Polyamine–Oligonucleotide Conjugates: 2′-OMe-Triazole-Linked 1,4,7,10-Tetraazacyclododecane and Intercalating Dyes and Their Effect on the Thermal Stability of DNA Duplexes

Oligonucleotides with the sequences 5′-GTG AUPA TGC, 5′-GCA TAUP CAC and 5′-GUPG ATA UPGC, where UP is 2′-O-propargyl uridine, were subjected to post-synthetic Cu(I)-catalyzed azide–alkyne cycloaddition to attach 1,4,7,10-tetraazacyclododecane (cyclen) and two well-known DNA intercalating dyes: thioxanthone and 1,8-naphthalimide. We propose a convenient cyclen protection–deprotection strategy that allows efficient separation of the resulting polyamine–oligonucleotide conjugates from the starting materials by RP-HPLC to obtain high-purity products. In this paper, we present hitherto unknown macrocyclic polyamine–oligonucleotide conjugates and their hybridization properties reflected in the thermal stability of thirty-two DNA duplexes containing combinations of labeled strands, their unmodified complementary strands, and strands with single base pair mismatches. Circular dichroism measurements showed that the B-conformation is retained for all dsDNAs consisting of unmodified and modified oligonucleotides. An additive and destabilizing effect of cyclen moieties attached to dsDNAs was observed. Tm measurements indicate that placing the hydrophobic dye opposite to the cyclen moiety can reduce its destabilizing effect and increase the thermal stability of the duplex. Interestingly, the cyclen-modified U showed significant selectivity for TT mismatch, which resulted in stabilization of the duplex. We conclude the paper with a brief review and discussion in which we compare our results with several examples of oligonucleotides labeled with polyamines at internal strand positions known in the literature.


Introduction
Macrocyclic polyamines and their transition metal complexes are attracting increasing interest due to their clinical potential in cancer and virus treatment and in magnetic resonance imaging. Chemical modifications involving covalent attachment of polyamines to oligonucleotides (ON) create zwitterionic functional groups that can significantly improve their biological and biophysical properties, such as target affinity and cell penetration, in a manner similar to polyamine transfection agents. The introduction of such modifications was carried out using several different strand positions, including the 3 and 5 -positions of the phosphate backbone, the 2 and 4 -positions on the ribose ring, and within the nucleobase itself [1,2]. In contrast to the 3 and 5 -positions, the stability of both oligonucleotides and duplexes is more sensitive to modifications of the ribose ring structure and conformation, although it is the 2 -position of the ribose ring that is particularly suitable for the covalent attachment of large molecules, such as polyamines, with minimal disruption of the base-paring potential. There are many examples of polyamine-oligonucleotide conjugates, but in most studies, polyamines are linear, while examples of macrocyclic polyamines that can form stable complexes with transition metals are rarer, and information on their effect on duplex stability is often lacking. Dubey et al. showed that cyclen-based transition metal complexes attached to the 5 -position of an oligo(dT) are able to hydrolyze the target oligo(dA) more efficiently; however, the effect of the cyclen moiety on the thermal stability of the duplex has not been described [3]. Steward et al. demonstrated a four-arm, lattice-bearing, single-stranded DNA bound to the central Ni(II)-cyclen complex, which improves self-assembly at the supramolecular level, but its effect on duplexes is also unknown [4]. On the other hand, it is known that macrocyclic polyamines such as 1,4,7,10tetraazacyclododecane (cyclen) are potential artificial nucleases, and their derivatives can cleave double-stranded DNA (dsDNA), even without metal ions, through hydrolysis or oxidative cleavage [5][6][7][8][9]. Thus, covalent attachment of a macrocyclic amine to ssDNA may provide new and useful models for studying the function and in vitro use of artificial nucleic acid-based nucleases.
The second type of modification that we introduced, intercalating dyes, have a considerable position in the chemistry of nucleic acids [10]. These planar and aromatic molecules can intercalate between the nucleobases of dsDNA changing its topology but can also be explored as fluorescent probes for in vitro applications. Typically, oligonucleotide-based probes consist of covalently attached fluorescent dyes, including perylene [11], pyrene and phenanthroline [12][13][14][15], or fluorescein [16], which are known to exhibit high fluorescence and can interact noncovalently with dsDNA, e.g., by intercalation or groove-binding, leading to its stabilization. We have previously shown that covalent attachment of a carbazole moiety to the 5 -end of a 9-mer sequence increases the thermal stability of the resulting 9-mer/15-mer dsDNA by +4.2 • C [17]. To date, the effect of the combined attachment of both molecules, an intercalator and cyclen, to double-stranded oligonucleotides on their thermal stability has not been investigated. The knowledge of the stabilizing (or destabilizing) effect will be helpful in the preparation of cyclen-containing oligonucleotides with tailored stability of the resulting hybridized duplexes. Telser et al. prepared several dsD-NAs with covalently attached labels, e.g., anthraquinone or pyrene, placed at the internal positions of both strands and showed that both label-duplex and label-label interactions affect the thermal stability of the resulting duplexes [18,19]. Following the above studies, we also examined the mutual influence of the introduced modifications on the stability of duplexes.
Herein, we present a preliminary study of a new methodology for the covalent attachment of cyclen moieties to oligonucleotides and the assessment of their effect on the thermal stability of the resulting DNA duplexes. For this purpose, we developed a new procedure for introducing N-TFA-protected cyclen via a 2 -OMe-triazole linkage, followed by purification and deprotection of the resulting conjugate to obtain a high-purity product that is well separated from the initial oligonucleotide. We were also interested in the mutual influence of the different labels placed on opposite positions of complementary strands on their stabilizing properties, which turn out to be of significant importance in the case of cyclen groups. In summary, we tested seven labeled oligonucleotides on examples of thirty-two dsDNA combinations formed between the labeled strands, their unmodified complementary strands, and strands with a single base pair mismatch.

Chemical Synthesis and Analysis
All reagents and anhydrous solvents were obtained from commercial sources and used without further purification except phenol distillation. Anhydrous solvents were dried over 4 Å molecular sieves and checked using a Karl Fisher titrator to determine if the water concentration was below 12 ppm before use. The progress of the chemical reactions was monitored by thin layer chromatography (TLC) on silica gel 60 F254 plates (Merc, Darmstadt, Germany). Spots on the TLC plate were visualized under UV light at 254 nm or by heating the plate after treatment with ninhydrin reagent made by dissolving 1.5 g of ninhydrin in 100 mL of n-butanol and adding 3.0 mL of acetic acid. Column chromatography was performed on Merck silica gel 60 (40-63 µm). Recycling preparative HPLC (prep-HPLC) was performed on a JAI LaboACE 5060 (Japan Analytic Industry, Tokyo, Japan). Depending on the type of compound to be purified, a tandem set of GPC JAIGEL-2HR+2.5 HR columns (∅20 mm × 600 mm) or a silica-based RP JAIGEL-ODS-AP-L SP-120-10 column (∅20 mm × 500 mm, 10 µm) was used for prep-HPLC. 1 H-NMR and 13 C-NMR spectra were recorded on a Varian NMR system 600 spectrometer (Agilent Technologies, Santa Clara, CA, USA) at 600 and 150 MHz, respectively. Peak multiplicity is expressed as follows: s = singlet, d = doublet, t = triplet, q = quartet, dd = doublet of doublets, ddd = doublet of doublets of doublets, m = multiplet. NMR chemical shifts are reported in ppm (δ), relative to residual nondeuterated solvents as internal standard and coupling constants (J) are given in Hz. Melting points (Mp) were determined using a Boethius microscope HMK type (Franz Küstner, Dresden, Germany). High-resolution electrospray ionization mass spectroscopy (HR-ESI-MS) analyses were performed on a Waters Xevo G2 QTOF apparatus (Waters-Micromass, Manchester, UK). Microwave-assisted reactions were carried out in a Biotage Initiator microwave reactor (Stockholm, Sweden) using 0.5-2.0 mL vials under the following conditions: 2 h, 90 • C, prestirring 30 s, high adsorption.

Ultraviolet Thermal Melting Studies
To determine the melting temperature (T m ) of the obtained duplexes, UV melting studies were performed on a Lambda 35 UV/Vis Spectrometer (Perkin-Elmer, Norwalk, CT, USA) using 10 mm path length Hellma SUPRASIL quartz cuvettes (Müllheim, Germany), monitoring at 260 nm with a complementary DNA/DNA strands concentration of 2.5 µM and a volume of 1.0 mL. Samples were prepared as follows: The modified strands and their corresponding complementary strands were mixed 1:1 (n/n) in 2.0 mL Eppendorf tubes before medium salt buffer (2×, 11.7 mM sodium phosphate, pH 7.0, 200 mM NaCl, 0.20 mM EDTA, pH 7.0, 500 µL) was added, which was completed in 1.0 mL using Milli-Q water. Thus, all samples were dissolved in 1× buffer condition (5.8 mM sodium phosphate, pH 7.0, 100 mM NaCl, and 0.10 mM EDTA). The samples were denatured by heating to 90 • C in a water bath and then slowly cooled to rt before transferring them to cuvettes. The absorbance at 260 nm was recorded as a function of time with a linear temperature increase from 6 to 80 • C at a rate of 1.0 • C/min programmed by a Peltier temperature controller. Two separate melting curves were measured, and T m values were calculated with the UV-WinLab software, taking the mean of the two melting curves with a deviation of no more than 0.5 • C.

Circular Dichroism Studies
Samples were prepared in the same way as for the T m measurement. The background spectrum of the buffer was recorded and subtracted from the corresponding spectra. Measurements were performed on a JASCO J-815 spectrometer (Tokyo, Japan) at 20 • C using quartz optical cells with a path length of 5 mm and a total volume of 1.0 mL. All CD spectra were recorded from 200-400 nm with a scan rate of 100 nm/min, employing 5 scans.

Oligonucleotides Purification and Analysis
RP-HPLC purification of crude oligonucleotides was performed by Waters 600 HPLC System with a Waters XBridge BEH C18-column (∅19 mm × 100 mm, 5 µm). Elution was performed by isocratic hold of A-buffer for 5.0 min, followed by a linear gradient to 70% of B-buffer for 16.5 min at a flow rate of 5.0 mL/min (A-buffer: 0.05 M TEAA buffer, pH 7.4; B-buffer: 25% A-buffer, 75% MeCN). IE-HPLC purification of oligonucleotides was caried on a DIONEX Ultimate 3000 system with a DNAPac PA100 Semi-Preparative column (∅9 mm × 250 mm, 13 µm) at 60 • C (Thermo Fisher Scientific, Darmstadt, Germany). Elution was performed with an isocratic hold of 10% C-buffer in Milli-Q water, starting with hold on 2% D-buffer in Milli-Q water for 2.0 min, followed by a linear gradient to 25% of D-buffer in Milli-Q water for 20.0 min at a flow rate of 2.0 mL/min (C-buffer: 0.25 M Tris-Cl, pH 8.0; D-buffer: 1.0 M NaClO 4 ). After purification, the appropriate fractions were combined and concentrated by purging with N 2 at 55 • C, and the obtained samples were dissolved in Milli-Q water (100 µL), then desalted with an addition of NaClO 4 solution (5.0 M, 15 µL), suspended in cold ethanol (1.5 mL) and stored at −20 • C for 1-2 h. After centrifugation (13,200 rpm, 5 min, 4 • C), the supernatant was filtered off and the pellet was washed with cold ethanol (2 × 1.0 mL), dried under N 2 flow at 55 • C, and dissolved in Milli-Q water (1.0 mL). Analytical RP-HPLC was performed on a Merck-Hitachi 7000 system (Hitachi Instruments, Tokyo, Japan) equipped with a Waters XBridge OBD C18-column (∅10 mm × 50 mm, 2.5 µm) at 60 • C. Elution was started with an isocratic hold of A-buffer for 2 min followed by a linear gradient to 85% of B-buffer for 30 min, keeping the flow rate at 1.3 mL/min. The structure and composition of oligonucleotides was verified by the MALDI-TOF MS method performed on an Ultraflex Extreme mass spectrometer (Bruker Daltonics, Bremen, Germany). Finally, the purified oligonucleotides were quantified by measuring OD as the absorbance at 260 nm of the sample in 1.0 mL of water in a 10 mm path length cuvette. The excitation coefficients for DNAs at 260 nm were estimated to be 1 × 10 4 M cm −1 residue −1 .

Synthesis and Purification of ON1-ON5
Target ON1-ON5 were synthesized at the 1.0 µmol scale on polystyrene beads (Amersham Biosciences, Piscataway, NJ, USA) using an automated synthesizer Expedite 8909 (PerSeptive Biosystems, Framingham, MA, USA) according to the manufacturer's standard protocol, except for the introduction of 2 -O-propargyl-uridine (U P ) into the ON3-ON5 sequence by the so-called "hand-coupling procedure", previously used by Wengel's group [17]. The stepwise coupling efficiencies were >95% for standard conditions and ∼85% for hand-coupling. Cleavage from the beads and nucleobase deprotection were performed by incubation with concd. aq. NH 3 in a screw cap vial at 55 • C overnight. The supernatant was filtered and evaporated to remove NH 3 by heating the filtrate to 55 • C and purging with N 2 for 4 h. The crude samples were purified DMT-on by RP-HPLC and the 5 -DMT group was cleaved with 2% aq. trifluoroacetic acid. The deprotected oligonucleotides were eluted with a 30% MeCN soln. in water (v/v) and purified by IE-HPLC, then the composition of the collected fractions was assessed by MALDI-TOF MS. Unmodified and 2 -O-propargylated oligonucleotides were isolated in overall yields of 80-88% and were >98% pure by IE-HPLC analysis. ) were subsequently added. The resulting mixture was vortexed and centrifuged after adding each of the reagents. The vial was equipped with a magnetic stirrer, purged with Ar, sealed with a Teflon-lined septum cap, and microwaved. After completion of the reaction, the volume was made up to 2.0 mL with dH 2 O and divided into two equal parts. Each sample was desalted through a NAP-10 column (GE Healthcare, Little Chalfont, UK) following manufacturer's protocol. The resulting solution contains a mixture of two major products, tris-N-TFA-protected ON6 and partially deprotected bis-N-TFA-protected ON6 . During RP-HPLC purification, the fractions ranging from t R 12.2 to 17.2 min were collected, evaporated together under a stream of N 2 , and used as a mixture in the next step. The resulting sample was incubated with 1.0 mL of satd. aq. NH 3 at 55 • C overnight and then evaporated by gentle blowing with N 2 at 30 • C for 4 h. The crude sample after deprotection was purified by IE-HPLC to give ON13 in 41% overall yield and 98% purity.
The synthesis of the intercalator-labeled ON9-ON12 and cyclen-labeled ON14 was performed under the same conditions as for ON13. After desalting through an NAP-10 column, ON9-ON12 samples were evaporated under a stream of N 2 and purified by RP-HPLC. These samples were obtained in high yield and purity (Table 1) and did not require further purification by IE-HPLC. After coupling 3 with ON4, a mixture of intermediates ON7" was obtained which was purified by RP-HPLC by collecting the fractions at t R from 12.2 to 17.2 min. After their joint deprotection and purification of the resulting sample by IE-HPLC, the ON14 conjugate was obtained in an overall yield of 40% and 98% purity.

Synthesis and Purification of ON15
For the synthesis of the double-functionalized ON15, the same procedure was used as for ON13, except that the following reaction system was used: ). The fractions ranging from t R 12.7 to 17.5 min were collected by RP-HPLC and evaporated together under a stream of N 2 . After complete deprotection, a single peak was observed at m/z 3428.227 in the MALDI-TOF MS spectra assigned to ON15 (calcd. as m/z 3428.957) and a single peak in the RP-chromatogram at t R 8.39 min. This fraction was collected and purified by IE-HPLC to give ON15 in 38% overall yield and 95% purity.

Chemical Synthesis of Labels
The structures of cyclen and selected intercalating dyes do not provide suitable functional groups for direct attachment to oligonucleotides, so we first synthesized their derivatives having an azide-terminated linker. The synthesis is shown in Scheme 1 and performed according to well-known methods with some modifications. First, three of the four cyclen amino groups were selectively N-protected as trifluoroacetamides (N-TFA) using ethyl trifluoroacetate (TFAEt) and purified by column chromatography in accordance with the method described previously [20]. The trifluoroacetamide protecting groups were chosen because of their easy and efficient removal in the last step of conjugate synthesis. During further steps, amine 1 was reacted with commercially available 5-bromovaleryl chloride in dry CH 2 Cl 2 , followed by treatment with NaN 3 in DMSO to form azide-terminated 3 with a 65% overall yield. Then, 2-Hydroxy-9H-thioxanthen-9-one 4 was synthesized by the reaction of phenol with thiosalicylic acid, which proceeds through successive EAS reactions and culminates in intramolecular Friedel-Crafts cyclization to form a tricyclic thioxanthone core [21]. Reaction of 4 and 1,8-napthalimide with 1,4-dibromobutane led to 5 and 7, respectively, in good yields. Further substitution of the terminal bromine for the azide group led to the formation of target compounds 6 and 8 with overall yield of 37 and 76%, respectively. The introduced linkers are expected to move the labels far enough and provide them sufficient flexibility to be close to the duplex backbone. The final products were purified by preparative HPLC before their conjugation with oligonucleotides.

Synthesis and Modification of Oligonucleotides
Scheme 2 shows a labeling method by post-synthetic coupling of azide-functionalized labels to 2 -O-propargylated oligonucleotides using Cu(I)-catalyzed azide-alkyne cycloaddition (CuAAC). Target oligonucleotides were prepared at the 1.0 µmol scale using an automated synthesizer and standard solid phase phosphoramidite chemistry. The 2 -Opropargyl-U (U P ) units were incorporated into the growing chains of ON3-ON5 in a sequence-specific manner using the so-called "hand-coupling protocol", which involves manual injection of U P phosphoramidite and increasing the coupling time to 25 min [22]. Covalent functionalization of mono-alkyne-modified ON3 and ON4 was carried out by microwave-assisted CuAAC using 0.4-fold molar ratio of CuSO 4 -TBTA complex with sodium ascorbate (n/n, 1:1:2.5) and a 4-fold ratio of azide-functionalized label to oligonucleotide [23]. Using the same reaction conditions to functionalize the di-alkyne analogue, ON5, did not give the expected double-clicked adduct, ON8", even after doubling the concentration of the catalyst system and azide label and increasing the MW reaction time to 6 h. We then sought to determine whether Cu(I)-THPTA or Cu(I)-BPS complexes could force the reaction to the double-clicked adduct better than Cu(I)-TBTA. We found that using THPTA instead of TBTA and BPS as a ligand, at the same molar ratio of the catalytic system, could promote the dual coupling of ON5, but showed no improvement in the single coupling of ON3 or ON4.
RP-HPLC analysis of the crude sample obtained after conjugation of 3 with ON3 followed by desalting on a NAP-column show the presence of two new fractions (shaded in gray in Figure 1a), well separated from each other and from the starting ON3. We collected both fractions separately and found that the one at t R 13.1 min gives a signal at m/z 3283.192, corresponding to ON6 (calcd. as m/z 3283.295) with a partially deprotected cyclen moiety, while the fraction at t R 16.3 min gives a signal at m/z 3379.920, corresponding to ON6 (calcd. as m/z 3379.303) with a fully protected cyclen moiety. Regardless of whether these fractions were collected and deprotected separately or together, in the RP-chromatogram, we observed the presence of only one fraction at t R 8.4 min with a signal at m/z 3091.684, coming from ON13 (calcd. as m/z 3091.278) having a fully deprotected cyclen moiety. After final IE-HPLC purification, ON13 was obtained in a total yield of 41% and a purity of 98% by IE-analysis (Supplementary Materials Figure S1). The overlay of RP-chromatograms in Figure 1a-e shows that omitting the isolation of the N-TFA-protected cyclen-oligonucleotide conjugates from the crude ON6-ON8 samples and proceeding directly to the deprotection step prevented further separation of the fully deprotected conjugates from the starting alkynylated oligonucleotides by RP-HPLC. In all cases, the retention time of the fully deprotected conjugates is almost equal to that of the starting alkynylated oligonucleotides and only a single peak is visible after the coinjection of both samples. IE-HPLC analysis of ON14 obtained by this procedure showed at least 17% content of the reaming fractions ( Figure S2). In turn, when ON14 was prepared by the same procedure as for ON13, i.e., by collecting fractions of differently protected ON7" conjugates (shaded in gray in Figure 1d) and deprotecting them together, we could easily obtain the final product in 41% overall yield and significant higher 98% purity ( Figure S3). We then applied this procedure to obtain ON15 by bifunctionalization of ON5, which, unlike monofunctionalization, resulted in a mixture of several overlapping fractions of differentially protected ON8" conjugates, seen in the RP-chromatogram at t R of 12.5 to 18.2 min (shaded in gray in Figure 1e). When these fractions were collected and deprotected together ( Figure S4), a single fraction was observed by RP-analysis at t R 8.4 min with a signal at m/z 3428.227 corresponding to ON15 (calcd. as m/z 3428.957).
Coupling of the thioxanthone derivative 6 to ON3 gave only one RP-fraction at t R 14.4 min with a signal at m/z 3118.754 corresponding to ON9 (calcd. as m/z 3119.595); this was accompanied by the disappearance of the initial ON3 peak at t R 8.4 min (Figure 1f). Purification by RP-HPLC gave ON9 with a purity of 98% and an overall yield of 56%; according to IE-analysis, the sample was sufficiently pure to be used in further duplex stabilization studies without the need for additional IE-HPLC purification ( Figure S5). A similar situation occurs for the conjugation of the 1,8-naphthalimide derivative 9 with ON3 and for other intercalating dyes; the products of these reactions are well separated from the starting materials and, after RP-HPLC purification, can be used directly for further studies (see Supplementary Materials, Figures S6-S8 show the results of IE-analysis for the remaining conjugates). Figure 2 shows circular dichroism (CD) spectra recorded to find possible changes in the secondary structure of the labeled duplexes. For all of them, the CD spectra showed intense negative and positive amplitudes at~250 nm and~280 nm, respectively, with no major differences relative to unmodified DU1 DNA duplex (black line in Figure 2a). The intensity of the bands also did not change significantly relative to unmodified DU1, suggesting than all modifications introduced do not induce any changes in the overall B-type duplex structure.  Figure 3 summarizes the duplex sequences along with the relative changes in melting temperatures (∆T m ) compared to the corresponding references. The unmodified duplex D1 has a reference T m of 32.5 • C, which is consistent with literature data [24]. In general, oligonucleotides with attached intercalating dyes have a positive effect on the thermal stability of all duplexes obtained, especially duplexes containing mismatches on one of the strands. The magnitude of this effect depends on the position and type of intercalating dye attached; in the case of duplexes containing only one modified strand, the highest increase in melting temperature was observed for DU6 and DU11, in which intercalatorlabeled U T and U N were adjacent to the GC base pair. For most duplexes containing two intercalator-labeled strands, an additive stabilizing effect was observed, although its magnitude depended on the combination and position of the labeled nucleotides. The largest stabilizing effect was observed for DU8, for which ∆T m is +10 • C. Interestingly, in the case of DU12 with an interchanged dye arrangement, compared to DU8, despite the stabilization of +5 • C compared to unmodified DU1, an antagonistic effect of lowering T m by −5 • C compared to DU8 was observed. The second regularity observed in almost all duplexes is a decrease in melting temperature by approximately the same value in the range from −4.0 to −5.5 • C, caused by the presence of a cyclen-labeled U C in one of the strands. The destabilizing effect of U C is additive and increases with increasing number of cyclen moieties attached to a single strand and their total number in the duplex, with ∆T m averaging −6 • C for each U C introduced. However, there are two exceptions to this regularity; one of which is DU28, where some selectivity against mismatch TT was observed, as evidenced by ∆T m of +1 • C. This result is not necessarily surprising, as previous work on Zn(II)-cyclen complexes has shown that cyclen is able to selectively recognize thymine by forming hydrogen bonds between the carbonyl oxygens of thymine and the cyclen amino groups [25,26]. The second exception is DU19, where two U C units are adjacent on complementary strands of the duplex, however, their destabilizing effect is not additive. In this case, only a thermal destabilization of −5 • C was observed, corresponding to the introduction of one U C unit instead of two. The ability of covalently coupled polyamines to thermally destabilize short duplexes has been reported and is discussed below. For mixed duplexes, consisting of one strand with a cyclen-modified U C and the other strand with an intercalator attached, additive effects affecting the thermal stability were also observed.

Discussion
Typically, when linear polyamines are introduced at the 2 or 4 -position of the ribose ring, either an adverse or no effect on the thermal stability of the DNA duplex is observed. Sund et al. showed that the introduction of C-branched spermine via ara-U-2 -phosphate (six-atom linker; the length of the linker is counted from the first atom attached to the ribose ring to the first polyamine atom) in the middle of DU33 (Figure 4) decreases the thermal stability of the duplex by as much as −28.5 • C, but the same study also showed that if spermine is attached to the 3 or 5 -end, then the thermal stability increases by +0.5 to +2.5 • C [27]. Winkler et al. attached linear polyamines via a 2 -N-succinylamido linker (four atoms long) to the internal and terminal nucleotides of DU34 (Figure 4), which also reduced the thermal stability by −4.5 • C [28]. Moreover, the introduction of further polyamine moieties into the internal positions of the 18-mer only led to further destabilization of the resulting duplexes. In contrast, Wengel's group has reported many examples of DNA oligonucleotides modified with linear or branched polyamines attached through 2 -amino-LNA motifs, including DU35-DU37 (Figure 4), which increase the thermal stability from +7.0 to +8.5 • C when present in the middle of the strand [23,29,30]. The modifications present in DU36 and DU37 are of particular interest to us because of their sequential and structural similarity to the duplexes obtained in this work, in particular to ON13. The cyclen-labeled U C present in DU4 consists of three protonable amine groups, similar to the spermidine moiety in DU37, and a 2 -methoxy-triazolyl-butyl linker is the same length (nine atoms long) as the linkers in DU36 and DU37. Despite these similarities, DU4 shows reduced thermal stability by −4.0 • C, while DU36 and DU37 show increased stabilization by +8.5 • C. The stabilization effect of +7.0 • C is also maintained by a shorter propanamide linker (three atoms long) conjugated with the piperazine ring in DU35, and even by the mere presence of the 2 -amino-LNA motif in D38, resulting in ∆T m of +4.0 • C. The results discussed here may suggest that the 2 -amino-LNA motif helps to adopt the correct conformation of the ribose ring, which may be critical for duplex stabilization by polyamines, especially when they are covalently attached to the internal positions of the strands. The results obtained for DU4 with those from the literature indicated a lower destabilization effect of cyclen in comparison with reported data for linear polyamines. The exceptions are polyamine-LNA conjugates, such as DU36 and DU37, unambiguously confirming the stabilizing effect of the LNA motif. Our study shows that the incorporation of cyclen-labeled U C in the middle of DU4 and other related duplexes led to a decrease in the thermal stability by an average value of −6.0 • C. An interesting exception to this rule is the introduction of two U C units directly opposite to each other on complementary strands, which causes much less duplex destabilization than would appear from the number of polyamines attached. Such an arrangement can be used to maximize the number of polyamines introduced with the least effect on the thermal stability of the duplex. Another exception to this rule is the presence of U C opposite the TT mismatch on a complementary strand; in this case, we observed a slight stabilization of the duplex, which can be used to design mismatch-selective DNA binders as useful models for understanding and modulating the action of DNA repair enzymes. We also showed that the presence of U T and U N modifications has a strong thermostabilizing effect on duplex formation, and the proximity of both modifications to each other and U C does not disturb their interaction with the duplex. This property can be useful to overcome the thermodestabilizing effect of cyclen moiety and to design hybrids possessing two functionalities. Shedding more light on the source of the observed effects will require additional studies on the interaction of polyamines with duplexes but will provide valuable insight into the key design requirements for such conjugates and their future applications in biological systems.

Conclusions
We have developed a new protocol for the synthesis of cyclen-containing oligonucleotides by post-synthetic coupling of azide-functionalized labels to 2 -O-propargylated oligonucleotides using Cu(I)-catalyzed azide-alkyne cycloaddition. All dye-containing oligonucleotides have a positive effect on the thermal stability of the obtained duplexes, especially those containing mismatches on one of the strands. The T m amplitude depends on the number and position of the attached dye molecules. The presence of the cyclen moiety in one of the strands decreases the melting temperature by approximately the same value in the range from −4.0 to −5.5 • C. This destabilization effect can, however, be diminished by the presence of a dye molecule in the complementary strand. Compensating for the destabilizing effect of cyclen (and possibly other polyamines) on dsDNA by inclusion of an intercalating dye is a promising tool for adjusting the thermal stability of polyamine-labeled DNA duplexes.

Data Availability Statement:
The data presented in this study are available upon request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.