Analyzing a single data set using multiple RNA informatics programs often requires a file format conversion between each pair of programs, significantly hampering productivity. To facilitate the interoperation of these programs, we propose a syntax to exchange basic RNA molecular information. This RNAML syntax allows for the storage and the exchange of information about RNA sequence and secondary and tertiary structures. The syntax permits the description of higher level information about the data including, but not restricted to, base pairs, base triples, and pseudoknots. A class-oriented approach allows us to represent data common to a given set of RNA molecules, such as a sequence alignment and a consensus secondary structure. Documentation about experiments and computations, as well as references to journals and external databases, are included in the syntax. The chief challenge in creating such a syntax was to determine the appropriate scope of usage and to ensure extensibility as new needs will arise. The syntax complies with the eXtensible Markup Language (XML) recommendations, a widely accepted standard for syntax specifications. In addition to the various generic packages that exist to read and interpret XML formats, an XML processor was developed and put in the open-source MC-Core library for nucleic acid and protein structure computer manipulation.

Optical melting was used to determine the stabilities of 11 small RNA oligomers of defined secondary structure as a function of magnesium ion concentration. The oligomers included helices composed of Watson–Crick base pairs, GA tandem base pairs, GU tandem base pairs, and loop E motifs (both eubacterial and eukaryotic). The effect of magnesium ion concentration on stability was interpreted in terms of two simple models. The first assumes an uptake of metal ion upon duplex formation. The second assumes nonspecific electrostatic attraction of metal ions to the RNA oligomer. For all oligomers, except the eubacterial loop E, the data could best be interpreted as nonspecific binding of metal ions to the RNAs. The effect of magnesium ions on the stability of the eubacterial loop E was distinct from that seen with the other oligomers in two ways. First, the extent of stabilization by magnesium ions (as measured by either change in melting temperature or free energy) was three times greater than that observed for the other helical oligomers. Second, the presence of magnesium ions produces a doubling of the enthalpy for the melting transition. These results indicate that magnesium ion stabilizes the eubacterial loop E sequence by chelating the RNA specifically. Further, these results on a rather small system shed light on the large enthalpy changes observed upon thermal unfolding of large RNAs like group I introns. It is suggested that parts of those large enthalpy changes observed in the folding of RNAs may be assigned to variations in the hydration states and types of coordinating atoms in some specifically bound magnesium ions and to an increase in the observed cooperativity of the folding transition due to the binding of those magnesium ions coupling the two stems together. Brownian dynamic simulations, carried out to visualize the metal ion binding sites, reveal rather delocalized ionic densities in all oligomers, except for the eubacterial loop E, in which precisely located ion densities were previously calculated.

Non-Watson–Crick base pairs mediate specific interactions responsible for RNA–RNA self-assembly and RNA–protein recognition. An unambiguous and descriptive nomenclature with well-defined and nonoverlapping parameters is needed to communicate concisely structural information about RNA base pairs. The definitions should reflect underlying molecular structures and interactions and, thus, facilitate automated annotation, classification, and comparison of new RNA structures. We propose a classification based on the observation that the planar edge-to-edge, hydrogen-bonding interactions between RNA bases involve one of three distinct edges: the Watson–Crick edge, the Hoogsteen edge, and the Sugar edge (which includes the 2′-OH and which has also been referred to as the Shallow-groove edge). Bases can interact in either of two orientations with respect to the glycosidic bonds, cis or trans relative to the hydrogen bonds. This gives rise to 12 basic geometric types with at least two H bonds connecting the bases. For each geometric type, the relative orientations of the strands can be easily deduced. High-resolution examples of 11 of the 12 geometries are presently available. Bifurcated pairs, in which a single exocyclic carbonyl or amino group of one base directly contacts the edge of a second base, and water-inserted pairs, in which single functional groups on each base interact directly, are intermediate between two of the standard geometries. The nomenclature facilitates the recognition of isosteric relationships among base pairs within each geometry, and thus facilitates the recognition of recurrent three-dimensional motifs from comparison of homologous sequences. Graphical conventions are proposed for displaying non-Watson–Crick interactions on a secondary structure diagram. The utility of the classification in homology modeling of RNA tertiary motifs is illustrated.

Anticodon hairpins are structural motifs with contradictory functions. The recognition by aminoacyl synthetases implies extended interactions with the anticodon base triplet and thus, usually, an unfolding of the anticodon loop. The recognition by the ribosome and cognate interaction with a mRNA codon implies, on the other hand, the formation of a mini-helix with a canonical anticodon hairpin structure as observed by crystallography and NMR. To be able to understand the various properties of this motif, a precise description of its structural conservation is required. Here, on the basis of phylogenetic, structural, and molecular dynamics data, we discuss a conserved interaction established between the ribose of the U33 and the base at position 35, either a purine or a pyrimidine. This interaction involves the hydrogen bonding donor or acceptor potential of the hydroxyl group of U33 and has to be integrated in an extended definition of the anticodon hairpin. The extended structural signature provides also an explanation for the role played by pseudouridines at position 35.

Subdomain IIId from the hepatitis C virus (HCV) internal ribosome entry site (IRES) has been shown to be essential for cap-independent translation. We have conducted a structural study of a 27-nt fragment, identical in sequence to IIId, to explore the structural features of this subdomain. The proposed secondary structure of IIId is comprised of two 3 bp helical regions separated by an internal loop and closed at one end by a 6-nt terminal loop. NMR and molecular modeling were used interactively to formulate a validated model of the three-dimensional structure of IIId. We found that this fragment contains several noncanonical structural motifs and non-Watson–Crick base pairs, some of which are common to other RNAs. In particular, a motif characteristic of the rRNA α-sarcin/ricin loop was located in the internal loop. The terminal loop, 5′-UUGGGU, was found to fold to form a trinucleotide loop closed by a trans-wobble U˚G base pair. The sixth nucleotide was bulged out to allow stacking of this U˚G pair on the adjacent helical region. In vivo mutational analysis in the context of the full IRES confirmed the importance of each structural motif within IIId for IRES function. These findings may provide clues as to host cellular proteins that play a role in IRES-directed translation and, in particular, the mechanism through which host ribosomes are sequestered for viral function.

The antisense RNA CopA binds to the leader region of the repA mRNA (target: CopT). Previous studies on CopA–CopT pairing in vitro showed that the dominant product of antisense RNA–mRNA binding is not a full RNA duplex. We have studied here the structure of CopA–CopT complex, combining chemical and enzymatic probing and computer graphic modeling. CopI, a truncated derivative of CopA unable to bind CopT stably, was also analyzed. We show here that after initial loop–loop interaction (kissing), helix propagation resulted in an extended kissing complex that involves the formation of two intermolecular helices. By introducing mutations (base-pair inversions) into the upper stem regions of CopA and CopT, the boundaries of the two newly formed intermolecular helices were delimited. The resulting extended kissing complex represents a new type of four-way junction structure that adopts an asymmetrical X-shaped conformation formed by two helical domains, each one generated by coaxial stacking of two helices. This structure motif induces a side-by-side alignment of two long intramolecular helices that, in turn, facilitates the formation of an additional intermolecular helix that greatly stabilizes the inhibitory CopA–CopT RNA complex. This stabilizer helix cannot form in CopI–CopT complexes due to absence of the sequences involved. The functional significance of the three-dimensional models of the extended kissing complex (CopI–CopT) and the stable complex (CopA–CopT) are discussed.

The wobble GoU pairs have been implicated in several biological processes where RNA molecules play a key role. We review the geometrical and conformational properties of wobble GoU pairs on the basis of available crystal structures of RNAs at high resolution. The similarities with the wobble A+oC pairs and UoU pairs are illustrated, while the differences with the recently discovered bifurcated G[bull ]U pairs are contrasted.

The crystal structure of the RNA duplex [r(CGUGAUCG)dC]2 has been solved at a resolution of 0.97 Å. The model has been refined to R-work and R-free of 14.88% and 19.54% for 23,838 independent reflections. The base-pairing scheme forces the 5′-rC to be excluded from the helix and to be disordered. In the crystals, the sequence promotes the formation of two GoU wobble pairs that cluster around a crystallographic threefold axis in two different ways. In the first contact type, the GoU pairs are exclusively surrounded by water molecules, whereas in the other contact type, the three amino groups of the guanine residues of the symmetry-related GoU pairs trap a sulfate ion. This work provides the first example of the interaction of a GoU pair with a sulfate ion in a helical context. Despite the negative charge on the polynucleotide backbone, the guanine amino N2 is able to attract negatively charged groups that could, in the folding of complex RNA molecules, belong to a negative phosphodiester group from a neighboring strand and, in a RNA–protein complex, to a negative carboxyl group of an aspartate or glutamate side chain.

Retroviruses encapsidate their genome as a dimer of homologous RNA molecules noncovalently linked close to their 5′ ends. The dimerization initiation site (DIS) of human immunodeficiency virus type 1 (HIV-1) RNA is a hairpin structure that contains in the loop a 6-nt self-complementary sequence flanked by two 5′ and one 3′ purines. The self-complementary sequence, as well as the flanking purines, are crucial for dimerization of HIV-1 RNA, which is mediated by formation of a “kissing-loop” complex between the DIS of each monomer. Here, we used chemical modification interference, lead-induced cleavage, and three-dimensional modeling to compare dimerization of subtype A and B HIV-1 RNAs. The DIS loop sequences of these RNAs are AGGUGCACA and AAGCGCGCA, respectively. In both RNAs, ethylation of most but not all phosphate groups in the loop and methylation of the N7 position of the G residues in the self-complementary sequence inhibited dimerization. These results demonstrate that small perturbations of the loop structure are detrimental to dimerization. Conversely, methylation of the N1 position of the first and last As in the loop were neutral or enhanced dimerization, a result consistent with these residues forming a noncanonical sheared base pair. Phosphorothioate interference, lead-induced cleavage, and Brownian-dynamics simulation revealed an unexpected difference in the dimerization mechanism of these RNAs. Unlike subtype B, subtype A requires binding of a divalent cation in the loop to promote RNA dimerization. This difference should be taken into consideration in the design of antidimerization molecules aimed at inhibiting HIV-1 replication.

Group I introns constitute excellent systems for analyzing the relationship between RNA tertiary folding and catalysis. Within a hierarchical framework interpretation of RNA folding, secondary structure motifs subtend RNA three-dimensional (3D) architecture. Thus, mutations in two-dimensional motifs are expected to have effects different from those disrupting 3D contacts. Using UV spectroscopy, we have studied the influence of nucleotide substitutions, in both secondary and tertiary structure elements, on the thermal stability of the tertiary folding of the bacteriophage T4 td group I intron. Further, we present a quantitative analysis of the relationship between the splicing efficiency in vivo and the stability of the intron structure as monitored by UV melting curves. We conclude that the stability of the tertiary structure of a group I intron as measured by UV melting is generally a good indication of its ability to splice in vivo.

1. INTRODUCTION 399

2. DEFINITIONS 401

3. CIS BASEPAIRS 410

3.1 Cis Watson–Crick/Watson–Crick 410

3.2 Wobble pairings 411

3.3 Cis Watson–Crick/Hoogsteen pairings 416

3.4 Bifurcated pairings 417

3.5 Cis open and water-inserted 421

4. TRANS BASEPAIRS 423

4.1 Trans Watson–Crick/Watson–Crick 423

4.2 Trans wobble pairs 424

4.3 Trans Watson–Crick/Hoogsteen pairs 424

4.4 Trans Hoogsteen/Hoogsteen pairs 430

4.5 Trans bifurcated pairings 432

5. SHALLOW-GROOVE PAIRINGS 432

5.1 Hoogsteen/Shallow-groove pairs 433

5.2 Watson–Crick/Shallow-groove pairings 438

5.3 Shallow-groove/Shallow-groove pairings 440

6. SIDE-BY-SIDE BASES 446

7. DEFINING A LIBRARY OF ISOSTERIC PAIRINGS 446

8. CONCLUSIONS 451

9. ACKNOWLEDGEMENTS 452

10. REFERENCES 452

RNA molecules fold into a bewildering variety of complex 3D structures. Almost every new RNA structure obtained at high resolution reveals new, unanticipated structural motifs, which we are rarely able to predict at the current stage of our theoretical understanding. Even at the most basic level of specific RNA interactions – base-to-base pairing – new interactions continue to be uncovered as new structures appear. Compilations of possible non-canonical base-pairing geometries have been presented in previous reviews and monographs (Saenger, 1984; Tinoco, 1993). In these compilations, the guiding principle applied was the optimization of hydrogen-bonding. All possible pairs with two standard H-bonds were presented and these were organized according to symmetry or base type. However, many of the features of RNA base-pairing interactions that have been revealed by high-resolution crystallographic analysis could not have been anticipated and, therefore were not incorporated into these compilations. These will be described and classified in the present review. A recently presented approach for inferring basepair geometry from patterns of sequence variation (Gautheret & Gutell, 1997) relied on the 1984 compilation of basepairs (Saenger, 1984), and was extended to include all possible single H-bond combinations not subject to steric clashes. Another recent review may be consulted for a discussion of the NMR spectroscopy and thermodynamic effects of non-canonical (‘mismatched’) RNA basepairs on duplex stability (Limmer, 1997).

A significant fraction of the bases in a folded, structured RNA molecule participate in noncanonical base pairing interactions, often in the context of internal loops or multi-helix junction loops. The appearance of each new high-resolution RNA structure provides welcome data to guide efforts to understand and predict RNA 3D structure, especially when the RNA in question is a functionally conserved molecule. The recent publication of the crystal structure of the “Loop E” region of bacterial 5S ribosomal RNA is such an event [Correll CC, Freeborn B, Moore PB, Steitz TA, 1997, Cell 91:705–712]. In addition to providing more examples of already established noncanonical base pairs, such as purine–purine sheared pairings, trans-Hoogsteen UA, and GU wobble pairs, the structure provides the first high-resolution views of two new purine–purine pairings and a new GU pairing. The goal of the present analysis is to expand the capabilities of both chemical probing and phylogenetic analysis to predict with greater accuracy the structures of RNA molecules. First, in light of existing chemical probing data, we investigate what lessons could be learned regarding the interpretation of this widely used method of RNA structure probing. Then we analyze the 3D structure with reference to molecular phylogeny data (assuming conservation of function) to discover what alternative base pairings are geometrically compatible with the structure. The comparisons between previous modeling efforts and crystal structures show that the intricate involvements of ions and water molecules in the maintenance of non-Watson–Crick pairs render the process of correctly identifying the interacting sites in such pairs treacherous, except in cases of trans-Hoogsteen A/U or sheared A/G pairs for the adenine N1 site. The phylogenetic analysis identifies A/A, A/C, A/U and C/A, C/C, and C/U pairings isosteric with sheared A/G, as well as A/A and A/C pairings isosteric with both G/U and G/G bifurcated pairings. Thus, each non-Watson–Crick pair could be characterized by a phylogenetic signature of variations between isosteric-like pairings. In addition to the conservative changes, which form a dictionary of pairings isosterically compatible with those observed in the crystal structure, concerted changes involving several base pairs also occur. The latter covariations may indicate transitions between related but distinctive motifs within the loop E of 5S ribosomal RNA.

Selenocysteine biosynthesis and its cotranslational incorporation into selenoproteins are achieved by a complex molecular machinery (reviewed in Hüttenhofer & Böck, 1998). A major wheel in this mechanism is the tRNASec, which plays a pivotal role. It is first charged with serine by the conventional SerRS, the seryl-residue being further converted in situ to the selenocysteyl-residue by the selenocysteine synthase enzyme. The charged selenocysteyl-tRNASec delivers selenocysteine to the nascent polypeptide chain in response to a reprogrammed UGA codon. The classical elongation factors EF-Tu (in bacteria) or EF1-α (in eukaryotes) do not intervene at this stage. Instead, this process requires a selenocysteine-specific translation factor, called SELB in bacteria, but for which no eukaryotic homologue has been cloned yet. Interestingly, antideterminants against EF-Tu binding were found in the Escherichia coli tRNASec (Rudinger et al., 1996). Based on the several functions that this tRNA has to accomplish, it is reasonable to expect that the tRNASec secondary structure should exhibit distinctive structural features deviating from classical elongator tRNAs. In this regard, functional studies, structural probing, and sequence comparisons confirmed the earlier proposal that the bacterial tRNASec needs an 8-bp long amino acceptor stem and a 6-bp long D-stem to function, instead of the canonical 7 bp and 3/4 bp, respectively (Leinfelder et al., 1988; Baron et al., 1990, 1993; Tormay et al., 1994).

A new category of self-splicing group I introns with conserved structural organization and function is found among the eukaryotic microorganisms Didymium and Naegleria. These complex rDNA introns contain two distinct ribozymes with different functions: a regular group I splicing-ribozyme and a small internal group I-like ribozyme (GIR1), probably involved in protein expression. GIR1 was found to cleave at two internal sites in an obligate sequential order. Both sites are located 3′ of the catalytic core. GIR1-catalyzed transesterification reactions could not be detected. We have compared all available GIR1 sequences and propose a common RNA secondary structure resembling that of group I splicing-ribozymes, but with some important differences. The GIR1s lack most peripheral sequence components, as well as a P1 segment, and, at approximately 160–190 nt, they are the smallest functional group I ribozymes known from nature. All GIR1s were found to contain a novel 6-bp pseudoknot (P15) within their catalytic core region. Experimental support of the proposed structure was obtained from the Didymium GIR1 by RNA structure probing and site-directed mutagenesis. Three-dimensional modeling indicates a compactly folded ribozyme with the functionally essential P15 exposed in the cleft between the two principal domains P3–P8 and P4–P6.

Our knowledge of RNA structure and function has progressed at an astonishingly fast pace in the last five years. Since then, we have seen the X-ray structure of a full ribozyme, the awesome structure of a large domain of a group I intron, and recently the striking structure of the loop E of 5S rRNA. NMR studies have yielded structures, not only of small recurrent loops, but also of several RNA aptamers and of a fragment of the 16S rRNA complexed with an antibiotic. Both methods have also contributed enormously to our understanding of RNA–protein interactions; the richness of the tRNA synthetase interactions is now recognized, whereas the structures of complexes from the spliceosome and the ribosome point to new horizons. The 22 chapters of this new book compose a tribute to the fascinating and ubiquitous versatility of the RNA molecule.