Anal. Chem. 2006, 78, 1515-1519
DNA-Directed Protein Immobilization for Simultaneous Detection of Multiple Analytes by Surface Plasmon Resonance Biosensor Christina Boozer, Jon Ladd, Shengfu Chen, and Shaoyi Jiang*
Department of Chemical Engineering, University of Washington, Seattle, Washington 98195
A versatile multichannel biosensor surface is prepared by site-directed immobilization of single-stranded DNAprotein conjugates onto a patterned self-assembled monolayer composed of ssDNA thiols and oligo(ethylene glycol)terminated thiols. The conjugates each consist of an antibody chemically linked to a unique ssDNA target with a sequence complementary to the surface-bound ssDNA probes and are immobilized on the surface via sequencespecific hybridization. The exceptional specificity of DNA hybridization combined with the diversity of available sequences makes this platform perfect for multichannel sensors. Once the surface is patterned with the appropriate probe sequences, sequence-specific hybridization sorts out the target conjugates and directs them to the appropriate spots on the surface. Previously (Boozer, C. L.; Ladd, J.; Chen, S.; Yu, Q.; Homola, J.; Jiang, S. Anal. Chem. 2004, 76, 6967-6972), we performed proof-ofconcept experiments demonstrating the feasibililty of using DNA-directed protein immobilization to produce a single channel biosensor. In this work, we extend this technique and employ DNA-directed protein immobilization to functionalize a multichannel biosensor, which was used for the simultaneous detection of a set of three fertility hormones: human chorionic gonadotropin, human luteinizing hormone, and follicle stimulating hormone by surface plasmon resonance sensor.
approaches such as covalent attachment,7,8 protein A,8-10 and biotin/streptavidin bridges8,11 are appropriate for the generation of single-channel sensor surfaces but require additional patterning steps for the preparation of multiplexed sensors. Although contact printing and spotting12,13 can be used to pattern proteins on a surface, these approaches frequently lead to decreased protein activity due to mechanical shearing, drying, and surface effects.13,14 To overcome some of these limitations, various investigators have begun employing DNA hybridization to spatially assemble an assortment of molecular components, such as nanoparticles, proteins, and polypeptides.15-18 Previously, we presented a method for preparing a biosensor via site-directed immobilization of protein-DNA conjugates onto a mixed self-assembled monolayer (SAM) composed of singlestranded DNA thiols and oligo(ethylene glycol) (OEG)-terminated thiols.19 The protein conjugates consist of an antibody chemically linked to a ssDNA target with a sequence complementary to the surface-bound ssDNA probes and are immobilized on the surface via sequence-specific hybridization. The resulting sensor surface has been shown to be specific and highly sensitive. The lower limit of detection for the pregnancy hormone human chorionic gonadotropin (hCG) was determined to be 0.1 ng/mL, 50-fold lower than we had measured using biotinylated antibodies. Although this improvement in sensitivity was encouraging, the true value of this approach lies in its potential for producing multichannel sensors.
There is an urgent need for robust multichannel sensors, capable of simultaneously detecting and identifying multiple analytes, for applications such as medical diagnostics,1,2 food safety,2,3 homeland security,4,5 and environmental monitoring.2,6 One of the most challenging aspects of developing multichannel sensors is patterning the antibodies (or other recognition elements) on the sensor surface. Common protein immobilization
(7) Cutler, P. Proteomics 2003, 3, 3-18. (8) Vijayendran, R. A.; Leckband, D. E. Anal. Chem. 2001, 73, 471-480. (9) Nisnevitch, M.; Firer, M. A. J. Biochem. Biophys. Methods 2001, 49, 467480. (10) Gersten, D. M.; Marchalonis, J. J. J. Immunol. Methods 1978, 24, 305309. (11) Boozer, C. L.; Yu, Q. M.; Chen, S. F.; Lee, C. Y.; Homola, J.; Yee, S. S.; Jiang, S. Y. Sens. Actuators, B: Chem. 2003, 90, 22-30. (12) MacBeath, G.; Schreiber, S. L. Science 2000, 289, 1760-1763. (13) Huang, R. P. Frontiers Biosci. 2003, 8, D559-D576. (14) Liotta, L. A.; Espina, V.; Mehta, A. I.; Calvert, V.; Rosenblatt, K.; Geho, D.; Munson, P. J.; Young, L.; Wulfkuhle, J.; Petricoin, E. F. Cancer Cell 2003, 3, 317-325. (15) Niemeyer, C. M. Trends Biotechnol. 2002, 20, 395-401. (16) Mirkin, C. A.; Letsinger, R. L.; Mucic, R. C.; Storhoff, J. J. Nature 1996, 382, 607-609. (17) Weng, S.; Gu, K.; Hammond, P. W.; Lohse, P.; Rise, C.; Wagner, R. W.; Wright, M. C.; Kuimelis, R. G. Proteomics 2002, 2, 48-57. (18) Ladd, J.; Boozer, C. L.; Chen, S.; Yu, Q.; Homola, J.; Jiang, S. Langmuir 2004, 20, 8090-8095. (19) Boozer, C. L.; Ladd, J.; Chen, S.; Yu, Q.; Homola, J.; Jiang, S. Anal. Chem. 2004, 76, 6967-6972.
* To whom correspondence should be addressed. E-mail: sjiang@ u.washington.edu. (1) Vo-Dinh, T.; Cullum, B. Fresenius J. Anal. Chem. 2000, 366, 540-551. (2) Rich, R. L.; Myszka, D. G. J. Mol. Recognit. 2001, 14, 273-294. (3) Leonard, P.; Hearty, S.; Brennan, J.; Dunne, L.; Quinn, J.; Chakraborty, T.; O’Kennedy, R. Enzyme Microb. Technol. 2003, 32, 3-13. (4) Cirino, N. M.; Musser, K. A.; Egan, C. Exp. Rev. Mol. Diagn. 2004, 4, 841857. (5) Paddle, B. M. Biosens. Bioelectron. 1996, 11, 1079-1113. (6) Rodriguez-Mozaz, S.; Marco, M. P.; de Alda, M. J. L.; Barcelo, D. Anal. Bioanal. Chem. 2004, 378, 588-598. 10.1021/ac051923l CCC: $33.50 Published on Web 01/13/2006
© 2006 American Chemical Society
Analytical Chemistry, Vol. 78, No. 5, March 1, 2006 1515
Figure 1. Schematic of DNA-directed immobilization of DNA-protein conjugates. Table 1. DNA Sequencesa
a
name
sequence
description
A c-A B c-B C c-C D c-D E c-E
5′ TCC TGT GTG AAA TTG TTA TCC GCT 3′ 5′ AGC GGA TAA CAA TTT CAC ACA GGA 3′ 5′ GTA ATC ATG GTC ATA GCT GTT 3′ 5′ AAC AGC TAT GAC CAT GAT TAC 3′ 5′ GTC CTC GCC TAG TGT TTC ATT G 3′ 5′ CAA TGA AAC ACT AGG CGA GGA C 3′ 5′ CAG CCA AGA TTC TTT TAC CGC C 3′ 5′ GGC GGT AAA AGA ATC TTG GCT G 3′ 5′ GGT CCG GTC ATA AAG CGA TAA G 3′ 5′ CTT ATC GCT TTA TGA CCG GAC C 3′
probe sequence used to prepare DNA SAMs target sequence conjugated to anti-hCG probe sequence used to prepare DNA SAMs target sequence conjugated to anti-hLH probe sequence used to prepare DNA SAMs target sequence conjugated to anti-FSH candidate sequence used to prepare DNA SAMS complementary sequence to D candidate sequence used to prepare DNA SAMS complementary sequence to E
All sequences are thiolated at the 5′ end.
The most useful feature of DNA as a linking agent is the diversity of available sequences combined with the specificity of base pairing. Effectively, each different sequence acts as an independent linker. If a set of proteins is conjugated to a corresponding set of target ssDNA sequences, and a surface is patterned with the appropriate complementary oligonucleotide probes, a mixture of the conjugates can be applied to the surface and sequence-specific hybridization will direct the target conjugates to the appropriate spots on the surface (see Figure 1). In one simple step, a DNA array is converted into a protein array, or multichannel biosensor. In this work, we use this concept to extend our previous work with DNA-directed immobilization to the generation of a multichannel SPR biosensor. After performing detailed control experiments to confirm the specificity of antibody immobilization, a set of three female hormones related to fertility was simultaneously detected from a common solution. Human luteinizing hormone (hLh) and follicle stimulating hormone (FSH) are markers of menstrual cycle phase and are used to test fertility and ovulation timing.20,21 hCG tests confirm pregnancy and may be used to diagnose pregnancy-related complications.22 (20) Berger, M. J.; Taymor, M. L. Am. J. Obstet. Gynecol. 1971, 111, 708-&. (21) Filicori, M.; Cognigni, G. E.; Taraborrelli, S.; Spettoli, D.; Ciampaglia, W.; de Fatis, C. T.; Pocognoli, P. J. Clin. Endocrinol. Metab. 1999, 84, 26592663.
1516
Analytical Chemistry, Vol. 78, No. 5, March 1, 2006
EXPERIMENTAL SECTION Materials. Three sets of oligonucleotides and antibodies were used to prepare three unique DNA-antibody conjugates. The DNA sequences and descriptions can be found in Table 1. Briefly, thiolated ssDNA A, B, and C were used to prepare the ssDNA probe surfaces, while the corresponding thiolated complements, c-A, c-B, and c-C were cross-linked to anti-hCG, anti-hLH, and antiFSH, respectively. Both unmodified plain and modified (thiolated) DNA sequences were custom synthesized by Synthegen. Thiolated oligos were shipped in a solution of 0.1 M dithiothreitol and repurified with a NAP-10 column as outlined by the vendor. After purification, the final oligo concentration was measured by UV spectrophotometer. OEG-terminated thiols (HS(CH2)11(OCH2)4OH) were purchased from ProChimia. Anti-β-hCG, anti-β-hLH, and anti-βFSH (all murine monoclonal, protein A purified, IgG1 antibodies) and their respective antigens were all purchased from Scripps Laboratory. Synthesis of DNA-Antibody Conjugates. The DNAantibody conjugates were synthesized by chemically cross-linking a thiolated single-stranded DNA with the respective antibody. The antibodies (5 mg/mL) were reacted with a 10-fold molar excess of sulfosuccinimidyl 4-(p-maleimidophenyl)butyrate (Pierce) in (22) Channing, C. P.; Kammerma, S. Biol. Reprod. 1974, 10, 179-198.
PBS (100 mM phosphate, pH 7.4, 138 mM NaCl, 2.7 mM KCl). After incubation for 30 min at room temperature, the derivatized antibodies were desalted by ultrafiltration (30 000 MW cutoff membrane; Millipore) and the buffer was changed to PBE (phosphate-buffered EDTA; PBS with 5 mM EDTA). Thiolated ssDNA was added to the antibodies at a 1:1 ratio, and the mixture was incubated at room temperature for 30 min. Unreacted ssDNA was removed by ultrafilatration (100 000 MW cutoff), and the purification was verified by nondenaturing PAGE. Product yields were quantified using a UV spectrophotometer at wavelengths of 260 and 280 nm. Substrate Preparation. Surface plasmon resonance (SPR) sensor substrates were prepared by coating clean BK-7 glass substrates with a 2-nm adhesion layer of chromium followed by a 50-nm layer of gold by electron beam evaporation at pressures below 1 × 10-6 Torr. Preparation of ssDNA Probe Surfaces. Prior to surface functionalization, all SPR chips were rinsed with ethanol and water, blown dry with nitrogen, and cleaned by 20 min of UV/ozone. After UV treatment, chips were rinsed again with water and ethanol and dried under a stream of N2. Chips were functionalized immediately following cleaning. For the uniform chips (not patterned), mixed ssDNA/oligo(ethylene glycol) (OEG) SAMs were formed by immersing clean Au chips in a 1.0 M KH2PO4 buffer solution of 100 nM ssDNA and 5 µM OEG thiol. Following overnight assembly, the samples were rinsed thoroughly with water and dried by nitrogen. Multichannel probe surfaces were patterned using a patterning assembly composed of a flow cell and an alignment base. Once the clean chip was sandwiched between the flow cell and alignment base, 100 µL of ssDNA/OEG (or just OEG) thiol solution was injected into each channel. The total thiol concentration was varied as described below, but all other assembly conditions were the same as for a uniform chip. To prevent solvent evaporation, the entire patterning assembly was sealed in a humidity chamber. After overnight assembly, the patterning flow cell was flushed with copious amounts of water before the chip was removed, rinsed further, dried, and mounted directly on the instrument. Alignment pins on the instrument ensured agreement between the patterned surface and the flow cell. Surface Plasmon Resonance Instrumentation and Operation. The SPR sensor setup has been described previously.11 Briefly, it is a custom-built instrument based on the Kretschmann configuration of the attenuated total reflection method. The glass side of the gold-coated substrate is index matched to the prism coupler while the functionalized surface is mechanically pressed against an acrylic flow cell with a Mylar gasket. A polychromatic light beam is directed through the prism and the glass substrate and excites surface plasma waves at the metal-dielectric interface. The reflected light is analyzed with a spectrograph. For the multichannel detections, an eight-channel SPR based on wavelength division multiplexing (described elsewhere) was used.23,24 Only four channels were measured at any given time, and each chip was used for two independent experiments. All SPR experiments began with a buffer baseline (TE-NaCl: 10 mM Tris-HCl with 1 mM EDTA, NaCl as described below, pH 7.2), and buffer was run again after each protein or DNA step. (23) Homola, J.; Vaisocherova, H.; Dostalek, J.; Piliarik, M. Methods 2005, 37, 26-36. (24) Dostalek, J.; Vaisocherova, H.; Homola, J. Sens. Actuators, B: Chem. 2005, 108, 758-764.
Table 2. Comparison of the Hybridization Efficiency of DNA Sequence Candidatesa probe sequence
SPR response (nm) to complement hybridization
A B C D E
2.4 1.9 2.0 1.8 1.0
a ssDNA/OEG SAMs were prepared with each probe sequence, and SPR was used to measure hybridization of the corresponding complement.
For the complement hybridization experiments, TE with 1 M NaCl was used. To preserve protein activity, all experiments involving antibody conjugates were performed in TE with 150 mM NaCl. Solutions were flowed at a rate of 50 µL/min. Antibody conjugate and antigen solutions were prepared at 20 and 1 µg/mL, respectively. All SPR sensorgrams show the change in SPR wavelength (nm) relative to the initial SPR baseline. For the instruments used in this work, an SPR wavelength shift of 1 nm corresponds to a protein coverage of ∼17 ng/cm2. RESULTS AND DISCUSSION Sequence Selection. Simultaneous detection of a set of three analytes requires a set of three independent DNA probe sequences. The sequences should display high hybridization efficiency (for maximum conjugate immobilization) and no crossreactivity (for maximum specificity). Sequences A and B were used in the previous single-channel measurements and were found to perform well. To find a third sequences, three additional sequence candidates (C-E) were chosen from the literature25 and screened for hybridization efficiency and cross-reactivity. Probe surfaces were prepared with each sequence, and hybridization of the respective complements was measured by SPR. Table 2 compares the hybridization of the candidate sequences to the original sequences, A and B. As C was found to hybridize the most, this sequence was chosen. To test for cross-reactivity, chips were prepared with the three chosen sequences (A-C), and their response to all of the complement sequences was measured. Figure 2 shows the resulting SPR spectrographs indicating no cross-hybridization between the three sequences. Conjugate Immobilization. Once a complete set of DNA sequences was chosen, the DNA-antibody conjugates were prepared. Antibodies corresponding to the three analytes (hCG, hLH, FSH) were each conjugated to one of the ssDNA target sequences (c-A, c-B, c-C) (Table 1). Table 3 compares the SPR response of each of the ssDNA probe surfaces to each of the DNA-antibody conjugates. Each probe surface immobilizes large amounts of its corresponding complementary conjugate and negligible amounts of the noncomplementary conjugates. This set of experiments clearly demonstrates that immobilization of the conjugates is controlled exclusively by sequence-specific hybridization. The nonfouling OEG background prevents the protein segment of the conjugates, or any other protein for that matter, (25) Feldkamp, U.; Wacker, R.; Schroeder, H.; Banzhaf, W.; Niemeyer, C. M. Chemphyschem 2004, 5, 367-372.
Analytical Chemistry, Vol. 78, No. 5, March 1, 2006
1517
Table 4. Effect of Thiol Concentration on DNA Patterning and Conjugate Immobilizationa SPR response (nm) to target conjugate concn ratio
c-A/anti-hCG
c-C/anti-FSH
1 2 5 10
9.2 5.8 6.0 10.7
2.2 1.0 1.3 0
a Probe A DNA surfaces were patterned with various thiol concentration ratios, and the SPR response to complementary (c-A/anti-hCG) and noncomplementary (c-C/anti-FSH) conjugates was measured.
Figure 2. Test of cross-hybridization between DNA sequences. Probe surfaces were prepared with each of the three DNA sequences (A, B, C). Hybridization of each surface to all of the complements (c-A, c-B, c-C) was measured by SPR. Note that there is no crosshybridization between the sequences. Table 3. Test of Cross-Reactivity between Conjugatesa
probe sequence A B C
SPR response (nm) to target conjugates c-A/anti-hCG c-B/anti-hLH c-C/anti-FSH 11.0 0.3 0.1
0.0 8.1 0.0
0.0 0.0 9.4
a Probe surfaces were prepared with each of the three DNA sequences (A-C) and were exposed to all three conjugates. Only the complementary conjugates were immobilized on each respective probe surface.
from nonspecifically binding to the surface, and the specificity of DNA hybridization ensures proper placement of the conjugates. Patterning the DNA Probe Surfaces. Multichannel sensor chips were patterned with the three probe sequences and an OEG reference channel. Initially, the patterned chips were prepared in the same manner as the single-channel chips, with one simple change to the protocol. Rather than submerging the entire chip into a single thiol solution, a flow cell was used to deliver each of the various thiol solutions to different spots on the chip surface. All other assembly conditions were maintained. Unfortunately, it was found that the patterned chips behaved much differently from their single-channel counterparts prepared by the standard immersion protocol. The patterned probe surfaces exhibited lower conjugate hybridization and significantly higher nonspecific binding. To improve the performance of the patterned chips, it was necessary to adjust the patterning protocol. We speculated that the small volume of thiol solution (100 µL) used to pattern each spot did not contain enough thiol molecules to form a complete SAM, thus resulting in low conjugate immobilization and high nonspecific binding. To test this hypothesis, the total thiol concentration was increased, while maintaining the relative concentrations of ssDNA and OEG thiols. A ssDNA probe surface was patterned with sequence A using 1, 2, 5, and 10 times the total thiol concentration used in the standard immersion protocol. Table 4 shows the SPR responses to the complementary conjugate-A and a noncomplementary conjugate-C. At concentration 1518 Analytical Chemistry, Vol. 78, No. 5, March 1, 2006
Figure 3. Conjugate immobilization on a patterned DNA chip. A chip was patterned as described, and one by one, each conjugate was sequentially flowed over all of the channels. The conjugates only bound in the channel with the corresponding ssDNA probe sequence (probe A + c-A/anti-hCG, probe B + c-B/anti-hLH, probe C + c-C/ anti-FSH).
ratios up to 5, we observe poor immobilization of the complementary conjugate and high nonspecific binding of the noncomplementary conjugate. At a concentration ratio of 10, the surface immobilizes as much conjugate-A as a nonpatterned singlechannel chip, and nonspecific binding of noncomplementary conjugate is eliminated. For the multichannel detections, all probe surfaces are patterned using a concentration ratio of 10. Control Experiments for Simultaneous Detection of Multiple Analytes. Using the modified patterning protocol developed above, probe surfaces were patterned with the three probe sequences and an OEG reference channel. One by one, each of the conjugates was sequentially run over all of the probe surfaces. As shown in Figure 3, conjugate-A only bound to the probe-A surface, conjugate-B only bound to the probe-B surface, and conjugate-C only bound to the probe-C surface. This confirms the sequence specificity of the antibody immobilization and the integrity of the patterning procedure. Such thorough control experiments are crucial to ensure that the behavior observed on the single-channel chips can be reproduced on the patterned probe surfaces. Since it has been clearly shown that antibody immobilization is strictly controlled by DNA hybridization, it is not necessary to immobilize the conjugates individually. Rather, a cocktail of all three conjugates can be applied to all of the channels simulta-
Figure 4. Antigen detection on a patterned DNA chip after simultaneous conjugate immobilization. After simultaneously immobilizing all three conjugates from a common solution, each of the respective antigens were sequentially flowed over all of the channels. Because the conjugates were only immobilized in their proper channels (probe A + c-A/anti-hCG, probe B + c-B/anti-hLH, probe C + c-C/anti-FSH), the antigens were only detected in those channels (A + hCG, B + hLH, C + FSH).
Figure 5. Simultaneous conjugate immobilization followed by simultaneous antigen detection on a patterned DNA chip. A cocktail of conjugates (c-A/anti-hCG, c-B/anti-hLH, c-C/anti-FSH) is used to convert a patterned probe surface into a patterned antibody surface in one step, followed by simultaneous detection of all three antigens (hCG, hLH, FSH) from a common solution.
neously and sequence-specific hybridization will sort out the conjugates and direct them to the appropriate spots (see Figure 1). Figure 4 shows the detection of hCG, hLH, and FSH after simultaneously immobilizing all three conjugates (not shown). As with the previous control experiments, each analyte is sequentially run over all of the probe surfaces, but only binds in the appropriate channel. Clearly, for the analytes to display this level of specificity, two criteria must be met. First, the respective antibody conjugates were immobilized only in the intended channels, and second, the OEG background prevents nonspecific binding of the analytes. Simultaneous Detection of Multiple Analytes. Figure 5 shows the simultaneous immobilization of all three conjugates, followed by the simultaneous detection of all three analytes: hCG, hLH, and FSH. Note that this type of detection was only possible
after all of the preceding control experiments were performed to test and confirm the specificity of DNA-directed immobilization and the protein-resistant OEG background. For this proof-of-concept study, high analyte concentrations (1 µg/mL) were used. In our previously reported single-channel paper, hCG was used to test the sensor sensitivity and the lower limit of detection was found to be 0.1 ng/mL hCG. Comparison of the SPR spectrographs for the detection of 1 µg/mL hCG obtained using the old single-channel sensor19 and our new patterned multichannel sensor shows similar SPR shifts, suggesting that the two sensors should exhibit similar levels of sensitivity. Advantages of DNA-Directed Protein Immobilization. DNA-directed immobilization offers many advantages over standard antibody immobilization techniques. First, a ssDNA chip is much more stable than a standard protein chip, where the antibodies may become denatured and lose activity over time. By storing the protein conjugates in solution and not immobilizing them until the time of use, protein activity can be preserved. Second, the ssDNA probe surface acts as a universal surface, with an infinite number of potential applications. The desired end use of the chip dictates its functionalization, not the other way around. Whereas a sensor surface that has been prespotted with prescribed antibodies can only be used to detect the associated antigens, a ssDNA chip can be used to create a custom protein array surface by immobilizing any desired protein conjugates. Last, DNA hybridization is reversible, allowing for the chips to be regenerated. A protein surface can be converted back into DNA surface by dehybridizing the DNA, thus removing the associated proteins from the surface. CONCLUSIONS In this work, we used DNA-directed immobilization to generate a multichannel biosensor surface and demonstrated the simultaneous detection of multiple analytes. The patterned ssDNA/OEG probe surfaces exhibit high specificity for the respective ssDNAprotein conjugates, ensuring that conjugate immobilization is controlled strictly by sequence-specific DNA hybridization. As a result, it was possible to simultaneously immobilize the conjugates from a single solution, containing a mixture of all of the conjugates. In one simple step, the patterned ssDNA chip was converted into a multichannel biosensor surface. Once the sensor surface was functionalized, the superior specificity of the sensor allowed for simultaneous detection of a set of three analytes from a mixed solution. Although we employ a patterned mixed ssDNA/OEG SAMs on gold as our probe surface, the DNA-directed protein immobilization method described in this work is universal and potentially could be utilized in conjunction with any commercially available DNA array chip. We use gold substrates because they are compatible with our surface plasmon resonance sensor, but other substrates such as silicon or glass would also be appropriate. ACKNOWLEDGMENT This work has been funded by the NSF (CTS-0528605) and FDA (FD-U-002250). C.B. was supported by an Intel Ph.D. Fellowship and a Ford Motor Company Fellowship. Received for review December 8, 2005.
October
27,
2005.
Accepted
AC051923L Analytical Chemistry, Vol. 78, No. 5, March 1, 2006
1519
