Formation of silver-metalated base pairsIn our investigation, we employed the self-complementary oligonucleotide sequence 5′-d(YYX CTC YXY TCC)−3′ (ODN1, forms duplex I), akin to the one utilized in previous research that reported the consecutive formation of silver-metallized base pairs within the DNA double helix14. However, in this particular study, we introduced a modification by replacing the canonical A and G bases with 7-deazaadenine (X) and 7-deazaguanine (Y) bases, respectively. Additionally, we substituted the C base in the ninth position with the Y base to create a self-complementary C-Y base pair, in contrast to the C-C mismatch pair in the previous research. To facilitate a comparison with results obtained using ODN1, we employed the equivalent canonical sequence 5′-d(GGA CTC GAG TCC)−3′ as a reference (ODN2, forms duplex II).We conducted initial CD titration experiments to investigate conformational changes in duplexes I and II upon interaction with AgI ions. Both free duplexes displayed distinct CD features, with a maximum 280 nm and a minimum at 245 nm (Fig. 1). For duplex I, the addition of AgI resulted in a gradual reduction in ellipticity and flattening of the 280 nm band, with changes stabilizing upon adding 1 equivalent of AgI per base pair (Ag/bp), thus indicating a saturation point (Fig. 1a). In contrast, duplex II showed a decrease in ellipticity and a complete inversion of the entire CD spectrum, manifesting a pronounced descending band at 280 nm, also stabilizing soon after adding 1Ag/bp (Fig. 1b). The results indicate that both duplexes interact with AgI ions but leading to different conformational changes. A comparative analysis of the CD spectra for duplex I and duplex I-Ag at 1Ag/bp suggests that these structures may share structural similarities. This conclusion arises from the observation that the CD profile of duplex I-Ag exhibits reduced ellipticity while maintaining a very similar profile to that of duplex I, with both spectra exhibiting a decline around 245 nm and an ascent around 280 nm, indicating closely superimposable profiles (Fig. 1c). These similarities highlight the close conformational relationship between duplex I and I-Ag. This assertion gains further support when examining the differences between canonical duplex II and II-Ag at the same point. In this case, the CD curves exhibited complete inversion with respect to each other, resembling mirror images (Fig. 1d), indicating that the II-Ag adopted a completely new organization. According to previous studies on canonical sequences, this new organization arises through the formation of non-canonical metallated base pairs, where Ag-binding to purine N7-positions plays a key role13,21,22. However, duplex I-Ag cannot form metallated base pairs involving the purines N7-atom, implying that its metalation process must involve the formation of alternative silver-metallized base pairs.Fig. 1: CD spectra in the absence and presence of AgI ions.a, b CD curves registered for duplex I and II, respectively, upon adding a controlled amount of AgI. c, d Ellipticity variation at the bands centered at 250 and 280 nm for duplex I and II, respectively. Conditions: 2 µM duplex, 5 mM MOPS pH 8.5, 100 mM NaClO4, AgI equivalent (equiv.) per base pair (bp) indicated in the inset.UV-variable temperature measurements were conducted to investigate the denaturation behavior of duplexes I and II in the presence and absence of AgI ions. In the absence of AgI ions, both duplex I and II exhibited sigmoidal melting curves with a melting temperature of 42 and 51 °C, respectively (Supplementary Fig. 1). The variation between them is attributable to the weaker hydrogen bonds formed by 7-deazapurine nucleobases in duplex I. The addition of AgI ions caused a gradual flattening of the melting curves during the titration experiment until they disappeared in the presence of 1Ag/bp. The primary conclusion derived from these findings is the formation of metallated species, where metal ions replace canonical hydrogen bonds, increasing the duplexes’ thermal stability. This enhanced stability surpasses the experimental measurement range in the presence of 1Ag/bp. It is important to exercise caution if attempting to draw conclusions from the melting curves recorded before reaching the saturation point at 1Ag/bp. Before saturation and throughout the titration experiment, fully metallized duplexes coexist with mixed-partially or non-metallized duplexes (see NMR data discussion below), complicating the interpretation of the data. Therefore, we did not attempt to calculate the melting temperature variations, and our subsequent studies will be conducted at the saturation point to ensure more precise and reliable results.To gain deeper insights into the denaturing process, we conducted CD-melting experiments for duplex I and II in the absence and presence of 1.1 equivalent of AgI per base pair, slightly exceeding the saturation point. In the absence of metal ions, the CD curves displayed a reversible behavior during the melting and alignment process, aligning with a denaturing reversible process (Supplementary Fig. 2). The ellipticity intensity variances of the CD curves at 20 °C before and after heating is attributed to condensation events occurring on the cuvette’s surface, slightly altering the concentration without affecting CD curves shape. A plot of the CD variation with temperature showed the expected enhanced thermal stability for duplex II compared to duplex I, mirroring the results of the UV-melting experiment (Supplementary Fig. 2d, 2d’). However, in the presence of 1 equiv. of AgI ions, different outcomes were observed for each duplex. Duplex I-Ag experienced reversible changes during the heating-cooling process, whereas duplex II-Ag underwent a non-reversible conformational change. This difference is particularly evident when comparing the CD curves at 90 and 20 °C (after cooling), as I-Ag displayed markedly different CD profiles at the two temperatures, whereas II-Ag displayed similar curves (Supplementary Fig. 3d, 3d’). The CD and UV experiments clearly demonstrated that the binding of AgI to duplex I and II leads to more stable species. However, the resulting conformational changes were markedly different for each duplex. Since it is well-known that the purine N7-position plays a crucial role in stabilizing silver-metallized canonical duplexes, the absence of 7-deazapurines in duplex II could promote the formation of silver-modified Watson–Crick base pairs.The electrospray ionization mass spectrometry (ESI-MS) experiment unequivocally provided evidence for the formation of I-Ag complex (Supplementary Fig. 4). The deconvoluted mass spectrum for the reaction between duplex I and AgI ions showed the peaks for duplex I (7281.0 g/mol) with 7-12 AgI ions (106.9 a.u) bound, demonstrating the formation of I-Ag complexes, including the saturated system with twelve metal ions (Supplementary Table 1).SAXS and ASAXS studies of silver-DNA duplexesSmall-angle X-ray scattering (SAXS) and anomalous small-angle X-ray scattering (ASAXS) provided structural information about duplexes I and II in the absence and presence of AgI ions. The scattering data for duplex I (at 0.7 mM) is consistent with the formation of a double helix structure, exhibiting an excellent match with the simulated scattering, up to q value of 0.2 Å−1 (Supplementary Fig. 5). Above this q value, the experimental curve starts to incorporate solvent scattering, primarily due to the relatively weak scattering of the low-concentration and light-atom species. Nonetheless, the match is convincing within the entire Guinier region (q ~0.01 to 0.2 Å−1), confirming the expected size and shape of the double helix (vide infra). Adding 1.1 equiv. AgI per base pair to duplex I leads to only small changes in the scattering data, suggesting the incorporation of AgI does not lead to significant conformal changes. We performed a PDDF (pair distance distribution function) analysis of the scattering data, which revealed the differences and similarities of the scattering, with and without AgI ions (Fig. 2 and Supplementary Figs. 6, 7). The PDDF is a probability distribution of scattering vectors through the dissolved species, obtained by a Fourier transform of the reciprocal space scattering data. The maximum diameter (r) where probability goes to 0 describes the maximum diameter of the scattering species; 37.4 Å for duplex I and 36.1 Å for I-Ag, precisely equivalent to the length of the duplex I. A slightly smaller Rg (radius of gyration, a shape independent, root mean square average of the distance of the electrons from the center of the particle) is reported for I-Ag than for duplex I (Supplementary Figs. 6, 7). This is consistent with the incorporation of the AgI ions in the center of the helix, more thoroughly evaluated with ASAXS, discussed below. The addition of AgI also increases the P(r), also due to the presence of heavy atoms in the center, and a low shoulder to the right appears (Fig. 2). The low shoulder indicates a bimodal electron density, with lower scattering intensity from the shell (the DNA surrounding the Ag-chain).Fig. 2: Scattering data from benchtop SAXS (Cu-Kα radiation, 8.04 keV) of duplex I, in the absence and presence of 1.1 equivalent Ag/bp (I-Ag).a scattering plot and b pair distance distribution function (PDDF) analysis.To determine the precise location of AgI within both the I-Ag and II-Ag complexes, we conducted ASAXS measurements at a synchrotron source. This involved collecting SAXS data using X-rays at 20 different energy levels, each situated just 1 keV below the silver K-edge at 25.514 keV (Supplementary Figs. 8–24, Supplementary Tables 2,3 for I-Ag, and Table 4 for II-Ag). For this investigation, we performed detailed modeling of various scattering components based on the solid cylinder model for these complexes (Fig. 3a). A comparative analysis of the data revealed that both duplexes contained AgI within the cylindrical region, meaning the double helices encapsulate them. However, I-Ag saturates at approximately 1 equivalent and II-Ag does not reach saturation within our experimental range of 3 equivalents (Fig. 3b). The length of I-Ag structure remained relatively consistent at around 40 Å. In contrast, II-Ag structure substantially increased, indicative of polymerization events (Fig. 3c and Supplementary Table 4). Overall, the modeling data indicated that AgI interacts with the interior regions of the double helices in both I-Ag and II-Ag complexes but yielding different structural outcomes, as also revealed by the herein described CD experiments.Fig. 3: ASAXS modeling and analysis using a solid cylinder model of I-Ag and II-Ag complexes.a Solid cylinder model of the Ag-DNA complexes. The AgI ions can bind the base pairs or be free within the cylindrical regions. Metrical information derived from the Ag-DNA complexes modeling, namely: b the number of AgI per base pairs of the complex (Ag/bp); c the lengths (H). The data were represented using red symbols with error bars for I-Ag complexes and blue symbols with error bars for II-Ag complexes, along with auxiliary lines to highlight the trend of the points.High-resolution NMR studies of silver-DNA duplexesNMR spectroscopy measurements were employed to further study the interaction of AgI ions with duplex I and II. For this, a solution of duplex I was titrated with AgI in the range between 0 and 2.5Ag/bp. 1H NMR spectra were recorded at various points during the titration to monitor the interaction (Fig. 4a). Based on signals corresponding to hydrogen-bonded imino (δ 12–15 ppm) and amino (δ 8–9 ppm) protons, the spectra indicate a double-stranded structure for the oligonucleotide at the start of titration. After adding AgI ions, the signals characteristic for the starting duplex structure began to decrease. Then, several new signals appeared in the spectra, indicating an interaction between duplex I and AgI ions. Finally, at about 1.5 equivalents, the signals converged into a single set, which stayed unchanged until the endpoint of the titration (Fig. 4a, b). The presence of four signals originating from amino protons indicates the formation of hydrogen bonds within the final structure. The number of non-exchangeable signals suits a symmetric antiparallel double-stranded structure similar to the one adopted natively by duplex I. Notably, the imino proton signals are absent in line with the expected position of the coordinated AgI ions in place of T-H3 and Y-H1 protons in the modified base pairs. The double-stranded nature of the final structure was further confirmed by DOSY NMR measurements, which gave similar translational diffusion coefficients (DT) of 1.15 and 1.16 · 10−10 m2 s−1 for the AgI free and bound structures, respectively, under comparable conditions. This is also similar to the DT measured for the canonical duplex II analog at 1.21 · 10−10 m2s−1, indicating a comparable hydrodynamic radius for all three structures.Fig. 4: NMR spectra of the I/I-Ag system.a Imino, aromatic, and methyl regions of 1H NMR spectra of the titration, AgI concentration presented in equivalents per base pair. b Normalized signal intensities of T10 H7 proton (methyl group) in I and I-Ag structures in 1H NMR spectra of the titration. c Aromatic-anomeric region of NOESY spectrum of I-Ag at 25 °C showing the characteristic sequence walk. d H1-H2’/H2” DQF-COSY spectra of I and I-Ag showing patterns characteristic of S sugar puckers.The appearance of additional signals in 1H NMR spectra during the titration with AgI indicates that the incorporation of silver proceeds through several partially metallated intermediates with silver ions bound at various positions within the duplex. The presence of these intermediates is consistent with minor structural rearrangements associated with the incorporation of silver ions that permit the coexistence of silver-mediated and canonical base pairs. On the other hand, it has been observed previously that single-step transitions are typical where larger structure rearrangements, like base bulging or strand slipping, are required13,14.The entire process described for duplex I is in marked contrast to titration of the standard analog duplex II, where the number of signals in 1H NMR spectra increased throughout the course of titration and did not converge to a single well-defined structure by ten equivalents AgI ions per base pair (Supplementary Fig. 28). These results are consistent with previous observations that AgI ions tend to promote the formation of non-canonical base pairs in DNA, such as C-AgI-C, T-AgI-T, or G-AgI-G homodimers, while the standard metallized Watson–Crick A-T and G-C are disfavored, leading to loss of base pairing pattern, emergence of an ensemble of structures with degenerate folding energies, and ultimately the aggregation of the refolded oligonucleotides13,14,21. This is reflected by the progressive line broadening and signal loss in 1H NMR spectra with increasing concentrations of silver ions in the solution.NMR reveals the structural characteristics of duplex Ag-7dDNAA NOESY walk was completed in the aromatic-anomeric and the aromatic-H2’/H2” region for the entire length of the I-Ag complex (Fig. 4b), which is consistent with a right-handed double-helical structure with all χ torsion angles in anti-orientation. Cross peaks involving protons of the Y7 residue were weaker than expected, indicating a possible distortion, or increased structural flexibility in the I-Ag complex at the C6-Y7 step.H1’-H2’ and H1’-H2” cross-peak patterns in DQF-COSY spectra indicate South sugar pucker for most residues, except for C4, T10, C11, and C12, where higher 3JH1’H2” values indicate a significant fraction of North sugar pucker for the respective ribose moieties. The same observation is true for duplex I and I-Ag and is a known feature of pyrimidine residues in general (Fig. 4b, c).31P NMR chemical shifts for duplex I are similar to the canonical analog II in the absence of AgI ions. After the formation of the complex, the chemical shift of X3P, X8P and Y9P shift upfield by 0.5 to 1.1 ppm (Supplementary Fig. 29). This observation indicates a backbone rearrangement from the canonical BI type toward BII type. On the other hand, patterns in HP-COSY spectra mainly indicate similar H3’ and H5’/H5” J coupling values, which imply similar γ and β torsion angles for duplex I in the presence and absence of AgI ions, except for the C6-Y7 step, where the C6H3’ to Y7P cross-peak is changed upon addition of the silver ions. This could be due to a backbone distortion at this step or due to the signal broadening observed for the Y7 residue.We applied NMR restraints derived from NOESY and DQF-COSY spectra to construct structural models of the I and I-Ag duplexes in AMBER. The silver-modified model was constructed by replacing imino protons within Watson–Crick base pairs with silver ions. The appropriate parameters for the metal ion binding terms were determined based on DFT calculations in Gaussian software, and parameters for the metal-mediated base pairs were obtained with the help of MCPB software23. A total of 250 NOE-derived distance restraints were applied for the native structure of duplex I and 320 for the silver-mediated duplex I-Ag structure. Additionally, sugar pucker was restrained to 140–180° range for residues 1-3 and 5-9 in both structures and backbone torsions were restrained in duplex I based on 31P NMR chemical shifts and HP-COSY cross-peaks.The NMR structure shows duplex I-Ag forms a right-handed helical structure resembling the B-type DNA helix (Fig. 5a), with an average twist of 34° and an average rise of 3.3 Å (compared to 33° and 3.3 Å in the structure of duplex I, Supplementary Table 5). The main difference compared to the canonical structure is the reorientation of the bases required to accommodate the silver ions within the base pairs (Fig. 5b). The average base opening parameter for the silver-modified duplex is −35°, compared to values in the structure of duplex I without the silver ions (the latter is also the same as in a model B-type duplex). The propeller twist of the base pairs is also increased to an average value of −20°, compared to −8° in the model of duplex I or −11° for B-DNA type structures in general (Fig. 5c). The change in base pair geometry forces the breaking of one of the hydrogen bonds in the C-Y base pair, leaving the base pair connected with one hydrogen bond and the silver coordination bond in place of the original three hydrogen bonds. Our model suggests that the 2-amino groups of Y residues, which are released from hydrogen bonding in C-Y base pairs, tend to form inter-base pair hydrogen bonds when a purine residue precedes the residue. Four such pairs are observed in our model: Y2-C12*, T5-Y9*, Y9-T5*, and C12-Y2* (asterisks denote complementary strand residues). Inter-planar hydrogen bonding has been observed in silver ion-mediated DNA structures before and has been proposed as an important factor for structure stabilization14. The differences observed between C-Y and T-X base pairs, and between different base steps, could have implications for the stability of various sequences of the 7dDNA system. Likewise, this structural detail could factor into the reasons why A-T base pairs have not been observed in native sequence contexts to date, while G-C pairs have.Fig. 5: NMR structure of I-Ag complex and structural parameters for the duplex I and I-Ag complex.a Side and top views of the lowest energy structure following NMR restrained MD. b Selected structural parameters for I and I-Ag. While most structural parameters remain similar, the base pairs are perturbed, as reflected in the propeller twist and base pair opening. Typical values for B-DNA and A-DNA type structures are indicated in the graphs, where applicable. c Base pair geometry for the metal-modified base pairs in I-Ag complex and duplex I.The silver ions within duplex I-Ag form a helical arrangement mirroring the pattern of base pairs. The overall helix retains a narrow central cavity similar to the native B-type helix, and the silver ions thread among the DNA bases, appearing obscured by the nucleobases when viewed from the top (Fig. 5a). The average distance between consecutive AgI ions along the structure is 3.8 ± 0.3 Å, exceeding the sum of the ionic radii (3.44 Å), suggesting a general absence of argentophilic interactions24. However, the distances between silver ions (Ag-Ag) vary depending on the sequence, reducing to 3.27 Å at steps 3 and 9. This proximity seems primarily influenced by base pair shift, slide and atom distance within the base pairs and reflects the leading role of interactions between the DNA nucleotides in determining the position of silver ions. This observation suggests the possibility of directing potential electronic properties of the assemblies throughout DNA sequence. Notably, the AgI arrangement differs from that seen in previously published X-ray structures of silver-mediated structures of canonical DNA13,14. This could reflect differences in the relative contribution of argentophilic interactions to the structures due to differences in nucleotide chemistry of applied conditions. A previously published solution structure of a DNA duplex with three consecutive silver-mediated base pairs did not show significant argentophilic interactions, similar to our findings here25.DFT calculations verified the structure of I-Ag complexTo study the positioning of the AgI ions more precisely, we performed DFT-based (PBEh-3c) calculations on the I-Ag complex in the presence of sodium counterions and water molecules. To our knowledge, no comprehensive DFT study has been conducted on an entire metal-DNA system with twelve consecutive metal-modified base pairs. Previous theoretical DFT (M06-2X) investigations in this area have been limited to fragments of up to six units in size21.Notably, the DFT-optimized geometry of the I-Ag complex closely resembles the structure derived from the NMR data (Fig. 6 and Supplementary Table 5). These structures align in several critical aspects, suggesting a robust agreement between the two structure determination methods. While subtle differences in specific parameters exist, these distinctions do not reduce the broader concordance between the NMR and DFT-optimized structure. For instance, the twist angle is approximately 34° in the NMR and DFT structures, showcasing a high level of agreement. Furthermore, parameters like the minor groove width, slide, rise, and tilt exhibit noteworthy alignment.Fig. 6: NMR vs. DFT Structural Comparison.Structures for I-Ag complex obtained by different methodologies: a NMR spectroscopy and b DFT calculations. c Superposition of the NMR (blue) and DFT (orange) structures.In addition, the parameters of buckle, opening, and major groove width appear to show minor variations considering the standard errors. However, the propeller angle (−8° ± 9) appears to be more distinct from that revealed in the NMR (−21° ± 4). It is essential to recognize that these differences are subtle and fall well within the range of experimental computational approximations. Thus, they do not detract from the overarching consensus on the core helical NMR structure of I-Ag.Similar to the NMR-derived model, the Watson–Crick pairing is preserved, with silver ions substituting a hydrogen bond with a linear coordination bond. These ions align according to the helix turn dictated by the base pairs, with an average Ag···Ag distance of 3.6 ± 0.3 Å, slightly smaller than observed in the NMR model (Supplementary Table 6). This situation excludes the formation of contiguous argentophilic interactions, confining them to specific regions throughout the sequence, particularly at steps 2, 3, and 9, in close accordance with the NMR data. This stands in contrast to the observations made in crystal structures of Ag-DNA systems featuring non-canonical base pairings, where a continuous array of argentophilic interactions occurs13,14. However, in some sections of the double strand, certain silver ions are observed with distances less than 3.44 Å, suggesting the potential existence of argentophilic interactions.
