Large Davydov Splitting and Strong Fluorescence Suppression: An Investigation of Exciton Delocalization in DNA-Templated Holliday Junction Dye Aggregates

Abstract

Exciton delocalization in dye aggregate systems is a phenomenon that is revealed by spectral features, such as Davydov splitting, J- and H-aggregate behavior, and fluorescence suppression. Using DNA as an architectural template to assembledye aggregates enables specific control of the aggregate size and dye type, proximal and precise positioning of the dyes within the aggregates, and a method for constructing large, modular two and three-dimensional arrays. Here, we report on dye aggregates,organized via an immobile Holliday junction DNA template, that exhibit large Davydov splitting of the absorbance spectrum (125nm, 397.5 meV), J- and H-aggregate behavior, and near-complete suppression of the fluorescence emission (~97.6% suppression).

Because of the unique optical properties of the aggregates, we have demonstrated that our dye aggregate system is a viable candidate as a sensitive absorbance and fluorescence optical reporter. DNA-templated aggregates exhibiting exciton delocalization may find application in optical detection and imaging, light-harvesting, photovoltaics, optical information processing, and quantum computing.

Graphical abstract

INTRODUCTION

Exciton delocalization, first discussed theoretically by Frenkel in the 1930s, and later by Davydov and Kasha, is an optical phenomenon that has been observed in optically active molecular (i.e., dye) aggregates.^1–6 Exciton (i.e., electron− hole pair) delocalization is the spatial extension of the excited electronic eigenstates across the aggregate, with the resultant excitonic state described by a coherent linear superposition of excited states.^7,8 Among the observable optical manifestations of excitonic delocalization are Davydov splitting,^2,3 J- and H-aggregate behavior,^4,5,7,9,10 motional narrowing,^11–13 Dicke superradiance, resonance fluorescence,¹⁴ fluorescence quenching,^15–19 and excitonically coupled circular dichroism (ECCD).^20,21 The optical behavior of such aggregates is largely dictated by the stacking arrangement of the closely spaced dye molecules (2 nm or less apart). A single dye, termed a monomer, has a single one-exciton energy state, which is the lowest lying excited singlet state. A transition between this excited state and the ground state is allowed (Figure 1a). A two-dye aggregate system, termed a dimer, has two one-exciton states in which the degeneracy of the two excited states has been broken by the coupling between the transition dipoles, thus resulting in two energy eigenstates: E⁺ and E⁻. The exciton is delocalized between the two dyes in these eigenstates. Transitions strengths between these states and the ground state, E₀, are strongly dependent on geometry. In the simplest geometric configurations, the two dyes can stack such that the transition dipoles are oriented parallel (H-dimer, Figure 1b), head-to-tail (J-dimer, Figure 1c), or at an intermediate configuration (oblique, Figure 1d). For these three possible cases, the allowed transitions are indicated by solid black arrows, while the forbidden transitions are indicated by dashed arrows in Figure 1. H-type aggregates exhibit a blue-shifted absorbance peak relative to the monomer, significantly reduced fluorescence emission intensity due to an optically forbidden energy transition from the lower energy exciton excited state (E⁻), and a large Stokes shift.^{9,15–17,22–30} Conversely, J-type aggregates, discovered by Jelley^31,32 and Scheibe,^33–35 exhibit a red-shift in the absorbance relative to the monomer since the only allowed optical transition is to the E⁻ state, which consequently produce a nearly resonant fluorescence emission(i.e., small Stokes shift) with a sharp, high-intensity emission peak.^{7,9,12,14,27,36–42} Oblique aggregates display band (i.e., Davydov) splitting of the absorbance spectrum in which optical transitions to both E⁻ and E⁺ are allowed and reflect a mix of H- and J-aggregate optical properties.^2,3,27,43,44

Figure 1.

Schematics and energy level diagrams for various dye aggregate arrangements: (a) monomer, (b) H-dimer, (c) J-dimer, and(d) oblique dimer. The top schematic shows dye molecules represented as purple spheres, while the dipole orientation is depicted by black arrows. Each energy level diagram shows a ground energy state (E₀), the first excited energy state (E) for the monomer, and the excited energy state’s split energy levels (higher energy, E⁺ and lower energy, E⁻) for dimers due to exciton delocalization. Allowed energy transitions are provided as solid black arrows and forbidden transitions are shown as dashed black arrows. Adapted from ref 4 with permission.

The manifestation of excitonic delocalization has been observed in both naturally occurring photosynthetic systems^45–47 and in synthetic dye-solvent systems.^{29,39,47–51} Photosynthetic systems have evolved such that the molecular chromophores (i.e., dyes) are proximally positioned via a protein scaffold. However, using a protein scaffolding architecture in synthetic systems is difficult owing to complex folding mechanisms. The aggregation of dyes in solvents has been found to occur spontaneously and preferred dye stacking arrangements are largely driven by kinetics and thermodynamics, which is problematic for controlling the excitonic delocalization behavior. To assert greater control over the assembly of dyes into aggregates of well-defined structure, we exploit DNA as an alternative biomolecular scaffolding material. DNA assembly is governed by simple and predictable design rules due largely to Watson−Crick hybridization. In addition, DNA can be functionalized with molecular dyes through covalent attachment at precisely specified locations anywhere along its length. Thus, covalent attachment of dyes onto a DNA substrate enables (i) proximal and precise positioning of the dye molecules,^27,52–58 (ii) specific and selective control of dye number and type,^24,52,54,59 and (iii) the self-assembly of large two- and three-dimensional arrays.^60–62

Capitalizing on the modularity enabled by DNA self-assembly allows larger, more complex arrays and networks of precisely controlled dye aggregates (i.e., dye assemblies) to be realized. Though there have been many studies that demonstrate exciton delocalization in DNA-templated dye aggregation within linear duplex structures,^{17,18,22–25,27,52–59,63–66} only a few studies have exploited DNA constructs of more complex geometry. Most of these have been limited to studies employing three-armed junctions for dye assembly, which have been constructed to explore light-harvesting complexes and excimer behavior.^52,67 In our recent work, we were able to facilitate and control the transition of Cy5 dimers templated within a linear DNA duplex structure to a Cy5 tetramer in a mobile 4-arm junction (i.e., 4AJ) structure, thereby inducing a change in exciton delocalization from J-type to H-type behavior.⁶⁸ The transition was controlled by varying salt and/or DNA concentrations. Here, we exploit an immobile 4AJ (i.e., Holliday junction)^69,70 to explore exciton delocalization in multimers, including dimers, trimers, and tetramers in order to better control the effects of exciton delocalization such as Davydov splitting and J- and H-type behavior at ambient temperatures. The excitonic delocalization between the four Cy5 dyes (i.e., tetramer) located within the core of a 4AJ exhibits an unusually large Davydov splitting (125 nm, 397.5 meV) that results in a striking color change of the solution, excitonically coupled circular dichroism (EC-CD) signals, and near-complete quenching (i.e., strong suppression) of the fluorescence emission. To obtain further insight into the 4AJ-guided assembly of excitonically coupled aggregates, a variety of aggregate types were examined. It was found that increasing the number of dyes within the aggregate core of the 4AJ substantially increased the energy splitting of the system. Because of its unique optical signatures, this immobile tetramer offers two characterization modes for optical detection that have applications as optical reporters, namely a large shift in absorbance and strong suppression of the fluorescence. The construction, by DNA self-assembly, of dye aggregates exhibiting the strong optical effects reported here may pave the way for novel practical applications of exciton delocalization.

METHODS

Synthesis.

Short 26 nucleotide (nt) DNA oligomers were purchased lyophilized from Integrated DNA Technologies, Inc. Oligomers internally functionalized with Cy5 dyes were purified via high-performance liquid chromatography (HPLC), while non dye-functionalized strands were purified through standard desalting. All oligomers were rehydrated with ultrapure water (Barnstead Nanopure, Thermo Scientific) to produce a 100 μM stock solution. DNA four-armed junctions (4AJs) were prepared by combining equimolar amounts of complementary oligomers in a 1× TAE (40 mM tris-(hydroxymethyl)aminomethane, 20 mM acetic acid, 1 mM ethylenediaminetatraacetic acid) buffer solution (pH 8.0) with 15 mM added magnesium chloride (MgCl₂) to obtain a final DNA concentration of 10 μM. Resulting solutions were left for 24−48 h at room temperature to allow for hybridization and self-assembly. TAE (10× stock solutions) and MgCl₂ (99% purity) were purchased from Fisher Scientific and used as received. Solutions of 1.375 M MgCl₂ were added to a 1× TAE buffer solution to yield final magnesium (Mg²⁺) concentrations of 0, 15, and 100 mM. The initial salt concentration due to the TAE buffer (<1 mM Na⁺ arising from the EDTA counterion) was found to be negligible compared to the subsequently added salt concentrations.

Purification.

Self-assembled DNA 4AJ structures were further purified using polyacrylamide gel-electrophoresis (PAGE). Solutions of 4AJs were prepared with various dye arrangements in a 1× TAE buffer at 10 μM double-stranded DNA (dsDNA) concentrations and 15 mM added MgCl₂. A 6× New England Biolabs loading buffer was added in a 1:5 ratio with the 4AJs to give a resulting 1× concentration of the loading buffer (1× buffer: 11 mM ethylenediaminetetraacetic acid, 3.3 mM tris-hydrochloric acid, 0.017% sodium dodecyl sulfate, 0.015% bromophenol blue, 2.5% Ficoll-400). The prepared sample solutions were then injected into a 12% gel, prepared with 0.25× TAE with 3.75 mM added MgCl₂ and allowed to migrate for approximately 90 min with a 150 V applied voltage. 1× TAE with 12 mM added MCl₂ was used in the buffer reservoirs of the electrophoresis apparatus. Using a multiplexed gel imaging and documentation system (FluorChem Q, ProteinSimple), completed gels were imaged by brief illumination with a 254 nm ultraviolet (UV) light source while placed on a phosphor imaging plate. Well-formed 4AJ structures were identified, excised from the gels with a razor blade, and extracted. Extraction was conducted by placing select excised gel pieces into a vial of 1× TAE buffer containing 15 mM added MgCl₂. The DNA 4AJs were allowed 48 h to diffuse out of the gel and into the surrounding buffer solution. The resulting buffer solution was then extracted from the vial and any remaining gel fragments removed by centrifuging the sample at 1000 relative centrifugal force (rcf) for 10 min. A further description of the gel-electrophoresis process and gel images are provided in the Supporting Information, part S5.

Optical Characterization.

All optical characterization was performed at room temperature. UV−Vis spectra were obtained using a dual-beam Cary 5000 UV−Vis−NIR spectrophotometer (Agilent Technologies). Solution samples (150 μL) were transferred into a 1 cm path length submicro, self-masking quartz spectrophotometer cell (Starna). Static absorbance spectra were monitored over a range of wavelengths and normalized by DNA concentrations. Static fluorescence spectra were obtained using a Fluorolog-3 spectrofluorometer (Horiba Scientific) and monitored as a function of wavelength. PAGE purified solution samples of each 4AJ construct (150 μL) were transferred to a 1 cm path length submicro, self-masking fluorometer cell (Starna) and were excited at or near their respective absorption maxima listed in Figure 2. The excitation and emission bandwidths used were 5.0 and 2.5 nm, respectively. Circular dichroism (CD) measurements were performed using a JACSO-J810 spectropolarimeter. Samples of purified 4AJ constructs (100 μL) were transferred to a 1 cm path length micro, self-masking quartz cuvette (Thorlabs). Spectra were monitored over a series of wavelengths and normalized by the DNA concentration.

Figure 2.

Absorbance (black traces in left column), fluorescence (blue traces), and circular dichroism (black traces in right column) spectra corresponding to the 4AJ templated (a, b) monomer, (c, d) adjacent dimer, (e, f) transverse dimer, (g, h) trimer, and (i, j) tetramer dye aggregates. The inset in part i shows the suppressed fluorescence emission peak at 664 nm that occurs due to a forbidden energy transfer mechanism. All spectra have been normalized by DNA concentration. The fluorescence studies were performed by monitoring the emission over a range of wavelengths and exciting the aggregates at their respective absorbance maxima. All samples were prepared at 10 μM DNA concentration in a 1× TAE buffer solution with 15 mM MgCl₂ added, PAGE purified, and normalized by the resulting DNA concentrations (~0.5−2.5 μM).

Optical Reporter Studies.

Optical reporter studies were performed using two characterization techniques: fluorescence and absorbance. For fluorescence studies, DNA-templated 4AJs were designed with a 10 nt toehold on one of the four arms. Structures were synthesized and purified as described above. Spectra were collected using a Horiba Fluorolog-3 spectrofluorometer. Static fluorescence spectra of the samples were obtained prior to and subsequently following the reaction. Samples (250 μL) were monitored at the fluorescence peak maximum (666 nm) over 70 min while being excited at the tetramer absorbance peak maximum (565 nm). At 10 min, an invasion strand-designed to be complementary to the strand with the 10 nt toehold extension-was introduced into the system in 1:1 stoichiometry with the 4AJ and pipette mixed for ~30 s. The change in fluorescence intensity was then monitored over 70 min. The initial dsDNA concentrations of the 4AJ were ~0.4 μM, and all reactions were held at 22 °C.

Absorbance studies were conducted by initially synthesizing the 4AJ structure in two parts, described as dimers A and B, such that the two dimers are complementary and could join together to form the full 4AJ structure. Dimers A and B were purified via PAGE and diluted to ~0.1 μM dsDNA concentration. Using a Cary 5000 UV−Vis−NIR spectrophotometer, changes in absorbance at the H-tetramer peak maximum (565 nm) were monitored. Initially, a sample of dimer A was monitored. At 10 min, dimer B was introduced in stoichiometry and pipette mixed for ~30 s. Upon hybridization, the 4AJ tetramer formed, inducing an absorbance signal at 565 nm. All reactions were held at room temperature (22 °C).

Stability Studies.

Results from studies performed to monitor the stability of the 4AJs with changes in solution salt and DNA concentrations are described in Supporting Information, part S6. Samples of tetramer 4AJ constructs used for salt studies were prepared in a 1× TAE buffer solution containing 0−100 mM MgCl2 and held at a constant DNA concentration of 1.0 μM. Samples of the tetramer used for DNA concentration studies were prepared in a 1× TAE buffer solution with 0.1−3.0 μM dsDNA concentrations and held at a constant salt concentration of 15 mM MgCl2

RESULTS AND DISCUSSION

Optical Characterization of 4AJ Templated Dye Aggregates.

Optical characterization using spectroscopy techniques, including absorbance, fluorescence, and circular dichroism (CD) were performed on a variety of dye-labeled DNA 4AJ structures to identify spectral signatures of exciton delocalization. Although the majority of the data reported here were obtained from samples in a 1× TAE buffer solution containing 15 mM of MgCl₂, we additionally performed absorbance measurements of samples prepared in a 1× TAE buffer solution containing a variety of salt concentrations: 0 and 100 mM MgCl₂ with a constant 1 μM DNA concentration and a variety of DNA concentrations: 0.1, 1.0, and 3.0 μM with a constant 15 mM salt concentration (Supporting Information, part S6) to investigate the salt and DNA concentration dependence that might indicate higher order aggregation formation. The absorbance spectra exhibited only a weak dependence on MgCl₂ and DNA concentration, indicating that the results obtained at 15 mM MgCl₂ concentration are representative of a broad range of salt concentrations and that our 4AJ structures have little tendency to form higher-ordered aggregates. The minimal dependence on MgCl₂ and DNA concentration to drive changes in dye aggregate formation is in contrast to what was found in our previous work with linear DNA duplex constructs that combined to form mobile 4AJ structures.⁶⁸ For spectroscopic studies, all 4AJ structures were purified via polyacrylamide gel-electrophoresis (PAGE). Dye structure and strand information are provided in the Supporting Information, part S1. Fluorescence suppression is computed using the area of the emission spectrum; a general equation is given in the Supporting Information, part S2. As a control, spectral analysis of the monomer (Figure 2a) reveals single absorbance and emission peak maxima at 653 and 666 nm, respectively. Although the monomer absorbance peak maximum is slightly red-shifted from the literature value, the shift is most likely a result of the increased rigidity of the 4AJ structure, dye−DNA interactions, and surrounding base sequences.⁷¹ As expected, the dye monomer did not produce a CD signal within the visible range (Figure 2b), indicating the absence of molecular chirality and excitonic coupling. Likewise, the CD signals appearing in the UV range at 246 and 276 nm, respectively, result from the expected right-handed macro-molecular structure of the β-DNA helix.

As shown in the schematics in Figure 2, the 4AJ structure was used to organize and create four different dye aggregates in addition to the monomer: two dimer configurations, a trimer, and a tetramer. The two-dye (dimer) aggregates are distinguished by the location of the dyes, either positioned on adjacent arms (adjacent dimer, Figure 2c) or on opposing arms (transverse dimer, Figure 2e). A pronounced difference in optical properties was observed between the two dimers. The adjacent dimer shows predominantly J-type aggregate behavior, with an absorbance peak maximum at 662 nm that is slightly red-shifted relative to the monomer at 653 nm. This red-shift in absorbance relative to the fluorescence peak at 667 nm yields a small Stokes shift of 5 nm, which is near resonance fluorescence behavior. A second, smaller absorbance peak also appears at 602 nm. The appearance of the two peaks is indicative of a small Davydov splitting of 58 nm (Table 1). The adjacent dimer exhibits a split CD signal (Figure 2c) at 600 and 675 nm. The ± of the CD peaks (read from right to left), signifies right-handed chirality and is anticipated owing to the right-handed nature of the DNA helix. Because of the predominant J-type behavior and presence of Davydov splitting and CD signal, the dyes are most likely arranged head-to-tail with some break in planarity between the molecules, or oblique-like arrangement, that favors J-aggregate stacking. In contrast, the transverse dimer (Figure 2e) shows optical behavior that is roughly opposite that of the adjacent dimer. The primary absorbance peak is blue-shifted, with an absorbance maximum at 600 nm and a much smaller red-shifted absorbance peak (maximum at 636 nm) corresponding to an even smaller 36 nm Davydov splitting. A 68 nm Stokes shift is observed when comparing the primary absorbance peak to the fluorescence peak at 668 nm, which is much reduced in intensity (~97.6%). The blue-shifted primary absorbance peak, the large Stokes shift, and the reduced fluorescence intensity are all indicative of predominately H-type aggregate behavior. The small Davydov splitting and a relatively strong ∓ CD signal at 600 and 669 nm, respectively, of the transverse dimer indicates left-handed chirality and an oblique stacking arrangement that does not stack in a perfectly parallel or H-type arrangement (Figure 2f). Interestingly, the small absorbance peak at 636 nm corresponds to a split CD signal with peaks at 630 and 669 nm, which implies excitonic coupling between the dyes. The considerably decreased fluorescence emission is indicative of fluorescence suppression due to a forbidden optical transition.

Table 1.

| Optical Excitonic Delocalization Behavior per 4AJ Construct

optical behavior	monomer	adjacent dimer	transverse dimer	trimer	tetramer
Davydov splitting (nm/meV)	0/0	58/179.8	36/117.0	58/186.3	125/397.5
abs fwhm (nm)	34	30	24	49	16
FL fwhm (nm)	33	32	34	34	48
Stokes shift (nm)	13	5	68	73	99
% FL suppression	0.0	53.3	89.6	84.5	97.6

The trimer (Figure 2g) shows absorbance and fluorescence properties similar to the transverse dimer. Most notably, the trimer absorbance spectrum shows an absorbance peak at 593 nm that is blue-shifted from the monomer by 60 nm and a smaller peak at 651 nm that is slightly blue-shifted by 2 nm. The absorbance spectrum also reveals broadening of the primary peak by 15 nm compared to the monomer, as indicated by an increase in the full-width at half-maximum (fwhm) value listed in Table 1. A detailed procedure for calculating the fwhm is described in the Supporting Information, section S2. The broadening may result from an irresolvable third peak; however, because the CD spectrum (Figure 2h) does not exhibit a third peak, the broadening most likely arises from a distribution of trimer dye configurations (i.e., dye positions and orientations). In combination with the blue-shifted absorbance peak, the trimer shows ~85% suppression of the fluorescence emission intensity (relative to the monomer), indicative of net H-aggregate behavior. Similar to the other aggregate configurations, the trimer has an observed EC-CD signal that, though small, indicates a slightly imperfect stacking arrangement with minor obliqueness. Additionally, the aggregate displays right-handedness, similar to the adjacent dimer.

The tetramer dye aggregate configuration produces the most interesting optical spectra. A large Davydov splitting of 125 nm (397.5 meV), extensive enough to induce a visible color change in the solution, was observed. See the Supporting Information, Figure S2.1, for spectral data and peak fitting analysis. To our knowledge, this is the largest reported splitting for DNA-templated dye aggregates (Figure 2i). The Davydov splitting is characterized by a significantly blue-shifted intense absorbance peak at 565 nm and a red-shifted much less intense peak at 690 nm. Though difficult to resolve in the absorption spectrum, the absorbance peak at 690 nm is further substantiated by a large signal in the CD spectrum (Figure 2j) at 700 nm. Note that the difference in extinction coefficients of the various dye aggregates result from two key effects: (1) the differences in dye number and (2) exciton delocalization and exciton-vibrational interactions. The tetramer also exhibits strong fluorescence suppression behavior at 664 nm, with a 97.6% decrease in the fluorescent emission relative to the monomer, as determined by peak area. The large Davydov splitting, strong fluorescence suppression, and 99 nm Stokes shift (Table 1) provide solid evidence of a dye assembly with predominantly H-type stacking. The pronounced ∓ CD signal of the tetramer configuration indicates strong exciton coupling between the dyes and reveals that the dyes are oriented predominantly in a parallel manner with some obliqueness that is supported by the absorbance peak at 690 nm. Comparing the optical spectra of the immobile 4AJ-templated tetramer presented here with the mobile 4AJ-templated tetramer observed in our prior study, the most notable difference is in the CD spectrum. Interestingly, the immobile tetramer shows right-handedness (Figure 2J), while the mobile tetramer shows left-handedness.⁶⁸ Additionally, an absorbance peak at 665 nm was observed for the mobile 4AJ-templated tetramer. In contrast, for the immobile tetramer, a very subtle peak was observed for at 690 nm that was further supported by a strong CD signal, indicating obliqueness within the immobile tetramer. These differences are most likely due to the base-pair stacking of the immobile tetramer locking the aggregate core such that the DNA junction does not undergo restacking. The mobile tetramer undergoes DNA breathing and base-pair restacking and was found to partition into two J-dimer pairs that are displaced either horizontally or vertically along the arms, which accounts for the observed 665 nm absorbance peak.⁶⁸

Theoretical Spectral Modeling of 4AJ Templated Dye Aggregates.

Theoretical modeling of the absorbance and CD spectra was performed to determine the geometrical stacking configurations of the five dye aggregate configurations since specific dye positioning and orientation beyond the above simple analysis is not readily apparent from visual inspection of experimental spectral data. Following the approach used in our previous work,⁶⁸ modeling of the experimental absorbance and CD spectra using a method based on the work of Kühn, Renger, and May, referred to as the Kühn−Renger−May (KRM) model, enables the extraction of the position and orientation of the dyes within each type of aggregate.⁷² The KRM model proceeds by constructing a system Hamiltonian that considers N dye molecules with arbitrary positions and orientations. Through judicious choice of a truncated Hilbert space, a perturbational treatment of the exciton-vibrational coupling is avoided in the diagonalization of the Hamiltonian. Information regarding the computation of the eigenvalues and eigenvectors is described in the Supporting Information, parts S3. An in-house software simulation tool that incorporates the KRM model was created to fit the absorbance and CD spectra. Smooth theoretical absorbance and CD spectra were obtained by calculating the optical transition rates for each eigenstate and convolving the transition rate line spectra with a Gaussian line shape to approximate the effect of the vibronic continuum on spectral line shape.

In Figure 3, theoretical KRM model fits of the absorbance and CD spectra are illustrated by red dashed curves, while the experimental data are shown as black open circles. To maximally constrain the theoretical modeling, the absorbance and CD data were simultaneously fit such that the two dominant peaks of the absorption and CD spectra were both matched and the mean-square difference between the experimental and theoretical absorption and CD spectra were minimized. From this fitting procedure, the relative positions and orientations of each dye are extracted (see the Supporting Information, Table S3.2).

Figure 3.

Absorbance (a, d, g, and j) and CD (b, e, h, and k) spectral fits with corresponding structures (c, f, i, and l) of the adjacent dimer, transverse dimer, trimer, and tetramer aggregates, respectively. Experimental data sets are shown as black open circles, while KRM theoretical fits are given as red curves. All samples were prepared at 10 μM DNA concentration in a 1× TAE buffer solution with 15 mM MgCl₂ added, PAGE purified, and normalized by the resulting DNA concentrations (~0.5−2.5 μM). Fluorescence measurements were obtained by exciting the samples at their respective absorption maxima.

Fitting the experimental absorbance (Figure 3a) and CD (Figure 3b) spectra for the adjacent dimer revealed that the dyes are oriented with predominantly J-aggregate (or head-to-tail) stacking but in a somewhat oblique manner, which is manifested in the increased vibronic peak at 600 nm. This configuration supports the evidence provided by the red-shifted absorbance spectrum and demonstration of excitonic coupling in the CD spectrum, which only occur with an oblique stacking arrangement. Figure 3c visualizes the dimer stacking configuration within the 4AJ construct based on the fitting of the theory to the absorbance and CD data (see the Supporting Information, Table S3.2). The figure was created using UCSF Chimera visualization software.^73,74 We note that only the position and orientation of the long axis of the Cy5 molecule is determined by the fitting procedure. Thus, how the molecules are oriented about the axis (i.e., rotated about the axis) and relative to the DNA is conjectural in Figure 3, parts c, f, i, and l. In contrast to the adjacent dimer, theoretical fitting of the transverse dimer absorbance (Figure 3d) and CD (Figure 3e) spectra using the KRM model determined that the dyes are indeed stacked parallel to one another (i.e., H-type) with a center-to-center distance of ~1.0 nm. The slight obliqueness of the dyes is exhibited in the EC-CD spectrum and a vibronic peak at 636 nm in the absorbance spectrum. Figure 3f illustrates the transverse dimer stacking configuration obtained from the theoretical fit of the absorbance and CD data.

Using the KRM model to fit the trimer absorbance (Figure 3g) and CD (Figure 3h) spectra substantiates our hypothesis that there are not three underlying absorbance peaks but rather that the peak at 593 nm is simply broadened. Though the theoretical fit does not provide an explanation for the peak broadening, we can conjecture the existence of a population of trimer constructs having slightly different dye configurations(i.e., dye positions and orientations within the core of the 4AJ), such that the spectrum, consisting of an average of these populations, exhibits a broadened absorption peak. The fit of the theory to the experimental absorbance and CD data indicates that the dyes of the trimer are stacked roughly parallel to each but with some twist, as depicted in Figure 3i. The configuration of the trimer dyes is predominantly H-type similar to the transverse dimer, which explains why their respective absorbance spectra appear so similar. Finally, fitting of the tetramer aggregate’s absorbance (Figure 3j) and CD (Figure 3k) spectra incorporates the red-shifted peak at 690 nm. Additionally, due to the strong EC-CD signals, obliqueness in the dye stacking arrangement was confirmed. The dye configuration obtained by the fitting procedure is illustrated in Figure 3l. Input parameters, output positions and orientations for each dye, and a quantitative characterization of the quality of the theoretical fits are provided in the Supporting Information, part S3.

Optical Reporter Functionality.

To examine the viability of the 4AJ tetramer as a highly sensitive optical reporter, we constructed and studied the reaction kinetics of two reporter systems that exploit changes in fluorescence and optical absorption, respectively. In the system that exploited fluorescence changes, strand invasion was employed to disassemble a tetramer into a one dye and three dye complex. Strand invasion was facilitated by a designed 10 nt extension (toehold) on one of the 4 arms to which the Invasion Strand can first attach by Watson−Crick base pairing. The Invasion Strand is fully complementary to the 4AJ strand with the extension and thus excises this strand from the 4AJ.⁷⁵ As illustrated in Figure 4a, this 4AJ tetramer (tetramer-1), would have nearly zero fluorescence while assembled. Upon invasion, the coherent excitonic delocalization within the aggregate would cease and greatly increased fluorescence would ensue from the dyes on both the single dye (monomer) product and the three dye complex (trimer). Conversely, the system that exploited absorbance changes consisted of two halves of the 4AJ synthesized separately that can hybridize when brought together to form a full 4AJ tetramer (tetramer-2). While disassembled, the two halves of the 4AJ (dimer A and dimer B) would exhibit minimal absorption at 565 nm (the H-tetramer absorbance peak), yielding a cyan solution color. When assembled, the absorption of the complete 4AJ (tetramer-2) at 565 nm greatly increases, yielding a solution color change to violet. This color change is depicted schematically in Figure 4 by the color of the represented dye molecules (cyan and violet spheres).

Figure 4.

(a) Schematic illustration and (b) experimental demonstration of the 4AJ tetramer undergoing strand-invasion using fluorescence detection (λ_fl = 666 nm). (c) Schematic illustration and (d) experimental demonstration of the 4AJ tetramer formation via DNA hybridization using UV−Vis−NIR absorbance, or colorimetric, detection (λ_abs = 666 nm).

The operation of the fluorescence scheme (Figure 4b) yielded a 25-fold increase in fluorescence intensity upon disassembly of the 4AJ into a monomer and trimer. Similarly, the absorbance reaction (Figure 4d) gave a 10-fold increase in absorbance at 565 nm. For the DNA concentrations at which the experiments were carried out, ~ 0.4 μM for the Fluorescence scheme and ~0.1 μM for the absorbance scheme, the time to half completion of the reactions was short compared to the 30 s required to thoroughly pipette mix the reactants in the cuvette. This implies a lower bound of 8 × 10⁵ M⁻¹ s⁻¹ for the rate constant of the strand invasion reaction of Figure 1a and a lower bound of 3 × 10⁵ M⁻¹ s⁻¹ for the rate constant of the hybridization reaction of Figure 1c. These are consistent with the rate constants reported in the literature for these types of reactions,^75,76 which are typically in the 10⁶ M⁻¹ s⁻¹ range. The observed near complete (>97.6%) fluorescence suppression and large Davydov splitting (125 nm, 397.5 meV) in the 4AJ tetramer (Figure 2) signals its potential usefulness as a highly sensitive optical reporter.

CONCLUSION

We have explored a two-dimensional DNA template for proximally positioning molecular dyes to induce exciton delocalization. The four-armed junction (4AJ) DNA construct enabled the analysis of a variety of dye aggregate types (i.e., adjacent and transverse dimers, trimer, and tetramer) that revealed various exciton delocalization spectral signatures. Through static spectral characterization of the 4AJ-templated dye aggregates, it was found that the type of aggregate (i.e., H-aggregate, J-aggregate, or oblique) depended largely on the position of the dyes within the 4AJ construct. The specificity of the dye position was demonstrated explicitly by the difference in spectral properties of the two dimer constructs, such that the dimer with dyes positioned on adjacent arms (adjacent dimer) demonstrated predominantly J-aggregate behavior, whereas the transverse dimer with dyes positioned on opposite arms exhibited H-aggregate behavior. The tetramer, with four dyes positioned at each node of the 4AJ, expressed a strikingly large Davydov splitting and strong fluorescence suppression. Exploiting the distinctive optical properties of the tetramer afforded the opportunity to fabricate and operate two dynamic dye aggregate schemes for potential use as a sensitive optical reporter using either absorbance or fluorescence characterization methods. Further exploration of exciton delocalization in DNA-templated dye assemblies may yield applications in optical detection and imaging schemes, light-harvesting, photovoltaics, optical information processing, and quantum computing.