7514
Langmuir 2003, 19, 7514-7520
Discerning Biomolecular Interactions Using Kelvin Probe Technology Douglas C. Hansen,*,† Karolyn M. Hansen,‡ Thomas L. Ferrell,‡ and Thomas Thundat‡ Princeton Applied Research, Oak Ridge, Tennessee 37831, and Life Sciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6123 Received February 25, 2003. In Final Form: May 29, 2003 Biomolecular conformational differences of oligomeric nucleic acids were discerned using scanning Kelvin probe (SKP) analysis in an array format on gold-coated silicon substrates. Variations in work function, measured as contact potential difference, of single-stranded DNA (ssDNA), double-stranded DNA (dsDNA), and dsDNA with internal mismatches were clearly distinguished under ambient conditions without the use of an external label. We propose that the effectiveness of SKP analysis is based upon subtle changes in biomolecular conformation resulting from variations in molecule polarizability and induced dipole properties. This sensing technique can be readily incorporated into microarray format for discrimination of specific biological and chemical molecular interactions.
1. Introduction The burgeoning disciplines of genomics and proteomics require the sensitive and specific discrimination of biomolecular interactions in high throughput microarray format. Ideally, the detection technique selected would be a direct measure of specific interactions without the use of exogenous labels, and it would be readily incorporated into a scanning array format. Detection technologies currently used for nucleic acid and protein molecule interaction analysis typically employ incorporated fluorescent labels. Marshall and Hodgson1 report in a prescient article that while fluorescent tags are most popular for readout, the method remains relatively insensitive and requires greater densities of hybridized target for reliable detection. They conclude that if downstream readout technology remains limiting, the value of chips will be restricted. Problems inherent in the use of fluorescent labels and fluorescence imaging have, not surprisingly, encouraged the development of alternate molecular interaction detection technologies, for example, mass spectrometry (specifically MALDI-TOF),2 optical fibers,3 nanoparticle colorimetry,4 optical deflection assays,5-7 surface plasmon resonance,8 quartz crystal microbalance (QCM),9 and electrochemistry/charge transduction utilizing intercalators.10-14 Using scanning Kelvin probe technology to measure inherent electrical properties associated with
biomolecular interactions, we have been able to discriminate fully complementary nucleic acid hybridization and the presence of single nucleotide polymorphisms (SNPs). We propose that the selective discrimination of DNA hybridization by Kelvin probe analysis is due to a change in biomolecule conformation and polarizability of the DNA oligomer layer, and that these parameters are characteristic of discreet molecular interactions. The Kelvin probe15 has been used in ultrahigh vacuum environments and in ambient atmospheric conditions to measure the difference in work function between a conducting probe and conducting or semiconducting metals, nonmetals, polymers, thin films, organic films, and biological samples.16 The work function, φ, is defined as the potential that an electron at the Fermi level must overcome to reach the level of zero kinetic energy away from a solid surface to infinity in a vacuum.17 When two metals (such as the Kelvin probe and a metal surface) having different work functions are electrically connected, electrons will distribute themselves such that an equilibrium of charge will be established. This redistribution of electrons establishes a contact potential, Ψ, and any contact potential difference (CPD) between the work functions of two metals in contact at thermal equilibrium is defined as18
∆Ψ ) φprobe - φmetal surface
(1)
†
Princeton Applied Research. ‡ Oak Ridge National Laboratory. (1) Marshall, A.; Hodgson, J. Nature Biotechnol. 1998, 16, 27-31. (2) Isola, N. R.; Altman, S. L.; Golovlev, V. V.; Chen, C. H. Anal. Chem. 2001, 73, 2126-2131. (3) Healey, B. G.; Matson, R. S.; Walt, D. R. Anal. Biochem. 1997, 251, 270-279. (4) Storhoff, J. J.; Elghanian, R.; Mucic, R. C.; Mirkin, C. A.; Letsinger, R. L. J. Am. Chem. Soc. 1998, 120, 1959-1964. (5) Fritz, J.; Baller, M. K.; Lang, H. P.; Rothuizen, H.; Vettiger, P.; Meyer, E.; Guntherodt, H.-J.; Gerber, Ch.; Gimzewski, J. K. Science 2000, 288, 316-318. (6) Hansen, K. M.; Ji, H.-F.; Wu, G.; Datar, R.; Cote, R.; Majumdar, A.; Thundat, T. G. Anal. Chem. 2001, 73, 1567-1571. (7) Wu, G.; Ji, H.-F.; Hansen, K. M.; Thundat, T. G.; Datar, R. H.; Cote, R. J.; Hagan, M. F.; Chakraborty, A. K.; Majumdar, A. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 1560-1564. (8) Thiel, A. J.; Frutos, A. G.; Jordan, C. E.; Corn, R. M.; Smith, L. M. Anal. Chem. 1997, 69, 4948-4956. (9) Okahata, Y.; Kawase, M.; Niikura, K.; Ohtake, F.; Furusawa, H.; Ebara, Y. Anal. Chem. 1998, 70, 1288-1296.
However, in the present case of ambient atmospheric conditions, a layer of oriented dipoles (such as water (10) Carter, M. T.; Rodriguez, M.; Bard, A. J. J. Am. Chem. Soc. 1989, 111, 8901-8911. (11) Kelley, S. O.; Boon, E. M.; Barton, J. K.; Jackson, N. M.; Hill, M. G. Nucleic Acids Res. 1999, 27, 4830-4837. (12) Kelley, S. O.; Barton, J. K. Science 1999, 283, 375-381. (13) Williams, T. T.; Odom, D. T.; Barton, J. K. J. Am. Chem. Soc. 2000, 122, 9048-9049. (14) Boon, E. M.; Salas, J. E.; Barton, J. K. Nature Biotechnol. 2002, 20, 282-286. (15) Lord Kelvin. Philos. Mag. 1898, 46, 82-120. (16) Baikie, I. D.; Smith, P. J. S.; Porterfield, D. M.; Estrup, P. J. Rev. Sci. Instrum. 1999, 70, 1842-1850. (17) Bare, S. R.; Somarjai, G. A. In Encyclopedia of Physical Science and Technology; Myers, R. A., Ed.; Academic Press: New York, 2002; Vol. 18, pp 401-403. (18) Janata, J.; Josowicz, M. Anal. Chem. 1997, 69, 293A-296A.
10.1021/la034333w CCC: $25.00 © 2003 American Chemical Society Published on Web 07/17/2003
Discerning Biomolecular Interactions
Langmuir, Vol. 19, No. 18, 2003 7515
molecules or immobilized DNA, for example) produces an electric double layer, giving rise to the surface potential, χ.18 The work function of a solid material at a solid/liquid interface can be divided into two components, the contact potential (also defined as the Volta potential19) and the surface potential, which when added together are defined as the Galvani potential,20 Φ:
Φsurface ) Ψ + χ
(2)
Note that this relationship describes the electrostatic potential between two condensed phases in contact and that the Kelvin probe measures the work function difference in this case between the probe and a surface based upon the vibrating condenser method.21-23 The relationship between work function and the Galvani potential for a liquid/solid interface is well defined24 and will not be discussed here. However, providing that the work function of the Kelvin probe is constant, variations in the contact potential difference (CPD) can be attributed to changes in the surface potential on the sample substrate directly below the scanning Kelvin probe tip.16 The measurement of the CPD can provide information about molecular conformation in adsorbed organic layers at the air/solution interface and on solid substrates.25,26 Lu et al.27 have attributed the CPD of self-assembled monolayers (SAMs) of alkanethiols on gold substrates to a net difference in dipole-charge distribution (the surface potential) of the SAMs from the surface potential of the gold. The relative differences in CPD values demonstrated a correlation with alkyl chain length and/or different terminal groups. Similarly, Wang and Bard,28 using the atomic force microscope, demonstrated that the surface potential of immobilized single-stranded DNA was half the value of corresponding hybridized DNA strands. Finally, in a preliminary study, Cheran et al.29 demonstrated the ability of a scanning Kelvin microprobe to measure the CPD values of both single-stranded and double-stranded DNA oligonucleotides immobilized to a silicon substrate. These studies strongly suggest that CPD measurements can be a useful tool for the quantitative detection of a variety of biomolecular interactions. Clearly then, the use of a scanning Kelvin probe to measure the surface potential of adsorbed organic molecules under ambient atmospheric conditions in a noncontact and nondisruptive manner is compelling. 2. Experimental Section Preparation of Silicon Wafers. Silicon wafers (110 silicon, ESPI, Ashland, OR) were prepared by coating with chromium and then gold (2.5 and 25 nm, respectively, by vapor deposition) (19) IUPAC Compendium of Chemical Terminology, 2nd ed.; McNaught, A. D., Wilkinson, A., Eds.; Blackwell Science Inc.: Malden, MA, 1997. (20) Bergveld, P.; Hendrikse, J.; Olthuis, W. Meas. Sci. Technol. 1998, 9, 1801-1808. (21) Kohlrausch, F. Praktische Physik, 22nd ed.; 1968; Vol. 2, Band 2, p 320. (22) Stratmann, M.; Feser, R.; Leng, A. Electrochim. Acta 1994, 39, 1207-1214. (23) Liess, H. D.; Maeckel, R.; Ren, J. Surf. Interface Anal. 1997, 25, 855-859. (24) Grunmeier, G.; Juttner, K.; Stratmann, M. In Materials Science and Technology; Cahn, R. W., Haasen, P., Kramer, E. J., Eds.; Vol. 19, Corrosion and Environmental Degradation, Vol. I; Schutze, M., Ed.; Wiley-VCH: Weinheim, 2000; pp 340-352. (25) Peterson, I. R. Rev. Sci. Instrum. 1999, 70, 3418-3424. (26) Bastide, S.; Gal, D.; Cahen, D.; Kronik, L. Rev. Sci. Instrum. 1999, 70, 4032-4036. (27) Lu¨, J.; Delamarche, E.; Eng, L.; Bennewitz, R.; Meyer, E.; Gu¨ntherodt, H.-J. Langmuir 1999, 15, 8184-8188. (28) Wang, J.; Bard, A. J. Anal. Chem. 2001, 73, 2207-2212. (29) Cheran, L.-E.; McGovern, M. E.; Thompson, M. Faraday Discuss. 2000, 116, 23-34.
to provide a gold surface for immobilization of 5′-end-thiolated DNA oligomers. Coated wafers were cleaned immediately prior to use by sequential washing in acetone, ethanol, piranha solution (3 parts 30% H2O2:7 parts concentrated H2SO4; CAUTIONs piranha solution is a caustic solution), water, and ethanol, and they were dried at 80 °C. DNA Probe Functionalization. Cleaned gold-coated wafers were exposed to phosphate buffer or 5′-end-thiolated DNA oligomers (Oligos Etc., Wilsonville, OR) of known length and nucleic acid sequence (25 µg/mL (42 µM) in sodium phosphate buffer (100 mM PO4-, 150 mM Na+), pH 7.0) in discreet spots (approximately 0.5 mm diameter) by placing 500 nL drops by hand onto the wafer and allowed to incubate for 1 h in a humid atmosphere at room temperature (hereafter known as probe DNA). Probe DNA spots were washed with an excess volume of phosphate buffer for 10 min. Target DNA aliquots (1000 nL of 20 µg/mL (33µM) DNA solution) in phosphate buffer containing complementary DNA of known length and nucleic acid sequence were placed over the probe DNA spots. Other probe DNA and buffer spots were exposed to 1000 nL drops of phosphate buffer solution containing either partially complementary target DNA (2 nucleic acid mismatches) or total noncomplementary target DNA (which has no matching nucleic acids to the probe DNA on the wafer). The spots were allowed to sit on the wafer for 1 h in a humid atmosphere at room temperature. Aliquots were removed by aspiration, and the wafers were washed with phosphate buffer followed by a deionized water wash. Excess water was removed, and the wafers were stored in a humid environment until they were analyzed. Probe and target DNA are presented in Table 1. Estimated densities of DNA oligonucleotides per spot on the goldcoated silicon macroarrays are in the picomole per square centimeter range.13 Control arrays featured the wafer spotted with phosphate buffer only, wafer spotted with phosphate buffer followed by noncomplementary DNA, or a nontreated wafer. Kelvin Probe Measurements. A SKP100E scanning Kelvin probe (SKP) system (Princeton Applied Research, USA) was used for the CPD analysis of the DNA arrays. Electrical connection to the wafer arrays was made using carbon tape under the chip allowing for connection of a wire lead to the SKP control unit. Wafers were attached to a glass plate using double-sided adhesive tape and oriented below the scanning probe. The glass plate/ sample was mounted on an adjustable triangular base and was leveled by adjustment of three leveling screws (Figure 1). The Kelvin probe scanning head was leveled, and the probe, a commercially available (Princeton Applied Research, USA), 500 µm diameter platinum wire in glass, was then lowered to approximately 100 µm above the wafer surface by eye. Final determination of the probe height was achieved by positioning the probe directly above an area of the gold-coated wafer that did not contain any immobilized DNA spots and using a capacitance measurement by applying a -10 V bias to the chip. Since the vibrating probe acts as a varying parallel plate capacitor with the chip, the measured voltage at the probe tip is proportional to the capacitance, which is inversely proportional to the probe/ wafer distance. The vertical height position of the probe (“z” axis) is then varied over a 50 µm increment, and the probe voltage is measured. The software controlling the SKP system calculates a curve fit to determine the absolute value of the distance between the probe tip and the chip surface. Once the probe height was determined using this technique, the probe was then positioned at 80 µm above the chip for all work function measurements. Prior to making any measurements, a stream of humidified N2 (44% relative humidity) gas was allowed to pass over the sample/ SKP interface to facilitate dissipation of any charge buildup due to probe height calibration. CPD measurements were made with the Kelvin probe by making stepwise measurements along the x-axis of a rectangular area containing the DNA spot and the adjacent gold surface on the chip. The number of lines scanned along the x-axis constitutes the y-axis, as each line scan is “stacked” to form a 2-dimensional area map. The aspect ratio of the number of steps along the x-axis to the number of lines scanned (the y-axis) is maintained at 4:3, respectively, resulting in the rectangular scan area. The number of steps along each x-axis line scan was 64, yielding 48 “stacked” line scans in total. At each step along the x-axis scan there was a 10 ms delay prior to acquisition of the data to allow for settling of the baseline
7516
Langmuir, Vol. 19, No. 18, 2003
Hansen et al.
Table 1. Sequences of Probe and Target Oligonucleotides (Noncomplementary Bases Are Indicated in Italics) probe DNA (20-mer) target DNA (20-mer comp) PNC target DNA (20-mer degenerate comp) 1MM target DNA (20-mer comp w/ 1 mismatch) 2MM target DNA (20-mer comp w/ 2 mismatches) NC target DNA (poly-T noncomp) probe DNA (30-mer) target DNA (30-mer comp) probe DNA (40-mer) target DNA (40-mer comp)
5′-thiol-TT AAG GTC TGG ACT GGC CTG-3′ 3′-AA TTC CAG ACC TGA CCG GAC-5′ 3′-TC ATG ACA GAT CTA CTC GTA-5′ 3′-AA TTC CAG ACG TGA CCG GAC-5′ 3′-AA TTC CAG AGG TGA CCG GAC-5′ 3′-TT TTT TTT TTT TTT TTT TTT-5′ 5′-thiol-TTA AGC TCT GGA CTG GCC TGA GAT CTC TAT-3′ 3′-AAT TCG AGA CCT GAC CGG ACT CTA GAG ATA-5′ 5′-thiol-TTA AGG TCT GGA CTG GCA AGC TTC TGT CAT TCA ATA GAT-3′ 3′-AAT TCC AGA CCT GAC CGT TCG AAG ACA GTA AGT TAT CTA-5′
present in the two-dimensional scans in Figures 2 and 4 may be a result of an insufficient amount of recovery time of the Kelvin probe measurement between points along the x-axis. This is the amount of time that is required for the Kelvin probe to return to its original work function as a result of the backing potential being applied. A 10 ms delay was used at each point along the x-axis prior to making a measurement. A longer delay was employed with better results (i.e. minimal streaking); however, this resulted in extended overall scan times and was not used, in the interest of expediting the measurements. The system may be characterized by the potential distribution diagram in Figure 5. The potential differences can be described as follows: Figure 1. Schematic of the SKP100E scanning Kelvin probe system. Table 2. CPD Values of Control Arrays Spotted with Phosphate Buffer Only, Phosphate Buffer and Partially Noncomplementary DNA, and No Treatment treatment
no. of measurements
mean CPD (meV)
std dev (meV)
buffer only buffer/partial noncomp. no treatment
7 6 14
-750 -722 -871
20 34 19
signal from the previous step. Five hundred measurements were made at each step, at a frequency of 30 000 Hz. The resulting area scan maps of the immobilized DNA spots and the surrounding gold wafer substrate displaying the CPD values as a work function (millielectronvolts, or meV) were used for analysis.
3. Results and Discussion Gold-coated silicon wafer substrates were exposed to 5′-end thiolated DNA oligomer probes of varying length or the appropriate buffer control in a macroarray format. Probe DNA arrays were exposed to fully complementary, mismatched, and noncomplementary target DNA oligomers (Table 1). For the control measurements, there were no discernible spots detected on the wafers by SKP, indicating the absence of nonspecific adsorption of buffer or nonthiolated target oligonucleotides on the gold surface (Table 2). Representative Kelvin probe scans of ssDNA and dsDNA spots are presented in Figure 2. CPD values for each spot of immobilized DNA were determined by scanning the probe over a spot at a specific rate, as well as the gold surface immediately adjacent to the spot on the chip as shown in Figure 2; replicate measurements (n ) 3) were taken at each spot location. CPD values of each DNA spot and the surrounding gold substrate were tabulated, and normalized CPD values of the DNA spots (DNACPD - AuCPD) were calculated for each spot on a wafer array, along with their respective means and standard deviation values. A total of eight arrays were analyzed, and the resulting averages of the normalized CPD values for the various spots were combined and plotted as a function of treatment in Figure 3. A representative scan of an array is presented in Figure 4. The “streaking”
DNA gold DNA ∆Φprobe ) ∆Φgold + ∆Ψprobe - χprobe DNA + χ
(3)
gold where ∆Φprobe is the Galvani potential difference between gold is the Galvani the gold and the Kelvin probe, ∆ΦDNA potential difference between the gold and the immobilized DNA spot, χDNA is the surface potential of the immobilized DΝA is the CPD between the immobilized DNA spot, ∆Ψprobe DNA spot and the Kelvin probe, and χprobe is the surface potential of the Kelvin probe. On the basis of the relationship described in eqs 1 and 2 and the potential distribution diagram in Figure 5, it is clear that the “normalization” of the data as described above can be represented by
gold - ∆Φgold ∆Φprobe DNA ) DNA air + χwater + ∆Ψwater + χair + ∆Ψprobe ) (χDNA + ∆Ψwater air
(4) thus demonstrating that the SKP is measuring the potential of the interface between the probe tip and the gold surface. Hybridization of fully complementary 20-mer target DNA to 20-mer immobilized probe DNA resulted in a normalized CPD value twice that of nonhybridized single stranded DNA (Figures 2-4). Discrimination between full complementarity of DNA strands (complete hybridization) and target strands with 1 or 2 internal mismatches (partial hybridization) is evident in the difference between the normalized CPD values of those respective spots. These values were consistently between those of the ssDNA and dsDNA. In addition, a target of degenerate complementarity (i.e. having a short stretch of base complementarity, designated 20-PNC, Figure 3) was similar in CPD value to targets with 1 or 2 internal mismatches and readily distinguished from nonhybridized and fully complementary 20-mer targets (Figures 3 and 4, spot B). Noncomplementary 20-mer poly-T target (designated 20-NC, Figure 3) CPD values were similar to those of 20-mer ssDNA, as expected, since no hybridization should occur. The length of immobilized ssDNA oligomers was readily distinguished: ss30-mer oligonucleotides resulted in an
Discerning Biomolecular Interactions
Langmuir, Vol. 19, No. 18, 2003 7517
Figure 2. Representative Kelvin probe scans of ssDNA and dsDNA spots. Right-hand images are area scan maps; left-hand images are plots of the line scan data as indicated by the horizontal line in the right-hand image.
average CPD value approximately 1.5 times that of the ss20-mer; ss40-mer CPD values were twice that of the 20-mer probes; hybridization of fully complementary 30-mer and 40-mer probe DNA resulted in normalized CPD values twice their respective ssDNA CPD values (Figure 3). The magnitude of the SKP signal strength is dependent upon the amount of DNA immobilized on the assay surface. In this study, we have estimated (on the basis of published reports) that the DNA density on the gold-coated wafer surface is in the picomole per square centimeter range, since thiolated oligomers were immobilized without the presence of Mg2+ in the buffer solution.13 Addition of Mg2+ to the reaction buffer would provide greater counterelectron shielding of the negatively charged DNA and allow for a tighter packing density at the surface. SKP geometry may also affect the magnitude of the CPD measurement. If the DNA spot size is less than or equal to the diameter of the SKP tip or the radius of the DNA spot is less than or equal to the probe height, then the accuracy of the CPD values measured could be limited by the spatial resolution of the system.30 Since the spot sizes were approximately the same as the diameter of the probe tip (500 µm), then it is possible that the CPDAu values could be making a significant contribution to the CPDDNA values measured. (30) McMurray, H. N.; Williams, G. J. Appl. Phys. 2002, 91, 16731679.
The converse may also be the reason there are differences in the CPD of the gold on either side of the immobilized DNA spots due to their proximity to each other. It is recognized that the present study was not conducted under optimal conditions and further investigation into the relationship between the DNA spot size, SKP height, and the measured CPDDNA values is warranted. The presence of immobilized biomolecules between the Kelvin probe and the gold surface has two significant effects. First, the biomolecule can lower the effective work function.31 Additionally, the capacitance is raised due to the presence of a dielectric (the DNA film), this effect depending strongly upon the polarizability of the sample. Capacitance, in the case of a parallel-plate capacitor, is proportional to the dielectric constant of the medium between the plates, κ, the permittivity of free space, 0, and the area, A, of the plates, and it is the inverse of the distance between the plates. In the case of a gap of distance b partially filled near one electrode with a dielectric of thickness z and the dielectric constant, the capacitance is32
C ) κ0A/[κb - z(κ - 1)]
(5)
(31) Stratmann, M.; Streckel, H. Corros. Sci. 1990, 30, 681-696. (32) Halliday, D.; Resnick, R.; Walker, J. Physics, 4th ed.; John Wiley and Sons: New York, 1993.
7518
Langmuir, Vol. 19, No. 18, 2003
Hansen et al.
Figure 3. Average CPD values for DNA spots on various chip arrays analyzed by a scanning Kelvin probe. Error bars are standard errors (ss ) single stranded; ds ) double stranded; 1MM ) 1 internal mismatch; 2MM ) 2 internal mismatches; PNC ) degenerate complementarity; NC ) 20-mer poly-T; 20, 30, or 40 ) length of DNA oligomer in bases (ss) or base pairs (ds)).
For field strengths of the size relevant to this discussion, the dielectric constant is linearly related to the molecular polarizability and density of the dielectric material. The field strengths estimated for these measurements are quite low, approximately 1 × 104 V/m (calculated as the approximate gold value divided by the gap distance: 870 mV/80 µm). An electric field strength of this magnitude is below that which is essential for ionic conductivity in the immobilized DNA oligomers. At high field strengths, DNA can act as a conductor.33 The polarizability, R, of a molecule is proportional to the volume of the molecule, and the bulk polarizability is proportional to the number density, n, of contributing charges as well as to the molecular polarizability. To detect the presence of a dielectric layer that is thin relative to the gap size, the layer must have a high dielectric constant. In our experiment, the Kelvin probe is at a distance of 80 µm and the thickness of the DNA layer is approximately 6 nm. Although there are reports of a dielectric constant as high as 103 for bulk DNA,34 more recent results indicate much smaller values; the dielectric constant of oligomeric DNA is composed of dielectric values associated with discrete areas of the hydrated DNA molecule (ranging from 20 for the minor groove to 55 for the major groove to 80 for the bulk solvent/ oligomer interface).35,36 The CPD measurements were made under a humid stream of air (44% RH). Assuming that DNA binds water molecules and the intercalation is in a ratio of 0.6 to 1 g of water to 1 g of DNA,37 the dielectric properties of a fully hydrated oligomeric DNA film cannot (33) Porath, D.; Bezryadin, A.; de Vries, S.; Dekker, C. Nature 2000, 403, 635-638. (34) Cole, R. H. Ann. N. Y. Acad. Sci. 1977, 303, 59-73. (35) Young, M. A.; Jayaram, B.; Beveridge, D. L. J. Phys. Chem. B 1998, 102, 7666-7669. (36) Tavernier, H. L.; Fayer, M. D. J. Phys. Chem. B 2000, 104, 1154111550. (37) Mashimo, S. In Dielectric Spectroscopy of Polymeric Materials; Runt, J. P., Fitzgerald, J. J., Eds.; American Chemical Society: Washington, DC, 1997; pp 201-225.
be responsible for the observed changes in CPD. Therefore, the CPD values are solely due to the presence of a hydrated oligomeric DNA film on the gold surface (see eq 4). The random coiled conformation of single-stranded oligomers along with the resulting increased density of polarized moieties (vs the bare gold surface) produces a variation in the surface potential of the DNA/probe interface (χDNA). Presumably, in the case of single mismatches, while hybridization does occur, the conformation of the mismatched DNA hybrid, the oligomer polarizability, and the volume occupied by the same number of molecules change. Therefore, the CPD is altered, since it depends on the density of polarized moieties. The hypothesis presented here assumes a difference in surface potential possibly due to the conformation of ssDNA, dsDNA, and mismatched dsDNA, with concomitant changes in inducible dipole properties and polarizability. Surface analysis of thiol-immobilized ssDNA versus dsDNA provides supporting evidence for such conformational changes: scanning tunneling microscopy (STM) imaging revealed a 0.22 nm height per base for ssDNA and 0.34 nm per base pair for dsDNA (e30-mer oligomers);38 The DNA film thickness measured by ellipsometry revealed a 0.132 nm height per base for 25mer ssDNA.39 In addition, variations in film dielectric properties, measured as capacitance40 and time domain reflectometry (TDR),41were shown for hybridized versus single-stranded oligomers, indicating an increase in film thickness upon hybridization. Variations in surface charge and surface potentials of alkanethiols, ssDNA, dsDNA, and dsDNA with mismatches have been documented using (38) Rekesh, D.; Lyubchenko, Y.; Shlyakhtenko, L. S.; Lindsay, S. M. Biophys. J. 1996, 71, 1079-1086. (39) Steel, A. B.; Levicky, R. L.; Herne, T. M.; Tarlov, M. J. Biophys. J. 2000, 79, 975-981. (40) Berney, H.; West, J.; Haefele, E.; Alderman, J.; Lane, W.; Collins, J. K. Sens. Actuator, B 2000, 68, 100-108. (41) Lee, R. S.; Bone, S. Biochim. Biophys. Acta 1998, 1397, 316324.
Discerning Biomolecular Interactions
Langmuir, Vol. 19, No. 18, 2003 7519
Figure 4. Representative Kelvin probe scan of a DNA array containing spots of (A) ssDNA, (B) ssDNA (partial complementarity), and (C) dsDNA (full complementarity), with line scan data for spots A, B, and C.
force measurements (AFM)28 and cantilever optical deflection assays,5-7,42 which are also indicative of immobilized biomolecule conformational change. Remarkably, the relative surface potential differences determined in solution using AFM force curve measurements by Wang and Bard28 were essentially the same as the relative SKP contact potential differences measured in air and reported here: dsDNA surface charge values were twice that of ssDNA (-120 versus -60 mV), and mismatched dsDNA surface charge values were between those of dsDNA and ssDNA (-80 mV). In that study, it was postulated that
the density of the phosphate groups present within the DNA oligomer was responsible for the resulting surface charge of an immobilized DNA oligomer film; the surface charges increased upon hybridization, thus allowing discrimination of ss- and dsDNA. While appearing to be straightforward at the outset, further consideration of this model in terms of partial complementarity involving single or multiple nucleic acid mismatches reveals such discrimination proves to be problematic. That is, the number of phosphate groups contained in fully hybridized dsDNA will be quite similar to that for partially hybridized or
7520
Langmuir, Vol. 19, No. 18, 2003
Figure 5. Potential distribution diagram of the SKP-DNAgold surface system (symbols as defined in the text).
single-base mismatched dsDNA. If the number of phosphate groups is similar, then how can the resulting surface charge/surface potential be different? The answer may be that while phosphate charge density plays a significant role, more importantly it is the relative change in conformation of the DNA strands, resulting in changes in inducible dipole properties, and polarizability as described above. The packing density at the gold surface may also influence the conformation of both single- and doublestranded DNA. Given that we used 100 mM phosphate buffer without added Mg2+, our packing densities are assumed to be low; this would allow for greater rotational freedom of thiol-immobilized ssDNA and the potential for interaction of the individual chains with the surrounding Au surface. However, Steel et al.39 demonstrated that ssDNA oligomers of
