Measurement of flexibility change of dsDNA using AFM method
Tracing and image analysis
The obtained AFM images were analyzed by a custom unbiased program written with MATLAB. Briefly, the original image without background slope was converted into a binary image and randomly selected a pixel in each dsDNA and subsequently searched the extensions from the selected pixel to both sides till the ends. The distance between two pixels was determined by their position coordinates \(\left( {\Delta {\text{l}} = \sqrt {(dx)^{2} + (dy)^{2} } } \right)\), and the contour length (the actual length of the polymeric chain from one end to another end along the polymer skeleton) of each dsDNA was obtained by summation of the distance \((\varSigma \Delta {\text{l)}}\) of all pixels on the DNA skeleton line. The process of dealing with the raw image is shown in Fig. 2.
Morphology and binding property analysis
In order to gain deeper insight into the influence of Ru(bpy)2dppz2+ intercalators on the conformational properties of dsDNA, it is critical to imaging individual DNA molecule on a mica surface under different ratios of Ru(bpy)2dppz2+ to DNA base pairs (Ru/DNAbps). Quantitative AFM images of dsDNA–ruthenium complexes on the mica surface were successfully and systematically captured in our study. In detail, the DNA fragments appeared like squiggly lines because of the resolution limit of the AFM (Fig. 3). Neither grooves nor helical twist could be resolved, but the overall shape could be seen and the length could be measured with reasonable accuracy. With increasing the ratio of Ru/DNAbps from 0.1 to 1, it was clearly observed that the contour length increased distinctly, which confirmed the intercalation of Ru(bpy)2dppz2+ and induced the local changes of dsDNA (Additional file 1). In Fig. 3A–D, dsDNA molecules showed very few crossings, whereas in Fig. 3E, F, dsDNA molecules were imaged in excess of Ru(bpy)2dppz2+ (ratio 2 and 3 of Ru/DNAbps) resulting in the condensation of many crossings. From these images, we can clearly observe the conformational change of dsDNA molecules with different ruthenium concentration. At saturated concentration, the dsDNA molecules appeared to be more condensed than compared with the low ratio of Ru/DNAbps, indicating that the dsDNA molecules was more flexible at high intercalator concentration than at low intercalator concentration which was consistent with the results of flexibility analysis. For determining the binding property between them, we used the converted equation of McGhee and von Hippel to obtain a relationship between the fractional extension L/L0 and the ratio of Ru/DNAbps (the plot is shown in Fig. 4) [30]. The obtained relation can be described as Eq. 1:
$$\frac{{C_{Ru} }}{{C_{{DNA{\text{bps}}}} }} = \nu + \frac{\nu }{{K_{a} (1 - p\nu )C_{{DNA{\text{bps}}}} }} \left [\frac{{1 - \left( {p - 1} \right)\nu }}{1 - p\nu }\right]^{p - 1}$$
(1)
$$\nu = \frac{{L - L_{0} }}{{L_{0} }}\frac{{\delta_{bp} }}{{\delta_{Ru} }}$$
(2)
where CRu and CDNAbps are the concentration of total ruthenium and DNAbps, respectively. ν is the ratio between the bound intercalator and the total concentration of DNAbps, p is binding site size i.e., the number of base pair sites occluded by one bound Ru(bpy)2dppz2+ and Ka is the affinity constant. The relation between ν and contour length can be described with Eq. 2, where L0 is contour length without ruthenium intercalation, L is the contour length after its intercalation, \(\delta_{bp}\) is the rise parameter per DNAbps of 0.34 nm and \(\delta_{Ru}\) is the DNA elongation per bound one Ru(bpy)2dppz2+ of 0.5 nm.
It was feasible to derive the relative extension of DNA double stand as a function of ruthenium concentration per base pair concentration from the contour length measurement, and the results are plotted in Fig. 4. Ka of ruthenium complex and dsDNA interactions was estimated using the comparison between the experimental data and the equation converted from the theory developed by McGhee and von Hippel in the case of non-cooperative ligand binding [30]. Figure 4 shows the comparison between our experimental data and theoretical model. The best fitting data was obtained for the p and Ka values of 2.87 bp and 5.9 * 107 M−1, respectively. These values are in good agreement with the values obtained by Williams et al. which was measured by stretching experiments with classical intercalator. In addition, as indicated by the plot of Fig. 4, the measured contour length of DNA increased with increasing Ru/DNAbps ratio. For the ratios of Ru/DNAbps is over 1, the observed contour length became saturated and the plateau indicated the 50% increase of the relative length. We speculated that ruthenium intercalation induced a change in the local structure of DNA helix resulting in a lengthening of the DNA strand. Which was consistent with previous results of classical intercalators [29, 31]. In cells, many DNA-distorting proteins used side chain intercalation to distort the DNA backbone which plays important roles for processing information in DNA and organizing chromosome DNA [32].
The flexibility of dsDNA analysis
In this part, we mainly presented the physical property of probed DNA molecules and how the binding of Ru(bpy)2dppz2+ affects the extension and the flexibility of DNA molecules. The flexibility of DNA can be characterized by its persistence length. Normally, there are two major methods to estimate the persistence length of adsorbed macro-molecules on the mica surface. The first method is based on the measurements of the end-to-end distance and the contour length which is very reliable when used on a large number of molecules [33]. The second method is based on the measurements of the angle between two small segments separated by a certain distance. This method is very sensitive to the local bending of the DNA molecules and does not require measurements of hundreds of molecules. However, AFM tip can easily lead to the local bending along the scanning direction. Herein, the first method was employed to measure the persistence length of dsDNA.
The straightforward approach is to consider the root mean square of end-to-end distance (⟨R2⟩) of an ensemble of identical polymers. Assuming the DNA to be at thermal equilibrium state, the 2D (two dimension) worm-like chain (WLC) model provides a relationship between the mean end to-end distance and contour length, and it can be described by the following equation [27]:
$$\left\langle {R^{2} } \right\rangle = 4L_{P} L_{C} \left( {1 - \frac{{2L_{P} }}{{L_{C} }} \left (1 - {\text{e}}^{{ - \frac{{{\text{L}}_{\text{C}} }}{{ 2L_{P} }}}} \right)} \right)$$
(3)
where LP is the persistence length and LC is the contour length. This nonlinear relationship needs to be inverted for LP after measuring the mean-square end-to-end distance of DNA chain and the contour length, and the persistence length can be obtained by the Eq. 3. For longer chains ⟨R2⟩ ≈ 4LPLC, at least fifty DNA fragments were analyzed under every ratio in our study.
The effect of the level of ruthenium intercalation with DNA molecule on the Lp and ⟨R2⟩ were plotted in Figs. 5, 6. The persistence length decreased with the fractional extension (L/L0) which was consistent with the recent results obtained by Maaloum et al. [25]. The measured value of the persistence length of dsDNA in Ru(bpy)2dppz2+ free solution was Lp = 54 ± 1.3 nm. This value was in good agreement with the expected value obtained from single molecule stretching experiments under physiological conditions [22]. A large decrease in the persistence length was observed on increasing the fraction extension L/L0 from 1.03 to 1.31 (corresponding to the ratio of Ru/bps from 0.1 to 0.2). The obvious change of the persistence length was mainly induced by the ruthenium intercalation, and the decrease of persistence length indicated the increase in dsDNA flexibility. But the plot of Fig. 6 suggested that the value of ⟨R2⟩ decreased with the increasing the single DNA molecule length under different ruthenium concentration. It would be speculated that the DNA helix might be bent due to ruthenium intercalation. However, by combing the increase of relative elongation of dsDNA (Fig. 4) when the ruthenium intercalation reached saturation, it would be concluded that ruthenium intercalation induced the local deformation of DNA duplex. Specifically, the DNA helix might detwisting along the axis of DNA helix at a certain degree, but we could not find the direct evidence due to limit of AFM resolution [9].
Finally, by comparing the best fitted results with the equation of McGhee and Von Hippel, the value of p was found to be 2.87, which simply meant that the saturated intercalation of Ru(bpy)2dppz2+ occurred after every 3 base pairs. In other words, the persistence length would change in a small range when the intercalation reached saturation. This was in good agreement with the persistence length results in Ru/bps solution containing ratio of Ru(bpy)2dppz2+ from 1 to 3, where the ruthenium intercalation reached saturation and the persistence length decreased to approximately 20 nm.
Thermal dynamic stability of dsDNA measurement with HRM method under different ratios of Ru/DNAbps
In order to investigate the thermal dynamic stability of dsDNA after ruthenium intercalation, the high resolution melting method was employed to detect the change of melting point of dsDNA as ruthenium intercalated into the duplex DNA which is a robust technique for detecting the thermal dynamic stability of dsDNA even if there is a tiny change in the structure of DNA helix [34, 35]. The melting profiles of dsDNA are shown in Fig. 7 (left) under different ratios of Ru/DNAbps. The initial relative fluorescence intensity of 80 bp duplexes was 23,000 which decreased on increasing the ratio of Ru/DNAbps because more and more ruthenium complex intercalated into DNA helix which could not be stained by the EvaGreen. The melting points of dsDNA were obtained from the differential curves and are shown in Fig. 8. The melting point of dsDNA without ruthenium compound was found to be 84.7 °C. The Tm values significantly increased from 84.7 to 89.9 °C with increasing the ratio of Ru/DNAbps, which could be interpreted that the thermal dynamic stability of DNA strand increased due to stacking of planar dppz ligand with DNA base pairs, but that may not disturb the integral structure of double helix. In other words, the increase of Tm was due to the increase of the interaction area between consecutive base pairs, which resulted in the increase of free energy [22].