Members of the serine/arginine-rich (SR) protein family have multiple functions in the pre-mRNA splicing reaction. In addition to being required for the removal of constitutively spliced introns, SR proteins can function to regulate alternative splicing both in vitro and in vivo (Ge & Manley, 1990; Krainer et al., 1990a; Fu et al., 1992; Zahler et al., 1993a; Caceres et al., 1994; Wang & Manley, 1995). In the cell, SR proteins migrate from speckles—subnuclear domains that may function as storage sites for certain splicing factors—to sites of active transcription (Misteli et al., 1997; Misteli & Spector, 1999) and some SR proteins have been found to shuttle in and out of the nucleus (Caceres et al., 1998). The subcellular localization of SR proteins can be modulated by phosphorylation (Misteli & Spector, 1998; Misteli et al., 1998) and this undoubtedly underlies some regulated splicing events. However, once in the nucleus and localized to the nascent pre-mRNA, exactly how SR proteins engage the general splicing machinery to recognize specific splice sites is unclear and is an area of intense investigation.

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.

The inherent chemical instability of RNA under physiological conditions is primarily due to the spontaneous cleavage of phosphodiester linkages via intramolecular transesterification reactions. Although the protonation state of the nucleophilic 2′-hydroxyl group is a critical determinant of the rate of RNA cleavage, the precise geometry of the chemical groups that comprise each internucleotide linkage also has a significant impact on cleavage activity. Specifically, transesterification is expected to be proportional to the relative in-line character of the linkage. We have examined the rates of spontaneous cleavage of various RNAs for which the secondary and tertiary structures have previously been modeled using either NMR or X-ray crystallographic data. Rate constants determined for the spontaneous cleavage of different RNA linkages vary by almost 10,000-fold, most likely reflecting the contribution that secondary and tertiary structures make towards the overall chemical stability of RNA. Moreover, a correlation is observed between RNA cleavage rate and the relative in-line fitness of each internucleotide linkage. One linkage located within an ATP-binding RNA aptamer is predicted to adopt most closely the ideal conformation for in-line attack. This linkage has a rate constant for transesterification that is ∼12-fold greater than is observed for an unconstrained linkage and was found to be the most labile among a total of 136 different sites examined. The implications of this relationship for the chemical stability of RNA and for the mechanisms of nucleases and ribozymes are discussed.

Members of the double-stranded RNA- (dsRNA) specific adenosine deaminase gene family convert adenosine residues into inosines in dsRNA and are involved in A-to-I RNA editing of transcripts of glutamate receptor (GluR) subunits and serotonin receptor subtype 2C (5-HT2CR). We have isolated hADAR3, the third member of this class of human enzyme and investigated its editing site selectivity using in vitro RNA editing assay systems. As originally reported for rat ADAR3 or RED2, purified ADAR3 proteins could not edit GluR-B RNA at the “Q/R” site, the “R/G” site, and the intronic “hot spot” site. In addition, ADAR3 did not edit any of five sites discovered recently within the intracellular loop II region of 5-HT2CR RNAs, confirming its total lack of editing activity for currently known substrate RNAs. Filter-binding analyses revealed that ADAR3 is capable of binding not only to dsRNA but also to single-stranded RNA (ssRNA). Deletion mutagenesis identified a region rich in arginine residues located in the N-terminus that is responsible for binding of ADAR3 to ssRNA. The presence of this ssRNA-binding domain as well as its expression in restricted brain regions and postmitotic neurons make ADAR3 distinct from the other two ADAR gene family members, editing competent ADAR1 and ADAR2. ADAR3 inhibited in vitro the activities of RNA editing enzymes of the ADAR gene family, raising the possibility of a regulatory role in RNA editing.

The crystal structure of the monoclinic form of yeast phenylalanine tRNA has been redetermined at a resolution of 1.93 Å. The structure of yeast tRNAphe described here is more accurate than its predecessors not only because it incorporates higher resolution data, but also because it has been refined using techniques that had not been developed when its predecessors were determined more than 20 years ago. The 1.93 Å resolution version of this structure differs interestingly from its predecessors in its details. In loop regions particularly, the backbone torsion angles in the new structure are not the same as those reported earlier. Several new divalent cation binding sites have been identified, and the water structure that has emerged is also different.

The poly(C) binding proteins (PCBPs) are encoded at five dispersed loci in the mouse and human genomes. These proteins, which can be divided into two groups, hnRNPs K/J and the αCPs (αCP1-4), are linked by a common evolutionary history, a shared triple KH domain configuration, and by their poly(C) binding specificity. Given these conserved characteristics it is remarkable to find a substantial diversity in PCBP functions. The roles of these proteins in mRNA stabilization, translational activation, and translational silencing suggest a complex and diverse set of post-transcriptional control pathways. Their additional putative functions in transcriptional control and as structural components of important DNA-protein complexes further support their remarkable structural and functional versatility. Clearly the identification of additional binding targets and delineation of corresponding control mechanisms and effector pathways will establish highly informative models for further exploration.

Eukaryotic initiation factor (eIF) 4A functions as a subunit of the initiation factor complex eIF4F, which mediates the binding of mRNA to the ribosome. eIF4A possesses ATPase and RNA helicase activities and is the prototype for a large family of putative RNA helicases (the DEAD box family). It is thought that the function of eIF4A during translation initiation is to unwind the mRNA secondary structure in the 5′ UTR to facilitate ribosome binding. However, the evidence to support this hypothesis is rather indirect, and it was reported that eIF4A is also required for the translation of mRNAs possessing minimal 5′ UTR secondary structure. Were this hypothesis correct, the requirement for eIF4A should correlate with the degree of mRNA secondary structure. To test this hypothesis, the effect of a dominant-negative mutant of mammalian eIF4A on translation of mRNAs with various degrees of secondary structure was studied in vitro. Here, we show that mRNAs containing stable secondary structure in the 5′ untranslated region are more susceptible to inhibition by the eIF4A mutant. The mutant protein also strongly inhibits translation from several picornavirus internal ribosome entry sites (IRES), although to different extents. UV crosslinking of eIF4F subunits and eIF4B to the mRNA cap structure is dramatically reduced by the eIF4A mutant and RNA secondary structure. Finally, the eIF4A mutant forms a more stable complex with eIF4G, as compared to the wild-type eIF4A, thus explaining the mechanism by which substoichiometric amounts of mutant eIF4A inhibit translation.

The translation machinery recognizes codons that enter the ribosomal A site with remarkable accuracy to ensure that polypeptide synthesis proceeds with a minimum of errors. When a termination codon enters the A site of a eukaryotic ribosome, it is recognized by the release factor eRF1. It has been suggested that the recognition of translation termination signals in these organisms is not limited to a simple trinucleotide codon, but is instead recognized by an extended tetranucleotide termination signal comprised of the stop codon and the first nucleotide that follows. Interestingly, pharmacological agents such as aminoglycoside antibiotics can reduce the efficiency of translation termination by a mechanism that alters this ribosomal proofreading process. This leads to the misincorporation of an amino acid through the pairing of a near-cognate aminoacyl tRNA with the stop codon. To determine whether the sequence context surrounding a stop codon can influence aminoglycoside-mediated suppression of translation termination signals, we developed a series of readthrough constructs that contained different tetranucleotide termination signals, as well as differences in the three bases upstream and downstream of the stop codon. Our results demonstrate that the sequences surrounding a stop codon can play an important role in determining its susceptibility to suppression by aminoglycosides. Furthermore, these distal sequences were found to influence the level of suppression in remarkably distinct ways. These results suggest that the mRNA context influences the suppression of stop codons in response to subtle differences in the conformation of the ribosomal decoding site that result from aminoglycoside binding.

Sm and Sm-like (LSm) proteins form heptameric complexes that are involved in various steps of RNA metabolism. In yeast, the Lsm1–7 complex functions in mRNA degradation and is associated with several enzymes of this pathway, while the complex LSm2–8, the composition of which largely overlaps with that of LSm1–7, has a role in pre-mRNA splicing. A human gene encoding an LSm1 homolog has been identified, but its role in mRNA degradation has yet to be elucidated. We performed subcellular localization studies and found hLSm1 predominantly in the cytoplasm. However, it is not distributed evenly; rather, it is highly enriched in small, discrete foci. The endogenous hLSm4 is similarly localized, as are the overexpressed proteins hLSm1–7, but not hLSm8. The foci also contain two key factors in mRNA degradation, namely the decapping enzyme hDcp1/2 and the exonuclease hXrn1. Moreover, coexpression of wild-type and mutant LSm proteins, as well as fluorescence resonance energy transfer (FRET) studies, indicate that the mammalian proteins hLSm1–7 form a complex similar to the one found in yeast, and that complex formation is required for enrichment of the proteins in the cytoplasmic foci. Therefore, the foci contain a partially or fully assembled machinery for the degradation of mRNA.

Many viruses and certain cellular mRNAs initiate protein synthesis from a highly structured RNA sequence in the 5′ untranslated region, called the internal ribosome entry site (IRES). In hepatitis C virus (HCV), the IRES RNA functionally replaces several large initiation factor proteins by directly recruiting the 43S particle. Using quantitative binding assays, modification interference of binding, and chemical and enzymatic footprinting experiments, we show that three independently folded tertiary structural domains in the IRES RNA make intimate contacts to two purified components of the 43S particle: the 40S ribosomal subunit and eukaryotic initiation factor 3 (eIF3). We measure the affinity and demonstrate the specificity of these interactions for the first time and show that the high affinity interaction of IRES RNA with the 40S subunit drives formation of the IRES RNA[bull ]40S[bull ]eIF3 ternary complex. Thus, the HCV IRES RNA recruits 43S particles in a mode distinct from both eukaryotic cap-dependent and prokaryotic ribosome recruitment strategies, and is architecturally and functionally unique from other large folded RNAs that have been characterized to date.

We describe characterization of spliceosomes affinity purified under native conditions. These spliceosomes consist largely of C complex containing splicing intermediates. After C complex assembly on an MS2 affinity-tagged pre-mRNA substrate containing a 3′ splice site mutation, followed by RNase H digestion of earlier complexes, spliceosomes were purified by size exclusion and affinity selection. This protocol yielded 40S C complexes in sufficient quantities to visualize in negative stain by electron microscopy. Complexes purified in this way contain U2, U5, and U6 snRNAs, but very little U1 or U4 snRNA. Analysis by tandem mass spectrometry confirmed the presence of core snRNP proteins (SM and LSM), U2 and U5 snRNP-specific proteins, and the second step factors Prp16, Prp17, Slu7, and Prp22. In contrast, proteins specific to earlier splicing complexes, such as U2AF and U1 snRNP components, were not detected in C complex, but were present in similarly purified H complex. Images of these spliceosomes revealed single particles with dimensions of approximately 270 × 240 Å that assort into well-defined classes. These images represent an important first step toward attaining a comprehensive three-dimensional understanding of pre-mRNA splicing.

Vertebrate TAP and its yeast ortholog Mex67p are involved in the export of messenger RNAs from the nucleus. TAP has also been implicated in the export of simian type D viral RNAs bearing the constitutive transport element (CTE). Although TAP directly interacts with CTE-bearing RNAs, the mode of interaction of TAP/Mex67p with cellular mRNAs is different from that with the CTE RNA and is likely to be mediated by protein–protein interactions. Here we show that Mex67p directly interacts with Yra1p, an essential yeast hnRNP-like protein. This interaction is evolutionarily conserved as Yra1p also interacts with TAP. Conditional expression in yeast cells implicates Yra1p in the export of cellular mRNAs. Database searches revealed that Yra1p belongs to an evolutionarily conserved family of hnRNP-like proteins having more than one member in Mus musculus, Xenopus laevis, Caenorhabditis elegans, and Schizosaccharomyces pombe and at least one member in several species including plants. The murine members of the family directly interact with TAP. Because members of this protein family are characterized by the presence of one RNP-motif RNA-binding domain and exhibit RNA-binding activity, we called these proteins REF-bps for RNA and export factor binding proteins. Thus, Yra1p and members of the REF family of hnRNP-like proteins may facilitate the interaction of TAP/Mex67p with cellular mRNAs.

Arginine residues in RG-rich proteins are frequently dimethylated posttranslationally by protein arginine methyltransferases (PRMTs). The most common methylation pattern is asymmetrical dimethylation, a modification important for protein shuttling and signal transduction. Symmetrically dimethylated arginines (sDMA) have until now been confined to the myelin basic protein MBP and the Sm proteins D1 and D3. We show here by mass spectrometry and protein sequencing that also the human Sm protein B/B′ and, for the first time, one of the Sm-like proteins, LSm4, contain sDMA in vivo. The symmetrical dimethylation of B/B′, LSm4, D1, and D3 decisively influences their binding to the Tudor domain of the “survival of motor neurons” protein (SMN): inhibition of dimethylation by S-adenosylhomocysteine (SAH) abolished the binding of D1, D3, B/B′, and LSm4 to this domain. A synthetic peptide containing nine sDMA-glycine dipeptides, but not asymmetrically modified or nonmodified peptides, specifically inhibited the interaction of D1, D3, B/B′, LSm4, and UsnRNPs with SMN-Tudor. Recombinant D1 and a synthetic peptide could be methylated in vitro by both HeLa cytosolic S100 extract and nuclear extract; however, only the cytosolic extract produced symmetrical dimethylarginines. Thus, the Sm-modifying PRMT is cytoplasmic, and symmetrical dimethylation of B/B′, D1, and D3 is a prerequisite for the SMN-dependent cytoplasmic core-UsnRNP assembly. Our demonstration of sDMAs in LSm4 suggests additional functions of sDMAs in tri-UsnRNP biogenesis and mRNA decay. Our findings also have interesting implications for the understanding of the aetiology of spinal muscular atrophy (SMA).

From 50 genomes of the three domains of life (7 eukarya, 13 archaea, and 30 bacteria), we extracted, analyzed, and compared over 4,000 sequences corresponding to cytoplasmic, nonorganellar tRNAs. For each genome, the complete set of tRNAs required to read the 61 sense codons was identified, which permitted revelation of three major anticodon-sparing strategies. Other features and sequence peculiarities analyzed are the following: (1) fit to the standard cloverleaf structure, (2) characteristic consensus sequences for elongator and initiator tDNAs, (3) frequencies of bases at each sequence position, (4) type and frequencies of conserved 2D and 3D base pairs, (5) anticodon/tDNA usages and anticodon-sparing strategies, (6) identification of the tRNA-Ile with anticodon CAU reading AUA, (7) size of variable arm, (8) occurrence and location of introns, (9) occurrence of 3′-CCA and 5′-extra G encoded at the tDNA level, and (10) distribution of the tRNA genes in genomes and their mode of transcription. Among all tRNA isoacceptors, we found that initiator tDNA-iMet is the most conserved across the three domains, yet domain-specific signatures exist. Also, according to which tRNA feature is considered (5′-extra G encoded in tDNAs-His, AUA codon read by tRNA-Ile with anticodon CAU, presence of intron, absence of “two-out-of-three” reading mode and short V-arm in tDNA-Tyr) Archaea sequester either with Bacteria or Eukarya. No common features between Eukarya and Bacteria not shared with Archaea could be unveiled. Thus, from the tRNomic point of view, Archaea appears as an “intermediate domain” between Eukarya and Bacteria.

A major pathway of mRNA turnover in eukaryotic cells initiates with deadenylation, leading to mRNA decapping and subsequent 5′ to 3′ exonuclease digestion. We show that a highly conserved member of the DEAD box family of helicases, Dhh1p, stimulates mRNA decapping in yeast. In dhh1Δ mutants, mRNAs accumulate as deadenylated, capped species. Dhh1p's effects on decapping only occur on normal messages as nonsense-mediated decay still occurs in dhh1Δ mutants. The role of Dhh1p in decapping appears to be direct, as Dhh1p physically interacts with several proteins involved in mRNA decapping including the decapping enzyme Dcp1p, as well as Lsm1p and Pat1p/Mrt1p, which function to enhance the decapping rate. Additional observations suggest Dhh1p functions to coordinate distinct steps in mRNA function and decay. Dhh1p also associates with Pop2p, a subunit of the mRNA deadenylase. In addition, genetic phenotypes suggest that Dhh1p also has a second biological function. Interestingly, Dhh1p homologs in others species function in maternal mRNA storage. This provides a novel link between the mechanisms of decapping and maternal mRNA translational repression.

Although the primary structures of class 1 polypeptide release factors (RF1 and RF2 in prokaryotes, eRF1 in eukaryotes) are known, the molecular basis by which they function in translational termination remains obscure. Because all class 1 RFs promote a stop-codon-dependent and ribosome-dependent hydrolysis of peptidyl-tRNAs, one may anticipate that this common function relies on a common structural motif(s). We have compared amino acid sequences of the available class 1 RFs and found a novel, common, unique, and strictly conserved GGQ motif that should be in a loop (coil) conformation as deduced by programs predicting protein secondary structure. Site-directed mutagenesis of the human eRF1 as a representative of class 1 RFs shows that substitution of both glycyl residues in this motif, G183 and G184, causes complete inactivation of the protein as a release factor toward all three stop codons, whereas two adjacent amino acid residues, G181 and R182, are functionally nonessential. Inactive human eRF1 mutants compete in release assays with wild-type eRF1 and strongly inhibit their release activity. Mutations of the glycyl residues in this motif do not affect another function, the ability of eRF1 together with the ribosome to induce GTPase activity of human eRF3, a class 2 RF. We assume that the novel highly conserved GGQ motif is implicated directly or indirectly in the activity of class 1 RFs in translation termination.

DNA templates modified with C2′-methoxyls at the last two nucleotides of the 5′ termini dramatically reduced nontemplated nucleotide addition by the T7 RNA polymerase from both single- and double-stranded DNA templates. This strategy was used to generate several different transcripts. Two of the transcripts were demonstrated by nuclear magnetic resonance spectroscopy to be unaffected in their sequence. Transcripts produced from the modified templates can be purified with greater ease and should be useful in a number of applications.

7-methylguanosine (m7G) modification of tRNA occurs widely in eukaryotes and bacteria, is nearly always found at position 46, and is one of the few modifications that confers a positive charge to the base. Screening of a Saccharomyces cerevisiae genomic library of purified GST-ORF fusion proteins reveals two previously uncharacterized proteins that copurify with m7G methyltransferase activity on pre-tRNAPhe. ORF YDL201w encodes Trm8, a protein that is highly conserved in prokaryotes and eukaryotes and that contains an S-adenosylmethionine binding domain. ORF YDR165w encodes Trm82, a less highly conserved protein containing putative WD40 repeats, which are often implicated in macromolecular interactions. Neither protein has significant sequence similarity to yeast Abd1, which catalyzes m7G modification of the 5′ cap of mRNA, other than the methyltransferase motif shared by Trm8 and Abd1. Several lines of evidence indicate that both Trm8 and Trm82 proteins are required for tRNA m7G-methyltransferase activity: Extracts derived from strains lacking either gene have undetectable m7G methyltransferase activity, RNA from strains lacking either gene have much reduced m7G, and coexpression of both proteins is required to overproduce activity. Aniline cleavage mapping shows that Trm8/Trm82 proteins modify pre-tRNAPhe at G46, the site that is modified in vivo. Trm8 and Trm82 proteins form a complex, as affinity purification of Trm8 protein causes copurification of Trm82 protein in approximate equimolar yield. This functional two-protein family appears to be retained in eukaryotes, as expression of both corresponding human proteins, METTL1 and WDR4, is required for m7G-methyltransferase activity.

Messenger RNAs are exported from the nucleus as large ribonucleoprotein complexes (mRNPs). To date, proteins implicated in this process include TAP/Mex67p and RAE1/Gle2p and are distinct from the nuclear transport receptors of the β-related, Ran-binding protein family. Mex67p is essential for mRNA export in yeast. Its vertebrate homolog TAP has been implicated in the export of cellular mRNAs and of simian type D viral RNAs bearing the constitutive transport element (CTE). Here we show that TAP is predominantly localized in the nucleoplasm and at both the nucleoplasmic and cytoplasmic faces of the nuclear pore complex (NPC). TAP interacts with multiple components of the NPC including the nucleoporins CAN, Nup98, Nup153, p62, and with three major NPC subcomplexes. The nucleoporin-binding domain of TAP comprises residues 508–619. In HeLa cells, this domain is necessary and sufficient to target GFP-TAP fusions to the nuclear rim. Moreover, the isolated domain strongly competes multiple export pathways in vivo, probably by blocking binding sites on the NPC that are shared with other transport receptors. Microinjection experiments implicate this domain in the export of specific CTE-containing RNAs. Finally, we show that TAP interacts with transportin and with two proteins implicated in the export of cellular mRNAs: RAE1/hGle2 and E1B-AP5. The interaction of TAP with nucleoporins, its direct binding to the CTE RNA, and its association with two mRNP binding proteins suggest that TAP is an RNA export mediator that may bridge the interaction between specific RNP export substrates and the NPC.

During RNA interference (RNAi), long dsRNA is processed to ∼21 nt duplexes, short interfering RNAs (siRNAs), which silence genes through a mRNA degradation pathway. Small temporal RNAs (stRNAs) and micro-RNAs (miRNAs) are ∼21 nt RNAs that are processed from endogenously encoded hairpin-structured precursors, and function to silence genes via translational repression. Here we report that synthetic hairpin RNAs that mimic siRNAs and miRNA precursor molecules can target a gene for silencing, and the mechanism of silencing appears to be through mRNA degradation and not translational repression. The sequence and structural configuration of these RNAs are important, and even slight modification in structure can affect the silencing activity of the hairpins. Furthermore, these RNAs are active when expressed by DNA vectors containing polymerase III promoters, opening the possibility for new approaches in stable RNAi-based loss of function studies.