Initial DNA Interactions of the Binuclear Threading Intercalator Λ,Λ-[μbidppz(bipy)4Ru2]4+: An NMR Study with [d(CGCGAATTCGCG)]2

Binuclear polypyridine ruthenium compounds have been shown to slowly intercalate into DNA, following a fast initial binding on the DNA surface. For these compounds, intercalation requires threading of a bulky substituent, containing one RuII, through the DNA base-pair stack, and the accompanying DNA duplex distortions are much more severe than with intercalation of mononuclear compounds. Structural understanding of the process of intercalation may greatly gain from a characterisation of the initial interactions between binuclear RuII compounds and DNA. We report a structural NMR study on the binuclear RuII intercalator Λ,Λ-B (Λ,Λ-[μ-bidppz(bipy)4Ru2]4+; bidppz=11,11′-bis(dipyrido[3,2-a:2′,3′-c]phenazinyl, bipy = 2,2′-bipyridine) mixed with the palindromic DNA [d(CGCGAATTCGCG)]2. Threading of Λ,Λ-B depends on the presence and length of AT stretches in the DNA. Therefore, the latter was selected to promote initial binding, but due to the short stretch of AT base pairs, final intercalation is prevented. Structural calculations provide a model for the interaction: Λ,Λ-B is trapped in a well-defined surface-bound state consisting of an eccentric minor-groove binding. Most of the interaction enthalpy originates from electrostatic and van der Waals contacts, whereas intermolecular hydrogen bonds may help to define a unique position of Λ,Λ-B. Molecular dynamics simulations show that this minor-groove binding mode is stable on a nanosecond scale. To the best of our knowledge, this is the first structural study by NMR spectroscopy on a binuclear Ru compound bound to DNA. In the calculated structure, one of the positively charged Ru2+ moieties is near the central AATT region; this is favourable in view of potential intercalation as observed by optical methods for DNA with longer AT stretches. Circular dichroism (CD) spectroscopy suggests that a similar binding geometry is formed in mixtures of Λ,Λ-B with natural calf thymus DNA. The present minor-groove binding mode is proposed to represent the initial surface interactions of binuclear RuII compounds prior to intercalation into AT-rich DNA.

Section 1: Impact of pH, buffer on the DNA-?,? -B mixture Phosphate buffer yields some unwanted effects by showing affinity to ?,? -B (see Section 6). Therefore, a test spectrum was recorded in the absence of any buffer. This resulted in a pH of 5, and an unstable DNA duplex with weak or missing peaks for the imino protons of the guanines. Consequently, all other spectra were recorded with 20 mM phosphate buffer at pH 6 Figure S1. Two overlayed NOESY spectra with the imino proton region of 1:1 DNA-?,? -B (1mM) mixtures. The orange spectrum was recorded at pH 6.5 with the presence of 20 mM pH=6.5 sodium phosphate buffer; the magenta spectrum was recorded without any buffer (pH 5). The guanine imino cross peaks are much weaker or completely missing in the magenta spectrum. Section 2: Assignment of both symmetric and asymmetric DNA Table S1 lists the chemical shifts for both free DNA and DNA binding ?,? -B. For the latter, several protons on terminal base-pairs could not be assigned. All assignments were obtained in 20mM sodium phosphate buffer at 25 0 C. Missing H3', H4', H5' and H5" protons overlap with the water signal. Figure S2 displays the chemical shift differences of strands a and ß compared to the free DNA. The ß-strand shows larger differences than the a-strand. Table S1. Chemical shifts (ppm) of free d[(CGCGAATTCGCG)] 2 (on the lines with the nucleotide identification) and bound DNA (on the lines with "α" for the α-strand, "β" for the β-strand respectively).    (Table  S2). For bound ?,? -B, only four resonances can be identified; the remaining ones are covered by the DNA resonances. Two of these four show intermolecular NOEs: 9.14 ppm and 9.35 ppm (Fig. S3). The unambiguous assignment of these two resonances is explained below.  Figure S3 shows specific patterns of cross peaks for the two resonances in question (9.14 and 9.35 ppm): two NOESY cross peaks that coincide with corresponding TOCSY peaks. From the nuclei with chemical shifts larger than 8.1ppm for the free ?,? -B, H3D and H4D can be excluded because only one NOESY peak with a coinciding TOCSY peak is expected. Similarly, a ll H4, H6B, H6C and all H5 resonances should yield more than two NOESY/TOCSY peaks; overlap of peaks can be excluded due to the large shift difference for free ?,? -B (>1 ppm), and/or due to NOESY peak intensities (very weak peaks for distances of 2.5Å). This leaves the following protons as candidates for the four resonances >9 ppm: H4Ca, H4'Ca, H4Cb and H4'Cb. For all possible assignment combinations (six due to symmetry reasons), CYANA calculations were performed, yielding the results summarized in Table S3. The assignment H4'Cb to the resonance at 9.35 ppm and H4Ca to the resonance at 9.14 ppm results in negligible residual violations of constraints and a CYANA target function near zero.
MD simulations were performed according to the following protocol, using as a starting model a structure provided by CYANA calculations. The system (DNA and ruthenium(II)-complex, neutralized by sodium ions and solvated by 10 Å octahedron of explicit TIP3P waters [12]) was initially minimized by 1000 steps of steepest descent followed by 1000 steps of conjugate gradient, followed by fast heating (50 ps) from 0 to 300 K with the Langevin thermostat [13] temperature control scheme with collision frequency of 2 ps -1 , with harmonic restraints of 20 kcal/mol/Å 2 on the heavy atoms of the solutes, performed in constant volume. The restraints were then gradually reduced to zero in a series of equilibration runs of 100 ps each, at constant pressure (1 bar) and temperature (300 K) sustained using Langevin thermostat but with collision frequency of 1 ps -1 . The system was further equilibrated using steered MD (based on intermolecular NOE restraints) during 2 ns, followed by productive, unrestrained MD trajectories of 5 ns, both recorded with the following parameters. An integration time step of 2 fs was used and all bond lengths involving hydrogen atoms were constrained using SHAKE [14]. Long-range interactions were treated using the PME approach with a 10 Å direct space cut-off. Steered MD run was coupled to Replica Exchange MD for better sampling of the conformational space. Taking into account the properties of DNA as well as that all simulations were performed with explicit solvent model, the temperature span was only 9 degrees (F), starting from 300 to 309 K, resulting i n total 10 replicas for each of the steered MD runs. The fully unrestrained MD trajectories of 5 ns length recorded for the ten replica were used to represent the structure of the DNA-?,? -B complex.

Section 5: Other interaction modes
In contrast to natural long DNA, our system consists of a short DNA fragment. The present experimental observations suggest that the terminal DNA base pairs of the symmetric DNA (i. e. of a binding mode different from the minor groove binding state of ?,? -B) interact with the ?,? -B in fast exchange process (Fig. S5). This may be explained by a hydrophobic interaction due to the stacking of the aromatic ring system of the ?,? -B with the terminal DNA bases.
When varying the temperature, it was also discovered that some resonances of ?,? -B were not temperature dependent, while others were reduced with increasing temperature, together with those of the melting DNA (melting temperature ca. 50°C; Fig. S6). Comparison with a nonbuffered sample showed that the temperature-independent peaks are correlated to the presence of phosphate buffer, suggesting a buffer-mediated aggregation of ?,? -B and oligomer, which observation was not investigated further (buffered and non-buffered NOESY spectra are compared for ?,? -B shown in Fig. S6a).
Comparison between NOESY spectra of DNA-?,? -B with and without buffer. The presence of buffer results in additional peaks that are temperature independent; some of these are shown in the red boxes in Fig. S6. A few more are found between 7 and 9 ppm; these peaks make no NOE contacts except to the resonances shown in the red boxes of