Efficient access to 3′-O-phosphoramidite derivatives of tRNA related N6-threonylcarbamoyladenosine (t6A) and 2-methylthio-N6-threonylcarbamoyladenosine (ms2t6A)

An efficient method of ureido linkage formation during epimerization-free one-pot synthesis of protected hypermodified N6-threonylcarbamoyladenosine (t6A) and its 2-SMe analog (ms2t6A) was developed. The method is based on a Tf2O-mediated direct conversion of the N-Boc-protecting group of N-Boc-threonine into the isocyanate derivative, followed by reaction with the N6exo-amine function of the sugar protected nucleoside (yield 86–94%). Starting from 2′,3′,5′-tri-O-acetyl protected adenosine or 2-methylthioadenosine, the corresponding 3′-O-phosphoramidite monomers were obtained in 48% and 42% overall yield (5 step synthesis). In an analogous synthesis, using the 2′-O-(tert-butyldimethylsilyl)-3′,5′-O-(di-tert-butylsilylene) protection system at the adenosine ribose moiety, the t6A-phosphoramidite monomer was obtained in a less laborious manner and in a remarkably better yield of 74%.


Introduction
Transfer RNAs (tRNAs) are known for having a substantial content of modied nucleoside units. 1,2 To date, in tRNAs from all domains of life, more than 130 modied units have been identied, which differ in chemical structure, 3-6 distribution within the tRNA molecules, 7 and their biological activity. [8][9][10][11][12][13] The majority of modied units are present in the anticodon loop and stem domain of tRNAs, particularly at position 34 (the wobble position) and at position 37, i.e. adjacent to the anticodon at its 3 0 -side. 3,7,14,15 Considering the latter modications, special interest has been paid to several N 6 -threonylcarbamoyladenosines (depicted in Fig. 1), which are widely involved in the decoding of the A-starting codons (ANN). 15,16 Among them, the most abundant N 6 -threonylcarbamoyladenosine (t 6 A) 17 and its analogs containing either the -SMe group at the purine C2 atom (ms 2 t 6 A), 18 or the methyl substituent at the N 6 -atom (m 6 t 6 A) 19 have been known for many years and their diverse functions during the protein biosynthesis were intensively studied. 6,9,15,16 Recently, next members of the t 6 A family have been identied in the tRNA anticodon loops, i.e. cyclic N 6threonylcarbamoyladenosine (ct 6 A) 20,21 cyclic 2-methylthio-N 6threonylcarbamoyladenosine (ms 2 ct 6 A), 22 and a t 6 A derivative having the threonine methyl group converted into a hydroxymetyl one (hydroxy-N 6 -threonylcarbamoyl-adenosine, ht 6 A). 23 Recognition of the structural aspects and biological functions of the t 6 A nucleoside family is highly dependent on the synthetic availability of these nucleosides, as well as their 3 0 -Ophosphoramidite derivatives, which are essential for fast and efficient synthesis of model oligonucleotides with the sequence of the appropriate tRNA anticodon stems and loops (ASL of tRNAs). To date, several procedures have been developed to modify adenosine or 2-methylthioadenosine (ms 2 A) at the N 6 position with a threonylcarbamoyl chain (a ureido system is formed) using either a carbamate or isocyanate approach (Scheme 1, paths A and B, respectively).
The isocyanate approach to the synthesis of t 6 A/ms 2 t 6 A (Scheme 1, path B1,2) was shown to have limited applicability in the preparation of "free" nucleosides. 24 Because this method required a threonine derivative protected on the OH and COOH functions, it was considered inferior to the carbamate approach in which unprotected amino acid can be used. [24][25][26] However, the isocyanate route was recently postulated by the Carell's group as a possible pathway for the formation of t 6 A under prebiotic conditions. 42 In the synthesis of threonine protected t 6 A/ms 2 t 6 A derivatives for the subsequent preparation of the corresponding 3 0 -Ophosphoramidites, the isocyanate approach 39,43 is much less explored than the carbamate procedures. 29,[34][35][36][37][38][39][40] Initially, the isocyanate derivative was generated from the N 6 -amine function of sugar protected adenosine (Scheme 1, path B1), but its condensation with the free amine function of L-threonine was ineffective and the ureido-nucleoside product was obtained in a low 19% yield. 43 Noticeably better results were obtained in our recently published method (Scheme 1, path B2), based on the reaction of isocyanate derivative of the amino acid substrate (prepared by removing of Boc-protection and phosgene treatment of the free amine function of L-threonine appropriately blocked on the OH and COOH functions) with the sugar protected nucleoside (overall yield of this three steps procedure $55%). 39 This result of isocyanate procedure turned our attention to the methods of synthesis of unsymmetrical ureas involving the formation of the isocyanate functionality directly from the carbamate type protecting groups of amino acids (e.g. N-Boc protecting group). [44][45][46][47][48][49][50][51][52][53] Most likely, such variant of the isocyanate method (Scheme 1, path B3) applied in the synthesis of the 3 0 -O-phosphoramidite derivatives of t 6 A/ms 2 t 6 A would be greatly advantageous in comparison to our previous isocyanate route (Scheme 1, path B2) owing to a smaller number of synthesis steps in the preparation of threonine derivative (the removal of N-Boc protection is unnecessary) and escaping the use of toxic phosgene.
Here we report a new one-pot procedure for the introduction of an ureido linkage into t 6 A/ms 2 t 6 A using a Tf 2 O-mediated generation of the isocyanate derivative directly from N-Bocprotecting group of L-threonine, followed by its straight reaction with the N 6 exo-amine function of the sugar protected nucleoside. We have also showed that this approach is compatible with the use of the recently introduced 2 0 -O-(tertbutyldimethylsilyl)-3 0 ,5 0 -O-(di-tert-butylsilylene) ribonucleoside sugar protection system, 54-56 that allows to prepare the 3 0 -Ophosphoramidite monomeric unit more effectively and in a less laborious manner.

Results and discussion
Search for the best conditions leading to the formation of the ureido compound 4a was performed using trimethylsilylethyl (TMSE) ester of N-Boc-O-tert-butyldimethylsilyl (TBDMS) protected L-threonine 39 (1) and 2 0 ,3 0 ,5 0 -tri-O-acetyladenosine (3a) ( Table 1, see ESI for spectroscopic data of 1, 2, 3a, Fig. S1-S7 †). In all cases, the nal condensation of isocyanate 2 with 3a was performed in the presence of Et 3 N (2-fold molar excess over the Tf 2 O activator used for isocyanate formation) in boiling toluene for 16 h. It was reported that addition of Et 3 N, which is an effective scavenger of triuoromethanesulfonic acid (generated in the step of isocyanate formation), helps to maintain a concentration of the unprotonated amine component sufficient for effective nucleophilic attack on the isocyanate moiety. 48,50 To optimize the triic anhydride (Tf 2 O) mediated conversion of N-Boc-protected threonine 1 (a dichloromethane solution) into the isocyanate derivative 2 (the rst step of the one-pot synthesis of 4a) we were changing the amount of Tf 2 O activator, basicity of amine, temperature and reaction time (entries 1-7). When the amount of 2 reached the plateau (TLC monitoring) the reaction mixture was concentrated, the residue was dissolved in toluene and Et 3 N and the nucleoside substrate 3a was added. The reaction 1 / 2 for 15 min at room temp. (entry Scheme 1 Approaches for the formation of the ureido linkage in t 6 A modified nucleoside. 1), followed by reaction with 3a, afforded the nal product 4a in a low 19% yield and several by-products were detected by TLC analysis. An experiment conducted at lower temperature (0 C) for much shorter time (5 min) (entry 2) was more productive (46% yield) but the yield did not further increase when higher concentration of Tf 2 O (2 equiv.) was used (entry 3). Compound 1 did not react when more common bases such as pyridine, 4dimethylaminopyridine or triethylamine were used (entries 4-6). In the case of 2,6-lutidine, some isocyanate 2 was generated aer 30 min at rt, but the nal product 4a was formed in only 16% yield (entry 7). Neither acetic anhydride nor triuoroacetic anhydride were able to promote the formation of isocyanate 2 regardless of the temperature applied.
In so far reported procedures for the one-pot syntheses of ureas from carbamates, the use of an excess of amine substrate, usually up to 3 equivalents (or more for less nucleophilic amines) is recommended to obtain the higher efficiency of the process. 48,50,51 However, in the case of t 6 A/ms 2 t 6 A synthesis, the amine nucleoside substrate, especially non-native 2-methylthioadenosine (ms 2 A) is a very costly reagent. Therefore, in the second step of optimizations we examined an excess of N-Boc protected L-threonine derivative 1 to nucleoside 3a in a range 1.5-2.5 (entries 8-10), yet the concentrations of Tf 2 O and 2-Cl-Py against 1 were kept as determined previously (entry 2). We were glad to see that 1.5 molar excess of 1 to 3a led to a signicantly better yield of 4a (71%, entry 8), while very high conversion of 3a to 4a was observed when 2.5 equivalents of 1 was applied (92%, entry 10). Unfortunately, further increase in the excess of 1 (3 equiv. or more) did not lead to a higher isolated yield of product 4a. Finally, the use of toluene instead of dichloromethane for the formation of 2 allowed us to carry out the whole process in the same solvent (92% yield, entry 11) which facilitate the preparative procedure for the one-pot synthesis of t 6 A derivative 4a.
The optimized method described above was used in synthesis of the phosphoramidite derivatives of t 6 A and ms 2 t 6 A (6a, and 6b, respectively; Scheme 2). Starting from 2.5 mmol of appropriately protected Boc-L-threonine 1 39 and 1 mmol of adenosine derivative 3a 57 or 3b, 39 the modied nucleosides 4a and 4b were obtained in 92% and 86% yield, respectively. Next, the acetyl groups in 4a/4b were removed under conditions safe for the installed N 6 -threonylcarbamoyl chain (Et 3 N/MeOH, rt, 24 h) and the resultant 5a/5b were appropriately protected and Scheme 2 Preparation of t 6 A and ms 2 t 6 A 3 0 -O-phosphoramidities and samples of modified nucleosides t 6 A and ms 2 t 6 A. phosphitylated according to the previously reported procedures 39 to give t 6 A/ms 2 t 6 A-phosphoramidites (6a/6b) in 48% and 42% overall yield, respectively (see ESI † for details). Also, the nucleosides 4a/4b were deprotected to yield 8a/8b (Scheme 2), to be used as standards in analysis of enzymic hydrolysates of t 6 Aor ms 2 t 6 A-containing oligomers. The silyl protecting groups (TBDMS, TMSE) were removed with excess 1 M tetrabutylammonium uoride (TBAF) in THF (4 h, rt), and the acetyl groups were cleaved off with NH 3 /MeOH (2 h, rt) (see experimental details in ESI †). The reactions were virtually quantitative and the HPLC proles recorded for the reaction mixtures (Fig. 2) contained single, slightly tailing peaks (proles in panels (A), part I for t 6 A and part II for ms 2 t 6 A). The tailing was not observed, when the highly lipophilic tetrabutylammonium cations were replaced with H + ions using DOWEX, H + / CaCO 3 treatment. 58 The resultant acidic forms 8a/8b had the same HPLC mobility as genuine L-t 6 A/L-ms 2 t 6 A standards 21,22,38,39 (compare proles in panels (B) and (C)). The proles recorded for 8a co-injected with D-allo-t 6 A 21,38 and for 8b co-injected with D-allo-ms 2 t 6 A 22,39 (panels (E)) indicate that the new procedure for ureido linkage formation is safe in terms of the stereochemistry at the Ca of the amino acid component. The proles for the D-allo nucleoside standards are shown in panels (D).

Conclusions
The presented here modication of the isocyanate method of formation of the ureido linkage between adenosine and threonine greatly facilitates synthesis of fully protected L-threonylcarbamoyl modied adenosines 4a,b rendering subsequent preparation of the t 6 A/ms 2 t 6 A phosphoramidite monomers 6a,b much more efficient. The developed one-pot procedure for 4a,b synthesis, consisting in the epimerization-free formation of Lthreonine isocyanate directly from the N-Boc-Thr upon activation with Tf 2 O in the presence of 2-Cl-Py, followed by its straight reaction with the N 6 exo-amine function of the sugar protected nucleoside, eliminates the use of toxic phosgene and provides a shorter protocol for the preparation of the protected t 6 A/ ms 2 t 6 A derivatives compared to the previously reported isocyanate and carbamate routes. In addition, the protected nucleosides 4a,b were efficiently deprotected yielding free nucleosides 8a,b to be used as the standards, e.g. in HPLC analysis of enzymatically digested oligomers bearing t 6 A/ms 2 t 6 A units. Moreover, the in situ formed threonine isocyanate reacted efficiently with 2 0 -O-(tert-butyldimethyl-silyl)-3 0 ,5 0 -O-(di-tert-butylsilylene)adenosine and the resultant conjugate was conveniently transformed into the t 6 A-phosphoramidite in a very good overall yield 74%. Developed procedures for the synthesis of t 6 A/ms 2 t 6 A 3 0 -O-phosphoramidities will signicantly facilitate the  availability of monomeric units for the chemical synthesis of various model tRNA fragments suitable for the structureactivity-relationship and biological studies of the t 6 A family nucleosides.

General remarks
Commercial reagents and analytical grade solvents were used without additional purication unless otherwise stated. Analytical thin layer chromatography (TLC) was done on silica gel coated plates (60 F254, Supelco) with UV light (254 nm) or the ninhydrin test (for amino acids) detection. The products were puried by chromatography on a silica gel 60 (mesh 230-400, Fluka) column eluted with the indicated solvent mixtures. NMR spectra were recorded using a 700 MHz (for 1 H) instrument, 176 MHz for 13 C and 283 MHz for 31 P. Chemical shis (d) are reported in ppm relative to residual solvent signals CDCl 3 : 7.26 ppm for 1 H NMR, 77.16 ppm for 13 C NMR; DMSO-d 6 : 2.50 ppm for 1 H NMR, 39.52 ppm for 13 C NMR. The signal multiplicities are described as s (singlet), d (doublet), dd (doublet of doublets), ddd (doublets of doublets of doublets), dq (dublet of quartets), t (triplet), td (triplet of doublets), q (quartet), qd (quartet of doublets), m (multiplet), and br s (broad singlet). High-resolution mass spectra were recorded on Synapt G2Si mass spectrometer (Waters) equipped with an ESI source and quadrupole-time-of-ight mass analyzer. HPLC analysis of nucleosides was performed on a Shimadzu Prominence HPLC system equipped with an SPD-M20A spectral photodiode array detector using a Kinetex® column (RP, C18, 5 mm, 4.6 Â 250 mm, 100Å, Phenomenex). Analyses were run at 30 C and the elution proles were UV monitored at l ¼ 254 nm.
General procedure for the one-pot synthesis of 4a, 4b and 10 from Boc-L-threonine 1 To a stirred solution of Boc-L-threonine 1 (1.08 g, 2.5 mmol) in dry toluene (30 mL) 2-chloropyridine (2-Cl-Py, 0.7 mL, 7.5 mmol) was added, followed by triuoromethanesulfonic anhydride (Tf 2 O, 0.64 mL, 3.75 mmol) and aer stirring for 15 min at room temperature triethylamine (Et 3 N, 1.04 mL, 7.5 mmol) and sugar-protected adenosine (3a, 3b or 9, 1.0 mmol) were added. The reaction mixture was stirred under reux for 16 h. Then the solvent was evaporated under reduced pressure and 4a,b or 10 were isolated by silica gel column chromatography.
One-pot synthesis of 4b from Boc-L-threonine 1 and adenosine derivatives 3b.

Preparation of nucleoside standards 8a and 8b
Fully-protected adenosine 4a or 4b (0.02 g, 0.03 mmol) was dissolved in 1 M solution of TBAF in THF (0.4 mL, 0.40 mmol) and the reaction mixture was stirred for 4 h at room temperature. Aer this time NH 3 in dry MeOH (8 M solution, 0.2 mL) was added for deprotection of all acetyl groups from ribose moiety. The reaction was carried out for 2 h and then NH 3 was removed under reduced pressure to obtain tetrabutylammonium salts 7a/7b. To exchange Bu 4 N + counterion to H + , CaCO 3 (0.28 g), dry DOWEX 50WX8 H + form (0.84 g) and distilled methanol (0.6 mL) were added and the reaction mixture was stirred for 1 h at room temperature. 58 Aer this time the resulting mixture was ltered through Celite plug and washed with MeOH. The ltrate was analysed by HPLC and the presence of fully-deprotected only one isomer of 8a/8b with natural Lthreonine residue was conrmed (for 8a R t ¼ 22.121 min, for 8b R t ¼ 28.745 min, see Fig. 2 panel (B)). RP-HPLC conditions for analysis of t 6 A derivatives: C18 column with linear gradient of buffer A (0.1% AcOH in H 2 O) and buffer B (ACN) with a ow of 1 mL min À1 as follows: 0-15 min from 2% to 8% B, 15-30 min from 8% to 25% B, 30-35 min 2% B. RP-HPLC conditions for analysis of ms 2 t 6 A derivatives: C18 column with linear gradient of buffer A (0.1% AcOH in H 2 O) and buffer B (ACN) with a ow of 1 mL min À1 as follows: 0-30 min from 2% B to 15% B, 30-40 min from 15% B to 30% B, 40-45 min 2% B.

Conflicts of interest
There are no conicts to declare.