Portal de Periódicos da CAPES

Another Bridge between Kingdoms: tRNA Splicing in Archaea and Eukaryotes

1997; Cell Press; Volume: 89; Issue: 7 Linguagem: Inglês

10.1016/s0092-8674(00)80287-1

1097-4172

Marlene Belfort, Alan L. Weiner,

Genomics and Phylogenetic Studies

There are three major branches, also called kingdoms or domains, in the generally accepted Woesian tree of life (20Woese C.R. Kandler O. Wheelis M.L. Proc. Natl. Acad. Sci. USA. 1990; 87: 4576-4579Crossref PubMed Scopus (4262) Google Scholar): Bacteria, Archaea, and Eucarya. Constructing a phylogeny based on rRNA sequences, Woese proposed that the bacteria branched first from the common trunk of the archaea and eucarya (Figure 1). This branching order predicts the greatest commonality of molecular processes between archaea and eukaryotes, despite obvious similarities between the archaea and bacteria in morphology, lack of nuclei, genomic organization, and metabolic pathways. Common elements in the complex molecular machinery of processes as diverse as transcription and ubiquitin- directed proteolysis in the archaea and eukaryotes indeed support their evolutionary relatedness. Similar conclusions can be drawn from shared themes in replication and translation. The presence in both archaea and eukaryotes of fibrillarin, a protein associated with the small nucleolar RNAs that direct rRNA processing and methylation in eukaryotes, suggests that commonalities may extend to RNA processing also (see other minireviews in this issue of Cell). What then of splicing? Neither group I nor group II self-splicing introns have been reported in archaea, despite their existence in both bacteria and eukaryotes. There is also no trace to be found of spliceosome-dependent pre-mRNA introns characteristic of many unicellular eukaryotes and most, if not all, metazoans. Archaeal mRNAs are not known to contain introns of any type. However, introns have been found in stable RNAs such as the 16S and 23S rRNAs, and tRNAs. While these introns have some tantalizing similarities to the nuclear tRNA introns of eukaryotes, they also display some striking differences (8Garrett R.A. Aagaard C. Andersen M. Dalgaard J.Z. Lykke-Andersen J. Phan H.T.N. Trevisanato S. Ostergaard L. Larsen N. Leffers H. System. Appl. Microbiol. 1994; 16: 680-691Crossref Scopus (16) Google Scholar). Many eukaryotes contain introns in their tRNAs, typified by the yeast tRNA introns, which are the best characterized. In budding yeast about one-fifth of the tRNAs contain introns; these introns vary in length from 14 to 60 bases, and always reside in the anticodon loop between the first two bases 3′ to the anticodon. Splicing of the intron from the pre-tRNA requires three enzymes: a site-specific endonuclease, an RNA ligase, and a phosphotransferase (Figure 2A). The endonuclease excises the intron at the 5′ and 3′ splice sites, leaving tRNA half-molecules with 5′-OH and 2′,3′ cyclic phosphate termini. The tRNA ligase then covalently joins the half-molecules in a multistep process requiring ATP and GTP, leaving a 2′-phosphate at the splice junction. In a remarkable final step, the phosphotransferase removes the 2′ phosphate by transfer to NAD, generating a mature tRNA with a standard 3′–5′ linkage at the ligation junction (14Phizicky E.M. Greer C.L. Trends Biochem.Sci. 1993; 18: 31-34Google Scholar). Archaeal tRNA introns are surprisingly similar to those of eukaryotes. The introns are generally small, ranging from 14 to 106 nt, and most reside in the very same place as tRNA introns in eukaryotes, just 3′ to the anticodon, fueling speculations that these introns have a common, ancient origin. However, some archaeal tRNA introns are found in the extra arm and anticodon stem, suggesting that archaeal introns may also have arisen de novo. Consistent with their recent acquisition is the occurrence of similar introns in archaeal rRNA, some of which contain lengthy open reading frames encoding proteins that can function as homing endonucleases (8Garrett R.A. Aagaard C. Andersen M. Dalgaard J.Z. Lykke-Andersen J. Phan H.T.N. Trevisanato S. Ostergaard L. Larsen N. Leffers H. System. Appl. Microbiol. 1994; 16: 680-691Crossref Scopus (16) Google Scholar). The archaeal tRNA and rRNA intron sequences are variable, except for conservation of a few nucleotides that form a bulge-helix-bulge motif at the exon–intron boundaries (17Thompson L.D. Daniels C.J. J. Biol. Chem. 1990; 265: 18104-18111Abstract Full Text PDF PubMed Google Scholar) (Figure 1). These distinctive characteristics notwithstanding, archaeal tRNA splicing occurs by a process that at least superficially resembles that of the eukaryotic tRNA introns. An endonuclease cleaves the splice sites to generate 5′-OH and 2′,3′ cyclic phosphate termini, and the exons are subsequently ligated to yield 3′–5′ phosphodiester bonds. Despite similarities between eukaryotic and archaeal tRNA splicing in chemistry (both proceeding through 5′-OH and 2′,3′ cyclic phosphate intermediates) and in position (many introns residing just 3′ to the anticodon), substrate recognition by the tRNA endonucleases appears to be different. The yeast and frog enzymes use a ruler mechanism to measure from some position in the mature domain of the tRNA to the cleavage sites at the exon–intron boundaries (15Reyes V.M. Abelson J. Cell. 1988; 55: 719-730Abstract Full Text PDF PubMed Scopus (97) Google Scholar). Additionally, a conserved purine residue in the intron three nucleotides from the 3′ splice site (Figure 1, R in box) must pair with a pyrimidine in the anticodon loop 6 nucleotides upstream of the 5′ splice site (Figure 1, Y in box) to form the A–I (for anticodon–intron) interaction (1Baldi M.I. Mattoccia E. Bufardeci E. Fabbri S. Tocchini-Valentini G.P. Science. 1992; 255: 1404-1408Crossref PubMed Scopus (54) Google Scholar). In contrast, the archaeal endonucleases seem not to use a ruler mechanism, but rather to recognize a specific structure, the bulge-helix-bulge, at the exon–intron boundaries. Consistent with recognition of a structural element, introns at different positions within archaeal tRNAs, and even in the rRNAs, could be accurately excised by the archaeal endonucleases. Intriguingly, the presumed secondary structures of archaeal and eukaryotic tRNA precursors are somewhat similar, especially near the 3′ splice site, where the A–I interaction in eukaryotes creates a 3-nucleotide bulge that is identical to that in the bulge-helix-bulge motif recognized by the archaeal enzyme (Figure 1). Yet, the yeast and archaeal enzymes are unable to process each other's tRNA precursors (13Palmer J.R. Nieuwlandt D.T. Daniels C.J. J. Bacteriol. 1994; 176: 3820-3823Crossref PubMed Scopus (18) Google Scholar), suggesting distinct substrate recognition mechanisms, or even separate processing systems with independent evolutionary origins. However, these differences between the eukaryotic and archaeal endonucleases now seem more apparent than real; recently it has been shown, using the frog tRNA endonuclease, that the ruler mechanism applies to the 5′ splice site, but cleavage of the 3′ splice site (where the A–I pair is important) can be dictated by local structure (7Di Nicola Negri N.E. Fabbri S. Bufardeci E. Baldi M.I. Gandini Attardi D. Mattoccia E. Tocchini-Valentini G.P. Cell. 1997; 89: 859-866Abstract Full Text Full Text PDF PubMed Google Scholar). Clearly lacking from these earlier studies was an appreciation of the processing machinery. Some of these gaps in knowledge have now been bridged by characterization of the tRNA splicing endonucleases from the yeast Saccharomyces cerevisiae and the archaeon Haloferax volcanii (18Trotta C.R. Miao F. Arn E.A. Ho C.K. Rauhut R. Abelson J.N. Cell. 1997; 89: 849-858Abstract Full Text Full Text PDF PubMed Scopus (163) Google Scholar, 11Kleman-Leyer K. Armbruster D.A. Daniels C.J. Cell. 1997; 89: 839-847Abstract Full Text Full Text PDF PubMed Scopus (88) Google Scholar). Subunit composition in the two systems differs, the archaeal enzyme being a homodimer of 37 kDa subunits, whereas the yeast enzyme is a heterotetramer consisting of 54, 44, 34, and 15 kDa subunits (Figure 2B). The surprise, after years of painstaking biochemistry on the scarce and membrane-bound yeast enzyme, is that two of the subunits, the 44 kDa Sen2 and 34 kDa Sen34 subunits, are homologous to each other and to the homodimeric H. volcanii enzyme. The homology, which spans some 115 amino acids, is also shared by potential endonucleases inferred from the genome sequences of the archaea Methanococcus jannaschii and Methanobacterium thermoautotrophicum, and a gene of unknown function from Zea mays. The homologous region is presumed to be the cleavage domain, mainly because the sen2-3 mutation, which selectively blocks 5′ cutting, is caused by a nonconservative mutation (G292D) within this region. When these facts are put together—the related endonuclease subunits and the distinctive 5′-OH and 2′,3′ cyclic phosphate processing intermediates—they smack of a common origin for the archaeal and eukaryotic tRNA splicing pathways. The archaeal endonuclease is a homodimer, and the 5′ and 3′ splice sites of archaeal tRNA are positioned within a 2-fold symmetric bulge-helix-bulge structure (Figure 1). This obvious correspondence points to a one-subunit–one-splice-site hypothesis with two active sites (Figure 2B). By analogy, one subunit of the heterotetrameric yeast enzyme is hypothesized to cleave the 5′ splice site and a second to cleave the 3′ splice site (Figure 2B). But cleavage of symmetrical splice sites by a homodimeric endonuclease in archaeal tRNA precursors does not guarantee the one-subunit–one-splice-site scenario. Several homodimers including reverse transcriptase, some tRNA synthetases, and thymidylate synthase have functionally or structurally nonequivalent subunits of identical sequence. However, experiments on yeast tRNA intron excision provide additional support. First, the one-subunit–one-splice-site model is favored by the random cleavage order of yeast tRNA precursors, since a one-active-site model relying on conformational changes between cleavage reactions would predict an obligatory order of reaction. Second, the G292D mutation in the Sen2–3 mutant inactivates 5′ splice-site cleavage, suggesting that Sen2 acts at the 5′ splice site. Third, mutation of a conserved histidine of the Sen34 subunit to alanine (H242A) results in accumulation of the intron-3′ exon two-thirds molecule, suggesting that Sen34 acts at the 3′ splice site (18Trotta C.R. Miao F. Arn E.A. Ho C.K. Rauhut R. Abelson J.N. Cell. 1997; 89: 849-858Abstract Full Text Full Text PDF PubMed Scopus (163) Google Scholar). These results provide strong support for the one-subunit–one-splice-site model. The divergence of a more ancient homodimeric endonuclease into two different subunits might then explain the different 5′ and 3′ splice-site cleavage structures in yeast. The sequences conserved among the phylogenetically diverse splicing endonucleases are likely to form the catalytic center of the different subunits. The divergent sequences of each subunit might then be associated with the catalytic cartridge to effect recognition of different splice sites or to perform such accessory functions as subunit–subunit association or membrane binding. Indeed in the yeast system there is evidence for all three functions, namely, distinct splice-site specificity, differential subunit interactions, and association with the inner nuclear membrane (possibly mediated by a domain just amino-terminal to the catalytic cartridge in Sen2). Although it seems likely that the catalytic cartridge of the endonuclease participates directly in the chemistry of cleavage, it is within the realm of possibility that the enzyme functions instead as a "maturase" to increase the efficiency or accuracy of a self-cleavage reaction. The human and Arabidopsis tRNATyr have indeed been observed to undergo self-cleavage at the authentic 3′ splice-site bulge, as well as one nucleotide downstream of the authentic 5′ splice-site, generating 2′,3′-cyclic phosphate and 5′-OH termini (19Weber U. Beier H. Gross H.J. Nucleic Acids Res. 1996; 24: 2212-2219Crossref PubMed Scopus (15) Google Scholar). The distinction between a maturase and an enzyme, however, may be artificial. A catalyst by definition accelerates a spontaneous reaction, whether it does so by performing the chemistry directly, by straining the reactants so they can perform the electronic maneuvers themselves, or by simply arranging the reactants in space. Detailed biochemistry will be required to determine whether the tRNA endonuclease participates directly in the chemistry of phosphoester bond cleavage, or helps the tRNA to process itself. Interestingly, the concept of the catalytic cartridge also applies to DNA endonucleases of several different families. For example, in the GIY-YIG family of intron endonucleases, the conserved GIY-YIG catalytic domain is tethered to distinct DNA-binding domains. In the H-N-H family of endonucleases, the catalytic H-N-H domain is again joined to nonconserved sequences, so that the basic chemistry of phosphodiester-bond cleavage is tailored to specific situations (2Belfort M. Roberts R.J. Nucleic Acids Res., in press. 1997; Google Scholar). Catalytic cartridges as independent protein folding domains that perform a basic chemical reaction, but are associated with other domains that impart specificity and novel molecular associations, provide a versatile strategy for diversification of molecular machines in many different biological processes. The related endonucleases and similar tRNA splicing pathways in archaea and eukaryotes bespeak the antiquity of the cleavage reaction. Although a homodimeric endonuclease is likely to predate the divergence of archaea and eukaryotes (there is no evidence for genetic exchange between these two kingdoms), the new results do not necessarily imply that tRNA introns are ancient. Neither do the data necessarily suggest that the primordial function of the endonuclease was to remove introns from tRNA. Given the difference in splice site recognition, and the disparate nature and locations of the introns themselves (in archaea they are found in rRNA as well as in different locations within tRNA), one can easily envision that a preexisting endonuclease was recruited independently for different processing reactions. The function of such an endonuclease in the last common ancestor might even have been processing of rRNA precursors as found in modern archaea. These new findings therefore tell us little of the origin of the tRNA introns. Several scenarios can be imagined to account for the origin and location of the protein-dependent tRNA introns extant in archaea and eukaryotes. The two major pathways involve de novo acquisition of foreign sequences or degeneration of preexisting self-splicing introns (3Cavalier-Smith T. Trends Genet. 1991; 7: 145-148Abstract Full Text PDF PubMed Scopus (323) Google Scholar, 16Shub D.A. Curr. Opin. Genet. Devel. 1991; 1: 478-484Crossref PubMed Scopus (8) Google Scholar). The prior existence of an endonuclease with specificity for a bulge-helix-bulge structure would create a permissive environment where such introns could arise spontaneously as expansion loops in tRNA or rRNA, or preexisting self-splicing tRNA introns (group I in bacteria, groups I and II in chloroplasts; Figure 1) could degenerate into a protein-dependent form. Both hypotheses are consistent with the lack of sequence conservation among the different tRNA introns, and with the diverse locations of such introns in archaeal tRNA and rRNA. Spontaneous generation of tRNA introns and degeneration of self-splicing introns are not mutually exclusive pathways. As in the great debate about the antiquity of mRNA introns, there is likely to be a middle ground. Some mRNA introns may be ancient, others afterthoughts. Some tRNA introns may be degenerate autocatalytic introns while others may be spontaneous expansion loops. Regardless, one is moved to ask, what is it about the tiny tRNA molecules that makes them attractive homes for introns? The sheer abundance of the stable RNAs would favor insertion of foreign sequences by providing many targets for illegitimate RNA integration events. Subsequent retroposition would immortalize these RNA recombination events in DNA genomes. Additionally, the strong promoter activity of genes encoding abundant RNAs would open the DNA for integration events of various kinds. For this very reason, it is believed that tRNA genes are favored integration sites of eubacterial pathogenicity islands, and bacterial and eukaryotic retroelements (5Curcio M.J. Morse R.H. Trends Genet. 1996; 2: 436-438Abstract Full Text PDF Scopus (29) Google Scholar, 9Hacker J. Blum-Oehler G. Muhldorfer I. Tschape H. Mol. Microbiol. 1997; 23: 1089-1097Crossref PubMed Scopus (854) Google Scholar). However, the facilitated entry of introns into tRNAs is one thing; persistence of introns in tRNAs is another. The maintenance of tRNA introns might be favored if the intrinsic structure of tRNA could keep exons aligned, provide a scaffold for endonuclease recognition, and juxtapose the splice sites to allow ready healing of the ends by preexisting RNA ligases. The structure of the tRNA may also facilitate self-excision (19Weber U. Beier H. Gross H.J. Nucleic Acids Res. 1996; 24: 2212-2219Crossref PubMed Scopus (15) Google Scholar). The favored location of tRNA introns just 3′ to the anticodon could then be explained by selective retention at a site that readily maintains tRNA structure. We should, however, be careful not to draw too many conclusions from intron habitats and sites of residence; there is still no rationale for the distribution of group I introns in eukaryotic nuclei (rRNA only), in organelles (mRNA, tRNA, and rRNA), and in bacteria (mRNA and tRNA, but not rRNA). The intron, once ensconced in the tRNA, may ultimately be used to advantage in its new environment. Pseudouridylation in the anticodon of certain tRNAs, for example, is dependent on the presence of an intron in the anticodon loop (10Johnson P.F. Abelson J. Nature. 1983; 302: 681-687Crossref PubMed Scopus (100) Google Scholar). Thus, introns can restructure tRNA to allow novel, beneficial nucleotide modifications. As a result, the intron becomes subject to positive selection and fixed at a specific site within the tRNA, rather than remaining a neutral passenger or suffering the fate of expulsion. One last not necessarily far-fetched possibility is that introns might, by altering the structure of the immature tRNA, prevent the cell from mistaking an immature tRNA for an uncharged tRNA. This could be important because uncharged tRNAs are key physiological regulators: uncharged tRNAs regulate transcription antitermination of genes and operons in Gram-positive bacteria (4Condon C. Putzer H. Grunberg-Manago M. Proc. Natl. Acad. Sci. USA. 1996; 93: 6992-6997Crossref PubMed Scopus (61) Google Scholar), and uncharged histidine tRNA regulates GCN2 kinase in yeast (21Zhu S. Sobolev A.Y. Wek R.C. J. Biol. Chem. 1996; 271: 24989-24994Crossref PubMed Scopus (91) Google Scholar). Such scenarios might also explain why intron removal is the last step in tRNA maturation, or even why the processing enzymes are bound to the nuclear membrane between nucleus and cytoplasm. is the most dangerous question in molecular biology. is much safer. Still, it is difficult to resist asking why so many of the molecular themes that are shared between archaea and eukaryotes appear to involve the Central Dogma—the flow of information from DNA to RNA to protein. Studies of molecular function and the results of genome sequencing projects clearly indicate that the DNA replication and transcription in archaea and eukaryotes have more in common than either kingdom does with the bacteria (Olsen and Woese, 1997 [this issue of Cell]). Now the same seems to be true of the endonuclease components of the tRNA splicing machinery, which are closely allied in archaea and eukaryotes. Translation is more of a patchwork across all three kingdoms (Dennis, 1997 [this issue of Cell]), and the tRNA ligation machinery may well turn out to be another phylogenetic mosaic (18Trotta C.R. Miao F. Arn E.A. Ho C.K. Rauhut R. Abelson J.N. Cell. 1997; 89: 849-858Abstract Full Text Full Text PDF PubMed Scopus (163) Google Scholar). On the other hand, archaea appear more closely related to bacteria than to eukaryotes from a metabolic standpoint (12Olsen G.J. Woese C.R. Cell, this issue. 1997; 89: 991-994Scopus (191) Google Scholar). What is going on here? Intermediary metabolism is a delicately balanced network of reactions and one may have to inherit much of the network or suffer the consequences as the delicate balance comes undone. Components of the Central Dogma may also be a package deal. While there are clearly some interchangeable parts, many steps are physically coupled to others, driving selection for integrated pathways. As for the patchwork, perhaps tRNA processing and translation have more interchangeable parts than we imagined.