A protonated base pair participating in rRNA tertiary structural interactions

Abstract

In the recently published X-ray crystallographic structure for the 50S subunit of Haloarcula marismortui ribosomes, residue U2546 of the 23S rRNA forms a non-Watson–Crick base pair with U2610. The corresponding residues in the secondary structure of the Escherichia coli 23S molecule are U2511 and C2575, and it follows that the latter base (C2575) should be protonated in order to form a base pair that is isostructural with its counterpart in H.marismortui. This prediction was demonstrated experimentally by reduction with sodium borohydride followed by primer extension analysis; borohydride is able to reduce positively charged bases, yielding products which block reverse transcription. In the course of the analysis a further charged base pair (AH⁺1528-G1543) was identified in the E.coli 23S molecule. Both charged pairs (U2511-CH⁺2575 and AH⁺1528-G1543) were only observed in the context of the intact ribosomal subunit and were not seen in deproteinized rRNA.

INTRODUCTION

Atomic structures for both the large and small ribosomal subunits have recently been determined by X-ray crystallography, that of the 50S subunit using ribosomes from Haloarcula marismortui (1) and that of the 30S subunit using ribosomes from Thermus thermophilus (2,3). On the other hand, the vast majority of the biochemical data relating to ribosomal structure has been obtained with ribosomes from Escherichia coli. For this organism, the best structural information so far available has been obtained by cryo-electron microscopy and, prior to the advent of X-ray crystallographic structures, computer models for the E.coli 16S and 23S rRNA were derived by combining the biochemical data with these cryo-electron microscopic structures (4,5). As part of a program to correlate the structures of the E.coli ribosomal subunits with those of the H.marismortui and T.thermophilus subunits, we have recently compared (6) the locations of several helices in the model for the E.coli 50S subunit (5) with their corresponding locations in the atomic structure (1). Furthermore, we have made use of the atomic structures (1,2) in order to evaluate the reliability of the data obtained for the E.coli ribosome from the various crosslinking and other chemical approaches which were exploited in the model-building studies (7).

Here we extend our correlation of the E.coli and H.marismortui structures to a more detailed level and describe the detection of a non-Watson–Crick CH⁺-U base pair in E.coli that is isostructural to a U-U pair in H.marismortui. The existence of this pair was at first predicted from a comparison of the respective secondary structures of the 23S rRNA molecules with the atomic structure for H.marismortui. Experimental proof was then obtained by reduction of the protonated base by treatment with sodium borohydride, followed by primer extension analysis to identify the locations of the reduction products. In addition to the CH⁺-U pair, a complete scan of the borohydride-treated 16S and 23S rRNAs by primer extension revealed an AH⁺-G protonated base pair in another region of the 23S rRNA. This technique should be generally applicable for the identification of base pairs containing protonated adenine or cytidine residues in any RNA or RNP.

MATERIALS AND METHODS

Ultrapure rNTPs and dNTPs were obtained from Pharmacia, alkaline phosphatase, T4 polynucletide kinase and reverse transcriptase from Boehringer Mannheim, [γ-³²P]ATP from Amersham, sodium borohydride from Sigma.

Sodium borohydride treatment

23S and 16S rRNA were treated with sodium borohydride in both 70S ribosomes and in the deproteinized state in solution. For this purpose an aqueous solution of sodium borohydride (15 µl, 3.3 mg/ml) was added to 60 pmol 23S plus 16S rRNA or 70S ribosomes in 135 µl of buffer (50 mM Tris–HCl pH 8.0, 70 mM NH₄Cl, 30 mM KCl, 7 mM MgCl₂, 1 mM DTT). The mixture was kept on ice for 30 min in the dark and the reaction was then stopped by addition of 15 µl of 3 M sodium acetate pH 5.5, and 450 µl of ethanol. For the ribosomal samples, after precipitation the pellets were dissolved in 50 µl of buffer (0.3 M NaOAc pH 7.0, 0.5% SDS, 5 mM EDTA) and rRNA was purified by phenol treatment (twice with 50 µl of phenol and once with 50 µl of chloroform) and precipitated with ethanol for 2 h at –20°C. For the 16S/23S rRNA samples this phenol treatment was carried out prior to the reaction with borohydride.

Primer extension

Full-length screening of the 23S rRNA was made using 12 primers complementary to nt 310–330, 559–568, 765–785, 1040–1059, 1320–1339, 1619–1636, 1831–1850, 2081–2100, 2281–2301, 2487–2503, 2730–2749 and 2886–2903, that of the 16S rRNA with eight primers complementary to nt 162–178, 324–340, 481–497, 684–700, 838–854, 1047–1063, 1310–1326 and 1457–1473. The primer extension reaction was carried out as described (8).

RESULTS AND DISCUSSION

In the atomic structure of H.marismortui 23S rRNA (1) there is a U-U base pair between residues U2610 and U2546 in helix 90, which is located close to the peptidyltransferase region (Fig. 1). This pair has the same structure as that reported (9) for the universally conserved U-U pair in the A site region of the rRNA from the small subunit. A search through the known secondary structures of the large subunit rRNAs of Archaea, Eubacteria and Eucaria reveals that in the Archaea and Eucaria the U residues corresponding to U2610 and U2546 in H.marismortui are both highly conserved. On the other hand, in the Eubacteria the upstream U is highly conserved but the downstream residue is a C; in E.coli these residues are U2511 and C2575, respectively. From the level of conservation and its location in the 50S subunit, it seems likely that this base pair could be important for maintenance of the tertiary and quaternary structure of the peptidyltransferase region of the 23S rRNA. However, for the C-U pair in E.coli to be isostructural with the U-U pair in H.marismortui the cytosine moiety would have to be protonated at position N3 (Fig. 2A). In fact, a protonated CH⁺-U pair with exactly this geometry was recently discovered by Blanchard and Puglisi in the structure of an oligonucleotide analog of the ‘A loop’ region of E.coli 23S rRNA (10); the pair existed at pH 5.5 but was rearranged at higher pH (pH 7.5). It should be noted that the structure of this CH⁺-U pair (Fig. 2A) is radically different from the single hydrogen bonded C-U pair found in the self-complementary dodecamer duplex (GGACUUCGGUCC)₂ (11) or that in the bifurcated hydrogen-bonded C-U pair described by Auffinger and Westhof (12).

Figure 1.

The base pair U2610-U2546 in H.marismortui and its location in the 3D structure of the 50S subunit. The view of the 50S subunit is from the interface side and the positions of the peptidyltransferase center (PTC), central protuberance (CP) and L1 and L7/L12 proteins are indicated.

Figure 2.

Structures of base pairs and reduction products. (A) Structures of the U2610-U2546 base pair in H.marismortui and of the isostructural C2575-U2511 pair in E.coli. (B) Protonated bases and their products of reduction by sodium borohydride. From top to bottom, N1-methyladenine, adenine, cytosine. (C) Structure of the G-AH⁺ base pair.

Sodium borohydride was chosen as a reagent to probe for the existence of such protonated residues by analogy with its ability to reduce N1-methylated adenine (Fig. 2B) (13). As a result of the methylation the adenine residue carries a positive charge, which is distributed between the N1 and N6 nitrogen atoms. When this substrate is reduced with sodium borohydride the six-membered ring loses its planar conformation and aromatic character. Both protonated adenine (unmethylated) and cytidine should have a similar structure to that of N-methylated adenine; in the case of cytidine the positive charge is distributed between the N3 and N4 nitrogen atoms (Fig. 2B). In the reduced form these residues should be unable to form normal Watson–Crick base pairs and, as a consequence, should be detectable by primer extension analysis.

Accordingly, we made a reverse transcriptase scan of the complete 16S and 23S rRNA molecules from E.coli, after sodium borohydride reduction of either 70S ribosomes or isolated 16S/23S rRNA (see Materials and Methods). As can be seen from Figure 3A, there is a strong reverse transcriptase stop at position 2576 after borohydride treatment of 23S rRNA in 70S ribosomes (lanes 3 and 3′), indicating that residue C2575 was indeed modified. In contrast, no corresponding signal was observed when the borohydride treatment was carried out using isolated rRNA (Fig. 3A, lane 1). Thus, the CH⁺2575-U2511 base pair would appear to be stabilized in the ribosomal structure by RNA–RNA or RNA–protein interactions and this stabilization does not occur with isolated rRNA.

Figure 3.

Primer extension analyses of 23S rRNA after sodium borohydride treatment. (A) The area around nucleotide C2575. The sequencing lanes are marked A, C, G and U, respectively. Lanes 1 and 2 are from an experiment with deproteinized 23S rRNA in solution, lane 1 with borohydride treatment, lane 2 without. Lanes 3, 3′, 4 and 4′ are from two independent experiments with 70S ribosomes, lanes 3 and 3′ with borohydride treatment, lanes 4 and 4′ without. (B) The area around nucleotide A1528. The sequencing lanes are marked as in (A). Lanes 1 and 2 are from an experiment with deproteinized 23S rRNA, with and without borohydride treatment, respectively, and lanes 3 and 4 from an experiment with 70S ribosomes, again with and without borohydride treatment.

The apparent pK_a of protonated cytidine in the nucleoside is 4.3 (14). Although a CH⁺-U base pair involving protonated cytosine has only recently been observed (1,9), its existence is not surprising, since it is well known that the involvement of cytosine in base pairing strongly increases its pK_a; for example, structures containing CH⁺ in d(C₄) (15) or in a triplex such as CH⁺GC (2,16,17) can exist at pH values close to 7.

The primer extension scan of the 23S rRNA revealed a second minor site of reaction with sodium borohydride at position A1528 in helix 59 of the E.coli 23S rRNA (Fig. 3B). As with CH⁺2575, this modification was only observed when the borohydride reduction was made with 70S ribosomes, but not with free rRNA (lanes 3 and 1, respectively). The phylogeny of A1528 is somewhat more complex than that of C2575, because in the Archaea and in some Eucaria helix 59 is absent. However, in many Eucaria there is an A-U pair between the residues that correspond to A1528 and G1543 in E.coli. In the Eubacteria the respective residues are either A and G (as in E.coli) or occasionally A and A, although in some cases helix 59 is missing here too. Taken together with the borohydride data of Figure 3B, this suggests that in E.coli there is an AH⁺1528-G1543 base pair, and indeed such a pair (Fig. 2C) has recently been reported (18), which is essentially isostructural to a normal A-U Watson–Crick pair. Since helix 59 is absent in H.marismortui it is difficult to draw any conclusions with regard to the importance of this pair for maintenance of the ribosomal quaternary structure. Furthermore, the weakness of the primer extension signal (Fig. 3B) suggests that the A1528 residue may not be fully protonated and hence only partially reduced by the borohydride treatment.

It has been shown (19) that the peptidyltransferase center of E.coli contains an adenine residue (A2451) with an unusually high pK_a value of ∼7.5, which has been proposed to be directly involved in the catalysis of peptide bond formation. In our experiments we did not find this residue to be protonated, but, since we carried out the borohydride reduction at pH 8.0, there is no inconsistency here. We suggest that the method we have described in this paper should prove useful as a direct and sensitive technique for the detection of protonated base pairs in any RNA molecule or RNP complex.