New N4-Donor Ligands as Supramolecular Guests for DNA and RNA: Synthesis, Structural Characterization, In Silico, Spectrophotometric and Antimicrobial Studies

The present work reports the synthesis of new N4-donor compounds carrying p-xylyl spacers in their structure. Different Schiff base aliphatic N-donors were obtained synthetically and subsequently evaluated for their ability to interact with two models of nucleic acids: calf-thymus DNA (CT-DNA) and the RNA from yeast Saccharomyces cerevisiae (herein simply indicated as RNA). In more detail, by condensing p-xylylenediamine and a series of aldehydes, we obtained the following Schiff base ligands: 2-thiazolecarboxaldehyde (L1), pyridine-2-carboxaldehyde (L2), 5-methylisoxazole-3-carboxaldehyde (L3), 1-methyl-2-imidazolecarboxaldehyde (L4), and quinoline-2-carboxaldehyde (L5). The structural characterisation of the ligands L1-L5 (X-ray, 1H NMR, 13C NMR, elemental analysis) and of the coordination polymers {[CuL1]PF6}n (herein referred to as Polymer1) and {[AgL1]BF4}n, (herein referred to as Polymer2, X-ray, 1H NMR, ESI-MS) is herein described in detail. The single crystal X-ray structures of complexes Polymer1 and Polymer2 were also investigated, leading to the description of one-dimensional coordination polymers. The spectroscopic and in silico evaluation of the most promising compounds as DNA and RNA binders, as well as the study of the influence of the 1D supramolecular polymers Polymer1 and Polymer2 on the proliferation of Escherichia coli bacteria, were performed in view of their nucleic acid-modulating and antimicrobial applications. Spectroscopic measurements (UV–Vis) combined with molecular docking calculations suggest that the thiazolecarboxaldehyde derivative L1 is able to bind CT-DNA with a mechanism different from intercalation involving the thiazole ring in the molecular recognition and shows a binding affinity with DNA higher than RNA. Finally, Polymer2 was shown to slow down the proliferation of bacteria much more effectively than the free Ag(I) salt.


Introduction
As all living organisms evolve, it is natural for pathogenic bacteria to develop drug resistance since antimicrobials are commonly used to eradicate them, making them able to counteract by developing resistant strains [1]. In this regard, the excessive and inappropriate use of antibiotics in medicine, animal husbandry, and agriculture accelerates the phenomenon [2]. Managing the increasing number of infections caused by multidrugresistant bacteria is a serious challenge for modern medicine [1]. It is estimated that in 2019 alone, about 5 million deaths were associated with bacterial antimicrobial resistance [3]. In addition, there are several issues concerning antibiotics already in use, such as their limited

Design and Synthesis
The five ligands L1-L5 were designed so as to possess two separate N2-donor binding moieties able to coordinate the tetrahedral Ag(I) and Cu(I) ions for their potential use as antibacterial agents. The linker between the two coordinating sites in all ligands was the p-xylyl, while the capping units varied-we used 2-thiophenyl (L1), 2-pyridinyl (L2), 5-methylisoxazol-3-yl (L3), 1-methylimidazol-2-yl (L4), and quinoline-2-yl (L5, Scheme 1). To gain some insights into the drug-likeness of the proposed molecules, we performed preliminary in silico studies, whose results are summarised in Table 1. A good bioavailability could be achieved with an appropriate balance between solubility and partitioning properties. Thus, in order to achieve good oral drugs, we subjected our compounds to the prediction of the Lipinski "Rule of Five" [29] and other properties for filtering compounds for subsequent synthesis and antimicrobial screening. The most important predictors for the good bioavailability of potential therapeutic agents were the ones given by the abovementioned Lipinski "Rule of Five." The rule states that good candidates should have a logP ≤ 5, a molecular weight ≤ 500 g/mol, no more than 10 hydrogen bond acceptors, and a maximum of 5 hydrogen bond donors. The ligands L1-L4 fulfil the Lipinski's rule [29]. They all also fulfil the rules given by Ghose, Veber, Egan, and Muegge [30][31][32][33]. The only exception was L5, which had slightly too high cLogP, according to Lipinski, WLogP and molar refractivity, according to Ghose, and XLogP3, according to Muegge. The calculated polar surface area (tPSA) values of ligands L1-L5 varied from 50.50 to 106.98 Å3, so they were potentially able to cross the membranes.
Molecules 2023, 28, x FOR PEER REVIEW 3 of 21 (Polymer1) and Ag(I) (Polymer2) ions. The bioactivity against a bacterial strain of Escherichia coli of the polymers, appropriately Ag(I) and Cu(I) salts, and the ligand itself, were also assessed.

Design and Synthesis
The five ligands L1-L5 were designed so as to possess two separate N2-donor binding moieties able to coordinate the tetrahedral Ag(I) and Cu(I) ions for their potential use as antibacterial agents. The linker between the two coordinating sites in all ligands was the p-xylyl, while the capping units varied-we used 2-thiophenyl (L1), 2-pyridinyl (L2), 5-methylisoxazol-3-yl (L3), 1-methylimidazol-2-yl (L4), and quinoline-2-yl (L5, Scheme 1). To gain some insights into the drug-likeness of the proposed molecules, we performed preliminary in silico studies, whose results are summarised in Table 1. A good bioavailability could be achieved with an appropriate balance between solubility and partitioning properties. Thus, in order to achieve good oral drugs, we subjected our compounds to the prediction of the Lipinski "Rule of Five" [29] and other properties for filtering compounds for subsequent synthesis and antimicrobial screening. The most important predictors for the good bioavailability of potential therapeutic agents were the ones given by the above-mentioned Lipinski "Rule of Five." The rule states that good candidates should have a logP ≤ 5, a molecular weight ≤ 500 g/mol, no more than 10 hydrogen bond acceptors, and a maximum of 5 hydrogen bond donors. The ligands L1-L4 fulfil the Lipinski's rule [29]. They all also fulfil the rules given by Ghose, Veber, Egan, and Muegge [30][31][32][33]. The only exception was L5, which had slightly too high cLogP, according to Lipinski, WLogP and molar refractivity, according to Ghose, and XLogP3, according to Muegge. The calculated polar surface area (tPSA) values of ligands L1-L5 varied from 50.50 to 106.98 Å3, so they were potentially able to cross the membranes. After this preliminary evaluation, we decided to synthesise all 5 ligands since the coordination of metal ions could improve the solubility and other parameters. All ligands were characterised by 1 H NMR, 13 C NMR, ESI-MS, and EA (Figures S1-S10). Moreover, ligands L1, L2, L4, and L5 were crystallised, and their solid-state structure was confirmed using X-ray (cf. Section 2.2). However, only ligand L1 formed coordination polymers with the initially planned Ag(I) and Cu(I) tetrahedral ions, as revealed by the X-ray diffraction on the single crystals (cf. Section 2.2). The polymeric structure of {[CuL1]PF6}n and {[AgL1]BF4}n was also established from the 1 H NMR spectra (Figures S11-S12). In the spectra of Polymer1, the appropriate peaks were shifted and broader compared to the spectrum of L1 ( Figure S11). It needs to be noted that we took precautions so that the Cu(I) did not oxidise to Cu(II); additionally, the reddish colour of the solution was retained for several weeks. In the case of the Ag(I) polymer, the coordination was evidenced by the Scheme 1. Synthesis of ligands L1-L5 with molecular representations of all derivatives including the respective R moieties.
After this preliminary evaluation, we decided to synthesise all 5 ligands since the coordination of metal ions could improve the solubility and other parameters. All ligands were characterised by 1 H NMR, 13 C NMR, ESI-MS, and EA (Figures S1-S10). Moreover, ligands L1, L2, L4, and L5 were crystallised, and their solid-state structure was confirmed using X-ray (cf. Section 2.2). However, only ligand L1 formed coordination polymers with the initially planned Ag(I) and Cu(I) tetrahedral ions, as revealed by the X-ray diffraction on the single crystals (cf. Section 2.2). The polymeric structure of {[CuL1]PF 6 } n and {[AgL1]BF 4 } n was also established from the 1 H NMR spectra (Figures S11 and S12). In the spectra of Polymer1, the appropriate peaks were shifted and broader compared to the spectrum of L1 ( Figure S11). It needs to be noted that we took precautions so that the Cu(I) did not oxidise to Cu(II); additionally, the reddish colour of the solution was retained for several weeks. In the case of the Ag(I) polymer, the coordination was evidenced by the shifts of the c, d, and e proton signals compared to the parent spectrum of L1 ( Figure S12). In the literature, there are some reports on the decomposition of Schiff bases upon the addition of metal salt [34], but in the case of our acyclic and macrocyclic Schiff bases, our experimental results indicated that they were highly stable [11,[35][36][37].
Ligands L1-L5 were examined for their interactions with DNA and RNA using UV-Vis titration, while Polymer1 and Polymer2 were tested for their antibacterial activity toward an E. coli strain.

Description of Structures
Perspective views of the ligands L and cations from the structures Polymer1 and Polymer2 are shown in Figures 1-3.
All four ligand molecules in their crystal structures are C i -symmetrical as they lie across the inversion centres in their respective space groups (L1 P2 1 /c, all others P-1). As a consequence, the two symmetry-related peripheral rings are exactly parallel.
Interestingly, in the crystal structures of two cationic complexes, the ligand molecules lie across symmetry elements as well: inversion centres in the case of both ligand molecules in the structure of Polymer1 and one of the molecules in Polymer2 and twofold screw for the second molecule in the latter case. As in both structures, the coordination polymers are formed, and the subsequent monomers are related by the inversion centres-two different ones in the case of L1 and additionally by a twofold screw in L2.
In the Cu complex, Polymer1, the Cu ions are four-coordinated by two ligand molecules and by the solvent-the acetonitrile molecule (N 4 ) in a distorted tetrahedral environment (Table S1 lists the relevant geometrical data). The two independent ligand molecules display different orientations of the thiazole ring (NCCN disposition is trans in molecule A and cis in B, cf. Table S1). In consequence, the coordinating potentials of these molecules are different-molecule A acts as a two-dentate ligand (connected to two subsequent Cu ions in the polymeric chain by ring nitrogen atoms only), while molecule B is tetradentate, utilizing all four nitrogen atoms as coordination centres ( Figure 2). Therefore, the acetonitrile molecule acts as the fourth coordination place.
Polymer2 is more typical: in both symmetry-independent ligand molecules, nitrogen atoms are in cis disposition (cf. Table S1), and both ligands are tetra-dentate; the Ag ions are coordinated only by the ligand nitrogen atoms, with coordination number 4 and a very severely distorted tetrahedral geometry ( Figure 3).
In both polymers, the counterions (PF 6 − and BF 4 − ) balance the overall charge and lie between the positively charged polymeric chains. It might be safely assumed that the electrostatic interactions between charged species are a main factor for crystal cohesion and the details of crystal architectures. In the structure of Polymer1, an additional solvent electron density was found and interpreted as a disordered water molecule.

Interactions with Nucleic Acids
In general, the enhanced DNA binding ability of a molecule can be achieved by increasing its planarity with ligands such as bipyridine and phenanthroline. Aromatic compounds, due to their planar structure, have the ability to slide between adjacent nucleic base pairs and induce high destabilization of the DNA double helix (for example with the transition from the B form to the Z form). Previous studies revealed that their planarity promotes intercalative interactions due to π-stacking between the plane of the aromatic rings and DNA base pairs [38,39]. Another important factor is the presence of potential donors and acceptors of H-bonding in the molecules that are directional and specific in the binding with molecules of interest [40].
In order to determine whether ligands interact with the DNA helix, a spectrophotometric titration was performed in our study. The method consists of measuring the UV absorbance of the ligands after each of the subsequent portions of CT-DNA (calf thymus DNA) is added (Figures 4 and S13-S16). The mechanisms of action of many bioactive compounds depend on their binding with DNA; hence, determining the interaction between the molecules and DNA is crucial for understanding, at a molecular level, the origin of their possible therapeutic effects. Chemical compounds may interact with DNA variously-via covalent bonding, intercalation between base pairs, electrostatic interactions with the phosphate-rich backbone, or binding to either minor or major grooves [41,42].
The manner in which a compound interacts with DNA affects the binding's reversibility, strength, specificity, and cytotoxic effect. When an aromatic compound binds to DNA, its absorbance decreases as the compound is no longer in its free form. In the case of the L1 ligand, there is a significant decrease in absorbance as the CT-DNA concentration rises, which may indicate the strong interaction with the DNA helix. The bonding stability constant Kb of the L1 ligand equals 1.48 × 10 4 , which is lower than the Kb of a standard intercalator ethidium bromide (Kb = 1.4 × 10 6 ). [14] Due to the conformational lability of the L1 ligand structure, it is possible that the compound adjusts its conformation to the shape of a major groove with the aid of the thiazole moiety. There are no significant changes in the spectra of the other ligands (Figures S13-S16).

Interactions with Nucleic Acids
In general, the enhanced DNA binding ability of a molecule can be achieved by increasing its planarity with ligands such as bipyridine and phenanthroline. Aromatic compounds, due to their planar structure, have the ability to slide between adjacent nucleic base pairs and induce high destabilization of the DNA double helix (for example with the transition from the B form to the Z form). Previous studies revealed that their planarity promotes intercalative interactions due to π-stacking between the plane of the aromatic rings and DNA base pairs [38,39]. Another important factor is the presence of potential donors and acceptors of H-bonding in the molecules that are directional and specific in the binding with molecules of interest [40].
In order to determine whether ligands interact with the DNA helix, a spectrophotometric titration was performed in our study. The method consists of measuring the UV absorbance of the ligands after each of the subsequent portions of CT-DNA (calf thymus DNA) is added (Figures 4 and S13-S16). The mechanisms of action of many bioactive compounds depend on their binding with DNA; hence, determining the interaction between the molecules and DNA is crucial for understanding, at a molecular level, the origin of their possible therapeutic effects. Chemical compounds may interact with DNA variouslyvia covalent bonding, intercalation between base pairs, electrostatic interactions with the phosphate-rich backbone, or binding to either minor or major grooves [41,42].
The manner in which a compound interacts with DNA affects the binding's reversibility, strength, specificity, and cytotoxic effect. When an aromatic compound binds to DNA, its absorbance decreases as the compound is no longer in its free form. In the case of the L1 ligand, there is a significant decrease in absorbance as the CT-DNA concentration rises, which may indicate the strong interaction with the DNA helix. The bonding stability constant K b of the L1 ligand equals 1.48 × 10 4 , which is lower than the K b of a standard intercalator ethidium bromide (K b = 1.4 × 10 6 ). [14] Due to the conformational lability of the L1 ligand structure, it is possible that the compound adjusts its conformation to the shape of a major groove with the aid of the thiazole moiety. There are no significant changes in the spectra of the other ligands (Figures S13-S16).
The next step of the study included an experiment determining the affinity of the ligands to the ribonucleic acid using the RNA from Saccharomyces cerevisiae as a model of RNA (Figures 5 and S17-S20). Given the key structural differences between RNA and DNA, it was expected that the ligands would interact with both types of nucleic acids in non-identical ways. The most important dissimilarity is that RNA is typically single-stranded, while DNA is typically found as a double-strand. RNA molecules play a prominent role in biological processes and evolve as an important target of therapeutic intervention. Molecules that specifically bind to RNA prevent its folding and the formation of RNA-protein complexes. They can also affect cellular functions and have therapeutic potential. Studies conducted over several decades strengthened the role of RNA as a central biomolecule that is considered a structurally and functionally sophisticated biopolymer that participates in key cellular events. For example, RNA can be used to control cell functions via interactions with exogenous ligands and in therapeutics. [43] In accordance with these assumptions, evident differences in the interaction of the ligands with DNA and RNA, stemming from the significant structural dissimilarities of the nucleic targets, were observed by us. The next step of the study included an experiment determining the affinity of the ligands to the ribonucleic acid using the RNA from Saccharomyces cerevisiae as a model of RNA (Figures 5 and S17-S20). Given the key structural differences between RNA and DNA, it was expected that the ligands would interact with both types of nucleic acids in non-identical ways. The most important dissimilarity is that RNA is typically singlestranded, while DNA is typically found as a double-strand. RNA molecules play a prominent role in biological processes and evolve as an important target of therapeutic intervention. Molecules that specifically bind to RNA prevent its folding and the formation of RNA-protein complexes. They can also affect cellular functions and have therapeutic potential. Studies conducted over several decades strengthened the role of RNA as a central biomolecule that is considered a structurally and functionally sophisticated biopolymer that participates in key cellular events. For example, RNA can be used to control cell functions via interactions with exogenous ligands and in therapeutics. [43] In accordance with these assumptions, evident differences in the interaction of the ligands with DNA and RNA, stemming from the significant structural dissimilarities of the nucleic targets, were observed by us.
Per the results of the UV-Vis titration studies, only the ligand L1 bound to RNA; its bonding stability constant equalled 5.74 × 10 3 . In the case of the L2, L4, and L5 ligands, the same phenomenon was not observed. The nonlinear changes in the absorbance in the case of the L3 ligand might indicate some weak and non-specific electrostatic interactions between the compound and the nucleic acid. The next step of the study included an experiment determining the affinity of the ligands to the ribonucleic acid using the RNA from Saccharomyces cerevisiae as a model of RNA (Figures 5 and S17-S20). Given the key structural differences between RNA and DNA, it was expected that the ligands would interact with both types of nucleic acids in non-identical ways. The most important dissimilarity is that RNA is typically singlestranded, while DNA is typically found as a double-strand. RNA molecules play a prominent role in biological processes and evolve as an important target of therapeutic intervention. Molecules that specifically bind to RNA prevent its folding and the formation of RNA-protein complexes. They can also affect cellular functions and have therapeutic potential. Studies conducted over several decades strengthened the role of RNA as a central biomolecule that is considered a structurally and functionally sophisticated biopolymer that participates in key cellular events. For example, RNA can be used to control cell functions via interactions with exogenous ligands and in therapeutics. [43] In accordance with these assumptions, evident differences in the interaction of the ligands with DNA and RNA, stemming from the significant structural dissimilarities of the nucleic targets, were observed by us.
Per the results of the UV-Vis titration studies, only the ligand L1 bound to RNA; its bonding stability constant equalled 5.74 × 10 3 . In the case of the L2, L4, and L5 ligands, the same phenomenon was not observed. The nonlinear changes in the absorbance in the case of the L3 ligand might indicate some weak and non-specific electrostatic interactions between the compound and the nucleic acid.
, experimental data points; solid line, linear fitting of the data.
Per the results of the UV-Vis titration studies, only the ligand L1 bound to RNA; its bonding stability constant equalled 5.74 × 10 3 . In the case of the L2, L4, and L5 ligands, the same phenomenon was not observed. The nonlinear changes in the absorbance in the case of the L3 ligand might indicate some weak and non-specific electrostatic interactions between the compound and the nucleic acid. The binding of L1 to DNA and RNA was also explored computationally by using blind molecular docking with the HDOCK program (see, Experimental section) that revealed the ability of L1 to interact with DNA (using PDB ID: 1BNA [44], i.e. a model of double-stranded DNA) and RNA (1U2A, a model of RNA from Saccharomyces cerevisiae) [45].
In the complex with DNA, no stacking interaction was predicted, but mainly the groove binding of the ligand was evidenced (Figure 6), confirming the hypothesis coming from our experimental findings. Remarkably, the experimental trend of binding affinities  The next step of the study included an experiment determining the affinity of the ligands to the ribonucleic acid using the RNA from Saccharomyces cerevisiae as a model of RNA (Figures 5 and S17-S20). Given the key structural differences between RNA and DNA, it was expected that the ligands would interact with both types of nucleic acids in non-identical ways. The most important dissimilarity is that RNA is typically singlestranded, while DNA is typically found as a double-strand. RNA molecules play a prominent role in biological processes and evolve as an important target of therapeutic intervention. Molecules that specifically bind to RNA prevent its folding and the formation of RNA-protein complexes. They can also affect cellular functions and have therapeutic po-, experimental data points; solid line, linear fitting of the data.
The binding of L1 to DNA and RNA was also explored computationally by using blind molecular docking with the HDOCK program (see, Experimental section) that revealed the ability of L1 to interact with DNA (using PDB ID: 1BNA [44], i.e., a model of doublestranded DNA) and RNA (1U2A, a model of RNA from Saccharomyces cerevisiae) [45].
In the complex with DNA, no stacking interaction was predicted, but mainly the groove binding of the ligand was evidenced (Figure 6), confirming the hypothesis coming from our experimental findings. Remarkably, the experimental trend of binding affinities was also found in the docking simulation with DNA that formed with L1 complexes more stable than RNA, as can be deduced by comparing the HDOCK scores in Table 2. This can be explained in terms of stronger interactions involving the thiazole moiety in the case of L1-DNA but not L1-RNA (Figure 7).
[RNA]; ▪, experimental data points; solid line, linear fitting of the data.
The binding of L1 to DNA and RNA was also explored computationally by using blind molecular docking with the HDOCK program (see, Experimental section) that revealed the ability of L1 to interact with DNA (using PDB ID: 1BNA [44], i.e. a model of double-stranded DNA) and RNA (1U2A, a model of RNA from Saccharomyces cerevisiae) [45].
In the complex with DNA, no stacking interaction was predicted, but mainly the groove binding of the ligand was evidenced (Figure 6), confirming the hypothesis coming from our experimental findings. Remarkably, the experimental trend of binding affinities was also found in the docking simulation with DNA that formed with L1 complexes more stable than RNA, as can be deduced by comparing the HDOCK scores in Table 2. This can be explained in terms of stronger interactions involving the thiazole moiety in the case of L1-DNA but not L1-RNA (Figure 7).

Effect of Selected Compounds on Bacterial Proliferation
The effect of ligand L1, Polymer1 {[CuL1]PF 6 } n , and Polymer2 {[AgL1]BF 4 } n on bacterial growth was tested, and, in particular, the effect of the compounds on the change in the number of bacteria after times of 1, 4, and 24 h was investigated, as shown in Figure 8.

Effect of Selected Compounds on Bacterial Proliferation
The effect of ligand L1, Polymer1 {[CuL1]PF6}n, and Polymer2 {[AgL1]BF4}n on bacterial growth was tested, and, in particular, the effect of the compounds on the change in the number of bacteria after times of 1, 4, and 24 h was investigated, as shown in Figure  8. The highest activity against E. coli bacteria was shown by the Polymer2 {[AgL1]BF4}n, whose MIC50 (minimum inhibitory concentration of a compound that inhibits the growth of microorganisms by 50%) was found to be 92 µM after 24 h (Figure 9). At the same time, the activity of this complex highly exceeded that of silver salt (AgBF4) used as a control, ruling out the exclusive influence of the silver ion (Figure 9, left). From a biomedical perspective, the Polymer2 containing tetradentate ligand L1 in cis disposition, coordinating Ag(I) in a tetrahedral fashion, was shown to slow down the proliferation of bacteria much more effectively than the free Ag(I) salt. In this respect, we hypothesize that this could be an effect of the increased lipophilicity of the material or of a prolonged "metal-drug" release from the coordination polymer. Increased activity of complexes might be explained by the Tweedy chelation theory: it is observed that the positive charge in the complex is partially reduced due to the overlap of the ligand orbital. Further, it increases both the delocalization of π-electrons over the whole chelate structure and the lipophilic character of the complex, which enhances the penetration of the compound into the lipid layer of the bacterial cell membranes and blocks the metal binding sites in enzymes. [46] Previous preliminary results suggest that the bactericidal mechanism of Ag(I) ions occurs via DNA condensation and that the diminished replication abilities are a reaction against the denaturation process. Moreover, silver ions may interact with thiol groups in proteins, which induces the deterioration of cellular functions. [14] At the same time, ligand L1 and the The highest activity against E. coli bacteria was shown by the Polymer2 {[AgL1]BF 4 } n , whose MIC50 (minimum inhibitory concentration of a compound that inhibits the growth of microorganisms by 50%) was found to be 92 µM after 24 h (Figure 9). At the same time, the activity of this complex highly exceeded that of silver salt (AgBF 4 ) used as a control, ruling out the exclusive influence of the silver ion (Figure 9, left). From a biomedical perspective, the Polymer2 containing tetradentate ligand L1 in cis disposition, coordinating Ag(I) in a tetrahedral fashion, was shown to slow down the proliferation of bacteria much more effectively than the free Ag(I) salt. In this respect, we hypothesize that this could be an effect of the increased lipophilicity of the material or of a prolonged "metaldrug" release from the coordination polymer. Increased activity of complexes might be explained by the Tweedy chelation theory: it is observed that the positive charge in the complex is partially reduced due to the overlap of the ligand orbital. Further, it increases both the delocalization of π-electrons over the whole chelate structure and the lipophilic character of the complex, which enhances the penetration of the compound into the lipid layer of the bacterial cell membranes and blocks the metal binding sites in enzymes. [46] Previous preliminary results suggest that the bactericidal mechanism of Ag(I) ions occurs via DNA condensation and that the diminished replication abilities are a reaction against the denaturation process. Moreover, silver ions may interact with thiol groups in proteins, which induces the deterioration of cellular functions [14]. At the same time, ligand L1 and the Polymer1 {[CuL1]PF 6 } n showed only moderate strain-directed activity, which was comparable to that of silver and copper salts.

Polymer1 {[CuL1]PF6}n
showed only moderate strain-directed activity, which was comparable to that of silver and copper salts.

Conclusions
In conclusion, five new ligands were synthesized, characterized and investigated in biomolecular interaction assays, revealing the ability of L1 to bind DNA and RNA, as evidenced by UV titration. The interaction with DNA was mediated by interactions with one

Conclusions
In conclusion, five new ligands were synthesized, characterized and investigated in biomolecular interaction assays, revealing the ability of L1 to bind DNA and RNA, as evidenced by UV titration. The interaction with DNA was mediated by interactions with one of the two thiazole moieties, while the L1-RNA binding was driven by weaker forces involving atoms out of the thiazole ring, as suggested by the molecular docking simulation.
Remarkably, L1 was able to form coordination polymers with Ag(I) and Cu(I) that were crystallized and described in a detailed form in our work. Polymer2 containing Ag(I) was shown to slow down the proliferation of bacteria more effectively than the free Ag(I) salt, while L1 and Polymer1 containing Cu(I) showed only moderate activity.
Overall, all the reported findings of this work concur to depict L1 as a thiazole-based heteroaromatic derivative with important characteristics as a binder of nucleic acids (as we showed with the two DNA and RNA models) and as a ligand of metals endowed with antimicrobial properties, as we demonstrated in particular in the case of the L1-based polymer containing Ag(I). Therefore, new studies involving both the synthesis of new modified L1 analogues and biophysical/biological assays with other biomolecular targets are clearly desirable.

Materials and Methods
CT-DNA was purchased from Merck (Darmstadt, Germany), while the baker's yeast RNA from S. cerevisiae was Alfa Aesar (Heysham, UK). All reagents (Merck, Darmstadt, Germany) and substrates were used without further purification. ESI mass spectra for MeCN solutions~10 −4 M were measured using a Waters Co. Micromass ZQ spectrometer (Milford, CT, USA) and QTOF type mass spectrometer Impact HD, Bruker (Billerica, MA, USA). NMR spectra were run on a Varian Gemini 300 MHz spectrometer (Oxford, UK) and were calibrated against the residual protonated solvent signals with chemical shifts represented in ppm. Microanalyses were performed using a VarioEL III CHN element analyzer (Thermo Scientific, Waltham, MA, USA).
CT-DNA and RNA were dissolved in a PBS buffer, pH = 7.4, prior to use. The CT-DNA solution gave a ratio of UV absorbance of 1.82:1 at 260 and 280 nm, indicating that the CT-DNA sample was sufficiently free from protein [47]. CT-DNA and RNA concentrations per nucleotide were determined from the UV absorbance at 260 nm using the extinction coefficient ε 260 = 6600 dm 3 ·mol −1 ·cm −1 and ε 260 = 7800 dm 3 ·mol −1 ·cm −1 , respectively [48]. Electronic absorption spectra were performed on UV-Vis JASCO V-770 equipped with a Peltier Thermo Cell Holder (water) PAC-743R (Jasco International Co., Tokyo, Japan).
It needs to be emphasized that compounds are stable in this medium for several weeks (after this time, some precipitate starts to occur).

Synthesis of Ligands L1-L5
The ligands were synthesized according to the procedure depicted in Scheme 1.

L1-C 16 H 14 N 4 S 2
The 2-thiazolecarboxaldehyde (343.94 mg, 3.03 mmol) was dissolved in 6 mL absolute EtOH. Then p-xylylenediamine (200 mg, 1.46 mmol) was added. The reaction was carried out for 24 h under an argon atmosphere at 78 • C. The product was precipitated with diethyl ether and was filtered, washed with cold absolute EtOH, and dried under reduced pressure for 5 h. A yellow product was obtained with a 72.0% yield. Single crystals suitable for X-ray diffraction analysis were formed by a slow diffusion of diisopropyl ether into the sample solution in acetonitrile at 4 • C over a period of 6-8 weeks.

L2-C 20 H 18 N 4
The pyridine-2-carboxaldehyde (293.2 µL, 3.1 mmol) was dissolved in 8 mL absolute EtOH. Then p-xylylenediamine (200.00 mg, 1.46 mmol) was added. The reaction was carried out for 24 h under an argon atmosphere at 78 • C. The product was precipitated with diethyl ether and was filtered, washed with cold absolute EtOH and dried under reduced pressure for 5 h. A white product was obtained with a 70.0% yield. Single crystals suitable for X-ray diffraction analysis were formed by a slow diffusion of diisopropyl ether into the sample solution in acetonitrile at 4 • C over a period of 6-8 weeks.

L4-C 18 H 20 N 6
The 1-methyl-2-imidazolecarboxaldehyde (339.57 mg, 3.1 mmol) was dissolved in 6 mL absolute EtOH. Then p-xylylenediamine (200.00 mg, 1.46 mmol) was added. The reaction was carried out for 24 h under an argon atmosphere at 78 • C. The product was precipitated with diethyl ether and was filtered, washed with cold absolute EtOH and dried under reduced pressure for 5 h. A white product was obtained with a 68.0% yield. Single crystals suitable for X-ray diffraction analysis were formed by a slow diffusion of diisopropyl ether into the sample solution in acetonitrile at 4 • C over a period of 6-8 weeks.

L5-C 28 H 22 N 6
The quinoline-2-carboxaldehyde (600 mg, 3.8 mmol) was dissolved into 10 mL absolute EtOH. Then p-xylylenediamine (258 mg, 2.098 mmol) was added. The reaction was carried out for 24 h under an argon atmosphere at 78 • C. The product was precipitated with diethyl ether and was filtered, washed with cold absolute EtOH and dried under reduced pressure for 5 h. A white product was obtained with a 64.0% yield. Single crystals suitable for X-ray diffraction analysis were formed by a slow diffusion of diisopropyl ether into the sample solution in acetonitrile at 4 • C over a period of 6-8 weeks.

Polymer1-{[CuL1]PF 6 } n
The L1 ligand (50.00 mg, 153.1 µmol) was dissolved in acetonitrile (15 mL), and then CuPF 6 salt (57.10 mg, 153.0 µmol) was added. Argon was passed through the solution to remove air to prevent the oxidation of copper ions. The reaction was carried out for 24 h at room temperature. The solution was then reduced to a volume of about 3 mL, and the precipitate was obtained by the addition of the diethyl ether. The orange product was filtered under reduced pressure and dried in a vacuum desiccator, with a yield of 58%. By the method of slow diffusion in the acetonitrile/diisopropyl ether solvent system, a monocrystal suitable for diffractometric measurements was obtained.
ESI  (15 mL), and then AgBF 4 salt (29.80 mg, 153.0 µmol) was added. The reaction was carried out for 24 h at room temperature, protecting the reaction flask from the sunlight. The solution was then reduced to a volume of about 5 mL, and a white precipitate was obtained with diethyl ether. The off-white product was filtrated under reduced pressure and dried in a vacuum desiccator with a yield of 35%. By the method of slow diffusion of diethyl ether into the acetonitrile solution of the complex, monocrystals suitable for X-ray structural analysis were obtained. 1

X-ray Crystallography
Diffraction data were collected using the ω-scan technique for L1 and Polymer2 at 130(1) K on a Rigaku SuperNova four-circle diffractometer with an Atlas CCD detector, equipped with Nova microfocus CuK α radiation source (λ = 1.54178 Å), and for L2, L4, L5, and Polymer1 at 100(1) K, on a Rigaku XCalibur four-circle diffractometer with an Eos CCD detector, with a graphite-monochromatised MoK α radiation source (λ = 0.71073 Å). The data were corrected for Lorentz-polarization as well as for absorption effects [49]. The structures were solved with SHELXT [50] and refined with the full-matrix least-squares procedure on F 2 by SHELXL-2013 [51]. All non-hydrogen atoms were refined anisotropically, and hydrogen atoms were placed in idealised positions and refined as 'riding models' with isotropic displacement parameters set at 1.2 (1.5 for methyl and hydroxyl groups) times U eq of appropriate carrier atoms. In the structure of Polymer1, the relatively high residual density far from the rest of the structure was interpreted as a disordered water molecule, with s.o.f's set at 1/3 and 2/3 on the basis of keeping similar displacement parameters. The relevant crystallographic data, together with the details of structure refinement, are listed in Table 3.

In Silico Pharmacokinetic Property Studies
The SMILES codes for compounds (Simplified Molecular Input Line Entry System) were obtained and applied to the calculations using the SwissADME [52] (http://www. swissadme.ch, accessed on 20 November 2022): the molecular weights (MW), the logP values (octanol-water partition coefficient), in six variants (ILOGP, XLOGP3, WLOGP, MLOGP, SILICOS-IT and consensus LogP, cLogP, which was an average of five mentioned predictions), tPSAs (topological polar surface area), number of hydrogen-bond acceptors and donors, number of atoms, rotatable bonds, ring, carbon and heteroatoms. The cLogP values were obtained using the SwissADME web service [47], together with ILOGP, obtained with the in-house physics-based method implemented by Daina et al. [53]; XLOGP3 values, predicted with the atomistic and knowledge-based method calculated using the XLOGP program, version 3.2.2, courtesy of CCBG, Shanghai Institute of Organic Chemistry; WLOGPs, obtained with the atomic method implemented by Wildman SA and Crippen GM [54]; MLOGPs, calculated with the topological method implemented by Moriguchi et al. [55,56] and Lipinski et al. [57]; SILICOS-IT scores, obtained with the hybrid fragmental/topological method calculated using the FILTER-IT program, version 1.0.2, courtesy of SILICOS-IT (http://ww1.silicos-it.com/, accessed on 20 November 2022); other parameters not mentioned here using the SwissADME web service [52]. The tPSA value was calculated according to Ertl et al. [58] SMILES codes: where [DNA] is the concentration of CT-DNA in base pairs, ε a is the observed extinction coefficient, ε f is the extinction coefficient of the compound in its free form, and ε b is the extinction coefficient of the compound fully bound to CT-DNA [59].

Ligand-RNA Interactions
Spectrophotometric titration of the ligands with RNA was also performed. Similarly to DNA, starting solutions of the ligands in DMSO (c = 2 mM) were prepared, and the PBS buffer with pH = 7.4 was used in the measurements. An amount of 2.45 mL of the buffer and 50 µL of the ligand solution were placed in a quartz cuvette (with dimensions 1 cm × 1 cm), and then the absorption spectrum was measured (at c RNA = 0 µM). Before each subsequent measurement, a portion of RNA was added, increasing its concentration by 40 µM until the final concentration of RNA was reached (c RNA = 200 µM). Six measurements were conducted for each of the ligands individually. The baseline was measured before every measurement for each of the RNA concentrations in the buffer. The ligand-RNA binding constant K b was calculated for the L1 ligand according to the equation: where [RNA] is the concentration of RNA in base pairs, ε a is the observed extinction coefficient, ε f is the extinction coefficient of the compound in its free form, and ε b is the extinction coefficient of the compound fully bound to RNA [59].

Effect of Compounds on Bacterial Proliferation
An antimicrobial assessment was conducted to determine the effect of ligand L1 and its complexes {[CuL1]PF 6 } n and {[AgL1]BF 4 } n on the proliferation of a bacterial strain of Escherichia coli. Stock solutions of the compounds were prepared with concentrations of c = 10 mM, c = 5 mM, and c = 2.5 mM. To the sets of liquid cultures (5 mL), 50 µL each of the previously prepared compound solutions was added. After 100× dilution, the concentrations in the test samples was 100 µM, 50 µM, and 25 µM, respectively. The OD (optical density) was measured both before and after incubation periods of 1 h, 4 h, and 24 h. The initial OD for the bacteria was 0.38. Measurements were also made for the corresponding metal salts used in the synthesis. Bacteria in a medium (without any of the tested compounds) were the control sample.

Molecular Docking Studies
We performed molecular docking simulations [60][61][62][63] with the HDOCK server (http: //hdock.phys.hust.edu.cn, accessed on 20 September 2022) [64,65], suitable for both macromolecule-to-macromolecule [64] and macromolecules-to-small molecule [66] rigid dockings, using default parameters. The PDB entries 1BNA [44], relative to the DNA dodecamer d(CpGpCpGpApApTpTpCpGpCpG), and 1U2A, relative to the stem-loop IIa from yeast U2 small nuclear RNA [45], were used as models of DNA and RNA, respectively, for our blind dockings. They were furnished to HDOCK as targets, while the 3D structure of L1 saved as a .pdb file was uploaded into the server as the ligand. Thanks to the iterative knowledge-based scoring function ITScore-PP, the HDOCK server ranked the top ten poses obtained after the dockings. The energy score (HDOCK score) values given by the program, and predicted by ITScore-PP, were dimensionless, with larger negative numbers being linked to higher affinity interactions between the interacting ligand and the target macromolecule, which was also reported to correlate well to experimental binding affinities showing a correlation coefficient of R = 0.71 [67]. More details on the procedures for the docking and the HDOCK docking server itself are available at http://hdock.phys.hust.edu.cn (accessed on 20 September 2022). We analysed the topranked pose (Top-1) for the complexes predicted using HDOCK according to the energy scores provided by the program, as explained in the Results and discussion section.