pubs.acs.org/NanoLett
Acoustic Characterization of Nanoswitch Structures: Application to the DNA Holliday Junction George Papadakis,†,¶ Achilleas Tsortos,†,¶ and Electra Gizeli*,†,‡ †
Institute of Molecular Biology and Biotechnology, Foundation for Research and Technology Hellas, 100 N. Plastira, Vassilika Vouton, 70013 Heraklion, Greece, and ‡ Department of Biology, University of Crete, Vassilika Vouton, 71409 Heraklion, Greece ABSTRACT A novel biophysical approach in combination with an acoustic device is demonstrated as a sensitive, rapid, and labelfree technique for characterizing various structures of the DNA Holliday Junction (J1) nanoswitch. We were successful in discriminating the “closed” from the “open” state, as well as confirming that the digestion of the J1 junction resulted in the two, anticipated, rodshaped, 20 bp long fragments. Furthermore, we propose a possible structure for the ∼10 nm long (DNA58) component participating in the J1 assembly. This work reveals the potential of acoustic devices as a powerful tool for molecular conformation studies. KEYWORDS Acoustic biosensor, QCM-D, nanomachines, intrinsic viscosity, structural characterization, conformational transition
S
tructural DNA nanotechnology is an emerging new field that exploits the ability of DNA to self-assemble into a variety of two and three-dimensional nanostructures and nanomachines. DNA can be used as a bottomup construction medium of artificial nanodevices and as a building block of periodic assemblies and complex nanostructures.1 DNA scaffolding was initially suggested to be used for the construction of crystalline cages that could orient other biological molecules inside those cages.2 More recently, the controlled folding of a viral DNA into various nanoscale shapes and patterns was reported, a procedure named DNA origami.3 A few years later this approach led to the creation of a 3D box with a controllable lid that can potentially carry and release nanocargos inside a cell.4 DNA origami was used to make a ruler to measure distances between single molecules and calibrate superhigh resolution microscopes.5 DNA is also a candidate for molecular computing6 and for nanoelectronic circuits, for example, in plasmonic devices.7 Finally, synthetic DNA nanomachinery was constructed in order to create switches or actuators that move between two states in response to a molecular or environmental trigger. Such nanomachines display potential both in medical8 and computational applications. Characterization of newly designed and assembled DNA nanostructures is of great importance but sometimes difficult to perform (e.g., requiring high-resolution AFM images).3 In the current study, we combine the experimental technique of quartz crystal microbalance (QCM-D) with a novel theo-
retical approach for the characterization of a DNA switchstructure comprising a four-way DNA junction with a central branch point. In contrast to the standard QCM-D approach where the surface bound biomolecules are modeled as a film and energy dissipation changes are (quantitatively) attributed to changes related to the amount of water entrapped within this film, our treatment is based on a “discrete molecule” approach. This approach was recently shown to be perfectly capable of detecting and quantifying DNA structural differences, such as in length, intrinsic curvature, and externally forced DNA bending.9-12 The details and principles of operation have been described before. Briefly, the presence of an analyte at the sensor’s surface affects the propagation characteristics of an acoustic wave, that is, its velocity and amplitude, which in turn, are monitored as changes in frequency (∆f) and energy dissipation (∆D). In our case, frequency changes are proportional to adsorbed mass13 and dissipation is sensitive to the viscoelastic properties of the bound molecules. Theoretical treatment of experimental data revealed that energy dissipation per unit mass, ∆D/∆f, can be used as a direct measure of the intrinsic viscosity [η] of the surface-attached DNA molecules.11,12 The significance of this finding can be appreciated when one realizes that the intrinsic viscosity is a parameter characteristic of a molecule and depends on its hydrodynamic volume and shape. While the QCM-D is applied here as the measuring technique, it should be noted that any liquid-operating acoustic sensor providing f (or phase) and D (or amplitude or resistivity) measurements could also be used in combination with the “discrete-molecule” model. In this work, the J1 form14 of the Holliday junction (Figure 1), known to be immobile due to the specific sequence around the branch point,15,16 was chosen as a
* To whom correspondence should be addressed. Tel: (30) 2810 394373 (Office). Tel: (30) 2810 394093 (Lab). Fax: +30 2810 394408. E-mail address: gizeli@ imbb.forth.gr. ¶ These authors contributed equally. Received for review: 10/5/2010 Published on Web: 11/01/2010
© 2010 American Chemical Society
5093
DOI: 10.1021/nl103491v | Nano Lett. 2010, 10, 5093–5097
FIGURE 3. The open conformation of the structure (without Mg2+ ions) is switched to the closed state by pumping over the surface of the sensor a 20 mM Tris, pH 7.5, 50 mM NaCl buffer including 5 mM MgCl2. The transition was monitored in real time by measuring the changes in both frequency and energy dissipation of the acoustic sensor. Alternating the buffer (with or without Mg2+) over the surface allows for multiple switching.
FIGURE 1. The J1 Holliday junction consists of two parts, a 58 base long DNA probe (DNA-58, black color) that was synthesized with a biotin molecule attached at the 5′ end and an 18 base long (DNA18, red color) target oligo. The assembly of the J1 (DNA-58 hybridized with DNA-18) on the sensor’s surface was performed in a 20 mM Tris, pH 7.5, 50 mM NaCl with 5 mM MgCl2 buffer or in bulk solution in the same buffer with or without the 5 mM MgCl2.
vidin layer in Tris/Na buffer in the absence of divalent Mg ions and in adequately law salt solutions.22 In our case (50 mM of NaCl), these molecules are known to adopt an open conformation (Figure 2A(i)). Magnesium ions were subsequently pumped over the sensor surface. This buffer exchange, which resulted in a significant decrease in dissipation and a minor increase in frequency (Figure 3), is attributed, primarily, to the transition toward a closed conformation of the junction (Figure 2A(ii)). The phenomenon could be reversed by switching to the Mg2+ free buffer. Conformational transitions could be repeated many times by continuous buffer exchange. Buffer exchange over the neutravidin surface caused a very small change; these values were subtracted from the total signal changes so as to calculate the values that correspond solely to conformational switching between the two Holliday junction forms. The acoustic ratio ∆D/∆f (6-10 measurements) for both the open and the closed states is then calculated. This was found to be 0.0139 ( 0.0016 (10-6/Hz) for the closed and ∼17%
model system of importance to both biology and nanotechnology. Holliday junctions are intermediate structures that are formed and resolved during the process of genetic recombination.17,18 In addition, DNA nanotechnology utilizes Holliday junctions as molecular nanomachines capable of moving between distinct states after being triggered by various electrochemical, optical, or thermal stimuli. This particular molecular switch, existing in either an open (extended) or closed (coaxially stacked) conformation,19 has also been used as a prototype for sequence-specific nucleic acid recognition.20,21 The first line of experiments studied conformational switching and discrimination between structures that are identical in terms of their mass content but exhibit distinct structural differences. Solution preassembled and biotinylated J1 molecules were initially immobilized on a neutra-
FIGURE 2. Schematic illustration of different forms of the J1 Holliday junction immobilized on a QCM device Au surface through biotinneutravidin interactions. (A) prehybridized open (i) and closed (ii) forms of the J1 molecule in the absence and presence of Mg2+, respectively. (B) In situ hybridization of DNA-58 (i) to give the closed (ii) J1 structure; the latter is subsequently digested to produce a captured biotinylated double stranded (iii) and a free, not sensed, nonbiotinylated (iv) DNA fragment (all the above in the presence of Mg2+). (C) DNA-58 in the absence (i) and presence (ii) of Mg2+. © 2010 American Chemical Society
5094
DOI: 10.1021/nl103491v | Nano Lett. 2010, 10, 5093-–5097
ments. Looking at the biotinylated DNA-58 molecule now, it should be noted that not much is known about its actual structure in solution. In the absence of Mg2+ the molecule should adopt an extended, openlike conformation with one missing single strand; the remaining strand has of course the great flexibility of any single-stranded DNA due to its small persistence length. On the other hand, in the presence of Mg2+ the persistence length becomes even smaller and the single-chain arms should approach adopting a more closedlike conformation. According to theory (see ref 19 and references therein), for hydrodynamic reasons, compact/ small structures are less “lossy” energetically speaking and consequently give smaller acoustic ratios compared to more open and extended conformations. This should be reflected in the corresponding acoustic ratios for each structure under different conditions. Acoustic experiments produced the following values for the DNA-58 fragment: ∆D/∆f ≈ 0.0189 ( 0.0017 (10-6/Hz) and ≈ 0.0150 ( 0.002 in the absence and presence of Mg2+, respectively. These numbers indeed support our expected/proposed structures. The presence of Mg2+ causes the single-stranded arms of DNA-58 to contract and bring the acoustic ratio closer to that of J1-closed form; its absence, makes it more lossy with a value for the acoustic ratio even greater than that of J1-open form (probably due to its higher side chain flexibility compared to the rather stiffer intact open junction). It is tempting and justified, then, to propose the forms shown in Figure 2C for the two structures of DNA-58 depending on buffer composition. Many studies have looked at the actual structures of Holliday junctions with various techniques, most commonly with microscopy and X-ray crystallography. For example, rather long sequences of cis-trans isomers have been observed by transmission electron microscopy (TEM),23 in recombination intermediates24and as predesigned sequences for protein interactions.25,26 Scanning confocal fluorescence microscopy has been used on a small junction27 but for dynamic measurements only. The results have shown that in TEM studies the majority of the DNAs may have altered structures due to the mounting procedure;25 even when experimental conditions favoring, for example, the closed form (resulting in a ∼40° interduplex crystallographic angle19) were chosen, in reality a variety of structures (open cross, Y-shape, kshape, straight rod) were observed. Our own attempts at observing microscopically (TEM) the very small (∼10 nm) J1 junction were not successful in distinguishing the various forms with the detail features/differences shown in Figure 2. It is then apparent that the acoustic ratio can be used as an alternative tool to differentiate the corresponding structures before and after hybridization and even predict their conformation based on the corresponding acoustic ratio of similar molecules of known conformation. In a third type of experiments, the J1 junction was resolved (digested) in solution by T7 endonuclease I. The structure of the DNA branchpoint is the key factor responsible for the strong selectivity of the junction resolving
FIGURE 4. Real time monitoring of frequency and dissipation changes corresponding to the addition of DNA-58 and DNA-18 on the acoustic sensor device. Here, ∆D/∆f ) (∆DDNA-58 + ∆DDNA-18)/ (∆fDNA-58 + ∆fDNA-18).
higher, that is, 0.0167 ( 0.0007 (10-6/Hz) for the open conformation. In agreement with previous studies, this ∆D/∆f ratio represents a way to identify the immobilized molecule that is characterized by a particular size and shape. In accordance with the theory12stating that ∆D/∆f ∼ [η], this difference is indeed validated here by direct measurement of the intrinsic viscosity of the two Holliday structures. Microviscometry results showed that the open form is approximately 8% different from the closed one, that is, [η]closed/[η]open ≈ 0.92, which is in good agreement with the above stated difference. The second line of experiments describes hybridization in situ, that is, the assembly of J1 on the device surface in the Mg2+ buffer. First, DNA-58 was immobilized on the neutravidin layer via its biotin linker (Figure 2B(i)) and changes in both the frequency and dissipation signals were recorded (Figure 4). Addition of the complementary single stranded DNA-18 resulted in further changes in both signals; as a control, addition of a noncomplementary 18 base DNA oligo had no measurable effect, as expected. Frequency changes correspond primarily to bound mass;12,13 consequently, direct comparison of the two frequency values can give an estimate of the hybridization efficiency for the two molecules. Given the molecular weights of DNA-58 and DNA18, 18.17, and 5.55 kD, respectively, the resulting weight ratio is ∼3.3. Calculating the ratio from the frequency changes ∆fDNA-58/∆fDNA-18, we get an average of ∼3.5 ( 0.4; comparison of the two values suggests a hybridization efficiency of better than 94%. Turning now to the calculation of acoustic ratios we note the following. For in situ hybridization, the newly formed complex should, obviously, have the exact same structure as that of the (bulk solution) prehybridized J1 structure; according to our approach, very close (if not identical) values for the corresponding acoustic ratios should be obtained. Indeed, the ∆D/∆f ratio measured for the in situ hybridization product is ≈0.0135 ( 0.0015 (10-6/Hz), almost identical to the 0.0139 ( 0.0016 value characterizing the preformed Holliday junction conformation used in the switching experi© 2010 American Chemical Society
5095
DOI: 10.1021/nl103491v | Nano Lett. 2010, 10, 5093-–5097
tion of this platform in monitoring the switching from the open to the closed form of a DNA nanostructure by measuring the changes in frequency and dissipation. The switching, caused by buffer exchange, was fully reversible and could be repeatedly obtained. Differences in the conformation of the open and closed J1 structures were clearly discriminated using the ∆D/∆f ratio derived from the acoustic measurements. This ratio was previously reported to be related to the intrinsic viscosity of immobilized molecules that reflects on the hydrodynamic size and shape of the molecules. The third novel finding is related to viscosity measurements of the junction in the presence and absence of magnesium ions; for the first time, it is experimentally proven that the ∆D/∆f value is directly related to the intrinsic viscosity/ conformation of the molecular switch. One should note that the acoustic technique could not discriminate between the closed J1 structure from the enzymatically produced rodshaped fragment. Discriminating between two ∼7 nm long, rod-shaped molecules differing in their diameter by ∼2 nm is quite a challenge even for high-resolution imaging techniques. However, it was shown that the acoustic approach can be effective even for such small structures in the case where the compared molecules differ significantly hydrodynamicaly. The importance of the acoustic approach described in this paper should be further evaluated in relation to information derived from standard, conformation-sensitive techniques. FRET has been extensively used for the dynamic study of conformational switching of various Holliday junction nanomachines in a milliseconds range and for single molecules.30,31 This technique relies on the use of two or more fluorophores whose relative distance can be translated (indirectly) to information related to the molecule’s conformation. The acoustic approach presented in this work goes a step beyond; through the direct measurement of molecular hydrodynamic volume (intrinsic viscosity) it is possible, not only to discriminate between various molecular structures and monitor conformational-switching events, but also to define quantitatively (through the acoustic ratio) the molecule’s shape and size (for example, rod-shape, 20 bp). One should also compare our method to gel electrophoresis, the standard lab-based technique used for structural analysis and differentiation. A major difference and significant advantage of the acoustic approach is the ability to perform real time analysis of structural changes, something not possible with gel electrophoresis. In fact, one could view the acoustic method described in the current paper as a combination of the two aforementioned techniques. Given the above, we anticipate that acoustic wave devices in combination with the “discrete-molecules” approach (as opposed to the older “film” one) will find significant applications in structural nanotechnology, opening up the possibility for the development of conformation-based biosensors.
FIGURE 5. The assembled J1 structure was successfully resolved by adding T7 endonuclease I and incubating at 37 °C for 1 h. The produced DNA fragments were loaded on a 2% agarose gel (lane 2) along with the unresolved structure (lane 3) and a DNA marker (lane 1).
enzymes,15,28 although cleavage efficiency can vary slightly according to DNA sequence.28 T7 endonuclease I is a basic protein containing 149 amino acids. It consists of two identical subunits that can contact the backbones of the junction over seven nucleotides. This enzyme recognizes and cleaves Holliday structures by introducing nicks in strands that are opposed across the branchpoint.29 As a result, linear duplex DNA molecules with half the size of the original molecule are produced. The electrophoretic behavior of such digested products is equivalent to oligonucleotide dimers.28 Indeed, the solution-resolved structures displayed this icon upon electrophoresis with 2% agarose gels (Figure 5). In our case, the cleaved duplexes derived from the resolved J1 structure consisted of one biotinylated (Figure 2B(iii)) and one nonbiotinylated (Figure 2B(iv)) product while no traces of undigested J1 were identified. The isolated DNA fragments were subsequently taken and loaded on a neutravidin-modified sensor surface and the acoustic ∆D/∆f ratio for the captured biotinylated fragment was measured. Since the size of the cleaved biotinylated duplex is about 18-20 bp long, the expected value should be near the previously measured one for a 20 bp molecule, that is, ∼0.0138 (10-6/Hz).12 Indeed, the ratio obtained here for the cleaved duplex is 0.0132 ( 0.0013 (10-6/Hz). This result further demonstrates the ability of our methodology to characterize the resolved Holliday junction structure and in general, confirm the formation of the correct product of a biochemical reaction based on its expected structure. Interestingly, acoustic ratios indicate that the structures of the cleaved duplex (Figure 2B(iii)) and closed J1 (Figure 2B(ii)) are similar, suggesting that the two arms in the closed form are rather tightly arranged next to each other, pretty much resembling a straight rod. This observation is in agreement with microscopy images of an other Holliday junction.25 To summarize, three novel findings are demonstrated in this work. First, the development of a detection platform, not sensitive to the commonly probed mass, electrical, or optical properties of the target analyte, but instead very sensitive to the conformation of the attached nanostructure is demonstrated. Second, we showed the successful applica© 2010 American Chemical Society
Acknowledgment. The authors acknowledge the Human Frontier Science Program and the Institute of Molecular 5096
DOI: 10.1021/nl103491v | Nano Lett. 2010, 10, 5093-–5097
Biology and Biotechnology-FORTH (Greece) for financially supporting this work.
(15) Hadden, J. M.; De´clais, A.-C.; Carr, S. B.; Lilley, D. M. J.; Phillips, S. E. V. Nature 2007, 449, 621–625. (16) Picksley, S. M.; Parsons, C. A.; Kernper, B.; West, S. C. J. Mol. Biol. 1990, 212, 723–735. (17) Holliday, R. Genet. Res. 1964, 5, 282–304. (18) Liu, Y.; West, S. C. Nat. Rev. Mol. Cell Biol. 2004, 5, 937–946. (19) Hays, F. A.; Jeffrey Watson, J.; Shing Ho, P. J. Biol. Chem. 2003, 278, 49663–49666. (20) Buck, A. H.; Campbell, C. J.; Dickinson, P.; Mountford, C. P.; Stoquert, H. C.; Terry, J. G.; Evans, S. A. G.; Keane, L. M.; Su, T.J.; Mount, A. R.; Walton, A. J.; Beattie, J. S.; Crain, J.; Ghazal, P. Anal. Chem. 2007, 79, 4724–4728. (21) Ferapontova, E. E.; Mountford, C. P.; Crain, J.; Buck, A. H.; Dickinson, P.; Beattie, J. S.; Ghazal, P.; Terry, J. G.; Walton, A. J.; Mount, A. R. Biosens. Bioelectron. 2008, 24, 422–428. (22) Mount, A. R.; Mountford, C. P.; Evans, S. A. G.; Su, T.-J.; Buck, A. H.; Dickinson, P.; Campbell, C. J.; Keane, L. M.; Terry, J. G.; Beattie, J. S.; Walton, A. J.; Ghazal, P.; Crain, J. Biophys. Chem. 2006, 124, 214–221. (23) Yakubovskaya, M. G.; Neschastnova, A. A.; Humphrey, K. E.; Babon, J. J.; Popenko, V. I.; Smith, M. J.; Lambrinakos, A.; Lipatova, Z. V.; Dobrovolskaia, M. A.; Cappai, R.; Masters, C. L.; Belitsky, G. A.; Cotton, R. G. Eur. J. Biochem. 2001, 268, 7–14. (24) Hsu, M.-T. Nucleic Acids Res. 1991, 19, 7193–7199. (25) Lee, S.; Cavallo, L.; Griffith, J. D. J. Biol. Chem. 1997, 272, 7532– 7539. (26) Compton, S. A.; Tolun, G.; Kamath-Loeb, A. S.; Loeb, L. A.; Griffith, J. D. J. Biol. Chem. 2008, 283, 24478–24483. (27) Zheng, H.; Goldner, L. S.; Leuba, S. H. Methods 2007, 41, 342– 352. (28) Dickie, P.; McFadden, G.; Morgan, A. R. J. Biol. Chem. 1987, 262, 14826–14836. (29) Muller, B.; Jones, C.; West, S. C. Nucleic Acids Res. 1990, 18, 5633– 5636. (30) Buranachai, C.; McKinney, S. A.; Ha, T. Nano Lett. 2006, 6 (3), 496–500. (31) McKinney, S. A.; De´clais, A.-C.; Lilley, D. M. J.; Ha, T. Nat. Struct. Biol. 2003, 10 (2), 93–97.
Supporting Information Available. Additional information regarding the DNA sequence, J1 assembly, T7 endonuclease and the viscosity measurements. This material is available free of charge via the Internet at http://pubs.acs.org. REFERENCES AND NOTES (1) (2) (3) (4)
(5) (6) (7)
(8) (9) (10) (11) (12) (13) (14)
Niemeyer, C. M. Curr. Opin. Chem. Biol. 2000, 4, 609–618. Seeman, N. C. Mol. Biotechnol. 2007, 37, 246–257. Rothemund, P. W. K. Nature 2006, 440, 297–302. Andersen, E. S.; Dong, M.; Nielsen, M. M.; Jahn, K.; Subramani, R.; Mamdouh, W.; Golas, M. M.; Sander, B.; Stark, H.; Oliveira, C. L. P.; Pedersen, J. S.; Birkedal, V.; Besenbacher, F.; Gothelf, K. V.; Kjems, J. Nature 2009, 459, 73–77. Steinhauer, C.; Jungmann, R.; Sobey, T. L.; Simmel, F. C.; Tinnefeld, P. Angew. Chem., Int. Ed. 2009, 48, 8870–8873. Seelig, G.; Soloveichik, D.; Zhang, D. Y.; Winfree, E. Science 2006, 314, 1585–1588. Kershner, R. J.; Bozano, L. D.; Micheel, C. M.; Hung, A. M.; Fornof, A. R.; Cha, J. N.; Rettner, C. T.; Bersani, M.; Frommer, J.; Rothemund, P. W. K.; Wallraff, G. M. Nat. Nanotechnol. 2009, 4, 557–561. Modi, S.; Swetha, M. G.; Goswami, D.; Gupta, G. D.; Mayor, S.; Krishnan, Y. Nat. Nanotechnol. 2009, 4, 325–330. Papadakis, G.; Tsortos, A.; Gizeli, E. Biosens. Bioelectron. 2009, 25, 702–707. Papadakis, G.; Tsortos, A.; Mitsakakis, K.; Gizeli, E. FEBS Lett. 2010, 584, 935–940. Tsortos, A.; Papadakis, G.; Mitsakakis, K.; Melzak, A. K.; Gizeli, E. Biophys. J. 2008, 94, 2706–2715. Tsortos, A.; Papadakis, G.; Gizeli, E. Biosens. Bioelectron. 2008, 24, 836–841. Bender, F.; Roach, P.; Tsortos, A.; Papadakis, G.; Newton, M. I.; McHale, G.; Gizeli, E. Meas. Sci. Technol. 2009, 20, 124011. Miick, S. M.; Fee, R. S.; Millar, D. P.; Chazin, W. J. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 9080–9084.
© 2010 American Chemical Society
5097
DOI: 10.1021/nl103491v | Nano Lett. 2010, 10, 5093-–5097