DNA-Directed Immobilization: Efficient, Reversible, and Site-Selective Surface Binding of Proteins by Means of Covalent DNA–Streptavidin Conjugates

Analytical Biochemistry 268, 54 – 63 (1999) Article ID abio.1998.3017, available online at http://www.idealibrary.com on

DNA-Directed Immobilization: Efficient, Reversible, and Site-Selective Surface Binding of Proteins by Means of Covalent DNA–Streptavidin Conjugates Christof M. Niemeyer, 1 Larissa Boldt, Bu¨lent Ceyhan, and Dietmar Blohm Department of Biotechnology and Molecular Genetics, University of Bremen, FB 2-UFT, Leobener Strasse, 28359 Bremen, Germany

Received July 27, 1998

Covalent DNA–streptavidin conjugates have been utilized for the reversible and site-selective immobilization of various biotinylated enzymes and antibodies by DNAdirected immobilization (DDI). Biotinylated alkaline phosphatase, b-galactosidase, and horseradish peroxidase as well as biotinylated anti-mouse and anti-rabbit immunoglobulins have been coupled to the DNA– streptavidin adapters by simple, two-component incubation and the resulting preconjugates were allowed to hybridize to complementary, surface-bound capture oligonucleotides. Quantitative measurements on microplates indicate that DDI proceeds with a higher immobilization efficiency than conventional immobilization techniques, such as the binding of the biotinylated proteins to streptavidin-coated surfaces or direct physisorption. These findings can be attributed to the reversible formation of the rigid, double-stranded DNA spacer between the surface and the proteins. Moreover, BIAcore measurements demonstrate that DDI allows a reversible functionalization of sensor surfaces with reproducible amounts of proteins. Ultimately, the simultaneous immobilization of different compounds using microstructured oligonucleotide arrays as immobilization matrices demonstrate that DDI proceeds with site selectivity due to the unique specificity of Watson–Crick base pairing. © 1999 Academic Press

Highly specific, functional biological compounds, such as enzymes and immunoglobulins, are often part of diagnostic and sensoric applications. To extend the applicability of biomolecular elements, a general need for reversible and site-specific immobilization methods is apparent to allow fabrication of reusable bio- and immunosensor 1

To whom correspondence should be addressed. Fax.: (49) 421 218-7578. E-mail: [email protected]. 54

devices. Strategies for reversible immobilization of proteins (1) include reversible chemical interactions (2– 4), in particular metal chelation (5–9) or disulfide cleavage (10–15), protein–ligand interactions (16, 17), and nucleic acid hybridization (18, 19). Moreover, a number of efforts are currently underway to develop methods for siteselective immobilization of biomolecules on surfaces. This should facilitate the fabrication of spatially defined receptor arrays for biosensors and parallel ligand-binding assays. As an example, immobilization of immunoglobulins was achieved by photolithographic techniques (20). A severe problem, however, results from the limited physicochemical stability of many biological macromolecules, preventing a stepwise, successive immobilization of multiple delicate proteins on different surface sites. Thus, the spatially addressing of many different protein compounds requires a single site-selective process under mild chemical conditions. Nucleic acid-directed immobilization of proteins provides a solution of this problem (19). Due to the exceptionally high physicochemical stability of DNA oligomers, high-density oligonucleotide arrays, “DNA chips,” are prepared by successive attachment or spatially separated DNA synthesis using automated photolithographic procedures and are state-of-the-art in today’s nucleic-acid analysis (21). To effectively facilitate DNA-directed immobilization (DDI), 2 biomolecules of interest need to be coupled with a single-stranded DNA moiety, providing a specific recognition site for complementary nucleic acids, and thus a molecular handle for selective immobilization on 2

Abbreviations used: DDI, DNA-directed immobilization; STV, streptavidin; bGal, b-galactosidase; AP, alkaline phosphate; POD, peroxidase; GAM, antibodies from goat directed against mouse; GAR, antibodies from goat directed against rabbit; IgG, immunoglobulin; ONPG, 2-nitrophenyl-b-D-galactopyranoside; pNPP, 4-nitrophenyl phosphate; ABTS, 2,29-azino-di-[3-ethylbenzthiazole sulfonate]; BIA, biomolecular interaction analysis; RU, resonance units; NTP, nanotiter plate; SPR, surface plasmon resonance. 0003-2697/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.

DNA-DIRECTED IMMOBILIZATION

DNA arrays. Since a direct crosslinking of DNA oligomers with protein components, such as antibodies or enzymes, is tedious, biomolecular adapters, covalent conjugates from single-stranded DNA and streptavidin (STV), have been developed (19). Due to the STV’s extraordinary specificity for the small water-soluble molecule biotin (vitamin H) and its high chemical and thermal stability (22), as well as the variety of mild chemical procedures to incorporate biotin into basically any molecular component, the biotin–STV system forms the basis of innumerable diagnostic and analytical tests (23, 24). Therefore, DNA–STV conjugates are versatile connectors for the generation of protein arrays, supramolecular bioconjugates (19), and the fabrication of biometallic nanostructures (25). In this report, these adapters are utilized for the reversible and site-selective immobilization of various enzymes and antibodies. Biotinylated b-galactosidase (bGal), alkaline phosphatase (AP) or horseradish peroxidase (POD), and biotinylated antibodies from goat directed against mouse (GAM) or rabbit (GAR) immunoglobulin G (IgG) have been coupled to DNA–STV adapters by simple, two-component incubation, and the resulting preconjugates were allowed to hybridize to surfacebound oligonucleotide arrays. Quantitative measurements indicate that DDI proceeds with high immobilization efficiency, is completely reversible, and also offers excellent site-selectivity. MATERIALS AND METHODS

Oligonucleotides and DNA–Streptavidin Conjugates Various 59-thiol modified oligonucleotides, purchased from NAPS, Go¨ttingen, Germany, were synthesized by standard methods using C6 Thiolmodifier reagent (Glen Res.) and purified by HPLC. The sequences are as follows: 59-thiol-TCC TGT GTG AAA TTG TTA TCC GCT-39 (tA24), 59-thiol-AGC GGA TAA CAA TTT CAC ACA GGA-39 (tA24as), 59-thiol-GTA ATC ATG GTC ATA GCT GTT-39 (tB21), 59-thiol-CAG GTC GAC TCT AGA GGA TCC-39 (tD21). The following biotinylated oligomer complements (NAPS) have been used as antisense capture oligonucleotides: 59-biotin-AGC GGA TAA CAA TTT CAC ACA GGA-39 (bA24as), 59-biotinTCC TGT GTG AAA TTG TTA TCC GCT-39 (bA24), 59-biotin-AAC AGC TAT GAC CAT GAT TAC-39 (bB21as), 59-biotin-GGA TCC TCT AGA GTC GAC CTG-39 (bD21as). Synthesis and purification of covalent DNA–STV hybrids, HA24, HA24as, HB21, and HD21, were carried out from the corresponding thiolated oligonucleotides, tA24, tA24as, tB21, and tD21, respectively, and recombinant streptavidin (Boehringer Mannheim), as previously described (19). In brief, STV (10 nmol) was derivatized with maleimido groups using a heterobispecific crosslinker (sulfoSMPB, Pierce), reacted with a 59-thiolated oligonucleotide (10 nmol) and subsequently purified by anion-

55

exchange chromatography. A one-to-one molar ratio of nucleic-acid and protein moiety of the hybrids was verified by gel-electrophoretic and photometric analysis, and concentrations of the preparations were determined by absorbance measurements (19). Formation of Preconjugates and DDI Preconjugates of the DNA–STV hybrids and biotinylated proteins were prepared by mixing 0.1 mM stock solutions of the hybrids and a 0.6- to 7-fold molar excess (37.8 mM stock solution) of biotinylated AP (Sigma), biotinylated horseradish POD (Pierce), biotinylated b Gal (Sigma), biotinylated goat anti-mouse F(ab) 2-fragments (GAM, Coulter Immunotech) or biotinylated goat anti-rabbit F(ab) 2 -fragments (GAR, Coulter Immunotech) in buffer A (20 mM Tris–Cl buffer, pH 7.5, 150 mM NaCl, 5 mM EDTA, 0.01% (w/v) Tween 20) containing 0.1 mg/ml bovine-serum albumin. After incubation for 30 min at room temperature, the mixtures were diluted to final concentrations, typically in the range of 25 to 5 nM, with buffer A containing 0.1 mg/ml bovine-serum albumin and 10 mM Dbiotin (Sigma) and the dilutions were incubated for additional 10 min. STV-coated microplates were prepared as previously described (26), and 50 ml of a 240 nM solution of the biotinylated oligomer complements (capture oligomers) in buffer A was allowed to bind for 60 min. After being washed with buffer A containing 10 mM D-biotin, 50 ml of the preconjugate solution was allowed to hybridize for 60 min at room temperature. Negative controls were run in parallel by applying preconjugates to wells, coated with noncomplementary oligonucleotides. In the case of immunoglobulins, the plate was washed four times with buffer A and 1 pmol mouse or rabbit IgG (Sigma) diluted in buffer A containing 0.1 mg/ml bovine-serum albumin was added and the plate was incubated for 60 min. The plate was washed four times with buffer A and 50 ml of protein A–AP conjugate (Sigma) diluted 1:200 in buffer B (20 mM Tris, pH 7.5, 150 mM NaCl) was added. After incubation and four washes with buffer A and two washes with buffer B, AP substrate was added as described below. For enzyme immobilization, the following substrates were prepared by manufacturers’ instructions: AP, pNPP (4-nitrophenyl phosphate, Boehringer Mannheim), l Abs 5 405 nm, BluePhos, (Kirkegaard & Perry Laboratories), l Abs 5 650 nm and AttoPhos (Boehringer Mannheim), l Ex 5 440 nm, l Em 5 550 nm; POD, ABTS (2,2´-azino-di-[3-ethylbenzthiazoline sulfonate], Boehringer Mannheim), l Abs 5 405 nm; b-Gal, ONPG (2-nitrophenyl-b-D-galactopyranoside, Boehringer Mannheim), l Abs 5 405 nm. The absorbances were determined with a Victor 1420 MultilabelCounter (Wallac) after incubation for 20 min at 37°C. Standard deviations were calculated from the signal intensities of interest. Typically, duplicate determina-

56

NIEMEYER ET AL.

tion of a particular sample revealed a percent error of about 10%. BIAcore Measurements Biospecific interaction analysis was carried out with BIAcore 1000, Pharmacia Biosensor AB (Uppsala, Sweden), as previously described (27). In brief, biotinylated capture oligomers, 10 mM in buffer C (150 mM NaCl, 30 mM sodium phosphate, 3 mM EDTA, 0.25% Triton X-100, pH 7.4) was immobilized by injection of the samples for 4 min over STV-coated sensorchips (SA-Chips, Pharmacia Biosensor AB). Loosely attached material was removed by three cycles of 10 ml 50 mM NaOH. An amount of 500 to 1000 RU of the biotinylated oligonucleotides was immobilized. Hybridization analysis was carried out with a continuous flow of buffer C at 10 ml/min and 25°C. Various concentrations of preconjugate HB21–GAM, 200 –25 nM in buffer C, were injected over the DNA-specific surface at 10 ml/ min for 4 min. After binding, the surface was regenerated by injection of 10 ml 50 mM NaOH. Controls, carried out by injecting preconjugates across surfaces, modified with noncomplementary oligonucleotides, revealed no significant increase in RU response. Injections of antigen (mouse IgG, Sigma) across chips, modified with HB21–GAM, led to an increase in RU response, typically of about 50% of RU values obtained from HB21–GAM binding. Site-Selective Immobilization on Oligonucleotide Arrays Biotinylated capture oligonucleotides were immobilized on STV-coated microplates as described above. Reiterating, nine columns were alternately coated with bA24as; bB21as, and bD21as (total of three wells per column, see also Fig. 6). The plate was subsequently washed with buffer A containing 0.1 mg/ml bovineserum albumin and 10 mM D-biotin. Nine different preconjugates were prepared by mixing 50 ml of the DNA–STV hybrids, HA24, HB21, and HD21 (100 nM in buffer A) with the same volume of biotinylated enzymes AP, bGal, or POD (100 nM in buffer A) as described above. To test nine possible permutations, three of the preconjugate solutions were combined to the following three mixtures: HA24 –bGal 1 HB21– POD 1 HD21–AP (mixture I), HA24 –POD 1 HB21– AP 1 HD21–bGal (mixture II), and HA24 –AP 1 HB21–bGal 1 HD21–POD (mixture III). Each of the mixtures was diluted with buffer A containing 0.1 mg/ml bovine-serum albumin to a final preconjugate concentration of 5 nM, and 50 ml was applied to different rows of the microplate containing a total of nine wells (see Fig. 6). After incubation and washing, successful binding was determined by applying 200 ml of three different substrates generating distinct colors,

blue for AP detection (BluePhos), green for peroxidase detection (ABTS) and yellow for b-galactosidase detection (ONPG). The reactions were carried out for 40 min at 37°C. A nanotiter plate-based oligonucleotide array was prepared on a gold-coated nanotiter plate (GeSim GmbH, Dresden) by spotting thiolated oligonucleotides tA24 and tA24as with a nanoliter-dispensing robot (28). The plates were blocked and washed as described above and subsequently incubated for 60 min in a solution containing preconjugate HA24as–AP, 2 nM in buffer A. Following a wash, AttoPhos was added as a substrate and fluorescent wells were determined on a standard UV illumination table equipped with a CCD camera (Herolab). Nanotiter plate surfaces were regenerated by treatment with aqueous NaOH solution (50 mM) and the plates were used for subsequent hybridization of preconjugate HA24 –AP, similar to the description above. RESULTS AND DISCUSSION

Coupling of DNA–STV Adapters with Biotinylated Proteins Covalent conjugates of DNA and STV, HA24, HA24as, HB21, and HD21, were synthesized from 59-thiolated oligonucleotides and recombinant streptavidin by chemical crosslinking and the chief products, hybrids containing a single oligonucleotide moiety per STV, were purified by chromatography (19). To initially study the adapter properties, the DNA–STV hybrids were coupled to various biotinylated enzymes, and the preconjugates formed were analyzed in a microplatebased, solid-phase hybridization assay (Fig. 1). In a first set of experiments, the variation of conjugation stoichiometry has been determined by coupling fixed amounts of hybrid HA24 with varying quantities of biotinylated enzymes, bGal, AP, or horseradish POD. The resulting preconjugates, HA24 –bGal, HA24 –AP, and HA24 –POD, respectively, were then allowed to bind to their corresponding antisense oligomer complement bA24as, previously immobilized on a STV-coated microplate (26) by use of STV– biotin interaction (see Materials and Methods for details). After hybridization, the amount of immobilized material was estimated from the enzymatic activity, using appropriate colorimetric substrates ONPG, pNPP, and ABTS, respectively. Negative controls were carried out by incubation of the preconjugates in wells containing a noncomplementary capture oligonucleotide (e.g., bB21as), as well as by incubation of the biotinylated enzymes only. As shown in Fig. 2, the low background signals obtained for the controls confirmed that binding occurred exclusively via specific single-stranded DNA hybridization. A stoichiometry-dependent course of signal intensities was obtained for the three enzymes: In the case of voluminous proteins, bGal and AP with

DNA-DIRECTED IMMOBILIZATION

57

FIG. 1. Schematic representation of DNA-directed immobilization (DDI). Covalent DNA–STV adducts are coupled with a biotinylated enzyme or antibody by mixing of the two compounds. As an example, biotinylated horseradish peroxidase (POD) and DNA–STV hybrid HA24 are coupled to form preconjugate HA24 –POD. Biotinylated antisense capture-oligonucleotide bA24as was immobilized on STV-coated microplates, and remaining free biotin-binding sites of the surface-bound STV are blocked with D-biotin, represented by shaded spheres. The preconjugates are then allowed to bind to their complement by formation of specific Watson–Crick base pairs. The amount of hybridized material is determined by enzyme-dependent color reaction using an appropriate substrate. In the case of antibody immobilization, the signals are generated by use of protein A–alkaline phosphatase conjugate.

relative molecular weights of 480 and 130 kDa, respectively, a 1:1 molar ratio of biotinylated enzyme and DNA–STV conjugate caused the most intense signals that are not significantly enhanced by increasing the amount of the enzyme. These findings indicate that— probably due to steric hindrance—an approximately equimolar conjugation stoichiometry of the two components occurs despite the tetravalent binding capabilities of STV. With the smaller POD (relative molecular weight of 43 kDa), maximal signal intensities are reached at also an approximately 1:1 molar coupling stoichiometry. However, the immobilization efficiency and thus the signal intensities significantly decrease upon elevating the amount of biotinylated POD during preconjugate formation. This effect may be attributed to the formation of preconjugates with a 1:2 or higher coupling ratio in which the binding of the DNA-moiety is sterically hindered. Immobilization Efficiency To estimate the immobilization efficiency, DDI was compared with a direct immobilization of the biotinylated proteins on STV-coated microplates. For this purpose, preconjugates of a 1:1 molar ratio of HA24 and biotinylated AP, bGal, or POD were allowed to hybridize to their oligomer complements. Thereby, an amount of 1.25 pmol of preconjugate was applied per well to ensure that excess analyte is present compared to approximately 0.5 pmol biotin-binding sites per well (29). In parallel reactions, similar amounts of the biotinylated enzymes were immobilized directly on STV-coated microplates via STV– biotin interaction. Controls were

carried out to confirm that enzyme immobilization is exclusively due to either DNA–DNA or STV– biotin interaction, respectively. As indicated in Fig. 3, a comparison of the two methods revealed a surprisingly high immobilization efficiency of DDI, only slightly affected from the size of the enzyme employed. The increase in signal intensity observed for DDI was 1.7fold for bGal, 1.9-fold for AP, and 2.7-fold for POD immobilization. Consistent with these findings, DNAdirected immobilization of a preconjugate of HB21 and POD led to a 2.7-fold higher efficiency than that with the direct binding of biotinylated POD to STV-coated microplates (not shown). The results indicate that reversible DDI exceeds irreversible STV– biotin-mediated immobilization. This can be a consequence of the reversibility of DNA hybridization, enabling a denser packing on formation of the enzyme layer. Furthermore, the lean structure of the rigid double-helical DNA spacer between the surface and the protein may also contribute to a larger effective surface area. Also, a higher biological activity may result from the larger distance between surface and enzyme, enabling a more homogeneous type of reaction during enzymatic substrate transformation. To further study this phenomenon, the immobilization of immunoglobulins was investigated. As carried out for the enzymes, the coupling stoichiometry-dependent immobilization efficiency of preconjugates from HA24 and F(ab) 2 fragments of either biotinylated GAM or GAR IgG have been determined, and the signal intensities were compared to those obtained from IgG immobilization via STV– biotin interaction or a direct

58

NIEMEYER ET AL.

FIG. 3. Comparison of the immobilization efficiencies obtained from biotin–STV interaction (open bars) and DDI (gray bars). Preconjugates of biotinylated enzymes, bGal, AP, or POD, and DNA– STV conjugate HA24 were prepared from equimolar amounts of the two compounds and the preconjugates were then allowed to bind to their oligomer complement bA24as, immobilized on microplates. In parallel reactions, the same amounts of the biotinylated enzymes were allowed to directly bind to STV-coated microplates. Negative controls of DDI, carried out as described in Fig. 2, and negative controls for direct STV– biotin immobilization, carried out by incubation in wells previously blocked with biotin, gave uniform signals of less than 0.2 absorbance units.

physisorption of the IgGs to microplates. To verify the functionality of immobilized antibodies, a capture step was carried out in which antigens, IgG from mouse or rabbit, were allowed to bind to surface-attached antibodies. Following incubation of a commercially available conjugate of AP and protein A that specifically binds to the Fc portion of IgG, the amounts of material immobilized were estimated from AP-dependent color reaction. As shown in Fig. 4, the results are consistent with those obtained for enzyme immobilization. Due to the size of the F(ab) 2 fragments (relative molecular weight of 100 kDa), a plateau of immobilization efficiency is reached at an approximately 1:1 molar ratio of antibodies and DNA–STV hybrid. Also, increasing the

FIG. 2. Investigation of the preconjugation stoichiometry. Fixed amounts of DNA–STV hybrid HA24 were mixed with varying amounts of the biotinylated enzymes, bGal (A.), AP (B.), or POD (C.). The molar ratio of hybrid to biotinylated AP was in the range of 1: 0.063 up to 1:7. Hybridization efficiencies of the preconjugates were determined as shown in Fig. 1 and negative controls (nc) were carried out by incubation of the highest coupling ratio in wells coated with noncomplementary capture oligomers. The histograms represent relative signal intensities, normalized as a percentage fraction of the signal obtained for a 1:1 molar coupling ratio.

DNA-DIRECTED IMMOBILIZATION

59

FIG. 4. DNA-directed immobilization of goat antibodies directed against mouse IgG (GAM, left) or rabbit IgG (GAR, right). Comparison of immobilization efficiencies: The histograms represent signal intensities obtained from the binding of protein A–AP conjugate to microplate-bound mouse or rabbit IgG, immobilized by capturing with GAM or GAR. The latter were immobilized by direct physisorption, biotin–STV interaction, or DDI. For both antibodies, DDI induces the highest binding capacities for antigens. Negative controls (nc) for DDI and biotin–STV were carried out as described in the legend to Fig. 3; for direct physisorption, noncomplementary antigens were applied during antigen capture. The variation of preconjugation stoichiometry during DDI is shown in the rightmost sets of bars. The 1:1 molar ratios of HA24 and biotinylated IgGs are indicated by arrowheads.

amount of biotinylated antibodies during the coupling did not result in signal decrease, suggesting that the formation of preconjugates containing two or more antibodies is scarce. A comparison of the direct immobilization methods with DDI confirms the superior performance of the latter. An approximately twofold efficiency enhancement was observed compared to immobilization via biotin–STV interaction on STV-coated microplates. Moreover, an even greater improvement was observed compared to an immediate coating of IgG directly onto the polystyrene microplate surfaces (Fig. 4). In addition to confirming the data obtained from enzyme immobilization, these results suggest a general superiority of immobilization methods based on specific biomolecular interactions rather than physisorption-based techniques. This is possibly due to the milder adsorption conditions and also the binding properties of STV-coated surfaces. The latter allow a fixation of an IgG by single-point contacts and therefore minimize damages from multiple surface contactinduced denaturation that occurs during conventional physisorption.

Reversibility To elucidate the reversibility of DDI, the hybridization of a preconjugate from DNA–STV hybrid and biotinylated IgG has been studied by biomolecular interaction analysis (BIA) using a commercial biosensor based on surface plasmon resonance (BIAcore) (30). This method has previously been used to characterize the hybridization properties of the DNA–STV conjugates (27). With this technique, the refractive index within a gold sensor chip-immobilized dextran matrix is continuously monitored, plotted against time, and presented in a sensorgram. The change of SPR response, measured in resonance units (RU), allows a quantification of interaction between chip-immobilized and liquid phase-diluted binding partners. Here, a biotinylated antisense capture oligonucleotide, bB21as, was immobilized on a streptavidin-coated sensor chip and binding measurements were carried out with the preconjugate HB21–GAM at several analyte concentrations (Fig. 5). Controls carried out by injecting the preconjugate over sensor chips modified with noncomplementary capture oligonucleotide bA24as re-

60

NIEMEYER ET AL.

FIG. 5. BIAcore measurements of reversibility and reproducibility of DNA-directed immobilization of immunoglobulins. Hybridization of a preconjugate of biotinylated immunoglobulin and DNA–STV conjugate (HB21–GAM) to SPR sensor chip-immobilized complementary capture oligonucleotide (bB21as) at analyte concentrations ranging from 50 to 6.3 nM. Each hybridization cycle was followed by regeneration of the sensor chip using NaOH treatment. Subsequent hybridizations carried out with a second set of analyte samples (dashed lines) indicate excellent reproducibility of DNA-directed antibody immobilization.

vealed no significant changes in SPR response (not shown), indicating that the binding observed is due exclusively to specific single-stranded DNA hybridization, a result similar to that obtained from the microplate assays described above. As shown in Fig. 5, the amount of preconjugate immobilized increases linearly with the concentration of HB21–GAM injected over the sensor chip and is highly reproducible. The binding capacity of the antibody-functionalized sensor chip for antigen, determined by subsequent injection of mouse IgG, was in the range of 50% of the RU obtained from preconjugate binding (not shown). Also, series of more than 50 cycles of hybridization were carried out with various enzymes and antibodies to demonstrate that the immobilized proteins can be completely removed by alkaline denaturation of the DNA double helix. These results show that DDI allows for loading distinct amounts of biomolecules and also complete regeneration of DNA-coated surfaces. Although not yet encountered in our laboratory, potential problems of this

method could arise in some applications as a result from nucleases in particular analyte samples that could degrade the DNA–STV conjugates or sensorbound capture oligonucleotides. Nevertheless, the advance in immobilization technique described should be especially advantageous for recovery and reconfiguration of expensive BIAcore sensor chips. Site-Selective Immobilization of Enzymes on DNA Arrays To demonstrate site selectivity of DDI, a model DNA array was simulated on a microplate using three different antisense capture oligonucleotides, bA24as, bB21as, and bD21as, immobilized via biotin–STV interaction. This array was used for selective immobilization of three different preconjugates consisting of HA24, HB21, or HD21, each of which was previously conjugated to a different biotinylated enzyme, bGal, AP, or POD. Mixtures of three preconjugates, consist-

DNA-DIRECTED IMMOBILIZATION

61

FIG. 6. Site selectivity of DDI, demonstrated by parallel immobilization of enzyme preconjugates to a model DNA array on a microplate. Each three-by-three block contains three columns of a different capture oligonucleotide, bA24as, bB21as, or bD21as. Mixtures, containing three distinct preconjugates of HA24, HB21, and HD21 coupled to AP, bGal, or POD were applied in rows to the three blocks. Successful binding was tested by the addition of an enzyme-specific substrate to the three blocks: ABTS for POD, ONPG for bGal, or BluePhos for AP detection. The fact that only one dominant signal appears in each row and each column of a block demonstrates the site selectivity of this immobilization procedure.

ing of, e.g., HA24 –AP, HB21–bGal, and HD21–POD, were then allowed to bind to the DNA-coated microplate. The results of this experiment are illustrated in Fig. 6. Each of the 3 3 3 blocks contain three different capture oligonucleotides, arranged in columns, to which the different preconjugate mixtures have been applied in rows. Different enzyme-specific substrates were then applied to the blocks to detect POD, bGal, or AP activity. For individual 3 3 3 blocks, each row and each column, is expected to have only one dominant signal if the self-sorting via DDI is successful. In fact, this result was obtained. Relative signal intensities, however, vary between the blocks as a consequence of the substrate-characteristic signal-to-noise ratios. Highest values were obtained with AP-substrate BluePhos and lowest with bGal-substrate ONPG. The signals derived from the binding of HD21-containing preconjugates are generally lower than HA24 or HB21 preconjugates, indicating lower immobilization rates due to the sequence-specific hybridization characteristics of D21 oligonucleotide (27). For optimization of DDI, this point is of particular interest since both chain length and GC content of the oligonucleotides influence the immobilization efficiency. As previously shown

(27), a shortening of the oligonucleotide’s length results in decreased binding. Moreover, the solid-phase hybridization efficiency is decreased with increasing stability of sequence-specific secondary structures, i.e., the formation of homodimers and intramolecular hairpin loops, which preferentially occur in GC-rich oligomers. As a further example, site-selectivity was demonstrated using a microstructured DNA array based on a nanotiter plate (NTP). These plates contain 617 wells, each of a volume of 400 nl, arranged on a silica plate of 2 3 2 cm edge lengths. Thiolated capture oligonucleotides have been spotted on a gold-coated NTP with a nanoliter-pipetting robot (28), 3 generating an array consisting of two different 24-mer oligonucleotides, tA24 and tA24as, respectively. A pattern of letters was spotted with oligonucleotide tA24 and a background 3

The nanoliter-dispensing robot was developed by Bremer Institut fu¨r Angewandte Strahltechnik (BIAS GmbH, Bremen). Details have recently been presented (28). Further information is available from M. Wolf, Bremer Institut fu¨r Angewandte Strahltechnik GmbH, Klagenfurter Str. 2, D-28359 Bremen, Fax: 149 (421) 218 –5063, E-mail: [email protected].

62

NIEMEYER ET AL.

FIG. 7. Site selectivity of DDI, demonstrated by successive immobilization of AP preconjugates to a model DNA array on a nanotiter plate (NTP), containing 617 wells with a volume of 400 nl on a 2 3 2-cm area. By use of a nanoliter-dispensing robot (28), an array consisting of two different 24-mer oligonucleotides, tA24 and tA24as, was generated by spotting individual patterns of letters with oligonucleotide tA24 and a background pattern with oligonucleotide tA24as. Hybridization of preconjugate HA24as–AP was used for the visualization of NTP-attached oligonucleotide tA24 (left image). Regeneration of the surface and subsequent hybridization with preconjugate HA24 –AP leads to the image shown at the right.

pattern was spotted with oligonucleotide tA24as. Subsequently, the NTP was incubated in a solution of preconjugate HA24as–AP and the visualization of complementary NTP-attached oligonucleotides, and thus the pattern of letters, was achieved by addition of fluorogenic AP-substrate Attophos (Fig. 7, left image). Regeneration of the surface by NaOH treatment and subsequent hybridization with preconjugate HA24 – AP, which is complementary to tA24as, led to the image of the background pattern shown at the right-hand side in Fig. 7.

mobilization of multiple compounds in a single reaction. This site selectivity should allow the fabrication of dense, highly functionalized sensor surfaces useful, e.g., for parallel detection assays. Since DDI is not limited in terms of the number and nature of molecular compounds to be arrayed, one may anticipate the generation of a wide variety of micro- and nanostructures useful not only in biomedical and analytical applications but also for other fields of biotechnology and synthetic chemistry. ACKNOWLEDGMENTS

CONCLUSIONS

DNA–STV hybrid molecules were used as coupling agents for the noncovalent attachment of oligonucleotide fragments to various biotinylated enzymes or antibodies. The single-stranded DNA-tagged proteins were immobilized to complementary surface-bound capture oligonucleotides by means of specific nucleic acid hybridization. The exploration of DNA-directed immobilization, carried out in this study, revealed a variety of striking advantages compared to conventional immobilization techniques: DDI allows a reversible functionalization of surfaces with biomolecules and thus improves the opportunities to fabricating reusable bio- and immunosensor chips. Due to short reaction times and also mild adsorption conditions even sensitive biomolecules seem to retain their full biological activity, and the immobilization efficiency of DDI is superior to comparable standard methods, such as direct binding to STV-coated microplates or physisorption. Furthermore, DDI permits the simultaneous im-

The authors thank Deutsche Forschungsgemeinschaft (DFG), Fonds der Chemischen Industrie and To¨njes-Vagt-Stiftung, Bremen, for financial support. We thank F. Holtkamp and M. Wolf from BIAS GmbH, Bremen, for collaboration on the NTP-based arrays, R. M. J. Hoedemakers, University of Groningen, for BIAcore measurements, and Michael Adler for valuable suggestions and discussions.

REFEERENCES 1. Scouten, W. H. (1995) Bioprocess Technol. 20, 233–265. 2. Tyagi, R., Singh, D. K., and Gupta, M. N. (1994) Biotechnol. Appl. Biochem. 20, 93–99. 3. Tyagi, R., and Gupta, M. N. (1994) Process. Biochem. 29, 443– 448. 4. Liu, X. C., and Scouten, W. H. (1996) J. Mol. Recognit. 9, 462– 467. 5. Schmitt, L., Dietrich, C., and Tampe, R. (1994) J. Am. Chem. Soc. 116, 8485– 8491. 6. Gritsch, S., Neumaier, K., Schmitt, L., and Tampe, R. (1995) Biosens. Bioelectron. 10, 805– 812. 7. Gershon, P. D., and Khilko, S. (1995) J. Immunol. Method. 183, 65–76.

DNA-DIRECTED IMMOBILIZATION 8. Ho, C. H., Limberis, L., Caldwell, K., and Stewart, R. (1997) Abstr. Am. Chem. Soc. 214, 44-Coll. 9. Anspach, F. B., and Altmannhaase, G. (1994) Biotechnol. Appl. Biochem. 20, 323–333. 10. Batistaviera, F., Barbieri, M., Ovsejevi, K., Manta, C., and Carlsson, J. (1991) Appl. Biochem. Biotech. 31, 175–195. 11. Carlsson, J., and Batistaviera, F. (1991) Biotechnol. Appl. Biochem. 14, 114 –120. 12. Sara, M., and Sleytr, U. B. (1992) Appl. Microbiol. Biotechnol. 38, 147–151. 13. Brena, B. M., Ovsejevi, K., Luna, B., and Batistaviera, F. (1993) J. Mol. Catal. 84, 381–390. 14. Batistaviera, F., Manta, C., and Carlsson, J. (1994) Appl. Biochem. Biotech. 44, 1–14. 15. Diaz, T., Stahl, U., Batistaviera, F., and Carlsson, J. (1995) Biotechnol. Tech. 9, 533–538. 16. Phelps, M. R., Hobbs, J. B., Kilburn, D. G., and Turner, R. F. B. (1994) Biotechnol. Progr. 10, 433– 440. 17. Phelps, M. R., Hobbs, J. B., Kilburn, D. G., and Turner, R. F. B. (1995) Biotechnol. Bioeng. 46, 514 –524. 18. Scouten, W. H., and Konecny, P. (1992) Anal. Biochem. 205, 313–318. 19. Niemeyer, C. M., Sano, T., Smith, C. L., and Cantor, C. R. (1994) Nucleic Acids Res. 22, 5530 –5539.

63

20. Rozsnyai, L. F., Benson, D. R., Fodor, S. P. A., and Schultz, P. G. (1992) Angew. Chem. Int. Ed. Engl. 31, 759. 21. Lockhart, D. J., Dong, H., Byrne, M. C., Follettie, M. T., Gallo, M. V., Chee, M. S., Mittmann, M., Wang, C., Kobayashi, M., Horton, H., and Brown, E. L. (1996) Nature Biotechnol. 14, 1675–1680. 22. Gonza´lez, M., Bagatolli, L. A., Echabe, I., Arrondo, J. L. R., Argaran˜a, C. E., Cantor, C. R., and Fidelio, G. D. (1997) J. Biol. Chem. 272, 11288 –11294. 23. Wilchek, M., and Bayer, E. A. (1990) Methods Enzymol. 184, 51– 67. 24. Wilchek, M., and Bayer, E. A. (1990) Methods Enzymol. 184, 14 – 45. 25. Niemeyer, C. M., Bu¨rger, W., and Peplies, J. (1998) Angew. Chem. Int. Ed. Engl. 37, 2265–2268. 26. Niemeyer, C. M., Adler, M., and Blohm, D. (1997) Anal. Biochem. 246, 140 –145. 27. Niemeyer, C. M., Bu¨rger, W., and Hoedemakers, R. M. J. (1998) Bioconjugate Chem. 9, 168 –175. 28. Boldt, L., Gersdorf, H., Niemeyer, C. M., Holtkamp, F., Bischoff, R., Sa¨lter, W., Adler, M., Kayser, O., Wolf, M., Ju¨ptner, W., and Blohm, D. (1998) BIOSENSORS ’98, 5th World Congress on Biosensors, Berlin , June 3–5. 1998, poster presentation. 29. Adler, M. (1996) Diploma thesis, Universita¨t Bremen. 30. Malmquist, M. (1993) Nature 361, 186 –187.