Site-specific strand breaks in RNA produced by ¹²⁵I radiodecay

Abstract

Decay of ¹²⁵I produces a shower of low energy electrons (Auger electrons) that cause strand breaks in DNA in a distance-dependent manner with 90% of the breaks located within 10 bp from the decay site. We studied strand breaks in RNA molecules produced by decay of ¹²⁵I incorporated into complementary DNA oligonucleotides forming RNA/DNA duplexes with the target RNA. The frequencies and distribution of the breaks were unaffected by the presence of the free radical scavenger dimethyl sulfoxide (DMSO) or by freezing of the samples. Therefore, as was the case with DNA, most of the breaks in RNA were direct rather than caused by diffusible free radicals produced in water. The distribution of break frequencies at individual bases in RNA molecules is narrower, with a maximum shifted to the 3′-end with respect to the distribution of breaks in DNA molecules of the same sequence. This correlates with the distances from the radioiodine to the sugars of the corresponding bases in A-form (RNA/DNA duplex) and B-form (DNA/DNA duplex) DNA. Interestingly, when ¹²⁵I was located close to the end of the antisense DNA oligonucleotide, we observed breaks in RNA beyond the RNA/DNA duplex region. This was not the case for a control DNA/DNA hybrid of the same sequence. We assume that for the RNA there is an interaction between the RNA/DNA duplex region and the single-stranded RNA tail, and we propose a model for such an interaction. This report demonstrates that ¹²⁵I radioprobing of RNA could be a powerful method to study both local conformation and global folding of RNA molecules.

INTRODUCTION

Methods based on probing the chemical reactivity of DNA and RNA bases provide valuable information on conformation of nucleic acids and are widely utilized to complement and verify structural data obtained by X-ray crystallography and NMR. One of the most popular approaches is hydroxyl radical footprinting, which is based on measuring breaks in nucleic acids chains produced by OH radicals (1). Closely related to hydroxyl radical footprinting is analysis of breaks in DNA and RNA produced by ionizing radiation, which has been applied to study structural changes in DNA and folding of RNA (2–4). The above techniques utilize nucleic acid attack by diffusible OH radicals generated in solution or delivered to the site of interest. Recently, we suggested a methodologically similar, but in principle different, approach that we called radioprobing (5). It is based on the analysis of breaks in DNA produced close to the site of decay of an Auger electron emitting radioisotope. The advantage of radioprobing is that the breaks produced by Auger electrons are mostly from the direct action of radiation and, thus, the frequency and distribution of these breaks provide information on the distance from the decay site to the corresponding bases.

Auger electron emitters are a class of radioisotopes that decay by electron capture or internal conversion producing a number of low energy electrons (6–8). The range of these low energy electrons in biological matter is very short and most of their energy is deposited within nanometers of the decay site. It is generally assumed that the strand breaks are due to energy deposition in the sugar–phosphate moiety since damage to the DNA bases generally does not lead to strand scission (9). Therefore, the frequency of DNA strand breaks at any given base should be inversely related to the distance from the decay site to the sugar of this base. In practice, this simple distance dependence could be altered by two other components of Auger emitter decay: the positively charged daughter nucleus and the diffusible OH radicals generated in solution. The positive charge on the daughter nucleus strips electrons from the neighboring bonds and can ultimately result in breakage of DNA strands (8). This effect is limited mostly to the bases located next to the decay site (10). The contribution of OH radicals generated in water to the total frequency of breaks becomes comparable with that of Auger electrons only at distances beyond 7 nt from the decay site (10).

We previously tested radioprobing on a DNA–protein complex with a well-known three-dimensional structure (11). As the Auger electron emitter we used ¹²⁵I, easily incorporated into DNA and widely available. ¹²⁵I was incorporated as [¹²⁵I]deoxycytosine in a single position of a short DNA duplex containing the Escherichia coli cAMP receptor (CRP)-binding sequence (11). We demonstrated that the frequencies of breaks produced by decay of ¹²⁵I in the complementary strand of the duplex correctly reflected the distances to the corresponding sugars calculated from the three-dimensional structure in the range 10–30 Å. These frequencies were also sensitive to the few angstrom changes in the distances caused by binding of CRP to the duplex. We also applied radioprobing to study the spatial arrangement of DNA strands within the E.coli RecA protein synaptic complex (12) and T7 RNA polymerase transcription elongation complex (13).

We recognize that radioprobing could be an informative tool to study folding of RNA molecules as well as conformational changes in RNA upon binding with proteins. However, there are currently no specific reports on breaks in RNA produced by decay of Auger emitters, although some authors noted that RNA is more resistant to attack by free radicals than DNA (14,15). Since strand breaks produced by ionizing radiation are ultimately products of free radical reactions, one may expect that the yield and distribution of breaks produced by Auger electrons will be different in DNA and RNA. To address this question, we studied strand breaks in RNA molecules produced by decay of ¹²⁵I incorporated into complementary DNA oligonucleotides forming RNA/DNA duplexes with the target RNA and compared them with the breaks in the reciprocal DNA/DNA duplexes.

MATERIALS AND METHODS

Materials

RNA phosphoroamidate monomers, containing a 2′-O-triisopropyloxymethyl (TOM) protecting group [A-TOM phosphoroamidate, C-TOM β-cyanoethyl (CE) phosphoroamidate, G-TOM-CE phosphoroamidate and U-TOM-CE phosphoroamidate], ABI style columns (0.2 µmol) and other chemicals for the ABI-394 DNA synthesizer were purchased from Glen Research. Tetrabutylammonium fluoride solution (1 M in tetrahydrofuran) and 33% ethanolic methylamine solution were purchased from Fluka Chemika-BioChemika. Aqueous methylamine (40%) and Spin-X centrifuge tube filters (0.22 µm, cellulose acetate) were obtained from Fisher Scientific. MicroSpin™ G-25 and G-50 columns were purchased from Amersham Pharmacia Biotech Inc. The SequaGel Kit and UltraPure acrylamide/bisacrylamide solution were from National Diagnostic, TEMED and dimethyl sulfoxide (DMSO) from Sigma, ammonium persulfate from Bio-Rad, 10× TBE from Advanced Biotechnologies, DEPC-treated water from Research Genetics and TE buffer (pH 7.4) from Quality Biological Inc. PBS solution (10×) and RNase T1 were purchased from Gibco BRL. T4 RNA ligase, DNA polymerase I large (Klenow) fragment, T4 polynucleotide kinase and alkaline phosphatase (CIP) were obtained from New England BioLabs, terminal transferase was from Promega and the MEGAscript in vitro Transcription kit was from Ambion. Streptavidin magnetic beads (Dynabeads M-280 Streptavidin) were from Dynal Inc. [5-¹²⁵I]dCTP (2200 Ci/mmol), [γ-³²P]ATP (6000 Ci/mmol) and [5′-³²P]pCp (3000 Ci/mmol) were from Du-Pont NEN Research Products.

Synthesis and purification of oligonucleotides

DNA oligonucleotides were synthesized by standard phosphoramidate chemistry and 48 nt length RNA (corresponding to the +1428/+1476 region of bcl-2 mRNA) was synthesized using the TOM-protected RNA method according to the Glen Research protocol on an ABI-394 DNA synthesizer (Applied Biosystems). Oligonucleotides after synthesis and deprotection were purified from 12 or 20% urea–polyacrylamide gels as follows. Oligonucleotide bands in the gel placed on a white plate were visualized under a UV lamp (UVGP-58; UVP Inc., San Gabriel, CA). DNA or RNA was eluted from the pieces of gel by shaking them in 0.5 ml of TE buffer (pH 7.4) for 2–4 h. After removing residual polyacrylamide by centrifugation in Spin-X centrifuge tube filters, oligonucleotide solutions were concentrated in a vacuum concentrator (ATR, Laurel, MD) until the final volume was ∼100 µl. The purification was completed by desalting (twice) on water-equilibrated G-25 spin columns. For enzymatically produced bcl-2 RNA (51 nt) we synthesized two complementary single-stranded (ss)DNA oligonucleotides 74 nt in length. They were annealed to form a duplex containing the T7 promoter and the DNA duplex was used for RNA synthesis by multiple rounds of transcription by T7 RNA polymerase. The length of the T7 RNA transcript was 3 nt longer than the length of the chemically synthesized RNA due to an extra GGG on the 5′-end of the transcribed RNA.

Labeling of targets

Labeling at the 5′-end of chemically synthesized RNA and DNA oligonucleotides was performed with [γ-³²P]ATP and T4 polynucleotide kinase. RNAs, obtained by T7 RNA polymerase synthesis, were 5′-labeled with [γ-³²P]ATP and phage T4 polynucleotide kinase following dephosphorylation by alkaline phosphatase. The 3′-ends of the RNAs were labeled by ligation of [5′-³²P]pCp using T4 RNA ligase. DNA molecules of the same length and sequence were 3′-end-labeled with 3′-[α-³²P]dATP using terminal transferase from calf thymus. All labeling reactions were performed according to the manufacturer’s recommendations for the enzymes.

Sequencing of DNA and RNA

To determine break positions in RNA, we hydrolyzed ³²P-labeled RNA using RNase T1, which cleaves ssRNA with high specificity on the 3′-side of guanylic residues (16). An aliquot of 1 pmol of ³²P-labeled RNA was mixed with 10 µg of total non-labeled E.coli RNA and incubated with 0.1 U of RNase T1 at 37°C for 1–3 min in 50 mM Tris–HCl (pH 8.0), 10 mM MgCl_2, 50 mM NaCl. Maxam–Gilbert G reactions were performed for DNA hydrolysis (17).

Iodine-125 antisense oligonucleotides (¹²⁵I-AO) synthesis

The antisense oligonucleotides AO-I, AO-II and A-IL (Fig. 1) were labeled with ¹²⁵I by primer extension reactions with the Klenow fragment of DNA polymerase I on template DNA oligonucleotides. The biotinylated template for AO-I and AO-II was d(ATGGCGCACGCTGGGAGAAGTAGGAB ACG)-Bio. This template was removed with Streptavidin magnetic beads as described (18). For incorporation of ¹²⁵I in AO-L we used the extended template oligonucleo tide d(GGGAAGGATGGCGCACGCTGGGAGAAGTAGG TTACGA). ¹²⁵I-labeled AO-L was purified by urea–PAGE as described above.

Figure 1.

Targets and antisense oligonucleotides used for investigation of strand breaks in RNA. Positions of [¹²⁵I]dC are underlined. The AUG translation start codon is shown in bold.

Binding of ¹²⁵I-AO with RNA targets

Target [³²P]RNA or [³²P]DNA (50–100 nM) was annealed to ¹²⁵I-AO (75–200 nM) in 1× PBS at 75°C for 3 min. Then the binding mixtures (total volume 10–20 µl) were slowly cooled to room temperature for 20–30 min, samples were evenly divided and DMSO was added to half to a final concentration of 10%. Samples were aliquoted and kept at –70°C or at +5°C for accumulation of decays.

Break analysis

Breaks in RNA were analyzed in 12 or 20% denaturing polyacrylamide gels. The gels were vacuum dried on Whatman 3MM Chr paper. Data were analyzed using a Bioimaging Analyzer (BAS 1500; Fuji). The positions of the breaks in target DNA and RNA were determined by comparison with the positions of bands in corresponding sequencing DNA and RNA lines. The break frequencies at each particular target position were calculated as described (5).

RESULTS

Strand breaks in RNA target

For the experiments on RNA cleavage by ¹²⁵I, we chemically synthesized a 48 nt long RNA target corresponding to the +1428/+1476 region of bcl-2 mRNA. This region contains an AUG translation start codon. Several publications have demonstrated successful targeting of this region by AO (19,20).

AO were labeled with [¹²⁵I]dCTP at one or two positions (termed AO-I and AO-II, respectively, Fig. 1) by primer extension as described in Materials and Methods. Binding of purified ¹²⁵I-AOs to RNA and DNA targets in all the following experiments was always confirmed by native PAGE. A typical band shift assay presented in Figure 2 shows that all the targets were bound to AOs; there were no bands corresponding to free ³²P-labeled targets in lanes 1, 2, 5 and 6. To ensure that all target molecules formed duplexes, binding was usually performed in an excess of ¹²⁵I-AO, thus the samples in lanes 1, 2, 5 and 6 contain some free ¹²⁵I-AO.

Figure 2.

Binding of antisense oligonucleotides to target RNA and DNA analyzed by 12% native PAGE. Lanes 3 and 7, [³²P]DNA and [³²P]RNA targets, respectively; lanes 1, 2, 5 and 6, [³²P]DNA and [³²P]RNA targets incubated with [¹²⁵I]AO-I (lanes 2 and 6 with DMSO); lanes 4 and 8, [³²P]DNA and [³²P]RNA targets incubated with excess cold AO-I.

To increase the yield of breaks, initial experiments were carried out using AO-II labeled with two [¹²⁵I]dC and chemically synthesized RNA. Duplexes of AO-II with 5′-³²P-labeled RNA and control samples were kept at –70°C or at +5°C. After 15 days of decay accumulation the strand breaks in target RNA were analyzed by denaturing PAGE. The results are shown in Figure 3B. Increased intensity of RNA fragments in the region 26–36 nt is observed in all the samples containing AO-II/RNA duplexes (lanes 2, 4, 10 and 12). These fragments correspond to the breaks in RNA opposite the positions of [¹²⁵I]dC in AO-II (Fig. 3A).

Figure 3.

Strand breaks in target DNA and RNA from two ¹²⁵I atoms incorporated in antisense oligonucleotides. (A) Scheme of the target and antisense oligonucleotides. (B) Autoradiogram of the 12% urea–PAGE showing RNA fragments after incubation of duplexes at –70°C (lanes 1–8) and at +5°C (lanes 9–12). Lanes 1, 3, 5, 7, 9 and 11, RNA targets without [¹²⁵I]oligonucleotides; lanes 2, 4, 6, 8, 10 and 12, RNA targets with [¹²⁵I]oligonucleotides, complementary AO-II (lanes 2, 4, 10 and 12) and non-complimentary AO-N (lanes 6 and 8). Samples in lanes 3, 4, 7, 8, 11 and 12 contained 10% DMSO. Lane 13, [³²P]DNA marker, 8–32 nt.

Incubation of target RNA with non-complementary ¹²⁵I-labeled oligonucleotide, AO-N, did not result in such breaks (lanes 6 and 8). Distributions of breaks were similar after incubation at –70°C and at +5°C (lanes 2 and 10). The presence of 10% DMSO, a free radical scavenger, did not significantly affect the frequency or distribution of breaks (lanes 4 and 12). This implies that we observed sequence-specific strand breaks in target RNA resulting from decay of ¹²⁵I incorporated in the antisense oligonucleotide and that the main reason for these breaks is the non-scavengeable component of ¹²⁵I decay.

This observed distribution of breaks in RNA, however, is different from the published distribution of ¹²⁵I-produced breaks in DNA (11); it is considerably wider. One possible explanation of this wider distribution of breaks is the presence of two consecutive [¹²⁵I]dC in AO-II.

Distributions of breaks in RNA and DNA targets

To further investigate the fine structure of the break distribution in RNA and compare it with that in the DNA target of the same sequence, we carried out experiments with single labeled AO-I. Target RNA for these experiments was synthesized by T7 RNA polymerase. This allowed us to significantly reduce the background in the denaturing PAGE lanes due to random fragmentation of the target. RNA and DNA targets were purified essentially identically after ³²P-labeling and annealed to AOs. Formation of duplexes was confirmed by gel shift assay. Figure 4 shows the results of the analyses of the breaks in DNA and RNA targets produced by AO-II (lanes 2 and 6) and AO-I (lanes 3 and 7). To determined the exact locations of the breaks relative to DNA or RNA sequence, the positions of the bands produced by ¹²⁵I decay were compared with a Maxam–Gilbert G sequencing ladder (lane 4) and with hydrolysis of target RNA with RNase T1 (lane 8). The intensities of the bands were measured and the frequencies of breaks at individual bases were calculated as described in Materials and Methods. The results are presented as graphs in Figure 4B.

Figure 4.

Analysis of break distribution in target DNA and RNA. (A) 12% urea–PAGE. Lane 1, [³²P]DNA; lane 2, [³²P]DNA with [¹²⁵I]-AO-II; lane 3, [³²P]DNA with [¹²⁵I]AO-I; lane 4, Maxam–Gilbert G sequencing line of [³²P]DNA; lane 5, [³²P]RNA; lane 6, [³²P]RNA with ¹²⁵I-AO-II; lane 7, [³²P]RNA with [¹²⁵I]-AO-I; lane 8, RNase T1 G sequencing line of [³²P]RNA. (B) Percentages of breaks at individual bases of the target molecules calculated from the data of the gel in (A) for DNA (open squares) and RNA (filled circles).

From the graph shown in Figure 4B it is clear that the distribution of breaks in RNA is not only wider than in the DNA target but has two distinct maxima. These maxima are even more apparent in the case of AO-I (Fig. 4B) and therefore could not result from the two [¹²⁵I]dC present in AO-II. One maximum of frequency of breaks is opposite the position of [¹²⁵I]dC in AO and another is shifted 6 nt to the 5′-end. The distribution of breaks in DNA in both cases has only one maximum that is shifted one base 5′ from the position of the G opposite to [¹²⁵I]dC. Since the intensity of ¹²⁵I-induced breaks is inversely proportional to the distance from the decay site, the second maximum means that the single-stranded 5′-ends of the RNA target fold into a tertiary structure (inter- or intra-molecular) such that the corresponding bases come into close proximity to the ¹²⁵I. To test this hypothesis we synthesized AO-L, which extends the DNA/RNA duplex to the 5′-end of the target, thus preventing RNA folding in this region.

Distribution of breaks in DNA/DNA and DNA/RNA duplexes

AO-L was labeled by primer extension and annealed with [5′-³²P]DNA and RNA targets. Duplex formation was confirmed by band shift assay. The results of the break analysis are shown in Figure 5. The positions of the breaks relative to the sequences of the targets were determined by comparison with DNA and RNA sequencing ladders as before (only the DNA sequencing ladder is shown). The frequencies of breaks at the individual bases were determined by measuring the intensities of the corresponding bands and are shown in Figure 5B. The frequencies of breaks in both the DNA and RNA targets are distributed around one maximum. The distribution of breaks in RNA is narrower than in DNA and has a pronounced maximum that is shifted one base towards the 3′-end from the position of the G opposite to [¹²⁵I]dC in AO-L. The overall yield of breaks, i.e. the ratio of the sum of the intensities of all the bands corresponding to breaks to the intensity of the band of the undamaged fragment, is very close for DNA and RNA. Thus we believe that the data presented in Figure 5B represent a true distribution of breaks in the RNA target within the DNA/RNA duplex, and the two maxima distribution of breaks in the RNA target shown in Figures 3 and 4 was the result of folding of the RNA molecule.

Figure 5.

Analysis of breaks in RNA and DNA with extended antisense oligonucleotide AO-L. (A) 12% urea–PAGE. Lane 1, [³²P]DNA; lane 2, [³²P]DNA with [¹²⁵I]AO-L; lane 3, [³²P]RNA; lane 4, [³²P]RNA with [¹²⁵I]AO-L; lane 5, Maxam–Gilbert G sequencing line of [³²P]DNA. (B) Percentages of breaks at individual bases of the target DNA (open squares) and RNA (filled circles) calculated from the data of the gel in (A). Error bars show standard deviations calculated from three independent measurements.

DISCUSSION

The data presented in this paper show that decay of ¹²⁵I in AO produces strand breaks in the complementary RNA molecule with a yield and distribution similar to those in a corresponding DNA target. Presence of the free radical scavenger DMSO and the temperature at which the decays accumulated had little effect on the yield of breaks in RNA. Therefore, we conclude that under our experimental conditions the majority of the breaks in RNA resulted from direct damage produced by low energy Auger electrons and, perhaps, the highly positively charged daughter nucleus (8), but not by diffusible OH radicals. Previously we found that most of the ¹²⁵I-induced breaks in DNA were also direct (5). It is important to note that in a control experiment with a non-complementary AO that was not bound to the target (Fig. 3, lanes 6 and 8), the same number of ¹²⁵I decays occurring in the same volume did not result in an increase in detectable breaks over background in RNA or DNA targets. Therefore, only the breaks produced by ¹²⁵I-AO bound to a target can be called sequence-specific, as opposed to the random breaks produced by dispersed ionizing radiation (IR).

A close analysis of the distributions of the frequencies of breaks in RNA/DNA and DNA/DNA duplexes reveals interesting details (Fig. 5). The distribution of breaks in RNA is narrower than in DNA and has a very pronounced maximum. These distributions correlate inversely with the distances from ¹²⁵I to the sugars in A-form (RNA/DNA duplex) and B-form (DNA/DNA duplex) double helices (Fig. 6). It is worth mentioning that the breaks at the bases located immediately next to the ¹²⁵I decay site mostly resulted from direct electron stripping due to the short-range effect of the highly positively charged daughter atom, the so-called ‘hot atom’ effect (10,21). Closest to the ¹²⁵I is the sugar of the base at position 1 (Fig. 6). Even though the difference in distance to this position in A-form and B-form duplexes is small, a very sharp distance dependence of the ‘hot atom’ effect may result in the noticeable increase in the frequency of breaks in RNA observed on Figure 5B. The distances to the bases distal to the 3′-side (nos 3–7) are longer in A-form than in B-form duplexes. This is in good agreement with the lower frequencies of breaks in the RNA/DNA hybrid and the overall narrower distribution of breaks in RNA.

Figure 6.

Distances from deoxyriboses to the ¹²⁵I atom in A-form (open squares) and B-form (filled circles) duplexes, calculated from the regular optimal structures obtained by Kosikov et al. (24).

IR-induced breakage in the polynucleotide chain is the final step in a sequence of free radical reactions. In the case of ¹²⁵I, the complete sequence of the reactions is not known, but most likely includes direct abstraction of protons from sugars by Auger electrons and attack by non-diffusible OH radicals created in DNA-bound water. We cannot rule out the possibility that the ribose of RNA is more resistant to attack by free radicals than the deoxyribose of DNA. There are several reports supporting this possibility (14,15). If this is the case, it would also help to explain the narrowness of the breaks distribution in RNA. To address the question one could study the reactivity of DNA and RNA to γ-radiation, for example, or, alternatively, compare breaks distributions in DNA/RNA hybrids and A-form DNA/DNA duplexes. We have shown previously that radioprobing, i.e. analysis of the distribution of breaks, was sensitive enough to detect DNA bending induced by binding with CRP (11). The results presented here show that radioprobing may also be utilized to detect intramolecular conformational changes such as a B→A transition.

Analysis of the distribution of breaks produced by decay of ¹²⁵I in AO-I and AO-II (Figs 3 and 4) reveals information on folding of the RNA target. There are two maxima of the frequency of breaks, one opposite [¹²⁵I]dC in the AO and another five bases towards the 5′-end, in the region beyond the AO/target duplex. This second maximum was found only in the RNA target and disappeared when the AO was extended to that region of RNA (AO-L, Fig. 5). This increase in the frequency of breaks indicates that the bases came into close proximity with the ¹²⁵I atom. This could, for example, happen if the single-stranded 5′-end of the target RNA forms a triplex-like structure with the RNA/AO duplex. Since the sequence in this region does not represent a polypurine-polypyrimidine motif and, thus, the bases cannot form canonical triads, it is possible that the triplex is stabilized by base–sugar interactions in the minor grove. Such triplexes were described previously for Group I introns (22) and were recently found to be the most abundant tertiary structure interaction in the large ribosomal subunit (23). They were called the ‘A-minor motif’ because they are often stabilized by interaction of adenines with 2′-hydroxyls of G-C base pairs. Two such triads could be formed between the target RNA tail and the AO/RNA duplex. We hypothesize that the single-stranded RNA tail folds back and binds to the duplex part of the same molecule, resulting in the bimodal breaks distributions observed in Figures 3 and 4. Even though further studies are required to prove or disprove this model, the above example demonstrates that radioprobing could be a powerful tool to study complex tertiary interaction and folding of RNA molecules.