Avatar pre-tRNAs help elucidate the properties of tRNA-splicing endonucleases that produce tRNA from permuted genes

Abstract

Unusual tRNA genes, found in some algae, have their mature terminal 3′ portion in front of their 5′ portion in the genome. The transcripts from such genes must be cleaved by a pre-tRNA endonuclease to form a functional tRNA. We present a mechanism for the generation of “corrected” tRNAs from such a “permuted” pre-tRNA configuration. We used two avatar (av) or model pre-tRNAs and two splicing endonucleases with distinct mechanisms of recognition of the pre-tRNA. The splicing results are compatible with an evolutionary route in which permuted genes result from a duplication event followed by DNA rearrangement. The model pre-tRNAs permit description of the features that a transcript, derived from a rearranged duplicated gene, must have to give rise to functional tRNA. The two tRNA endonucleases are a eukaryal enzyme that normally acts in a mature domain-dependent mode and an archaeal enzyme that acts in a mature domain-independent mode. Both av pre-tRNAs are able to fold into two conformations: 1 and 2. We find that only conformation 2 can yield a corrected functional tRNA. This result is consistent with contemporary algae representing snapshots of different evolutionary stages, with duplicated genes preceding recombinatorial events generating a permutated gene. In a scenario elucidated by the use of the av pre-tRNAs, algal permuted tRNA genes could have further lost one of two mature domains, eliminating steric problems for the algal tRNA endonuclease, which remains a typical eukaryal enzyme capable of correcting the permuted transcript to a functional tRNA.

Transfer RNA is one of the most conserved RNA molecules in structure and function, with a typical secondary structure, the cloverleaf, and a characteristic L-shaped tertiary structure. All mature tRNAs are also recognized and aminoacylated specifically by their cognate aminoacyl-tRNA synthetases, fitting the tRNA binding sites of the ribosome to transfer amino acids to a growing polypeptide chain. Nevertheless, different architectures of tRNA genes have been described (1).

Two types of tRNA genes have been identified to date: intron-lacking and intron-containing. The latter group includes two recently discovered subgroups: split and permuted tRNAs (2–5). Split tRNA genes are encoded at two or three separate regions of the genome, and the corresponding transcripts are processed into single RNA molecules. Split tRNAs are characterized by a structural motif called bulge–helix–bulge (BHB) or the more relaxed bulge–helix–loop (BHL).

Permuted tRNA genes were discovered in the ultrasmall unicellular red alga Cyanidioschyzon merolae, which inhabits an extreme environment. A search for tRNA genes using the tRNAscan-SE program (6) predicted tRNA genes encoding only 30 species of anticodon, a number that is insufficient to decode all 61 codons. Computational analysis of the nuclear genome of the red alga identified 11 tRNA genes in which the 3′ half of the tRNA lies upstream of the 5′ half in the genome. Such a gene arrangement is accomplished by circular gene permutation. These permuted genes are expressed and produce mature tRNAs that are aminoacylated. Features characteristic of such novel gene organization are a TATA-like sequence found upstream of the 3′ half of the genes, which instead of the intragenic bipartite promoter consisting of an A-box and a B-box, presumably helps to effect the initiation of transcription. Downstream of the 5′ half, a run of Ts is likely to provide a termination signal for RNA polymerase III (5). It appears, therefore, that the transcript is a linear RNA molecule that we refer to as the actual permuted pre-tRNA (ap pre-tRNA).

An intervening sequence separates the 3′ end of the 5′ half from the 5′ end of the 3′ half. The sequences adjacent to those junctions form the BHB motif. Maturation of the ap pre-tRNA starts with processing of a BHB motif by the tRNA-splicing endonuclease and ligase, resulting in formation of a circular RNA intermediate. The intervening sequence is then removed by RnaseP and tRNAZ (7, 8), followed by CCA addition to create the aminoacylatable terminus.

How could permuted tRNA genes have arisen?

Results and Discussion

The evolution of the permuted tRNA family, and the reason why it is present in algae, is unknown. A recent study of the distribution of permuted tRNA genes in the available genome sequences of green algae revealed tRNA genes duplicated in tandem (9). Duplicated genes and, in general, multiple copies are one of the major features of genome evolution. A classic model predicts that one member of a duplicated pair is lost through the accumulation of degenerative mutations or occasionally preserved because of beneficial mutations that confer a novel function or an improvement of an existing function. An alternative model, termed subfunctionalization, predicts that both members accumulate mutations and that the function of the ancestral gene is partitioned between the duplicates. We found a striking example of subfunctionalization in Archaea, in which a previously unrecognized oligomeric form of tRNA-splicing endonuclease, a heterotetramer, arises by gene duplication and subsequent subfunctionalization (10, 11).

Duplicated genes have been proposed to be at the origin of the permuted rRNA and tmRNA genes (12–14). Permuted tRNA genes could similarly result from DNA rearrangement through alternative pathways that followed a duplication event (5, 9, 15).

All such alternative pathways must allow the correct processing of the transcript to generate a functional tRNA. The primary product of transcription of a typical short intronless tRNA gene (pre-tRNA) has to go through processing in the form of trimming at both termini and modification (16, 17). Similarly, a rearranged tRNA duplicated gene must produce a single transcript capable of being cleaved by the preexisting tRNA endonuclease to form a permuted, functional tRNA.

To see how the transcript of a duplicated gene can be thus processed to generate a permuted tRNA, we developed a strategy involving two tRNA-splicing endonucleases characterized by two distinct mechanisms of recognition of the pre-tRNA and of two avatar (av) or model pre-tRNAs (av pre-tRNA) (Figs. 1 and 2). We call them avatars because they are not actual and may never have existed or ever exist in algae, but rather are the incarnation of a model version. The goal is to see if they help elucidate the features that a transcript derived from a rearranged duplicated gene must have to give rise to permuted tRNA. This approach is valid even if the hypothesis that individual av pre-tRNAs have a role in the production of permuted tRNA is false or improbable (18).

Fig. 1.

Predicted structures of possible conformation of the duplicate pre-tRNABHB (FA*FB). Each Pre-tRNA^BHB consists of two regions derived from yeast pre-tRNA^Phe (nucleotides 1–31 and 38–76), joined by a 25-nt insert that corresponds to the BHB motif of archaeal tRNA^Trp; the sequence is written in capital letters. The 5′ halves and the 3′ halves are colored in gray and the intron in black. The leader sequence, the trailer, and the spacer sequence are written in lowercase letters. A rectangular box delimits the spacer sequence. Triangles indicate the splicing sites for METJA and SCHPO tRNA endonuclease; black arrows indicate the size of the two synthetized transcripts (FA and FA*FB). The products of the reactions and the size in nucleotides are indicated.

Fig. 2.

Predicted structures of possible conformation of the duplicate pre-tRNABHB (FMA*FMB). tRNA FMA is a mutant tRNA FA lacking the 5′ bulge and FMB is a mutant tRNA FB lacking the 3′ bulge. Gray triangles indicate the splicing sites for both METJA and SCHPO tRNA endonuclease; the white triangle indicates the splicing site of the SCHPO tRNA endonuclease. The black arrows indicate the size of the two synthetized transcripts (FMA and FMA*FMB). The products of the reactions and the size in nucleotides are indicated.

In both Eukarya and Archaea, the specificity for recognition of pre-tRNA resides in the splicing endonucleases (19). These enzymes remove the intron by making two independent endonucleolytic cleavages. The eukaryal enzyme normally acts in a mature–domain–dependent mode (20–23). There is, however, a single striking exception: a specific structure consisting of two 3-nt bulges separated by a 4-bp helix, the so-called BHB motif, is recognized and cleaved in vitro by the eukaryal enzyme with no involvement of the mature domain. This finding presents interesting evolutionary implications. The BHB is a universal substrate that is also cleaved by the archaeal enzymes (24, 25). In fact, many intron-containing archaeal pre-tRNAs are characterized by a BHB constructed from the intron and from exonic sequences.

Because the archaeal splicing endonucleases, in general, remove the intron in a mature–domain–independent mode, the intron in archaeal pre-tRNAs can be located in different positions relative to the mature domain. Several archaeal pre-tRNAs contain relaxed forms of the BHB motif, one being the BHL consisting of a single 3-nt bulge and an internal loop separated by a 4-bp helix (26, 27).

We use two tRNA endonucleases with a different architecture: the Methanocaldococcus jannaschii (METJA) homotetramer from the archaeal kingdom and the Schizosaccharomyces pombe (SCHPO) heterotetramer from the eukaryal kingdom. The METJA enzyme is a homotetramer in which two of the subunits play a structural role and two contain catalytic sites. The METJA enzyme cleaves the BHB twice and the BHL once at the only existing bulge at the 3′ end.

The SCHPO enzyme is composed of the four subunits Sen2, Sen54, Sen34, and Sen15. Sen2 and Sen34 are the catalytic subunits responsible, respectively, for the cleavage of the 5′ and 3′ splicing sites. Sen15 and Sen54 are the two structural subunits necessary to form the active complex. Interactions between Sen2 and Sen54 and between Sen34 and Sen15, respectively, were detected by the two-hybrid system. The sen54 subunit is presumed to mediate the interaction with the mature domain of the tRNA (28).

Both av pre-tRNAs used in this study are transcribed from an external promoter and are able to fold into two conformations, conformation 1 and conformation 2. Only conformation 2 can generate a permuted tRNA; the tRNA endonuclease of the cell must therefore be able to cleave the competent conformation. To produce av pre-tRNA I, we generated tandem repeat genes of a pre-tRNA^BHB separated by a 50-nt-long spacer sequence (Fig. 1A). The pre-tRNA^BHB consists of two regions derived from yeast pre-tRNA^Phe (nucleotides 1–31 and 38–76), joined by a 25-nt insert that corresponds to the canonical BHB motif present in the anticodon loop of archaeal tRNA^Trp and is a universal substrate for all of the tRNA-splicing endonucleases. To date there are no reports of a permuted tRNA^Phe, so our work represents an idealized model. The spacer sequence placed between the two genes was derived from the intergenic spacer present between the multiple copies of tRNA^Ser on chromosome 3 of Ostreococcus lucimarinus (OSTLU) (5).

A pre-tRNA^BHB gene was generated using two partially overlapping oligomers. This gene was then used as a template to generate two pre-tRNA genes by PCR. The gene FA has a T7 promoter followed by a 17-nt leader sequence at the 5′ end and a spacer sequence at the 3′ end (Fig. 1). The gene FB has the spacer sequence at the 5′ end and a trailer sequence of 21 nt at the 3′ end (Fig. 1). These two genes were then used as templates in a single PCR, using as primers two oligomers complementary to the leader sequence at the 5′ end and the trailer sequence at the 3′ end. Because of the overlapping spacer sequence between the two templates (at the 3′ end of FA and the 5′ end of FB), the amplification product resulted in a single DNA fragment containing two tRNA^Phe (FA*FB) under the control of a unique T7 promoter.

Fig. 1 schematically shows how the same transcript av pre-tRNA I can fold into two different conformations. Conformation 1 has the two pre-tRNAs (FA*FB) separated by a spacer sequence (Fig. 1A). Conformation 2 comprises one pre-tRNA formed by the 5′ half of FA and the 3′ half of FB and a second one formed by the 3′ half of FA interacting with the 5′ half of FB with a loop generated by the spacer sequence closing the acceptor stem (Fig. 1B). The FA gene and the duplicated genes FA*FB were transcribed as described in Materials and Methods.

Fig. 3 shows that the archaeal enzyme (METJA) could cleave the FA substrate correctly (lane 5), generating the 13-nt intron (IVS) of 16 nt as in the case of pre-tRNA^BHB (lane 1), a 53-nt 5′ half, and a 89-nt 3′ half. Incubation of the FA substrate with the eukaryal enzyme (SCHPO) results in exactly the same products [lane 2 (FA) and lane 5 (pre-tRNA^BHB) of Fig. 4].

Fig. 3.

Processing by METJA tRNA endonuclease. Lanes: 1, pre-tRNA^BHB; 2, FA*FB; 3, FMA*FMB; 4, FMA; 5, FA; 6, pre-tRNA^BHB; 7, tRNA^Phe. The reaction products are indicated.

Fig. 4.

Processing by SCHPO tRNA endonuclease. Lanes: 1, FMA; 2, FA; 3, FMA*FMB; 4, FA*FB; 5, pre-tRNA^BHB; 6, pre-tRNA^BHB; 7, tRNA^Phe. The reaction products are indicated.

For the duplicated gene FA*FB, both possible conformers (Fig. 1) are expected to be cleaved to give exactly the same gel bands. Incubation of FA*FB with the METJA endonuclease (Fig. 3, lane 2) gave four products: a permuted tRNA of 125 nt, a 3′ half of 60 nt, a 5′ half of 53 nt, and an intron of 16 nt. However, in sharp contrast to the reaction with METJA enzyme, the FA*FB substrate (lane 4) was not cleaved at all by the SCHPO enzyme. Because both the SCHPO and METJA enzymes cleave the pre-tRNA^BHB and FA substrates without difficulty, we argue that only conformation 2, and not conformation 1, is used. We suggest that the mature–domain–dependent SCHPO enzyme cannot cleave FA*FB because of steric problems caused by the presence of two mature domains in the same molecule.

To probe these inferences further, we generated a tandem repeat gene with two mutated pre-tRNA^BHB moieties. The spacer sequences between the two genes and the 5′ leader and the 3′ trailer were unchanged (Fig. 2A). However, the mutant FMA gene is a pre-tRNA^BHB gene lacking the 3-nt bulge at the 5′ end, and the mutant FMB gene is a pre-tRNA^BHB gene lacking the 3-nt bulge at the 3′ end (Fig. 2A).

Each of the two genes was generated using two partially overlapping oligomers. The FMA gene has a T7 promoter followed by a 17-nt leader sequence at the 5′ end and a spacer sequence at the 3′ end (Fig. 2A). The FMB gene has the spacer sequence at the 5′ end and a trailer sequence of 21 nt at the 3′ end (Fig. 2A). These two genes were then used as templates in a single PCR with two oligomer primers complementary to the leader sequence at the 5′ end and the trailer sequence at the 3′ end. Because of the overlapping spacer sequence between the two templates at the 3′ end of FMA and the 5′ end of FMB, the amplification product resulted in a single DNA fragment comprising two tRNA (FMA*FMB) genes under the control of a single T7 promoter.

Fig. 2 shows schematically how the av pre-tRNA II transcript can fold into two different conformations. Conformation 1 is represented by two pre-tRNAs (FMA, FMB) separated by a spacer sequence (Fig. 2A), whereas conformation 2 consists of two pre-tRNAs: one formed by the 5′ half of FMA and the 3′ half of FMB and the second formed by of the 3′ half of FMA interacting with the 5′ half of FMB, with a loop generated by the spacer sequence closing the acceptor stem (Fig. 2B). Only the latter conformation can form a canonical BHB motif, so in this case the transcript can fold into either of two conformations that are not expected to produce the same cleavage products.

The FMA gene and the duplicated genes FMA*FMB were transcribed as described in Materials and Methods. Incubation of the substrate FMA with the METJA endonuclease resulted in two bands: the 3′ half 89-nt and the 5′ half 66-nt products (Fig. 3, lane 4), indicating that this enzyme cleaves exclusively at the 3′ site. This result is in accord with the notion that the METJA homotetramer cleaves the BHL only at the 3′ site.

Instead, when the FMA substrate was incubated with the SCHPO endonuclease, the reaction products consisted of four bands: the 89-nt 3′ half, the 53-nt 5′ half, the IVS, and a 66-nt product corresponding to the 5′ half of a tRNA spliced only at the 3′ bulge (Fig. 4, lane 1). The SCHPO tRNA endonuclease processes this substrate by two mechanisms. One mechanism, similar to that used by the METJA endonuclease, cleaves only at the 3′ site, whereas the other mechanism (that is, mature–domain–dependent) cuts at the 5′ half to release an intron that is 3 nt shorter because of the 5′ 3-nt deletion compared with the FA substrate (Fig. 4, lane 2).

The METJA endonuclease yields three products from the FMA*FMB transcript (Fig. 3, lane 3): the 73-nt 3′ half, the 125-nt permuted tRNA, and the 66-nt 5′ half. This result indicates that the substrate can be cleaved by the METJA enzyme only if it is folded in conformation 2 (Fig. 2B): the METJA enzyme does not cleave conformer 1.

Lane 3 (Fig. 4) shows the products generated by the SCHPO enzyme. The processing results in a 66-nt 5′ half, the 53-nt 5′ half, and the IVS similar to the FMA substrate (Fig. 4, lane 1). There is also a band of 73 nt that corresponds to the cleavage of the bulge site in the tRNA FMB. The presence of an IVS and a 53-nt 5′ half is consistent with processing of the substrate only when it is folded in conformation 1 (Fig. 2A). Once more the mature–domain–dependent enzyme fails to generate a 125-nt permuted tRNA.

In summary, the SCHPO enzyme does not cleave av pre-tRNA I (FA*FB) and cleaves only conformer 1 of av pre-tRNA II (FMA*FMB). The METJA enzyme cleaves av pre-tRNA I (FA*FB) and av pre-tRNA II (FMA*FMB) only in conformation 2.

What about the tRNA endonuclease in cryptophytes and chlorarachniophytes?

To identify homologs of the pre-tRNA endonuclease, we performed BLASTP searches on two available complete sequences for C. merolae (http://merolae.biol.s.u-tokyo.ac.jp/) and O. lucimarinus (http://genome.jgi-psf.org/).

There are two homologs of the sequence of METJA endonuclease (Q58819) in each organism. Using each retrieved sequence as a probe, we performed a BLASTP search of the SWISS-PROT database that confirmed that these two homologs correspond to either the sen34 or the sen2 catalytic subunits of yeast. These subunits are highly homologous throughout the entire length of the protein (Figs. S1 and S2). The sequence alignment units demonstrate the highest degree of similarity in the region harboring the endonuclease active sites.

A search with the sequence of the YEAST subunit sen54 (Q02825) revealed one homolog in each organism with an E-value 0.013 in CYAME and 3.77e-1 in OSTLU. Using each sequence as a probe in a BLASTP search of the SWISS-PROT database, we confirmed that these two homologs correspond to sen54. These putative homologs show similarities in the amino- and carboxyl-terminal regions of the protein when aligned with the subunits of the YEAST, SCHPO, and HUMAN enzymes (Fig. S3).

HUMAN (Q8WW01) and YEAST (Q04675) sen15 sequences found no homologs, but this was expected because the sequence of the HUMAN sen15 could not be found by homology search either; it was determined by mass spectroscopy once the enzyme was purified (29). We conclude that the pre-tRNA endonucleases of these two algae are typical eukaryotic heterotetramers, likely to function as mature–domain–dependent tRNA endonucleases.

In general, algae present either duplicated or permuted genes, which are snapshots at different evolutionary stages. The duplicated genes in contemporary algae represent evolutionary stages that preceded recombinatorial events thought to be responsible for the formation of a gene that could generate a single transcript producing permuted tRNAs. These duplicated genes give rise to two distinct transcripts. A single transcript would not, like av pre-tRNA I (FA*FB), be cleaved by the eukaryal enzyme. By contrast, the archaeal enzyme cleaves conformer 2 of av pre-tRNA I (FA*FB), giving rise to permuted tRNA.

Algal permuted tRNA genes lost one of the two mature domains, eliminating the steric problems that impede the eukaryal enzyme action. As a consequence of the loss of internal sequences in eukaryal pre-tRNA, the function of the canonical internal promoter, with its characteristic A-box and B-box features, could be impaired. During evolution, the liberation from requirement for an internal promoter through the acquisition of an external promoter with a TATA-box would thus be essential for the appearance of permuted tRNA.

An analysis of the sequences of permuted tRNA genes in OSTLU revealed genes presenting these features. The gene coding for the permuted tRNA^Cys_GCA has lost one of the two 3′ halves to permit folding into a single mature tRNA domain (5).

A recent study of atypical tRNA transcripts in Thermofilum pendens revealed permuted tRNA genes in Archaea that are organized as are the genes in algae. Although analysis of the conservation of the structure of the intron in the permuted tRNA and pre-tRNA in the green algae revealed that they can form either a BHB motif or a relaxed BHL motif (typical of archaeal pre-tRNAs), the archaeal permuted genes have a CCA at the 3′ end and show no sequence similarity to the algae, excluding the possibility of horizontal gene transfer of permuted tRNA genes from other organisms (15).

Compelling evidence favoring the role of an endonuclease-dependent pathway in the evolutionary process resulting in the extinction of group I introns in Archaea has been presented previously (30). The current results support a further scenario in which the tRNA endonucleases have a central role in the generation of permuted tRNA.

Genomes contain records of past scenarios that permit the reconstruction of how things developed. It has been pointed out that “biocosmology,” a new science, can attempt to elucidate the evolution of the biological universe (31).

Materials and Methods

tRNA Splicing Endonuclease Purification.

The enzymes from Methanocaldococcus jannaschii (METJA), and the enzyme from Schizosaccharomyces pombe (SCHPO) were purified as described (30).

RNA Synthesis in Vitro and RNA Cleavage Reactions.

DNA templates, synthesized as described in ref. 11, were used in T7 RNA polymerase transcription reactions carried out with the Ambion T7-Megashortscript kit. [α-32P]UTP (800 Ci/mmol; Perkin–Elmer) (1 Ci 1/4 37 GBq) was included in the reaction to label the transcription products. Transcripts of the correct size were purified by electrophoresis on a 10% (wt/vol) denaturing polyacrylamide gel, followed by elution, phenol extraction, and ethanol precipitation. The same conditions were used for all of the reactions: 25 mM Tris·HCl (pH 7.5), 5 mM MgCl2, 100 mM NaCl, and 10% (vol/vol) glycerol, 20 fmol of substrate, incubation for 1 h at 65 °C upon addition of the endonuclease. The reaction including the SCHPO enzyme was carried out in the same buffer in the presence of 7 mM MgCl2, and incubated at 30 °C. All reactions were stopped by phenol extraction and ethanol precipitated (30). The products were separated on 10% (wt/vol) denaturing polyacrylamide gels and analyzed on a Molecular Dynamics model Storm 860 PhosphorImager using ImageQuant software, version 4.

The sequence alignments are described in SI Materials and Methods.