The structure of the ITS2-proximal stem is required for pre-rRNA processing in yeast.

Accurate and efficient processing of pre-rRNA is critical to the accumulation of mature functional ribosomal subunits for maintenance of cell growth. Processing requires numerous factors which act in trans as well as RNA sequence/ structural elements which function in cis. To examine the latter, we have used directed mutagenesis and expression of mutated pre-rRNAs in yeast. Specifically, we tested requirements for formation of an ITS2-proximal stem on processing, a structure formed by an interaction between sequences corresponding to the 3' end of 5.8S rRNA and the 5' end of 25S. Pre-rRNA processing is inhibited in templates encoding mutations that prevent the formation of the ITS2-proximal stem. Compensatory, double mutations, which alter the sequence of this region but restore the structure of the stem, also restore processing, although at lower efficiency. This reduction in efficiency is reflected in decreased levels of mature 5.8S and 25S rRNA and increased levels of 35S pre-rRNA and certain processing intermediates. This phenotype is reminiscent of the biochemical depletion of U8 snoRNA in vertebrates for which the ITS2-proximal stem has been proposed as a potential site for interaction with U8 RNP. Thus, formation of the ITS2-proximal stem may be a requirement common to yeast and vertebrate pre-rRNA processing.

Location of N2,N2-dimethylguanosine-specific tRNA methyltransferase.

Most steps in the maturation of nuclear coded tRNAs occur in the nucleus in eukaryotic cells, but little is known as to the intranuclear location of this RNA maturation pathway. Indirect immunofluorescence experiments using antibody to N2,N2 dimethylguanosine-specific tRNA methyltransferase, a tRNA processing enzyme, and to Nup1p, a nuclear pore protein, show that both locate to the nuclear periphery in wild type cells. Staining of the nuclear membrane is more uniform with anti-Trm1p than the punctate staining observed with antibodies recognizing Nup1p. Biochemical fractionation experiments comparing fractionation of Trm1p with Nup1p, tRNA splicing ligase, and tRNA splicing endonuclease show that Trm1p behaves more like the known peripheral nuclear membrane proteins, Nup1p and tRNA splicing ligase, than like the integral membrane protein, tRNA splicing endonuclease. Cells overproducing Trm1p also concentrate it to the nuclear periphery. Thus, the site(s) of interaction of Trm1p are not easily saturable and are likely to be in excess to Trm1p. Trm1p is shared by mitochondria and the nucleus. Cells transformed with a gene coding Trm1p with a mutant nuclear targeting signal display cytoplasmic staining and an enzyme with increased solubility when compared to the solubility of wild type enzyme. Thus, mutations that prevent the enzyme from entering the nucleus result in an increase in its cytosolic but not mitochondrial concentration suggesting that the mitochondrial/nuclear distribution of Trm1p is not due solely to competition of mitochondrial and nuclear targeting information.

The functional analysis of nonsense suppressors derived from in vitro engineered Saccharomyces cerevisiae tRNA(Trp) genes.

Nonsense suppressors derived from Saccharomyces cerevisiae tRNA(Trp) genes have not been identified by classical genetic screens, although one can construct efficient amber (am) suppressors from them by making the appropriate anticodon mutation in vitro. Herein, a series of in vitro constructed putative suppressor genes was produced to test if pre-tRNA(Trp) processing difficulties could help to explain the lack of classical tRNA(Trp)-based suppressors. It is clear that inefficient processing of introns from precursor tRNA(Trp), or inaccurate overall processing, may explain why some of these constructs fail to promote nonsense suppression in vivo. However, deficient processing must be only one of the reasons why classical tRNA(Trp)-based suppressors have not been characterized, as suppression may still be extremely weak or absent in instances where the in vitro construct can lead to an accumulation of mature tRNA(Trp). Furthermore, suppression is also very weak in strains transformed with an intronless derivative of a putative tRNA(Trp) ochre (oc) suppressor gene, wherein intron removal cannot pose a problem.

Minimum intron requirements for tRNA splicing and nuclear transport in Xenopus oocytes.

The presence or absence of an intron defines two classes of eukaryotic nuclear tRNA genes whose transcripts differ in a requirement for splicing. Using quantitative nuclear microinjection, we have previously found that nucleocytoplasmic transport of these two classes of tRNAs involves pathways which differ in one or more limiting components. To examine substrate features which distinguish these two pathways, a series of variants of a Xenopus tRNA(Tyr) gene were constructed in which the intron size was altered. The splicing and transport properties of the resulting transcripts were examined in oocyte microinjection and in vitro processing assays. The addition of one or two nucleotides at the splice site equivalent in an intronless gene produced transcripts which could be transported without splicing. However, transport was reduced relative to the mature-sequence tRNA, suggesting the anticodon loop (interrupted in pre-tRNAs) may be recognized by the intronless tRNA transport apparatus. Transcripts with four- or six-nucleotide intervening sequences were incompletely spliced with cleavage at only the 3' splice site. Neither unspliced precursor nor partially processed intermediates were efficiently transported. The results of coinjection experiments using tRNA and pre-tRNA competitors suggest that simple retention by the splicing apparatus may not account for failure to export these RNAs. Finally, a requirement for splicing is not unique to transport of pre-tRNA(Tyr) since a pre-tRNA(3Leu) variant which was not spliced was also not exported.

Novel activity of a yeast ligase deletion polypeptide. Evidence for GTP-dependent tRNA splicing.

Yeast tRNA ligase possesses multiple activities which are required for the joining of tRNA halves during the tRNA splicing process: cyclic phosphodiesterase, kinase, adenylylate synthetase, and ligase. A deletion polypeptide of a dihydrofolate reductase-ligase fusion protein, designated DAC, was previously shown to join tRNA halves although ATP-dependent kinase activity was not measurable in the assay used. We describe here a characterization of the mechanism of joining used by DAC and the structure of the tRNA product. DAC produces a joined tRNA and a splice junction with a structure identical to that produced by DAKC, the full-length dihydrofolate reductase-ligase fusion. Furthermore, DAC can use GTP as the sole cofactor in the joining reaction, in contrast to DAKC, which can only complete splicing in the presence of ATP. Both enzymes exhibit GTP-dependent kinase activity at 100-fold greater efficiency than with ATP. These results suggest that a potential function for the center domain of tRNA ligase (missing in DAC) is to provide structural integrity and aid in substrate interactions and specificity. They also support the hypothesis that ligase may prefer to use two different cofactors during tRNA splicing.

Multiple nucleotide cofactor use by yeast ligase in tRNA splicing. Evidence for independent ATP- and GTP-binding sites.

We have examined multiple cofactor usage by yeast tRNA ligase in splicing in vitro. The ligase mechanism of action requires expenditure of two molar equivalents of nucleotide cofactor per mole of tRNA product. Recent evidence (Westaway, S.K., Belford, H.G., Apostol, B.L., Abelson, J., and Greer, C.L. (1993) J. Biol. Chem. 268, 2435-2443) demonstrated that the ligase-associated kinase activity is more efficient with GTP as cofactor than with ATP. Employing a ligase fusion construct with dihydrofolate reductase (Apostol, B.L., Westaway, S.K., Abelson, J., and Greer, C.L. (1991) J. Biol. Chem. 266, 7445-7455) for purposes of enzyme purification, we performed joining assays demonstrating that ATP and GTP are the most effective combination of cofactors. ATP was essential to the joining reaction, while UTP, CTP, or ATP replaced GTP inefficiently. Specific and functionally independent binding sites were confirmed for ATP and GTP by direct binding measurement. A third site was implicated in UTP- and CTP-ligase interactions. Comparison of binding constants with Kapp values determined for nucleotide-dependent joining suggested both that nucleotide triphosphate binding may be limiting in tRNA joining and that tRNA ligation occurs most efficiently using GTP for the kinase reaction and ATP as the adenylylate synthetase cofactor.

Pre-tRNA splicing: variation on a theme or exception to the rule?

There has been a growing recognition that there are many conserved features among apparently diverse RNA splicing systems, suggesting that they may have a common origin. However, pre-tRNA splicing is an apparent exception in nearly all respects. Features of this unique class should be considered in any comprehensive discussion of the origin(s) of splicing and its implications for the evolution of gene structure.

Exon sequence and structure requirements for tRNA splicing in Saccharomyces cerevisiae.

A survey of exon sequence and structure requirements for splicing was undertaken using labeled pre-tRNA substrates prepared by in vitro transcription of bacterial promoter-yeast tRNA(Tyr) gene fusions. Transcription templates were assembled from oligonucleotide cassettes allowing analysis of 22 derivatives affecting each of the potential secondary and certain tertiary interactions in the pre-tRNA. Effects on both excision of the intervening sequence by yeast endonuclease and joining of exons by ligase were examined. Replacements within the D- and T-stems and anticodon stems revealed that while the primary sequences of these segments were not essential for splicing, formation of base-paired structures was required. Replacements which altered the primary sequence while retaining the secondary structure of the aminoacyl stem allowed efficient excision by endonuclease but reduced joining by ligase. Potentially, the effects of changes within these stems may be indirect through effects on adjacent or overall structure. The presence of either structured or unstructured 5' leader and/or 3' trailer sequences had no effect on either splicing step. Alterations in the conserved Levitt tertiary pair (G15/C48), previously implicated in splicing of pre-tRNA(Phe), did not alter splicing of pre-tRNA(Tyr). A precursor in which the small (type I) extra arm in pre-tRNA(Tyr) was replaced with the large (type II) extra arm sequence from tRNA(Ser) was efficiently spliced. These and previous results suggest that only limited features of exon sequence or structure are recognized by the splicing enzymes.

Conformational transition required for efficient splicing of transcripts from hybrid lambda promoter yeast tRNA gene fusion.

Fusion of a prokaryotic promoter to a yeast tRNA gene provides a means for uncoupling analyses of mutations affecting splicing from requirements for transcription and other processing steps. For this purpose, a phage lambda promoter was fused to the Saccharomyces cerevisiae tRNATyr(SUP3a) coding sequence. This fusion allows the synthesis of an end-mature precursor by in vitro transcription with Escherichia coli RNA polymerase. This precursor was accurately spliced by purified yeast endonuclease and ligase fractions. However, both the initial rate and the extent of the endonuclease cleavage reaction were reduced in comparison to those for substrates produced by yeast RNA polymerase III. Efficient splicing could be restored in a magnesium- and temperature-dependent renaturation step, suggesting a conformational transition was required. Enzymatic solution structure probing of transcripts from wild-type and intron-variant templates revealed that the essential conformational transition involved a segment of the tRNA-like portion of the precursor. These results (1) suggest that the primary sequence of the precursor alone may not be sufficient to ensure formation of the active conformer during synthesis, (2) provide direct evidence that endonuclease recognizes mature tRNA-like structure in the precursor, and (3) suggest a general caution for the use of semisynthetic transcripts in RNA processing reactions. Potentially, transcription and processing of tRNATyr in yeast may provide a useful paradigm for examining active control of conformation in RNA biosynthesis.

Preferential binding of yeast tRNA ligase to pre-tRNA substrates.

Joining of tRNA halves during splicing in extracts of Saccharomyces cerevisiae requires each of the three enzymatic activities associated with the tRNA ligase polypeptide. Joining is most efficient for tRNA as opposed to oligonucleotide substrates and is sensitive to single base changes at a distance from splice sites suggesting considerable specificity. To examine the basis for this specificity, binding of ligase to labeled RNA substrates was measured by native gel electrophoresis. Ligase bound tRNA halves with an association constant 1600-fold greater than that for a nonspecific RNA. Comparison of binding of a series of tRNA processing intermediates revealed that tRNA-structure, particularly in the region around the splice sites, contributes to specific binding. Finally, the ligase was shown to form multiple, discrete complexes with tRNA substrates. The basis for recognition by ligase and its role in a tRNA processing pathway are discussed.

Deletion analysis of a multifunctional yeast tRNA ligase polypeptide. Identification of essential and dispensable functional domains.

Splicing of tRNA precursors in extracts of Saccharomyces cerevisiae requires the action of two enzymes: a site specific endonuclease and a tRNA ligase. The tRNA ligase contains three distinct enzymatic activities: a polynucleotide kinase, a cyclic phosphodiesterase, and an RNA ligase. The polypeptide also has a high affinity pre-tRNA binding site based on its ability to form stable complexes with pre-tRNA substrates. To investigate the organization of functional enzymatic and binding elements within the polypeptide a series of defined tRNA ligase gene deletions were constructed and corresponding proteins were expressed in Escherichia coli as fusions with bacterial dihydrofolate reductase (DHFR). The DHFR/ligase derivative proteins were then efficiently purified by affinity chromatography. The complete ligase fusion protein retained enzymatic and binding activities which were unaffected by the presence of the DHFR segment. Examination of tRNA ligase deletion derivatives revealed that the amino-terminal region was required for adenylylation, while the carboxyl-terminal region was sufficient for cyclic phosphodiesterase activity. Deletions within the central region affected kinase activity. Pre-tRNA binding activity was not strictly correlated with a distinct enzymatic domain. A DHFR/ligase-derived protein lacking kinase activity efficiently joined tRNA halves. We postulate that this variant utilizes a novel RNA ligation mechanism.

Intron sequence and structure requirements for tRNA splicing in Saccharomyces cerevisiae.

Predicted single-stranded structure at the 3' splice site is a conserved feature among intervening sequences (IVSs) in eukaryotic nuclear tRNA precursors. The role of 3' splice site structure in splicing was examined through hexanucleotide insertions at a central intron position in the Saccharomyces cerevisiae tRNA gene. These insertions were designed to alter the structure at the splice site without changing its sequence. Endonuclease cleavage of pre-tRNA substrates was then measured in vitro, and suppressor activity was examined in vivo. A precursor with fully double-stranded structure at the 3' splice site was not cleaved by endonuclease. The introduction of one unpaired nucleotide at the 3' splice site was sufficient to restore cleavage, although at a reduced rate. We have also observed that guanosine at the antepenultimate position provides a second consensus feature among IVSs in tRNA precursors. Point mutations at this position were found to affect splicing although there was no specific requirement for guanosine. These and previous results suggest that elements of secondary and/or tertiary structure at the 3' end of IVSs are primary determinants in pre-tRNA splice site utilization whereas specific sequence requirements are limited.

Copy number and stability of yeast 2 mu-based plasmids carrying a transcription-conditional centromere.

For certain yeast plasmids, the presence of a centromere segment (CEN) enhances mitotic stability and results in low copy number. Transcription from an inducible promoter adjacent to a CEN segment has been shown to alter centromere function. The rate of loss of a conditional CEN-ARS plasmid was examined and the results suggest that segregation control was immediately and effectively inactivated upon shift to inducing conditions. The effect of a conditional centromere on stability and copy number of hybrid CEN3-2 mu plasmids was also examined. When transcription was repressed, copy number was low. Mitotic stability varied and was correlated with the presence of an intact 2 mu recombination system. When transcription was induced, plasmid copy number increased. However, plasmids became highly unstable. These results indicate that while centromere function is affected by transcription from an adjacent promoter the centromere remains incompatible with the 2 mu maintenance system and may retain partial function.

Substrate recognition and identification of splice sites by the tRNA-splicing endonuclease and ligase from Saccharomyces cerevisiae.

We have examined the substrate requirements for efficient and accurate splicing of tRNA precursors in Saccharomyces cerevisiae. The effects of Schizosaccharomyces pombe tRNASer gene mutations on the two steps in splicing, intron excision and joining of tRNA halves, were determined independently by using partially purified splicing endonuclease and tRNA ligase from S. cerevisiae. Two mutations (G14 and A46) reduced the efficiency of excision and joining in parallel, whereas two others (U47:7 and C33) produced differential effects on these two steps; U47:7 affected primarily the excision reaction, and C33 had a greater impact on ligation. These data indicate that endonuclease and ligase recognize both common and unique features of their substrates. Another two mutations (Ai26 and A37:13) induced miscutting, although with converse effects on the two splice sites. Thus, the two cutting events appear to be independent. Finally, we suggest that splice sites may be determined largely through their position relative to sites within the tRNA-like domain of the precursors. Several of these important sites were identified, and others are proposed based on the data described here.

Assembly of a tRNA splicing complex: evidence for concerted excision and joining steps in splicing in vitro.

Splicing of tRNA precursors in Saccharomyces cerevisiae extracts proceeds in two steps; excision of the intervening sequence and ligation of the tRNA halves. The ability to resolve these two steps and the distinct physical properties of the endonuclease and ligase suggested that the splicing steps may not be concerted and that these two enzymes may act independently in vivo. A ligase competition assay was developed to examine whether the excision and ligation steps in tRNA splicing in vitro are concerted or independent. The ability of either yeast ligase or T4 ligase plus kinase to join the tRNA halves produced by endonuclease and the distinct structures of the reaction products provided the basis for the competition assay. In control reactions, joining of isolated tRNA halves formed by preincubation with endonuclease was measured. The ratio of yeast to T4 reaction products in these control assays reflected the ratio of the enzyme activities, as would be expected if each has equal access to the substrate. In splicing competition assays, endonuclease and pre-tRNA were added to ligase mixtures, and joining of the halves that were formed was measured. In these assays the products were predominantly those of the yeast ligase even when the T4 enzymes were present in excess. These results demonstrate preferential access of yeast ligase to the endonuclease products and provide evidence for the assembly of a functional tRNA splicing complex in vitro. This observation has important implications for the organization of the splicing components and of the gene expression pathway in vivo.

RNA ligase in bacteria: formation of a 2',5' linkage by an E. coli extract.

Ligase activity was detected in extracts of Escherichia coli, Clostridium tartarivorum, Rhodospirillum salexigens, Chromatium gracile, and Chlorobium limicola. Ligase was measured by joining of tRNA halves produced from yeast IVS-containing tRNA precursors by a yeast endonuclease. The structure of tRNATyr halves joined by an E. coli extract was examined. The ligated junction is resistant to nuclease P1 and RNAase T2 but sensitive to venom phosphodiesterase and alkaline hydrolysis, consistent with a 2',5' linkage. The nuclease-resistant junction dinucleotide comigrates with authentic (2',5') APA marker in thin-layer chromatography. The phosphate in the newly formed phosphodiester bond is derived from the pre-tRNA substrate. The widespread existence of a bacterial ligase raises the possibility of a novel class of RNA processing reactions.

Enzymatic mechanism of an RNA ligase from wheat germ.

We have characterized the mechanism of action of a wheat germ RNA ligase which has been partially purified on the basis of its ability to participate in in vitro splicing of yeast tRNA precursors (Gegenheimer, P., Gabius, H-J., Peebles, C.L., and Abelson, J. (1983) J. Biol. Chem. 258, 8365-8373). The preparation catalyzes the ligation of oligoribonucleotide substrates forming a 2'-phosphomonoester, 3',5'-phosphodiester linkage. The 5' terminus of an RNA substrate can have either a 5'-hydroxyl or a 5'-phosphate. The 5'-phosphate, which for a 5'-hydroxyl substrate can be introduced by a polynucleotide kinase activity in the preparation, is incorporated into the ligated junction. The 3' terminus can have either a 2',3'-cyclic phosphate or a 2'-phosphate. 2',3'-Cyclic phosphates can be converted into 2'-phosphates by a 2',3'-cyclic phosphate, 3'-phosphodiesterase activity in the preparation. The 2'-phosphate of the ligated product is derived from the phosphate at the 3' terminus of the substrate. Ligation proceeds with the adenylylation of the 5'-phosphorylated terminus to form an intermediate with a 5',5'-phosphoanhydride bond.