Noncovalent binding and fluorogenic response of cyanine dyes to DNA homoquadruplex and PNA-DNA heteroquadruplex structures

Abstract

Two symmetrical cyanine dyes based on benzothiazole heterocycles and a trimethine bridge were found to bind to a parallel-stranded DNA guanine quadruplex based on the MYC oncogene promoter sequence with high nanomolar affinity and 1:1 stoichiometry. The dyes exhibited substantial fluorescence enhancements upon binding. In the presence of homologous guanine-rich peptide nucleic acid oligomers, PNA-DNA heteroquadruplexes were formed. The dyes retained their ability to bind to the heteroquadruplexes at low micromolar concentrations and with varying fluorescence enhancements, although indeterminate stoichiometries preclude quantitative comparison of the affinities with the DNA homoquadruplex precursor. The difference in fluorescence enhancement between DNA homoquadruplex and PNA-DNA heteroquadruplex allows the dyes to be used as fluorogenic indicators of hybridization in a facile method for determining PNA-DNA stoichiometry.

Introduction

1The growing recognition of the important roles of guanine quadruplex structures in regulating gene expression and telomere biology have motivated considerable efforts to design or discover synthetic molecules that can bind to quadruplexes with high affinity and selectivity.^1–3 Quadruplex-binding ligands might find applications in chemical biology, where they could report the presence or perturb the biological function of a quadruplex.^4,5 Clinical applications of drug-like small molecules targeted to G-quadruplexes are also envisioned.⁶

Small molecules bind to G-quadruplexes primarily through shape-based recognition. An alternative strategy relies on sequence-based recognition, where complementary C-rich oligonucleotides bind via Watson-Crick base pairing to form heteroduplexes^7–10 or homologous G-rich oligonucleotides bind via Hoogsteen-based G-tetrad formation to yield heteroquadruplexes.^11–15 Peptide nucleic acid (PNA) probes can bind to G-quadruplexes by either of these mechanisms, leading to stable hybrid structures.

Our recent focus has been on PNA-DNA/RNA heteroquadruplex formation, based on the high affinity (low nanomolar K_ds) of the PNA for its homologous DNA^11,13,14 and RNA^12,16 targets, as well as excellent sequence discrimination for homologous versus complementary targets that can be achieved through PNA backbone modification.¹⁴ Short PNAs consisting of two G₂ or G₃ tracts and various nucleobase or abasic loops successfully invade G-quadruplex targets under physiological conditions to form stable heteroquadruplex structures. The heteroquadruplex stoichiometries are typically 2 PNA: 1 DNA or RNA. Hence, the short PNAs disrupt the folded homoquadruplex then hybridize such that each of two PNA strands recognizes two of the four G-tracts that make up the quadruplex target.

The current report describes our first steps toward combining sequence- and shape-based recognition of quadruplexes through the simultaneous application of PNAs and small molecules. In our prior work, we used a covalently conjugated fluorogenic cyanine dye to provide a fluorescence response to PNA binding, as in the “light-up” probes originally reported by Ishiguro and coworkers for DNA¹⁷ and subsequently by Svanvik and coworkers for PNA.^18,19 Recently, Ladame and coworkers¹⁵ combined a PNA G₃ tract that could form one edge of a heteroquadruplex with an acridone stacking ligand previously shown to end-stack on G-tetrads.²⁰

Although the examples cited above clearly illustrate that a covalently conjugated dye can interact with a PNA-DNA heteroquadruplex, it was unclear to what extent the dye would interact with the heteroquadruplex if not for the covalent linkage to the PNA terminus. While still composed of G-tetrads found in DNA homoquadruplexes, PNA-DNA heteroquadruplexes offer unique structural characteristics that could be useful for small molecule recognition. For example, the negative charge density in a heteroquadruplex is less than half that of a homoquadruplex, due to the uncharged nature of the backbone and the presence of cationic termini on the PNA. Thus, electrostatic interactions offered by the two types of quadruplex will be quite different. Similarly, the van der Waals surfaces of the quadruplex grooves should vary considerably between homo- and heteroquadruplexes.

Cyanine dyes have found numerous applications in nucleic acid recognition,^21,22 with various reports on intercalation, groove binding and end-stacking on duplexes,²³ triplexes^24,25 and quadruplexes.^25–30 The wide range of absorption/emission wavelengths accessible to the cyanines, the ease with which the net charge on the molecule can be varied through substituents, and the versatility of the dyes, in terms of either having environmentally sensitive or insensitive quantum yields, have been exploited in these diverse studies. In this report, we describe two carbocyanine dyes that bind noncovalently to PNA-DNA heteroquadruplexes. These dyes can exhibit variable fluorescence enhancements upon binding, depending on the loop composition of the PNA strand. These results open the door to using cyanine dyes in combination with G-rich PNAs to form and detect heteroquadruplex structures.

Results and Discussion

Cyanine Dye-DNA quadruplex interactions.

The carbocyanine dyes DiSC₁(3) and β-Me-DiSC₁(3) (Scheme 1; these dyes are also known as Cyan 46 and Cyan 2, respectively³¹) have been shown to exhibit substantial increases in fluorescence in the presence of quadruplex DNA.^25,30 The quadruplexes studied in the prior reports were modeled on telomeric DNA sequences and other G-rich motifs. For our experiments, we focused on a DNA quadruplex formed by a 19-base sequence found in the promoter region of the MYC oncogene³² (Myc19, Scheme 1). This sequence is known to fold into a stable intramolecular G quadruplex with a parallel arrangement of the four strands.³³

Scheme 1.

Cyanine dye structures and DNA target and PNA probe sequences. Differences among PNA probes are underlined; “eg” = 8-amino-3,6-dioxa-octanoic acid.

The interaction of DiSC₁(3) with Myc19 is illustrated in Figure 1, where UV-vis (A) and fluorescence (B) spectra of the dye undergo significant perturbations in the presence of the quadruplex. As the DNA is titrated into the dye, the absorption spectrum exhibits batho- and hypochromism, while the fluorescence spectrum also exhibits a bathochromic shift and enhanced fluorescence. These observations are consistent with a binding interaction between the dye and the DNA quadruplex.

Figure 1.

Effect of Myc19 quadruplex DNA on UV-vis (A) and fluorescence (B) spectra of DiSC₁(3). Samples contained 1.0 µM DiSC₁(3) and 0, 0.25, 0.5, 1.0 and 2.0 µM DNA. Fluorescence spectra were excited at λ = 520 nm.

Data from a full fluorescence titration experiment were fit adequately by a simple 1:1 binding model, yielding K_D = 0.77 ± 0.06 µM (Fig. 2A). The stoichiometry was further supported by a continuous variations experiment (Fig. 2B). Thus, the carbocyanine dye DiSC₁(3) binds with high nanomolar affinity to the Myc19 quadruplex to form a stable 1:1 complex. Circular dichroism experiments indicated that the parallel morphology of the quadruplex (positive peak at 260 nm) is maintained in the presence of the dye (data not shown).

Figure 2.

Fluorescence titration (A) and continuous variations (B) experiments to determine K_D and stoichiometry of DiSC₁(3) binding to Myc19 quadruplex. For (A), DNA was titrated into a solution of 1.0 µM dye. For (B), a total concentration of dye + DNA = 1.0 µM was used.

The binding site for the cyanine dye on the DNA quadruplex is unknown. The parallel structure of the Myc19 quadruplex features three double-chain reversal loops, in which single- or double-nucleotide loops fold back across the groove to allow adjacent G-tracts to align in parallel orientations. This leaves one groove unblocked and potentially accessible to the cyanine dye. However, we cannot rule out the possibility that the dye stacks preferentially on one end of the quadruplex.

Similar experiments were undertaken with the bridge-substituted cyanine dye β-Me-DiSC₁(3). Absorption and fluorescence spectra recorded as a function of Myc19 concentration are shown in Figure 3. The shape of the absorption spectrum changes with DNA concentration, although the peak does not shift (Fig. 3A). Meanwhile, the fluorescence of the dye in solution is much lower than the non-bridge substituted analogue DiSC₁(3). Bridge-substitution of cyanine dyes is known to suppress fluorescence, possibly due to inducing a conformational change to a structure³⁴ with a lower quantum yield. The fluorescence of β-Me-DiSC₁(3) is strongly enhanced in the presence of the Myc19 quadruplex (Fig. 3B) similar to prior work by Yarmoluk and coworkers with duplex and quadruplex DNAs.^25,31,35 The spectra also exhibit minor variation in shape with DNA concentration, perhaps reflecting the presence of multiple binding modes. The shoulder observed at 640 nm in these spectra is also unexpected and is not observed if the KCl concentration is reduced from 100 mM to 1 mM (Sup. Fig. 1). This suggests that the shoulder is due to aggregation of the dye, since dimerization/aggregation of cyanines is known to be favored at elevated ionic strength.³⁶

Figure 3.

Effect of Myc19 quadruplex DNA on UV-vis (A) and fluorescence (B) spectra of β-Me-DiSC₁(3). Samples contained 1.0 µM β-Me-DiSC₁(3) and 0, 0.25, 0.5, 1.0 and 2.0 µM DNA. Fluorescence spectra were excited at λ = 520 nm.

A continuous variations experiment indicated that a 1:1 dye:DNA complex is formed (Fig. 4B), although the turnover point is not as well resolved as for DiSC₁(3). Fitting the titration data to a 1:1 model as shown in Figure 4A indicates that this dye binds slightly stronger to the quadruplex than does the unsubstituted analogue DiSC₁(3). Although both DiSC₁(3) and β-Me-DiSC₁(3) have similar quantum yields when fully bound to the quadruplex, the lower quantum yield for unbound β-Me-DiSC₁(3) leads to a much stronger enhancement.

Figure 4.

Fluorescence titration (A) and continuous variations (B) experiments to determine K_D and stoichiometry of β-Me-DiSC₁(3) binding to Myc19 quadruplex. For (A), DNA was titrated into a solution of 1.0 µM dye. For (B), a total concentration of dye + DNA = 1.0 µM was used.

Cyanine Dye-PNA/DNA heteroquadruplex interactions.

We previously demonstrated the ability of G-rich peptide nucleic acid (PNA) probes to form heteroquadruplexes with homologous DNA targets such as Myc19. For example, the PNA probe P_Myc forms a 2:1 heteroquadruplex with Myc19 (K_D = 5 nM).¹³ Similar to prior reports in which fluorogenic cyanine dyes were used as reporters for heteroduplex formation, the fluorogenic dye thiazole orange was a useful reporter molecule for PNA-DNA heteroquadruplex formation. (This dye has been used as the basis for a fluorescent displacement assay to screen quadruplex-binding ligands).²⁷ When conjugated to the PNA N-terminus (Scheme 1), the dye exhibits a substantial fluorescence enhancement upon heteroquadruplex formation, allowing determination of the stoichiometry and affinity. Thus, we were interested to determine whether the fluorogenic trimethine dyes DiSC₁(3) and β-Me-DiSC₁(3) would bind noncovalently to PNA-DNA heteroquadruplexes.

UV-vis and fluorescence titrations of the dyes with various PNA-DNA heteroquadruplexes did not produce well-defined stoichiometries, precluding determination of binding affinities and molecularities (data not shown). Nevertheless, UV-vis spectral perturbations and fluorescence enhancements were evident in many cases. Figure 5 shows the fluorescence intensities of the dyes alone, in the presence of the Myc19 DNA homoquadruplex and with a variety of PNA-DNA heteroquadruplexes at constant dye, DNA and PNA concentrations. Modest fluorescence enhancements, relative to unbound dye, are observed for DiSC₁(3) and the PNA-DNA heteroquadruplexes, with slightly larger enhancements observed for those PNAs having bases in their loops (_PA, P_C, P_T and P_TG) compared to the PNAs having abasic loops (P_eg1 and P_eg2). As a result of its lower background fluorescence, β-Me-DiSC₁(3) exhibits considerably larger fluorescence enhancements for all PNA-DNA heteroquadruplexes. In this case, the strongest fluorescence for the heteroquadruplexes is observed for those PNAs having the capacity to form longer loops when bound to DNA (P_eg2 and P_TG).

Figure 5.

Fluorescence of cyanine dyes DiSC₁(3) (left) and β-Me-DiSC₁(3) (right) in presence of various DNA and PNA-DNA quadruplexes. Samples contained 1.0 µM dye, 1.0 µM DNA and 2.0 µM PNA.

It is difficult to interpret the results of Figure 5 due to uncertainty in the affinities and binding modes for the two dyes. For example, differences in fluorescence intensity could arise from differences in affinity and/or quantum yield for the bound dye. Nevertheless, the fact that the intensity is substantially higher for the DNA homoquadruplex relative to the PNA-DNA heteroquadruplexes allows a useful experiment to be performed to determine the stoichiometry of the PNA-DNA quadruplexes. PNA and DNA are mixed in various ratios but constant total concentration. A constant amount of cyanine dye is also present in solution. The fluorescence of the dye is determined for each PNA:DNA ratio and is plotted versus mole fraction of PNA. As shown in Figure 6, the results of this continuous variations experiment reveal a 2:1 PNA:DNA stoichiometry for P_eg2-Myc19, consistent with our prior investigations of this heteroquadruplex.

Figure 6.

Job plot constructed by mixing together P_eg2 and Myc19 in varying ratios and 1.0 µM total concentration in the presence of 1.0 µM DiSC₁(3). Fluorescence was recorded with excitation at 520 nm.

We previously used two methods to determine PNA-DNA stoichiometry. In the first, a fluorogenic TO dye was conjugated to the end of the PNA and the difference in fluorescence intensity in the free versus hybridized states was used to construct a Job plot that revealed 2:1 binding. Alternatively, surface plasmon resonance experiments in which the total amount of immobilized DNA and captured PNA are known can provide the stoichiometry directly.¹³ Use of a noncovalent reporter dye such as the cyanines described herein provides an appealing alternative, since this bypasses synthesis of a separate PNA-dye conjugate for each system or the need for a specialized and costly SPR instrument, although, of course other methods such as electrophoretic mobility shift assay could provide the same information.

Materials and Methods

PNAs were synthesized using t-Boc-protected monomers (ASM chemicals) and solid phase methods, as reported previously in reference ³⁸ and ³⁹. Crude oligomers were purified by reverse-phase HPLC and characterized by MALDI-TOF mass spectrometry. PNAs P_myc, P_eg1 and P_eg2 were reported previously. (Expected m/z for P_A, 2168.1, found, 2171.2; expected m/z for P_C, 2144.1, found, 2147.2; expected m/z for P_T, 2159.1, found, 2157.9; and the expected m/z for P_TG, 2450.3, found, 2454.1). DNA oligonucleotides were purchased from Integrated DNA Technologies (www.idtdna.com) and used as received. Cyanine dyes DiSC₁(3) and β-Me-DiSC₁(3) are known compounds and were synthesized by Dr. Gloria Silva. Products were characterized using ESI mass spectrometry (Thermo-Fisher LCQ ESI/APCI Ion Trap). (β-Me-DiSC₁(3), expected m/z for M⁺, 351.5; found, 351.3. DiSC₁(3), expected m/z for M⁺, 337.4; found, 337.3). UV-vis spectra for the compounds matched those from the literature.³¹

All DNA and PNA concentrations were determined by measuring the absorbance at 260 nm at 85°C on a Cary 300 Bio UV-visible spectrophotometer. At high temperatures, the bases are assumed to be unstacked and the extinction coefficient of the oligomers is estimated as the sum of the extinction coefficients of the individual bases at that temperature. For the DNA oligomers, the extinction coefficients were used as reported in the literature.⁴⁰ The individual nucleobase extinction coefficients used to calculate concentrations were 13,700 M⁻¹cm⁻¹ (A), 11,700 M⁻¹cm⁻¹ (G), 6,600 M⁻¹cm⁻¹ (C) and 8,600 M⁻¹cm⁻¹ (T).

Absorption and emission spectroscopy.

Absorption spectra of pre-annealed (i.e., heated to 95°C for 5 min before slow cooling to room temperature) samples were collected using Varian Cary 300 Bio UV-Visible Spectrophotometer. Corresponding baselines corrections were made prior to absorbance measurement. Emission (fluorescence) spectra of pre-annealed samples were collected using a Varian Cary Eclipse Fluorescence Spectrophotometer. Samples containing DNA, PNA and dye were prepared in a buffer solution containing 10 mM Tris-HCl (pH 7), 0.1 mM Na₂EDTA and 100 mM KCl. (This buffer was used for all spectroscopic experiments).

Two types of continuous variations experiments were performed. For DNA-dye stoichiometries, DNA and dye were mixed at varying ratios but constant total concentration of 1.0 µM. Fluorescence spectra were recorded with excitation at 520 nm and the integrated fluorescence intensity was plotted versus dye mole fraction to determine the empirical stoichiometry of the DNA-dye complex. For PNA-DNA stoichiometries, PNA and DNA were mixed at varying ratios but constant total concentration of 1.0 µM. Dye was also present at 1.0 µM concentration. Fluorescence spectra were recorded with excitation at 520 nm and the integrated fluorescence intensity was plotted versus dye mole fraction to determine the empirical stoichiometry of the PNA-DNA complex.

Circular dichroism (CD) spectropolarimetry.

CD measurements were performed using a Jasco 715 CD spectropolarimeter, equipped with a water-circulating temperature controller. Samples were annealed prior to CD recording. Each CD spectrum represents an average of six scans, collected at a rate of 100 nm/min.

Conclusion

The cyanine dyes DiSC₁(3) and β-Me-DiSC₁(3) bind to the DNA quadruplex Myc19 in 1:1 stoichiometries and with high nanomolar affinities. While these dyes are unlikely to exhibit selective binding to a particular DNA quadruplex relative to other quadruplexes, double-stranded DNA or RNA, the cyanine scaffold is readily modified with various substituents on the heterocycle periphery, the (poly)methine bridge or the heterocycle nitrogens. Thus, libraries of cyanine dyes could be screened in order to identify substitution patterns that favor selective binding, as has been done with other scaffolds.^27,30,37 The low background fluorescence of β-Me-DiSC₁(3) makes it particularly attractive for further study since dyes based on this structure might exhibit selective staining of quadruplex structures in cells without having to wash out unbound dye.

In addition to the dye-DNA results presented above, this report provides the first examples of noncovalent binding of small molecules to PNA-DNA heteroquadruplexes. The variation in fluorescence intensities for the different heteroquadruplexes evident in Figure 5 indicates that the dyes sense subtle structural variations and this impacts the affinities and/or quantum yields of the bound dyes. These dyes are useful in the shortterm for determining PNA-DNA stoichiometries; in the longer view, we hope to identify dyes that will bind selectively to PNA-containing heteroquadruplexes over DNA or RNA homoquadruplexes, duplexes or other structures. Such dyes would provide sensitive indicators of successful PNA hybridization in complex biological samples, including within cells. Alternatively, covalent conjugation of the dyes to PNA could provide alternatives to the TO conjugates used previously, with longer wavelength emission and potentially larger fluorescence enhancements, particularly in the case of β-Me-DiSC₁(3).