J. Phys. Chem. B 2008, 112, 2439-2444
2439
Molecular Recognition with DNA Nanoswitches: Effects of Single Base Mutations on Structure C. P. Mountford,†,# A. H. Buck,‡ C. J. Campbell,‡,§ P. Dickinson,‡ E. E. Ferapontova,§ J. G. Terry,| J. S. Beattie,‡ A. J. Walton,| P. Ghazal,‡ A. R. Mount,§ and J. Crain*,†,⊥,# School of Physics, The UniVersity of Edinburgh, Mayfield Road, Edinburgh EH9 3JZ, United Kingdom, The DiVision of Pathway Medicine, The UniVersity of Edinburgh, The Chancellor’s Building, 49 Little France Crescent, Edinburgh, EH16 4SB Scotland, United Kingdom, School of Chemistry, The UniVersity of Edinburgh, West Mains Road, Edinburgh EH9 3JJ, United Kingdom, Institute of Integrated Systems, Scottish Microelectronics Centre, School of Engineering and Electronics, UniVersity of Edinburgh, Edinburgh, EH9 3JF, United Kingdom, National Physical Laboratory, Hampton Road, Teddington, Middlesex, TW11 0LW, and CollaboratiVe Optical Spectroscopy Micromanipulation and Imaging Centre (COSMIC), The UniVersity of Edinburgh, Mayfield Road, Edinburgh EH9 3JZ, United Kingdom ReceiVed: May 17, 2007; In Final Form: September 7, 2007
This paper investigates the properties of a simple DNA-based nanodevice capable of detecting single base mutations in unlabeled nucleic acid target sequences. Detection is achieved by a two-stage process combining first complementary-base hybridization of a target and then a conformational change as molecular recognition criteria. A probe molecule is constructed from a single DNA strand designed to adopt a partial cruciform structure with a pair of exposed (unhybridized) strands. Upon target binding, a switchable cruciform construct (similar to a Holliday junction) is formed which can adopt open and closed junction conformations. Switching between these forms occurs by junction folding in the presence of divalent ions. It has been shown from the steady-state fluorescence of judiciously labeled constructs that there are differences between the fluorescence resonance energy transfer (FRET) efficiencies of closed forms, dependent on the target sequence near the branch point, where the arms of the cruciform cross. This difference in FRET efficiency is attributed to structural variations between these folded junctions with their different branch point sequences arising from the single base mutations. This provides a robust means for the discrimination of single nucleotide mismatches in a specific region of the target. In this paper, these structural differences are analyzed by fitting observed time-resolved donor fluorescence decay data to a Gaussian distribution of donor-acceptor separations. This shows the closest mean separation (approximately 40 Å) for the perfectly matched case, whereas larger separations (up to 50 Å) are found for the single point mutations. These differences therefore indicate a structural basis for the observed FRET differences in the closed configuration which underpins the operation of these devices as biosensors capable of resolving single base mutations.
1. Introduction Molecular switches represent a diverse class of molecules that can exist in two or more distinct states, transitions between which involve mechanical movement or chemical changes. Among numerous examples are electronic switches,1 shuttles,2,3 nanovalves,4 molecular elevators,5 ratchets,6 and turnstiles.7 These can be activated by a variety of stimuli including, electrochemical, optical, thermal, or chemical. Despite this diversity, none of these nanoswitches have been explored with the view to performing sequence-specific nucleic acid recognition. Recently, it has been shown that changes in the switching characteristics of a device when bound to a particular target molecule can be used to differentiate between targets. This was * Corresponding author. E-mail:
[email protected]. † School of Physics, University of Edinburgh. ‡ Division of Pathway Medicine, University of Edinburgh. § School of Chemistry, University of Edinburgh. | School of Engineering and Electronics, Unversity of Edinburgh. ⊥ IBM T. J. Watson Research Center. # Collaborative Optical Spectroscopy Micromanipulation and Imaging Center, University of Edinburgh.
realized in a simple DNA-based device where the target molecule is also a nucleic acid and where the probe/target complex forms a switchable four-way junction; it was found that the switch characteristics sensitively depended on junction sequence such that it was possible to detect a single nucleotide polymorphism in the target sequence.8 The operational principle involves using a single DNA strand to form a probe molecule shown in Figure 1. Complementary sequences on this probe allow the formation of a partial cruciform structure, with two remaining unhybridized strands. These can then bind to a separate target molecule to form a complete cruciform switch complex as shown, whose state can be switched from open to closed by the addition of divalent cations. It has been shown that single nucleotide mutations in the target base sequence cause structural perturbations to the switch states of this complex which are sufficient to enable detection of these mismatches through steady-state fluorescence emission.8 The combined probe/target complex is similar to the DNA Holliday junction (HJ). This structure is a junction of four double helices first proposed in 1964.9 The distinctive topological element of the junction is the branch point discontinuity formed
10.1021/jp073817o CCC: $40.75 © 2008 American Chemical Society Published on Web 02/05/2008
2440 J. Phys. Chem. B, Vol. 112, No. 8, 2008
Mountford et al. This paper presents the fundamental characterization of the changes in the closed switch state structure which underpins this molecular recognition DNA nanoswitch. Information is obtained about the underlying structure in the closed form by using time-resolved fluorescence spectroscopy to probe the donor lifetime and utilizing a Gaussian analysis previously applied to similar DNA structures17-19 to determine the separation of the donor-acceptor fluorophore pair. Under the Gaussian approximation this is a two parameter fit, resulting in values for the mean separation distance and the width of the distance distribution. This analysis is used to investigate the effect of point mutations in the target on the closed conformer, giving insight into the origins of the different fluorescent emission characteristics of the closed switches. This enables an assessment of the extent to which molecular devices of this type may eventually be useful in resolving single nucleotide polymorphisms (SNPs) in biological assays. 2. Experimental Methods
Figure 1. (a) Schematic representation of the DNA nanoswitch. The probe (red) binds to the target (blue) by complementary Watson-Crick pairing. The recognition is achieved by a conformational change, which changes the FRET efficiency between the two fluorophores, represented here by the colored spheres. This allows mismatches in the target molecule to be detected. (b) A schematic of the DNA sequence layout in a complete DNA nanoswitch. The labeling system for target bases is indicated by the blue numbering.
at the intersection of the component strands. The overall structure of the junction is determined primarily by the strong electrostatic repulsion between the DNA backbone phosphate groups. Solvated under low cation concentration conditions, this repulsive Coulombic interaction favors an extended structure determined by maximum charge separation and an open branch point region, which forces the arms to be maximally extended.10 This open conformer has also been identified in certain crystal structures of the junction in complexes with recombination and repair proteins.11 Electrostatic screening of the interphosphate repulsion by the addition of (usually multivalent) cations above a critical concentration leads to instability of the extended structure which induces a conformational transition to a more compact, folded (closed) junction. Here, the branch point collapses to enable pairs of double helical arms to align coaxially and cross in the stacked-X conformation.10,12 Each stacked duplex has one nonexchanging strand running along the exterior of the junction and one sharply bent interior exchanging strand. Therefore, a four-way junction can form two alternative stackings depending on the stacking preference. The stacked-X junction undergoes millisecond isomerization transitions between these two forms as suggested by single molecule measurements.13,14 The stacked-X conformer has three further structural degrees of freedom corresponding to the interduplex angle (Jtwist) and the slide (Jslide) and rotation (Jroll) of the helices about their axes.15 The free energy surface of these junctions is clearly very complex and is only now beginning to be explored by atomistic computer simulation methods.16
2.1. Structural Information from Time-Resolved FRET: Models and Data Analysis. Fluorescence resonance energy transfer (FRET) is used to investigate the conformation of the probe molecule in ionic solutions, with target molecules that are either entirely complementary to the unhybridized strands or contain a single base point mutation near the branch point of the probe molecule, where the base sequence is expected to have the greatest effect on molecular conformation.18 FRET is a powerful tool, sensitive to distance changes on Ångstrom length scales, allowing the probe conformation to be investigated optically. The process involves nonradiative energy transfer via dipolar coupling between an excited-state donor fluorophore and an acceptor fluorophore separated by a distance typically between 10 and 100 Å.20 The fundamental relationship governing the process is the energy transfer efficiency E which depends on the donor-acceptor separation distance R according to
E)
1 1 + (R/R0)6
(1)
The distance R0 is defined as the distance at which the energy transfer process is 50% efficient. The FRET efficiency is experimentally accessible from steady-state spectra by measurement of the fluorescence intensity of the donor in the presence and in the absence of the acceptor (FDA and FD, respectively) according to
E)1-
FDA FD
(2)
Combining eqs 1 and 2 allows the donor-acceptor fluorophore separation distance to be calculated; however, this assumes that all fluorophores in the sample are at the same fixed separation. In reality, it is more likely that fluorophore separations will be better described by a distribution of distances, rendering this simple analysis inapplicable. Using time-resolved FRET analysis, however, allows information about this distribution to be obtained.19 Fluorescence decays of unquenched donor molecules should obey
ID(t) ) i(t) X K(t)
(3)
with
K(t) )
∑i Ri exp(-t/τi)
(4)
Molecular Recognition with DNA Nanoswitches
J. Phys. Chem. B, Vol. 112, No. 8, 2008 2441
Here, τi are individual donor lifetimes, each with their associated amplitude Ri. I(t) is the measured decay and i(t) is the instrument response function. In this system, we are observing one fluorophore exhibiting a complex decay, and therefore, the fluorophore has the same radiative decay rate in each environment. The Ri factors therefore represent the fraction of the molecules in each environment at t ) 0, corresponding to the ground state equilibrium.20 For donors attached to DNA, multiexponential lifetimes are typically observed.17,21 Each can be attributed to distinct donor environments, with the longest lifetime corresponding to a well solvated environment comparable to free donor and shorter lifetimes each corresponding to the donor in an environment close to the DNA backbone and/ or tether, when donor quenching to the molecules in this environment is significant. Such variation in configuration is expected when using flexible donor-DNA tethers.17,21 When the donor is quenched by FRET to the acceptor, the decay described by eqs 3 and 4 is modified, and the lifetime of the donor in each environment is reduced. When all distinct donor environments share a common donor-acceptor distribution function, P(R) (i.e., when changes in donor environment cause a negligible difference in donor-acceptor separation), then the fluorescence decay is described by:
I(t) ) i(t) X
∫RR ∑ Ri exp i max
min
[( ) { ( ) }] -t τi
1+
R0 R
6
P(R) dR (5)
In this expression, Ri and τi are the amplitudes and lifetimes obtained from a donor-only molecule and R0 is fixed by the choice of donor-aceptor pair. P(R) is to be determined and describes the distribution of fluorophore separation.17 This expression also simplifies the effect of the angular orientation of the donor and acceptor transition dipole moments to the FRET efficiency. Considerations of dipole orientations are explicitly contained in R0 according to
R0 ∝ 6
κ2φD 2
n
∫0∞ FD(λ)A(λ)λ4 dλ
(6)
where n is the refractive index of the transfer medium, φD is the quantum yield of the donor in the absence of an acceptor, FD(λ) is the donor emission fluorescence emission spectrum, and A is the extinction coefficient of the acceptor at wavelength λ. κ2 describes the relative orientations of the transition dipole moments; this averages to 2/3 in the approximation that they are free to rotate on a short time scale compared with the lifetime of the donor. In the probes used here, the fluorophores are attached using a six-carbon linker, making the angular insensitivity of FRET a reasonable approximation; this has been confirmed in similar systems.17-19 P(R), the distribution of fluorophore separation then contains information on the distribution of donor-acceptor separation (and hence molecular conformation). The analysis here assumes a Gaussian distance distribution
P(R) ) 4πR2c exp[-a(R - b)2]
0 < R e100 Å (7)
which has been shown to be useful in similar systems17 and oligopeptides22 and as such is used here as a useful distribution form. This allows donor fluorescence decays to be fitted to eq 5 by varying a and b to extract the mean value and width of P(R), respectively, under the assumption of a Gaussian distribution. The parameter c is a normalization factor, and so, there are only 2 fitting parameters, a and b.
2.2. Probe Design. The DNA nanoswitch of interest here uses the properties of a DNA probe molecule designed to recognize an unlabeled, complementary target molecule by binding and forming a synthetic HJ type structure, as depicted schematically in Figure 1. The probe comprises a DNA strand designed such that the primary base sequence leads to folding into a partially complete cruciform. There are two complete arms and two unhybridized single strands which are uncomplexed. The base sequence of these uncomplexed strands is designed to be complementary to a specific (matched) target sequence. The probe is labeled with two fluorophores; the location of these donor and acceptor fluorophores is also shown in Figure 1, chosen such that these molecules come into close proximity upon transition to the stacked-X conformer; this sequence has been chosen to maximize the proportion of the stacked X-conformer which gives rise to FRET compared with the alternative conformer.18 The resulting FRET increase on switch closure allows the conformational change to be monitored.8,21,23 2.3. DNA Switch Assembly. The DNA probe sequence (5-TGCATAGTGGATTGCATTTTTGCAATCCTGAGCACATTTTTGTGCTCACCGAATCCCA-3′) was synthesized by Eurogentec with tetramethylrhodamine (TAMRA) attached with a C6 linker at the 5′ end and carboxyfluorescein (FAM) attached with a C6 linker at a thymidine sited 19 nucleotides from the 5′ end. The value of R0 for these fluorophores free in solution is 50 Å.24,25 The probe was at 1 µM in the presence of the chosen 10 µM unlabeled target DNA oligonucleotide (Eurogentec), in 20 mM Tris/HCl (pH 7.5), 10 mM MgCl2. The sequences chosen were 5′-TGGGATTCGGACTATGCA-3′, the perfectly matched target, henceforth known as matched, and a series of targets which together show systematic target single nucleotide polymorphism at sites 10 and 11, from the 5′ end, in the branch point region: 5′-TGGGATTCGAACTATGCA-3′ (10A); 5′TGGGATTCGCACTATGCA-3′ (10C); 5′-TGGGATTCGTACTATGCA-3′ (10T); 5′-TGGGATTCGGCCTATGCA-3′ (11C); 5′-TGGGATTCGGTCTATGCA-3′ (11T); and 5′-TGGGATTCGGGCTATGCA-3′ (11G). The same nomenclature used in the literature is applied here,8 which is further illustrated in Figure 1b. The target was always at 10× excess to ensure full association of probe with target, thus negating the need for consideration of the uncomplexed probe in the analysis. This was confirmed by gel electrophoresis experiments, which showed the probe fully complexed with target and absent of free fluorophores in these experiments. Samples were annealed at 80 °C for 5 min followed by slow cooling to room temperature in 20 mM Tris/ HCl (pH 7.5) with or without 10 mM MgCl2. 10 mM has been shown in previous work8 to be a sufficient concentration of Mg2+ ions to cause adoption of the closed conformation. Donoronly lifetime emission in a comparable environment in the absence of FRET was measured using 1 µM of the probe without acceptor, synthesized by Eurogentec, in 20 mM Tris/ HCl (pH 7.5), 10 mM MgCl2. 2.4. Time-Resolved Fluorescence. Donor excitation was achieved using one of two sources: (1) a mode locked, frequency-doubled Coherent MIRA 900-F titanium sapphire laser, producing vertically polarized, 200 fs pulses of wavelength 450 nm, with a repetition rate externally reduced to 4.7 MHz pumped with a Coherent 10W Verdi or (2) laser pulses of wavelength 468 nm with a pulse at a rate of 10 MHz, from a picosecond diode laser head (Picoquant LDH-P-C-470). Emission was collected in an Edinburgh Instruments FL-920 spectrometer, with the output passed through a monochromator
2442 J. Phys. Chem. B, Vol. 112, No. 8, 2008
Mountford et al.
TABLE 1: Parameters from the Multiexponential Fit (eq 4) to the Fluorescence Decay for a Number of Molecules Studieda molecule
R1
τ1
R2
τ2
R3
τ3
χ2
donor only matched open matched closed 11C 11T 11G
0.80 0.32 0.24 0.20 0.28 0.41
4.5 3.2 3.3 3.4 3.3 3.2
0.17 0.53 0.40 0.45 0.43 0.37
2.3 1.8 1.2 1.3 1.4 1.5
0.03 0.16 0.36 0.35 0.30 0.22
0.5 0.4 0.3 0.3 0.3 0.3
1.068 1.029 1.045 1.069 1.068 1.019
a Donor only molecule is the probe molecule without an acceptor attached. In all cases, an excellent quality of fit is obtained, as demonstrated by the χ2 value. However, the changes in the Ri values are physically unrealistic without recourse to a FRET distance distribution model.
set to 517 nm, corresponding to the peak of the donor emission, and collected by a microchannel plate (Hamamatsu) using time correlated single photon counting (TCSPC). The emission polarizer was set at 54.7° to the vertical to avoid anisotropy effects. All data were collected to a peak of 10 000 counts with a delay window of 20 ns in 4096 channels. For each measurement, a corresponding instrument response was measured using a Ludox scattering sample. Fitting was achieved after deconvolution of the instrument response, i(t), from the lifetime measurement, I(t) to give K(t) in eq 3 by using a standard numerical re-convolution procedure (F900, Edinburgh Instruments), which has been used for this purpose in other work.26 Three exponents were used to model the decay and care was taken to ensure that fitting of more exponents did not improve the fit quality or change the recovered distribution form. The resulting K(t) decay was fitted to eq 5 using a custom-written nonlinear least-squares fitting routine (Microcal Origin). The quality of fit was assessed both visually, by plotting the residuals at all t , and by using chi-square statistics.
Figure 2. Inset shows the raw donor decay data (red) fitted by eq 4 (blue line), with associated residuals (shown beneath the inset). This was then fitted to eq 5, as shown in the main figure by the green line, to obtain fluorophore separation distributions.
3. Results
studies.19 These results are described in Table 2, with an example of the fit quality shown in Figure 2. Figure 3 shows P(R) distributions calculated from eqs 5 and 7 and the corresponding lifetime decays for each target. The results for the mean FRET separation distance and full width at half-maxima (FWHM) for each data set calculated from a and b are shown in Table 2. Equation 5 assumes that the sole contributions to changes in the observed donor lifetime are changes in the donor-acceptor separation distance due to differences in the folded structure of the molecule. However, it is known that a small population of alternative conformers of molecules of this type exist. The base pair sequence around the branch point is known to sensitively affect the preference.18 The J1 structure, with an identical branch point sequence to the molecule used here, has been shown to have a very small population of below 5% in alternative stacked conformers.18 It is therefore necessary to be careful in this work, with a lack of complementarity around the branch point, to check for significant populations of alternative conformers. Fits of eq 5 were of high quality in all cases studied, suggesting that the populations of alternative conformers is small. An example, demonstrating the fit quality, is shown in Figure 2. By using the analysis of Miick et al.,18 the population of alternative conformers is probed by adding a second Gaussian distribution to eq 5. This was also carried out for the samples analyzed here; however, again, the quality of fit was not improved and the fit appeared over parametrized. This also
The fit to the fluorescence decay for the donor-only molecule (eq 4) was multiexponential, with typical parameters given in Table 1. A good fit was found to require three components. This table shows that 80% of the fluorescent species contributing to the decay are the longest lifetime component, 1, which can be assigned to the lifetime of donor in a highly solvated configuration, comparable to free donor. However, there are also 17% and 3% of shorter lifetime emitting species (components 2 and 3); in similar molecules, these have been considered as being characteristic of the lifetimes of the donor in two alternative environments, where the donor can experience quenching due to DNA and linker interactions.17,21 By fitting eq 4 to the molecules studied, the remaining results in Table 1 are obtained, each with an excellent quality of fit from their χ2 values. This is not surprising, given that these are each six-parameter fits. However, although in all cases as expected, each component lifetime is shortened, there is significant variation in the proportions of the three lifetimes (from the Ri values) in the matched and unmatched targets. This seems physically unrealistic, as the free rotation of the dyes and relatively small changes brought about by the single base mutations would not be expected to lead to such drastic variation. In contrast, readily interpretable results and comparable fits are obtained by using a two-parameter fit to eq 5, which assumes that the donor decay is modified solely by changes in fluorophore separation, in accordance with previous
TABLE 2: Peaks and Full Widths at Half-Maxima (fwhm) of Probability Distributions Describing Fluorophore Separationa target
peak separation distance (Å)
FWHM (Å)
matched closed matched open 10T 10A 10C 11C 11T 11G
40.5 50.5 44.1 45.4 45.9 42.3 44.2 49.0
28.4 17.6 28.9 30.5 28.3 25.5 28.2 29.4
a Fits to eq 5 are more physically realistic (with no changes in Ri) despite the resulting fit quality being comparable to that obtained with eq 4.
Molecular Recognition with DNA Nanoswitches
J. Phys. Chem. B, Vol. 112, No. 8, 2008 2443
Figure 3. Fluorophore separation distributions for all possible targets with point mutations at the 10th (a) and 11th (b) target bases. The results indicate that the matched target forms a switch which adopts a closed conformation in which the fluorophores are in their closest possible separation, while point mutations in the target cause the fluorophores to be more widely separated. (c) Comparison of the open and closed matched molecule.
suggests that any population of unfolded molecules or alternative conformations, as have been observed in work with similar molecules,27 cannot be identified here, and will be included in the single enforced Gaussian distribution. To ensure that this was due to negligible alternative populations and not a limit on the analysis technique, a series of set concentrations of a molecule with known fluorescence decay were added. The proportion added was then recovered by adding a term to eq 5. It was shown that the spiked proportion could be recovered with high accuracy to below 20% of the total mixture, which sets a conservative upper limit on this analysis,
in addition to reassurance that the model and implementation are behaving physically. As a final control, a similar analysis was applied by adding a term to eq 5 which represented a donor only (i.e., nonFRETing) component. This again caused no improvement in the fit. With these results in mind, it was concluded that the molecules analyzed here are significantly biased into one closed conformer, with the populations of alternative conformers being very small and not contributing greatly to the data. This is further supported by unpublished steady-state measurements.
2444 J. Phys. Chem. B, Vol. 112, No. 8, 2008 The stability of the fit was tested by setting extreme initial conditions for each sample. Despite the high correlation between the parameters a and b, the fit proved to be remarkably stable, with identical parameters and satisfactory convergence obtained for all but the most extreme initial conditions. As such, eq 5 was fitted to the donor fluorescence decay for each target studied. Figure 3 shows the optimal P(R) distributions obtained for each of the four possible mutations at the 10th and 11th target base. The peaks and widths of these distributions are given in Table 2, the results which compare favorably with those obtained for similar molecules.17 These results are completely consistent with those presented using steady-state measurements.8 The target which produced the highest FRET efficiency, namely, the perfectly matched target, also shows a peak at the shortest separation distance here. Conversely, 11G shows the largest peak separation, again as expected. This trend can be extended throughout all targets studied here, with the closer separations resulting in a higher FRET efficiency throughout. The widths of the distributions show little variation. Repeated measurements, using different sample preparation cycles and both excitation sources, showed excellent repeatability with a maximum overall spread in peak position of 0.4 Å and fwhm of 5.5 Å. It is clear that single point mutations cause changes to the free energy surface for the overall molecule sufficient to cause a perturbation in the peak separation of up to 10 Å. This result is consistent with previous results suggesting noncomplementary base sequences disrupt the folding of DNA junctions.27 It is interesting that the matched, open conformation, shown in Figure 3c, has a larger fluorophore separation distance than all others, which is to be expected. In addition, the distribution is narrower, suggesting the positions of the junction arms are more tightly constrained here. This indicates that the open conformation displays FRET consistent with the expected donor-acceptor distance, which shows this analysis can be used for both states of the switch. This work provides new information on the operation of DNA nanoswitch biosensors. Despite the structural changes induced being very small, they are detectable using both time-resolved and steady-state spectroscopy. 4. Conclusion In this paper, the properties of a simple DNA-based nanodevice capable of recognizing unlabeled target nucleic acid sequences in solution and discriminating single base mutations have been explored. This functionality is enabled by a novel recognition mechanism that combines complementary-base hybridization of a target strand with sequence dependent switching. This sequence dependence is manifested by subtle changes in the structure of mismatched probe/target complexes which are structurally similar to the four-way cruciform junction formed by the perfectly matched probe/target complex. Previously published steady-state results are built upon in this paper by using time-resolved FRET and a Gaussian distribution of donor-acceptor separation distances to provide quantitative estimates of the structural perturbations caused by point mutations of the target, in particular, the relative contributions of changes in the mean and distribution of donor-acceptor separation, which give insight into changes in both molecular structure and rigidity. There are likely to be many other molecular architectures, in addition to cruciform junctions, where this recognition mode may be realized and this approach to FRET analysis will give valuable information on molecular structure. The general applicability of the Gaussian distribution,
Mountford et al. previously applied to DNA junctions, to a variety of molecular architectures will be the subject of future work. Acknowledgment. This work was carried out as part of a DTI funded Beacon project. The authors thank A. M. Haughey for help with supporting spectroscopy work. References and Notes (1) Emberly, E. G.; Kirczenow, G. Phys. ReV. Lett. 2003, 91 (18), 1883011-1883014. (2) Collin, J. P.; Dietrich-Buchecker, C.; Gavina, P.; Jimenez-Molero, M. C.; Sauvage, J. P. Acc. Chem. Res. 2001, 34 (6), 477-487. (3) Murakami, H.; Kawabuchi, A.; Matsumoto, R.; Ido, T.; Nakashima, N. J. Am. Chem. Soc. 2005, 127 (45), 15891-15899. (4) Nguyen, T. D.; Tseng, H.-R.; Celestre, P. C.; Flood, A. H.; Liu, Y.; Stoddart, J. F.; Zink, J. I. Proc. Nat. Acad. Sci. 2005, 102 (29), 1002910034. (5) Badjic, J. D.; Balzani, V.; Credi, A.; Silvi, S.; Stoddart, J. F. Science 2004, 303 (5665), 1845-1849. (6) Kelly, T. R.; De Silva, H.; Silva, R. A. Nature 1999, 401 (6749), 150-152. (7) Carella, A.; Jaud, J.; Rapenne, G.; Launay, J.-P. Chem. Commun. 2003, 19, 2434-2435. (8) Buck, A. H.; Campbell, C. J.; Dickinson, P.; Mountford, C. P.; Stoquert, H. C.; Terry, J. G.; Evans, S. A. G.; Keane, L.; Su, T. J.; Mount, A. R.; Walton, A. J.; Beattie, J. S.; Crain, J.; Ghazal, P. Anal. Chem. 2007, 79 (12), 4724-4728. (9) Holliday, R. Genet. Res. 1964, 5 (2), 282-304. (10) Lilley, D. M. J. Quart. ReV. Biophys. 2000, 33 (2), 109-159. (11) Hargreaves, D.; Rice, D. W.; Sedelnikova, S. E.; Artymiuk, P. J.; Lloyd, R. G.; Rafferty, J. B. Nat. Struct. Mol. Biol. 1998, 5 (6), 441-446. (12) Ortiz-Lombardia, M.; Gonzalez, A.; Eritja, R.; Aymami, J.; Azorin, F.; Coll, M. Nat. Struct. Mol. Biol. 1999, 6 (10), 913-917. (13) McKinney, S. A.; Declais, A.-C.; Lilley, D. M. J.; Ha, T. Nat. Struct. Mol. Biol. 2003, 10 (2), 93-97. (14) Buranachai, C.; McKinney, S. A.; Ha, T. Nano Lett. 2006, 6 (3), 496-500. (15) Hays, F. A.; Watson, J.; Ho, P. S. J. Biol. Chem. 2003, 278 (50), 49663-49666. (16) Yu, J.; Ha, T.; Schulten, K. Nucleic Acids Res. 2004, 32 (22), 66836695. (17) Eis, P. S.; Millar, D. P. Biochemistry 1993, 32 (50), 13852-13860. (18) Miick, S. M.; Fee, R. S.; Millar, D. P.; Chazin, W. J. Proc. Nat. Acad. Sci. 1997, 94 (17), 9080-9084. (19) Klostermeier, D.; Millar, D. P. Biopolymers 2002, 61 (3), 159179. (20) Lakowicz, J. R. The Principles of Fluorescence Spectroscopy, 2nd ed.; Klewer Academic/Plenum Publishers: New York, 1999; pp 13-14, 129-130. (21) Mountford, C. P.; Mount, A. R.; Evans, S. A. G.; Su, T. J.; Dickinson, P.; Buck, A. H.; Campbell, C. J.; Terry, J. G.; Beattie, J. S.; Walton, A. J.; Ghazal, P.; Crain, J. J. Fluorescence 2006, 16 (6), 839845. (22) Haas, E.; Wilchek, M.; Ephraim, K. Z.; Steinberg, I. Z. Proc. Nat. Acad. Sci. 1975, 72 (5), 1807-1811. (23) 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 (3), 214-221. (24) Lorenz, S. D. Electrophoresis 2001, 22 (6), 990-998. (25) Yuan, C.; Rhoades, E.; Lou, X. W.; Archer, L. A. Nucleic Acids Res. 2006, 34 (16), 4554-4560. (26) Neely, R. K.; Magennis, S. W.; Dryden, D. T. F.; Jones, A. C. J. Phys. Chem. B 2004, 108 (45), 17606-17610 (27) Duckett, D. R.; Lilley, D. M. J. J. Mol. Biol. 1991, 221, 147-161.
