Solid-Phase Synthesis of Boranophosphate/Phosphorothioate/Phosphate Chimeric Oligonucleotides and Their Potential as Antisense Oligonucleotides

In this study, we successfully synthesized boranophosphate (PB), phosphorothioate (PS), and phosphate (PO) chimeric oligonucleotides (ODNs) as a candidate for the antisense oligonucleotides (ASOs). The PB/PS/PO-ODNs were synthesized utilizing H-boranophosphonate, H-phosphonothioate, and H-phosphonate monomers. Each monomer was condensed with a hydroxy group to create H-boranophosphonate, H-phosphonothioate, and H-phosphonate diester linkages, which were oxidized into PB, PS, and PO linkages in the final stage of the synthesis, respectively. As for condensation of an H-phosphonothioate monomer, regulating chemoselectivity was necessary since the monomer has two nucleophilic centers: S and O atoms. To deal with this problem, we used phosphonium-type condensing reagents, which could control the chemoselectivity. In this strategy, we could synthesize PB/PS/PO oligomers, including a 2′-OMe gapmer-type dodecamer. The physiological and biological properties of the synthesized chimeric ODNs were also evaluated. Insights from the evaluation of physiological and biological properties suggested that the introduction of suitable P-modification and sugar modification at proper sites of ODNs would control the duplex stability, nuclease resistance, RNase H-inducing ability, and one base mismatch discrimination ability, which are critical properties as potent ASOs.


■ INTRODUCTION
Many efforts have been devoted to the development of antisense oligonucleotides (ASOs) since it is demonstrated that an oligonucleotide that is complementary to a target mRNA could control the translation of mRNA into a protein. 1,2 There are two kinds of ASOs according to a mechanism of translation regulation, namely, a steric blocking type and an RNase H-dependent type. RNase H is an endonuclease located mainly in the nucleus, recognizes DNA/ RNA duplexes, and selectively cleaves the RNA strand. 3,4 Required properties for potent RNase H-dependent ASO include high nuclease resistance, duplex stability, RNase Hinducing ability, high retention in tissues, and low cytotoxicity. Since the properties of ASOs can be modulated by introducing chemical modifications to phosphate (PO) moieties, sugar, and the nucleobase of nucleotides, 5 many groups have investigated a wide variety of modifications. A phosphorothioate (PS) backbone, in which one of the nonbridging oxygen atoms is replaced with a sulfur atom, is the most used chemical modification applied to ASOs. PS is still widely used as a chemically modified analogue owing to its high nuclease resistance and the facility of its synthesis. In addition to this, it has been demonstrated that PS linkages are crucial for improving the pharmacokinetics of ASOs due to their high affinity with certain kinds of protein such as serum albumin. 6,7 For example, it has been found that PS linkages prevent ASOs from glomerular filtration clearance, which led to the extension of blood retention. In addition to this, ASOs containing PS linkages bind specific proteins outside the cell, which leads to the uptake of ASOs into cells, 8 and intracellular proteins determining the intracellular distribution of ASOs. 9 However, it has been reported that PS linkages reduce duplex stabilities of ASOs with target mRNAs. Moreover, some PS oligonucleotides (ODNs) are cytotoxic and trigger undesired immunoresponse events, which are major obstacles for clinical trials. 10,11 To improve duplex stability, nucleotides with sugar modifications have been introduced, such as 2′-O-modifications including 2′-OMe and 2′-O-MOE, 12 and locked nucleic acids (LNAs). 13−15 These modifications are typically used for gapmer-type ASOs, namely, the nucleotides with a sugar modification are placed in the 5′-and 3′-end (wing) regions of an ASO to acquire a duplex stability with a complementary RNA while deoxyribonucleotides are located in the central (gap) region to maintain an RNase H-inducing ability 16 Some ASOs that were approved by the Food and Drug Administration (FDA) contain both PS linkages and sugar modifications. 17 This suggests that a combination of Pmodifications and sugar modifications dramatically improves their properties as ASOs. Alternatively, although some reports suggested that 2′-O-MOE gapmers containing PS linkages lower their cytotoxicity, 18 further suppression of cytotoxicity would be needed for the more secured ASOs.
Under these circumstances, boranophosphate (PB), in which a nonbridging oxygen atom of a PO linkage is replaced with a borano group, has received a lot of attention as another promising ASO candidate since PB modification offers higher nuclease resistance than the PS counterpart 19 and exhibits low cytotoxicity. 20,21 However, a full PB modification reduces the duplex stability of ASOs with target mRNAs and RNase Hinducing activities. 19,22−24 To overcome these problems, Caruthers et al. 24−26 and our group 27 introduced both PB and PO linkages in ODNs (PB/ PO chimeric ODNs). PB/PO chimeric ODNs have improved duplex stability and some of them show higher RNase Hinducing activity compared to the fully PB-modified counterparts. Notably, PB/PO chimeric ODNs have substantial nuclease resistance. 25 These indicate that P-modified chimeric ODNs can take advantage of each other's strength. This strategy utilizing chimeric ODNs also works for ODNs containing PS linkages. The research from IONIS Pharmaceuticals, Inc. revealed that replacing a few PS linkages with alkylphosphonate linkages significantly reduces toxicity while maintaining the antisense activity of ASOs. 28 Furthermore, replacing a few PS linkage with mesylphosphoramidate 29 linkages can also suppress toxicity and in certain cases improve the antisense activity of ASOs. 30 Taking that into consideration, we expected that introducing PB and PO linkages to PS-ODNs (PB/PS/PO and/or PB/PS chimeric ODNs) would be a promising way to modulate cytotoxicity along with maintaining the favorable pharmacokinetics of ODNs.
Although PB derivatives have been viewed as promising ASO candidates, examples for their synthesis are still limited. The biggest hurdle for the synthesis of PB derivatives is the fact that acyl-type amino protecting groups on nucleobases are not compatible with the synthesis of boranophosphate by the general phosphoramidite method since N-acyl groups were easily reduced with a boronating reagent to N-alkyl groups 31 which cannot be removed. To deal with this problem, Caruthers et al. and our group obtained PB derivatives in different synthetic strategies. Caruthers et al. have synthesized PB-ODNs via the phosphoramidite method using N-di-tertbutylisobutylsilyl (N-BIBS)-protected phosphoramidite monomers. This protecting group is stable under the general reaction conditions of the phosphoramidite method 26 and tolerant toward boronation. In contrast, we have developed the H-boranophosphonate method using an H-boranophosphonate monoester, which contains characteristic H−P → BH 3 groups as monomer units. 32,33 In this method, an Hboranophosphonate monomer unit is condensed with a 5′hydroxy group using a condensing reagent to form an Hboranophosphonate diester linkage followed by detritylation step without a transformation of the resultant internucleotidic linkages. These two steps are repeated and after the designed length is achieved, all internucleotidic H-boranophosphonate diesters are oxidized to PB linkages by treatment with CCl 4 and water in the presence of a base, followed by the removal of amino protecting groups and release from a solid support. Caruthers's and our synthetic strategies were also applicable to the synthesis of PB/PO chimeric ODNs. We have synthesized PB/PO chimeric ODNs using H-boranophosphonate monomers and H-phosphonate monomers. 27 34,36 From these insights, we expected that an introduction of PS linkages would be possible by utilizing H-phosphonothioate monoesters and the proper oxidation conditions. The major problem associated with using Hphosphonothioate monoesters is the chemoselectivity of a condensation reaction since H-phosphonothioate monoesters have two different nucleophilic centers, namely, sulfur and oxygen atoms. Desired PS derivatives are obtained by the Oactivation followed by oxidation while the S-activation results in the formation of PO derivatives as byproducts. Stawinski et al. solved the problem using a chlorophosphate derivative, as a condensing reagent since soft S-nucleophiles have low reactivity toward the hard phosphorus centers. 36 However, there are few reports using an H-phosphonothioate monomer for solid-phase synthesis. 37−39 Hence, we tried to synthesize H-PS internucleotidic linkages on a solid support using phosphonium-type condensing reagents that is expected to avoid the S-activation since these have a hard character. In this research, we demonstrated that this synthetic strategy was applicable to the synthesis of PB/PS/PO chimeric ODNs including gapmer-type ODNs, which have 2′-OMe-modified nucleotides in the wing region.
Solid-Phase Synthesis of Phosphorothioate Dimers. Next, we investigated the reaction conditions for solid-phase synthesis using the H-phosphonothioate monomer 5t (Scheme 3). The reaction conditions for the solid-phase synthesis using H-boranophosphonate and H-phosphonate monomers had already been optimized in our previous report. 27 First, the Hphosphonothioate monomer 5t (0.1 M) was condensed with the 5′-hydroxy group of thymidine on a highly cross-linked polystyrene (HCP) support 42 via a succinyl linker using CH 3 CN as a solvent and a phosphonium-type condensing reagent (0.25 M) in the presence of a base, then detritylation was conducted by treatment with 3% dichloroacetic acid (DCA) in CH 2 Cl 2 in the presence of Et 3 SiH as a trityl cation scavenger. 43 Afterward, oxidation of the resultant Hphosphonothioate linkage was carried out using a mixture of CCl 4 and H 2 O in the presence of triethylamine as a base and 2,6-lutidine as a cosolvent. Finally, deprotection of the N 3benzoyl group and cleavage of the linker by treatment with concentrated aqueous NH 3 and EtOH afforded the PS dimer. The crude mixture was analyzed by reversed-phase highperformance liquid chromatography (RP-HPLC). The chemoselectivity was estimated by the area ratios of the PS diester and the PO diester which were derived from the S-activation. In addition to this, the HPLC yield was estimated by the area ratios of the PS diester (8) to unreacted thymidine and the PO diester (7).
First, the effect of condensing reagents on the reaction was investigated ( Table 1, entries 1−3). In these experiments, 2,6lutidine was used as a base in the condensation reaction. When the H-phosphonothioate monomer was condensed by 2- 44 which has HOBt as a leaving group, only a trace amount of the PS and the PO diester was obtained (entry 1). On the other hand, the use of condensing reagents such as 1, 45 (entry 2) and 3-nitro1,2,4triazol-1-yl-tris(pyrrolidin-1-yl) phosphonium hexafluorophosphate (PyNTP) 45 (entry 3) which have 3-nitro 1,2,4-triazole (NT) as a leaving group afforded PS diester with over 90% HPLC yields. These results suggested that the presence of NT was critical for the condensation of the H-phosphonothioate monomer and the 5′-hydroxy group. Compared with MNTP, PyNTP gave better chemoselectivity and condensation efficiency. Therefore, PyNTP was chosen as a condensing reagent for H-phosphonothioate monomers in the following investigations.
Next, bases for the condensation reaction were examined (entries 3−7). It was found that the use of the weaker base (pyridine, entry 4, quinoline, entry 5) slightly improved chemoselectivities while maintaining high condensation efficiency. Raising the ratio of quinoline in the reaction solvent had a marginal effect on the reaction outcome (entry 6). Surprisingly, the condensation reaction in the absence of base improved both coupling yield and chemoselectivity (entry 7). Therefore, we concluded that the utilization of PyNTP in the presence of 1.8 M quinoline or without a base was the optimum conditions for the condensation reaction. It has been shown that MNTP had a higher activity as a condensing reagent for phosphorylation and phosphonylation reactions than PyNTP, 45 likely due to less steric hindrance and the strained cyclic structure of the phosphonium center. However, for the condensation reaction using the H-phosphonothioate 5t, PyNTP gave better chemoselectivity and condensation efficiency. In addition, the use of a stronger base for the condensation reaction was found to cause lower chemoselectivity and condensation efficiency. From these observations, an overactivation of the monomer 5t seemed to affect the reaction outcome. The activation of 5t by MNTP might be prone to lead the overactivation as shown in Scheme 4 and resulted in inferior chemoselectivity and condensation efficiency. High basicity of the reaction medium may also cause the overactivation of the monomer.
The optimized reaction conditions were then applied to monomers containing other nucleobases (Scheme 5 and Table  2). The deoxyadenosine 5a, deoxycytidine 5c, and deoxyguanosine 5g H-phosphonothioate monomers were used to     In addition to this, we carried out the solid-phase synthesis of PB dimers using 2′-O-methyl-3′-H-boranophosphonate monomers (6a, 6c, 6g, and 6u) for the synthesis of gapmertype PB/PS/PO chimeric oligonucleotides. 2′-O-Methyl-3′-Hboranophosphonate monomers were condensed with the 5′hydroxy group of thymidine on an HCP support via a succinyl linker under the same conditions with 2′-deoxynucleoside counterparts using MNTP as a condensing reagent in the presence of 2,6-lutidine (Scheme 6). The following procedures were the same as in the synthesis of T PS T. Dimers A PB T, C PB T, G PB T, and U PB T (underline indicates 2′-OMe nucleoside) were obtained in 95−98% HPLC yields (Table 3), indicating that the condensation reaction proceeded efficiently regardless of the presence of 2′-O-modification.

Solid-Phase Synthesis of PB/PS/PO Chimeric ODNs.
Next, to elucidate whether the synthetic strategy is applicable to the synthesis of PB/PS/PO chimeric ODNs, the synthesis of tetramers (d(C PO A PB G PS T) (19) and d(C PS A PB G PO T) (20)) containing PB, PS, and PO linkages was conducted using the H-phosphonothioate, H-boranophosphonate, and Hphosphonate monomers. The cycle consisting of the condensation and detritylation was repeated, and oxidation of internucleotidic linkages followed by removal of the amino protecting groups and cleavage of the linker afforded the tetramers. It was confirmed that the tetramers were formed as main products by the RP-HPLC analysis of the reaction mixtures ( Figure S4). These results indicated that these different internucleotidic linkages were simultaneously oxidized without side reactions, and thus this strategy enables the synthesis of PB/PS/PO chimeric ODNs.
These results prompted us to synthesize a PB/PS/PO c h i m e r i c D N A d o d e c a m e r , a n d d -(C PS A PS G PS T PS C PB A PB G PB T PB C PO A PO G PO T) (21) was chosen as a synthetic target to demonstrate the potential of the strategy, since 21 contains almost all of the potential combination of internucleotidic linkages and nucleobases. The dodecamer 21 was synthesized by repeating the condensation and detritylation cycles, oxidation, and the deprotection and the release step and was isolated in 6% yield (Table 4, entry 3). Encouraged by the success, we designed and synthesized types of chimeric ODNs containing PB modifications (Table 4, entries 4−6). These sequences were antisense sequences to apoB protein mRNA 46    . This result indicated that some 2′-O-modified gapmers would be synthesized by this synthetic strategy. It is worth noting that although we had purified PB containing ODNs by anion-exchange HPLC, it was found that RP-HPLC purification using a mixture of methanol and a buffer containing hexafluoroisopropanol and triethylamine as an eluent was effective for isolation of these ODNs. 47 Thus, all of the ODNs but 22 were purified by RP-HPLC.
Although there is room for further improvement of yields, some kinds of chimeric ODNs were successfully synthesized in the strategy. Hybridization Properties. Next, we moved on to the evaluation of properties that are crucial for ASOs. Nuclease Resistance. Second, nuclease digestion experiments were conducted using snake venom phosphodiesterase (SVPDE) from Crotalus adamanteus venom, a representative of 3′-exonuclease. Aqueous solutions of each ODN (22−29) were treated with SVPDE solution (0.4 U/mL) at 37°C for 12 h. After SVPDE was denatured at 95°C, the mixtures were analyzed by RP-HPLC. The results are shown in Figures 2 and  S21. The PO-ODN was completely digested while the PS-ODN was partially degraded in the same conditions, respectively ( Figure S21). On the other hand, the PB-ODN remained almost intact ( Figure S21). Therefore, it was suggested that the order of nuclease resistance toward SVPDE was PB-ODN > PS-ODN > PO-ODN, which was in good agreement with the previous report. 19 On the other hand, the PB/PO-ODN, PS/PO-ODN, PB/PS/PO-ODN, and PB/ PS/PO-gapmer, which have three or four consecutive PO linkages at the central positions, were completely digested and their full-length ODNs were not detected by RP-HPLC (Figures 2 and S21). Although SVPDE is known as a 3′exonuclease, there are reports that SVPDE also has endonuclease activity. 48−50 Hence, SVPDE may recognize consecutive PO linkages at the central position as endonuclease. In sharp contrast, there was a slight sign of decomposition of the PB/PS-ODN indicating that combination of PB and PS modifications confers substantial nuclease resistance to ODNs (Figure 2). In the nuclease digestion experiments, it was suggested that the ODNs containing a higher ratio of PO linkages would significantly reduce nuclease resistance.
RNase H Activity. Finally, we studied the effect of ODNs on RNase H activity, which is crucial for the efficacy of RNase H-dependent antisense therapeutics. Escherichia coli RNase H was used for the experiments. Aqueous solutions of each ODN (22−29) and 10 equiv of cRNA were treated with 25 or 50 U/ mL RNase H at 37°C for 30 min. After RNase H was denatured at 95°C, the mixtures were analyzed by RP-HPLC. In the assay using 25 U/mL RNase H, although almost all of the ODNs did not show clear effects on RNase H activities, the PO-ODN (26) and the PB/PS/PO-gapmer (25) were able to induce cleavage of the cRNA strand by RNase H to some extent since small fragments of cRNA were detected by RP-HPLC ( Figure S22). Compared with the PO-ODN, the PB/ PS/PO-gapmer showed slightly better RNase H activity. In the assay using 50 U/mL RNase H, almost all cRNA was digested by RNase H with all ODNs (Figures 4 and S23). Hence, we calculated the amount of the intact cRNA compared with benzamide as an internal standard based on the area ratio of each peak in HPLC profiles. In the presence of PB/PS/POgapmer, PB/PS/PO, PS/PO, PS, and PO-ODNs, over 95% of cRNA was cleaved by RNase H. In contrast, when using PB/ PS, PB/PO, and PB-ODN, approximately 90% of cRNA was cleared by RNase H. From these results, it was suggested that  On the other hand, in the RNase H assay, two pairs of cleaved fragments, namely, p4mer and 8mer (cleavage site a in Figure 3) and p5mer and 7mer (cleavage site b in Figure 3) (p indicates phosphate group at the 5′ end of fragments) were mainly detected by RP-HPLC and electrospray ionization mass spectrometry (ESI-MS) analysis. The peaks derived from the fragments were shown in Figure 3 (fragment A was from cleavage of RNA (30) at cleavage site a; fragment B was from cleavage of RNA (30) at cleavage site b). The area ratios of the peaks A and B in HPLC profiles were shown in Table 6 and the representative profiles were shown in Figure 4. In the presence of ODNs without PB modification, the area ratios of peaks A and B were approximately 1:2 (Table 6,

■ CONCLUSIONS
We developed an efficient synthetic strategy for PB/PS/PO chimeric ODNs without a limitation of the nucleobase and position of P-modification utilizing H-boranophosphonate, Hphosphonothioate, and H-phosphonate monomers. In addition to this, we showed that the strategy was applicable to the synthesis of a PB/PS/PO chimeric 2′-OMe gapmer. One of    The Journal of Organic Chemistry pubs.acs.org/joc Article the key points pertaining to the synthesis of these chimeric ODNs was regulating the chemoselectivity of the condensation reaction using an H-phosphonothioate monomer. It was found that phosphonium-type condensing reagents having NT as a leaving group, especially PyNTP, provided excellent chemoselectivity. The properties of obtained ODNs were also evaluated. As for a thermal denaturation study, introducing PO linkages to PS-ODN, PB-ODN, and PB/PS-ODN (PS/ PO, PB/PO, and PB/PS/PO chimeric ODNs) improved their duplex stability. An SVPDE assay revealed that the introduction of three or four consecutive PO linkages reduced their nuclease resistance significantly. In the RNase H activity assay, it was suggested that ODNs containing a higher ratio of the PB linkages reduced their RNase H activity to some extent, and altered the cleavage site preference of RNase H. From these results, it was expected that introducing proper Pmodifications at appropriate sites of ASOs would regulate the duplex stability, nuclease resistance, RNase H activity, and one base mismatch discrimination. In addition, this synthetic strategy would also work for the synthesis of any other gapmers containing LNA or 2′-O-MOE which are widely used sugar modifications of ASOs. Thus, we expect that a PB/PS/ PO chimeric ODN is one of the promising candidates of potent ASOs. Hence, the details of the biological and physiological properties of PB/PS/PO chimeric ODNs are under investigation.

■ EXPERIMENTAL SECTION
General Information. All reactions were conducted under an Ar atmosphere. Dry organic solvents were prepared by appropriate procedures. A thermostatic chamber was used for the reaction run at 50°C. 1 H NMR spectra were recorded at 400 MHz with tetramethylsilane (δ 0.0) as an internal standard in CDCl 3 . 13 C NMR spectra were recorded at 100 MHz with CDCl 3 (δ 77.0) as an internal standard in CDCl 3 . 31 P NMR spectra were recorded at 162 MHz with H 3 PO 4 (δ 0.0) as an external standard in CDCl 3 . Analytical thin-layer chromatography was performed on commercial glass plates with a 0.25 mm thickness silica gel layer. Silica gel column chromatography was performed using spherical, neutral, 63−210 μm silica gel. Manual solid-phase synthesis was carried out using a glass filter (10 mm × 50 mm) with a stopper at the top and a stopcock at the bottom as a reaction vessel. Synthesized dimers were analyzed by reversed-phase HPLC. Synthesized oligomers (tetramer and dodecamers) were analyzed by reversed-phase HPLC and/or anion-exchange HPLC, purified by reverse-phase HPLC or anionexchange HPLC, and identified by electrospray ionization (ESI) mass spectroscopy. Isolated yields of oligomers were estimated by measuring UV−vis spectra. PO-ODN (26), PS-ODN (27), PS/PO-ODN (29), and cRNA (30) were purchased and used for denaturation, nuclease resistance, and RNase H activity tests without further purification.
Copies of 1