The synthesis of 15N(7)-Hoogsteen face-labeled adenosine phosphoramidite for solid-phase RNA synthesis

Abstract We have developed an efficient route for the synthesis of 15N(7)-labeled adenosine as phosphoramidite building block for site- and atom-specific incorporation into RNA by automated solid-phase synthesis. Such labeled RNA is required for the evaluation of selected non-canonical base pair interactions in folded RNA using NMR spectroscopic methods. Graphical abstract Electronic supplementary material The online version of this article (doi:10.1007/s00706-016-1882-8) contains supplementary material, which is available to authorized users.


Introduction
Numerous reports on the synthesis of atom-specific 15 Nlabeled nucleosides exist in the literature; however, procedures for efficient access to the corresponding phosphoramidites for RNA solid-phase synthesis are rare [1][2][3]. We have recently described our preferred synthetic routes for 15 N(1)-adenosine, 15 N(1)-guanosine, 15 N(3)uridine, and 15 N(3)-cytidine phosphoramidites which allow base pair-specific labeling in RNA for direct monitoring of Watson-Crick base pairs by 1 H/ 15 N/ 15 N-COSY experiments [4]. The approach of individual Watson-Crick base pair labeling is particularly useful for the analysis of conformationally flexible RNAs when competing and interconverting secondary structures are encountered [5,6].
Along the same line, the evaluation of more complex base pair interactions such as base triplets or quartets, and the underlying dynamics in solution is important to understand functional RNA structures. Thereby, selective nucleobase labeling is again advantageous for NMR spectroscopic investigations to directly spot the interaction of interest which in many cases involves the Hoogsteen face of one or more purine nucleosides. This is especially true for larger functional RNAs with complex folding, such as riboswitch aptamer domains and ribozymes, where spectral crowding can make the assignment procedure very labor-intensive or even impossible. Here, we present an optimized procedure to synthesize 15 N(7)-labeled adenosine phosphoramidite. Additionally, a potential application to probe a cis Watson-Crick/Hoogsteen base pair interaction [7] is demonstrated for a base triplet that has been observed in the crystal structure of the env22 twister ribozyme [8].
Electronic supplementary material The online version of this article (doi:10.1007/s00706-016-1882-8) contains supplementary material, which is available to authorized users.

Results and discussion
To achieve 15 N(7)-labeled adenosine amidite 12, we conceived a strategy that employs a silyl-Hilbert-Johnson nucleosidation [9][10][11] and a recently introduced azido-toacetamido purine transformation [4] as key steps. Therefore, 15 N(7)-hypoxanthine 5 was synthesized following the protocol by Jones and coworkers (Scheme 1) [12]. We started with sodium ethoxide-mediated cyclization of thiourea and ethyl cyanoacetate to form 6-amino-2-mercapto-pyrimidone (1) in high yields [13]. Nitrosylation of compound 1 installed the 15 N-label by electrophilic substitution using the cost-effective isotope source Na 15 NO 2 in aqueous acid. The deep red nitroso compound 2 precipitated and was directly reduced to the colorless diamino mercapto pyrimidone 3 with dithionite [12]. Subsequent desulfurization with activated nickel sponge in dilute aqueous ammonia yielded compound 4 [12]. Treatment with formic acid and diethoxymethyl acetate at elevated temperature resulted in formation of the imidazo moiety to furnish the desired 15 N(7) hypoxanthine 5 [12]. In our hands, the 5-step reaction sequence proceeded in 75% overall yield and was conducted at multigram scales without the need for chromatographic purifications.
The nucleosidation reaction of 15 N(7)-hypoxanthine 5 with 1-O-acetyl-2,3,5-tri-O-benzoyl-ß-D-ribofuranose under Vorbrüggen conditions proceeded in the presence of N,O-bis(trimethylsilyl)acetamide and trimethylsilyl triflate to give tribenzoylated 15 N(7) inosine 6 as the major product, isolated after column chromatography in 56% yield. N1-, N3-, N7-, and/or O6-glycosylated isomers form as byproducts in minor amounts, consistent with previous reports [10,11] and in accordance to the qualitative comparison of H-C8 1 H NMR resonances observed in the product mixture (Supplementary Material). The integrity (C1 0 -N9 connectivity) of compound 6 was further supported by NMR spectroscopic comparison of 6 with an authentic sample that was prepared by direct benzoylation of commercial inosine (Supplementary Material). Next, we activated the carbonyl group of compound 6 in analogy to a procedure that was originally developed by Wan and coworkers [14][15][16] for the mild activation of cyclic amides with (benzotriazol-1-yloxy)tris(dimethylamino)phosphonium hexafluorophosphate (BOP). Thereby, O 6 -(benzotriazol-1-yl)inosine is formed which can be substituted by a nucleophilic displacement (S N Ar) reaction. These compounds can either be isolated [17] and used as convertible nucleosides after incorporation into DNA through solid-phase synthesis or directly derivatized with an appropriate electron-rich nucleophile [14][15][16]. For our route (Scheme 2), we used sodium azide to substitute the benzotriazolyl inosine intermediate. In the presence of Cs 2 CO 3 , this reaction proceeded smoothly in DMF to yield the 6-azido purine nucleoside 7 in 70% yield. Compound 7 was then reduced with thioacetic acid, resulting directly in the N 6 -acetyl protected 15 N(7) adenosine 8 in 84% yield. Then, selective deprotection of the ribose benzoyl groups was achieved in close to quantitative yield by treatment with aqueous NaOH in ethanol and pyridine. Functionalization of nucleoside 9 as building block for RNA solidphase synthesis started with the introduction of a 4,4 0dimethoxytrityl group on the ribose 5 0 -OH to give compound 10 (59% yield), followed by tert-butyldimethylsilylation of the ribose 2 0 -OH to furnish compound 11 (51% yield). Finally, phosphitylation was executed with 2-cyanoethyl-N,N-diisopropylchlorophosphoramidite (CEP-Cl) in the presence of N,N-diisopropylethylamine in CH 2 Cl 2 . Starting with the nucleosidation reaction of 15 N(7) hypoxanthine 6, our route provides building block 12 in a 10% overall yield in seven steps and with six chromatographic purifications; in total, 1.6 g of 12 was obtained in the course of this study.
Our motivation to synthesize 15 N(7) adenosine phosphoramidite 12 refers to the structural distinctions that have been found recently in the crystal structures of the twister ribozyme [8,[18][19][20]. Interestingly, while in the O. sativa twister ribozyme, the phylogenetically highly conserved four-base pair stem P1 was formed as predicted [18], in the env22 ribozyme; only two base pairs of stem P1 were observed; two nucleotides (U1 and U4) fold back to the core of the ribozyme and were involved in triplet interactions [8,19]. One of these triplets (U4ÁA49-A34) very close to the active site of the ribozyme (Fig. 1) can be considered to affect the active site conformation to support phosphodiester cleavage. It was this base triplet that we intended to verify in solution by direct monitoring of the A49-N(7)ÁHN(3)-U4 hydrogen bond interaction, applying HNN COSY NMR experiments. We, therefore, resorted to the same bimolecular twister RNA construct that we previously designed for pK a determination of the putative general acid A6 at the cleavage site [19]. This time, however, the RNAs were synthesized with 15 N(3)-labeled U4 and 15 N(7)-labeled A49. Indeed, in the HSQC NMR spectra, a resonance appeared at 161.5 ppm in the presence of saturating concentration of Mg 2? , indicating the reduced exchange of the H-15 N(3)-U4 with the solvent and thereby supporting a defined base pair interaction. Unfortunately, our attempts to verify this interaction directly by a correlation signal between H-15 N(3)-U4 and 15 N(7)-A49 in the HNN COSY experiments failed so far [21,22]. The 15 N(3) chemical shift value observed for U4 (158.8 ppm) is in the typical ppm range (between 155 and 165 ppm) of Watson-Crick UA, trans Watson-Crick/Hoogsteen UA, and also wobble UG pairs that were observed also in other RNAs [21,23,24]. The H-15 N(3)-U4 chemical shift value observed for U4 (11.92 ppm) is slightly higher compared to typical H-15 N(3)-U shifts in GU wobble base pairs (11.5-11.8 ppm) but also not in the typical range of reversed Hoogsteen AU pairs (12.5-13.0 ppm) [21,23,24]. At this time, we, therefore, cannot completely exclude alternative U4ÁG53 wobble base pair formation within stem P1 in solution although we favor the view of a preference for triplet formation. Further experiments and experimental designs are needed to answer this particular question in the context of twister ribozyme folding.