882
Langmuir 2006, 22, 882-886
Effects of Contact Force and Salt Concentration on the Unbinding of a DNA Duplex by Force Spectroscopy Mark Vander Wal, Sarah Kamper, Jennifer Headley, and Kumar Sinniah* Department of Chemistry & Biochemistry, CalVin College, 1726 Knollcrest Circle SE, Grand Rapids, Michigan 49546 ReceiVed August 29, 2005. In Final Form: December 6, 2005 We report that varying the contact force in force spectroscopy results in a significant shift in DNA unbinding forces, measured from short oligonucleotides using a PicoForce microscope. The contact force between a 30-mer complementary DNA-coated probe and surface was varied from 100 pN to 10 nN, resulting in a significant shift in the most abundant unbinding force measured between the duplex. When contact forces were set at 200 pN or less, which is generally considered to be a low contact force region for biomolecular force spectroscopy studies, the shift in DNA unbinding forces was significant with changes in contact force. The effect of the salt concentration on the DNA unbinding forces was also examined for a range of salt concentrations from 5 to 500 mM because the presence of salt ions is necessary to facilitate the hybridization process. Although an increase in salt concentration resulted in the facilitation of DNA multiple binding events during force spectroscopy measurements, no significant shift in unbinding forces was observed. Our experiment demonstrates that the wide variation in DNA unbinding forces reported in the literature (50-600 pN) for short oligonucleotides can be accounted for by the different contact forces used and shows little or no effect of the salt concentration used in those studies. Furthermore, this study demonstrates the importance of reporting contact forces in force spectroscopy measurements for quantitative comparisons between different biomolecular systems, especially for noncovalent-type interactions.
Introduction The past decade has seen the evolving of force spectroscopy as an analytical tool capable of measuring biomolecular interactions on the piconewton to the nanonewton scale.1-4 However, it has also become apparent that with noncovalent-type interactions typical of most biomolecular systems the interaction force between the recognition molecules is dependent on the loading rate at which the measurement is made. Whereas a number of researchers have shown this loading rate dependence on biomolecular systems, a few have also shown the lack of such a dependence, though admittedly over a limited loading range spanned by the atomic force microscope.5-8 Nonetheless, the importance of reporting loading rates at which force spectroscopy experiments are performed provides a measure of comparison between biomolecular systems. However, as recently described, several other factors may compound the direct quantitative comparison between different biological systems.9 Foremost are issues related to immobilization methods, cantilever calibrations, and solution conditions. In this letter, we examine two specific conditions that have a significant impact on force spectroscopy measurements, namely, the contact force and solution conditions. To test the effect of contact force and solution conditions, we chose a system that has been well studied over the past decade: the unbinding of short DNA duplexes, which form a part of DNA hybridization methods being developed for label-free detection by microarray-based techniques.10-13 Force spectro* Corresponding author. E-mail:
[email protected]. Phone: (616) 5266058. Fax: (616) 526-6501. (1) Heinz, W. F.; Hoh, J. H. Nanotechnology 1999, 17, 143. (2) Janshoff, A.; Netizert, M.; Oberdorfer, Y.; Fuchs, H. Angew. Chem., Int. Ed. 2000, 39, 3212. (3) Hugel, T.; Seitz, M. Macromol. Rapid Commun. 2001, 22, 989. (4) Allison, D. P.; Hinterdorfer, P.; Han, W. Curr. Opin. Biotechnol. 2002, 13, 47. (5) Evans, E. Annu. ReV. Biophys. Biomol. Struct. 2001, 30, 105. (6) Merkel, R.; Nassoy, P.; Leung, A.; Richie, K.; Evans, E. Nature 1999, 397, 50. (7) Dupres, V.; Menozzi, F. D.; Locht, C.; Clare, B. H.; Abbott, N. L.; Ceunot, S.; Bompard, C.; Raze, D.; Dufrene, Y. Nat. Methods 2005, 2, 515. (8) Auletta, T. J. Am. Chem. Soc. 2004, 126, 1577. (9) Sattin, B. D.; Pelling, A. E.; Goh, M. C. Nucleic Acid Res. 2004, 32, 4876.
scopy has been used to examine the annealing of short DNA oligonucleotides by quantitatively measuring the unbinding forces between complementary DNA strands from 12 to 30 base pairs. What is remarkable is that this simple system has shown the complexities that arise from force spectroscopy measurements. The seminal work of Lee and Colton14 inspired many others to investigate this system, and several other groups have reported unbinding forces ranging from 400 to 600 pN,14-17 whereas several recent measurements have shown the unbinding forces to be less than 100 pN.9,18,19 These differences in force measurements have been attributed to the possibility of different contact forces and solution conditions used in the measurements.9,19 The present study shows how varying the contact force and solution conditions can resolve these differences. As a basis for our study, we chose the 30-mer thiolated DNA sequence used by Strunz et al.18 because it provides an all-ornone complementary binding sequence, whereas the dynamic force spectroscopy study by the authors provides us a general range in which to expect the unbinding force between the DNA duplex. We examined probes prepared under two different conditions. In the first case, the complementary DNA oligonucleotides on the probe and surface were treated with 6-mercapto-1-hexanol (MCH) to minimize multiple contacts with the ultraflat gold surface.12,13,20 This technique enables the DNA to stand up on the surface and probe while facilitating the hybridization and the unbinding process. We also diluted the (10) O’Brien, J. C.; Stickney, J. T.; Porter M. D. Langmuir 2000, 16, 9559. (11) McKendry, R.; Zhang, J.; Arntz, Y.; Struntz, T.; Hegner, M.; Lang, H. P.; Baller, M. K.; Certa, U.; Meyer, E.; Guntherodt, H.-J.; Gerber, C. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 9783. (12) Zhou, D.; Sinniah, K.; Abell, C.; Rayment, T. Langmuir 2002, 18, 8278. (13) Zhou, D.; Sinniah, K.; Abell, C.; Rayment, T. Angew. Chem., Int. Ed. 2003, 42, 4934. (14) Lee, G. U.; Chrisey, L. A.; Colton, R. J. Science 1994, 266, 771. (15) Noy, A.; Vezenov, D. V.; Kayyem, J. F.; Meade, T. J.; Lieber, C. M. Chem. Biol. 1997, 4, 519. (16) MacKerell, A. D.; Lee, G. U. Eur. Biophys. J. 1999, 28, 415. (17) Schotanus, M.; Aumann, K.; Sinniah, K. Langmuir 2002, 18, 5333. (18) Strunz, T.; Oroszlan, K.; Schafer, R.; Guntherodt, H. J. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 11277. (19) Pope, L. H.; Davies, M. C.; Laughton, C. A.; Roberts, C. J.; Tendler, S. J. B.; Williams, P. M. Eur. Biophys. J. 2001, 30, 53.
10.1021/la0523560 CCC: $33.50 © 2006 American Chemical Society Published on Web 12/31/2005
Letters
DNA molecules on the probe to 1% surface coverage, using MCH as the diluent, to minimize multiple DNA interactions between the probe and surface.19 In the second case, the DNA molecules on the probe were not diluted with MCH, but the surface was treated with MCH. This enabled us to make a direct comparison with DNA rupture forces reported in the literature with a complete coverage of DNA molecules on the probe. We then examined how the contact force and changes in salt concentration affect the unbinding force of the 30-mer DNA duplex. We found that changes in contact force have a significant effect on the unbinding force measured by force spectroscopy irrespective of how the probe is treated. Because the range of forces we found covers the range of differences in unbinding forces reported in the literature, we have uncovered an important and overlooked variable that must be accounted for in binding studies. After controlling for the changes in contact force, we further demonstrate that changes in salt concentration have very little to no effect on the DNA unbinding force, while showing that multiple bindings are facilitated with increasing salt concentration. Experimental Section Materials. All chemicals and buffers were purchased from SigmaAldrich (St. Louis, MO) unless otherwise noted. Complementary nonrepeating oligonucleotides sequences 5′-thiol-GGCTCCCTTCTACCACTGACATCGCAACGG-3′ and its complement 5′-thiolCCGTTGCGATGTCAGTGGTAGAAGGGAGCC-3′ were purchased from Invitrogen (Carlsbad, CA) and used in all experiments reported here. Two buffers were used for force spectroscopy measurements and for all dilutions. A commercial phosphate buffer was purchased at a 10× concentration and diluted to 1× (10 mM phosphate, 500 µM NaCl, pH 7.5) using ultrapure water (18.2 MΩ cm resistivity) obtained from a Barnstead water purifier (Fisher Scientific). This buffer was used for all experiments investigating salt concentration dependence by adding variable amounts of NaCl. Another commercial buffer, sodium saline phosphate EDTA (SSPE), was purchased at a concentration of 20× and diluted to 1× (10 mM phosphate, 149 mM NaCl, 1 mM EDTA, pH 7.4) using ultrapure water. The SSPE buffer was used for all experiments related to the investigation of contact force. Substrate Preparation. Template stripped gold surfaces21 were prepared by gold (99.99% purity) evaporation of 200 nm onto mica surfaces that were annealed at 300 °C for ∼20 h in a vacuum evaporator (Denton Vacuum, NJ). During evaporation, the base pressure inside the vacuum chamber was held constant at around 8 × 10-6 Torr, and the mica surfaces were held at 300 °C. After coating, the Au/mica surfaces were cut into approximately 1 cm2 pieces and glued to Si(100) wafer pieces of slightly smaller area using 10 µL of EPO-TEK 377 epoxy (Epoxy Technology, MA). The sandwich surfaces were then heated in a vacuum oven (Salvis Lab, Switzerland) at 150 °C for 2 h. In preparation for DNA immobilization, the mica was stripped from the gold surface using tweezers after the application of tetrahydrofuran (THF, 99.9+%) for 3 to 4 min. The gold surfaces were then washed using THF followed by several milliliters of ultrapure water and placed immediately into a 1.0 µM thiolated single-stranded DNA solution for 20 h prior to use. Standard Si3N4 cantilevers (MLCT, Veeco Probes) were modified by 10 nm of chromium evaporation as an adhesion layer followed by 30 nm of gold evaporation (at a rate of 1 Å/s). They were then placed immediately into a solution composed of 100 µM 6-mercapto1-hexanol (97%) (MCH) and 1.0 µM thiolated single-stranded DNA that is a complement to the DNA placed on the surface or, in other experiments, into a solution consisting of only 1.0 µM thiolated DNA (without any MCH treatment). The tips were similarly incubated for 20 h prior to use in force measurements. Contact force variation (20) Steel, A. B.; Levicky, R. L.; Herne, T. M.; Tarlov, M. J. Biophys. J. 2000, 79, 975. (21) Hegner, M.; Wagner, P.; Semenza, G. Surf. Sci. 1993, 291, 39.
Langmuir, Vol. 22, No. 3, 2006 883 experiments were performed using both types of modified probes, whereas salt concentration studies were carried out using the 1:100 DNA/MCH probe. The gold-coated surfaces were placed in a 1.0 µM thiolated ssDNA for 20 h and then incubated in a 1.0 mM solution of MCH for 20 min. The DNA coverage on the surface is expected to decrease by 10-15% during the time of MCH exposure.20 The MCH minimizes DNA multiple contacts with the surface and displaces some of the immobilized DNA on the gold, effectively diluting the number of DNA molecules present in the contact area between the surface and tip. The MCH also facilitates binding by forcing the DNA oligonucleotides to stand up straighter on both the tip and surface for better interaction. Instrumentation. All force spectroscopy experiments were performed using a PicoForce atomic force microscope (Digital Instruments, Veeco Metrology, California) equipped with a 40 µm scanner (20 µm in the z direction). Spring constants of gold-coated Si3N4 cantilevers were measured using the thermal fluctuation method22 and were found to be 20 ( 5 pN/nm. Force Spectroscopy Measurements. Force spectroscopy measurements were performed in a glass fluid cell (Digital Instruments) at room temperature (22 °C). Force measurements were conducted in phosphate buffer with varying concentrations of added NaCl. The maximum contact force was maintained at 150 pN for all salt concentration studies. The ramp rate was 0.5 Hz with a ramp size of 250 nm. A constant loading rate of 5000 pN/s was applied for all force measurements. Measurements were taken over a period of 2 h until approximately 100 specific unbinding force curves were collected from approximately 1000-1800 force-distance data curves in at least 15-20 random spots on the surface. Special attention was paid to the shape of force curves both in data collection and analysis. Only force curves that indicated a change in the slope of the retraction force curve were used in the analysis. Control experiments were performed with identical DNA sequences on both the tip and surface to verify the specificity of the complementary DNA interaction. The majority of the force curves showed zero adhesion or nonspecific interactions, whereas a small fraction of force curves showed specific interactions similar to those reported by Strunz et al.18 Force spectroscopy measurements for contact force investigations were conducted in 1X SSPE buffer under the conditions described above. The contact force was varied from 100 pN to 10 nN, and approximately 100 specific unbinding force curves were collected at each selected contact force. All force-distance curves were analyzed by using the DI Nanoscope offline software provided by Veeco Metrology (version 6.1). Force data was analyzed using the freeware statistical package R (http://www.r-project.org/). Kernel density function plots were generated to examine the distribution of forces. All density curves reported were generated using 512 points and the Gaussian kernel function.
Results and Discussion Contact Force Effects. To test the effect of the varying unbinding force on a receptor-ligand type interaction, the 30mer DNA and its complement were examined by force spectroscopy at several different contact forces. Figure 1 shows a kernel density function (kdf) plot of the DNA unbinding forces at a contact force of 100 pN using a DNA-functionalized AFM probe that was diluted with MCH in a 1:100 ratio of 1 µM DNA/100 µM MCH. As has been reported previously, the kernel density functions are better for the visualization of force distribution data.23,24 On the basis of the kdf plot, the unbinding forces resulting from single-molecule interactions can be easily (22) Hutter, J. L.; Bechhoefer, J. ReV. Sci. Instrum. 1993, 64, 1868. (23) Baumgartner, W.; Hinterdorfer, P.; Schindler, H. Ultramicroscopy 2000, 82, 85. (24) Force distribution data have been typically shown in the form of a histogram. The ambiguity of the histogram shape resulting from assigning an arbitrary bin size or bin width complicates the interpretation of such plots, whereas the kernel density function is an improvement on the histogram.
884 Langmuir, Vol. 22, No. 3, 2006
Letters
Figure 1. Kernel density function of DNA unbinding force data from a 30-mer DNA duplex at a 100 pN contact force between the probe and surface. The graph also shows underlying Gaussian fits that best fit the kernel density function representing one, two, and three unbinding events, respectively.
Figure 2. Kernel density function of DNA unbinding force data taken over a range of contact forces from 100 pN to 10 nN with (a) a DNA-coated probe with no MCH treatment and (b) a DNA-coated probe with 1:100 dilution of 1 µM DNA/100 µM MCH.
extracted. By carefully eliminating nonspecific interactions and including only specific interactions (i.e., force curves that display a change in slope), Figure 1 demonstrates that even at a low contact force (100 pN) a significant number of multiple interactions are present. The kernel density function can be best fitted to three Gaussians, showing the unbinding forces (centered around 55 ( 8, 103 ( 9, and 170 ( 11 pN) arising from one, two, or three unbinding events. Uncertainties in the force measurements are reported at the 96% confidence level. The single-molecule unbinding event, 55 ( 8 pN, corresponds very well to the most probable unbinding force measured by Strunz et al., who obtained a value of ∼50 pN for the identical DNA sequence.18 Upon changing the contact force from 100 pN to 10 nN and measuring the unbinding force between the DNA duplex at each contact force, the peak of the most abundant feature of the kernel density function shifts to higher forces as shown in Figure 2b. This shift is most likely a result of the number of DNA unbinding events taking place within the contact area between the AFM probe and surface as well as the increase in nonspecific-type interactions at higher contact forces. It is also important to note that at low contact forces the distribution of forces is narrow whereas at higher contact forces the distribution of forces spans a wide range. This result is most likely due to the fact that at low contact forces the pull-off forces measured arise from a smaller contact area between the tip and
Figure 3. Representative force-distance curves at varying contact forces. (A) 100 pN, (B) 200 pN, (C) 500 pN, (D) 1 nN, (E) 5 nN, and (F) 10 nN.
surface, resulting in the unbinding of single, double, or triple unbinding events. Even at a small contact force of 100 pN, the kernel density function provides evidence for the presence of multiple unbinding events. At higher contact forces, the tipsurface contact area is much larger, resulting in an abundance of multiple interactions as well as the increased possibility of nonspecific-type interactions. Although some nonspecific-type interactions are easy to identify, small variations in force curves are not easy to identify without biasing the data. At very high contact forces (5 and 10 nN), multiple interactions are limited by the contact area of the radius of the tip (typically around 30-50 nm), resulting in only a marginal shift in the most abundant forces in the higher contact force regime. We also examined the effect of contact force using an undiluted DNA probe. Because DNA unbinding measurements have been reported with diluted and undiluted DNA probes, our hypothesis was that the undiluted probe would yield multiple interactions even at low contact forces, resulting in larger unbinding forces. However, we were surprised to find that even at low contact forces (i.e., 100 pN) the kernel density function appeared to be similar to those obtained with the probe molecule diluted (Figure 2a). In fact, the contact force trend observed with the DNA molecule diluted probe is repeated with the undiluted DNA probe, showing how increasing the contact force affects them both similarly. Representative force-distance curves at each contact force are shown in Figure 3. The peak force relating to the most abundant feature in the kernel density function was plotted against the contact force and is shown in Figure 4. This Figure demonstrates the importance
Letters
Langmuir, Vol. 22, No. 3, 2006 885
Figure 4. Maximum of the most abundant peak in the kernel density function plotted as a function of contact force. (a) DNA probe with no MCH treatment and (b) DNA probe diluted 1:100 with MCH.
of controlling contact forces even within the low contact force regime irrespective of whether the sensing molecule on the probe was diluted. The peak of the most abundant feature of the kdf plot shows slightly higher forces in the low contact force regime for the undiluted DNA probe presumably because of a greater number of multiple DNA interactions. The increase in the most abundant unbinding forces highlighted within the low contact force regime in Figure 4 suggests that multiple interactions tend to dominate even with marginal increases in contact forces. Because most noncovalent-type interactions show adhesion forces of less than 200 pN, control of the contact force is critical in force collection experimental procedures. Although the importance of the loading rate on the adhesion force has been recognized, the importance of the contact force on adhesion forces, especially for biomolecular interactions, is yet to be fully recognized. Figure 4 also provides the means to correlate the discrepancy of DNA unbinding forces reported in the literature. The high unbinding forces reported for DNA duplexes of varying lengths range from 330 to 600 pN at different salt concentrations.14-17 As we will demonstrate below, it is unlikely that the differences in salt concentration provide for this large variation in unbinding forces. It is, however, more likely that the influence of the contact force was not fully appreciated and was not, as a result, controlled in these experiments. Second, this large variation in unbinding forces reported may likely be influenced by the number of DNA molecules interacting at the contact point. Thus, one would expect to observe higher unbinding forces from the undiluted DNA probe resulting from a greater number of multiple contacts. However, according to Figure 4, at contact forces >1 nN we do not observe a significant change in the apparent unbinding forces between the two differently prepared probes. It is possible that at higher contact forces it is harder to discriminate between specific and nonspecific interactions, and it is this lack of discrimination that is evident in the data between the diluted and undiluted
Figure 5. Kernel density functions of DNA unbinding forces taken at varying salt concentrations from 5 to 500 mM NaCl at a 150 pN contact force.
DNA probes. In the low contact force regime, several published reports have shown the DNA unbinding forces to be less than 100 pN for a number of different length DNA oligomers.9,18,19 However, as seen in Figure 4, even under low contact forces a significant shift in unbinding forces is to be expected. Thus, a comparison of noncovalent-type interactions of different biomolecular systems is complicated by this important factor, which necessitates the reporting of contact forces used in experiments even for qualitative comparisons. Salt Concentration Effects. While controlling for contact forces, we have examined the salt concentration effects on the DNA unbinding force by varying the salt concentration in 1 µM DNA from 5 to 500 mM. Figure 5 shows the kernel density function of the DNA unbinding force at four different salt concentrations in 10 mM phosphate buffer: 5, 150, 250, and 500 mM NaCl. What is clearly noticeable from the data in Figure 5 is that single and multiple DNA interactions are present at all of the salt concentrations, whereas the abundance of multiple interactions is greater at higher salt concentrations. By fitting the kernel density function to the minimum number of Gaussians that provides for the best fit, the unbinding forces from single and multiple binding events were extracted and are shown in Table 1. Using a minimal contact force of 150 pN, we are able to measure the unbinding force of up to five unbinding
886 Langmuir, Vol. 22, No. 3, 2006
Letters
Table 1. Peak Force Values from the Best-Fit Gaussian Distributions to the KDF Functions Shown in Figure 4 with Uncertainties Reported at the 96% Confidence Level salt one two three four five concn unbinding unbinding unbinding unbinding unbinding (mM) event (pN) events (pN) events (pN) events (pN) events (pN) 5 150 250 500
60 ( 9 58 ( 14 54 ( 6
101 ( 9 103 ( 7 98 ( 6 101 ( 14
150 ( 5 150 ( 7
186 ( 23 197 ( 7 189 ( 11
277 ( 24 286 ( 14
events. Not all five unbinding events are clearly discernible at each salt concentration, but what is apparent is that there appears to be no discernible shift within experimental error in the unbinding forces measured for each unbinding event as the salt concentration of the buffer increases. Whereas increasing the salt concentration is expected to screen the DNA backbone, which facilitates the hybridization process, it does not appear to have a significant bearing on the unbinding force in force spectroscopy measurements other than facilitating multiple binding effects. Although most biomolecular interaction studies and DNA unbinding studies have been carried out at a physiologically relevant salt concentration (∼150 mM NaCl), discrepancies in the unbinding forces for noncovalent interactions have been generally thought to be due to environmental factors such as the variation in salt concentration. If there is an environmental effect at all, our study indicates that it is marginal at best and its contribution to the overall unbinding force is minimal, whereas differences in contact force and loading rates appear to dominate the unbinding or rupture forces measured in biomolecular systems.
Conclusions We have examined DNA unbinding forces by varying the contact forces between the tip and surface in force spectroscopy measurements and by varying the salt concentration in the buffer solution in which the DNA pull-off measurements were performed. Our results show that the unbinding forces between DNA strands are significantly affected by both high and low contact forces applied between the probe and surface, and this explains the variation in DNA unbinding forces reported in the literature. Additionally, we find that while holding the contact force constant, changes in salt concentration in the range from 5 to 500 mM show a minimal effect on the DNA unbinding forces. Our results suggest that the adhesion forces measured between biomolecular interactions by force spectroscopy are also dependent on contact forces in addition to loading rates and that strict control of these experimental parameters is required for quantitative comparisons between different biomolecular systems. Acknowledgment. We thank the donors of the American Chemical Society Petroleum Research Fund (grant 39585-B5) for supporting this research. We thank the NIH (grant 1R15GM073662-01) for supporting the purchase of the PicoForce atomic force microscope used in this study. We also thank Professor Michael Stob (Department of Mathematics, Calvin College) for helpful discussions on the statistical analysis of the data. M.V.W. was supported by a grant from Merck-AAAS. LA0523560
