Interactions of ruthenium(II) polypyridyl complexes with human telomeric DNA

Abstract

Eight [Ru(bpy)₂L]²⁺ complexes (where bpy = 2,2’-bipyridine, an ancillary ligand and L = a polypyridyl experimental ligand) were investigated for their G-quadruplex binding abilities. Fluorescence resonance energy transfer melting assays were used to screen these complexes for their ability to selectively stabilize human telomeric DNA variant, Tel22. The best G-quadruplex stabilizers were further characterized for their binding properties (binding constant and stoichiometry) using UV-vis, fluorescence spectroscopy, and mass spectrometry. The ligands’ ability to alter the structure of Tel22 was determined via circular dichroism and PAGE studies. We identified me₂allox as the experimental ligand capable of conferring excellent stabilizing ability and good selectivity to polypyridyl Ru(II) complexes. Replacing bpy by phen (1,10-phenanthroline) did not significantly impact interactions with Tel22, suggesting that binding involves mostly the experimental ligand. However, using a particular ancillary ligand can help fine-tune G-quadruplex-binding properties of Ru(II) complexes. Finally, the fluorescence “light switch” behavior of all Ru(II) complexes in the presence of Tel22 G-quadruplex was explored. All Ru(II) complexes displayed “light switch” properties, especially [Ru(bpy)₂(diamino)]²⁺, [Ru(bpy)₂(dppz )]²⁺, and [Ru(bpy)₂(aap)]²⁺. Current work sheds light on how Ru(II) polypyridyl complexes interact with human telomeric DNA with possible application in cancer therapy or G-quadruplex sensing.

Graphical Abstract

Eleven Ru(II) polypyridyl complexes that differ by their experimental ligands were studied for their interaction with human telomeric DNA, which forms G-quadruplex structure. The complexes displayed strong “light switch” effects, excellent G-quadruplex stabilizing abilities, good selectivities, and modest binding affinities.

1. Prologue

This article is dedicated to life and achievements of F. Ann Walker. LAY completed her PhD work in Ann’s lab at the University of Arizona from 1999–2003. I consider myself lucky to have chosen her group, as I found in Ann supportive, generous, and kind research advisor and an outstanding scientist. One of the very impressive things about Ann was her deep knowledge on the wide number of topics including experimental design and execution, although at the time, she has rarely, if ever, ran experiments. I remember collecting NMR data and observing an unexpected line broadening in spite of re-purifying the sample or varying all possible parameters. She looked at my data and asked if I verified that the solvent was free of contaminants. Although I was using a previously unopen solvent bottle, upon checking it, I discovered that it contained decomposition products, which caused problems in my work for weeks. I was blown away by Ann’s ability to quickly see the root of the problem. At the time, I was the only graduate student as our lab was populated by long-term postdoctoral fellows and technicians. Ann believed in “her people”. She worked tirelessly to assure that each of us is fully funded and when the grant money ran out, she financed the lab from her personal accounts. Ann was extremely kind. I remember, one day she was washing about 100 NMR tubes. I asked her how did she end up with so many of them? She said that she wanted her Inorganic course students to have more free time to prepare for their exam, so she was washing their NMR tubes. It must have taken her hours to complete the cleaning. I also want to comment on is Ann’s lifelong mission to get equal respect for women and their work. She hated when people called her Ma’am or madam and she was quick to point that she is not a madam, she is a Professor. I deeply respected her for that.

Ann supported me long after my Arizona years. I am here today at Swarthmore because of her influence on my education and scientific growth. She was my mentor, my friend, and my role model. She was also a human. She had a matching set of earrings for every outfit that she owned, she loved her dogs, she was a great fighter who never gave up, she believed that every human deserves respect and love. She is an example by which I try to live today.

2. Introduction

Guanine-rich DNA sequences fold into non-canonical DNA structures known as G-quadruplexes (GQs).¹ GQs are formed by the π-π stacking of planar G-tetrads, where each tetrad contains four guanines connected by eight Hoogsteen hydrogen bonds (Fig. 1). GQ structures are further stabilized by monovalent cations (e.g. K⁺ or Na⁺).² Recent studies have identified GQs as important players in a variety of biological processes,³ as well as potential cancer targets.⁴ To design effective anticancer therapies, many small-molecule ligands that interact selectively with GQs have been identified and thoroughly characterized.^5–7 Several ligands were successfully tested in cancer cell lines and a few have advanced to clinical trials.⁸ GQ ligands tend to share two common properties, a positive charge and planarity, allowing for electrostatic and π-π interactions with GQs,⁹ although neutral and moderately non-planar ligands were also shown to bind well GQ DNA.^10,11 To be therapeutically useful, GQ ligands must be highly selective for GQs over double-stranded (ds) DNA. The difference in size between a G-tetrad and a base pair allows for differential selectivity.

Fig. 1.

(A) Structure of a G-tetrad with a monovalent cation in the middle. (B) General structure of polypyridyl Ru(II) complexes under investigation. The complexes are synthesized as hexafluorophosphate salts. (C) The identity, structure, and abbreviations for the experimental ligands, L. In three cases, ligand names are followed by two numbers, corresponding to the [Ru(bpy)₂(L)]²⁺ and [Ru(phen)₂(L)]²⁺ complexes, respectively.

In this work we focus on ruthenium (Ru(II)) polypyridyl complexes which, besides their ability to bind and stabilize GQ DNA, contain a metal center, display rich photophysical properties, and can be prepared by a modular synthesis allowing for a facile fine-tuning of their structure and properties. They hold promise as GQ fluorescence sensors and anticancer therapeutic agents.^7,12–23 Typically Ru(II) complexes contain a hexacoordinated Ru(II) center bound to three bidentate N-donor ligands. Often, one ligand is responsible for photophysical and electrochemical properties of the complex (so called experimental ligand) and two ligands play structural role but otherwise are considered ‘spectators’ (so called ancillary ligands). Common ancillary ligands are phen (1,10-phenanthroline), bpy (2,2’-bipyridine), and their derivatives. Ru(II) complexes with a wide range of experimental ligands have been prepared and investigated for their ability to interact with GQ DNA.^{9,15,20,21,24,25} The most well characterized Ru(II) complexes are [Ru(bpy)₂(dppz)]²⁺ and [Ru(phen)₂(dppz)]²⁺ (dppz = dipyrido-[3,2-a:2’,3’-c]phenazine; also see List of abbreviations) which bind human telomeric DNA and display impressive fluorescence in its presence, yet, suffer from low selectivity for GQs over dsDNA.^18,26,27 A related complex where dppz was extended by imidazolone induced and stabilized an antiparallel GQ structure in the absence of metal cations and displayed significant emission intensity enhancement.¹⁹ Selectivity of Ru(II) complexes was improved by paying close attention to the design of the experimental ligand. Mononuclear and dinuclear Ru(II) complexes containing substituted pyridyl bidentate ligands displayed excellent selectivity for GQ vs. dsDNA, induced the formation and increased stability of human telomeric GQ in the absence and presence of 100 mM of K⁺, and inhibited telomerase.^15,28 [Ru(IP)₂(PIP)](ClO₄)₂·2H₂O complex exhibited anticancer properties comparable to cisplatin by inhibiting telomerase activity.⁹ [Ru(phen)₂CPIP]²⁺ and [Ru(TAP)₂CPIP]²⁺ demonstrated photo-cytotoxicity with 100% mortality rates against cancer cells which do not express telomerase suggesting an anticancer mechanism which does not depend on telomerase inhibition.²⁰

Ru(II) complexes are well known for their “light switch” behavior originally detected in the presence of dsDNA²⁹ but also demonstrated in the presence of GQ DNA.^24,30,31 The “light switch” behavior is governed by a metal-to-ligand charge transfer (MLCT) of the excited 4d electron on Ru(II) to a low lying π* molecular orbital of conjugated aromatic ligand, e.g., dppz.³² The “light switch” effect originating from MLCT has several advantages such as low energy excitation in the visible range, high quantum yield and photostability, long lifetime, etc. Several studies have used modified dppz ligands to produce a colorimetric “light switch” effect detectable without a fluorimeter.^19,31 The search continues for highly GQ-selective ligands, which display “light switch” properties that would allow for the visualization of GQ DNA in vivo and thus advance our understanding of GQ localization and biological functions.

The direct information on the binding modes of Ru(II) complexes to GQ DNA is scarce and often indirect methods are used. Typically it is accepted that Ru(II) binding occurs via π-stacking of an experimental ligand with a terminal G-tetrad. Molecular docking studies^20,28 as well as a growing number of GQ-Ru(II) structures (three X-ray structures from the Cardin lab, one as recent as 2022, and one NMR structure from the Thomas lab)^33–36 support this binding mode but also suggest stacking with loops and overhangs. For example, Λ-[Ru(TAP)₂(11-CN-dppz)]²⁺ π-stacks onto a terminal G-tetrad with 11-CN-dppz ligand contacting three of the four guanine bases, while the Δ isomer interacts only with the terminal T-T base pairs contributing to crystal packing.³⁴ Λ-[Ru(TAP)₂(dppz)]²⁺, the photooxidizing analogue of well-studied [Ru(phen)₂(dppz)]²⁺, displays three binding modes, none of which involves a G-tetrad, rather non-canonical T-T and T-A base pairs formed by overhangs.³⁵ NMR structures from the Thomas lab using binuclear [(Ru(phen)₂)₂tpphz]⁴⁺ complex indicate that both ΛΛ and ΔΔ isomers end-stack with G-tetrads but also interact extensively with diagonal or lateral loops, respectively, of antiparallel basket human telomeric GQ.³⁶ Structural studies therefore, confirmed that end-stacking of the experimental ligand is a likely binding mode but other binding modes need to be considered and stoichiometries of more than one ligand to one DNA should be expected.

Here we present a systematic screening and subsequent comprehensive characterization of Ru(II) complexes with different ancillary and experimental ligands that allows us to a) parse the effects of ancillary vs experimental ligands and b) understand the specific ligand properties required for effective and selective GQs binding and “light switch” properties. The successful approach has been to modify or extend the dppz ligand. Therefore, in this work we studied two classes of Ru(II) complexes, [Ru(bpy)₂(L)]²⁺ and [Ru(phen)₂(L)]²⁺, with eight experimental ligand, L (Fig. 1) including [Ru(bpy)₂(dppz)]²⁺ as a reference. For three of the experimental ligands, allox, me₂allox, and amino, we investigated the effect of ancillary ligands, bpy vs. phen, on the GQ binding properties. Our results provide deeper understanding of how the Ru(II) ligands interact with and light-up in the presence of GQ DNA.

3. Materials and Methods

3.1. Ruthenium complexes and oligonucleotides

All complexes used in this study were synthesized as described previously^37,38 and stored as dry powder at −20 °C. Each compound was dissolved directly into the desired buffer or water to a concentration of 0.5 –1.5 mM using sonication, and all complexes were stored at − 80 °C for further use. The concentration of each complex was determined using reported extinction coefficients^37,38 listed in the Supporting Information (SI) Table S1. The oligonucleotide, dAGGG(TTAGGG)₃ (Tel22), was purchased from Midland Certified Reagent Company (Midland, TX) or IDT (Texas, USA), dissolved in Millipore water to ~1 mM, and stored at 4 °C. The fluorescently-labeled oligonucleotide 5′-6-FAM-GGG(TTAGGG)₃-Dabcyl-3′ (F21D) was purchased from IDT, diluted to 100 μM in Millipore water, and stored at − 80 °C. Genomic calf thymus (CT) DNA was purchased from Sigma Aldrich (Missouri, USA), dissolved in 10 mM lithium cacodylate buffer pH 7.2 with 1 mM sodium EDTA and kept at 4 °C on a nutator. After four days, the sample was centrifuged for 10 min to remove undissolved DNA and the supernatant was stored at 4 °C for no more than four months. The concentration of DNA stock solutions was determined via UV-vis measurements using extinction coefficients listed in SI Table S2. Three lithium cacodylate-based buffers were used:10 mM lithium cacodylate, pH 7.2, 5 mM KCl, 95 mM LiCl (5K); 10 mM lithium cacodylate, pH 7.2, 50 mM NaCl, 50 mM LiCl (50Na); and 10 mM lithium cacodylate, pH 7.2, 100 mM LiCl (100Li), SI Table S3. Unless specified, experiments were conducted in 5K buffer.

GQ-forming sequences were heated alone or in the presence of up to five equivalents of Ru(II) complexes to 90 °C for 10 min to denature any non-specific secondary structures, cooled either on ice (fast cooling) or slowly over the course of 3–4 hours (slow cooling). Samples were stored at 4 °C overnight and subjected to circular dichroism (CD) studies to verify their secondary structures. Ligand interactions, tested via CD and polyacrylamide gel electrophoresis (PAGE), were found to be more efficient under slow cooling conditions that were subsequently used for all experiments. In some cases, samples were prepared by ligand addition to pre-annealed Tel22. CD experiments verified that such sample preparation did not alter the resulting GQ-Ru(II) complex structure.

3.2. Fluorescence resonance energy transfer (FRET) melting assays

FRET melting assays³⁹ were performed according to a protocol well established in our lab using doubly labeled (5’-FAM and 3’-Dabcyl) human telomeric DNA, F21D.⁴⁰ F21D was annealed at 10 μM in 5K buffer; these samples were diluted and mixed with varying amount of Ru(II) such that the final concentration of F21D was 0.2 μM and that of Ru(II) was 0–4 μM. The fluorescence of 6-FAM was measured at 519 nm when excited at 496 nm on an MJ research DNA Engine Real-Time PCR machine. Experiments were repeated in 50Na buffer using only the highest concentration of the Ru(II) complexes, 4 μM, as we expected low values of stabilization temperatures. Samples were incubated at room temperature for one hour before melting. Each experiment contained duplicate samples and was repeated at least three times. Melting curves were normalized, duplicate samples were averaged, the resulting curves were smoothed, and their first derivatives were obtained. The melting temperatures were determined manually from the maximum of the first derivatives (associated error was assumed to be ± 0.5 °C). The overall error presented here was calculated based on averaging the melting temperatures across trials and if needed using error propagation. FRET was also conducted on Ru(II) complexes alone to assure that they do not interfere with the melting data.

FRET competition experiments were conducted to establish the ability of Ru(II) complexes to selectively stabilize GQ DNA. The studies utilized unlabeled CT DNA as a duplex competitor. Increasing amounts of CT DNA (up to 96 μM per base pair), were added to a mixture of 0.20 μM F21D and 1.6 μM of each Ru(II) complex. In the control experiment, 96 μM of CT DNA was added to 0.20 μM F21D to assure that the presence of competitor did not alter the melting temperature of F21D. The selectivity of each ligand was quantified using a selectivity coefficient, S, which is proportional to the stabilization of GQs at the highest concentration of CT DNA tested (96 μM):

$$S=\frac{\Delta T_{CTDNA}}{\Delta T_{NoCTDNA}}$$

where ΔT_{CT DNA} and ΔT_{No CT DNA} are the stabilization temperatures in the presence and absence of CT DNA, respectively.

3.3. UV-vis measurements

We used an Agilent Cary 3500 or Cary 300 UV-vis spectrophotometer equipped with a Peltier block probe temperature controller (± 0.5 °C error). Spectra were collected in 1-cm quartz cuvettes in the range of 352–670 nm.

3.3.1. Binding Stoichiometry in 5K buffer determined by Job’s method

The binding stoichiometry was confirmed via Job’s method⁴¹. Ru(II) complexes and Tel22 GQ were prepared at equal concentrations in 5K buffer. Two sets of titrations were performed. In the first set, Ru(II) complex was placed in both sample and reference cuvettes. The sample cuvette was titrated with Tel22 while the reference cuvette was titrated with 5K buffer. In the second set of titrations, Tel22 was placed in the sample cuvette and 5K buffer was placed in the reference cuvette. Both cuvettes were titrated with the same Ru(II) complex. The two experiments combined spanned a molar fraction of Ru(II) complexes from 0 to 1. In each case, difference spectra were collected. Job plot titrations were repeated at least twice, and their results were averaged. The data were plotted as difference in absorbance at a particular wavelength against mole fraction of the Ru(II) complex. These plots were analyzed using two ways: tangents method and max/min of the curve⁴². In the tangent method, two straight lines are drawn through the left and right part of the data and their intercept is used as a molar ratio suggestive of the stoichiometry; it is known that this method works best for complexes with tight binding, K_a > 10⁶ M⁻¹. In the case of weaker binding (K_a < 10⁶ M⁻¹), binding stoichiometries suggested by min/max on the Job plots are deemed to be more accurate. The Ru(II) complexes under study have binding constants of ~10⁵ M⁻¹ (see Table 1); thus, we used the latter method to determine molar fraction of the ligand and binding stoichiometry.

Table 1.

Binding parameters and other spectroscopic data for Tel22-Ru(II) complexes. All binding constants are reported for 2:1 Ru(II) to Tel22 binding mode, except for phen-amino for which data are better fit with 3:1 stoichiometry.

Ligand	Job plot	UV-vis titrations			FL titration
	Mole fraction	Ka × 105, M−1	Δλ*	%H*	Ka × 105, M−1
Bpy-allox	nm	0.5 ± 0.1	6.3 ± 0.6	9 ± 3	nm
Bpy-me2allox	0.64 ± 0.03	1.1 ± 0.1	0	22 ± 1	2.1 ± 0.2
Bpy-diamino	0.71 ± 0.02	5.1 ± 1.7	16 ± 1	19 ± 1	1.53 ± 0.04
Bpy-amino	0.63 ± 0.04	1.0 ± 0.3	12 ± 1	10 ± 1	0.9 ± 0.1
Bpy-dppz	nm	6 ± 3	0	25 ± 3	nm
Bpy-aap	0.62 ± 0.03	3.4 ± 1.9	1 ± 1	17 ± 4	2.1 ± 0.3
Phen-allox	nm	0.8 ± 0.3	9 ± 1	13 ± 1	nm
Phen-me2allox	0.72 ± 0.08	4.6 ± 0.9	5 ± 1	17 ± 2	1.8 ± 0.6
Phen-amino	0.71 ± 0.02	4.1 ± 1.0	11 ± 1	16 ± 2	2.1 ± 0.3

Nm – not measured;

^*numbers are reported for the lower energy peak

3.3.2. Binding constant determination in 5K buffer

In UV-vis titrations, we held the concentrations of the Ru(II) complexes constant (typically 13–25 μM) while the concentration of Tel22 was incrementally increased.Tel22 samples were prepared at 0.7–1.1 mM and included the Ru(II) complex, added post-annealing, to maintain its constant concentration throughout the titration. After each addition of Tel22, the solution was mixed thoroughly, incubated for 2 min, and the UV-vis absorbance was recorded. Titration was considered complete when 2–3 spectra were fully superimposable, which typically required about a 5-fold excess of DNA. Titration data were used to obtain the change in wavelength of maximum absorption (Δλ) and percent hypochromicity, %H, as described elsewhere.⁴³ Titrations were repeated three times and the results were averaged.

In order to determine the number of significant components throughout titrations, a linear algebra method, Singular Value Decomposition (SVD), was used.⁴⁴ The analysis was performed using MatLab and Origin 9.1. When SVD confirmed that binding occurs in a simple two-component equilibrium, direct fit was used to determine the binding constant, K_a. The details of direct fit are presented in our earlier work.⁴³ Based on the results from Job plots (see Section 3.3.1), 2:1 and 3:1 Ru(II):Tel22 binding models were employed. It is important to note, that the accurate values of K_a could only be determined for 1:1 binding model. For higher stoichiometries an approximation needs to be made that the ligand binds to identical and independent binding sites on Tel22 in a non-cooperative manner. Such data treatment allows us to calculate binding sites on DNA by multiplying DNA concentration by 2 (for 2:1 binding) or by 3 (for 1:3 binding). The binding constant determination was performed in GraphPad Prism 4.0. SVD vector v2 was used for all the data fitting.

3.4. Fluorescence spectroscopy

Fluorescence experiments were conducted on a Photon Technology International QuantaMaster 40 spectrofluorometer. Spectra were collected at 20 °C, in 1 cm plastic cuvettes with four clear windows, using a step increment of 1 nm, integration time of 0.5 s, and 10 nm slits. The samples were excited at 440 nm and emission was collected from 550 to 800 nm.¹⁸ A Ru(II) complex peak was usually detected at 595 – 611 nm, depending on the specific complex.

3.4.1. Binding constant determination using fluorescence in 5K buffer

To confirm binding constants obtained from UV-vis titrations, fluorescence titrations were performed for selected ligands. Ru(II) complexes of 0.5 or 5 μM were titrated with up to 15 equivalents of Tel22 at 20 °C. Titrations were repeated twice and averaged. Direct fit was used, as described above, to determine K_a values which were compared to those determined from UV-vis titrations. SVD vector v1 was used for all the data fitting.

3.4.2. “Light switch” effect in the presence of Tel22 in 5K buffer

To test the “light switch” effect, fluorescence of 1–2 μM Ru(II) complexes was measured at 20 °C. The bpy-aap complex, which demonstrated particularly strong background fluorescence, was studied at 0.125 μM to ensure that the fluorescence intensity would be within the detectable range of the instrument after addition of Tel22. Ten equivalents of Tel22 GQ was added to the Ru(II) complexes, samples were mixed thoroughly and equilibrated for 5 min. This high amount of Tell22 needed to reach fluorescence saturation was determined via fluorescence titrations (Section 3.4.1). The emission spectra were collected for buffer alone, Ru(II) alone and Ru(II) in the presence of Tel22. All spectra were corrected for lamp fluctuation. The increase in fluorescence intensity was quantified using the fluorescence enhancement (F.E.), calculated according to the following formula:

$$\mathit{\text{Fluorescence Enhancement }}=\frac{F_{(Ru+Tel22)}-F_{Ru}}{F_{Ru}}\text{, where Ru denotes a Ru(II) complex}.$$

3.5. Gel electrophoresis

Samples of Tel22 at 40 μM were annealed at 90 °C for 5 min with a 5-fold excess of each Ru(II) complex. The samples were cooled either on ice (flash-cooling, FC) or slowly over the course of 3–5 h (slow-cooling, SC) and then placed at 4 °C overnight. Tel22 alone was used as a control. Samples were loaded onto a 20% non-denaturing polyacrylamide gel in 1× TBE buffer (89 mM Tris–borate, 2 mM EDTA, pH 8.3) with 5 mM KCl at 220 V and 16 °C for 4 h. Oligothymidylate markers 5′ dT_n (where n = 6, 15, 24, 30 and 57) were used as internal migration standards (and not necessarily as length markers) along with duplex markers dx9: 5′-GCGTATCGG-3’ + 5’-CCGATACGC-3’ and dx12: 5′-GCGTGACTTCGG-3’ + 5′-CCGAAGTCACGC-3’. DNA bands were visualized by UV-shadowing and ruthenium fluorescence using Typhoon (filter 670 nm, excitation at 488 nm, voltage 600 V).

3.6. Circular Dichroism (CD) spectroscopy

Samples for CD were prepared either by diluting PAGE samples to a final concentration of ~2–4 μM immediately before collecting CD scans or by annealing Tel22 typically between 3–10 μM in the presence of up to 5 eq of Ru(II) complexes in 5K or in 50Na buffers. CD wavelength scans were collected in 10 or 2 mm quartz cells at 20 °C using JASCO J-815 or J-1500 spectropolarimeter with a 2 nm band width, 50–200 nm min⁻¹ scan speed, 1 nm step, and 220–350 nm spectral window. Five to ten scans were recorded and averaged. The data were treated as described elsewhere.⁴³

3.7. Electrospray ionization mass spectrometry (ESI-MS)

Samples for ESI-MS were prepared using annealed (either via the flash-cooling or slow-cooling) Tel22 mixed with Ru(II) complexes to a final concentration of 5 μM Tel22 and 10 or 20 μM Ru(II) complexes. These solutions were prepared in 100 mM ammonium acetate. All measurements were performed on an Exactive Orbitrap instrument (Thermo, Bremen, Germany) using the electrospray ionization HESI probe in negative ion mode. The HESI probe was not heated, and the instrument conditions were optimized to avoid dissociation of the Tel22-Ru(II) complexes. The capillary temperature was set to 110 °C and the sheath gas to 70 °C. The skimmer voltage was set to 10 V and the tube lens voltage to 180 V. The flatapole and transfer multipole voltage offset were tuned to increase the sensitivity and avoid fragmentation.

We determine equilibrium association constants (M⁻¹) using ESI-MS for the 11 compounds and Tel22 sequence. The equilibrium association constants are defined in the following way:

$${\text{Ka}}_1=[1:1]/\left([\text{DNA}]*{[\text{L}]}_{\text{free }}\right)$$

$${\text{Ka}}_2=[2:1]/\left([1:1]*{[\text{L}]}_{\text{free }}\right)$$

$${\text{Ka}}_3=[3:1]/\left([2:1]*{[\text{L}]}_{\text{free }}\right)$$

Where 1:1, 1:2, and 1:3 represent Tel22 bound by one, two, and three Ru(II) complexes, respectively.

The variables in the equations above are defined in the following way:

$$[\text{DNA}]=\left[{\text{DNA}}_{\text{tot }}\right]\times {\text{I}}_{\text{DNA}}/\left({\text{I}}_{\text{DNA}}+{\text{I}}_{1:1}+{\text{I}}_{2:1}+{\text{I}}_{3:1}\right)$$

$$[1:1]=\left[{\text{DNA}}_{\text{tot }}\right]\times {\text{I}}_{1:1}/\left({\text{I}}_{\text{DNA}}+{\text{I}}_{1:1}+{\text{I}}_{2:1}+{\text{I}}_{3:1}\right)$$

$$[2:1]=\left[{\text{DNA}}_{\text{tot }}\right]\times {\text{I}}_{2:1}/\left({\text{I}}_{\text{DNA}}+{\text{I}}_{1:1}+{\text{I}}_{2:1}+{\text{I}}_{3:1}\right)$$

$$[3:1]=\left[{\text{DNA}}_{\text{tot }}\right]\times {\text{I}}_{3:1}/\left({\text{I}}_{\text{DNA}}+{\text{I}}_{1:1}+{\text{I}}_{2:1}+{\text{I}}_{3:1}\right)$$

$${[\text{L}]}_{\text{free }}={[\text{L}]}_{\text{tot }}-{[\text{L}]}_{\text{bound }}$$

$${[\text{L}]}_{\text{bound }}=[1:1]+2\times [2:1]+3\times [3:1]$$

The starting [DNA_tot] = 5 μM and [L]_tot. = 20 μM. The concentration of each species in solution at equilibrium is calculated using the ratio of intensities and the mass balance equations.⁴⁵ No binding model equations are used for the determination of the equilibrium binding constants because mass spectrometry is able to resolve each stoichiometries present at equilibrium in solution. The concentration of the species are determined using the relative intensities of the mass spectral peaks. Matlab software was used for peak integration. Sum of 5- and 4- ions are taken into account; windows of integration encompass up to five ammonium adducts.

4. Results and Discussion

Eight complexes of the general formula [Ru(bpy)₂L]²⁺ and three complexes of the general formula [Ru(phen)₂L]²⁺ (Fig. 1) were examined as racemic mixtures for their ability to interact with human telomeric DNA. Here, bpy and phen are ancillary ligands, while L is an experimental ligand. For the ease of presentation, the complexes will be referred to as bpy-L or phen-L. Experimental ligands in these complexes are all derivative of dppz except for aap which is the ring cleavage product of keto ligand.³⁸ As a GQ model we used a well-studied human telomeric DNA, Tel22, with the sequence 5’-AGGG(TTAGGG)₃-3’.^46,47 We chose Tel22 because its binding to numerous other Ru(II) polypyridyl ligands been investigated^18,19,36,48 allowing for a straightforward comparison with our data. We tested all Ru(II) complexes via: 1) FRET melting assays for their stabilizing ability and selectivity for GQ vs. dsDNA; 2) electrospray ionization mass spectrometry (ESI-MS) for their affinity toward Tel22; 3) fluorescence for their “light switch” property; and 4) CD and PAGE for their ability to cause structural changes in Tel22. The most selective and stabilizing complexes were further characterized by UV-vis and fluorescence to determine optimal binding conditions, binding stoichiometry, and affinity. [Ru(bpy)₂(dppz)]²⁺ and [Ru(bpy)₃]²⁺ served as controls. We implemented two different protocols for preparation of Tel22-Ru(II) complexes due to demands of the specific method. For FRET and spectroscopic titrations, Tel22 was annealed alone and then mixed with Ru(II) complexes. For ESI-MS and PAGE, Tell22 was annealed with the specified amount of Ru(II). Therefore, we conducted a CD study in 5K buffer on samples prepared in parallel via the two protocols and observed nearly identical signatures indicating similar structure of resulting Tel22-Ru(II) complexes (Fig. S1).

4.1. Thermal stabilization and selectivity of Ru(II) complexes via FRET assays

The first requirement for a good GQ ligand is its ability to stabilize GQ structures although recently the importance of GQ-destabilizers was also highlighted⁴⁹. The ability of a ligand to stabilize GQ DNA is typically quantified by ΔT_m - the increase in the melting temperature (T_m) of GQ DNA in the presence of a fixed amount of ligand. For screening purposes, we consider ligands that produce ΔT_m > 10 °C at reasonable ligand concentration to be promising thus warranting further investigation. We used a fluorescently labeled analog of Tel22, F21D, which contains the 5’-FAM fluorophore and the 3’-Dabcyl quencher. When F21D is folded into a GQ, the FAM dye is close to the Dabcyl quencher, and no fluorescence is expected. Melting of the GQ moves the dye and the quencher apart, leading to an increase in fluorescence signal. Analysis of the signal allowed us to determine T_m for DNA alone and in the presence of Ru(II) ligands, and thus we can calculate ΔT_m.

4.1.1. Thermal stabilization of F21D by [Ru(bpy)₂L]²⁺ in 5K buffer

F21D adopts hybrid conformation in 5K buffer and displays a T_m of 44.8 ± 0.8 °C, consistent with our earlier studies.⁴⁰ All [Ru(bpy)₂L]²⁺ complexes displayed a dose-dependent stabilization of F21D (Fig. 2). Representative raw data are shown in Fig. S2 and stabilization temperatures are listed in Table S4. Bpy-dppz, used as a reference, stabilized F21D by 14 ± 1 °C at 20 eq. (4 μM) and by 4 ± 1 °C at 2 eq. Consistent with our data, an earlier UV-vis study found that at 1 eq. bpy-dppz stabilized Tel22 in K⁺ and Na⁺ buffers by 4.7 and 5.3 °C, respectively.¹⁸ Bpy-me₂allox demonstrated the highest stabilization with a ΔT_m of 17.3 ± 0.5 °C at 4 μM ligand while the lowest stabilization was observed for bpy-pterin and bpy-keto, with a ΔT_m of 3.4 ± 0.5 and 3.7 ± 0.9 °C, respectively. Based on FRET melting assays, we can rate the ability of [Ru(bpy)₂L]²⁺ complexes to stabilize GQ DNA as follows: me₂allox > dppz ≈ diamino ≈ aap > amino ≈ allox > pterin ≈ keto. All [Ru(bpy)₂L]²⁺ complexes except those with L = pterin and keto displayed ΔT_m at or above 10 °C at ligand saturation, indicating that they are promising GQ binders.

Fig. 2.

Dose dependent stabilization of 0.2 μM F21D by 0–20 eq (0–4 μM) of Ru(II) complexes in 5K buffer. Data for eight [Ru(bpy)₂L]²⁺ complexes are shown with filled symbols and data for three [Ru(phen)₂L]²⁺ complexes are shown as open symbols connected by dash line. The dotted line indicates the 10 °C stabilization.

4.1.2. Effect of ancillary ligands, bpy vs. phen, on the GQ stabilizing ability of Ru(II) complexes

To explore the effect of the ancillary ligands on the stabilizing abilities of Ru(II) complexes toward human telomeric DNA, three [Ru(phen)₂L]²⁺ complexes, where L = me₂allox, allox, and amino, were tested in FRET assays (Fig. 2). The stabilization temperatures for [Ru(bpy)₂L]²⁺ and [Ru(phen)₂L]²⁺ complexes with the same experimental ligand were within experimental error of each other with phen-me₂allox, displaying the highest ΔT_m of 17.8 ± 0.7 °C comparable to that for bpy-me₂allox of 17.3 ± 0.5 °C at 4 μM. The similarity in stabilization temperature suggests that GQ binding is primarily mediated by the experimental and not the ancillary ligand.

Contrary to our observations, several studies have reported that stabilizing abilities of Ru(II) complexes depends strongly on its ancillary ligand. For example, [Ru(bpy)₂(ptpn)]²⁺ displayed superior stabilizing ability toward GQ as compared to [Ru(phen)₂(ptpn)]²⁺ in FRET experiments.¹⁵ The authors suggested that ancillary ligands can be used to fine-tune the Ru(II)-GQ interaction.¹⁵ Similarly, Yu et al. reported a higher stabilizing ability for [Ru(bpy)₂(mitatp)]²⁺ than [Ru(phen)₂(mitatp)]²⁺.²⁴ This difference was suggested to be due to the steric hindrance caused by phen’s larger surface. On the contrary, Wang et al. reported that [Ru(phen)₂(3-tppp)]²⁺ is a better GQ stabilizer than [Ru(bpy)₂(3-tppp)]²⁺.⁵⁰ The authors propose this difference to be due to phen’s stronger hydrophobicity and larger aromatic surface, which allows for better π-π interactions between the ligand and GQ. In general, while our data suggest that ancillary ligands likely are not engaged in stabilizing human telomeric DNA, such interaction may also depend on the specific GQ, and can be governed by its topology or conformation of its loops/overhangs. Recent structural studies demonstrate loop/overhang binding of Ru(II) complexes in place of, or in addition to, G-tetrad binding.^33,35,36

4.1.3. Thermal stabilization of F21D by [Ru(bpy)₂L]²⁺ in 50Na buffer

In sodium-rich buffers F21D adopts predominantly antiparallel topology.⁴⁰ To test whether the stabilizing ability of Ru(II) complexes depends on the nature of the monovalent cations (Na⁺ vs K⁺) we repeated FRET experiments in 50Na buffer. We included the higher amount of Na⁺ as compared to K⁺ (while maintaining the same ionic strength) because human telomeric DNA, and GQs in general, is more stable in K⁺ vs Na⁺ buffers.² Addition of Ru(II) ligands to Tel22 in 50Na buffer did not alter the GQ conformation and in a few cases (specifically for bpy-amino and bpy-diamino) lead to some increase in the signal intensity (Fig. S3). FRET data, collected in Table S4, indicate that the stabilization of F21D by Ru(II) complexes in 50Na buffer is weak and even the best stabilizers with me₂allox experimental ligand lead to only 4.4–7.5 °C stabilization at saturating concentrations of 4 μM. We reported earlier that a series of cationic porphyrins display significantly decreased ability stabilize F21D in Na⁺ vs K⁺ buffers.⁵¹Another porphyrin ligand, N-methylmesoporphyrin IX, stabilized F21D to a great extent in the 5K buffer but did not display any stabilization in the 50Na buffer⁴⁷.

Stabilizing abilities of Ru(II) complexes in both buffers correlate well according to the following equation ΔT_m(50Na) = −1.066 + 0.414 × ΔT_m(5K) (R² = 0.740), Fig. 3. Good correlation between stabilizing abilities of ligands in K⁺ vs Na⁺ buffers was reported for other Ru(II) GQ ligands.^50,52 Data in Fig. 3. indicates that phen-me₂allox displays the best stabilization of F21D in both buffers and underlines the importance of the ancillary ligand for fine-tuning the stabilizing properties of Ru(II) complexes. Due to the strong correlation across two buffers, understanding the stabilizing ability of Ru(II) complexes toward GQ DNA in one set of conditions could allow us to predict their behavior in a different set of conditions.

Fig. 3.

Comparison of stabilization temperatures in 5K and 50Na buffers. The dash line represents a linear fit to the data: ΔT_m(50Na) = −1.066 + 0.414 × ΔT_m(5K) (R² = 0.740).

4.1.4. Selectivity of Ru(II) complexes for GQ vs. dsDNA.

FRET assays allow facile determination of ligands’ selectivity for GQ vs. other DNA structures such as dsDNA. To determine the selectivity of Ru(II) complexes, we included an increasing amount of unlabeled dsDNA competitor, CT DNA, while keeping the concentration of F21D and [Ru(bpy)L]²⁺ constant (at 0.2 and 1.6 μM, respectively). We observed a gradual decrease of T_m for F21D (Fig. 4) indicating that at high concentration CT DNA outcompetes GQ DNA for Ru(II) binding. The [Ru(phen)₂L]²⁺ complexes displayed selectivity comparable to that of [Ru(bpy)₂L]²⁺ complexes (Fig. S4) suggesting that the ancillary ligand does not play a defining role in the binding of the Ru(II) complexes to GQ DNA.

Fig. 4.

Stabilization of 0.2 μM F21D with 1.6 μM of six [Ru(bpy)₂(L)]²⁺ complexes in the presence of increasing amounts of CT DNA in 5K buffer. The equivalents CT DNA are shown in the legend. We excluded bpy-keto and bpy-pterin because of their poor stabilizing properties.

The extent of selectivity can be quantified using the selectivity coefficient S (Table S4). To orient the reader, a compound with exceptional selectivity is characterized by S approaching 1; S of > 0.25 is reasonable. For Ru(II) complexes of this study, the selectivity coefficient ranged from 0.03 to 0.48. Bpy/phen-allox, and bpy/phen-me₂allox display the greatest selectivity maintaining ~50% of the stabilization in the presence of 480-fold excess of CT DNA and displaying S of 0.4–0.5. It was suggested based on biophysical studies and DFT calculation of ligand structures that the presence of two methyl groups in me₂allox experimental ligand may inhibit intercalation of bpy-me₂allox into CT DNA,⁵³ leading to improved selectivity. Yet, both allox and me₂allox performed similarly well, undermining the direct benefit of the two methyl groups. The least selective complexes are bpy-dppz and bpy-amino. Poor selectivity of bpy-dppz was previously reported.^18,26 Comparison of complexes with phen and bpy ancillary ligands indicates that they display similar selectivity (Table S4) pointing to the importance of the experimental ligands.

In summary, we screened eleven [Ru(bpy)₂L]²⁺ and [Ru(phen)₂L)]²⁺ complexes for their ability to selectively stabilize human telomeric DNA. Correlation diagrams of stabilizing ability vs. selectivity (Fig. S5) and of stabilizing ability in 5K vs. 50Na (Fig. 3) identified me₂allox as the experimental ligand capable of conferring excellent stabilizing ability and good selectivity to polypyridyl Ru(II) complexes. All complexes with ΔT_m ≥ 10 °C were further characterized for their GQ binding affinities and stoichiometry via UV-vis and fluorescence titrations as well as ESI-MS.

4.2. Spectroscopic and ESI-MS characterization of Tel22 binding to Ru(II) complexes

We measured binding constants for all Ru(II) complexes but bpy-keto and bpy-pterin (which displayed poor Tel22 stabilization in FRET) using UV-vis and fluorescence spectroscopy. We corroborated determined binding constants using ESI-MS.

4.2.1. Binding stoichiometry for Tel22-Ru(II) complexes via Job method.⁴¹

We began by determining model independent stoichiometry of Ru(II) binding to Tel22 via Job method. A representative Job Plot is shown in Fig. 5 and the rest of the data is shown in Fig. S6. The curvature of the plots (i.e., the absence of sharp point of change of slope) indicates modest binding. Thus, we determined a binding ratio using minima or maxima on the Job plot curves rather than employing a tangent method. Molar fractions for all complexes lie between 0.62–0.72, Table 1, suggesting that Ru(II) complexes bind Tel22 likely with 2:1 stoichiometry although 3:1 stoichiometry could also be considered. This observation is consistent with CD titration data for [Ru(bpy)₂(ptpn)]²⁺ and [Ru(phen)₂(ptpn)]²⁺, which bind to human telomeric DNA with a stoichiometry of 2:1.¹⁵ Even higher binding stoichiometry of 4:1 is observed in the crystal structures of [Ru(TAP)₂(11-CN-dppz)]²⁺ with tetramolecular model of human telomeric DNA, (dTAGGGTTA)₄³⁴ or [Ru(TAP)₂(dppz)]²⁺ with (dTAGGGTT)₄.³⁵ We will assume for the rest of this discussion that the binding sites are identical and independent; however, binding stoichiometry higher than 1:1 may also suggest existence of different binding sites. Both nearly identical and different binding sites for Ru(II) complexes on human telomeric DNA were detected in structural studies.^34,35

Fig. 5.

Stoichiometry of bpy-diamino binding to Tel22 determined via a Job plot method at 25 °C in 5K buffer. Job plot suggests a 2:1 stoichiometry.

4.2.2. Binding affinity via UV-vis and fluorescence titrations.

Representative UV-vis titrations are shown in Fig. 6 and Fig. S7 and data analysis of the titration spectra is collected in Table 1 and Table S5. UV-vis titrations rely on Ru(II) MLCT bands found between 360–600 nm.^37,38 Upon addition of Tel22, Ru(II) spectra display modest hypochromicities (decrease of signal intensity) of 10–25 % and only bpy-amino, bpy-diamino, and pehn-diamino display significant red shifts of 12–16 nm. Hypochromic and bathochromic (red shift) effects indicate that binding occurs due to π-π stacking between Ru(II) complexes and bases of GQ, likely G-tetrads. Titration data were analyzed using Singular Value Decomposition (SVD, Fig. S8). In all cases but three, SVD analysis indicates a two-state equilibrium between free and bound Ru(II) complexes without intermediates. For bpy-dppz, SVD suggests two-state biding in the region 0 < [Tel22]/[Ru(II)] < 2 and a more complicated binding model for conditions in which [Tel22]/[Ru(II)] > 2. Similarly, bpy/phen-allox display two-state behavior in the region 0 < [Tel22]/[Ru(II)] < (3–4). The two-state binding equilibrium is characterized by an isosbestic point, which is observed in all our systems and listed in Table S5. We analyzed titration data using a direct fit method. Informed by the Job method, we tested two different binding models, 2:1 and 3:1 Ru(II) to Tel22. The former model produced better fits for all complexes but phen-amino. Binding constants for 2:1 binding mode are in the range (0.5–6)×10⁵ M⁻¹, Table 1. The values of binding constants suggest that binding is modest.

Fig. 6.

Determination of K_a for the Tel22-bpy-diamino complex via UV-vis and fluorescence titrations in 5K buffer at 25 °C. (A) Representative UV-vis absorption spectra of 20.9 μM of bpy-diamino titrated with 980 μM Tel22. The final [GQ]/[Ru(II)] ratio was ~5.7. (C) Representative fluorescence spectra of 0.5 μM bpy-diamino titrated with 670 μM Tel22. The final [GQ]/[Ru(II)] ratio was 13.4. (B, D) The best fit of the titration data using direct fit and 2:1 Ru(II):Tel22 binding model is shown by the solid line and 95% confidence interval is displayed by the dashed lines. The concentration of the binding site here is 2×[Tel22].

We confirmed binding constants using fluorescence spectroscopy which relies on the “light switch” property of Ru(II) complexes. Representative fluorescence titration is shown in Fig. 6C. As expected, the fluorescence of Ru(II) complexes increased with increasing amounts of added DNA which protects Ru(II) complexes from aqueous medium. Binding constants, (0.9–2.1)×10⁵ M⁻¹ are consistent with those found in UV-vis titrations. Observed fluorescence enhancement is discussed further in Section 4.4.

4.2.3. Binding affinity via ESI-MS.

To verify binding stoichiometries and binding constants obtained in spectroscopic titrations and to screen efficiently all complexes for their binding to Tel22, we employed ESI-MS method (Fig. 7). We used soft ion handling conditions in the source and checked these conditions as described previously⁵⁴ (Fig. S9), to minimize risks that non-covalent complexes would be disrupted upon transfer from the solution to the gas phase. The risk of disrupting the ligand-DNA complex in the gas phase can be neglected, because we have a positively charged ligand bound to the negatively charged oligonucleotide. Because of the method requirements, 100 mM ammonium acetate was used in place of the 5K or 50Na buffers (note all buffers have similar ionic strength of 100–110 mM). The softness of the conditions used in this work is attested by the preservation of inter-quartet ammonium ions (Fig. 7B–C). In non-native conditions, ammonia would be lost.

Fig. 7.

Electrospray mass spectrum of Tel22 with bpy-aap in 100 mM NH₄OAc. The concentration of Tel22 is 5 μM and of bpy-aap is 20 μM. (A) Free Tel22, 1:1 and 2:1 complexes are observed at charge states 5- and 4-. (B, C) zoom on the 1:1^4- and 2:1^4- complexes, respectively. Two or three specific ammonium ions are conserved, indicating the preservation of the GQ structure of Tel22 and a putative binding mode by end-stacking and extra cation incorporation. Intercalation can be ruled out because the number of ammonium ions would be reduced (by displacement of some of the inner NH₄⁺).

Ru(II) complexes with Tel22 were prepared at 4:1 ratios and ESI-MS spectra were collected for each complex (Fig. 7 and Fig. S10). The anionic charge states of z = 4 and z = 5 were observed in each spectrum; peaks corresponding to free Tel22 and Tel22-Ru(II) complexes were seen for both charge states and allowed us to calculate the K_a values (Table S7) as well as determine the total concentration of bound ligand (Fig. S11). The most prominent species in the MS spectra are Tel22 and 1:1 Ru(II)-Tel22, although 2:1 and 3:1 complexes are also observed. Based on the K_a values and Fig. S11, bpy-allox, bpy-amino, bpy-dppz, and bpy-aap display the strongest binding to Tel22. It is interesting that all [Ru(phen)₂L]²⁺ complexes display a rather weak binding affinity in ESI-MS.

Binding constants determined by UV-vis and fluorescent titrations under relatively similar conditions agree well with each other (Table S6) and reflect a modest binding. ESI-MS data are also in good agreement, although the ESI-MS yielded significantly smaller K_a values for the two studied phen complexes, phen-me₂allox and phen-amino. For the classical spectroscopic methods, such as UV-vis and fluorescence, as soon as 2:1 complexes are present, the analytical solution for the binding equations are too difficult to be effectively used and approximations are employed instead (e.g., identical and independent binding sites and non-cooperative binding). In mass spectrometry, on the other hand, each specie (1:1, 2:1, etc) is independently resolved and their concentration can be accurately determined via integration of the observed peaks. Thus, it is not surprising that the K_a values between spectroscopic methods and mass spectrometry differ. The difference is compounded by the exact definition of K_a, difference in buffers, and different response factors between free and bound Ru(II) complexes (different response between the 4- and 5- charge states can be seen in Fig. 7). For the same spectroscopic technique, no significant differences were observed between [Ru(bpy)₂L]²⁺ and [Ru(phen)₂L]²⁺ complexes with the same experimental ligand, in agreement with our FRET data.

Binding of a variety of Ru(II) complexes using spectroscopic titrations (e.g., UV-vis, fluorescence, CD) and isothermal titration calorimetry has been reported. For example, binding of [Ru(bpy)₂(dppz)]²⁺ and [Ru(phen)₂(dppz)]²⁺ to Tel22 in 10 mM Tris 7.5, 100 mM KCl was measured to be 6.3×10⁵ and 2×10⁶ M⁻¹, respectively using the Scatchard analysis of UV-vis data.²⁶ [Ru(phen)₂(bppp)]²⁺ and [Ru(phen)₂(pppp)]²⁺ exhibited K_a values of 3.8×10⁵ and 8.3×10⁵ M⁻¹, respectively, in K⁺ buffer.⁵² Overall, these values are in good agreement our data. Some Ru(II) complexes display tighter binding constants. For example, two dinuclear [(Ru(bpy)₂)₂L]⁴⁺ complexes, where L is based on the tatpp ligand, interact with Tel22 with K_a of ~(4–10)×10⁶ M⁻¹ and 1:1 binding stoichiometry.⁵⁵ [Ru(bpy)₂L]²⁺ with large π-delocalized indoloquinoline moiety interacts with Tel22 with K_a on the order of ~10⁷ M⁻¹.⁴⁸ Similarly, dinuclear Ru(II) complex, [(Ru(bpy)₂)₂H₂bipt)]⁴⁺ displayed tight binding to human telomeric DNA (Tel22 without 5-A) with K_a of 1.1×10⁷ M⁻¹ and 2:1 Ru(II)/GQ binding mode.⁵⁶

Binding of [Ru(bpy)₂L]²⁺ complexes with all experimental ligands studied here to CT DNA has been reported earlier using similar UV-vis and fluorescence titrations.^37,38,53,57 The K_a values are determined to be (0.5–5)×10⁶ M⁻¹ and are comparable to K_as obtained here for Tel22. The binding stoichiometry was reported as 1:1 for all complexes but bpy-allox and bpy-me₂allox, which display high binding stoichiometry of 3–6 Ru(II)-to-CT DNA.^37,53,57 The fact that values of K_a for Tel22 and CT DNA are comparable, confirms modest selectivity of the Ru(II) complexes in this study. It is, of course, possible that these Ru(II) complexes may display a stronger affinity or higher selectivity for a different GQ target.

4.3. Effect of Ru(II) complexes on the Tel22 secondary structures via polyacrylamide gel electrophoresis (PAGE) and CD spectroscopy

In order to detect if binding of Ru(II) complexes to Tel22 causes structural rearrangement, PAGE and CD experiments were conducted in parallel on the same samples in 5K buffer. While CD spectroscopy can detect structural changes in DNA and, potentially, GQ stacking topology of the new structures,⁵⁸ it does not report on the number of species present in solution. For example, a CD signature for a mixed-hybrid GQ could resemble the CD signature of a mixture of parallel and antiparallel quadruplexes. Native PAGE yields information on the number of species present in solution and, with a proper standard, can indicate their conformation as well. Therefore, the combination of CD scans and PAGE conducted on the same samples can be a powerful way to investigate structures and homogeneity of DNA samples and monitor ligand effect on the secondary structure of DNA. Tell22 alone in 5K buffer adopts hybrid conformation and travels on PAGE as a fast-moving band between dT6 and dT15 markers. PAGE indicates that most of Ru(II) complexes did not alter the location of Tel22 band suggesting that these ligands likely bind to prefolded hybrid Tel22 GQ rather than altering its conformation (Fig. S12). Three complexes, bpy-pterin, bpy-diamino, and bpy-dppz, converted part or all of Tel22 into higher order species, which appeared on the PAGE as low mobility bands. These bands can be clearly seen using fluorescence in place of UV-shadowing (Fig. S12B). Few other complexes display new higher order bands, but those bands are relatively weak: bpy-keto, bpy-aap, bpy-amino, and phen-amino. We also tested whether the cooling rate has any effect on the interaction between Ru(II) and DNA. Fast cooling conditions, where samples are placed directly on ice after annealing at 95 °C, led to the same changes as observed for a slow cooling but to a lesser extent signifying that slow cooling used in all experiments throughout this paper is necessary for efficient Tel22-Ru(II) interactions. It is important to keep in mind that the absence of new bands on PAGE can be a result of dissociation of Tel22-Ru(II) complexes during migration and does not necessarily reflect a lack of interactions.

PAGE samples were diluted to ~3 μM Tel22 and were subjected to CD scans (Fig. 8). For bpy-keto, bpy-me2allox, and bpy-dppz the CD changes are relatively minor and include change to peak width and increase in intensity of the 260 nm trough (Fig. 8A). Bpy-pterin, bpy-amino, and bpy-diamino complexes lead to a new second peak at 280–285 nm, while maintaining the original feature at 293–295 nm; all three complexes also display new bands on PAGE and, thus, the new features potentially originate from the new higher order structures. Bpy-allox and bpy-aap induce a new trough at 259 nm potentially signifying increase in the antiparallel character of Tel22 GQ.

Fig. 8.

Representative CD scans of ~3 μM Tel22 in the presence of 5 eq of [Ru(bpy)₂L]²⁺ and [Ru(phen)₂L]²⁺ collected in 5K buffer at 20 °C (A-D). All graphs are shown on the same scale. Note, we verified that the absorbance of Ru(II) complexes in this region does not affect the observed CD signal.

Examination of the CD signature of Tel22 in the presence of [Ru(phen)₂L]²⁺ complexes and comparison of this data to that for Tel22-[Ru(bpy)₂L]²⁺ in the presence of the same experimental ligand (Fig. 8 and Fig. S13) indicate that changes observed are similar for the two ancillary ligands but are more pronounced for phen. Specifically, a new feature is observed at about 275 nm while the signal at 293 nm is preserved.

We also collected CD signatures with lower ligand equivalents, 2–2.5, (Fig. S1) and observed similar CD spectra albeit with some attenuated features.

Overall, the combined CD and PAGE data indicate that majority of Ru(II) complexes interact with hybrid conformation of Tel22. In some cases, Ru(II) complexes cause formation of higher order GQ structures which are highly prominent for bpy-pterin, bpy-diamino, and bpy-dppz. The new features that are visible in the CD signature of Tel22 could be explained by the presence of these new higher order species and/or by the stacking of Ru(II) complexes with the terminal G-tetrad or their interactions with loops or overhangs.

4.4. “Light switch” effect

The “light switch” effect refers to an increase in fluorescence of a Ru(II) ligand in the presence of DNA binding partner. The fluorescence of Ru(II) originates from the MLCT between excited 4d electrons on Ru(II) and low-lying π* orbital on the conjugated experimental ligand. In aqueous solutions, the excited state is quenched due to hydrogen bonding with water. Structural studies indicate that binding to GQ DNA shields the Ru(II) complex from the solvent leading to a dramatic increase in fluorescence intensity (so called solvent exclusion).^34,36 Higher encapsulation of Ru(II) complex (e.g. simultaneous binding to a G-tetrad and a loop) leads to a greater fluorescence enhancement. Good “light switch” probes display dark background fluorescence and strong fluorescence upon binding to their target. dsDNA-binding dyes used routinely for fluorescence microscopy (propidium iodide, ethidium bromide, etc) exhibit a 20- to 30-fold increase in fluorescence (i.e., fluorescence enhancement) when bound to DNA. A compound that fluoresces without binding to a target (i.e., with high background fluorescence) is prone to give false positive results, while a compound which does not have high fluorescence enhancement would likely give false negative results in vivo.

We tested all Ru(II) complexes in this work for the “light switch” effect using Tel22 in 5K buffer and exciting the complexes at their UV-vis maxima (Table S7). bpy-amino, phen-allox, and phen-amino have low background fluorescence equal to or lower than that of bpy-dppz (Fig. 9A). bpy-aap displays a rather strong background fluorescence, 15-fold higher than that of bpy-dppz. This observation could be due to the presence of the exocyclic amid and a smaller right system of aap as compared to other experimental ligands. The background fluorescence of other complexes is low and is only 3–7 times higher than that of the bpy-dppz.

Fig. 9.

“Light switch” properties of Ru(II) complexes in 5K buffer at 20 °C. (A) Fluorescent intensity of 1.0 μM Ru(II) complexes in 5K buffer, F₀, relative to F₀ of bpy-dppz. (B) Fluorescence enhancement observed upon addition of 10 equivalents of Tel22 GQ to 1 μM Ru(II) for all complexes but bpy-aap, whose concentration was 0.125 μM.

Upon addition of Tel22 GQ, all Ru(II) complexes exhibited “light switch” effect, Fig. 9B. Three complexes stand out, bpy-dppz with the fluorescence enhancement of 370 ± 20, bpy-diamino with fluorescence enhancement of 36 ± 4, and bpy-aap, with fluorescence enhancement of 53 ± 7. Thus, bpy-dppz remains the best “light switch” probe, but its low selectivity for GQs limits its applicability in the GQ field. At the other extreme, bpy-allox, bpy/phen-me₂allox, and bpy-amino displayed rather low fluorescent enhancement of 2–5. The rest of the complexes displayed modest fluorescence enhancement of ~10. It is important to recognize that the fluorescence enhancement depends strongly on the background fluorescence. Therefore, while bpy-aap in the presence of Tel22 displays the strongest fluorescence, its fluorescence enhancement is only 53 due to a rather high background fluorescence. Interestingly, all complexes with phen ancillary ligand display rather low fluorescence in the presence of Tel22; the only reason why fluorescence enhancement numbers for phen-allox and phen-amino look reasonable (13 and 10, respectively) is due to their extremely low background fluorescence. The fluorescence enhancement does not appear to correlate with the GQ stabilization ability of Ru(II) complex as determined by FRET in 5K buffer, Fig. S4B, although all three “light switch” complexes (bpy-dppz, bpy-aap, and bpy-diamino) display the best combination of GQ stabilizing ability and “light switch” properties.

The “light switch” effect of [Ru(bpy)₂L]²⁺ with L = dppz, allox, me₂allox, pterin, amino, and diamino was measured previously at 10 μM Ru(II) complexes in the presence of 12-fold excess of CT DNA (per base pair) in 10 mM phosphate buffer with 50 mM NaCl pH 7.0.³⁷ While the absolute value of the effect cannot be compared with our data due to the difference in conditions, the general trend is very similar in all cases but for bpy-amino, which displays rather large fluorescence enhancement in the presence of CT DNA, but not in the presence of Tel22 GQ.

4. Conclusions

In this work we investigated interactions between eleven octahedral polypyridyl Ru(II) complexes and human telomeric DNA. We determined that Ru(II) bind Tel22 with 2:1 or 3:1 stoichiometry and binding constants on the order of 10⁵ -10⁶ M⁻¹. Bpy-me₂allox and phen-me₂allox stabilize human telomeric DNA to a greater extent and are more selective compared to the well-studied bpy-dppz. Additionally, bpy-diamino and bpy-aap exhibit excellent “light switch” properties, making them promising GQ fluorescence probes for further characterization and use. Changing the ancillary ligand from bpy to phen had little effect on the stabilization or selectivity but fine-tuned “light switch” properties and effect of ligands on Tel22 secondary structure. Changing buffer conditions from K⁺ to Na⁺ resulted in uniformly lower stabilization of Tel22 by Ru(II) complexes. Further experiments, such as crystallization, are on the way in our laboratory to determine the mode of ligand binding and role that ancillary and experimental ligands play in the binding process. Given that numerous groups report enantioselective interaction of Ru(II) complexes with GQs,^{17,25,33,34,36,59,60} it could be worthwhile to develop a synthetic route for enantiomerically pure Ru(II) complexes and elucidate binding preferences for different enantiomers. The results could guide the synthesis of GQ selective ligands, which exhibit optimal GQ binding properties, allowing for the design of novel anticancer therapies and fluorescence probes.

Highlights.

• [Ru(bpy)₂L]²⁺ and [Ru(phen)₂L]²⁺ with eight experimental ligands, L, were studied

• Ru(II) complexes were investigated for their interaction with human telomeric DNA

• Ru(II) complexes displayed excellent stabilizing ability and good selectivity in FRET

• Two complexes displayed strong “light switch” comparable to that of [Ru(bpy)2dppz]²⁺

• Binding constants were measured in spectroscopic titrations and via ESI-MS to be ~10⁵ M⁻¹

List of abbreviations:

bppp

(2-bromo-pyrido[2′,3′:5,6]pyrazino[2,3-f][1,10]phenanthroline)

bpy

2,2’ bipyridine

dppz

dipyrido-[3,2-a:2’,3’-c]phenazine

dppz-idzo

dipyrido-[3,2-a:2’,3’-c]phenazine-imidazolone

ebipcH₂

2,2′-(9-ethyl-9H-carbazole-3,6-diyl)bis(1H-imidazo[4,5-f][1,10]phenanthroline))

hbpibH₂

2,6-bis(1H-imidazo[4,5-f][1,10]phenanthrolin-2-yl)-4-methylphenol)

H₂bipt

2,5-bis[1,10]phenanthrolin[4,5-f]-(imidazol-2-yl)thiophene

IP

imidazole [4, 5-f] [1,10] phenanthroline

mbpibH2

1,3-bis(1H- imidazo[4,5-f][1,10]phenanthrolin-2-yl)benzene)

mitatp

5-methoxy-isatino[1,2-b]-1,4,8,9-tetraazatriphenylene

phen

1,10-phenanthroline

pppp

(2-phenylpyrido[2′,3′:5,6]pyrazino[2,3-f][1,10]phenanthroline)

PIP

2-phenylimidazo-[4, 5-f][1,10] phenanthroline

ptpn

3-(1,10-phenanthroline-2-yl)-as-triazino[5,6-f ]1,10-phenanthroline

tap

1,4,5,8-tetraazaphenanthrene

tatpp

tetraazatetrapyrido[3,2-a:2′3′-c:3″,2″-l:2″′,3″′-n]pentacene

tpphz

tetrapyrido[3,2-a:2′,3′-c:3″,2″-h:2‴,3‴-j]phenazine

3-tppp

12-(3-thienyl)- pyrido[2′,3′:5,6]pyrazino[2,3-f] [1,10]phenanthroline

Note, phen, bpy, tap, IP, and PIP are common auxiliary ligands