Facile synthesis of a 3-deazaadenosine phosphoramidite for RNA solid-phase synthesis

Access to 3-deazaadenosine (c3A) building blocks for RNA solid-phase synthesis represents a severe bottleneck in modern RNA research, in particular for atomic mutagenesis experiments to explore mechanistic aspects of ribozyme catalysis. Here, we report the 5-step synthesis of a c3A phosphoramidite from cost-affordable starting materials. The key reaction is a silyl-Hilbert–Johnson nucleosidation using unprotected 6-amino-3-deazapurine and benzoyl-protected 1-O-acetylribose. The novel path is superior to previously described syntheses in terms of efficacy and ease of laboratory handling.


Introduction
The synthesis of 3-deazaadenosine building blocks for RNA solid-phase synthesis represents a severe bottleneck in modern RNA research, in particular for studies that aim at the mechanistic elucidation of site-specific backbone cleavage of recently discovered ribozyme classes, known as twister, twister sister, pistol, and hatchet RNA motives [1,2]. Selected adenines in their active sites have been discussed to participate in acid base catalysis, thereby contributing to accelerate the specific phosphodiester cleavage of these nucleolytic ribozymes. Concerning the twister ribozyme, structural analyses suggest that an adenine N3 atom plays a dominant role in catalysis [3][4][5]. Also for the pistol ribozyme, evidence exists that an adenine-N3 in the active site is significant for the cleavage activity, most likely by 5'-O-leaving group stabilization through proton shuttling [6,7]. Another example for a specific role of an adenine-N3 is associated with the catalysis during ribosomal peptide bond formation, a proposal about its role in proton transfer has been disputed heavily since the first ribosome crystal structures up to very recent investigations [8][9][10]. The involvement of N3, and not N1, is surprising with respect to basicity of these purine nitrogen atoms, because N1 represents the major protonation site, followed by N7 and N3. This order is deduced from the macroscopic pK a values that were measured for adenine, 9-methyladenine, and adenosine [11]. Importantly, there is growing evidence that the pK a values of nucleobases can be significantly shifted within a well-structured RNA fold [12][13][14][15]. Scheme 1: Synthesis of c 3 A described by Rousseau et al. in 1966 [22]. a) 1,2,3,5-Tetraacetyl-ß-D-ribofuranose, chloroacetic acid (cat.), 175 °C (melting until clear solution). b) NH 3 in CH 3 OH, 0 °C, 14 h. c) Anhydrous hydrazine, steam bath, 1 h, not isolated. d) Raney nickel, water, reflux, 1 h. Scheme 2: Synthesis of c 3 A described by Montgomery et al. in 1977 [23]. The final step, displacement of the 2-chlorine atom by a hydrogen atom, remains problematic [24][25][26]. a) 1,2,3,5-Tetraacetyl-ß-D-ribofuranose, p-toluenesulfonic acid (cat.), melt (160 °C), 5 to 10 min. b) NH 3 in ethanol (saturated at −30 °C), 140 °C, 89 h [23] or NH 3 (30% aq), 200 °C, 5 d, 80% [24].
To address RNA phenomena of that kind, comparative atomic mutagenesis is an indispensable means, and with respect to ribozymes, can deliver important insights into the RNA catalyzed chemical reactions and underlying mechanisms. Therefore, 1-deazaadenosine (c 1 A), 1-deaza-2'-deoxyadenosine (c 1 dA), 3-deazaadenosine (c 3 A), and 3-deaza-2'-deoxyadenosine (c 3 dA), and the corresponding phosphoramidites to prepare oligoribonucleotides are highly requested nucleoside modifications. Unfortunately, synthetic approaches to achieve them are troublesome and time consuming, in particular for c 3 A. To the best of our knowledge, only two papers have reported the synthesis of c 3 A phosphoramidites so far [16,17]. Thereby, the major bottleneck is access to the naked nucleoside. Although the c 3 A nucleoside is commercially available, prices in the hundreds of Euro range for low milligram amounts make this source unsatisfying. The previously reported c 3 A phosphoramidite synthesis from our laboratory [16], which took older reports by Matsuda, Piccialli, McLaughlin, Watanabe, Robins, and co-workers into account [17][18][19][20][21], started from inosine leading to c 3 A after 8 steps via a 5-amino-4-imidazolecarboxamide (AICA) riboside derivative with 8% overall yield. Another 4 steps followed to achieve a properly protected building block for RNA solid-phase synthesis [16]. With a total of 12 steps, the approach is not very attractive. Because of this frustrating situation, we set out to develop an efficient and easy-to-handle synthesis of a 3-deazaadenosine phosphoramidite building block.

Results and Discussion
Previously described synthetic routes to c 3 A via nucleosidation In 1966, Rousseau, Townsend, and Robins reported the nucleosidation of 4-chloroimidazo [4,5-c]pyridine and 1,2,3,5tetraacetyl-ß-D-ribofuranose in the presence of chloro acetic acid to yield the corresponding 6-chloro-3-deazapurine nucleoside (Scheme 1) [22]. Subsequent attempts to convert the chlorine atom directly by amination under various conditions failed. Only when treated with hydrazine, nucleophilic substitution was observed and after reduction with Raney nickel the desired 3-deazaadenosine was isolated. Our own attempts towards direct ammonolysis failed as well. Additionally, the limited commercial availability of hydrazine and its inconvenience in handling excluded this route for our purposes.

Efficient 5-step synthesis of 3-deazaadenosine phosphoramidite
The key step of our novel route to c 3  NMR spectroscopy which was consistent with the structure of the desired ß-N9 isomer 4, indicated by strong ROEs of the nucleobase C3-H with ribose C3'-H, C2'-H and C1'-H (see Supporting Information File 1). The benzoyl groups of nucleoside 4 were then cleaved with methylamine in ethanol and water to furnish the free c 3 A nucleoside 5. An authentic reference sample that was synthesized according to the previously established 12-step route was used for direct spectroscopic comparison (see Supporting Information File 1) and additionally confirmed its identity. Then, treatment with N,N-dibutylformamide dimethyl acetal [31] resulted in amidine protection of the exocyclic C6-NH 2 group. At the same time, the applied excess of the reagent allowed to transiently form the corresponding nucleoside 2',3'-O-acetal [32], leaving the primary 5'-OH group available for selective tritylation with 4,4'-dimethoxytrityl chloride to give compound 6. Selective protection of the 2'-OH was challenging. Initial attempts that focused on the introduction of the TBDMS group according to the procedure described by McLaughlin and co-workers [17] were unsuccessful. Also, attempts to introduce the [(triisopropylsilyl)oxy]methyl group (TOM) following standard procedures [32] unfortunately failed. We encountered these problems already in our previously published synthesis for N 6 -benzoyl protected c 3 A phosphoramidite [16], and therefore, we decided to apply triisopropylsilyl chloride (TIPS-Cl) and silver nitrate which resulted in the desired 2'-O-TIPS protected nucleoside 7 in 28% yield after chromatographic separation from the corresponding 3'-regioisomer. Finally, the 5'-O-DMTr-2'-O-TIPS protected 3-deazaadenosine derivative 7 was converted into the phosphoramidite building block 8 with 2-cyanoethyl diisopropylchlorophosphoramidite in the presence of N-dimethylethylamine. Starting from compound 3, our route provides 8 in a 6% overall yield in five steps with four chromatographic purifications; in total, 0.6 g of 8 was obtained in the course of this study.

Conclusion
With the reported 5-step synthesis of a c 3 A phosphoramidite we created a route that is superior to previously described syntheses in terms of efficacy and ease of laboratory handling.
The key reaction is a silyl-Hilbert-Johnson nucleosidation using unprotected 6-amino-3-deazapurine and benzoyl-protected 1-Oacetylribose, providing 3-deazaadenosine (c 3 A) in high yields for the subsequent functionalizations to yield a properly protected building block for RNA solid-phase synthesis.
The so-obtained c 3 A-modified RNAs are currently used for atomic mutagenesis experiments to explore mechanistic aspects of phophodiester cleavage of recently discovered ribozyme classes, such as twister, pistol, and hatchet ribozymes [1,2,33].

Experimental
General. Chemical reagents and solvents were purchased from commercial suppliers (Sigma-Aldrich) and used without further purification. 4-Aminoimidazo[4,5-c]pyridine (6-amino-3-deazapurine) and 4-chloroimidazo [4,5-c]pyridine (6-chloro-3-deazapurine) were purchased from Synthonix and Carbogen. Organic solvents for reactions were dried overnight over freshly activated molecular sieves (4 Å). The reactions were carried out under an argon atmosphere. Analytical thin-layer chromatography (TLC) was carried out on Marchery-Nagel Polygram SIL G/UV254 plates. Column chromatography was carried out on silica gel 60 (70-230 mesh). 1 H, and 13 C NMR spectra were recorded on Bruker DRX 300 MHz and Bruker Avance II+ 600 MHz instruments. Chemical shifts (δ) are reported relative to tetramethylsilane (TMS) and referenced to the residual proton or carbon signal of the deuterated solvent: CDCl 3 (7.26 ppm) or DMSO-d 6 (2.49 ppm) for 1 H NMR; CDCl 3 (77.0 ppm) or DMSO-d 6 (39.5 ppm) for 13 C NMR spectra. 1 H and 13 C assignments are based on COSY and HSQC experiments. MS experiments were performed on a Waters ESI TOF LCT Premier Serie KD172 or Bruker 7T FT-ICR instrument with an electrospray ion source. Samples were analyzed in the positive-ion mode.   Compound 4 (1.231 g, 2.13 mmol) was dissolved in a solution of 33% methylamine in ethanol (10 mL) and 40% methylamine in water (10 mL) and stirred for 18 hours at room temperature. All volatiles were evaporated and the residue was dried in high vacuum. The crude product was puri-  The reaction was quenched with methanol (1 mL) and all volatiles were evaporated, followed by coevaporation with toluene (2 × 10 mL). The residue was partitioned between dichloromethane (10 mL) and 5% aqueous citric acid solution (7 mL). The organic layer was separated, washed with water and saturated sodium bicarbonate solution (10 mL each), dried over Na 2 SO 4 and evaporated. The crude product was purified by flash chromatography (1% methanol in dichloromethane + 1.5% triethylamine, size: 18.0 × 2.0 cm, 21 g silica gel

Supporting Information
Supporting Information File 1 Synthetic procedures of compounds 1-3 and NMR spectra of compounds 1-8.