The eukaryotic translation initiation factor 4A (eIF4A) is a representative of the DEAD-box RNA helicase protein family. We have solved the crystallographic structure of the amino-terminal domain (residues 1–223) of yeast eIF4A. The domain is built around a core scaffold, a parallel α–β motif with five β strands, that is found in other RNA and DNA helicases, as well as in the RecA protein. The amino acid sequence motifs that are conserved within the helicase family are localized to the β strand →→→→ α helix junctions within the core. The core of the amino terminal domain of eIF4A is amplified with additional structural elements that differ from those of other helicases. The phosphate binding loop (the Walker A motif) is in an unusual closed conformation. The crystallographic structure reveals specific interactions between amino acid residues of the phosphate binding loop, the DEAD motif, and the SAT motif, whose alteration is known to impair coupling between the ATPase cycle and the RNA unwinding activity of eIF4A.

The availability of the complete sequence of the genome of Saccharomyces cerevisiae (Goffeau et al., 1996) and other organisms has provided a substantial amount of information on their chromosomal organization. We need now to understand the coordinated expression of the many genes present in a given genome and the function of the various encoded proteins. Microarrays (DNA chips) offer the opportunity to analyze nucleic acids at the genomic scale and have been used to perform global gene expression studies (Lander, 1999). However, these techniques also have the potential to give us a new perspective on posttranscriptional processes involved in the regulation of gene expression. We provide here a striking example through the investigation of yeast pre-mRNA splicing. Our analysis demonstrates that pre-mRNA splicing is quantitatively a much more important process in this species than previously thought and shows that splicing-signal conservation is correlated with transcription efficiency.

A set of free energy values is suggested for RNA H-pseudoknot loops. The parameters are adjusted to be consistent with the theory of polymer thermodynamics and known data on pseudoknots. The values can be used for estimates of pseudoknot stabilities and computer predictions of RNA structures.

The 3′ splice site of group I introns is defined, in part, by base pairs between the intron core and residues just upstream of the splice site, referred to as P9.0. We have studied the specificity imparted by P9.0 using the well-characterized L–21 ScaI ribozyme from Tetrahymena by adding residues to the 5′ end of the guanosine (G) that functions as a nucleophile in the oligonucleotide cleavage reaction: CCCUCUA5 (S) + NNG [lhard ][rharu ] CCCUCU + NNGA5. UCG, predicted to form two base pairs in P9.0, reacts with a (kcat/KM) value ∼10-fold greater than G, consistent with previous results. Altering the bases that form P9.0 in both the trinucleotide G analog and the ribozyme affects the specificity in the manner predicted for base-pairing. Strikingly, oligonucleotides incapable of forming P9.0 react ∼10-fold more slowly than G, for which the mispaired residues are simply absent. The observed specificity is consistent with a model in which the P9.0 site is sterically restricted such that an energetic penalty, not present for G, must be overcome by G analogs with 5′ extensions. Shortening S to include only one residue 3′ of the cleavage site (CCCUCUA) eliminates this penalty and uniformly enhances the reactions of matched and mismatched oligonucleotides relative to guanosine. These results suggest that the 3′ portion of S occupies the P9.0 site, sterically interfering with binding of G analogs with 5′ extensions. Similar steric effects may more generally allow structured RNAs to avoid formation of incorrect contacts, thereby helping to avoid kinetic traps during folding and enhancing cooperative formation of the correct structure.

A U5 snRNP protein, hPrp8, forms a UV-induced crosslink with the 5′ splice site (5′SS) RNA within splicing complex B assembled in trans- as well as in cis-splicing reactions. Both yeast and human Prp8 interact with the 5′SS, branch site, polypyrimidine tract, and 3′SS during splicing. To begin to define functional domains in Prp8 we have mapped the site of the 5′SS crosslink within the hPrp8 protein. Immunoprecipitation analysis limited the site of crosslink to the C-terminal 50–60-kDa segment of hPrp8. In addition, size comparison of the crosslink-containing peptides generated with different proteolytic reagents with the pattern of fragments predicted from the hPrp8 sequence allowed for mapping of the crosslink to a stretch of five amino acids in the C-terminal portion of hPrp8 (positions 1894–1898). The site of the 5′SS:hPrp8 crosslink falls within a segment spanning the previously defined polypyrimidine tract recognition domain in yPrp8, suggesting that an overlapping region of Prp8 may be involved both in the 5′SS and polypyrimidine tract recognition events. In the context of other known interactions of Prp8, these results suggest that this protein may participate in formation of the catalytic center of the spliceosome.

The signal recognition particle (SRP) is a phylogenetically conserved ribonucleoprotein required for cotranslational targeting of proteins to the membrane of the endoplasmic reticulum of the bacterial plasma membrane. Domain IV of SRP RNA consists of a short stem-loop structure with two internal loops that contain the most conserved nucleotides of the molecule. All known essential interactions of SRP occur in that moiety containing domain IV. The solution structure of a 43-nt RNA comprising the complete Escherichia coli domain IV was determined by multidimensional NMR and restrained molecular dynamics refinement. Our data confirm the previously determined rigid structure of a smaller subfragment containing the most conserved, symmetric internal loop A (Schmitz et al., Nat Struct Biol, 1999, 6:634–638), where all conserved nucleotides are involved in nucleotide-specific structural interactions. Asymmetric internal loop B provides a hinge in the RNA molecule; it is partially flexible, yet also uniquely structured. The longer strand of internal loop B extends the major groove by creating a ledge-like arrangement; for loop B however, there is no obvious structural role for the conserved nucleotides. The structure of domain IV suggests that loop A is the initial site for the RNA/protein interaction creating specificity, whereas loop B provides a secondary interaction site.

Positioning rRNA within the ribosome remains a challenging problem. Such positioning is critical to understanding ribosome function, as various rRNA regions interact to form suitable binding sites for ligands, such as tRNA and mRNA. We have used phenanthroline, a chemical nuclease, as a proximity probe, to help elucidate the regions of rRNA that are near neighbors of the stem-loop structure centering at nt 790 in the 16S rRNA of the Escherichia coli 30S ribosomal subunit. Using phenanthroline covalently attached to a DNA oligomer complementary to nt 787–795, we found that nt 582–584, 693–694, 787–790, and 795–797 were cleaved robustly and must lie within about 15 Å of the tethered site at the 5′ end of the DNA oligomer, which is adjacent to nt 795 of 16S rRNA.

“U-turns” represent an important class of structural motifs in the RNA world, wherein a uridine is involved in an abrupt change in the direction of the polynucleotide backbone. In the crystal structure of yeast tRNAPhe, the invariant uridine at position 33 (U33), adjacent to the anticodon, stabilizes the exemplar U-turn with three non-Watson–Crick interactions: hydrogen bonding of the 2′-OH to N7 of A35 and the N3-H to A36-phosphate, and stacking between C32 and A35-phosphate. The functional importance of each noncanonical interaction was determined by assaying the ribosomal binding affinities of tRNAPhe anticodon stem and loop domains (ASLs) with substitutions at U33. An unsubstituted ASL bound 30S ribosomal subunits with an affinity (Kd = 140 ± 50 nM) comparable to that of native yeast tRNAPhe (Kd = 100 ± 20 nM). However, the binding affinities of ASLs with dU-33 (no 2′-OH) and C-33 (no N3-H) were significantly reduced (2,930 ± 140 nM and 2,190 ± 300 nM, respectively). Surprisingly, the ASL with N3-methyluridine-33 (no N3-H) bound ribosomes with a high affinity (Kd = 220 ± 20 nM). In contrast, ASLs constructed with position 33 uridine analogs in nonstacking, nonnative, and constrained conformations, dihydrouridine (C2′-endo), 6-methyluridine (syn) and 2′O-methyluridine (C3′-endo) had almost undetectable binding. The inability of ASLs with 6-methyluridine-33 and 2′O-methyluridine-33 to bind ribosomes was not attributable to any thermal instability of the RNAs. These results demonstrate that proton donations by the N3-H and 2′OH groups of U33 are not absolutely required for ribosomal binding. Rather, the results suggest that the overall uridine conformation, including a dynamic (C3′-endo > C2′-endo) sugar pucker, anti conformation, and ability of uracil to stack between C32 and A35-phosphate, are the contributing factors to a functional U-turn.

The mRNA surveillance system is known to rapidly degrade aberrant mRNAs that contain premature termination codons in a process referred to as nonsense-mediated decay. A second class of aberrant mRNAs are those wherein the 3′ UTR is abnormally extended due to a mutation in the polyadenylation site. We provide several observations that these abnormally 3′-extended mRNAs are degraded by the same machinery that degrades mRNAs with premature nonsense codons. First, the decay of the 3′-extended mRNAs is dependent on the same decapping enzyme and 5′-to-3′ exonuclease. Second, the decay is also dependent on the proteins encoded by the UPF1, UPF2, and UPF3 genes, which are known to be specifically required for the rapid decay of mRNAs containing nonsense codons. Third, the ability of an extended 3′ UTR to trigger decay is prevented by stabilizing sequences within the PGK1 coding region that are known to protect mRNAs from the rapid decay induced by premature nonsense codons. These results indicate that the mRNA surveillance system plays a role in degrading abnormally extended 3′ UTRs. Based on these results, we propose a model in which the mRNA surveillance machinery degrades aberrant mRNAs due to the absence of the proper spatial arrangement of the translation-termination codon with respect to the 3′ UTR element as defined by the utilization of a polyadenylation site.

Proper deployment of Nanos protein at the posterior of the Drosophila embryo, where it directs posterior development, requires a combination of RNA localization and translational controls. These controls ensure that only the posteriorly-localized nanos mRNA is translated, whereas unlocalized nanos mRNA is translationally repressed. Here we describe cloning of the gene encoding Smaug, an RNA-binding protein that interacts with the sequences, SREs, in the nanos mRNA that mediate translational repression. Using an in vitro translation assay, we demonstrate that SRE-dependent repression occurs in extracts from early stage embryos. Immunodepletion of Smaug from the extracts eliminates repression, consistent with the notion that Smaug is involved. Smaug is a novel gene and the existence of potential mammalian Smaug homologs raises the possibility that Smaug represents a new class of conserved translational repressor.

Two studies have provided separate pieces of information that bear on the functional and evolutionary significance of introns in the budding yeast Saccharomyces cerevisiae (Holstege et al., 1998; Spingola et al., 1999). By the measure of the number of introns within its genes, budding yeast seems a disappointing eukaryote. Fewer than 250 of the more than 6,200 annotated genes are known to have introns, and fewer than 10 are known have more than one intron (Spingola et al., 1999). In contrast, metazoan genes are estimated to average nearly 10 introns, and the intronless gene is the exception rather than the rule. Although many essential yeast genes have introns, it would appear that introns are on the way out of the yeast genome, perhaps by a cDNA-directed homologous recombination mechanism proposed by Fink (1987).

We have identified a novel splicing factor, Isy1p, through two-hybrid screens for interacting proteins involved in nuclear pre-mRNA splicing. Isy1p was tagged and demonstrated to be part of the splicing machinery, associated with spliceosomes throughout the splicing reactions. At least a portion of the Isy1 protein population is associated with snRNAs; low levels of U5 and U6 snRNAs are coimmunoprecipitated specifically with Isy1p. When the ISY1 gene was knocked out, no defect in vegetative growth was observed. Using a sensitive in vivo splicing assay, however, we observed lower splicing efficiency in the isy1 null mutant compared to wild-type, indicating that Isy1p is important in the optimization of splicing.

A range of antibiotic inhibitors that act within the peptidyl transferase center of the ribosome were examined for their capacity to perturb the relative positioning of the 3′ end of P/P′-site-bound tRNA and the Escherichia coli ribosome. The 3′-terminal adenosines of deacylated tRNA and N-Ac-Phe-tRNA were derivatized at the 2 position with an azido group and the tRNAs were cross-linked to the ribosome on irradiation with ultraviolet light at 365 nm. The cross-links were localized on the rRNA within extended versions of three previously characterized 23S rRNA fragments F1′, F2′, and F4′ at nucleotides C2601/A2602, U2584/U2585 (F1′), U2506 (F2′), and A2062/C2063 (F4′). Each of these nucleotides lies within the peptidyl transferase loop region of the 23S rRNA. Cross-links were also formed with ribosomal proteins L27 (strong) and L33 (weak), as shown earlier. The antibiotics sparsomycin, chloramphenicol, the streptogramins pristinamycin IA and IIA, gougerotin, lincomycin, and spiramycin were tested for their capacity to alter the identities or yields of each of the cross-links. Although no new cross-links were detected, each of the drugs produced major changes in cross-linking yields, mainly decreases, at one or more rRNA sites but, with the exception of chloramphenicol, did not affect cross-linking to the ribosomal proteins. Moreover, the effects were closely similar for both deacylated and N-Ac-Phe-tRNAs, indicating that the drugs selectively perturb the 3′ terminus of the tRNA. The strongest decreases in the rRNA cross-links were observed with pristinamycin IIA and chloramphenicol, which correlates with their both producing complex chemical footprints on 23S rRNA within E. coli ribosomes. Furthermore, gougerotin and pristinamycin IA strongly increased the yields of fragments F2′ (U2506) and F4′ (U2062/C2063), respectively. The results obtained with an RNAse H approach correlate well with primer extension data implying that cross-linking occurs primarily to the bases. Both sets of data are also consistent with the results of earlier rRNA footprinting experiments on antibiotic–ribosome complexes. It is concluded that the antibiotics perturb the relative positioning of the 3′ end of the P/P′-site-bound tRNA and the peptidyl transferase loop region of 23S rRNA.

In the transcriptionally inert maturing oocyte and early embryo, control of gene expression is largely mediated by regulated changes in translational activity of maternal mRNAs. Some mRNAs are activated in response to poly(A) tail lengthening; in other cases activation results from de-repression of the inactive or masked mRNA. The 3′ UTR cis-acting elements that direct these changes are defined, principally in Xenopus and mouse, and the study of their trans-acting binding factors is just beginning to shed light on the mechanism and regulation of cytoplasmic polyadenylation and translational masking. In the marine invertebrate, Spisula solidissima, the timing of activation of three abundant mRNAs (encoding cyclin A and B and the small subunit of ribonucleotide reductase, RR) in fertilized oocytes correlates with their cytoplasmic polyadenylation. However, in vitro, mRNA-specific unmasking occurs in the absence of polyadenylation. In Walker et al. (in this issue) we showed that p82, a protein defined as selectively binding the 3′ UTR masking elements, is a homolog of Xenopus CPEB (cytoplasmic polyadenylation element binding protein). In functional studies reported here, the elements that support polyadenylation in clam egg lysates include multiple U-rich CPE-like motifs as well as the nuclear polyadenylation signal AAUAAA. This represents the first detailed analysis of invertebrate cis-acting cytoplasmic polyadenylation signals. Polyadenylation activity correlates with p82 binding in wild-type and CPE-mutant RR 3′ UTR RNAs. Moreover, since anti-p82 antibodies specifically neutralize polyadenylation in egg lysates, we conclude that clam p82 is a functional homolog of Xenopus CPEB, and plays a positive role in polyadenylation. Anti-p82 antibodies also result in specific translational activation of masked mRNAs in oocyte lysates, lending support to our original model of clam p82 as a translational repressor. We propose therefore that clam p82/CPEB has dual functions in masking and cytoplasmic polyadenylation.

Highly efficient cap-independent translation initiation at the 5′-proximal AUG is facilitated by the 3′ translation enhancer sequence (3′TE) located near the 3′ end of barley yellow dwarf virus (BYDV) genomic RNA. The role of the 3′TE in regulating viral translation was examined. The 3′TE is required for translation and thus replication of the genomic RNA that lacks a 5′ cap (Allen et al., 1999, Virology 253:139–144). Here we show that the 3′TE also mediates translation of uncapped viral subgenomic mRNAs (sgRNA1 and sgRNA2). A 109-nt viral sequence is sufficient for 3′TE activity in vitro, but additional viral sequence is necessary for cap-independent translation in vivo. The 5′ extremity of the sequence required in the 3′ untranslated region (UTR) for cap-independent translation in vivo coincides with the 5′ end of sgRNA2. Thus, sgRNA2 has the 3′TE in its 5′ UTR. Competition studies using physiological ratios of viral RNAs showed that, in trans, the 109-nt 3′TE alone, or in the context of 869-nt sgRNA2, inhibited translation of genomic RNA much more than it inhibited translation of sgRNA1. The divergent 5′ UTRs of genomic RNA and sgRNA1 contribute to this differential susceptibility to inhibition. We propose that sgRNA2 serves as a novel regulatory RNA to carry out the switch from early to late gene expression. Thus, this new mechanism for temporal control of translation control involves a sequence that stimulates translation in cis and acts in trans to selectively inhibit translation of viral mRNA.

In vitro transcription with T7 RNA polymerase is one of the most powerful tools in RNA research. This is mainly linked to the easy preparation of the enzyme, the quite unlimited range of sizes and sequences of the RNA that can be synthesized, as well as the efficiency and accuracy of synthesis. So far only two major limitations are common knowledge, namely poor transcription efficiency of non-G-rich initial sequences (Dunn & Studier, 1983) and 3′-end heterogeneities of the transcripts by 1 or 2 nt (Milligan et al., 1987; Draper et al., 1988; Kholod et al., 1998). These drawbacks have been overcome in most cases by improving the initiating sequence and/or purifying the transcription products to single-nucleotide resolution when necessary. The Uhlenbeck laboratory has published the first evidence for shifty errors by the T7 RNA polymerase at the very 5′ end of the synthesized RNA, an as yet unsuspected event (Pleiss et al., 1998). They demonstrated that in the particular case of tRNA sequences starting with G-rich stretches, the enzyme incorporates additional nontemplated G residues in up to 30% of the transcripts. The authors point to the fact that these errors, in combination with the 3′-end heterogeneity common to T7 transcripts, may be crucial, for instance, for functional studies of tRNAs, because some in vitro-transcribed tRNA molecules of rigorously correct final size may be erroneously considered as functional.

Adenosine deaminases that act on RNA (ADARs) convert adenosine to inosine in double-stranded regions of RNA. ADAR activity is in the nucleus in Xenopus laevis stage VI oocytes, and released into the cytoplasm at oocyte maturation. We previously demonstrated that a cytoplasmic double-stranded RNA (dsRNA) binding factor(s), cyto-dsRBP, protects microinjected dsRNA from the ADAR released at maturation. Here we describe experiments to determine whether an endogenous dsRNA, the duplex formed between sense and antisense transcripts of basic fibroblast growth factor (bFGF), is protected in a similar manner. Consistent with the presence of cyto-dsRBP, we observed that the majority of bFGF RNA was not deaminated, before or after maturation. However, a minor fraction of the bFGF RNA was deaminated whether the RNA was isolated from stage VI oocytes or matured oocytes. Since ADAR activity is in the nucleus in stage VI oocytes, our results suggest that a fraction of the bFGF RNAs are hybridized in the nucleus and are ADAR substrates. Adenosine deaminations result in A-to-G changes in cDNAs, so we quantified the fraction of modified molecules using restriction-enzyme assays of RT-PCR products. Caveats due to recombination during RT-PCR are discussed.

Translation termination in eukaryotes requires a stop codon-responsive (class-I) release factor, eRF1, and a guanine nucleotide-responsive (class-II) release factor, eRF3. Schizosaccharomyces pombe eRF3 has an N-terminal polypeptide similar in size to the prion-like domain of Saccharomyces cerevisiae eRF3 in addition to the EF-1α-like catalytic domain. By in vivo two-hybrid assay as well as by an in vitro pull-down analysis using purified proteins of S. pombe as well as of S. cerevisiae, eRF1 bound to the C-terminal one-third domain of eRF3, named eRF3C, but not to the N-terminal two-thirds, which was inconsistent with the previous report by Paushkin et al. (1997, Mol Cell Biol 17:2798–2805). The activity of S. pombe eRF3 in eRF1 binding was affected by Ala substitutions for the C-terminal residues conserved not only in eRF3s but also in elongation factors EF-Tu and EF-1α. These single mutational defects in the eRF1–eRF3 interaction became evident when either truncated protein eRF3C or C-terminally altered eRF1 proteins were used for the authentic protein, providing further support for the presence of a C-terminal interaction. Given that eRF3 is an EF-Tu/EF-1α homolog required for translation termination, the apparent dispensability of the N-terminal domain of eRF3 for binding to eRF1 is in contrast to importance, direct or indirect, in EF-Tu/EF-1α for binding to aminoacyl-tRNA, although both eRF3 and EF-Tu/EF-1α share some common amino acids for binding to eRF1 and aminoacyl-tRNA, respectively. These differences probably reflect the independence of eRF1 binding in relation to the G-domain function of eRF3 (i.e., probably uncoupled with GTP hydrolysis), whereas aminoacyl-tRNA binding depends on that of EF-Tu/EF-1α (i.e., coupled with GTP hydrolysis), which sheds some light on the mechanism of eRF3 function.

Comparative sequence analysis complements experimental methods for the determination of RNA three-dimensional structure. This approach is based on the concept that different sequences within the same gene family form similar higher-order structures. The large number of rRNA sequences with sufficient variation, along with improved covariation algorithms, are providing us with the opportunity to identify new base triples in 16S rRNA. The three-dimensional conformations for one of our strongest candidates involving U121 (C124:G237) and/or U121 (U125:A236) (Escherichia coli sequence and numbering) are analyzed here with different molecular modeling tools. Molecular modeling shows that U121 interacts with C124 in the U121 (C124:G237) base triple. This arrangement maintains isomorphic structures for the three most frequent sequence motifs (approximately 93% of known bacterial and archaeal sequences), is consistent with chemical reactivity of U121 in E. coli ribosomes, and is geometrically favorable. Further, the restricted set of observed canonical (GU, AU, GC) base-pair types at positions 124:237 and 125:236 is consistent with the fact that the canonical base-pair sets (for both base pairs) that are not observed in nature prevent the formation of the 121(124:237) base triple. The analysis described here serves as a general scheme for the prediction of specific secondary and tertiary structure base pairing where there is a network of correlated base changes.

The eukaryotic nucleolus contains a large number of small RNA molecules that, in the form of small nucleolar ribonucleoprotein complexes (snoRNPs), are involved in the processing and modification of pre-rRNA. One of the snoRNPs that has been shown to possess enzymatic activity is the RNase MRP. RNase MRP is an endoribonuclease involved in the formation of the 5′ end of 5.8S rRNA. In this study the association of the hPop1 protein with the RNase MRP complex was investigated. The hPop1 protein seems not to be directly bound to the RNA component, but requires nt 1–86 and 116–176 of the MRP RNA to associate with the RNase MRP complex via protein–protein interactions. UV crosslinking followed by ribonuclease treatment and immunoprecipitation with anti-Th/To antibodies revealed three human proteins of about 20, 25, and 40 kDa that can associate with the RNase MRP complex. The 20- and 25-kDa proteins appear to bind to stem-loop I of the MRP RNA whereas the 40-kDa protein requires the central part of the MRP RNA (nt 86–176) for association with the RNase MRP complex. In addition, we show that the human RNase P proteins Rpp30 and Rpp38 are also associated with the RNase MRP complex. Expression of Vesicular Stomatitis Virus- (VSV) tagged versions of these proteins in HeLa cells followed by anti-VSV immunoprecipitation resulted in coprecipitation of both RNase P and RNase MRP complexes. Furthermore, UV crosslinking followed by anti-Th/To and anti-Rpp38 immunoprecipitation revealed that the 40-kDa protein we detected in UV crosslinking is probably identical to Rpp38.