We previously showed that the deleterious effects from introducing abasic nucleotides in the hammerhead ribozyme core can, in some instances, be relieved by exogenous addition of the ablated base and that the relative ability of different bases to rescue catalysis can be used to probe functional aspects of the ribozyme structure [Peracchi et al., Proc Nat Acad Sci USA 93:11522]. Here we examine rescue at four additional positions, 3, 9, 12 and 13, to probe transition state interactions and to demonstrate the strengths and weaknesses of base rescue as a tool for structure–function studies. The results confirm functional roles for groups previously probed by mutagenesis, provide evidence that specific interactions observed in the ground-state X-ray structure are maintained in the transition state, and suggest formation in the transition state of other interactions that are absent in the ground state. In addition, the results suggest transition state roles for some groups that did not emerge as important in previous mutagenesis studies, presumably because base rescue has the ability to reveal interactions that are obscured by local structural redundancy in traditional mutagenesis. The base rescue results are complemented by comparing the effects of the abasic and phenyl nucleotide substitutions. The results together suggest that stacking of the bases at positions 9, 13 and 14 observed in the ground state is important for orienting other groups in the transition state. These findings add to our understanding of structure–function relationships in the hammerhead ribozyme and help delineate positions that may undergo rearrangements in the active hammerhead structure relative to the ground-state structure. Finally, the particularly efficient rescue by 2-methyladenine at position 13 relative to adenine and other bases suggests that natural base modifications may, in some instance, provide additional stability by taking advantage of hydrophobic interactions in folded RNAs.

Purine-rich exonic splicing enhancers (ESEs) have been identified in many alternatively spliced exons. Alternative splicing of several ESE-containing exons has been shown to depend on subsets of the SR protein family of pre-mRNA splicing factors. In this report, we show that purified SR protein family member SRp55 by itself binds a 30-nt ESE-containing exon, the alternatively spliced exon 5 of avian cardiac troponin T. We show that purified SRp55 binds specifically to this RNA sequence with an apparent Kd of 60 nM as assayed by gel mobility retardation experiments. Mutations in the exon 5 sequence that increase or decrease exon 5 inclusion in vivo and in vitro have correspondingly different affinities for SRp55 in our assays. The exon 5 sequence contains two purine-rich motifs, common to many ESEs, and both are required for SRp55 binding. Hill plot analysis of binding titration reactions indicates that there is a cooperative binding of at least two SRp55 proteins to the exon sequence. Chemical modification interference studies using kethoxal show that SRp55 binding to exon 5 requires the N1 and/or the N2 of almost every G residue in the exon. Dimethylsulfate modification interference studies indicate that none of the N1 positions of A residues in the exon are important for binding. We postulate that SRp55 may recognize both primary sequence and RNA secondary structural elements within pre-mRNA.

NanGIR1 is a member of a new class of group I ribozymes whose putative biological function is site-specific hydrolysis at an internal processing site (IPS). We have previously shown that NanGIR1 requires 1 M KCl for maximal activity, which is nevertheless slow (0.03 min–1). We used in vitro selection and an RNA pool with approximately nine mutations per molecule to select for faster hydrolysis at the IPS in 100 mM KCl. After eight rounds of selection, GIR1 variants were isolated that catalyzed hydrolysis at 300-fold greater rates than NanGIR1 RNA. Although not required by the selection, many of the resultant RNAs had increased thermal stability relative to the parent RNA, and had a more compact structure as evidenced by their faster migration in native gels. Although a wide spectrum of mutations was found in generation 8 clones, only two mutations, U149C and U153C, were common to greater than 95% of the molecules. These and one other mutation, G32A, are sufficient to increase activity 50-fold. All three mutations lie within or proximal to the P15 pseudoknot, a structural signature of GIR1 RNAs that was previously shown to be important for catalytic activity. Overall, our findings show that variants of the Naegleria GIR1 ribozyme with dramatically improved activity lie very close to the natural GIR1 in sequence space. Furthermore, the selection for higher activity appeared to select for increased structural stability.

Histone pre-mRNA 3′ processing is controlled by a hairpin element preceding the processing site that interacts with a hairpin-binding protein (HBP) and a downstream spacer element that serves as anchoring site for the U7 snRNP. In addition, the nucleotides following the hairpin and surrounding the processing site (ACCCA′CA) are conserved among vertebrate histone genes. Single to triple nucleotide mutations of this sequence were tested for their ability to be processed in nuclear extract from animal cells. Changing the first four nucleotides had no qualitative and little if any quantitative effects on histone RNA 3′ processing in mouse K21 cell extract, where processing of this gene is virtually independent of the HBP. A gel mobility shift assay revealing HBP interactions and a processing assay in HeLa cell extract (where the contribution of HBP to efficient processing is more important) showed that only one of these mutations, predicted to extend the hairpin by one base pair, affected the interaction with HBP. Mutations in the next three nucleotides affected both the cleavage efficiency and the choice of processing sites. Analysis of these novel sites indicated a preference for the nucleotide 5′ of the cleavage site in the order A > C > U > G. Moreover, a guanosine in the 3′ position inhibited cleavage. The preference for an A is shared with the cleavage/polyadenylation reaction, but the preference order for the other nucleotides is different [Chen F, MacDonald CC, Wilusz J, 1995, Nucleic Acids Res 23:2614–2620].

We have previously shown that a specific monoclonal antibody prepared against the U1A protein, MAb 12E12, is unique in its ability to recognize a form of U1A which is not associated with the U1snRNP. This unique form of U1A, termed snRNP-free U1A or SF-A, was found to be complexed with a novel set of non-snRNP proteins (O'Connor et al., 1997, RNA 3:1444-1455). Here we demonstrate that the largest protein in these SF-A complex(es), p105, is the polypyrimidine-tract binding protein-associated factor (PSF), an auxiliary splicing factor. We show that PSF copurifies and co-immunoprecipitates with SF-A from 293T cell nucleoplasm and that it interacts with SF-A in vitro. In addition, we show that MAb 12E12 inhibits both splicing and polyadenylation in an in vitro coupled splicing and polyadenylation reaction. This suggests that SF-A and/or the SF-A complex(es) perform an important function in both processing reactions and possibly in last exon definition.

Experimental results are presented suggesting that 23S rRNA is directly involved in the peptide bond formation usually performed on the ribosome. Although several reports have indicated that the eubacterial peptidyltransferase reaction does not necessarily require all the ribosomal proteins, the reconstitution of peptidyltransferase activity by a naked 23S rRNA without the help of any of the ribosomal proteins has not been reported previously. It is demonstrated that an E. coli 23S rRNA transcript synthesized by T7 RNA polymerase in vitro was able to promote peptide bond formation in the presence of 0.5% SDS. The reaction was inhibited by the peptidyltransferase-specific antibiotics chloramphenicol and carbomycin, and by digestion with RNases A and T1. Site-directed mutageneses at two highly conserved regions close to the peptidyltransferase center ring, G2252 to U2252 and C2507G2581 to U2507A2581, also suppressed peptide bond formation. These findings strongly suggest that 23S rRNA is the peptidyltransferase itself.

A novel three-dimensional model of tertiary interactions in the core region of the eukaryotic selenocysteine tRNA is proposed based on the analysis of available nucleotide sequences. The model features the 7/5 tRNASec secondary structure characterized by seven and five base pairs in the acceptor and T-stems, respectively, and four nucleotides in the connector region between the acceptor and D-stems. The model suggests a unique system of tertiary interactions in the area between the major groove of the D-stem and the first base pair of the extra arm that provides a rigid orientation of the extra arm and contributes to the overall stability of the molecule. The model is consistent with available experimental data on serylation, selenylation, and phosphorylation of different tRNASec mutants. The important similarity between the proposed model and the structure of the tRNASer is shown. Based on this similarity, the ability of some tRNASer mutants to be serylated, selenylated, and phosphorylated was evaluated and found to be in a good agreement with experimental data.

Self-splicing of the Tetrahymena group I intron is attenuated by an rRNA stem-loop in the 5′ exon, which competes with formation of the P1 splice site helix. The equilibrium between the P1 and P(−1) stem-loops is influenced by rRNA sequences upstream and downstream of the intron. To investigate the mechanism of this conformational switch, internal deletions and point mutations were introduced in the second rRNA stem-loop upstream of the 5′ splice site. Nuclease protection, native gel electrophoresis, and self-splicing results show that this helix is important for maintaining self-splicing activity. Co-axial base stacking of adjacent helices in the 5′ exon is proposed to enable exchange between inactive and active conformations of the pre-rRNA.

Dynamic rearrangement of RNA structure is crucial for intron recognition and formation of the catalytic core during pre-mRNA splicing. Three of the splicing factors that contain sequence motifs characteristic of the DExD/DExH-box family of RNA-dependent ATPases (Prp16, Prp22, and the human homologue of Brr2) recently have been shown to unwind RNA duplexes in vitro, providing biochemical evidence that they may direct structural rearrangements on the spliceosome. Notably, however, the unwinding activity of these proteins is sequence nonspecific, raising the question of how their functional specificity is determined. Because the highly conserved DExD/DExH-box domain in these proteins is typically flanked by one or more nonconserved domains, we have tested the hypothesis that the nonconserved regions of Prp16 determine the functional specificity of the protein. We found that the nonconserved N-terminal domain of Prp16 is (1) essential for viability, (2) required for the nuclear localization of Prp16, and (3) capable of binding to the spliceosome specifically at the step of Prp16 function. Moreover, this domain can interact with the rest of the protein to allow trans-complementation. Based on these results, we propose that the spliceosomal target of the unwinding activity of Prp16, and possibly other DExD/DExH-box splicing factors as well, is defined by factors that specifically interact with the nonconserved domains of the protein.

We have investigated the specificity of the eukaryotic enzymatic machinery that transforms adenosine at position 37 (3′ adjacent to anticodon) of several tRNAs into threonylcarbamoyladenosine (t6A37). To this end, 28 variants of yeast initiator tRNAMet and yeast tRNAVal, devoid of modified nucleotide, were produced by in vitro transcription with T7 polymerase of the corresponding synthetic tRNA genes and microinjected into the cytoplasm of Xenopus laevis oocytes. Threonylcarbamoyl incorporation was analyzed in tRNA transcripts mutated in the anticodon loop by substitution, deletion, or insertion of nucleotides, or in the overall 3D structure of the tRNA by altering critical tertiary interactions. Specifically, we tested the effects of altering ribonucleotides in the anticodon loop, changes of the loop size, perturbations of the overall tRNA 3D structure due to mutations disruptive of the tertiary base pairs, and truncated tRNAs. The results indicate that, in addition to the targeted A37, only U36 was absolutely required. However, A38 in the anticodon loop considerably facilitates the quantitative conversion of A37 into t6A37 catalyzed by the enzymes present in X. laevis. The anticodon positions 34 and 35 were absolutely “neutral” and can accept any of the four canonical nucleotides A, U, C, or G. The anticodon loop size may vary from six to eight nucleotides, and the anticodon stem may have one mismatch pair of the type A∗C or G∗U at location 30–40 without affecting the efficiency of t6A37 formation and still t6A37 is efficiently formed. Although threonylcarbamoylation of A37 occurred with tRNA having limited perturbations of 3D structure, the overall L-shaped architecture of the tRNA substrate was required for efficient enzymatic conversion of A37 to t6A37. These results favor the idea that unique enzymatic machinery located in the oocyte cytoplasm catalyzes the formation of t6A37 in all U36A37-containing tRNAs (anticodon NNU).

Microinjection of the yeast tRNAMeti into the cytoplasm of X. laevis oocytes also revealed the enzymatic activities for several other nucleotide modifications, respectively m1G9, m2G10, m22G26, m7G46, D47, m5C48/49, and m1A58.

The ability of different picornavirus internal ribosome entry site (IRES) elements to direct initiation of protein synthesis has been assayed in different cell lines in the presence and absence of viral proteases that inhibit cap-dependent protein synthesis. Reporter plasmids that express dicistronic mRNAs, containing different IRES elements, with the general structure CAT/IRES/LUC, have been assayed. In each plasmid, the CAT sequence encodes chloramphenicol acetyl transferase and the LUC sequence encodes luciferase. The poliovirus (PV) 2A protease and the foot-and-mouth disease virus (FMDV) Lb protease induce the cleavage of the translation initiation factor eIF4G and hence inhibit the activity of the cap-binding complex, eIF4F. In human osteosarcoma (HTK-143) cells, each of the various IRES elements functioned efficiently. In these cells, the co-expression of the viral proteases severely inhibited the expression of CAT, but the proteases had little effect on the activities of the various IRES elements. In contrast, in baby hamster kidney (BHK) cells, the efficiencies of the different IRES elements varied significantly, whereas, in normal rat kidney (NRK) cells, each of the IRES elements was relatively inefficient. In both BHK and NRK cells, the activities of those IRES elements that functioned inefficiently were strongly stimulated by the co-expression of the PV 2A or FMDV Lb proteases. This stimulation was independent of the loss of cap-dependent protein synthesis and was not achieved by the co-expression of the C-terminal fragment of eIF4G. The results suggest that the PV 2A and FMDV Lb proteases induce the cleavage of another cellular protein, in addition to eIF4G, which influences IRES function.

Many proteins with unusual structural properties are comprised of multiple repeating amino acid sequences and are often fractious to expression in recombinant systems. To facilitate recombinant production of such proteins for structural and engineering studies, we have produced circular messenger RNAs with infinite open reading frames. We show that a circular mRNA containing a simple green fluorescent protein (GFP) open reading frame can direct GFP expression in Escherichia coli. A circular mRNA with an infinite GFP open reading frame produces extremely long protein chains, proving that bacterial ribosomes can internally initiate and repeatedly transit a circular mRNA. Only the monomeric forms of GFP produced from circular mRNA are fluorescent. Analysis of the translation initiation region shows that multiple sequences contribute to maximal translation from circular mRNA. This technology provides a unique means of producing a very long repeating-sequence protein, and may open the way for development of proteinaceous materials with novel properties.

The RNA of satellite tobacco necrosis virus (STNV) is a monocistronic messenger that lacks both a cap and a poly(A) tail. Translation of STNV RNA in vitro is promoted by a 120-nt translational enhancer domain (TED) in the 3′-untranslated region. TED also stimulates translation of heterologous mRNAs. In this study, we show that TED stimulates translation of a cat mRNA by increasing translation efficiency to the level of capped mRNA. This stimulatory activity is not impaired by translation through TED. TED stimulates translation efficiency from different positions within the mRNA, varying from the 5′ end to 940 nt downstream of the coding region. Duplication of TED has an additive effect on translation stimulation only when located at both ends of the mRNA. On dicistronic RNAs, TED stimulates translation of both cistrons to the same extent. These data suggest that TED acts primarily by recruiting the translational machinery to the RNA.

Picornavirus RNAs are translated by an unusual mechanism of internal ribosome entry that requires a substantial segment of the viral 5′-untranslated region, generally known as the internal ribosome entry segment (IRES), and in some circumstances may require cellular trans-acting proteins, particularly polypyrimidine tract binding protein (PTB). It is shown here that for encephalomyocarditis virus (EMCV), the PTB dependence of IRES function in vitro is determined partly by the nature of the reporter cistron, and more especially by the size of an A-rich bulge in the IRES. With a wild-type EMCV IRES (which has a bulge of 6 As), translation is effectively independent of PTB provided the IRES is driving the synthesis of EMCV viral polyprotein. With an enlarged (7A) bulge and heterologous reporters, translation is highly dependent on PTB. Intermediate levels of PTB dependence are seen with a 7A bulge IRES driving viral polyprotein synthesis or a wild-type (6A) bulge IRES linked to a heterologous reporter. None of these parameters influenced the binding of PTB to the high-affinity site in the IRES. These results argue that PTB is not an essential and universal internal initiation factor, but, rather, that when it is required, its binding to the IRES helps to maintain the appropriate higher-order structure and to reverse distortions caused, for example, by an enlarged A-rich bulge.

Photocrosslinking has identified the joiner between domains 2 and 3 [J(23)] as folding near domain 5 (D5), a highly conserved helical substructure of group II introns required for both splicing reactions. D5 RNAs labeled with the photocrosslinker 4-thiouridine (4sU) reacted with highly conserved nucleotides G588 and A589 in J(23) of various intron acceptor transcripts. These conjugates retained some ribozyme function with the lower helix of D5 crosslinked to J(23), so they represent active complexes. One partner of the γ·γ′ tertiary interaction (A587·U887) is also in J(23); even though γ·γ′ is involved in step 2 of the splicing reaction, D5 has not previously been found to approach γ·γ′. Similar crosslinking patterns between D5 and J(23) were detected both before and after step 1 of the reaction, indicating that the lower helix of D5 is positioned similarly in both conformations of the active center. Our results suggest that the purine-rich J(23) strand is antiparallel to the D5 strand containing U32 and U33. Possibly, the interaction with J(23) helps position D5 correctly in the ribozyme active site; alternatively, J(23) itself might participate in the catalytic center.

The 60-kDa Ro autoantigen is normally complexed with small cytoplasmic RNAs known as Y RNAs. In Xenopus oocytes, the Ro protein is also complexed with a large class of variant 5S rRNA precursors that are folded incorrectly. Using purified baculovirus-expressed protein, we show that the 60-kDa Ro protein binds directly to both Y RNAs and misfolded 5S rRNA precursors. To understand how the protein recognizes these two distinct classes of RNAs, we investigated the features of Y RNA sequence and structure that are necessary for protein recognition. We identified a truncated Y RNA that is stably bound by the 60-kDa Ro protein. Within this 39-nt RNA is a conserved helix that is proposed to be the binding site for the Ro protein. Mutagenesis of this minimal Y RNA revealed that binding by the 60-kDa Ro protein requires specific base pairs within the conserved helix, a singly bulged nucleotide that disrupts the helix, and a three-nucleotide bulge on the opposing strand. Chemical probing experiments using diethyl pyrocarbonate demonstrated that, in the presence of the two bulges, the major groove of the conserved helix is accessible to protein side chains. These data are consistent with a model in which the Ro protein recognizes specific base pairs in the conserved helix by binding in the major groove of the RNA. Furthermore, experiments in which dimethyl sulfate was used to probe a naked and protein-bound Y RNA revealed that a structural alteration occurs in the RNA upon Ro protein binding.

Selection of pre-mRNA splice sites is a highly accurate process involving many trans-acting factors. Recently, we described a role for U6 snRNA position G52 in selection of the first intron nucleotide (+1G). Because some U2 alleles suppress U6-G52 mutations, we investigated whether the corresponding U2 snRNA region also influenced 5′ splice site selection. Our results demonstrate that U2 snRNAs mutated at position U23, but not adjacent nucleotides, specifically affect 5′ splice site cleavage. Furthermore, all U2 position U23 mutations are synthetic lethal with the thermosensitive U6-G52U allele. Interestingly, the U2-U23C substitution has an unprecedented hyperaccurate splicing phenotype in which cleavage of introns with a +1G substitution is reduced, whereas the strain grows with wild-type kinetics. U2 position U23 forms the first base pair with U6 position A59 in U2/U6 helix Ib. Restoration of the helical structure suppresses 5′ splice site cleavage defects, showing an important role for the helix Ib structure in 5′ splice site selection. U2/U6 helix Ib and helix II have recently been described as being functionally redundant. This report demonstrates a unique role for helix Ib in 5′ splice site selection that is not shared with helix II.

The translational apparatus very efficiently eliminates errors that would cause a spontaneous shift in frames. The probability of frameshifting can be increased dramatically by either cis or trans-acting factors. Programmed translational frameshift sites are cis-acting sequences that greatly increase the frequency of such errors, at least in part by causing a transient translational pause. Pausing during programmed +1 frameshifts occurs because of slow recognition of the codon following the last read in the normal frame. Frameshifting can also be elevated in strains carrying mutations in the homologous elongation factors EF-Tu in bacteria, and EF-1α in the yeast Saccharomyces cerevisiae. This phenotype implies that the factors contribute to frame maintenance. Because EF-Tu/EF-1α modulate the kinetics of decoding, it is possible that the frameshift suppressor forms of the factors transiently slow normal decoding, allowing spontaneous frameshifting to occur more efficiently, resulting in phenotypic suppression. We have used a set of frameshift reporter plasmids to test the effect of suppressor forms of EF-1α on constructs that differ widely in the efficiency with which they stimulate +1 shifting. When these results were compared to the effect of increased translational pausing, it was apparent that the mutations affecting EF-1α do not simply prolong the translational pause. Rather, they appear to generally increase the likelihood of frame errors, apparently by affecting the error correction mechanism of the ribosome.

The U1 snRNP is essential for recognition of the pre-mRNA 5′-splice site and the subsequent assembly of the spliceosome. Yeast U1 snRNP is considerably more complex than its metazoan counterpart, which suggests possible differences between yeast and metazoa in early splicing events. We have comprehensively analyzed the composition of yeast U1 snRNPs using a combination of biochemical, mass spectrometric, and genetic methods. We demonstrate the specific association of four novel U1 snRNP proteins, Snu71p, Snu65p, Nam8p, and Snu56p, that have no known metazoan homologues. A fifth protein, Npl3p, is an abundant cellular component that reproducibly co-purifies with the U1 snRNP, but its association is salt-sensitive. Therefore, we are unable to establish conclusively whether it binds specifically to the U1 snRNP. Interestingly, Nam8p and Npl3p were previously assigned functions in (pre-m)RNA-metabolism; however, so far, no association with U1 snRNP has been demonstrated or proposed. We also show that the yeast SmB protein is a U1 snRNP component. Yeast U1 snRNP therefore contains 16 different proteins, including seven snRNP core proteins, three homologues of the metazoan U1 snRNP-specific proteins, and six yeast-specific U1 snRNP proteins. We have simultaneously continued the characterization of additional mutants isolated in a synthetic lethal (MUD) screen for genes that functionally cooperate with U1 snRNA. Consistent with the biochemical results, mud10, mud15, and mud16 are alleles of SNU56, NAM8, and SNU65, respectively. mud10 and mud15 affect the in vivo splicing efficiency of noncanonical introns. Moreover, mud10p strongly affects the in vitro formation of splicing complexes, and extracts from the mud15 strain contain a U1 snRNP that migrates aberrantly on native gels. Finally, we show that Nam8p/Mud15p contributes to the stability of U1 snRNP.

The coding sequence for mammalian ornithine decarboxylase antizyme is in two different partially overlapping reading frames with no independent ribosome entry to the second ORF. Immediately before the stop codon of the first ORF, a proportion of ribosomes undergo a quadruplet translocation event to shift to the +1 reading frame of the second and main ORF. The proportion that frameshifts is dependent on the polyamine level and, because the product antizyme is a negative regulator of intracellular polyamine levels, the frameshifting acts to complete an autoregulatory circuit by sensing polyamine levels. An mRNA element just 5′ of the shift site and a 3′ pseudoknot are important for efficient frameshifting. Previous work has shown that a cassette with the mammalian shift site and associated signals directs efficient shifting in the budding yeast Saccharomyces cerevisiae at the same codon to the correct frame, but that the shift is −2 instead of +1. The product contains an extra amino acid corresponding to the shift site. The present work shows efficient frameshifting also occurs in the fission yeast, Schizosaccharomyces pombe. This frameshifting is 80% +1 and 20% −2. The response of S. pombe translation apparatus to the mammalian antizyme recoding signals is more similar to that of the mammalian system than to that of S. cerevisiae. S. pombe provides a good model system for genetic studies on the mechanism of at least this type of programmed mammalian frameshifting.