Selective Enzymatic Labeling To Detect Packing-Induced

Sep 19, 2008 - Dana Peled , Ron Naaman and Shirley S. Daube. The Journal of ... R. Kaufmann , D. Peled , R. Naaman and S. S. Daube. ACS Applied ...
0 downloads 0 Views 547KB Size
11842

Langmuir 2008, 24, 11842-11846

Selective Enzymatic Labeling To Detect Packing-Induced Denaturation of Double-Stranded DNA at Interfaces Dana Peled,† Shirley S. Daube,‡ and Ron Naaman*,† Department of Chemical Physics and Chemical Research Support, Weizmann Institute, RehoVot 76100, Israel ReceiVed May 9, 2008 The adsorption of DNA on surfaces is a widespread procedure and is a common way for fabrication of biosensors, DNA chips, and nanoelectronic devices. Although the biologically relevant and prevailing in vivo structure of DNA is its double-stranded (dsDNA) conformation, the characterization of DNA on surfaces has mainly focused on singlestranded DNA (ssDNA). Studying the structure of dsDNA on surfaces is of invaluable importance to microarray performance since their effectiveness relies on the ability of two DNA molecules to hybridize and remain stable. In addition, many of the enzymatic transactions performed on DNA require dsDNA, rather than ssDNA, as a substrate. However, it is not established that adsorbed dsDNA remains in its structure and does not denature. Here, two methodologies have been developed for distinguishing between surface-adsorbed single- and double-stranded DNA. We demonstrate that, upon formation of a dense monolayer, the nonthiolated strand comprising the dsDNA is released and the monolayer consists of mostly ssDNA. The fraction of dsDNA within the ssDNA monolayer depends on the length of the oligomers. A likely mechanism leading to this rearrangement is discussed.

Introduction The combination of biological molecules, especially DNA, with nanotechnology is viewed as holding high potential for futuristic “molecular electronics” and sensing applications. Today, the adsorption of DNA on surfaces is a widespread procedure and is a common way for fabrication of biosensors, DNA chips, and nanoelectronic devices.1-8 Several properties make DNA a popular molecule in nanotechnology. Most important is its ability to self-assemble by recognizing a complementary DNA sequence. In addition, there are many molecular biology techniques by which the DNA can be modified chemically and tailored structurally.9-14 Although the biologically relevant and prevailing in vivo structure of DNA is its double-stranded (dsDNA) conformation, the characterization of DNA on surfaces has mainly focused on single-stranded DNA (ssDNA),15 due to the simplicity of obtaining chemically synthesized ssDNA oligos and its utility as a probe in the widespread microarray technology. However, studying the structure of dsDNA at surfaces is of invaluable † ‡

Department of Chemical Physics. Chemical Research Support.

(1) Dekker, C.; Ratner, M. A. Phys. World 2001, 14, 29–33. (2) Hazani, M.; Hennrich, F.; Kappes, M.; Naaman, R.; Peled, D.; Sidorov, V.; Shvarts, D. Chem. Phys. Lett. 2004, 391, 389–392. (3) Jung, A. Anal. Bioanal. Chem. 2002, 372, 41–42. (4) Kasemo, B. Surf. Sci. 2002, 500, 656–677. (5) Keren, K.; Krueger, M.; Gilad, R.; Ben-Yoseph, G.; Sivan, U.; Braun, E. Science 2002, 297, 72–75. (6) Pirrung, M. C. Angew. Chem., Int. Ed. 2002, 41, 1276–1289. (7) Wang, J. Nucleic Acids Res. 2000, 28, 3011–3016. (8) Willner, I.; Willner, B.; Katz, E. ReV. Mol. Biotechnol. 2002, 82, 325–355. (9) Gerry, N. P.; Witowski, N. E.; Day, J.; Hammer, R. P.; Barany, G.; Barany, F. J. Mol. Biol. 1999, 292, 251–262. (10) Kim, J. H.; Hong, J. A.; Yoon, M.; Yoon, M. Y.; Jeong, H. S.; Hwang, H. J. J. Biotechnol. 2002, 96, 213–221. (11) Pena, S. R. N.; Raina, S.; Goodrich, G. P.; Fedoroff, N. V.; Keating, C. D. J. Am. Chem. Soc. 2002, 124, 7314–7323. (12) Frutos, A. G.; Smith, L. M.; Corn, R. M. J. Am. Chem. Soc. 1998, 120, 10277–10282. (13) Bamdad, C Biophys. J. 1998, 75, 1997–2003. (14) O’Brien, J. C.; Stickney, J. T.; Porter, M. D. J. Am. Chem. Soc. 2000, 122, 5004–5005. (15) Sassolas, A.; Leca-Bouvier, B. D.; Blum, L. J. Chem. ReV. 2008, 108, 109–139.

importance to microarray performance, since their effectiveness relies on the ability of two DNA molecules to hybridize and remain stable. In addition, many of the enzymatic reactions performed on DNA, which are relevant for sensing applications in disease detection, require dsDNA, rather than ssDNA, as a substrate.16 Moreover, to get highly sensitive detection of enzymatic activity, a large number of dsDNA molecules should be adsorbed on the surface. There is a lack of information about the dsDNA conformation on surfaces. Most of the relevant data are based on spectroscopy and electrochemistry,4 methods that are complicated and indirect and require elaborated data interpretation. There are few cases reporting the instability of adsorbed dsDNA, such as the local denaturation of the ends of long dsDNA adsorbed on surfaces17-19 and XPS analysis showing that ssDNA, in the form of a hairpin structure, harboring dsDNA regions in its structure, was denaturated upon adsorption.20 A more direct study aimed at stabilizing short dsDNA on surfaces by using an unnatural yet more stable version of DNA21,22 was unsuccessful in preventing DNA denaturation. To investigate the extent of the adsorption-induced denaturation and to establish its underlying mechanism, one has to quantitatively determine the amount of dsDNA and ssDNA molecules on a surface. Here we present two inventive straightforward radiolabeling techniques to selectively distinguish between ssDNA and dsDNA within a mixed monolayer of the two species. These simple techniques can be used with any DNA with no sequence or length limitations, as long as it is linear. The radioactive signal, unlike fluorescence, is independent of the (16) Han, A. S.; Takarada, T.; Shibata, T.; Nakayama, M.; Maeda, M. Anal. Sci. 2006, 22, 663–666. (17) Michalet, X. Nano Lett. 2001, 1, 341–343. (18) Allemand, J. F.; Bensimon, D.; Jullien, L.; Bensimon, A.; Croquette, V. Biophys. J. 1997, 73, 2064–2070. (19) Wooley, A. T.; Kelly, R. T. Nano Lett. 2001, 1, 345–348. (20) Opdal, A.; Petrovykh, D. Y.; Kimura-Suda, H.; Tarlov, M. J.; Whitman, L. J. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 9–14. (21) Wackerbarth, H.; Grubb, M.; Wengel, J.; Chorkendorff, I.; Ulstrup, J. Surf. Sci. 2006, 600L122-L127. (22) Wackerbarth, H.; Marie, R.; Grubb, M.; Zhang, J. D.; Hansen, A. G.; Chorkendorff, I.; Christensen, C. B. V.; Boisen, A.; Ulstrup, J. J. Solid State Electrochem. 2004, 8, 474–481.

10.1021/la801437n CCC: $40.75  2008 American Chemical Society Published on Web 09/19/2008

Packing-Induced Denaturation of DNA at Interfaces

Langmuir, Vol. 24, No. 20, 2008 11843

environment the DNA is in. Its signal is the same in both aqueous and dry environments. Moreover, the radioactive signal is insensitive to the type of linkage between the DNA and the surface: the DNA can be linked by any method (reviewed in ref 6), since it is placed on the DNA either prior to adsorption or postadsorption at the distal end of the DNA further from the surface. Because a gold surface is often being used in biosensing applications, we chose gold as a substrate for studying adsorption of dsDNA monolayers and used the radioactive signal to evaluate how many dsDNA molecules, adsorbed on a gold surface, remain in the dsDNA conformation and how many undergo denaturation and remain as ssDNA on the surface. On the basis of these findings, a microscopic model that describes the process is suggested.

Methods Substrate Preparation. Gold. Polished (100) silicon wafers were cleaned in a plasma asher (March Plasmod) for 5 min in an atmosphere of oxygen and argon. Immediately after cleaning, the wafers were placed in an electron beam evaporator. The deposition was performed at a base pressure of 5 × 10-6 mbar. A 10 nm chromium layer was deposited at a rate of 0.1 nm/s, and then a 150 nm gold layer was deposited at the same deposition rate. The wafer was cut into 1 × 4 cm samples using a diamond pen. These gold slides were then cleaned in an ultraviolet ozone cleaner (Uvocs, model T10 × 10/ OES/E) for 20 min followed by a 20 min incubation in absolute ethanol (Merck) with agitation. The slides were then rinsed with ethanol and stored for no more then 24 h in ethanol solution before adsorption. Silicon. Silicon slides (100; 1 × 2 cm) were boiled in acetone and ethanol for 5 min each, followed by UV/ozone oxidation (UVOCS) for 20 min and then 20 min of incubation in absolute ethanol with agitation. The slides were then rinsed with ethanol (Merck) and dried under a N2 stream. The samples were immediately placed in the adsorption solution. The monolayers were prepared from 1 µL of (3-mercaptopropyl)trimethoxysilane (MPTS; Flucka) dissolved in 1.5 mL of bicyclohexyl (Aldrich). Adsorption was carried out for 2 h in N2-filled vials. After adsorption the samples were sonicated twice in toluene for 30 s each and N2 dried. DNA Radiolabeling and Hybridization. DNA oligomers were treated and labeled at their 5′ end using the enzyme T4 polynucleotide kinase and [32P]-γ-ATP according to published protocols.23 The radiolabeled ssDNA strands were hybridized with their complementary nonlabeled strands at a ratio of 1:1.1 of thiolated to nonthiolated strand, respectively, to ensure that all of the thiolated strands were hybridized. The oligomers were incubated at 80 °C for 10 min, following slow cooling to room temperature. The efficiency of hybridization was evaluated by loading samples of the radiolabeled single strands and double strands on a 15% nondenaturing polyacrylamide gel, at a 1:19 ratio of bisacrylamide to acrylamide in 89 mM Tris borate, 2 mM EDTA (1 × TBE buffer). Under these conditions the migration of the double strands is retarded compared to that of the single strands. DNA Adsorption. On a Gold Substrate. Self-assembled DNA monolayers on gold films were prepared by deposition of 15, 26, or 50 nucleotide (nt) DNA oligomers either as ssDNA, full dsDNA, or in a primer/template (p/t) dsDNA configuration (Figure 1A-C). All DNA oligomers were designed to have random sequences with no pronounced secondary structure or special surface affinity (Table 1). A solution of 10 µM ssDNA, dsDNA, or p/t dsDNA (in 20 mM Tris-HCl, pH 7.5, 0.4 M NaCl) was spotted on the clean gold slide for a given time and at controlled humidity to avoid dryness. After adsorption at room temperature, the slide was rinsed in 20 mM Tris-HCl, pH 7.5, 0.4 M NaCl to wash away excess DNA. The slide was then soaked in the same buffer with shaking for 15 min. This wash was then followed by shaking for 15 min in 20 mM Tris-HCl, pH 7.5, 0.2 M NaCl, and soaking for 15 min in 20 mM (23) Aqua, T.; Naaman, R.; Daube, S. S. Langmuir 2003, 19, 10573–10580.

Figure 1. Schematic illustration of adsorption and detection protocols: (A) dsDNA hybridized in a solution prior to surface adsorption, (B) hybridization of a complementary nonthiolated strand to thiolated ssDNA preadsorbed on the surface, (C) incorporation of a complementary nucleotide to a dsDNA primer/template by DNA polymerase I, (D) schematic representation of packing-induced denaturization.

Tris-HCl, pH 7.5, and subsequently by a through rinse in 20 mM Tris-HCl, pH 7.5, and a quick rinse (10 s) in sterile deionized (Millipore) water to remove any excess salt left on the surface. The slide was then dried in air. Hybridization on the surface was performed by adsorbing a thiolated ssDNA of 28 nt (10 µM) for 1 h at room temperature. Following adsorption, the DNA was washed as described above. A 10 µM nonthiolated, complementary, DNA strand (either nonlabeled or radiolabeled) of 26 nt in hybridization buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl) was added to the monolayer to get dsDNA. The DNA strands were left to hybridize overnight following washing of the monolayer twice in hybridization buffer to wash away any excess of the nonthiolated complementary strand. The slides were then soaked in 20 mM Tris-HCl, pH 7.5, with shaking for 15 min. This was then followed by a quick rinse (10 s) in water. On a Silicon Substrate. The thiolated dsDNA was diluted to a concentration of 10 µM in 50 mM NaHCO3/NaH2CO3 buffer, pH 9, and spotted on the mercaptosilane-coated surfaces at a controlled humidity and temperature. After overnight adsorption the slides were rinsed in 50 mM NaHCO3/NaH2CO3, pH 9, to wash away excess DNA. The slide was then soaked in the same buffer with shaking for 15 min, followed by shaking for 15 min in 20 mM Tris-HCl, pH 7.5, 0.2 M NaCl, and soaking for 15 min in 20 mM Tris-HCl, pH 7.5, and subsequently by a thorough rinse in 20 mM Tris-HCl, pH 7.5, and a quick rinse in sterile deionized (Millipore) water to remove any excess salt left on the surface. The slide was then dried in the air. DNA Polymerase I SelectiVity Assays. The Escherichia coli DNA polymerase I Klenow fragment was used to incorporate selectively radiolabeled nucleotides into a p/t dsDNA structure (see the scheme in Figure 1C). In addition to its polymerization activity, this DNA polymerase has a 3′ f 5′ exonuclease activity that is repressed in the presence of nucleotides. The Klenow fragment of E. coli DNA polymerase I lacks the 5′ f 3′ exonuclease activity that is present in the full native enzyme. Therefore, the DNA oligomers on the surface are safe from the hydrolyzable activities of the enzyme. The enzymatic experiment was performed on a gold slide using an array of wells separated from one another with RTV polymer. The RTV matrix setup was molded by polymerization of RTV615 silicone rubber compounds (GE Bayer silicon) at 70 °C on top of an array of beads to create an empty-well matrix after polymerization. After 4 h of polymerization the RTV polymer was removed, cut to the desired shape, and placed on a clean gold slide.24 Self-assembled layers of either ssDNA or p/t dsDNA were prepared by spotting 10 µL (10 µM) of the DNA in adsorption buffer in each well for 1 h. (24) Kaufmann, R.; Averbukh, I.; Naaman, R.; Daube, S. S. Langmuir 2008, 24, 927–931.

11844 Langmuir, Vol. 24, No. 20, 2008

Peled et al.

Table 1. Sequences and Structures of the DNA Molecules Used

After the adsorption the wells were washed several times as described previously. Following the washes, a 10 µL mixture of 5 U/µL E. coli DNA polymerase I Klenow fragment (New England Biolabs) in the enzyme’s reaction buffer (10 mM Tris-HCl, pH 7.5, 5 mM MgCl2) and 100 µM of the nucleotide dCTP or rUTP, supplemented with ∼1 µM of the radioactive R-32P dCTP or R-32P rUTP, respectively, were added. After 2 h of incubation at 30 (C, the wells were washed several times with 20 mM Tris-HCl, pH 7.5, 200 mM NaCl. Following the washes, the RTV polymer was peeled from the gold surface and the slides were washed twice for 15 min each with 20 mM Tris-HCl, pH 7.5, 200 mM NaCl. The slides were then dried in air. RadioactiVe Measurements. Gold slides adsorbed with radioactive DNA were exposed to a phosphorimaging screen, and the screen was imaged by a phosphorimager scanner (FLA-5100, Fuji). The number of pixels per area was converted to the number of picomoles per area using the specific activity. The specific activity of each DNA sample is defined as the amount of radiation produced by 1 mol of that DNA. This number was determined empirically by spotting 1 µL of radioactive DNA with a known concentration on a gold slide. The samples were allowed to air-dry without any washing, followed by phosphorimaging. This value already takes into account gold enhancement/quenching.

Characterizing surface-induced denaturation presents a technical challenge due to the difficulty in accurately determining whether a DNA molecule placed on a surface is in a singlestranded or double-stranded conformation. We approached this problem by placing a radiolabeled phosphate at the 5′ end of one of two complementary DNA oligomers (Figure 2A). The radioactive phosphate was placed either on the strand harboring

the thiol (thiolated) or on the complementary nonthiolated strand (complementary). Each of the radiolabeled strands was hybridized to its nonlabeled counterpart, resulting in two types of dsDNA species (I and II, Figure 2A) that are chemically identical, except for the position of the radioactive phosphate. The two dsDNA species migrated in a similar fashion in a nondenaturing gel (Figure 2B, lanes 2 and 4), showing that each of the dsDNA solutions did not contain any excess ssDNA. If there was excess ssDNA, another band would have appeared in lanes 2 and 4, lower than the existing bands. This band would be at the same height as the band in lanes 1 and 3, which are the ssDNA signals. In the initial experiments, a 26 nt long DNA oligomer with a sequence lacking any major secondary structures (Table 1) was used. Following the adsorption of each of the two dsDNA types onto a gold surface, the ratio between the radioactive signals from the two types of dsDNA was monitored. If the adsorbed DNA remains in its original double-stranded conformation, a ratio of unity between the radioactive measurements of the two species is expected. Any denaturation and loss of the nonthiolated DNA should result in a decrease in the signal obtained from species II (Figure 2, AII) but not from species I (Figure 2, AI), resulting in a ratio of less than unity. When monitoring the radioactive signal as a function of adsorption time (Figure 3), we observed that the density of the labeled thiolated DNA strands increased with adsorption time until reaching saturation, while that of the labeled complementary strand was lower by almost an order of magnitude and remained constant with time. Hence, the ratio between thiolated and complementary strands is lower than unity at all time points, indicating that denaturation has occurred. In addition, assuming

Figure 2. (A) Two dsDNA species of the same sequences, but with different strands harboring the radiolabel: species I, the strand modified with a thiol (blue circle) at the 3′ end is labeled at the 5′ end with a radioactive phosphate (green star) (“thiolated”); species II, the nonthiolated complementary strand (red) is labeled at the 5′ end with a radioactive phosphate (“complementary”). (B) Polyacrylamide gel electrophoresis (PAGE) analysis of the two dsDNA species differing in radiolabeling: lanes 1 and 2, the thiolated strand is radiolabeled; lanes 3 and 4, the nonthiolated complementary strand is radiolabeled; lanes 1 and 3, ssDNA; lanes 2 and 4, dsDNA. This nondenaturing gel does not disrupt base-pairing between the strands. Therefore, the nonlabeled complementary strand retards the migration of the labeled single strand.

Figure 3. Number of molecules per unit area (molecules/mm2) as measured from radioactive labeled molecules after different adsorption times. The blue triangles represent the labeled thiolated strand of ssDNA (4) and dsDNA (2). The red triangles represent the labeled complementary strand of ssDNA (4) and dsDNA (2).

Results

Packing-Induced Denaturation of DNA at Interfaces

Langmuir, Vol. 24, No. 20, 2008 11845 Table 2. dsDNA Quantification on Surfaces radioactive labeling

density (pmol/mm2)

DNA Hybridized in Solution and Then Adsorbed to the Surface dsDNA (Figure 1A) dsDNA (Figure 1C) ssDNA (control)

thiolated complementary dCTP rUTP dCTP rUTP

0.233 ( 0.009 0.034 ( 0.002 0.050 ( 0.002 0.007 ( 0.003 0.001 ( 0.0002 0.003 ( 0.0005

DNA Hybridized on the Surface dsDNA (Figure 1B) dsDNA (Figure 1C) Figure 4. Ratio of labeled nonthiolated to labeled thiolated dsDNA of different lengths and therefore stabilities. The calculated free energy change (∆G; 25 °C, 0.4 M NaCl) for the 15, 26, and 50 bp oligimers are 25.4, 39.9, and 85.9 kcal/mol, respectively.

that nonspecific binding of DNA to gold is minimal, the amount of labeled complementary strand represents dsDNA molecules since this strand does not contain a thiol and could therefore remain on the surface and withstands the extensive washing only if it is hybridized to the thiolated strand. Therefore, it seems that some dsDNA species adsorb at early time points and remain as dsDNA on the surface, but their concentration does not increase with time. Additional dsDNA molecules approaching the surface undergo denaturation and adsorb as ssDNA (see the scheme in Figure 1D). The signal of the labeled complementary strand did not decrease when the washing time was increased from 5 min (Supporting Information), suggesting that the washing process itself did not induce dsDNA denaturation. The kinetics of thiolated DNA adsorption were very similar for both ssDNA and dsDNA conformations (Figure 3), suggesting that the adsorption process itself, rather than the denaturation step, is rate limiting and that the two strands are being separated quickly, during the adsorption on the gold surface, most likely on a millisecond time scale. By varying the length of the dsDNA adsorbed on the gold surface, so that 15, 26, and 50 bp oligomers were probed, we found that the ratio between the labeled complementary and labeled thiolated strands increased with the length of the dsDNA, in correlation with the increase in the free energy (Figure 4 and Table 1), yielding 95%, 85%, and 35% denaturation of the 15, 26, and 50 bp oligimers, respectively. Namely, the denaturation depends on the length and therefore on the stability of the doublestranded DNA. Since denaturation was so pronounced, it is possible that even the remaining nonthiolated DNA on the surface was bound nonspecifically to the surface and was not engaged in hydrogen bonds with the thiolated complementary strand. To directly determine at what conformation the nonthiolated DNA is found on the surface, we utilized the unique ability of enzymes to be highly selective for their substrates. That is, DNA polymerases extend a DNA oligomer by incorporating a nucleotide in a sequence-specific manner, only if the DNA is in a ds conformation known as a primer/template (Table 1). ssDNA and full dsDNA oligomers would not be extended by the enzyme. Therefore, if the DNA on the surface is in a primer/ template dsDNA conformation, incorporation of a radiolabeled deoxycytidine (dCTP) opposite the guanine at the primer/template junction is expected to occur (Figure 1C). However, if the ds primer/template has undergone denaturation and only the thiolated template strand remains on the surface, no radiolabeled nucleotide will be incorporated.

complementary dCTP rUTP

0.16 ( 0.04 0.14 ( 0.02 0.003 ( 0.001

To demonstrate that this selectively assay is reliable and sensitive in a surface reaction, we first formed, as a positive control, a DNA monolayer in a manner that must contain dsDNA. That is, if a gold surface is first covered with a thiolated ssDNA to form a densely packed monolayer, the surface becomes protected by the ssDNA. Therefore, addition of a complementary nonthiolated strand to this monolayer should result in dsDNA molecules on the surface (Figure 1B). Table 2 demonstrates that, when such a process was performed with a radiolabeled complementary strand, 0.16 ( 0.04 pmol/mm2 was bound to the DNA monolayer on the surface. When a mixture of the large fragment of the enzyme E. coli DNA polymerase I (Klenow fragment; see Methods) was added to such a monolayer, in the presence of radiolabeled dCTP, 0.14 ( 0.02 pmol/mm2 dCTP was incorporated. Since only one dCTP can be incorporated into one primer/template DNA molecule (Figure 1C and sequence of p/t in Table 1), the enzymatic activity was able to detect, within the experimental error, all those molecules having the primer/ template structure. To eliminate the possibility that the radioactive signal is stemming from nonspecific adsorption of nucleotides to the gold surface, we replaced the dCTP with rUTP. rUTP should have an affinity to gold surfaces similar to that of dCTP, but binds to DNA polymerase with very low efficiency. In addition, it does not match the particular sequence in our DNA. Therefore, the low signal obtained in the rUTP reaction (Table 2) represents the background nonspecific adsorption of nucleotides, validating that dCTP incorporation is specific. Next, this methodology was applied to probe monolayers prepared by direct adsorption of dsDNA prehybridized in solution (Figure 1A). The densities obtained for the thiolated strand and the complementary strand were 0.233 ( 0.009 and 0.034 ( 0.002 pmol/mm2, respectively (Table 2), yielding a reproducible ratio of 0.15 between the strands (Figures 3 and 4).When the DNA polymerase I Klenow fragment was added in the presence of radiolabeled dCTP (Figure 1C), the density of the radiolabel was found to be 0.05 ( 0.002 pmol/mm2 (Table 2), which is a value very similar to that of the complementary strand of the dsDNA. In this experiment, as in our positive control experiment, a noncomplementary nucleotide (rUTP) or an ssDNA monolayer resulted in a very low background signal, supporting the specificity of this enzymatic assay. In conclusion, the enzymatic assay verified that all the complementary strands found on the surface in our original experiments (Figures 3 and 4) were in a dsDNA conformation. Our results are also consistent with XPS thickness measurements in which the thickness of an ssDNA monolayer of 26 bp adsorbed on a gold surface is about 31 ( 3 Å. The thickness of the dsDNA monolayer should be 90 Å, but our XPS measurements resulted in a thickness of only 38 ( 6 Å. If we consider that for

11846 Langmuir, Vol. 24, No. 20, 2008

Figure 5. Ratio of labeled nonthiolated to labeled thiolated dsDNA adsorbed overnight on different surfaces.

the 26 bp long DNA only 15% of the monolayer was ds, while 85% was ssDNA, a simple calculation can yield the effective thickness of the DNA monolayer: (15% × 90 Å) + (85% × 31 Å) ) 40 Å. These results provide further support that adsorption of dsDNA on a gold surface results in a mixed monolayer of mostly ssDNA and only some dsDNA. To shed light on the involvement of the surface in the denaturation process, dsDNA was adsorbed on a silicon surface, rather than gold. The dsDNA was attached to the silicon surface via covalent attachment to an intermediate mercaptosilane monolayer through the 3′-thiol modification. Figure 5 demonstrates that, while on gold the dsDNA was denatured, it did not undergo denaturation on silicon. It is important to realize that the maximal DNA density obtained on the silicon surface was at most 1 × 1010 molecules/mm2, which is about 10% of the densities on the gold.

Discussion In the present study we aimed at characterizing self-assembled monolayers of double-stranded DNA adsorbed on gold surfaces in two selective radiolabeling techniques. The methods we developed for characterizing the structure of the adsorbed DNA, independent of the sequence and length, are simple to apply to every surface, and the interpretation of the results is straightforward. The radioactive measurements allow us to accurately quantify the DNA coverage on surfaces, while the enzymatic assay enabled us to accurately identify and confirm the conformation of the DNA on the surfaces. The results indicate that, when dsDNA was adsorbed on the gold surface and not on silicon, the adsorption process induced denaturation of the DNA (Figures 3-5 and Table 2), suggesting a surface effect. This denaturation occurred even when stable 50 bp dsDNA was used, although to a lesser extent in comparison to that of the shorter oligomers. The extent of denaturation was inversely proportional to the stability of the dsDNA (Figure 4). Since it is generally accepted that nonspecific interactions between the bases of DNA and a gold surface are highly dependent on the base,21,22,25 our strategy was to use DNA with random sequences to avoid any sequence dependency. That is, the denaturation observed is not due to special DNA sequences harboring a high affinity to gold, but is rather a general property of DNA of random sequences. In past studies, the local denaturation of short stretches of DNA on surfaces has mostly been attributed to the hydrophobic or electrostatic interactions of the bases with the surface. It was (25) Wolf, L. K.; Gao, Y.; Georgiadis, R. M. Langmuir 2004, 20, 3357–3361.

Peled et al.

assumed that these nonspecific interactions compete with the energy involved in hydrogen bonding and stacking of the bases within the dsDNA.17,20 This description does not depend on the actual formation of a monolayer and therefore should not depend on the density of the adsorbed molecules. A possible alternative model (alluded to in ref 21) suggests that the denaturation of dsDNA on the surface is due to the formation of a dense layer. The area occupied by an ssDNA is smaller than that of a dsDNA, and therefore, the denaturation allows for another DNA molecule to penetrate and a additional thiol bond to be formed in the vacant space created when the double strand dissociates. The kinetic analysis (Figure 3) combined with the thermodynamic analysis (Figure 4) suggests a two-step mechanism resulting in most of the DNA monolayer being single-stranded. The first, occurring at very short times, involves the adsorption of double-stranded DNA that does not undergo denaturation. In the second stage, additional thiolated molecules adsorb and the surface becomes increasingly denser. It is clear from Figure 2B that any additional molecules that are adsorbed at this second step are in the form of thiolated ssDNA. We suggest that the complementary strand is ejected to allow more thiolated strands to adsorb, reaching a thermodynamically stable mixed monolayer composed of ∼15% dsDNA and ∼85% ssDNA (for the 26 nt long strands). The ratio between the adsorbed ds and ssDNA depends of course on the stability of the double strand, as compared to the free energy gained upon the formation of a thiol-surface bond. This interpretation suggests a correlation between the density on the surface and the conformation of the adsorbed DNA and is in agreement with the dsDNA monolayers obtained on silicon. On silicon, the monolayer formed is not as dense as on gold. That is, even at very short adsorption times on gold, the density is already much higher than that obtained overnight on silicon. Therefore, the adsorption-induced denaturation does not occur on silicon. Applying electrochemical methods, Weckerbarth et al.21,22 showed that a 10 bp dsDNA adsorbed on a gold surface was denatured. Their results indicate that denaturation is very inefficient in the first 10 s of adsorption, but increases with time. However, even after a long time, only about 15% of the dsDNA remained in its native form. Our results suggest that denaturation is faster than 10 s and shows directly that hours after denaturation was completed, the monolayer rearranged itself to accommodate as many thiolated strands as possible. The fact that nonspecific binding to the gold surface was found to be minimal (Figure 3, nonthiolated ssDNA adsorption) further supports the packing-induced denaturation model over the surfacenonspecific interaction model. Hence, our results strongly support the model in which the gain in energy due to the formation of a thiol-gold bond balances the energy required to separate two DNA strands and that this balance is only manifested at high enough DNA densities on the surface. Acknowledgment. This work was partially supported by the Grand Center at the Weizmann Institute. We thank Dr. H. Cohen and Dr. A. Vilan for the XPS measurements and fruitful discussions. Partial support from the Schmidt Minerva Center is acknowledged. Supporting Information Available: Figure demonstrating that the signal of the labeled complementary strand did not decrease when the washing time was increased, suggesting that the washing process did not induce dsDNA denaturation. This material is available free of charge via the Internet at http://pubs.acs.org. LA801437N