Healing for destruction: tRNA intron degradation in yeast is a two-step cytoplasmic process catalyzed by tRNA ligase Rlg1 and 5'-to-3' exonuclease Xrn1.

In eukaryotes and archaea, tRNA splicing generates free intron molecules. Although ∼ 600,000 introns are produced per generation in yeast, they are barely detectable in cells, indicating efficient turnover of introns. Through a genome-wide search for genes involved in tRNA biology in yeast, we uncovered the mechanism for intron turnover. This process requires healing of the 5' termini of linear introns by the tRNA ligase Rlg1 and destruction by the cytoplasmic tRNA quality control 5'-to-3' exonuclease Xrn1, which has specificity for RNAs with 5' monophosphate.

tRNAs are essential for translation in all kingdoms of life. In addition, tRNAs play important roles in other cellular events, such as protein degradation, apoptosis, cellular response to stress, and cancer (Phizicky and Hopper 2010). tRNA biogenesis involves numerous post-transcriptional processes, including removal of 5′ leaders and 3′ trailers, addition of 3′ CCA nucleotides and nucleoside modifications, and, for tRNAs encoded by intron-containing genes, splicing of introns (Phizicky and Hopper 2010).

Splicing of introns is essential for cells to generate complete sets of tRNAs to decode all codons. The percentage of tRNA genes that harbor introns varies significantly, from 0% in bacteria to 5% in mice and humans to >60% in some archaea (Genomic tRNA Database, http://lowelab.ucsc.edu/GtRNAdb). In yeast Saccharomyces cerevisiae, 59 (21.5%) of 274 tRNA genes from 10 different isodecoder tRNA families contain an intron (Phizicky and Hopper 2010).

In yeast and vertebrates, tRNA splicing is catalyzed by the heterotetrameric tRNA splicing endonuclease (SEN) complex, yielding the 5′ half with a 2′, 3′-cyclic phosphate, the intron with a 5′ hydroxyl and 2′, 3′-cyclic phosphate, and the 3′ half with a 5′ hydroxyl (Knapp et al. 1979; Trotta et al. 1997; Paushkin et al. 2004). In vertebrates, tRNA splicing occurs in the nucleus (Melton et al. 1980; Lund and Dahlberg 1998; Paushkin et al. 2004), whereas it is a cytoplasmic process in yeast, since the SEN complex is located on the outer surface of mitochondria (Yoshihisa et al. 2003). Despite the difference in localization, the structure and function of the SEN complex are conserved from yeast to humans (Trotta et al. 1997; Paushkin et al. 2004).

In yeast and plants, after splicing, the 5′ and 3′ halves are ligated through a series of enzymatic reactions that require three activities of the tRNA ligase Rlg1 (Greer et al. 1983; Phizicky et al. 1986; Englert and Beier 2005): (1) a cyclic phosphodiesterase (CPDase) activity that opens the 2′, 3′ cyclic phosphate of the 5′ exon, (2) a polynucleotide kinase activity that phosphorylates the 5′ end of the 3′ exon, and (3) an ATP-dependent ligase activity that joins the two exons together. Finally, the 2′-phosphate remaining at the ligation junction is removed by the phosphotransferase Tpt1 (Culver et al. 1993) to form the mature tRNA.

Although much is known about the ligation of two tRNA halves, little is known about the fate of the other splicing product, the free tRNA intron. In contrast to the high abundance of mature tRNAs, free introns are barely detectable, indicating that they are degraded rapidly after splicing. To date, two tRNA degradation pathways have been identified in yeast as tRNA quality control mechanisms. The nuclear surveillance pathway provides one mechanism; pre-tRNAMeti lacking m1A58 and pre-tRNAs with aberrant 3′ ends are polyadenylated by the TRAMP complex and further degraded by the nuclear exosome (Kabada et al. 2004; Copela et al. 2008; Ozanick et al. 2009). The other pathway, the rapid tRNA decay (RTD) pathway, destroys mature tRNAs that lack certain combinations of modifications or tRNAs that bear unstable T or acceptor stems (Chernyakov et al. 2008; Whipple et al. 2011).

Although turnover pathways for aberrant precursor and mature tRNAs have been studied extensively, little is known about degradation of tRNA introns. Since ∼3,000,000 tRNAs are produced per generation in yeast (Waldron and Lacroute 1975) and >20% are intron-containing tRNAs, an enormous amount of tRNA introns (∼600,000) is produced during each generation. Cells may need to efficiently degrade tRNA introns to generate nucleotides or prevent introns from interacting with other RNAs and RNA-binding proteins and perhaps inhibiting their functions. There is precedent for this, as cytoplasmic accumulation of mRNA intron lariats in human cells sequesters and suppresses TDP-43, a key factor involved in amyotrophic lateral sclerosis (Armakola et al. 2012).

Here we uncovered the mechanism for tRNA intron decay in yeast. Through a genome-wide screen for novel gene products involved in tRNA biology, we found that tRNA introns accumulate in cells lacking the cytoplasmic RTD 5′-to-3′ exonuclease Xrn1. Degradation of introns via Xrn1 requires healing of tRNA intron 5′ termini by the tRNA ligase Rlg1. Neither defects in 3′-to-5′ nucleases nor a temperature-sensitive (ts) mutation of the nuclear RTD 5′-to-3′ exonuclease Rat1 affect tRNA intron turnover. We conclude that tRNA intron decay is a cytoplasmic two-step process that is catalyzed by the 5′-to-3′ nuclease Xrn1 and tRNA ligase Rlg1.

Results and Discussion

The 5′-to-3′ exonuclease Xrn1 degrades tRNA introns

To identify unknown gene products involved in tRNA biogenesis, degradation, and subcellular trafficking, we conducted an unbiased genome-wide screen of yeast genes using the S. cerevisiae MATa haploid deletion collection (results of this screen will be presented separately) (Winzeler et al. 1999; Giaever et al. 2002). One of the primary goals of our screen was to identify gene products involved tRNA intron turnover.

To conduct this screen, we assessed the impact of each gene product upon tRNAIleUAU, which contains the largest tRNA intron (60 nucleotides [nt]) in yeast, by Northern analysis (Wu et al. 2013). We found that cells deleted for XRN1 accumulate high levels of free tRNAIleUAU intron compared with wild-type cells. We verified this result using a probe complementary to the nucleotides 1–37 of the 5′ exon and nucleotides 1–30 of the intron of tRNAIleUAU (Fig. 1A [probe 1], B [lanes 1,2]). Similar results were obtained by analysis of xrn1Δ cells attained from the independent MATα collection (Fig. 1B, lane 3; Giaever et al. 2002). These results were further confirmed usng another probe (Fig. 1A, probe 2) that hybridizes solely to the 60-nt tRNAIleUAU intron (Fig. 1C). Accumulation of introns was suppressed when Xrn1 was exogenously expressed in xrn1Δ cells from a multicopy plasmid, whereas xrn1Δ cells transformed with vector alone accumulated introns (Fig. 1D, lanes 3,4). We also assessed intron levels for four additional RNAs encoded by tRNALeuCAA, tRNALysUUU, tRNATrpCCA, and tRNAProUGG genes using probes complementary solely to the entire intron of each tRNA. Deletion of XRN1 results in accumulation of all tested tRNA introns (Supplemental Fig. S1). Quantitative analysis of the signal intensity of tRNA introns compared with initial transcripts demonstrated 2.5-fold to 11-fold increases in xrn1Δ cells compared with control cells (Supplemental Fig. S1). These data provide evidence that the 5′-to-3′ exonuclease Xrn1 affects tRNA intron turnover in yeast.

Figure 1.

Deletion of the XRN1 gene leads to tRNA intron accumulation. (A) A schematic view of a primary tRNAIleUAU transcript with 5′ leader and 3′ trailer, exons, and an intron and the probes used for Northern hybridization. (B) Two independent xrn1Δ strains (MATa xrn1Δ and MATα xrn1Δ) accumulate tRNAIleUAU introns. Northern analysis was performed using probe 1. (C) Northern analysis of MATa wild-type (wt) and two independent xrn1Δ strains using probe 2. (D) Complementation of the phenotype of xrn1Δ cells by plasmid encoded Xrn1. Northern analysis of RNAs from MATa wild-type, xrn1Δ, and xrn1Δ cells transformed with the multicopy plasmid YEpXrn1 or vector alone were performed using probe 2. Ratios of the signal intensities of introns versus primary tRNA transcripts were calculated and normalized to the wild-type ratio. (P) Primary tRNA transcript (145 nt); (I) end-mature intron containing tRNA (136 nt); (2/3 with a question mark) unknown species, which is likely the 5′ 2/3 splicing intermediate according to its size (97 nt); (M) mature tRNA (76 nt); (IN) intron (60 nt).

5′ monophosphorylation is requisite prior to intron degradation by Xrn1

It is known that Xrn1 specifically hydrolyzes RNA molecules with a 5′ monophosphate group (Stevens 1980). However, tRNA splicing generates intron molecules with a 5′ hydroxyl group (Knapp et al. 1979). Therefore, we hypothesized that the 5′ termini of nascent spliced introns must be phosphorylated prior to degradation by Xrn1 (Fig. 2A).

Figure 2.

The 5′ end of tRNA introns is monophosphorylated prior to degradation by Xrn1. (A) Schematic view of tRNA splicing products and the strategy for determining the 5′ phosphorylation status of introns using CIP and TEX. (B) CIP and TEX treatments. Northern analysis of untreated or treated total small RNAs isolated from xrn1Δ cells was performed using probe 2. (P, white block) Intron with a 5′ phosphate; (HO, white block) intron with 5′ hydroxyl. 5.8S and 5S served as controls.

To test this hypothesis, we treated total small RNAs isolated from xrn1Δ cells with calf intestinal phosphatase (CIP), which catalyzes hydrolysis of 5′ phosphate groups from RNA, and/or a Terminator 5′-phosphate-dependent exonuclease (TEX), which specifically degrades RNAs with a single 5′ phosphate (Fig. 2A; Patrick et al. 2009). If the 5′ ends of tRNA introns that accumulate in xrn1Δ cells have a terminal phosphate, CIP treatment could lead to a change in electrophoretic mobility; in contrast, TEX treatment would result in intron degradation. 5.8S rRNA and 5S rRNA, which harbor 5′ monophosphate and triphosphate (Maxam et al. 1977; Henry et al. 1994), respectively, served as positive and negative controls, and their sensitivity/resistance to TEX treatments were as anticipated (Fig. 2B). Treatment of RNAs with CIP resulted in faster migration of the introns (Fig. 2B, lane 2) compared with mock-treated samples (Fig. 2B, lane 1), and TEX treatment resulted in intron degradation (Fig. 2B, lane 3). Introns from samples first treated with CIP to remove the 5′ phosphate were resistant to subsequent degradation by TEX (Fig. 2B, lane 4). Interestingly, end-matured intron-containing pre-tRNAs are resistant to TEX, suggesting that their secondary structure inhibits Terminator exonuclease. Together, the data demonstrate that tRNA introns are 5′ monophosphorylated prior to degradation by Xrn1; these results motivated a search for the kinase that phosphorylates tRNA introns.

The tRNA ligase Rlg1 phosphorylates tRNA introns prior to degradation by Xrn1

A well-characterized yeast RNA kinase is the tRNA ligase Rlg1. Rlgl contains three enzymatic activities required for tRNA ligation: CPDase, polynucleotide kinase, and ATP-dependent ligase activities. Prior in vitro studies proposed assembly of a SEN–Rlg1 complex for concerted tRNA splicing and ligation (Greer 1986). Moreover, Rlg1 mutant cells were found to accumulate an oligonucleotide of the same size as the tRNAIle intron (Phizicky et al. 1992). Therefore, we considered the possibility that free introns generated by SEN splicing may also be substrates of Rlg1.

If Rlg1 is required for intron degradation, loss of RLG1 function should cause accumulation of unphosphorylated introns. Since RLG1 is an essential gene for cell viability, there is no rlg1Δ strain in the yeast deletion collections. We therefore used ts mutants of RLG1, rlg1-4, and rlg1-10 (Phizicky et al. 1992). RNAs isolated from rlg1-4 cells were compared with RNAs from wild-type and xrn1Δ cells by Northern analysis using two probes, 2 and 3, which hybridize to the intron and the 5′ exon of tRNAIleUAU, respectively (Fig. 3A; Supplemental Fig. 2A). Two very low abundant species of introns with different mobilities were detected from RNAs isolated from wild-type cells grown at 23°C or after a 2-h shift to 37°C (Fig. 3B, lanes 1,3). RNA from rlg1-4 ts cells grown at 23°C (Fig. 3B, lane 5) show results similar to wild-type cells, but after a 2-h shift to the nonpermissive temperature (37°C), there was significant accumulation of tRNA exons (Fig. 3B, lane 7; Supplemental Fig. S2, lane 4), as previously reported (Phizicky et al. 1992). Likewise, tRNAs isolated from rlg1-10 also accumulated tRNA exons (Supplemental Fig. S2B, lane 6). Importantly, both rlg1-4 and rlg1-10 also accumulated tRNA introns (Fig. 3B, lane 7; Supplemental Fig. S2B, lanes 4,6), demonstrating that Rlg1 is required for tRNA intron turnover. In contrast to xrn1Δ cells that accumulate the 5′ monophosphorylated introns with slower mobility (Fig. 3B, lanes 9,11), the introns that accumulate in rlg1 ts mutants had faster mobility (Fig. 3B, lane 7; Supplemental Fig. S2B, lanes 4,6), supporting the idea that the introns accumulating in rlg1 ts mutants lack 5′ phosphate.

Figure 3.

Rlg1 phosphorylates tRNA introns. (A) Probes 3 and 2 were used in the Northern blot shown in B. (B) Northern analysis of TEX-treated and untreated small RNAs from wild-type (wt), rlg1-4, xrn1Δ, and rlg1-4 xrn1Δ cells grown at 23°C or after a 2-h shift to 37°C. (Black block) 5′ exon; other symbols are as described in Figures 1 and 2. 5.8S and 5S rRNAs served as controls.

If Rlg1 phosphorylates introns prior to Xrn1-mediated degradation, the double mutant of RLG1 and XRN1 genes would accumulate unphosphorylated introns. We therefore generated a rlg1-4 xrn1Δ strain. At 23°C, the rlg1-4 xrn1Δ cells (Fig. 3, lane 13) show the same phenotype as xrn1Δ cells—accumulation of introns with slow mobility. However, after a 2-h shift to 37°C, rlg1-4 xrn1Δ cells also accumulate introns with faster mobility (Fig. 3B, lane 15). The data are consistent with the hypothesis that when cells lose RLG1, introns fail to be properly phosphorylated and thus cannot be degraded by Xrn1, supporting the notion that Rlg1 acts upstream of Xrn1 in intron turnover.

To test the hypothesis that loss of RLG1 causes accumulation of introns lacking 5′ phosphate, we performed TEX treatment followed by Northern analysis for RNAs isolated from wild-type, rlg1-4, xrn1Δ, and rlg1-4 xrn1Δ cells (Fig. 3). If introns accumulating in rlg1 ts mutants are unphosphorylated, they would be resistant to TEX treatment. As anticipated, introns accumulating in rlg1-4 cells after a shift to 37°C were not degraded by TEX (Fig. 3B, lanes 7,8), whereas introns accumulating in xrn1Δ cells are degraded (Fig. 3B, lanes 9–12). Similar to introns in xrn1Δ cells, introns accumulating in rlg1-4 xrn1Δ cells grown at 23°C are also degraded by TEX (Fig. 3B, lanes 13,14). However, of the two intron species accumulating in rlg1-4 xrn1Δ cells after shift to 37°C (Fig. 3B, lane 15), only the pre-existing monophosphorylated species was sensitive to TEX treatment (Fig. 3B, lane 16). As anticipated, the unstructured 5′ tRNA exons bearing 5′ monophosphates that accumulate in rlg1-4 and rlg1-4 xrn1Δ cells upon shift to nonpermissive temperature were degraded by TEX (Fig. 3B, lanes 7,8,15,16). Together, the data show that loss of RLG1 leads to accumulation of the intron with 5′ hydroxyl, whereas loss of XRN1 causes accumulation of introns with 5′ phosphate. These studies provide evidence that Rlg1 possesses the kinase activity that phosphorylates introns prior to recognition and degradation of introns by Xrn1.

We also analyzed introns of four other tRNAs, tRNALeuCAA, tRNALysUUU, tRNATrpCCA, and tRNAProUGG in the rlg1-4 mutant. Loss of RLG1 lead to a fourfold to 15.6-fold increase in intron levels (Supplemental Fig. S1). These data suggest that the Rlg1- and Xrn1-mediated tRNA intron degradation pathway is generally used for all tRNAs.

Prior in vitro studies of Rlg1 substrate specificity showed that, in addition to catalyzing ligation of tRNA halves, Rlg1 is able to phosphorylate or ligate the ends of artificial linear RNA substrates (Phizicky et al. 1986; Apostol et al. 1991). Our finding that Rlg1 phosphorylates introns indicates that linear molecules can be the substrates for Rlg1 kinase activity in vivo. On the other hand, the question arises as to whether Rlg1 ligase activity is able to join the intron ends to generate circular introns. Rlg1 is capable of circularizing artificial substrate in vitro (Phizicky et al. 1986), and stable circular introns formed by 3′,5′-phosphodiester linkage have been reported for certain archaea (Salgia et al. 2003). However, since TEX is able to effectively degrade introns, it would appear that the majority of introns in yeast are linear, although there may be minor pools of circular introns that are beyond our detection level. In addition, we did not detect elevated levels of end-matured intron-containing tRNAs or 2/3 molecules consisting of an exon and an intron in xrn1Δ cells, indicating that Rlg1 does not join the spliced introns and exons back together even in the presence of high intron levels, whereas it clearly is able to phosphorylate free introns.

Although yeast and plants use Rlg1 and the 2′ phosphotransferase Tpt1 for tRNA ligation, vertebrates appear to contain two distinct tRNA ligation pathways: a direct ligation pathway and a yeast-like ligation pathway. In the first, a 3′-to-5′ ligase activity is used to directly join the 3′ phosphate of the 5′ exon to the 5′ hydroxyl of the 3′ exon (Popow et al. 2012). In the second, hClp1, a component of the mammalian SEN complex, catalyzes phosphorylation of the 5′ end of the 3′ exon of tRNAs, although the ligase is not identified (Zillmann et al. 1991; Weitzer and Martinez 2007; Ramirez et al. 2008). New studies reported that CLP1 mutation leads to severe neurodegenerative disease, and fibroblasts from human patients accumulate linear introns with a 5′ hydroxyl, although intron accumulation was not observed in in vitro analyses (Karaca et al. 2014; Schaffer et al. 2014). We wonder whether introns are targets of CLP1 in vertebrate cells and whether it is possible that defects in tRNA intron turnover might contribute to the neurological phenotypes caused by CLP1 mutations.

tRNA intron degradation is a 5′-to-3′ process in the cytoplasm

Although the above studies provide evidence that turnover of tRNA introns is a two-step process consisting of healing by Rlg1 followed by 5′-to-3′ degradation by Xrn1, they do not eliminate the possibilities that other RNA decay machineries also degrade introns and that this process may occur in other cellular compartments than the cytoplasm. To address these issues, we examined whether Rat1, the nuclear 5′-to-3′ exonuclease Xrn1 ortholog (Johnson 1997), affects intron turnover. We used a ts mutant of RAT1, rat1-1 (Li et al. 2011), since RAT1 is an essential gene that is not contained in the deletion collection. Similar to wild type (Fig. 4A, lanes 3,4), rat1-1 fails to accumulate tRNA introns after a 2-h shift to the nonpermissive temperature 37°C (Fig. 4A, lanes 1,2), arguing that tRNA introns are not Rat1 substrates and supporting the notion that tRNA intron degradation is restricted to the cytoplasm in yeast. Since splicing occurs in the nucleus in vertebrates, an intriguing question is: Where does intron degradation occur in vertebrates?

Figure 4.

tRNA intron turnover is a 5′-to-3′ cytoplasmic process. (A) Northern analysis of RNAs from rat1-1 and wild-type (wt) cells grown at 23°C or after a shift to 37°C using probe 2. (B) Northern analysis of RNAs from MATa rai1Δ, dxo1Δ, ski2Δ, ski3Δ, ski7Δ, ski8Δ, xrn1Δ, and wild-type cells. (C) Northern analysis of RNAs from MATα rai1Δ, dxo1Δ, ski2Δ, ski3Δ, ski7Δ, ski8Δ, xrn1Δ, and wild-type cells. Ratios of the signal intensities of introns versus primary tRNA transcripts were calculated and normalized to the wild-type ratio.

Through our screen of the deletion collection, we found that none of the mutants of the other 5′-to-3′ or 3′-to-5′ RNA decay enzymes accumulated tRNA introns. To verify that they are not involved in tRNA intron turnover, we examined their deletion mutants from the independent MATα deletion collection (Giaever et al. 2002).

Two other 5′-to-3′ RNA decay enzymes, Rai1, which is predominantly localized to the nucleus (Xue et al. 2000), and Dxo1, which is primarily cytoplasmic (Chang et al. 2012), were examined. There is no accumulation of tRNA introns in RNAs from dxo1Δ or rai1Δ cells (Fig. 4B,C, lanes 1,2). We also examined mutations of subunits of the cytoplasmic exosome, the SKI complex (Mitchell and Tollervey 2003). None of the SKI mutants (ski2Δ, ski3Δ, ski7Δ, and ski8Δ) accumulated tRNA introns (Fig. 4B,C, lanes 3–6). Intron levels of tRNALeuCAA were also examined in all of these mutants, and no accumulation was observed (data not shown). Our data show that the exosome does not contribute to tRNA intron turnover, suggesting that intron turnover may not occur via 3′-to-5′ exonucleases.

Conclusions

Here we report the discovery of the mechanism of intron turnover in yeast (Fig. 5). It is a process catalyzed by a Rlg1-mediated healing step prior to degradation catalyzed by Xrn1. Assessment of other nuclear or cytoplasmic 5′-to-3′ and 3′-to-5′ decay pathways provides evidence that intron turnover is a 5′-to-3′ process in the cytoplasm (Fig. 5).

Figure 5.

Mechanisms for tRNA splicing, ligation, and intron degradation in yeast. tRNA second structure is illustrated with the following color scheme of circles: exons in light blue, anti-codons in red, introns in yellow, modified nucleotides in pink, and CCA in orange. After nuclear export, end-matured intron-containing tRNAs are spliced by SEN. The CPDase and polynucleotide kinase activity of Rlg1 catalyze healing of the excised ends, and ligase activity catalyzes the sealing of two exons. During intron turnover, the 5′ end of newly spliced introns is phosphorylated by Rlg1, although whether the 3′ end is subjected to Rlg1 CPDase activity is unclear. After Rlg1-mediated phosphorylation, introns are recognized and degraded by Xrn1.

Materials and methods

Yeast strains

S. cerevisiae strains in the MATa/BY4741 background (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0) and/or the MATα/BY4742 background (MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0) were used: wild type, xrn1Δ, rai1Δ, dxo1Δ, ski2Δ, ski3Δ, ski7Δ, and ski8Δ (Open Biosystems). rlg1-4 and rlg1-10 ts strains were previously described (Phizicky et al. 1992). The XRN1 gene was deleted from the rlg1-4 strain by gene replacement with a bacterial hygromycin B resistance gene (Hph) to construct rlg1-4 xrn1Δ. The ts rat1-1 strain was a gift from Dr. Charlie Boone (Li et al. 2011).

CIP treatment

Twenty micrograms of small RNAs was treated with CIP (New England Biolabs) for 1 h at 37°C according to the manufacturer’s protocol. The reaction was terminated by phenol extraction and ethanol precipitation.

TEX treatment

Ten micrograms of small RNAs was treated with TEX (Epicentre) for 1 h at 30°C according to the manufacturer’s protocol. The reaction was terminated by phenol extraction and ethanol precipitation.

Northern analysis

Small RNAs were isolated by phenol extraction from cultures grown at the permissive temperature (23°C) to 0.4 OD600. The ts mutants were shifted for 2 h to the nonpermissive temperature (37°C) before RNA extraction. Seven micrograms of small RNAs was separated by electrophoresis through a 10% polyacrylamide gel and transferred onto a Hybond N+ membrane. tRNAs were detected using DIG-labeled probes as described (Wu et al. 2013). The membranes were exposed to a Lumi-Imager (Boehringer Mannheim), and the signal intensity of each band was quantified as absolute integrated Boehringer light units (BLU) using LumiAnalyst version 3.0 software.

Oligonucleotides

The sequences of the oligonucleotides used are provided in the Supplemental Material.