Article pubs.acs.org/ac
Carboxymethyl Cellulose Film as a Substrate for Microarray Fabrication Yuri M. Shlyapnikov,*,† Elena A. Shlyapnikova,† and Victor N. Morozov†,‡ †
Institute of Theoretical and Experimental Biophysics, Russian Academy of Sciences, Pushchino, Moscow Region, Russia 142290 The National Center for Biodefense and Infectious Diseases, George Mason University, Manassas, Virginia 20110, United States
‡
S Supporting Information *
ABSTRACT: Magnetic beads (MB) are widely used for quick and highly sensitive signal detection in microarray-based assays. However, this technique imposes stringent requirements for smoothness and adhesive properties of the surface, which most common substrates do not satisfy. We report here a new type of substrate for microarrays with a low adhesion to MBthermally cross-linked carboxymethyl cellulose (CMC) film. This substrate can be readily fabricated on a conventional glass slide. A highly cross-linked CMC film (∼1 cross-link per monomer unit) possesses a surface smooth on a nanometer scale and a low adhesion to protein-coated MB, which partly originates from electrostatic repulsion of MB from negatively charged CMC surface. The efficiency of the CMC substrate is demonstrated hereby in fabrication of microarrays for the detection of three bacterial toxins: cholera toxin, staphylococcal enterotoxin A, and toxic shock syndrome toxin. The assay employing a primary antibodies arrayed on a CMC surface and detection of the bound bacterial toxins with a biotinylated secondary antibodies and streptavidin-coated MB resulted in a limits of detection as low as 0.1 ng/mL. The CMC-based microarrays demonstrated very high storage stability; their activity did not change after one year storage at room temperature.
S
and effectively compete with strong but less numerous specific antigen−antibody bonds. That is why many conventional substrates cannot be used in immunoassay where MB are employed as labels. Several types of substrates have been reported so far to work well in the MB-based detection: membranes from regenerated cellulose,1−3 gold blocked by polyethylene glycol (PEG)−SH following immobilization of SH-functionalized biomolecules,7 PEG-grafted polystyrene,8,9 and neutravidin-coated glass slides.5,6 Except for the cellulose membranes, preparation of all other supports mentioned above involves complex and expensive procedures of surface modification, unsuitable for a large-scale industrial application. Here we describe a simple inexpensive method of coating of a solid surfaces, which makes them suitable for the manufacture of microarrays to be used in assays employing the detection with MB. Because unspecific binding presents a problem in other heterophase assays employing micro- and nanoparticles, we believe that the approach described here may also be helpful for reduction of unspecific binding in the assays using superconducting quantum interference devices (SQUIDs) for magnetic signal detection10−14 and other similar techniques.
uperparamagnetic beads are widely used as labels in microarray-based immunoassays1−9 and in polymerase chain reaction (PCR)-free detection of genomic DNA.5 Magnetic beads (MB) tethered to the microarray surface by specific intermolecular bonds may be further discovered by a magnetoelectronic detector5,6 or under a dark-field microscope.1−3 The detection procedure employing MB can be performed in many ways:2 from the simplest “push−pull” technique to the scan of microarray surface with MB pushed by a shear flow. The main advantages of the latter technique are its speed and sensitivity. Typically, the whole procedure takes about 1 min and provides a sensitivity to single molecules. Because a single antigen−antibody bond is capable of holding MB on the surface, individual analyte molecules on the microarray surface can be labeled and detected on the microarray surface.2 For example, this technology has been successfully applied in immunochemical assay,1 where active electrophoretic collection of analytes followed by the detection of the captured analyte with MB in the scanning mode allowed one to detect just a few hundred protein molecules and viruses within 3 min. Detection with MB imposes more severe requirements on the properties of microarray substrate than enzyme-linked immunosorbent assay (ELISA) and other heterogeneous immunoassay techniques. This is explained by a large area of the bead−surface contact where numerous weak nonspecific interactions (van der Waals, ionic, hydrophobic) come into play © 2014 American Chemical Society
Received: November 7, 2013 Accepted: January 22, 2014 Published: January 22, 2014 2082
dx.doi.org/10.1021/ac403604j | Anal. Chem. 2014, 86, 2082−2089
Analytical Chemistry
■
Article
EXPERIMENTAL SECTION Materials and Reagents. The following reagents were purchased from Sigma-Aldrich Co.: 3-aminopropyltriethoxysilane (APTES), Atto 655-streptavidin, protein A, bovine serum albumin (BSA), ovalbumin (Ova), carboxymethyl cellulose (CMC) sodium salt, 250 kDa, degree of substitution 0.7, cholera toxin (CT), N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC), ethanolamine, fat-free dry milk, glycine, glutaraldehyde, N-hydroxysuccinimide (NHS), 2(N-morpholino)ethanesulfonic acid (MES), new fuchsin dye, PBS tablets, poly(ethyleneimine) (PEI), poly(vinyl alcohol) (PVA, 124−186 kDa, 98−99% hydrolysis degree), poly(vinylpyrrolidone) (PVP, 360 kDa), sodium borohydride, trehalose, and Tween 20. Streptavidin (SA)-coated and carboxylic MB (Dynal MyOne, 1 μm diameter) were from the Invitrogen (Carlsbad, CA, U.S.A.). Microscope glass slides were the products of the Gerhard Menzel, Glasbearbeitungswerk GmbH & Co. (Braunschweig, Germany). Dialysis membranes from regenerated cellulose (Fisherbrand, MW cutoff 3.5 kDa) were obtained from Fisher Scientific (Pittsburgh, PA). Solutions were prepared using Milli-Q water. Staphylococcal enterotoxin A (SEA), toxic shock syndrome toxin (TSST), monoclonal antibodies to SEA, TSST, CT, prepared as described in ref 15, were kindly provided by Professor E. Grishin. CMC Film Preparation and Characterization. A 0.5% CMC solution in water was filtered through a 0.2 μm PVDF filter (Immobilon P, Sigma-Aldrich) and concentrated on a centrifugal evaporator to a final concentration of 1.5%, which was determined by weighting the dry residues. Microscope slides were cleaned overnight in 3% K2Cr2O7 in H2SO4, immersed in 1% APTES solution in 20% EtOH/H2O (v/v) or in 0.3% PEI water solution for 30 min at room temperature (RT), washed with water, and dried. CMC solution was spincoated on the slides at 3000 rpm, dried, and cured at 150 °C. Film thickness was estimated by scratching its surface to expose the glass surface and by measuring the scratch depth with the Linnik’s interferometer. To prepare CMC samples for swelling studies and for the titration of carboxyl groups, a thick (∼1 mm) layer of the CMC solution was dried on a glass slide. AFM Characterization of the CMC Surface. Atomic force microscopy (AFM) was performed using a SmartSPM1000 atomic force microscope (AIST-NT, Co., Moscow, Russia). The tapping mode with a resonance frequency of 300−350 kHz was used in all scanning experiments. Root mean square (rms) values of surface height variations were determined using the Gwyddion software (Czech Metrology Institute, Czech Republic) on 0.5 μm2 surface regions with plane subtraction filter applied. Titration of Base Groups. For titration of base groups, pieces of a film containing ∼3 mg of dry CMC were placed in a 0.1 M HCl solution for 10 min, washed with water until a constant pH value was obtained, and placed into a 0.5 mL 1 M KCl solution. After equilibration under stirring, pH change caused by the film introduction was titrated back with a 10.0 mM NaOH solution. Measurements of MB Adhesion to Different Substrates. These were performed by the “push−pull” technique and by scanning in the shear flow as described in ref 2. Experimental details of the procedures, as well as the surface
modification protocols are provided in the Supporting Information. ζ-Potential Measurements. The ζ-potential of the CMC film surface was measured using the rotating disk technique described in ref 16. A CMC film was manufactured on a round cover glass, 2 cm in diameter, as described above. The disk was attached to a rotating head, and the streaming potential was measured between two Ag/AgCl electrodes one of which was placed close to the center of the disk, while the other was at a distance from the disk. Potential measurements were performed in a 1 μM KCl solution at different speeds of the disk rotation. The ζ-potentials of SA and carboxylic MB were determined by measuring a streaming potential generated upon pumping a buffer solution through a microcolumn of beads. A homemade device used in this measurement is described in the Supporting Information. Microarray Fabrication. CMC films were first activated for 20 min in 0.4 M NHS dissolved in 50 mM MES buffer, pH 6.0, containing 1% EDC, washed with water, quickly dried, and used immediately for microarray manufacturing. Anti-CT antibodies were deposited by electrospraying a dialyzed solution of 0.1% monoclonal anti-CT-IgG containing 1% of trehalose as described in ref 17. Substrates with protein−sugar deposits were kept for 30 min in a humid chamber and then immersed into a 0.5 M glycine solution, pH 9.0, for 1 h. The microarrays were then washed with water, rinsed with a 1% trehalose solution, and dried. For storage stability studies, ∼0.1 μL drops of a 0.1% monoclonal anti-CT-IgG solution containing 1% of trehalose were manually spotted onto the CMC film, all other procedures being the same as described above. Anti-CT-IgG microarrays on a dialysis membrane were fabricated as described in ref 3. Multicomponent microarrays were fabricated by spotting 0.1% solutions of dialyzed proteins (Ova, protein A, antibodies to CT, SEA, TSST) containing 1% trehalose onto the CMC substrate activated as described above. An Xpress Lane (U.S.A.) spotter with pins, 140 μm diameter, was used for arraying. The printing hardware and software were adjusted to enable a noncontact spotting so that the pin tip stopped at a distance of 30−50 μm from the substrate surface. Then a pressure impulse was applied inside the tip to push the solution into contact with the substrate surface. After printing the microarrays were treated as described above. Bacterial Toxins Immunoassay with MB in the Scanning Mode. The flow cell used in the assay with MB detection is described in ref 2. An analyte solution (100 μL) in PBS containing 1% PVP and 0.1% Tween 20 (buffer A) was pumped for 3 min over the microarray in the flow cell. Then 100 μL of a biotinylated detecting antibodies solution (anti-CT antibody, 0.2 μg/mL, for single-component assay, or mixture of anti-CT, anti-SEA, and anti-TSST, 0.2 μg/mL each, for multiplex assay) in buffer A was pumped through the flow cell for 2 min. After that, a magnet was placed underneath the flow cell, and a suspension of SA-coated MB (0.001%, w/w, in the sodium phosphate buffer, pH 7.4, containing 10 mM NaCl, 1% PVP, and 0.1% Tween 20) was pumped through the flow cell at a rate of 8−12 μL/min for 2 min. The concentration of sodium phosphate was 1 mM unless otherwise specified. The image was then taken by a CCD camera. The number of tethered MB in each microarray spot was calculated as described in the Supporting Information. Measurement of CT by Using a Fluorescent Label. The microarrays were sequentially treated with a solution containing 2083
dx.doi.org/10.1021/ac403604j | Anal. Chem. 2014, 86, 2082−2089
Analytical Chemistry
Article
20 μg/mL of CT, then with biotinylated monoclonal anti-CTIgG (20 μg/mL), and finally with Atto 655-SA (20 μg/mL) in buffer A (see above), 2 h for each stage. Microarrays were washed with water for 2 min after each stage. The fluorescence intensity was measured using a VAF-1 chip detector (MMTech, Russia). The same experiments without the CT binding step were also carried out as negative controls, which always showed no unspecific signal.
■
SAFETY CONSIDERATIONS The staphylococcal and cholera toxins are poisonous and should be handled with care.
■
RESULTS AND DISCUSSION Figure 1. Strength of the specific (dashed line) and nonspecific (solid lines) bonds between SA-coated beads and different surfaces: ⧫ , APTES slide; ■, dry milk-blocked APTES slide; □, CMC film at 100 mM NaCl; ○, CMC film at 10 mM NaCl; ∗, dialysis membrane at 100 mM NaCl; ●, dry milk-blocked aldehyde slide; ▲, polystyrene; ×, dry milk-blocked polystyrene; Δ, the strength of the specific antigen− antibody binding evaluated in the CT microarray-based assay on a plasma-treated dialysis membrane, shown for comparison. Only the active zones of a microarray (spots) were used to calculate the density of bound beads. The CT concentration was 100 ng/mL.
MB Adhesion to Different Substrates. Our major goal was to find substrate materials and the techniques for treatment of their surfaces which would allow to manufacture microarrays with minimum background (density of MB bound nonspecifically) and maximum signal-to-noise ratio. Push−pull technique provides an easiest way to characterize bead−surface interaction by pressing first MB to the substrate surface with a magnet placed underneath, then pulling them from the surface with another magnet placed above.2 The density of MB left on the surface characterizes the bead−surface adhesion. Two types of MB were used in an adhesion study: carboxylic acid- and SA-coated beads. Our experimental data show that the latter demonstrate adhesion properties similar to those of beads covered with other slightly acidic or near-neutral proteins, i.e., antibodies. Several types of surfaces were tested, including those that have been reported to be successfully used with MB as labels. The results are summarized in Table S1 of the Supporting Information. As seen in Supporting Information Table S1, blocking polystyrene and coated glasses with BSA and dry milk reduces adhesion of both types of MB to 1/10−1/ 100 of its level before blocking, though substantial adhesion still persists. This “residual” adhesion may not essentially affect the “push−pull”-based assay, where each bead probes only a small area of microarray surface and the probability that a bead falls directly onto the strongly binding (“hot”) spot is low. A relatively low background in this case can still provide a good signal-to-noise ratio in the push−pull technique.4 However, in the shear flow scanning technique,2 when the beads scan all the surface area in search of a specific site, all “hot” spots will be eventually found and marked with MB, considerably increasing the background. This technique imposes more stringent requirements on substrate adhesive properties. Most common substrates, as we demonstrate below, do not satisfy the condition of low adhesion in scanning even after blocking their surface. The number of MB retained on the surface plotted versus the shear rate for different surfaces and different ways of their blocking is presented in Figure 1 (see details in the Supporting Information). It was found that the adhesion of beads to polystyrene, APTES, and aldehyde-coated glass did not depend on the salt concentration (10 or 100 mM) and on the way they were blocked: in most cases, the surfaces were very “sticky” to MB, no matter whether they were blocked or not. To realize, how important is the level of unspecific adhesion, one must compare its strength to the strength of specific antigen−antibody bonds which keep tethered MB on the active microarray spots. If nonspecific interactions are stronger than specific ones, the
signal would be indistinguishable from the background. It is seen in Figure 1 that the nonspecific adhesion of MB to the above-mentioned surfaces is substantially stronger than adhesion due to specific interactions making these surfaces unsuitable in assays employing the MB scanning technique. In contrast to this, membrane from regenerated cellulose showed no MB retained at the shear rates above 20 s−1, thus confirming its low adhesion properties already noted above in the “push− pull” mode. The data presented in Figure 1 and Supporting Information Table S1 show that cellulose membrane and agarose gel are characterized by extremely low nonspecific adhesion. We speculate that low adhesion of these substrates is due to entropy-driven steric repulsion of free polysaccharide chains and trains on a swollen surface.18 Unfortunately, the agarose cannot be used as microarray substrate in assays employing the MB detection due to its high permeability to proteins which permits their diffusion into the gel depth, out of contact with MB. As for cellulose film, its performance as microarray substrate is well-documented.1−3 Below we are using the microarrays fabricated on the cellulose membrane as a reference, to which CMC-based microarrays are compared. At the same time, the usage of free unsupported cellulose membrane as a microarray substrate is often associated with many technical complications (it buckles and changes its dimensions upon drying and wetting, whereas complex and expensive procedures are needed to manufacture layers of these materials on a solid support.18 The use of a charged surface to compensate for the hydrophobic and van der Waals attraction forces by electrostatic repulsion between substrate and beads is explored here as simple alternative to cellulose membrane. CMC Film Manufacturing. One possible embodiment of the approach based on electrostatic repulsion of MB employs a cross-linked CMC film. CMC is a readily available hydrophilic anionic polymer. An easy and effective method to cross-link carboxylic acid-containing films consists in their curing at a temperature of 120−160 °C,19 which leads to the formation of ester bonds between alcohol and carboxylic acid groups. In the 2084
dx.doi.org/10.1021/ac403604j | Anal. Chem. 2014, 86, 2082−2089
Analytical Chemistry
Article
Figure 2. Chemical scheme of CMC film cross-linking and linking CMC film to APTES-coated glass surface.
Figure 3. (A) Swelling of CMC films as a function of the curing time. (B) The density of cross-links in CMC (determined from film swelling and from base titration data) as a function of the curing time. The curing temperature was 150 °C.
line. The density of cross-links independently determined from the base titration of cured films provides a very similar dependence on the curing time presented by a solid line in Figure 3B. As seen from Figure 3A, the cross-linking process is completed within 1.5 h at 150 °C. No matter which method was used to estimate the density of cross-links, we should admit that a rather dense network holds the CMC. The density of cross-links is essential both for microarray fabrication and for assay with MB detection because it ensures low permeability of the film for antibody and analyte molecules. Permeability of cross-linked CMC films was characterized by the rate of diffusion of small dye molecules inside the film (see the Supporting Information). As we found, the diffusion coefficient of fuchsin in CMC film is by 50% lower than that in the dialysis membrane with the cutoff of 3.5 kDa, indicating that cross-linked CMC is denser and should be less penetrable for proteins than the dialysis membrane. The base titration data shown in Figure 3B were also used to estimate the density of free carboxylic groups left in the crosslinked CMC film. Taking into account the degree of substitution in the initial uncured CMC polymer (0.7, according to the manufacturer data), and the data in Figure
present work, this thermal cross-linking technique was used to stabilize CMC films on glass slides. Because CMC films do not adhere to glass surface, an additional adhesive layer was introduced. An adsorbed layer of PEI and a layer of covalently bound polymerized APTES were tested here as means to strongly bind the CMC film to the glass surface. While the use of PEI coatings sometimes resulted in the film detachment, APTES layers after curing provided a very strong covalent bonding via siloxane and amide bonds between the glass surface and CMC layer (see Figure 2). Typically, a film thus prepared had thickness of about 300 nm as measured by interferometric microscopy. Film Characterization. We studied the dependence of the CMC film swelling on the curing temperature (see Figure S2 in the Supporting Information). The curing temperature of 150 °C was chosen for subsequent experiments to minimize the curing time and, on the other hand, to prevent the polymer from destruction at higher temperatures. The swelling of cured CMC films as a function of curing time is presented in Figure 3A. The density of cross-links estimated from the swelling ratio by using the Flory−Rehner theory (see the Supporting Information for details) is plotted in Figure 3B as a dotted 2085
dx.doi.org/10.1021/ac403604j | Anal. Chem. 2014, 86, 2082−2089
Analytical Chemistry
Article
Figure 4. (A) AFM image of a cured CMC film surface. The white arrow shows the position of the cross section presented on panel B. (B) Profile of the CMC surface on a cross section indicated on panel A.
Figure 5. (A) Adhesion of MB to a CMC film measured by the “push−pull” technique as a function of NaCl concentration. (B) The electrostatic repulsive force between a magnetic bead, 1 μm diameter, and a CMC film in 1 mM and 10 mM salt concentration (see the text for details of calculations).
the film surface significantly inhibits the adhesion of MB to the surface by counterbalancing hydrophobic and van der Waals forces. At high ionic strength, electrostatic interactions are suppressed, which results in the strengthening of nonspecific adhesion. Estimation of Electrostatic Interactions between CMC Film and MB. To compare magnetic and electrostatic forces involved in the bead−surface interactions, we calculated the electrostatic repulsion force as a function of distance between the bead and film. According to the Derjaguin approximation, the force acting on a sphere with radius R at a distance h, from a planar surface is21,22
3B indicating that 0.5 carboxyl groups were spent for crosslinking, we conclude that ∼0.2 free carboxylic groups are still left per monosaccharide unit. This corresponds to ∼1 mol of carboxylic groups per liter of film. AFM studies revealed a highly smooth surface in cross-linked CMC films as images in Figure 4 illustrate. The smoothness of the microarray surface is essential for detection by scanning with MB because the strength of both specific and nonspecific bead−surface interactions greatly varies with the surface roughness.20 The roughness of the CMC surface at a scale comparable with the size of MB (0.5 μm2) is described by rms of 0.8 ± 0.2 nm, which is substantially lower than the roughness of dialysis membranes, 6.5 ± 1 nm.20 Adhesion of MB to CMC Films. The adhesion of carboxylic and SA-coated MB (both having a negatively charged surface at neutral pH) to a CMC film measured by push−pull technique shows a significant dependence on ionic strength as illustrated in Figure 5A. Whereas in a buffer solution with a low salt concentration (
