen captured by an antibody monolayer does not increase linearly and asymptotically reaches a limit at around 4 × 107 BSA/spot, so that saturation is already reached at a printed antibody concentration of 0.5 g L−1 . The linear behavior observed in the hydrogel system implies that a constant fraction of the antibodies is still active after immobilization, while the ratio of active antibodies in the monolayer decreases for antibody concentrations above 0.25 g L−1 . One question, which needs to be addressed, is why the number of BSA molecules captured is much smaller than that of the immobilized anti-BSA. It has been reported before, that the immobilization of antibodies in monolayers can result in a decrease in affinity or binding capacity [24]. This has been explained by steric hindrance of the antigen’s access to the paratopes at the antibody. In agreement with this in other work it has been shown, that with increasing antibody density on the surface, their efficiency to bind large antigens decreases significantly [25]. Especially, when the antibodies are randomly oriented, as in the case of amino groups binding to epoxide silane surfaces, some of the paratopes may be too close to the surface to be accessible for antigens [26]. The maximum antibody density measured in the monolayer is comparable to that reported by Kusnezow and coworkers with twenty thousand per m2 [27]. They assumed at least half of the antibodies inaccessible, but did not give quantitative data. When the results shown in Fig. 7 are analyzed, it has to be considered, that the number of capture proteins (anti-BSA) in the hydrogels and the monolayer are the same. The higher binding capacity in the hydrogels seem to indicate, that more paratopes are accessible to bind antigens, if the antibodies are immobilized in the flexible, highly swollen hydrogel compared to immobilization in a dense monolayer. This indicates that the immobilization reaction of the rare benzophenone moieties in the hydrogel does not affect the biofunctionality of the antibodies.
The benzophenone containing hydrogels allow a simple photochemical immobilization of antibodies on polymer substrates. Since the radical-based photoreaction of benzophenone is not specific, various polymer protein mixtures can be attached to essentially any polymer substrate provided, that are wetted by the aqueous print solution. The only additional prerequisite is that the substrate exhibits a sufficiently low background fluorescence to allow facile read-out. Hence, this process for protein chips opens the road to the use of simple polymer substrates that can be fabricated by injection molding in high numbers at extremely low cost. On top of it, our strategy enables to produce protein chips in a very simple way. The polymer and protein solutions are mixed and printed on (clean) unmodified polymer substrates followed by a short UV-exposure to induce the photoreaction. The polymer protein chips can be printed under ambient conditions. There is neither the need for an inert atmosphere nor for a specific chemical laboratory environment for chip production, which makes the process very simple and fast. Since the chosen hydrogel binds proteins via epoxide and benzophenone groups, this approach is not limited to antibody arrays, but can bind many kinds of proteins including enzymes, which is currently elucidated in other studies. The resolution limit of the process depends only on the printer used and the wettability of the chip surface. Hence, the 400 pL droplets printed here gave spots of about 150 m in diameter. If desired, even smaller hydrogel structures can be produced by photolithography if the protein–polymer mixture is UV-irradiated through a fine grid and then non-exposed areas of polymer and protein are washed off. It is a very interesting feature of this method of chip fabrication, that it permits a detailed quality control of the chip production process. As it is essentially a one-step process, the measurement of the volume of a dispensed droplet is sufficient to determine the number of immobilized probe proteins, because they are covalently bound by both benzophenone and epoxide side groups of the polymer and therefore cannot bleed out during the assay. Despite the simple production process proteins immobilized in these hydrogels show a high accessibility to capture antigens from a sample solution. We will show in a following communication, that the described method can be used to generate chips with extremely high signal-to-noise ratios and accordingly high sensitivity. Conflict of interest statement The authors declare no financial or commercial conflict of interest. Acknowledgment We thank Natalia Schatz for valuable technical support. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.aca.2010.05.008. References [1] T. Kodadek, Chemistry & Biology 8 (2001) 105. [2] M.F. Templin, D. Stoll, J.M. Schwenk, O. Pötz, S. Kramer, T.O. Joos, Proteomics 3 (2003) 2155. [3] S. Sauer, B.M.H. Lange, J. Gobom, L. Nyarsik, H. Seitz, H. Lehrach, Nature Reviews Genetics 6 (6) (2005) 465. [4] G. MacBeath, Nature Genetics 32 (2002) 526.
98
M. Moschallski et al. / Analytica Chimica Acta 671 (2010) 92–98
[5] H. Zhu, M. Bilgin, R. Bangham, D. Hall, A. Casamayor, P. Bertone, N. Lan, R. Jansen, S. Bidlingmaier, T. Houfek, T. Mitchell, P. Miller, R.A. Dean, M. Gerstein, M. Snyder, Science 293 (2001) 2101. [6] P. Bertone, M. Snyder, FEBS Journal 272 (21) (2005) 5400. [7] J. Qiu, J. Madoz-Gurpide, D.E. Misek, R. Kuick, D.E. Brenner, G. Michailidis, B.B. Haab, G.S. Omenn, S. Hanash, Journal of Proteome Research 3 (2004) 261. [8] P. Wagner, R. Kim, Current Drug Discovery 5 (2002) 23. [9] B.B. Haab, M.J. Dunham, P.O. Brown, Genome Biology 2 (2001) 13. [10] B.A. Stillman, J.L. Tonkinson, BioTechniques 29 (2000) 630. [11] S. Oehler, R. Alex, A. Barker, Analytical Biochemistry 268 (1999) 330. [12] W. Kusnezow, A. Jacob, A. Walijew, F. Diehl, J.D. Hoheisel, Proteomics 3 (2003) 254. [13] G. MacBeath, S. Schreiber, Science 289 (2000) 1760. [14] W. Kusnezow, J.D. Hoheisel, Journal of Molecular Recognition 16 (2003) 165. [15] O. Prucker, C.A. Naumann, J. Ruhe, W. Knoll, C.W. Frank, Journal of the American Chemical Society 121 (1999) 8766. [16] D.-N. Kim, W. Lee, W.-G. Koh, Analytica Chimica Acta 609 (2008) 59. [17] W. Lee, D. Choi, Y. Lee, D.-N. Kim, J. Park, W.-G. Koh, Sensors and Actuators B: Chemical 129 (2008) 841.
[18] A.L. Hook, H. Thissen, N.H. Voelcker, Biomacromolecules (2009). [19] D.G. Swan, M.J. Swanson, www.freepatentsonline.com, vol. 09/521545, SurModics, Inc., Eden Prairie, MN, United States, 2004. [20] R. Toomey, D. Freidank, J. Ruhe, Macromolecules 37 (2004) 882. [21] T. Neumann, Wiederverwendbare 3D-Polymer-DNA-Chips zur Echtzeitanalyse und kompartimentierten Paralleldiagnostik, dissertation, University of Freiburg, 2006. [22] S. Golze, Oberflächengebundene Polymermonolagen für die Herstellung von DNA-Chips, dissertation, University of Mainz, 2001. [23] B. Berchtold, Surface attached polymers for the reendothelialization of porcine heartvalves, dissertation, University of Freiburg, 2005. [24] T.M. Spitznagel, J.W. Jacobs, D.S. Clark, Enzyme and Microbial Technology 15 (1993) 919. [25] Y. Kwon, Z. Han, E. Karatan, M. Mrksich, B.K. Kay, Analytical Chemistry 76 (2004) 5713. [26] R.A. Vijayendran, D.E. Leckband, Analytical Chemistry 73 (3) (2001) 471. [27] W. Kusnezow, Y.V. Syagailo, S. Ruffer, K. Klenin, W. Sebald, J.D. Hoheisel, C. Gauer, I. Goychuk, Proteomics 6 (2006) 794.