The synthesis, conformation and hydrolytic stability of an N,S-bridging thiophosphoramidate analogue of thymidylyl-3’,5’-thymidine†

The full-text may be used and/or reproduced, and given to third parties in any format or medium, without prior permission or charge, for personal research or study, educational, or not-for-pro t purposes provided that: • a full bibliographic reference is made to the original source • a link is made to the metadata record in DRO • the full-text is not changed in any way The full-text must not be sold in any format or medium without the formal permission of the copyright holders. Please consult the full DRO policy for further details.

The alkylating agent, 5′-deoxy-5′-iodothymidine 5, showed poor solubility at neutral pH, however, deprotonation of the thymine base ( pK a ∼10) increased solubility markedly. Thus pH 12 was maintained throughout the addition of 5′-deoxy-5′iodothymidine 5 to thiophosphoramidate 4. Under these conditions, 98% of the crude thiophosphoramidate 4 was converted to TnpsT 1 as determined by 31 P NMR spectroscopy. TnpsT 1 was isolated by ion exchange chromatography in 64% yield (based on starting amine 3) as a triethylammonium salt, which was converted to the K + salt for subsequent conformational and kinetic studies.

C Conformational analysis of TnpsT 1
Modified (deoxy)riboses display perturbed north/south conformational preferences (Fig. 2). 27 Given our long-term intention to employ our synthetic approach in nucleic acid templated ligations, that would generate ligated products containing a thiophosphoramidate motif, we sought to gauge the conformational preference of TnpsT 1 using NMR methods.
The elucidation of solution-state ribose conformations by means of 1 H NMR J-coupling values was pioneered by Altona and Sundaralingam, 28,29 and the effect on the conformation of thiophosphate-containing dinucleosides was investigated by Beevers et al. 30 using similar methods. Rinkel and Altona found the sum of the J 1′-2′ and J 1′-2″ coupling constants, ΣH1′, to be linearly correlated with the proportion of the south conformer through eqn (1), where P S is the proportion of the 'south' conformer: Many of the 1 H NMR signals of TnpsT 1 are coincident, and second-order couplings complicate the extraction of J values of the deoxyribose protons. The signals for both C1′ protons, however, are well separated, and the J 1′-2′ and J 1′-2″ coupling values were 7.5 and 4.0 Hz, respectively for the 3′-amino-3′deoxythymidine (Tnp) residue of 1. The C1′-H signal at 6.28 ppm for the 5′-deoxy-5′-thiothymidine fragment ( psT) pre-sents as an apparent triplet with a coupling constant of 6.7 Hz (Fig. 3).
Application of eqn (1) yields values of 29% south for the Tnp ribose ring, and 61% south for the psT ribose ring. Comparison with TpT 2 (Tp: 74.2% south; pT: 62.7% south) 30 indicates that the substitution of nitrogen for oxygen brings about a greater population of the 'north', C3′-endo, or 'RNA-like' conformer in the Tnp fragment, while the psT ribose ring retains its 'DNA-like' conformation. This is not surprising, as the conformation of the furanose ring is largely dictated by the anomeric and gauche effects, where the lower electronegativity of C3′-nitrogen compared to oxygen reduces the gauche effect of donation from the C2′-H bonding orbital into the C3′-N/O antibonding orbital and thus disfavours the south conformation. The thiophosphoramidate group is not linked directly to the ribose ring of the psT fragment and so has no apparent influence on its conformation. The Tnp conformational change is similar to that observed by Beevers et al. in the dideoxynucleoside 3′-phosphorothiolate analogue (TspT), 30 however, a more detailed analysis, 31 particularly in the context of extended and double-stranded nucleic acid structures will be required to confirm this result.

D Kinetics and mechanism of TnpsT hydrolysis
Hydrolysis experiments on TnpsT 1 were carried out at 90°C in buffered aqueous solutions with pHs ranging from 1.32 to 10.91, at intervals of approximately one pH unit ( pH values calculated at 90°C based on values measured at 25°C, see ESI †). Aliquots of substrate 1 in each buffered solution were sealed into vials, and heated at 90°C. Samples were removed from the heating bath or block at suitable intervals, the remaining starting material and products were resolved by HPLC, and the ratios of the integrals of the substrate and hydrolysis products relative to an internal standard were assessed.
At each pH, separate experiments were performed, using 10 and 100 mM buffer concentrations in order to check for buffer-promoted hydrolysis pathways. Acetate and formate buffered-experiments afforded rate constants that appeared to be dependent on buffer concentration. Thus, additional experiments were performed using 40 and 70 mM buffers, and k obs -buffer concentration plots were extrapolated to estimate the buffer-dependent and buffer-independent rate constants (see ESI †).
The log k obs -pH profile for the hydrolysis of TnpsT 1 displayed a pH-independent 'plateau' from ∼pH 7 to 10 with t 1/2 ∼13 days, while at pH 1.32 the half-life was nine seconds (red trace in Fig. 4). The rapidity of reaction at lower pH values put practical limits on our ability to further explore this region. At the high pH extreme, the appearance of insoluble materials in the reaction mixture suggested etching of the glass vials, which precluded the straightforward exploration of the higher pH extreme within the format of our experimental design.
Data for the disappearance of TnpsT 1 were found to fit eqn (2), where rate coefficients k 0 and k H represent the contributions to k obs of the neutral/zwitterionic forms 1(neutral) and monocationic form 1H + respectively (Scheme 2). The acid dissociation constant between 1H + and the kinetically indistinguishable neutral forms 1(neutral) is captured in K a1 . A single data point at pH 10.91 suggests a potential downward trend in reactivity towards higher pHs. This may be associated with Fig. 4 The pH-log(k obs ) profiles for the hydrolyses at 90°C of: TnpsT 1; and related systems Tnp(s)T 6 and TnpT 7 studied by Ora et al. 32 Kinetic data are fitted to eqn (2). Fig. 3 The signals corresponding to the C1' protons in TnpsT 1. The signal at 6.27 ppm corresponds to the 5'-linked nucleoside ( psT), while the signal at 6.19 ppm corresponds to the 3'-linked nucleoside (Tnp).
nucleobase ionisation (K a2 ) of 1(neutral), however, there are insufficient data to substantiate this hypothesis.
The reactivity on the pH plateau, with k 0 = 6.3 × 10 −7 s −1 , is similar to the related analogues Tnp(s)T 6 and TnpT 7 that were studied by Ora et al. (blue and green traces in Fig. 4). 32 The lack of an observed plateau in reactivity at lower pH values limits our ability to unequivocally assign values to the interdependent variables k H and K a1 . Based on the fitting of the available data, however, values of k H < 0.15 M −1 s −1 and pK a1 < 1 were estimated, which align with those observed for TnpT 7.
The similarity in reactivity profiles of TnpsT 1 and TnpT 7 suggests that mechanisms are likely to be similar. This is borne out in product analysis studies, and illustrative examples are discussed below (Scheme 3). At pH 7.0 and 7.7 the largest detected peak by HPLC is thymine 8, derived from initial depyrimidinylation (route A) of either thymidine site within TnpsT 1 and subsequent fragmentation of the resulting species, as seen by Ora et al. for Tnp(s)T 6 and TnpT 7. 32 Given the remoteness of the phosphoryl-sites from the C1′ sites where de-pyrimidininylations occur, it is unsurprising that the reactivities of TnpsT 1, Tnp(s)T 6 and TnpT 7 in the pH independent regions are similar. At pH 6, some depyrimidinylation is observed, however, P-N cleavage is now evident, with amine 3 being observed at a similar reten-   A and B). Another product appears at a much longer retention time, with a lag in its formation. We believe this to be disulfide 11 formed from thiol 10 through oxidation, which is expected to be relatively facile at this pH, and has been reported in a similar system. 33 We were, however, unable to confirm this by HPLC in a MS-compatible buffer system. Thiol 10 is formed by rapid dephosphorylation of phosphorothiolate 9, which arises from acid promoted P-N scission. At pH 3.2, the product chromatograms are simpler, displaying only two major peaks. Amine 3 represents one of these peaks, derived from P-N scission, whereas the second peak is consistent with thiol 10, which is formed rapidly from phosphorothiolate 9 (route B). Thiol 10 is expected to be relatively stable towards oxidation under these conditions. The overlap of the pH-log k obs profiles of TnpsT 1 and TnpT 7 in the acidic region suggests that either the values of k H and K a1 are identical for these species, or that ionisation and reactivity compensate each other to arrive at identical k obs values.

E Conclusions
TnpsT 1, which is an analogue of thymidyl-3′,5′-thymidine 2, was successfully synthesised under aqueous conditions, without protecting groups. NMR-based analyses revealed a predominantly 'north', 'RNA-like' C3′-endo conformational preference for the 3′-aza-substituted deoxyribose (Tnp) fragment of TnpsT 1. Hydrolytic studies on TnpsT 1 yielded a near-identical profile to non-thio-analogue TnpT 7, where for pH > 7, depyrimidinylation dominates, and P-N scission is dominant for lower pHs. The combination of simple aqueous synthesis, knowledge of conformational preference and stability of the linkage will allow us to exploit N,S-bridging nucleotide systems in future applications.