Synthesis of Aminoglycoside-2′-O-Methyl Oligoribonucleotide Fusions

Phosphoramidite building blocks of ribostamycin (3 and 4), that may be incorporated at any position of the oligonucleotide sequence, were synthesized. The building blocks, together with a previously described neomycin-modified solid support, were applied for the preparation of aminoglycoside-2′-O-methyl oligoribonucleotide fusions. The fusions were used to clamp a single strand DNA sequence (a purine-rich strand of c-Myc promoter 1) to form triple helical 2′-O-methyl RNA/DNA-hybrid constructs. The potential of the aminoglycoside moieties to stabilize the triple helical constructs were studied by UV-melting profile analysis.


Introduction
Aminoglycosides are well-known small molecular ligands for a variety of RNA targets [1] (including bulges and internal loops at ribosomal decoding site [2][3][4][5], several ribozymes [6,7] and important regions of HIV RNAs [8][9][10][11]), and they also show relatively high affinities as groove binders for DNA-and RNA-triple helices and for their hybrids [12][13][14]. Thanks to these binding properties, they may be attractive conjugate groups for oligonucleotide-based probes to provide an extra binding motif in the recognition of the target DNA or RNA [15][16][17][18][19][20]. For example, neomycin has been used to enhance affinity of oligonucleotides to an α-sarcin loop RNA sequence [21] and to HIV-1 trans activation response element (TAR) models [22] via binding to known binding sites for neomycin on these RNA targets. In a favorable case, the cooperative recognition via combined small molecular binding and hybridization may take place [22]. Triple helical recognition of DNA has also been enhanced by appropriately conjugated neomycin ligands [23]. Furthermore, aminoglycoside moieties may improve cellular uptake via lipid-mediated delivery of oligonucleotides [24]. All together, these beneficial properties of the conjugated aminoglycosides may find applications in developing of modern antigene and antisense therapies.
We have previously described aminoglycoside-derived phosphoramidite building blocks (1 and 2) [17] and solid supports (5-7) [22], which may be used for the automated synthesis of 5 -and 3 -aminoglycoside conjugated oligonucleotides, respectively ( Figure 1). Herein, the set of the useful building blocks is expanded by appropriately modified ribostamycins (3 and 4) that may be incorporated at any position of the oligonucleotide sequence. The building blocks (3 and 4), together with a previously described neomycin-derived LCAA-CPG-support 7 (long chain alkylamine controlled pore glass), were used for the synthesis of pure fusions of the aminoglycosides and

Synthesis of Aminoglycoside 2′-O-Methyl Oligonucleotide Fusions Using 3 and 4
To evaluate the applicability of 3 and 4 in the automated chain assembly, intra-chain fusions of the ribostamycins and a random 2′-O-methyl RNA sequence (5′-GCUCA-R-UCUG-3′, ON1: R = α-ribostamycin residue (R α ) and ON2: R = ribostamycin residue (R β )) were first synthesized ( Figure 2). The coupling efficiency was evaluated by DMTr-assay. A double phosphoramidite coupling using 0.1 mol L −1 3 and 4 in acetonitrile, benzylthiotetrazol as an activator and a 600 s coupling time (2 × 600 s), followed by the standard oxidation step, gave ca. 95% coupling yield for both building blocks. Otherwise, the oligonucleotides were assembled using the standard RNA coupling cycle (a 300 s coupling time used for the 2′-O-methyl nucleoside building blocks). After the chain assembly, the levulinoyl groups of the ribostamycin moieties were removed on support with a mixture of hydrazinium acetate (NH2NH2·OH2, pyridine, AcOH, 0.124/4/1, v/v/v, 2 × 10 min at 25 °C), and the supports were then subjected to concentrated ammonia (overnight at 55 °C) [17]. The released ON1 and ON2 were analysed by ion exchange HPLC. As seen in the HPLC profiles of the crude products (Figure 2a,b), the automated synthesis could be successfully carried out. Interestingly, the chiral

Synthesis of Aminoglycoside 2 -O-Methyl Oligonucleotide Fusions Using 3 and 4
To evaluate the applicability of 3 and 4 in the automated chain assembly, intra-chain fusions of the ribostamycins and a random 2 -O-methyl RNA sequence (5 -GCUCA-R-UCUG-3 , ON1: R = α-ribostamycin residue (R α ) and ON2: R = ribostamycin residue (R β )) were first synthesized ( Figure 2). The coupling efficiency was evaluated by DMTr-assay. A double phosphoramidite coupling using 0.1 mol L −1 3 and 4 in acetonitrile, benzylthiotetrazol as an activator and a 600 s coupling time (2 × 600 s), followed by the standard oxidation step, gave ca. 95% coupling yield for both building blocks. Otherwise, the oligonucleotides were assembled using the standard RNA coupling cycle (a 300 s coupling time used for the 2 -O-methyl nucleoside building blocks). After the chain assembly, the levulinoyl groups of the ribostamycin moieties were removed on support with a mixture of hydrazinium acetate (NH 2 NH 2 ·OH 2 , pyridine, AcOH, 0.124/4/1, v/v/v, 2 × 10 min at 25 • C), and the supports were then subjected to concentrated ammonia (overnight at 55 • C) [17]. The released ON1 and ON2 were analysed by ion exchange HPLC. As seen in the HPLC profiles of the crude products ( Figure 2a,b), the automated synthesis could be successfully carried out. Interestingly, the chiral integrity of the ribostamycin moieties also affects the retention times of the conjugates in the HPLC profiles ( Figure 2c). Conjugates ON4-ON6, which aimed to clamp a purine-rich DNA strand (a sequence of C-Myc promoter 1), were then synthesized on our previously described neomycin-derived LCAA-CPG support (7) [22]. A manual coupling of 3 and 4 to support 7 was carried out (see the material and methods) and then the 2′-O-methyl oligoribonucleotide chain, including the 2′-deoxy oligonucleotide turn (TCTCT), was assembled in a standard manner. After the chain assembly, the levulinoyl groups of the ribostamycin moieties were removed on support using the hydrazine acetate treatment as mentioned above. The solid-supported conjugates were then exposed to a mixture of NaOMe in methanol (0.1 mol L −1 , for 2 h at 25 °C, i.e., acetyl removal of the neomycin moieties and cleavage of the succinyl linker), and the deprotection was continued by ammonolysis (overnight at 55 °C, i.e., removal of the Tfa-protections and of the Bz-protections of cytosine bases) [22]. Conjugates ON7 and ON8 were synthesized following the procedure described for ON1 and ON2.  Table S1), respectively. Isolated yields ranged from 10-26%.  Conjugates ON4-ON6, which aimed to clamp a purine-rich DNA strand (a sequence of C-Myc promoter 1), were then synthesized on our previously described neomycin-derived LCAA-CPG support (7) [22]. A manual coupling of 3 and 4 to support 7 was carried out (see the material and methods) and then the 2 -O-methyl oligoribonucleotide chain, including the 2 -deoxy oligonucleotide turn (TCTCT), was assembled in a standard manner. After the chain assembly, the levulinoyl groups of the ribostamycin moieties were removed on support using the hydrazine acetate treatment as mentioned above. The solid-supported conjugates were then exposed to a mixture of NaOMe in methanol (0.1 mol L −1 , for 2 h at 25 • C, i.e., acetyl removal of the neomycin moieties and cleavage of the succinyl linker), and the deprotection was continued by ammonolysis (overnight at 55 • C, i.e., removal of the Tfa-protections and of the Bz-protections of cytosine bases) [22]. Conjugates ON7 and ON8 were synthesized following the procedure described for ON1 and ON2. RP HPLC and MS (ESI-TOF) data of the conjugates ON1, ON2, ON4-ON8 are shown in Figures 2a,b and 3 (and Table S1), respectively. Isolated yields ranged from 10-26%.

UV-Melting Profile Analysis
The effect of the aminoglycosides moieties (ribostamycin and neomycin) to stabilize the clamp structures targeted to a purine-rich DNA single strand (a sequence of c-Myc promoter 1) (cf. Scheme 2) has been studied by UV-melting profile experiments ( Figure 4 and Table 1). The measurements were carried out using 2 µmol L −1 of each oligonucleotide in a mixture of 10 mmol L −1 sodium cacodylate and 0.1 mol L −1 NaCl at pH 6.0 and 7.0. The temperature was changed at a rate of 0.2 °C min −1 . In each case, a biphasic melting curve was observed (the duplex melting range, Tm = 65 °C, excluded in Figure 4) and the Tm 3 -values were extracted from the first inflection point. The Tm 3 -values of the

UV-Melting Profile Analysis
The effect of the aminoglycosides moieties (ribostamycin and neomycin) to stabilize the clamp structures targeted to a purine-rich DNA single strand (a sequence of c-Myc promoter 1) (cf. Scheme 2) has been studied by UV-melting profile experiments ( Figure 4 and Table 1). The measurements were carried out using 2 µmol L −1 of each oligonucleotide in a mixture of 10 mmol L −1 sodium cacodylate and 0.1 mol L −1 NaCl at pH 6.0 and 7.0. The temperature was changed at a rate of 0.2 °C min −1 . In each case, a biphasic melting curve was observed (the duplex melting range, Tm = 65 °C, excluded in Figure 4) and the Tm 3 -values were extracted from the first inflection point. The Tm 3 -values of the

UV-Melting Profile Analysis
The effect of the aminoglycosides moieties (ribostamycin and neomycin) to stabilize the clamp structures targeted to a purine-rich DNA single strand (a sequence of c-Myc promoter 1) (cf. Scheme 2) has been studied by UV-melting profile experiments ( Figure 4 and Table 1). The measurements were carried out using 2 µmol L −1 of each oligonucleotide in a mixture of 10 mmol L −1 sodium cacodylate and 0.1 mol L −1 NaCl at pH 6.0 and 7.0. The temperature was changed at a rate of 0.2 • C min −1 .
In each case, a biphasic melting curve was observed (the duplex melting range, T m = 65 • C, excluded in Figure 4) and the T m 3 -values were extracted from the first inflection point. The T m 3 -values of the triple helical clamps were expectedly higher (ca. 10 • C) at pH 6.0 than at pH 7.0. As shown, the overhanging 3 -aminoglycoside moiety of the conjugates increased the stability of the clamps in each case. The acidic conditions (pH 6.0 vs. 7.0) did not show a marked role in ∆T   triple helical clamps were expectedly higher (ca. 10 °C) at pH 6.0 than at pH 7.0. As shown, the overhanging 3′-aminoglycoside moiety of the conjugates increased the stability of the clamps in each case. The acidic conditions (pH 6.0 vs. 7.0) did not show a marked role in ∆Tm 3 -values. The 3′-neomycin conjugate (ON4) increased the stability of the clamps (Target 1 and Target 2) by ∆Tm 3 = +4.0-+5.8 °C. The incorporation of the ribostamycin unit to the 3′-aminoglycosides overhang (ON5 and ON6) seemed to elicit, however, slightly decreased ∆Tm 3 -values (e.g., ON4: +5.8 °C vs. ON5: +4.8 °C and ON6: +5.5 °C with Target 2 at pH 7.0). ON6 did an exception with Target 1 at pH 7.0 (ON6: ∆Tm 3 = +5.1 vs. ON4: ∆Tm 3 = +4.0). A slightly increased stability was also observed, when the ribostamycin was incoporporated into the 2′-deoxy oligoribonucleotide turn (ON7 and ON8: ∆Tm 3 = +2.6 °C-+4.3 °C).

Synthesis of the Building Blocks 3 and 4 and of the Aminoglycoside-Oligonucleotide Conjugates ON1, ON2, ON4-ON8
Phosphoramidite building blocks 3 and 4 could be synthesized in relatively high yields (Scheme 1). Despite the multistep synthesis, the overall yields of 3 and 4 (calculated from 6,3 ,4 -tri-O-acetyl-1,3,2 ,6 -tetraazido neamine 9 [26]) were 19% and 10%, respectively. The chiral integrity of the anomers was confirmed by a NOESY spectrum that showed a correlation between the H1 and 2 -O-methoxy groups of the β 1 −5 -ribostamycins (e.g., 13). Building blocks 3 and 4 could be efficiently incorporated into oligonucleotide sequences using either an automated double phosphoramidite coupling (2 × 600 s coupling time) or a manual coupling (see details in the Materials and Methods section). In the manual coupling, the concentration of the building blocks could be increased to 0.11 mol L −1 (after the addition of benzylthiotetrazol) that improved the coupling efficiency (Note: In the synthesizer, the initial concentration of the phosphoramidites is 0.1 mol L −1 that is diluted to one half by the solution of benzylthiotetrazol). The O-levulinoyl/N-trifluoroacetyl (Lev/Tfa)-protecting group combination for 3 and 4 was applied. The reason for this protecting group scheme was the selective on-support removal of the Lev protections (by a hydrazinium acetate treatment in the presence of N-Tfa groups) that suppressed plausible O→N acyl migration. The NaOMe-catalyzed methanolysis was sufficient to selectively remove O-acetyl groups from the neomycin moiety (7) and to eliminate N-acylated side products (ON4) [22], but the Lev/Tfa-combination is evidently [17] the superior protecting group scheme, when the number of 1,2-aminoethanol-moieties increases. Thus, two-step-(ON1, ON2, ON4, ON7 and ON8) and three-step-treatments (ON5 and ON6) with hydrazinium acetate, NaOMe/MeOH and ammonolysis were used to release the conjugates.

The Effect of the Aminoglycoside Moieties of ON1, ON2, ON4-ON8 on the Triple Helical Constructs
According to UV-melting profiles, the aminoglycoside moieties (ribostamycin and neomycin) increased the stability of the clamp structures in each case (∆T m = +2.2-+5.8 • C), but the effect remained modest. The Watson-Hoogsteen groove (the groove between the pyrimidine strands) of the DNA-triple helix may bind multiple aminoglycosides (neomycin primarily) and the binding has been proposed to be involved in the amino groups of neomycin rings II and IV [27]. The elongated 3 -aminoglycoside overhang of ON5 and ON6 (containing one biosamine and two neamines in the ribose-phosphodiester backbone) probably could not reach the optimal binding contact needed for the groove binding and the stability did not increase compared to 3 -neomycin moiety (ON4: ∆T m 3 = +4.0-+5.8 • C).
The phosphodiester bond between the neomycin and ribostamycin units may also disturb the binding. The incorporation of the ribostamycin units into the 2 -deoxy oligonucleotide turn increased the stability of the clamps by ∆T m = +2.6-+4.3 • C. The stability of the clamps may hence be further increased by incorporation of aminoglycosides at both terminus of the clamp. Further studies may also be needed to evaluate the influence of the longer spacers between the oligonucleotide and the aminoglycoside overhang.

General Remarks
MeCN, pyridine and dichloromethane were dried over 3Å molecular sieves and triethylamine over CaH 2 . NMR spectra were recorded using a 500 MHz instrument. The chemical shifts for 1 H and 13 C NMR resonances are given in parts of million from the residual signal of the deuterated solvents (CD 3 OD and CD 3 CN). 31 P NMR resonance shifts are compared to external H 3 PO 4 . Mass spectra were recorded using electrospray ionization (ESI-TOF).

UV-Melting Temperature Studies
The melting curves (absorbance vs. temperature) were measured at 260 nm on a PerkinElmer Lambda 35 UV-Vis spectrometer equipped with a multiple cell holder and a Peltier temperature controller. An internal thermometer was also used. The temperature was changed at a rate of 0.2 • C min −1 . Each T m 3 -value was determined to be the maximum of the first derivative of the melting curve.

Conclusions
In this primarily synthetic study, synthesis of phosphoramidite building blocks of 2 -O-methyl ribostamycins 3 and 4 that may be incorporated at any position of the oligonucleotide sequence, have been described. According to DMTr-assay and HPLC analysis of the released conjugates, the building blocks (3 and 4) could be efficiently incorporated into oligonucleotide sequences using a double phosphoramidite coupling (2 × 600 s) and using benzylthiotetrazol as an activator. 3 and 4 and a neomycin-derived solid support (7) were used for the preparation of aminoglycoside conjugates of 2 -O-methyl and 2 -deoxy oligoribonucleotide hybrids that were aimed to clamp a purine-rich DNA single strand (a sequence of c-Myc promoter 1). The potential of the intrachain ribostamycin and 3 -multiaminoglycoside overhangs to act as groove binders to stabilize triple helical region of the DNA-2 -O-methyl RNA clamps was demonstrated. According to UV-melting profile analysis, slightly increased clamp stability was observed.
Supplementary Materials: The supplementary materials are available online.