Nucleic Acids Research

Structures of oligonucleotides

Figure 1 records the structures of the chimeric ribo/2[prime]-O-methylribo oligonucleotides studied. All these oligomers contain one ribonucleoside unit, the rest of the nucleoside units being 2[prime]-O-methylated. Accordingly, only one of the phosphodiester bonds may be cleaved in each oligomer by intramolecular nucleophilic attack by the neighbouring 2[prime]-hydroxy function, which allows an accurate determination of the cleavage rate of this particular bond. The underlying assumption for the use of this type of chimeric oligomer as a mimic of oligoribonucleotides is that 2[prime]-O-methylation does not severely alter the secondary structure. This assumption received considerable support from a recent NMR spectroscopic study on the duplex r(CGCGCG)₂ and its 2[prime]-O-methyl counterpart (50).

Figure 1. Structures of the chimeric oligonucleotides studied. Bold letters refer to ribonucleosides, the other nucleosides being 2[prime]-O-methylated. The position of the scissile phosphodiester bond is indicated by an arrow. With 4, the scissile bond is a 2[prime],5[prime]-bond. All the other phosphodiester bonds in structures 1-10 are 3[prime],5[prime]-bonds.

Oligomer 1 serves in the present study as a model for a linear single-stranded ribonucleotide having no ordered secondary structure. Oligomers 2 and 3 mimic the widely occurring (51,52) and well-established 5[prime]-GCAA-3[prime] tetraloop structure (53,54), closed in this particular case with a double helical stem of four GC base pairs. They differ from each other only with respect to the position of the scissile 3[prime]-ribo phosphodiester bond: with 2 this bond is situated in the loop, with 3 in the stem. The melting point of 2 was observed to be 90 $ 1°C in the concentration range 1-10 µM at pH 7 (I = 0.1 M with NaCl). At a pH as high as 10, the melting point was 64.9 $ 0.5°C. This high thermal stability of the hairpin structure is consistent with a previous study (54) on a 1 bp shorter hairpin, 5[prime]-GGCGCAAGCC-3[prime], which exhibited a concentration-independent melting point of 71.0°C at the same pH and ionic strength. Accordingly, 2 and 3 undoubtedly exist as hairpins under the experimental conditions used in the kinetic measurements(T = 50°C, pH 4.8-7.5). The same conclusion in all likelihood also applies to oligomer 4, having the same base sequence but a scissile 2[prime],5[prime]-CpA phosphodiester bond in the loop instead of the 3[prime],5[prime]-CpA bond of 2. The melting point of this hairpin proved to be even higher than that of 2 (~95°C).

Oligomers 5-10 may also be expected to exist as hairpins under the conditions of the kinetic measurements (T = 50°C). They all contain the same highly stable stem structure as 2-4. We believe that with 5 the closing base pair really is the CG pair, although formation of an additional AU base pair is in principle possible. For comparison, the recent NMR spectroscopic studies of Sich et al. (55) show that the hairpin 5[prime]-GGCGUACGUUUCGUACGCC-3[prime] contains a pentaloop (indicated by bold letters), not a 3 nt loop, which could be obtained by closing the next GC base pair. Accordingly, it appears reasonable to assume that 5 also contains a pentaloop, not a 3 nt loop. The melting point of this hairpin was observed to be very similar to that of 2 (89 $ 1°C). However, a small change in hypochromicity was observed at 42°C, which could possibly refer to an additional AU interaction. With 6-10 formation of an additional AU base pair appears more probable and the kinetic results discussed below also suggest that this kind of weak interaction may exist. Accordingly, 6 and 7 may be regarded as tetralooped structures that rather easily undergo conversion to hexaloops and 8-10 as pentalooped structures that may be converted to heptaloops. The melting curves of these oligomers exhibited only one transition: 6 at 87 $ 1°C and 8 at 89 $ 1°C. For comparison, the melting temperature of a duplex of a fully complementary 15mer 2[prime]-O-methyl oligoribonucleotide containing seven CG pairs (but being unable to form a hairpin) has been reported to be 69°C under similar experimental conditions (56). Accordingly, the melting temperatures of nearly 90°C observed for 6 and 8 may be taken as an indication of a stable hairpin structure.

The cleavage experiments

The metal ion-promoted cleavage of oligomers 1-10 was studied under three different conditions: (i) in 5 mM Zn(NO₃)₂ at pH 6.5; (ii) in 2 mM Zn²⁺[12]aneN₃ (11) at pH 7.5; (iii) in 5 mM Pb(OAc)₂ at pH 4.8. Moreover, some measurements were carried out in 30 mM MgCl₂ at pH 7.4. All the reactions were carried out at 50°C and at an ionic strength of 0.1 M adjusted with NaNO₃. The compositions of the aliquots withdrawn at appropriate intervals were analysed by ion exchange HPLC. The pseudo first order rate constants observed for cleavage of 1-10 to two shorter oligomers by rupture of the scissile 3[prime],5[prime] (or 2[prime],5[prime])-ribo phosphodiester bond are listed in Table 1.

Table 1. Pseudo first order rate constants of the cleavage of chimeric ribo/2[prime]-O-methylribo oligonucleotides (1-10) at 50°C in excess Zn²⁺, Zn²⁺[12]aneN₃ and Pb²⁺ (I = 0.1 M with NaNO₃)

Oligonucleotide	k(Zn2+)/ (10-6/s)a	k(Zn2+[12]aneN3)/ (10-6/s)b	k(Pb2+)/ (10-6/s)c
1	6.3 ± 0.1	9.2 ± 0.4	d
2	<0.2	2.5 ± 0.1	6.2 ± 0.3
3	<0.2	<0.2	<0.2
4	1.1 ± 0.1	3.0 ± 0.2	7.8 ± 0.4
5	<0.2	2.4 ± 0.1	<0.2
6	<0.2	3.7 ± 0.1	<0.2
7	<0.2	0.5 ± 0.1	<0.2
8	2.2 ± 0.1	7.3 ± 0.3	11.0 ± 0.4
9	0.9 ± 0.1	5.9 ± 0.2	8.6 ± 0.2
10	<0.2	1.4 ± 0.1	<0.2

^aIn 5 mM Zn(NO₃)₂ at pH 6.5 adjusted with HEPES buffer.
^bIn 2 mM Zn[12]aneN₃(NO₃)₂ at pH 7.5 adjusted with HEPES buffer.
^cIn 5 mM Pb(NO₃)₂ at pH 4.8 adjusted with acetate buffer.
^dCould not be determined owing to precipitation.

The results of the present study fully corroborate previous observations (19 and references therein, 24,39,40), according to which phosphodiester bonds within a double helix are less readily cleaved than those within a random single-stranded region. No cleavage of the scissile 3[prime],5[prime]-CpC ribo phosphodiester bond within the stem of 3 was observed over 5 days with any of the catalysts used, while the half-lives for the cleavage of single-stranded 1 with Zn²⁺ and Zn²⁺[12]aneN₃ were 31 and 21 h respectively. Solubility problems prevented accurate measurements on Pb²⁺-induced cleavage of 1. Consistent with previous observations on cleavage of 3[prime],5[prime]-UpU (57) and poly(U) (58), Mg²⁺ was observed to be catalytically virtually inactive. No cleavage of either 1 or 2 was observed in 30 mM MgCl₂ solution at pH 7.4 over 5 days.

The kinetic data in Table 1 also clearly show that various cleaving agents exhibit rather dissimilar susceptibilities to the secondary structure of RNA, although with each of them the phosphodiester bonds within loops were observed to be less reactive than those in a linear single strand. The cleaving activity of various metal ion catalysts is discussed below in more detail.

Cleavage of hairpin loops by Zn²⁺[12]aneN₃

Zn²⁺[12]aneN₃ cleaved the scissile 3[prime],5[prime]-ribo phosphodiester bond within all the hairpin loops studied (2 and 5-10), the half-lives ranging from 26 (8) to 140 h (10). As mentioned above, cleavage within a loop is always slower than that of the linear single-stranded oligomer 1, but the rate retardation is only moderate. With 2, having the scissile bond in the middle of the well-established tetraloop 5[prime]-GCAA-3[prime], the cleavage rate was 27% of that of the similar 3[prime],5[prime]-CpA bond within the linear structure 1. In pentalooped hairpin 5, the 5[prime]-linked nucleoside of the scissile 3[prime],5[prime]-ApG bond participates in the base pairing that closes the pentaloop structure. This fact does not, however, seem to bring about any marked extra stabilization: the cleavage rate of the AG bond is very similar to that of the CA bond within the tetraloop of 2. It is also worth noting that the 2[prime],5[prime]-CpA bond in the tetraloop of 4 is cleaved approximately as rapidly as the corresponding 3[prime],5[prime]-CpA bond in 2.

With hairpins 6-10, the cleavage rates of scissile phosphodiester bonds within the loop range from 5 to 79% of that in the linear reference oligomer 1. As mentioned above, the stem-loop structure of these oligomers is not completely unambiguous, since at least one UA base pair may possibly be formed in addition to the four GC pairs of the stem. In other words, 6 and 7 may be either tetra- or hexalooped and 8-10 either penta- or heptalooped. The kinetic data suggest that the additional UA interaction exists, but this base pairing is weak compared with the GC and CG pairs in the stem. As discussed above, phosphodiester bonds within duplex regions are hydrolysed much less readily than those within single-stranded regions. In oligomers 7 and 10, the scissile phosphodiester bond is situated between a 5[prime]-linked G that is certainly engaged in a double helix and a 3[prime]-linked A that possibly participates in formation of an additional UA pair. Consistent with the assumed weak UA base pairing, this 3[prime],5[prime]-ApG bond is cleaved less readily than the corresponding 3[prime],5[prime]-ApG bond within the pentaloop of hairpin 5, but still at a measurable rate. The cleavage rates of 7 and 10 are 5 and 15% of that of the linear reference oligomer 1 respectively. When the scissile bond is situated farther away from the GC stem, the cleavage rate is higher than within the tetraloop of 2 or pentaloop of 5, namely 40, 79 and 64% of that of 1 with hairpins 6, 8 and 9 respectively. In summary, Zn²⁺[12]aneN₃ is able to cleave all the scissile phosphodiester bonds within the loops, even when the 3[prime]-linked ribonucleoside is engaged in weak additional base pairing. Only when both the 3[prime]- and 5[prime]-linked nucleosides participate in formation of a stable double helix is a really marked stabilization (>2 orders of magnitude) towards Zn²⁺ chelate-promoted cleavage encountered.

Cleavage of hairpin loops by Zn²⁺

The Zn²⁺ aquo ion appears to have considerably more rigorous geometric requirements for the target oligonucleotide than Zn²⁺[12]aneN₃. Although the linear oligomer 1 is cleaved nearly as readily by Zn²⁺ and Zn²⁺[12]aneN₃, Zn²⁺ was observed to cleave only 8 and 9 among the entirely 3[prime],5[prime]-bonded hairpins studied, and even these less readily than the Zn²⁺ chelate. As discussed above, these two oligomers may be expected to rather easily adopt a flexible heptaloop structure with opening of the partially closing weak UA base pair. The higher cleavage rate of 8 compared with that of 9 suggests that the reactivity of phosphodiester bonds is increased on approaching the central region of the loop. Cleavage by Zn²⁺ hence appears to be severely retarded as soon as the flexibility of single-stranded RNA is somehow restricted. While Zn²⁺ did not cleave the 3[prime],5[prime]-CpA bond within the tetraloop of 2, the corresponding 2[prime],5[prime]-CpA bond in 4 was cleaved at a measurable rate. The behaviour of the Zn²⁺ aquo ion differs also in this respect from that of its chelate 11.

The dissimilar susceptibility of Zn²⁺- and Zn²⁺[12]aneN₃-promoted cleavage to RNA secondary structure is consistent with our previous suggestions on the binding mode of different metal ion catalysts. We have proposed that the Zn²⁺ aquo ion binds bidentately to oligonucleotides, bridging two phosphate functions, while Zn²⁺[12]aneN₃, three coordination sites of which are filled with nitrogen atoms, may bind only monodentately (38,44). The inability of the Zn²⁺ aquo ion to cleave phosphodiester bonds within loop structures could hence be attributed to the fact that it cannot bridge two phosphate groups within a conformationally constrained loop. Hence, the Zn²⁺ aquo ion binds only monodentately and the phosphodiester bonds within loops become less reactive that those in linear RNA molecules. As the loop becomes larger and the RNA strand more relaxed, bidentate binding becomes easier and the observed reactivity approaches that of a linear RNA molecule. Zn²⁺[12]aneN₃-induced cleavage, in turn, is less sensitive to the structure, because this cleaving agent is bound to only one phosphate group even in the case of a linear molecule.

Cleavage of hairpin loops by Pb²⁺

Cleavage by Pb²⁺ also seems to be more sensitive to the secondary structure of RNA than that by Zn²⁺[12]aneN₃. However, several of the hairpins studied, namely 2, 4, 8 and 9, were observed to be cleaved by Pb²⁺. In contrast to Zn²⁺[12]aneN₃, Pb²⁺ was unable to cleave 5, 7 and 10, where the 3[prime]-ribonucleotide of the scissile 3[prime],5[prime]-ApG bond may be engaged in weak additional UA base pairing. As with Zn²⁺, cleavage within the larger loop of 8 and 9 is faster than that within the tetraloop of 2, although the reactivity difference is rather modest.

The differences in cleavage patterns between Zn²⁺ and Pb²⁺ may well result from different affinities of these cations for the base moieties. The binding of Zn²⁺ to the nucleoside bases is weak (59) and it does not seem to have any significant effect on the reactivity of phosphodiester bonds of RNA (60). One may tentatively assume that Pb²⁺, as a relatively soft Lewis acid, interacts with the base moieties more extensively than Zn²⁺ and, hence, the cleavage rates observed may be more dependent on the identity of the neighbouring bases. It has very recently been suggested that the susceptibility of RNA terminal loops to Pb²⁺-induced hydrolysis strongly depends on their base composition (61).

Conclusions

On the basis of the results obtained in this work, it seems clear that a loop structure as such does not enhance metal ion-promoted cleavage of RNA phosphodiester bonds, but a similar bond within a linear RNA sequence is always more reactive. In the presence of Zn²⁺-based catalysts, the reactivity of phosphodiester bonds within loops approaches that of a linear molecule as the size of the single-stranded region increases and the structure becomes more relaxed. Zn²⁺-promoted cleavage appears to be more susceptible to the secondary structure of RNA than the reaction induced by Zn²⁺[12]aneN₃. This dissimilarity is consistent with the previously suggested difference between the binding mode of these species: bidentate binding of Zn²⁺ to two different phosphates and monodentate binding of Zn²⁺[12]aneN₃. A flexible single strand is needed to meet the geometric requirements of Zn²⁺ aquo ion binding. The cleavage pattern of Pb²⁺ resembles that of Zn²⁺. Compared with these cleaving agents, Mg²⁺ is rather ineffective.

Materials

The chimeric ribo/2[prime]-O-methylribo oligonucleotides (1-3 and 5-10) were assembled from commercial 2[prime]-O-methylated (Glenn Research) and 2[prime]-O-[1-(2-fluorophenyl)4-methoxypiperidin-4-yl] (2[prime]-O-Fpmp)-protected (Cruachem) building blocks by conventional phosphoramidite strategy, applying the standard RNA coupling protocol of the ABI 392 DNA/RNA Synthesizer. The completed sequences (DMTr-on) were released from the support and the base and phosphate moiety protections were removed by treating the support with concentrated aqueous ammonia for 2.5 days at room temperature. The crude oligomer was isolated by RP HPLC on a LiChroCART column (250 × 10 mm, 5 µm). The conditions for separation were: flow rate 3 ml/min, buffer A 0.05 M AcONH₄ in water, buffer B 0.05 M AcONH₄ in aqueous MeCN (65% v/v), 0-5 min isocratic elution with a 4:1 mixture of A and B, 5-35 min a linear gradient to a 1:4 mixture of A and B. After desalting on the same column, the 5[prime]-terminal DMTr group and the 2[prime]-O-Fpmp group were removed with sterile 0.01 M aqueous hydrogen chloride (20 h at room temperature). The reaction was stopped by adjusting the pH to 5.7 with triethylamine. The deprotected oligonucleotide was isolated by ion exchange HPLC on a SynChropak A×300 IC column (250 × 4.6 mm, 5 µm) and desalted. The conditions for the ion exchange separation were: flow rate 1 ml/min, buffer A 0.05 M phosphate buffer (pH 5,6) in aqueous formamide (50% v/v), buffer B as buffer A but containing 0.6 M (NH₄)₂SO₄, 0-5 min isocratic elution with a 9:1 mixture of A and B, 5-35 min a linear gradient to a 3:7 mixture of A and B. The desalting was performed on the LiChroCART column described above. The elution pattern was as follows: 0-5 min 0.1 M AcONH₄, 5-21 min water, 21-42 min aqueous MeCN (36% v/v).

Oligomer 4, containing a 2,5-phosphodiester bond, was assembled using N⁴-benzoyl-5[prime]-O-(4,4[prime]-dimethoxytrityl)-3[prime]-O-tert-butyldimethylsilylcytidine 2[prime]-O-(2-cyanoethyl)-N,N-diisopropylphosphoramidite as a building block instead of the 2[prime]-O-Fpmp-protected cytidine phosphoramidite used to obtain 2. Silyl protection was removed by conventional treatment with tetrabutyl-ammonium fluoride. The deprotected oligomer was purified and desalted as indicated above.

1,5,9-Triazacyclododecane was a product of Aldrich and was used as received to prepare the Zn²⁺[12]aneN₃ chelate (11). All the other reagents were of reagent grade.

Melting temperature measurements of the oligonucleotides

The melting curves were recorded on a Perkin-Elmer Lambda 2 UV spectrophotometer 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. The change in UV absorption was followed at 260 nm in 10 mM Tris buffer at pH 7.0 (I = 0.1 M with NaCl). The results of triplicate measurements usually agreed within $1°C. The melting points could not be obtained at higher accuracy, since, owing to the high melting temperature, the curves did not sufficiently level off to a constant value in the temperature range that could be employed (T < 100°C).

Kinetic measurements

The reactions were carried out in tightly sealed glass tubes immersed in a water bath, the temperature of which was maintained at 50.0 $ 0.1°C. Zn²⁺-, Zn²⁺[12]aneN₃- and Mg²⁺-induced cleavage were studied in HEPES buffers and Pb²⁺-induced cleavage in an acetate buffer. The metal ions were introduced as nitrates. The ionic strength was adjusted to 0.1 M with sodium nitrate. The total volume of the reaction solution was 1.5 ml and it contained 1-2 OD units of the appropriate oligonucleotide.

Aliquots (100 µl) withdrawn at suitable intervals were immediately cooled to 0°C and excess EDTA was added to stop the reaction. Typically from 10 to 12 aliquots were taken. The composition of the samples was determined by ion exchange HPLC. The chromatographic conditions were similar to those described above for purification of the synthesized oligonucleotides (SynChropak A×300 IC column). The pseudo first order rate constants (metal ion in excess) for cleavage of the starting oligonucleotide into two fragments were calculated by applying the integrated first order rate law to disappearance of the starting material.