The reactivity of phosphodiester bonds within linear single-stranded oligoribonucleotides is strongly dependent on the base sequence

Abstract

The cleavage of short chimeric oligonucleotides containing only one reactive ribonucleoside unit, all other nucleosides being 2′-O-methylated, has been studied at pH 8.5 and 35°C. Among the 20 different sequences that did not exhibit any tendency to form a defined secondary structure, the scissile 5′-UpA-3′ and 5′-CpA-3′ phosphodiester bonds experienced >100- and up to 35-fold reactivity differences, respectively. Compared with dinucleoside monophosphates, both rate accelerations and retardations of more than one order of magnitude were observed. Even a change of a single base several nucleosides away from the scissile bond markedly affected the reaction rate. Duplex formation at the 3′- and/or 5′-side of the scissile bond was also studied and observed to be strongly rate retarding. The origin of the high sensitivity of phosphodiester bonds to the molecular environment is discussed.

INTRODUCTION

Transesterification of RNA phosphodiester bonds has been intensively studied over the last decades. The systems examined vary from the cleavage of RNA and RNA mimetics by low molecular weight inorganic and organic agents (1–3) to ribozyme (4–8) and DNA enzyme (9,10) reactions. The reasons for the interest lay in the potential applications of these reactions. The sequence-selective cleavage of RNA phosphodiester bonds is hoped to offer a method of selective elimination of intracellular mRNA molecules (2,11).

Under physiological conditions, the transesterification of RNA phosphodiester bonds (Scheme 1) is very slow: the kinetic data obtained at 90°C and pH 7.0 (12–14) suggest that at 37°C the half-life of the cleavage reaction is in the order of years. Efficient catalysis is hence required to utilise the reaction. Factors that enhance the transesterification have been studied extensively, but rather little is still known about the effect of base sequence on the reactivity of phosphodiester bonds. The base composition is known to have only a modest effect on the alkaline cleavage of the phosphodiester bond of dinucleoside monophosphates (12,15,16) and diphosphates (17) or short chimeric DNA–RNA oligonucleotides (14): up to 5-fold differences have been reported. In the context of natural RNA polymers, 5′-UpA-3′ and 5′-CpA-3′ phosphodiester bonds have frequently been observed as less stable than those flanked by other nucleosides (18–22). Efficient and selective cleavage of the 5′-UpA-3′ bond within synthetic oligoribonucleotides in the presence of organic cofactors, such as polyvinylpyrrole, has been extensively documented (23–25).

Scheme 1. Cleavage of RNA by intramolecular transesterification.

The effect of base composition on the stability of RNA phosphodiester bonds has been frequently attributed to stacking interactions between the adjacent nucleic acid bases (14,18,24). The exceptionally slow cleavage within A-rich sequences (26,27), for example, has been explained (14) in this manner. Strong stacking is assumed to hinder the cleavage of the intervening phosphodiester bond by preventing the attacking 2′-OH, the phosphorus atom and the departing 5′-oxygen to adapt a co-linear in-line conformation that is believed to be essential for efficient transesterification. Experimental evidence for the importance of stacking interactions as a factor determining the reactivity is, however, contradictory: there is practically no correlation between the rate constants of alkaline cleavage of dinucleoside monophosphates (15,16) and the stacking efficiency predicted by molecular dynamics simulations (28). Similarly, no clear correlation between the stacking and reactivity could be obtained by C5 alkylation of the uridine adjacent to the scissile phosphodiester bond (25). Although the reactivity was expectedly decreased with the increasing size of the substituent, the 5-methyl group that enhances stacking proved to be clearly rate accelerating. Possibly more important than the stacking between the neighbouring bases is the stacking farther in the molecule and its influence on the overall structure of the oligomer (14,24). In fact, stacking-related changes in the hydration pattern in the vicinity of the scissile bond have been offered as a possible explanation for the exceptionally rapid cleavage of some 5′-UpA-3′ bonds in the presence of organic cofactors (18,25). The hydrogen bonding network around the scissile bond, attacking 2′-OH and neighbouring bases, may in principle accelerate the reaction in three different ways: (i) by accepting a proton from the 2′-OH; (ii) by donating a proton to the negatively charged phosphodiester; and (iii) by donating a proton to the leaving 5′-oxyanion. In particular, a hydrogen bond, either direct or water mediated, between the phosphate and the base of a 5′-linked nucleoside has been suggested to enhance the electrophilicity of the phosphate (25).

The present work is aimed at providing quantitative data on the influence of base sequence on the reactivity of phosphodiester bonds within linear single-stranded oligoribonucleotides that do not form noticeably stable secondary structures. No such data on reactions in the absence of any organic or inorganic cofactors has been published thus far. In addition, the effect of duplex formation at the 3′- and/or 5′-side of the scissile bond have been studied. Short chimeric oligonucleotides containing one reactive ribonucleoside unit, all other nucleosides being 2′-O-methylated, have been used as model compounds. Special attention is paid to the dependence of the reactivity of 5′-UpA-3′ and 5′-CpA-3′ bonds, i.e. the preferred cleavage sites in RNA, on the sequence farther in the oligonucleotide chain. Rate constants for the cleavage have been determined in mildly alkaline buffer solutions (pH 8.5) in the absence of any cofactors and, for comparative purposes, in aqueous alkali. The origin of the significant reactivity differences between various sequences is discussed.

MATERIALS AND METHODS

Materials

Chimeric ribo/2′-O-methylribo oligonucleotides (2, 4–24) were synthesised from commercial 2′-O-methylated (Glenn Research) and 2′-O-[1-(2-fluorophenyl)-4-methoxypiperidin-4-yl] (2′-O-Fpmp) protected (Cruachem) building blocks by conventional phosphoramidite strategy, according to the standard RNA-coupling protocol of ABI 392 DNA/RNA Synthesizer. The protecting groups were removed and the crude oligonucleotides were purified as described earlier (29). UpU (1) and UpA (3) were products of Sigma, and they were used as received. All the buffer solutions were prepared in sterilised water, and sterilised equipment was used for handling the solutions.

Melting temperature measurements of oligonucleotides

The melting curves were recorded on a Perkin-Elmer Lambda 2 UV spectrometer equipped with a PTP-6 temperature programmer that consisted of two electronic control units and Peltier cell housing blocks. The temperature was increased at a rate of 1°C/min over the temperature range 15–90°C. The change in the UV absorption was followed at 260 nm in 0.1 M CHES buffer (cyclohexylaminoethanesulfonic acid, pK_a = 9.5 at 25°C) at pH 8.5 (I = 0.1 M with NaNO₃).

Kinetic measurements

The reactions were carried out in Eppendorf tubes immersed in a water bath, the temperature of which was maintained at either 35.0 ± 0.1°C or 90.0 ± 0.1°C. The reactions were carried out in 300–400 µl of a CHES buffer (0.1 M, pH 8.5, I = 0.1 M with NaNO₃) or aqueous sodium hydroxide (10 mM), containing 0.5–0.9 OD units of the oligonucleotide. p-Toluenesulfonate ion was used as an internal standard. Its concentration was adjusted to give a peak of similar size to that of the substrate at the beginning of the reaction. Aliquots (20 and 50 µl in the case of oligonucleotides and dinucleotide monophosphates, respectively) withdrawn at suitable intervals were immediately cooled to 0°C. The samples were kept in a freezer until analysed.

The analysis of samples

The aliquots of the reactions of UpA (3) were analysed by RP-HPLC on a Hypersil RP-18 column (250 × 4 mm, 5 µm particle size). Aqueous acetic acid [2.5 µl; HOAc:H₂O 1:3 (v/v)] was added to the samples withdrawn from the CHES buffer, and 6.25 µl of aqueous hydrogen chloride was added to the samples withdrawn from the sodium hydroxide solutions. An acetic acid buffer (0.1 M, pH 4.3, I = 0.1 M with NH₄Cl) containing 3% acetonitrile (v/v) was used as an eluent. At the flow rate 1 ml min^–1, the retention time of UpA was 20 min. The detection wavelength was 260 nm. The aliquots of the reactions of UpU (1) were analysed by RP-HPLC as described previously (30).

The aliquots of the reactions of oligonucleotides were analysed by capillary zone electrophoresis (CZE; Hewlett Packard ^3DCE). Tetrameric oligonucleotides 19 and 20 were analysed using a fused silica capillary [75 µm inner diameter (i.d.), 112 cm total length, 105 cm effective length] and borate buffer (0.5 M, pH 8.5) as a background electrolyte. The voltage applied was +30 kV. Between each analytical run, the capillary was flushed with water, 10 mM aqueous sodium hydroxide and the background electrolyte buffer for 2, 3 and 3 min, respectively. Longer oligonucleotides were analysed on a fused silica capillary having an extended light path (75 µm i.d., optical path length 200 µm, 112.5 cm total length, 104 cm effective length), making use of inverted polarity and a citrate buffer (0.2 M, pH 3.2) as a background electrolyte. The voltage applied was –30 kV. The capillary was flushed between analytical runs as described above, but by using 10 mM aqueous hydrogen chloride instead of aqueous sodium hydroxide. The temperature of the capillary was kept at 25°C. To adjust the pH close to that of the background electrolyte buffer, 1.0 µl aqueous acetic acid [HOAc:H₂O 1:3 (v/v)] was added to the aliquots withdrawn from the CHES buffer, and 2.0 µl aqueous hydrogen chloride to those withdrawn from the sodium hydroxide solutions. The aliquots of the tetramers (19 and 20) were injected by hydrodynamic injection with 50 mbar for 30 s, whereas with longer oligonucleotides (2, 5–18 and 21–24) the injection time varied from 10 to 15 s. The absorbance of the oligonucleotides and the internal standard was measured at 260 and 220 nm, respectively. The migration times of the tetramers were 33 min, while those of longer oligonucleotides ranged from 20 to 43 min. The migration time of the internal standard was 51 and 23 min, on applying normal and inverted polarity, respectively. The migration times depended on the chain length and base composition when citrate buffer (0.2 M, pH 3.2) was used as the background electrolyte. The initial cleavage products, namely the 5′-fragment bearing a terminal 2′,3′-cyclic monophosphate group (31) and the 3′-fragment having a free 5′-OH group, appeared as separate peaks. The separation of the oligonucleotide 2′- and 3′-phosphates formed upon hydrolysis of the 2′,3′-cyclic phosphate group could not be achieved under the conditions described. In CHES buffers, partial dephosphorylation of the 2′- and 3′-phosphates to the 3′-OH free oligomer took place.

Calculation of rate constants

The pseudo first-order rate constants for the cleavage of the starting material into two fragments were calculated by applying the integrated first-order rate law to the disappearance of the starting material. The peak area was first normalised by dividing the area observed by the migration time. The normalised area of the starting material was then divided by the normalised area of the internal standard material.

RESULTS

Structures of the oligonucleotides studied

The compounds studied are depicted in Figure 1. Oligonucleotides 2 and 4–24 contain only one scissile phosphodiester bond, the rest of the nucleoside units being 2′-O-methylated. This allows accurate determination of the reactivity of one particular phosphodiester bond in different molecular environments. 2′-O-Methylated oligoribonucleotides may be expected to mimic the structure of oligoribonucleotides reasonably well. They, for example, are known to form duplexes that closely resemble the natural A-form RNA (32). Uridylyl-3′,5′-uridine (1, UpU) and a 13mer uracil homooligomer (2) have been used previously as model compounds of linear random coils (30), as stacking interactions between uracil bases are known to be of minor importance (28,33). The same compounds are also used here for reference purposes. Among the oligonucleotides studied, 4–20 are single-stranded oligomers, where the base sequence, the length of the sequence and the position of the scissile phosphodiester bond are varied. Compounds 21 and 22 are hairpin loops where a sequence identical to that of 6 is engaged in an intramolecular duplex at either the 5′- (21) or 3′-side (22) of the scissile bond. In 23 and 24, the sequence of 6 forms a bulged double helix with a complementary all 2′-O-methylated sequence containing one (23) or five (24) extra nucleotides opposite to the scissile phosphodiester bond.

Figure 1.

Structures of the compounds studied. Bold letters refer to ribonucleosides, the other nucleosides being 2′-O-methylated. An arrow indicates the position of the scissile bond.

Possible secondary structures of oligonucleotides 4–18 were examined by calculating the ΔG° values for the duplex formation of their all ribo analogues by the MFOLD program (MFOLD program is available at http://bioinfo.math.rpi.edu/~zukrm/rna/) (34,35). In all cases the Gibbs free energy values were positive (4, 5, 7–9 and 12–18), or no secondary structure was possible (6, 10 and 11). Although the program employed is based on parameters generated from the melting of RNA in 1 M NaCl, i.e. not in 0.1 M CHES buffer used in the present study, and although 2′-O-methylated oligoribonucleotides hybridise more efficiently than their non-methylated counterparts (36), these calculations still argue against the existence of any markedly stable secondary structure. To exclude the possible occurrence of such structures under the conditions used in the kinetic experiments, the melting curves for oligonucleotides 4, 5, 7–9 and 12–18 were measured in the reaction solution (0.1 M CHES, I = 0.1 M, pH 8.5). In all cases, the melting curves levelled at <35°C. The melting temperatures (T_m) of hairpins 21 and 22, and duplexes 23 and 24 were, in turn, determined to verify the existence of the assumed secondary structures under the conditions of kinetic experiments. The T_m values of hairpins 21 and 22 were 75°C, and those of duplexes 23 and 24 were 55°C. In summary, it appears clear that oligonucleotides 4–20 do not exhibit any secondary structure under the conditions used in the cleavage experiments, while 21 and 22 are present as hairpins, and 23 and 24 as duplexes.

Cleavage experiments

The cleavage of 1–24 was studied in 0.1 M CHES buffer at pH 8.5 and 35°C. For comparative purposes, the alkaline cleavage was studied in 10 mM aqueous sodium hydroxide at 35 and 90°C. The ionic strength was adjusted to 0.1 M with sodium nitrate. Under these experimental conditions, the predominant reaction is the hydroxide ion catalysed transesterification, although the buffer-dependent cleavage may also noticeably contribute to the cleavage in the CHES buffer (37). Isomerisation of 3′,5′-phosphodiester bonds to 2′,5′-bonds is insignificant under these conditions (12,13,37). In order to exclude the contribution of metal ion catalysis, 2 mM EDTA was added to the CHES buffer. The rate constants obtained in the presence and absence of EDTA were identical. The composition of the aliquots withdrawn at appropriate intervals was analysed by CZE. The pseudo first-order rate constants were calculated from the disappearance of the starting material.

The rate constants obtained are shown in Table 1. As seen, the phosphodiester bond within various sequences is cleaved at pH 8.5 at a very different rate. More than 100-fold reactivity difference is observed between oligonucleotides 5′-GGGUAN↓AAGUGC-3′ (4 and 13–15), all having the same base sequence with the exception of 5′-ribonucleoside of the scissile phosphodiester bond. The 5′-ApA-3′ bond linkage within 13 is, as expected, very stable: no reaction was observed in 3 months. For comparison, the half-lives for the cleavage of the 5′-UpA-3′ (4), 5′-CpA-3′ (14) and 5′-GpA-3′ (15) bonds are 5, 270 and 160 days, respectively. The 5′-ApA-3′ bond is very stable also within oligonucleotide 5′-GGGUAUA↓AGUGC-3′ (16). Compared with the cleavage of UpU (1), oligo(U) (2) and UpA (3), all of which are assumed to contain a freely rotating phosphodiester bond, oligomer 4 having a scissile 5′-UpA-3′ bond hence exhibits a rate acceleration by one order of magnitude, whereas oligomers 13–16 having a 5′-NpA-3′ bond (N = C, G, A), all show a rate retardation.

Table 1. Pseudo first order rate constants of the cleavage of the compounds 1–24 in CHES buffer at 35°C and in 10 mM NaOH at 35 and 90°C (I = 0.1 M with NaNO₃).

The ribonucleoside is indicated with a bold letter and the position of the strand scission with a vertical line.

^aIn 0.1 M CHES buffer, pH 8.5.

^bNo reaction after 3 months.

It is, however, important to note that the 5′-UpA-3′ bond is not inherently exceptionally reactive, but it shows enhanced reactivity only within some oligonucleotide substrates. The cleavage of UpA (3) is even slightly slower than that of UpU (1), and within tetramer 19 the 5′-UpA-3′ bond is cleaved approximately as rapidly as in these dimers (1 and 3) or the 5′-CpA-3′ bond in tetramer 20. Nevertheless, when inserted into a dodecameric substrate, the cleavage of the 5′-UpA-3′ bond may experience an up to 30-fold acceleration (compare 4 and 8), which is strongly sequence dependent. The half-life for the cleavage of 5′-UpA-3′ bond in various dodecamers ranges from 2 to >400 days depending on the base sequence. A variation almost as large is observed among the cleavage rates of the 5′-CpA-3′ bond in different oligomers: 17 is cleaved 30 times as fast as 14. In other words, the 5′-UpA-3′ phosphodiester bond as such is not significantly less stable than the other phosphodiester bonds, but the base sequence of the whole molecule determines the reactivity.

Interestingly, modification at the near neighbourhood of the scissile 5′-UpA-3′ bond within 4 does not have any remarkable influence on the reactivity. When the 5′-UAU↓AA-3′ sequence in 4 is replaced with a 5′-UUU↓AA-3′ (5), 5′-AAU↓AA-3′ (6) or 5′-UAU↓AU-3′ (7) sequence, the reactivity of the central 5′-UpA-3′ bond is only slightly increased. Somewhat unexpectedly, modifications farther in a molecule have a more significant effect: replacement of G2 in 4 with U to give 9 decreases the reactivity of the 5′-UpA-3′ bond by a factor of 10, and within 5′-CCCCAAU↓AACCCC-3′ (10) this bond is so stable that no reaction was detected in 3 months. Accordingly, in the latter case the 5′-UpA-3′ bond is at least 10 times less reactive than in UpA (3) or within 5′-AU↓AA-3′ (19). However, when every second C in the terminal CCCC-sequences in 10 are replaced with U to obtain 5′-UCUCAAU↓AACUCU-3′ (11), the cleavage rate of the 5′-UpA-3′ bond again receives a value comparable with that in the short model compounds 3 and 19. As the whole sequence so markedly affects the stability of an individual phosphodiester bond, it is only natural that the position of the scissile bond within a given sequence also influences the reactivity. The 5′-UpA-3′ bond within 5′-GGGUAU↓AAGUGC-3′ (4) is, for example, cleaved 15 times more readily than that within 5′-GGGU↓AUAAGUGC-3′ (12).

As mentioned above, the molecular environment also affects the cleavage rate of the 5′-CpA-3′ bond. Up to a 35-fold difference in reactivity may be observed, but the effects do not parallel those observed with the 5′-UpA-3′ bond. In 5′-GGGUAN↓AAGUGC-3′, where N is either U (4) or C (14), the 5′-UpA-3′ bond is 50 times more reactive than the 5′-CpA-3′ bond, whereas in 5′-GGGUAN↓AAGUUC-3′ [N = U (8), N = C (17)], the corresponding reactivity ratio is only 3, and in 5′-GUGUAN↓AAGUGC-3′ [N = U (9), N = C (18)] the 5′-UpA-3′ and 5′-CpA-3′ bonds are approximately as reactive. When G11 in 14 is replaced with U (17), the cleavage rate of the 5′-CpA-3′ bond is increased by a factor of 35, while this modification has no particular influence on the reactivity of the 5′-UpA-3′ bond (compare 4 and 8). Replacing G2 with U near the 5′-end of 4 and 14, in turn, slightly enhances the cleavage of the 5′-CpA-3′ bond (18) but clearly retards the rupture of the 5′-UpA-3′ bond (9). In other words, the same changes in the base sequences have rather different influences on the reactivity of these two 5′-pyrimidine–purine-3′ bonds.

The large differences in the stability of internucleosidic linkages in a buffer solution at 35°C are greatly diminished when the reactions are carried out at the same temperature but at a higher pH. For example, the 5′-UpA-3′ bond within 5′-GGGAAU↓AAGUGC-3′ (6) in the CHES buffer is 15 times as reactive as the 5′-UpA-3′ bond within 5′-AU↓AA-3′ (19), but in 10 mM aqueous sodium hydroxide the reactivity ratio is only 3. Increasing to 90°C, the cleavage rates become almost independent of the base sequence.

Comparison of the cleavage rate of the 5′-UpA-3′ bond within 6 with that within hairpin 22 shows that when the 3′-terminal sequence of 6 is engaged in a duplex, the 5′-UpA-3′ bond is stabilised by a factor of 7. Hybridisation of the 5′-terminal sequence, as in 21, exerted such a significant rate retarding effect that no cleavage was observed in 3 months. This difference can hardly be accounted for by slightly different structures of the 5′- and 3′-dangling ends. In duplexes 23 and 24, sequence 6 forms a double helix with complementary sequences having either 1 or 5 extra nt. The scissile phosphodiester bond within 23 and 24 lies just opposite to the one or five base bulge. When the reaction solution contains equal amounts of 6 and its complementary sequence, the 5′-UpA-3′ bond in either of the duplexes (23 or 24) is at least 100 times less reactive compared with that in a linear single strand (6) not engaged in a duplex formation.

DISCUSSION

The reactivity of phosphodiester bonds within RNA oligonucleotides having no defined secondary structure seems to be strongly dependent on the base sequence of the substrate at a low temperature, whereas no such differences are observed with dinucleotide monophosphates or tetrameric oligonucleotides. Although some of the oligonucleotides used could be argued to form self-complementary duplexes by G-U and canonical base pairing, it should be noted that neither MFOLD free energy calculations nor melting experiments lent any support to noticeably stable secondary structure under the conditions used in the kinetic experiments. The variation in cleavage rate is much larger in buffer solutions at pH 8.5 than under highly alkaline conditions. The reactivity is not particularly sensitive to the nearest neighbourhood of the scissile bond, but the sequence farther within the chain often plays an even more important role. Both rate accelerations and retardations greater than one order of magnitude depending on sequence are observed on inserting a dinucleotide monophosphate into a dodecameric oligonucleotide. It is also clear that the sequence required to obtain the highest possible cleavage rate of a particular phosphodiester bond depends on the identity of nearest neighbours: the optimal sequence for the cleavage of a 5′-UpA-3′ is different from that giving the highest lability for 5′-CpA-3′. Hence, the results obtained in this work do not lend support to the view that 5′-UpA-3′ bonds as such are inherently weaker than the other phosphodiester bonds and, hence, cleavage at such bonds could be expected to be always preferred.

The marked sensitivity of the cleavage of phosphodiester bonds to the base sequence results from the fact that the oligonucleotide chain may adapt a structure that either facilitates or retards the transesterification. The structure most probably is stabilised by intramolecular interactions, such as base stacking or hydrogen bonding. Rate retardations may at least be partly attributed to base stacking. Base stacking is known to stabilise the A-form helical conformation of single-stranded oligonucleotides (33,38,39). It is known that in double-helical structures the cleavage of phosphodiester bonds is markedly retarded in all likelihood due to the fact that the 5′-linked nucleoside cannot easily adapt to a position required for in-line displacement (30,40–42). Accordingly, interactions that favour the formation of helical-type conformations are rate retarding. Consistent with formation of a semi-stable ordered oligonucleotide structure, the reactivity differences almost completely disappear at an elevated temperature. The reactivity difference between the 5′-UpA-3′ phosphodiester bonds within 10 and 11 serves as an illustrative example of an effect of stacking. Consistent with the known self-stacking tendency of the cytosine bases (39), the C-rich sequences of 10 stack upon each other, which forces the scissile phosphodiester bond in an unfavourable conformation. With 11, such continuous stacking is not possible and, consequently, the conformation of the scissile phosphodiester bond is less restricted allowing a more efficient cleavage. The interactions within the 5′-terminal sequence appear to be more important than those within the 3′-terminal sequence: the hybridisation in 21 has a more significant effect on the reactivity than that in 22.

Rate accelerations compared with dinucleotide monophosphates or entirely random coils are more difficult to explain. A rough correlation has been established previously between the ‘in-line fitness’ and the rate of the cleavage of phosphodiester bonds within ATP-binding RNA aptamers (43). Reactivity differences of 500-fold were observed, and phosphodiester bonds having an initial state conformation that resembles the transition state of the in-line displacement were approximately 10 times more reactive than a linear reference molecule. Attribution of marked rate accelerations observed with relatively short oligonucleotides to a favourable initial state conformation is not, however, fully consistent with the Curtin–Hammett principle. As long as the conformational changes take place at a rapid pre-equilibrium step they should not influence the reaction rate. Evidently the transition state must somehow be stabilised compared with that derived from a fully flexible reference compound. The tentative suggestion given by Kierzek and coworkers (25) for rapid cleavage of 5′-NpA-3′ (N = C, U) in the presence of polyvinylpyrrole appears attractive. The underlying idea is that the hydrogen bonding network around the scissile bond, attacking 2′-OH and neighbouring bases may be the source of rate accelerations. Such hydrogen bonding may accelerate the reaction by abstracting the proton from the attacking 2′-OH, and/or by protonating the anionic phosphodiester or the leaving 5′-nucleoside oxyanion (3,44,45). The required hydrogen-bonding network is undoubtedly very sensitive to the identity and mutual orientation of the participating groups, including the H-bond donors and acceptors of the base moiety of the 5′-linked nucleoside which may well play a major role in the formation of a proper hydration pattern, as discussed by Kierzek and coworkers (25). Hence, changes in the base sequence even far away from the scissile bond may still play a role. The reactivity differences largely disappear at high alkality, suggesting that the hydrogen bonding becomes less important on approaching the pK_a value of the entering and leaving nucleophile. At high pH, the 2′-oxyanion attacks and the 5′-oxyanion leaves without any proton transfer in the transition state. While this mechanism is undoubtedly utilised also at pH 8.5 (3,13), a concurrent reaction that involves rate-limiting water-mediated proton transfer from the 2′-hydrogen group to the departing 5′-oxygen might be responsible for the observed rate accelerations.

Conclusions

The base sequence has a significant effect on the reactivity of phosphodiester bonds. Not only the neighbouring nucleic acid bases, but also those farther in the molecule contribute to the reactivity. Compared with fully flexible reference compounds, both rate acceleration and retardation of more than one order of magnitude are observed. The oligonucleotides that exhibit rate retardations in all likelihood adopt, due to base stacking and hydrogen bonding, a structure that hinders the free rotation of the phosphodiester bond and hence the reaction by an in-line mechanism. The exceptionally readily cleaved oligonucleotides are possibly making use of a preferential hydration pattern around the scissile bond that, by hydrogen bonding interactions, increases the nucleophilicity of the 2′-OH, increases the electrophilicity of the phosphate and/or facilitates the departure of the leaving group by proton transfer.