Genetic bypass of essential RNA repair enzymes in budding yeast.

RNA repair enzymes catalyze rejoining of an RNA molecule after cleavage of phosphodiester linkages. RNA repair in budding yeast is catalyzed by two separate enzymes that process tRNA exons during their splicing and HAC1 mRNA exons during activation of the unfolded protein response (UPR). The RNA ligase Trl1 joins 2',3'-cyclic phosphate and 5'-hydroxyl RNA fragments, creating a phosphodiester linkage with a 2'-phosphate at the junction. The 2'-phosphate is removed by the 2'-phosphotransferase Tpt1. We bypassed the essential functions of TRL1 and TPT1 in budding yeast by expressing "prespliced," intronless versions of the 10 normally intron-containing tRNAs, indicating this repair pathway does not have additional essential functions. Consistent with previous studies, expression of intronless tRNAs failed to rescue the growth of cells with deletions in components of the SEN complex, implying an additional essential role for the splicing endonuclease. The trl1Δ and tpt1Δ mutants accumulate tRNA and HAC1 splicing intermediates indicative of RNA repair defects and are hypersensitive to drugs that inhibit translation. Failure to induce the unfolded protein response in trl1Δ cells grown with tunicamycin is lethal owing to their inability to ligate HAC1 after its cleavage by Ire1. In contrast, tpt1Δ mutants grow in the presence of tunicamycin despite reduced accumulation of spliced HAC1 mRNA. We optimized a PCR-based method to detect RNA 2'-phosphate modifications and show they are present on ligated HAC1 mRNA. These RNA repair mutants enable new studies of the role of RNA repair in cellular physiology.

INTRODUCTION

RNA repair enzymes catalyze rejoining of an RNA molecule after cleavage of phosphodiester linkages. RNA repair is carried out by enzymes that prepare the RNA termini for ligation, ligate the termini, and “clean-up” the ligated product when required. RNA repair systems are present in all domains of life and catalyze rejoining of the 5′-hydroxyl and 2′,3′-cyclic phosphate products of cleavage. RNA repair catalyzes the maturation of endogenous tRNAs and mRNAs (Schwer et al. 2004), and can also counteract the action of endonucleases during inter-organismal conflicts (Amitsur et al. 1987; Nandakumar et al. 2008; Burroughs and Aravind 2016). E. coli RtcB can repair ribosomal RNA cleaved during stress by the endonuclease MazF, thereby reversing ribosomal heterogeneity and restoring translational activity to MazF-processed ribosomes (Temmel et al. 2016). The Pnkp–Hen1 RNA repair complex, which is present in more than 250 bacterial species, combines the enzymatic activities of the bacteriophage T4 RNA repair system with the Hen1 methyltransferase, which installs a 2′--methyl group to “immunize” repaired RNA molecules against future cleavage by the same ribotoxin (Wang et al. 2015). In light of these and other recent developments in our understanding of the functional and genetic diversity of RNA ligases (Burroughs and Aravind 2016), we developed a genetic bypass of the Saccharomyces cerevisiae RNA repair system, enabling direct study of the roles of RNA repair proteins in budding yeast.

In budding yeast, S. cerevisiae, after tRNA intron excision, ligation of the exons is carried out in successive steps (Phizicky et al. 1986). First, the splicing endonuclease (SEN) complex cleaves the intron at two sites, creating two exon products, one with a 2′,3′-cyclic phosphate terminus and one with a 5′-hydroxyl terminus, and an intron product with 5′-hydroxyl and 2′,3′-cyclic phosphate termini that is degraded (Wu and Hopper 2014). In a second step, the two exons are ligated by the RNA ligase Trl1. Trl1 joins substrates with 2′,3′-cyclic phosphate and 5′-hydroxyl termini via: (i) conversion of the 2′,3′-cyclic phosphate to a 2′-phosphate/3′-hydroxyl; (ii) phosphorylation of the 5′-OH; (iii) adenylylation of the 5′-phosphate, and (iv) nucleophilic attack on the adenylylate by the 3′-OH, producing AMP and a new 5′ → 3′ phosphodiester linkage with a 2′-phosphate (Greer et al. 1983). The 2′-phosphate left at the ligation junction is removed by the NAD-dependent 2′-phosphotransferase Tpt1, creating a canonical 5′ → 3′ phosphodiester linkage (Fig. 1A;McCraith and Phizicky 1990; Culver et al. 1993).

FIGURE 1.

Genetic bypass of essential components of tRNA splicing with intronless tRNAs. (A) Functions of tRNA splicing enzymes. Introns from pre-tRNAs are removed by the SEN complex, which contains the Sen2 and Sen34 endonucleases that cleave the 5′ and 3′ splice sites, respectively, as well as Sen15 and Sen54. Following cleavage, the tRNA exons are ligated by the multifunctional RNA ligase Trl1, producing a ligated tRNA with a 2′-phosphate at the splice junction. Tpt1 removes the 2′-phosphate in an NAD+-dependent reaction, producing ADP-ribose-1″,2″ cyclic phosphate and nicotinamide. (B) Schematic of a plasmid encoding the 10 S. cerevisiae tRNAs in intronless form. The tRNAs are expressed from a high-copy 2µ TRP1 plasmid containing a SUP4 promoter and terminated by the RPR1 terminator. (C) Genetic bypass of RNA repair mutants by expression of intronless tRNAs (“10×-tRNA” plasmid). Strains expressing a URA3 covering plasmid were transformed with an empty vector (TRP1 2µ) or a high-copy plasmid encoding intronless tRNAs (B). URA+ TRP+ colonies (top row) were selected and struck on FOA media (bottom row), which selects against cells with the covering plasmid, to assess intronless tRNA-mediated bypass. Plates were photographed after 5–7 d of incubation at 30°C. Intronless tRNAs complement deletion of TRL1 and TPT1 but do not rescue deletion of SEN components (bottom right). (D) Cells with genomic deletions of SEN genes (SEN2, SEN15, SEN34, SEN54) and a single-copy URA3 plasmid expressing the deleted gene were individually transformed with high-copy LEU2 plasmids containing the genomic locus of each of the SEN genes. LEU+ colonies were plated on −LEU media (top), and FOA media (bottom), to select against the covering plasmid. Only those plasmids that contain the cognate genomic locus of the deleted SEN gene were able to rescue growth on FOA (bottom).

Trl1 has a second known role in the cell: activating the unfolded protein response (UPR). Trl1 ligates the two exons of the HAC1 mRNA after cleavage by Ire1, activating the UPR (Gonzalez et al. 1999). Ire1 excises an intron from the HAC1 pre-mRNA, and Trl1 subsequently ligates the HAC1 exons together, enabling its translation into a transcription factor that localizes to the nucleus and drives transcription of hundreds of stress response genes (Sidrauski et al. 1996). Yeast cells that lack Trl1 and Tpt1 and that express RNA repair enzymes from T4 bacteriophage are viable, but they exhibit low-fidelity HAC1 mRNA cleavage and ligation, suggesting the 2′-phosphate/3′-hydroxyl terminus produced by the cyclic phosphodiesterase domain of Trl1 directs precise ligation (Schwer et al. 2004). Trpt1, the mammalian 2′-phosphotransferase, was shown to be dispensable for UPR activation in mammals (Harding et al. 2008). However, subsequent studies showed that the HSPC117/RtcB RNA ligase—which does not create 2′-phosphate ligation products (Chakravarty et al. 2012)—activates the mammalian UPR (Lu et al. 2014), explaining why Trpt1 2′-phosphotransferase activity is dispensable (Harding et al. 2008). The role of Tpt1 in budding yeast in the UPR has not been previously explored.

Using a genetic bypass strategy, RNA repair was previously shown to be essential only for tRNA splicing in C. elegans (Kosmaczewski et al. 2014) and in trypanosomes (Lopes et al. 2016). Using a similar strategy, we designed and tested a genetic bypass for deletion of the essential RNA repair enzymes Trl1 and Tpt1 in budding yeast and show that rescued trl1Δ and tpt1Δ cells (the “RNA repair mutants”) have unique phenotypes for both tRNA splicing and HAC1 mRNA splicing during the unfolded protein response.

RESULTS AND DISCUSSION

Genetic bypass of essential RNA repair genes in budding yeast

Ten S. cerevisiae tRNA isodecoders are encoded with introns (Chan and Lowe 2009), which must be accurately processed for cells to faithfully translate messenger RNA (Hopper 2013). We adapted a strategy first documented in C. elegans (Kosmaczewski et al. 2014) to express these 10 tRNAs in “prespliced” form (Fig. 1B, the 10×-tRNA plasmid) and found that expression of these intronless tRNAs rescues the growth of cells with deletions in the essential genes TRL1 and TPT1 (Figs. 1C, 2A). This result is consistent with previous findings that TPT1 is essential only in the context of the generation of 2′-phosphorylated tRNAs by Trl1 (Schwer et al. 2004) and that a growth defect caused by TRL1 knockdown in the trypanosome Trypanosoma brucei is rescued by expressing intronless tRNATyr (Lopes et al. 2016).

FIGURE 2.

Growth phenotype of RNA repair mutants. (A) RNA repair mutants rescued by prespliced tRNAs expressed from the (10×-tRNA TRP1 2µ) plasmid grow on -TRP and FOA media, but not on -URA media. Plates were photographed after 3 d of growth at 30°C. (B) Growth of RNA repair mutants is temperature-sensitive. Cells were serially diluted, spotted on YPD media, and incubated at the indicated temperatures. Wild-type and complemented trl1Δ (TRL1) and tpt1Δ (TPT1) cells grow at all temperatures, whereas bypassed trlΔ and tpt1Δ cells fail to grow at 37°C. (C) Sensitivity of RNA repair mutants to translational inhibitors. Cells were serially diluted and spotted on YPD media or YPD media supplemented with sublethal doses of the translational inhibitors anisomycin, hygromycin B, and cycloheximide at the concentrations indicated. Wild-type cells are viable under all conditions, whereas RNA repair mutants fail to grow in the presence of inhibitors of translation. (D) Deletion of XRN1 does not suppress the temperature-sensitive phenotype of trl1Δ rescued with prespliced tRNAs. All plates were photographed after 3–4 d of incubation at 30°C or 37°C.

The SEN complex cleaves introns from pre-tRNAs (Fig. 1A), and each component of the heterotetrameric endonuclease (Sen2, Sen15, Sen34, and Sen54) is essential for growth (Trotta et al. 1997). We created cells with single genomic deletions in each of the SEN2, SEN15, SEN34, and SEN54 genes and found that in each case cells were only viable when complemented by plasmid-mediated expression of the cognate SEN gene (Fig. 1D). We expressed the 10 prespliced tRNAs in cells containing deletions of each of the SEN genes and found that this failed to rescue deletion of any component of the SEN complex (Fig. 1C), consistent with this complex having an essential function beyond splicing of pre-tRNA (Dhungel and Hopper 2012). The gene encoding the only other known substrate of the SEN complex, CBP1, is non-essential (Tsuboi et al. 2015), suggesting that the SEN complex has another essential function, possibly to process unknown RNA substrates.

RNA repair mutants have defects in translation

Genetically bypassed trl1Δ and tpt1Δ cells share common growth phenotypes that may reflect translational defects in the absence of RNA repair or incomplete rescue by intronless tRNAs. We found that rescued trl1Δ and tpt1Δ cells grow slowly at 30°C and fail to grow at 37°C on rich media (Fig. 2B). These growth defects could be due to differences in the abundance or functionality of prespliced tRNAs or due to the accumulation of endogenous intermediates of tRNA splicing. Alterations in tRNA pools and post-transcriptional modifications of specific tRNA species have been implicated in regulating translational dynamics in both prokaryotes and eukaryotes (Kirchner and Ignatova 2015), providing a possible explanation for the temperature sensitivity. Rescued trl1Δ cells should accumulate unligated pre-tRNA half-molecules that have been cut by the SEN complex, whereas tpt1Δ cells should accumulate spliced and ligated pre-tRNAs that retain a 2′-phosphate at the ligation junction (Fig. 1A). Accumulation of these molecular species could have distinct effects on translation. However, both RNA repair mutants exhibit broad sensitivity to sub-lethal doses of translational inhibitors (Fig. 2C).

It is possible that the prespliced tRNAs used to rescue trl1Δ and tpt1Δ mutants are suboptimal due to defects in their expression levels, amino acid charging, or post-transcriptional modification. S. cerevisiae tRNAs have an average of 13 post-transcriptional modifications per tRNA species (Phizicky and Hopper 2010), and the diverse set of modifications added to tRNAs can impact maintenance of reading frame, translational fidelity, and tRNA stability (Hopper 2013). The rapid tRNA decay pathway (RTD) comprises a quality control mechanism that degrades aberrant tRNA molecules via the 5′ → 3′ exonucleases Rat1 and Xrn1. Structural stability is a major determinant of RTD substrates: Both the introduction of destabilizing mutations in tRNA sequences and the deletion of modification enzymes that act to enhance tRNA stability cause temperature sensitivity in budding yeast. However, growth at 37°C can be rescued by compensatory mutations that restore tRNA structural stability or by deletion of the Xrn1 exonuclease (Whipple et al. 2011). Based on these prior observations, we tested whether RTD is responsible for temperature sensitivity by deleting XRN1 in a rescued trl1Δ mutant and found that the double deletion, trl1Δ xrn1Δ, grows no better at 37°C than the single trl1Δ mutant (Fig. 2D), suggesting that the temperature sensitivity of RNA repair mutants is not a consequence of RTD.

Whereas our results suggest that prespliced tRNAs are not rapidly degraded due to structural instability, these intronless tRNA species may still be hypomodified or otherwise suboptimal. Previous studies of tRNA introns identified two types of post-transcriptional modifications that are only added at the pre-tRNA stage, when an intron is still present. Site-specific introduction of 5-methylcytidine (m5C) or pseudouridine in the anticodon or at position 40 is dependent on the presence of the intron, and these intron-dependent modifications occur in four of the 10 intron-containing tRNAs in S. cerevisiae (Table 1; Grosjean et al. 1997). In particular, incorporation of m5C at the wobble position in the pre-tRNALeu(CAA) is catalyzed by the tRNA-specific methyltransferase Trm4, and disruption of TRM4 causes sensitivity to the antibiotic paromomycin, an aminoglycoside that interferes with translational fidelity (Wu et al. 1998). This observation is consistent with our finding that RNA repair mutants rescued with prespliced tRNAs fail to grow on rich media containing low doses of the translational inhibitors anisomycin, hygromycin B, or cycloheximide (Fig. 2C). Thus, this phenotype could be indicative of hypomodification of one or more of the intronless tRNAs from the construct. Alternatively, it is possible that bypass of trl1Δ and tpt1Δ with intronless tRNAs provides a complete rescue of tRNA splicing, but causes reduced fitness due to an additional but nonessential role of RNA repair related to translation.

TABLE 1.

Intron-containing tRNA copy number (Chan and Lowe 2009) and intron-dependent modifications (Grosjean et al. 1997) in S. cerevisiae

RNA repair mutants accumulate intermediates and products of tRNA splicing

To evaluate the ability of cells lacking RNA repair to produce mature tRNAs, we analyzed the processing of tRNAIle(UAU) by northern blot using total RNA from rescued trl1Δ and tpt1Δ cells, unshuffled control strains, wild-type, and xrn1Δ mutants. Wild-type, trl1Δ (TRL1), and tpt1Δ (TPT1) cells have expected tRNA processing intermediates, including primary transcript (145 nt), pre-tRNA with 5′- and 3′-processing by RNaseP, Rex1, RNase Z, and Lph1 (136 nt), mature tRNA at 76 nt, excised intron at 60 nt, and 5′-exon at 38 nt (Fig. 3A,B). Total RNA from trl1Δ mutants complemented with intronless tRNAs contains a band of the same size as mature tRNA from wild-type controls, and this tRNA must arise from the intronless tRNA construct because these cells cannot ligate exons processed from genomic intron-containing tRNAs (Fig. 3A, lane 4). Cells that contain the prespliced tRNA plasmid exhibit high molecular weight species to which the northern probe hybridizes, which are likely tandem tRNA transcripts that are not completely processed. This result is consistent with the construct functioning as a single transcriptional unit, owing to its design with a single SUP4 Pol III promoter and RPR1 Pol III terminator; additionally, the intronless tRNA genes may also be transcribed individually via internal promoters (Galli et al. 1981).

FIGURE 3.

tRNA processing phenotypes of RNA repair mutants. Diagrams of tRNA primary structure, annotations, and probe locations are depicted next to each northern blot. SCR1 loading control blots are below, and were the basis of loading for each lane. (A) Northern blot using probe that hybridizes to the 5′-exon of tRNAIle(UAU) identifies preprocessed tRNA intermediates derived from primary transcripts of intronless tRNAs (lanes 4,5, and 7, bracket annotation), as well as intermediates arising from the intron-containing endogenous tRNAs (at ∼150 nt). Each strain produces mature tRNAs (∼76 nt). The trl1Δ mutant (lane 4) is unable to ligate tRNAIle(UAU) exons arising from chromosomal copies of the gene, and thus mature tRNAs in these cells occur from processing of the intronless tRNA transcript. A band at ∼32 nt in lanes 4 and 7 is likely a product of SEN cleavage of chromosomally encoded intron-containing tRNAs that are not re-ligated. (B) Northern blot with probe to tRNAIle(UAU) intron shows increased accumulation of the intron in trl1Δ mutants (lanes 4 and 7), as well as in xrn1Δ and in double deletion trl1Δ xrn1Δ mutants (lanes 6 and 7, respectively). A band at ∼90 nt putatively represents 5′-exon/intron (question mark). Densitometry quantifications of intron signal relative to wild-type (lane 1) and normalized to SCR1 signal are displayed below lane numbers. (C) Northern blot with probe to the 5′-exon of tRNATyr(GUA), a tRNA with a copy number of eight in the budding yeast genome, reveals mature tRNA bands at approximately the same intensity across wild-type (lane 1) and repair mutant strains (lanes 4 and 5). (D) Northern blot with probe to the 5′-exon of tRNAPhe(GAA), a tRNA with a copy number of 10 in the genome, exhibits approximately equal density of mature tRNA band intensity in wild-type (lane 1) and deletions in RNA repair genes (lanes 4 and 5). (E) Northern blot with probe to the 5′-exon of tRNALeu(CAA), a tRNA with a copy number of 10 in the genome, shows a decrease in mature tRNA band intensity in the trl1Δ mutant (lane 4), and also in the tpt1Δ mutant (lane 5), when compared to wild-type and covered strains (lanes 1–3). (F) Northern blot with probe to the 5′-exon of tRNAPro(UGG), a tRNA with a copy number of 10 in the genome, reveals a decrease in mature tRNA band intensity in both trl1Δ and tpt1Δ RNA repair mutants (lanes 4 and 5, respectively). All lanes show a doublet of bands, with the upper band being consistent in size with mature tRNAPro(UGG), and the lower band is annotated with an asterisk. Lane 5 displays a strong tpt1Δ-dependent band, annotated with a question mark.

The S. cerevisiae genome contains tRNA genes at varying copy numbers (Table 1; Chan and Lowe 2009), raising the possibility that rescue by a high-copy plasmid expressing intronless tRNAs could alter the levels of specific tRNA isodecoders. To assess the ability of the genetic bypass to produce quantities of tRNAs similar to those found in wild-type cells, we performed northern blotting with probes hybridizing to the 5′-exons of tRNAs with varying genomic copy number. In the case of low copy number, tRNAIle(UAU) is encoded at two copies per genome, and the intronless tRNA plasmid produces quantities of mature tRNAIle(UAU) similar to wild-type cells (Fig. 3A, lane 1 versus 4 and 5). In contrast, the isodecoders tRNATyr(GUA), tRNALeu(CAA), tRNAPhe(GAA), and tRNAPro(UGG) are encoded in the genome at relatively high copy number (8, 10, 10, and 10 copies, respectively) (Table 1). The intronless tRNA plasmid produces tRNATyr(GUA) and tRNAPhe(GAA) at abundances similar to wild-type (Fig. 3C,D, lane 1 versus 4 and 5), but the plasmid fails to produce equivalent amounts of tRNALeu(CAA) and tRNAPro(UGG) (Fig. 3E,F, lane 1 versus 4 & 5). These results show that there is not a monotonic relationship between copy number in the genome and the ability of the intronless tRNA plasmid to produce wild-type quantities of each tRNA. However, these results do not comment on the modification status of intronless tRNAs. Furthermore, the discrepancy in quantities of tRNALeu(CAA) and tRNAPro(UGG) produced by the intronless tRNA plasmid could contribute to the slow growth phenotype of the RNA repair mutants (Fig. 2).

The northern blot for tRNAPro(UGG) (Fig. 3F) reveals several additional features of interest that appear to be unique to this tRNA isodecoder. Doublet bands of spliced tRNAPro(UGG), which we also detected in all strains tested, have been previously observed by denaturing polyacrylamide gel (Winey et al. 1986). The upper of these two bands has a length consistent with the 75 nt mature tRNAPro(UGG) (Chan and Lowe 2009). The lower band, ∼10 nt shorter, cannot be explained as an intron-containing tRNA half or processing intermediate, as the intron for tRNAPro(UGG) is 31 nt and a processing intermediate that contained the intron and both exons would migrate at ∼103 nt. Because these intermediates accumulate in the RNA ligase mutant—and ligase is required for intron splicing—the prespliced tRNA construct is also competent to produce both species (i.e., the putative processing intermediate marked with an asterisk and mature tRNA). In addition, both tpt1Δ (TPT1) and tpt1Δ cells (Fig. 3F, lanes 3 and 5) contain a third tRNAPro(UGG) band (indicated by a question mark) that accumulates to high levels in tpt1Δ cells and is not seen in other genotypes, which may correspond to an additional tRNA species resolved via 2D PAGE analysis of tRNA from a tpt1 conditional mutant (Spinelli et al. 1997). It is possible that this third band is derived from endogenous spliced tRNAPro(UGG) that retains a 2′-phosphate at the splice junction in the absence of TPT1; however, the precise identity of these three species of tRNAPro(UGG) remains unresolved.

In addition to analysis of mature tRNA production, we further investigated the fate of the introns from endogenous tRNA genes in the context of the RNA repair mutants. Northern blot analysis of the tRNAIle(UAU) intron recapitulated the finding that decay of excised tRNA introns requires both Xrn1 and Trl1 (Wu and Hopper 2014). Bypassed trl1Δ cells show an increase in levels of tRNA intron (47-fold compared to wild-type, normalized to SCR1 levels), as do xrn1Δ cells (6.3-fold) and double mutant trl1Δ xrn1Δ cells (22-fold) (Fig. 3B, lanes 4,6, and 7). These observations are in line with kinase-mediated decay of cleaved tRNA introns, in which the 5′-RNA kinase activity of Trl1 (Fig. 1A) is required to phosphorylate the 5′-hydroxyl intron products of SEN cleavage to enable their 5′ → 3′ exonucleolytic decay by Xrn1, which specifically degrades 5′-phosphorylated RNA substrates (Stevens 2001).

RNA repair mutants have defects in unfolded protein response activation

Induction of the unfolded protein response (UPR) requires the ligation of two exons after intron excision from the HAC1 mRNA (Gonzalez et al. 1999). We tested cells lacking RNA repair for their ability to activate the UPR. Bypassed trl1Δ mutants fail to grow on media containing the UPR-inducing drug tunicamycin (Fig. 4A), confirming that Trl1 is required for UPR activation in response to protein folding stress (Sidrauski et al. 1996). In contrast, tpt1Δ mutants grow equally well in the presence and absence of tunicamycin (Fig. 4A), suggesting that the 2′-phosphate remaining on HAC1 mRNA after ligation in tpt1Δ cells does not interfere with its translation or that partial HAC1 translation is sufficient for UPR activation (see below). Cells with Hac1-FLAG (in wild-type, trl1Δ, and tpt1Δ backgrounds) were also spotted onto tunicamycin containing media and had similar growth to cells with untagged Hac1, confirming that Hac1-FLAG can function in the UPR (Fig. 4A).

FIGURE 4.

UPR-related phenotypes of RNA repair mutants. (A) Growth assay of RNA repair mutants on UPR-inducing media. Yeast cells (wild-type, trl1Δ, and tpt1Δ) were serially diluted and spotted onto rich media (YPD) and tunicamycin-containing media (80 ng/mL tunicamycin) to induce the UPR. Plates were imaged after 3 d of growth at 30°C. Wild-type and tpt1Δ cell growth is unaffected by tunicamycin, whereas trl1Δ cells fail to grow on media containing tunicamycin. Serial dilution growth assays for cells of the same genetic background as above, but with C-terminal FLAG tags on Hac1, are shown below. (B) Analysis of HAC1 splicing in RNA repair mutants. Total RNA from untreated and tunicamycin-treated wild-type, trl1Δ, and tpt1Δ cells was analyzed by RT-PCR using primers specific for HAC1, producing products at 499 bp (unspliced HAC1) and 247 bp (spliced HAC1). A no-template (NT) control is shown in lane 7. The proportion of spliced HAC1 upon tunicamycin treatment increases in wild-type (lanes 1, and 2) and tpt1Δ cells (lanes 5 and 6), and spliced HAC1 is visible in RNA from both cells (lanes 2 and 6, 247 bp). Spliced HAC1 is undetectable in trl1Δ cells (lane 4) upon tunicamycin treatment (lanes 3 and 4), owing to the inability of trl1Δ cells to ligate HAC1 exons. Asterisk marks an unknown PCR product dependent on tunicamycin treatment. Cells with C-terminal FLAG tags of Hac1 were also analyzed for splicing in the same manner (below) with a no-template (NT) control in lane 9. Reactions lacking reverse transcriptase (RT−) were negative for amplification (data not shown). (C) Hac1 protein levels in RNA repair mutants. Whole cell lysates were prepared from wild-type, trl1Δ, and tpt1Δ cells expressing C-terminal Hac1-FLAG and grown in the presence and absence of tunicamycin. Lysates were analyzed by SDS–PAGE and nitrocellulose transfer followed by Ponceau S staining and cross-reaction with anti-FLAG and anti-GAPDH antibodies. Scale to the left is nominal molecular mass of a protein ladder (kDa); the expected mass of Hac1-FLAG is 31 kDa. Hac1-FLAG is detected in wild-type cells upon tunicamycin addition, but is undetectable in trl1Δ and tpt1Δ cells. (D) Induction of the UPR-responsive KAR2 gene in RNA repair mutants. Amounts of KAR2 mRNA (normalized to PGK1 mRNA abundance) were measured by RT-qPCR in wild-type, trl1Δ, and tpt1Δ cells in the presence and absence of tunicamycin. Error bars are 95% confidence intervals, n = 3. Relative abundance of KAR2 mRNA increased 20-fold in wild-type cells treated with tunicamycin, whereas the corresponding levels of KAR2 did not increase in trl1Δ and increased 1.4-fold in tpt1Δ cells. (E) Detection of ligated and 2′-phosphorylated HAC1 mRNA. Total RNA was treated with calf intestinal phosphatase (CIP) to remove 2′-phosphates (diagram) and reverse-transcribed using HAC1-specific primer under high (500 µM) concentrations of dNTPs. The cDNA products were PCR amplified, yielding products for unspliced (456 bp) and spliced (204 bp) HAC1 mRNA. Using high dNTP concentrations, we find that splicing of HAC1 in wild-type cells increases upon tunicamycin treatment (compare lanes 1 and 2 versus 3 and 4), similar to B, but is unaffected by CIP treatment (compare lanes 1 versus 2, and 3 versus 4). Likewise, we find that spliced HAC1 mRNA in tpt1Δ cells increases in response to tunicamycin (cf. lanes 9 and 10 versus lanes 11 and 12), albeit to a lesser extent than wild-type cells, and is unaffected by CIP treatment (cf. lanes 9 versus 10 and lanes 11 versus 12). An asterisk marks an unknown PCR product dependent on tunicamycin treatment. (F) Detection of ligated and 2′-phosphorylated HAC1 mRNA. Using low (1 µM) dNTP concentrations, we find that spliced HAC1 is preferentially amplified in wild-type cells over unspliced HAC1 (lanes 1–4). The abundance of spliced HAC1 mRNA from wild-type cells increases in response to tunicamycin but is unaffected by CIP treatment (cf. lanes 1 versus 2 and 3 versus 4). RT-PCR analysis of HAC1 mRNA from tpt1Δ cells reverse transcribed under low dNTP concentrations shows both unspliced and spliced forms of HAC1 mRNA, and spliced HAC1 mRNA increases in response to tunicamycin treatment (cf. lanes 10 and 12). However, in contrast to wild-type, amplification of spliced HAC1 mRNA from tpt1Δ is strongly dependent on prior treatment with CIP. In the absence of tunicamycin and CIP treatment, spliced HAC1 mRNA is undetectable, whereas treatment with CIP enables reverse transcription (cf. lane 9 to lane 11; see panel with enhanced contrast to the right). Similarly, the abundance of spliced HAC1 mRNA from tpt1Δ cells increases in response to tunicamycin, and its abundance is further increased upon CIP treatment (lane 11 versus 12).

We corroborated the growth assay by analyzing HAC1 splicing in trl1Δ and tpt1Δ cells using RT-PCR (Fig. 4B). The wild-type strain catalyzes cleavage and ligation of HAC1 upon UPR stimulation with tunicamycin, whereas the trl1Δ mutant exhibits no detectable spliced HAC1 and a reduction in unspliced HAC1. The decrease in unspliced HAC1 observed in lanes 4 and 6 (Fig. 4B) when compared to their tunicamycin-null control lanes can be explained by two facts: First, both strains express Ire1, the endonuclease responsible for cleaving HAC1 mRNA prior to ligation; second, the Hac1 protein is a transcription factor that activates transcription of the HAC1 gene, creating a positive feedback loop to sustain UPR activation (Ogawa and Mori 2004). When Ire1 cleaves HAC1 mRNA, but Trl1 is unavailable for ligation, Hac1 protein is not translated and cells fail to activate the positive feedback loop. Then, without the newly transcribed HAC1 mRNA, the remaining pool of HAC1 is efficiently cleaved so that the cDNAs synthesized do not contain both PCR priming sites and thus fail to amplify in this PCR assay. These simultaneous deficits could lead to a decrease in unspliced HAC1 in RNA repair mutants treated with tunicamycin when compared to their null-treatment controls.

We found that tpt1Δ cells accumulate spliced HAC1, but to a lesser extent than wild-type, suggesting that the 2′-phosphorylated products of ligation interfere with processing or stability of the HAC1 mRNA (Fig. 4B, compare lane 2 versus 6). Cells with Hac1-FLAG spliced HAC1 mRNA with the same pattern as their untagged controls (Fig. 4B). An immunoblot for Hac1-FLAG (Fig. 4C) showed that wild-type cells showed no detectable Hac1 until the UPR was induced with tunicamycin, but Hac1-FLAG could not be detected in lysate from trl1Δ or tpt1Δ cells.

In addition, trl1Δ cells failed to increase expression of the chaperone KAR2 (Fig. 4D), a representative UPR-responsive gene strongly induced by Hac1 transcription factor protein (Mori et al. 1992; Nikawa et al. 1996), indicating that trl1Δ cells fail to induce UPR-responsive genes. Consistent with reduced spliced HAC1 mRNA accumulation (Fig. 4B), tpt1Δ mutants also show reduced expression of KAR2 (Fig. 4D). However, this reduced degree of UPR induction, as observed by HAC1 mRNA splicing and KAR2 expression, is nonetheless sufficient for growth of tpt1Δ cells in the presence of tunicamycin (Fig. 4A). Despite accumulation of spliced HAC1 mRNA (Fig. 4B), accumulation of KAR2 mRNA (Fig. 4D) and growth on tunicamycin (Fig. 4A), we were unable to detect Hac1-FLAG by immunoblot from tpt1Δ cells (Fig. 4C). It is possible that the 2′-phosphate allows only partial translation of the spliced HAC1 mRNA 5′-exon, leading to production of Hac1 N-terminal bZIP domain and low-level UPR activation, but precluding translation of the C-terminal FLAG epitope. Consistent with this possibility, translation of unspliced HAC1 mRNA is sufficient to restore growth on tunicamycin-containing media in the absence of the Duh1 ubiquitin ligase (Di Santo et al. 2016).

We found that tpt1Δ mutants accumulate less spliced HAC1, less Hac1 protein, and less KAR2 mRNA than wild-type yeast despite having a functional RNA ligase. To determine whether ligated HAC1 mRNA retains a 2′-phosphate at the ligation junction in tpt1Δ cells, we adapted a reverse-transcriptase-based method from previous reports that show 2′-phosphates inhibit reverse transcription (Schutz et al. 2010; Dhungel and Hopper 2012). We used calf intestine phosphatase (CIP) treatment, which removes 2′-phosphates from RNA (Fig. 4E schematic; McCraith and Phizicky 1990), and RT-PCR to test whether 2′-phosphates were present in ligated HAC1 mRNA. Using low concentrations of dNTPs (1 µM) (Fig. 4F) and a reverse transcription primer adjacent to the expected site of 2′-phosphorylation, CIP treatment substantially enhanced detection of spliced HAC1 mRNA in tpt1Δ cells, but levels of amplified HAC1 in RNA from wild-type cells were unaffected, indicating that a 2′-phosphate remains at the ligation junction in the HAC1 mRNA from tpt1Δ cells. Reverse transcription reactions with typical dNTP concentrations (500 µM) (Fig. 4E) restored amplification of unspliced HAC1 mRNA. We surmise that reduced dNTP concentrations lower the processivity of reverse transcriptase, favoring synthesis of shorter cDNA substrates. The presence of spliced HAC1 mRNA in tpt1Δ cells in the absence of CIP treatment (Fig. 4F, lane 11) could indicate that the assay is not quantitatively sensitive to sites of 2′-phosphorylation or that not all molecules of ligated HAC1 mRNA retain 2′-phosphates despite the absence of Tpt1. In any case, 2′-phosphorylated RNA reduces cDNA synthesis under these conditions, enabling their detection via this RT-PCR strategy.

Conclusions

We showed the one essential function of RNA repair in budding yeast is catalyzing the ligation of tRNA halves resulting from splicing; however, the reduced-growth phenotypes of RNA repair mutants caused by various translational inhibitors, a UPR-inducing drug, and elevated temperature suggest either that the prespliced tRNAs do not function at the same efficiency/abundance as endogenously encoded tRNAs or that RNA repair is generally helpful to cells, albeit not essential. We demonstrated that “prespliced” tRNA genes are transcribed and/or processed to a size consistent with wild-type mature tRNA. Furthermore, we provided evidence of dysfunction of the UPR in both the ligase mutant and the 2′-phosphotransferase mutant, suggesting that 2′-phosphorylated HAC1 mRNA contributes to UPR induction dynamics. Lastly, we showed that RNA retaining 2′-phosphate residues is detectable when the enzyme responsible for removing them is deleted.

These new genetic reagents enable studies to identify other targets of RNA repair. For example, cells lacking the Tpt1 2′-phosphotransferase accumulate 2′-phosphates at sites of RNA ligation, enabling their possible identification by methods to identify 2′-O-modifications by RNA-seq (Birkedal et al. 2015) or by affinity purification by tagging with mutant Tpt1 enzyme and biotin-NAD (Steiger et al. 2005). These approaches address the diversity of RNA repair in biology and are the subject of further study.

MATERIALS AND METHODS

General methods

Saccharomyces cerevisiae W303 strains (Table 2) were cultured in YPD and synthetic “drop-out” media (Sherman 2002) for experimental cultures and for plasmid selection, respectively. Single-copy URA3 SEN plasmids were created by recombining SEN-containing genomic fragments from the yeast tiling collection (Jones et al. 2008) into Advanced Gateway plasmids (Alberti et al. 2007) via Gateway LR reactions (Life Technologies). Plasmid counterselection was performed on synthetic complete solid media containing 5-fluoroorotic acid (FOA, US Biologicals) (1 mg/mL). Yeast transformations were performed using a Lithium Acetate PEG-3350 Sheared Salmon Sperm protocol (Gietz and Schiestl 2007). Genotypes of yeast were confirmed by PCR using forward and reverse primers flanking the disrupted locus and outward-facing primers within the disruption cassette. Success of FOA plasmid counterselections was confirmed using RT-PCR for the shuffled gene in DNased total RNA isolated from shuffled strains and positive controls.

TABLE 2.

Strain numbers and genotypes

Experiments with tunicamycin (Sigma) treatment were performed at final concentrations of 0.08 µg/mL in solid media and at 2.5 µg/mL in liquid media, with negative controls using DMSO (solvent). Cultures were incubated at 30°C unless otherwise indicated. UPR inductions ± tunicamycin were carried out for 2 h during exponential phase growth. For growth assays, cells were grown in YPD overnight and liquid cultures were normalized to OD600 = 0.2 before plating 10-fold serial dilutions on indicated media. For translation inhibition assays, dilutions were spotted onto YPD agar supplemented with 2 µg/mL anisomycin (Sigma), 100 ng/mL hygromycin B (Invivogen), or 100 ng/mL cycloheximide (Sigma).

Northern blotting

Oligonucleotide probes (IDT) were designed to hybridize with RNA species to be analyzed (Table 3). Oligonucleotide probes were 5′-radiolabeled with [γ-32P]-ATP (PerkinElmer) using T4 PNK (Enzymatics), and excess unincorporated label was removed with G-25 Sephadex (GE) spin columns. Probes were heated to 100°C and diluted into 10 mL ULTRAhyb Oligo hybridization buffer (Life Technologies). Total RNA (normalized to SCR1 loading) was electrophoresed, along with RiboRuler Low Range RNA Ladder (Thermo) and ss10 ssDNA ladder (Simplex Sciences) on precast 10% polyacrylamide urea TBE Novex gels (Life Technologies) at 160 V for 90 min. Gels were stained with SybrGold Nucleic Acid Stain (Life Technologies), imaged, and electrotransferred with a Genie electroblotter (Idea Scientific) onto Hybond N+ nylon membranes (GE) at 19 V for 60 min submerged in 1× TBE. RNA was crosslinked to the membranes with 0.12 J of 254 nm UV light (Stratalinker 1800, Stratagene). Nylon membranes were blocked with 10 mL ULTRAhyb Oligo for 1 h at 42°C and hybridized with probe at 42°C for 18 h. Blots were washed at 42°C for 30 min each with buffer (2× SSC, 0.1% SDS) and were developed by phosphorimaging (GE, Molecular Dynamics).

TABLE 3.

Oligonucleotide sequences

Detection of 2′-phosphate linkages by RT-PCR

A reverse primer that anneals directly downstream from the 2′-phosphate site (Table 3, HAC1-R0 2-P) was designed so that the first deoxyribonucleotide incorporated by the reverse transcriptase is complementary to the 2′-phosphorylated ribonucleotide. Total RNA (2 µg) was DNased (Turbo DNase Kit, Ambion), treated ± with calf intestinal phosphatase (NEB), and acid-phenol extracted. DNased and ±CIP-ed total RNA was denatured at 65°C for 5 min with 10 pmol of the reverse primer and then transferred to ice. The concentration of dNTPs in reverse transcription reactions was reduced from 500 to 1 µM to inhibit reverse transcriptase (SuperScript III, Life Technologies, lot #1826824) at sites of 2′-phosphorylation (Fig. 4E, schematic). PCR was performed with the same reverse primer and a forward primer that anneals to the 5′-exon of HAC1; 28 cycles were performed on low dNTP reactions and 22 cycles were performed on RT reactions with normal dNTP concentrations. Amplified DNA was electrophoresed in a 2% agarose TBE 1× SybrGold (Life Technologies) gel and imaged.

Expression vector for intronless tRNAs

We designed a sequence to express 10 prespliced tRNAs using the SUP4 promoter and RPR1 terminator, respectively (Good and Engelke 1994). The full sequence is available in Supplemental Material and Addgene record 70125. Two restriction sites at the 5′ and 3′ ends of the sequence for SacI and ClaI are underlined. Ten tRNA genes are encoded with an intervening spacer (“CTTTGT”) derived from a dicistronic tRNA (Engelke et al. 1985). From 5′ to 3′, the tRNAs are: Phe(GAA), Leu(CAA), Lys(UUU), Ser(GCU), Ile(UAU), Trp(CCA), Tyr(GUA), Pro(UGG), Ser(CGA), and Leu(UAG). The Pol III promoter contains sequence elements within the SUP4 Tyr(GUA) tRNA (Good and Engelke 1994), and thus tRNATyr(GUA) occurs twice in the construct. The sequence was synthesized as a gBlock from IDT and recombined into pDONR221 in a Gateway BP reaction (Life Technologies). The product plasmid (pDONR221-10X-tRNA) was recombined by a Gateway LR reaction into pAG424-ccdB (created by removing the GPD promoter from pAG424-GPD-ccdB[Alberti et al. 2007]), which has the TRP1 selectable marker and a 2µ origin for high-copy number per cell (Chan et al. 2013). The pDONR221-10X-tRNA, pAG424-ccdB (empty vector), and pAG424-10X-tRNA plasmids are available from Addgene (plasmids 70125, 70124, and 70123).

HAC1 epitope tagging and western blotting

Oligonucleotide probes HAC1-pML104-plus and HAC1-pML104-minus (IDT) (Table 3) form an insert that was ligated into the Cas9 and sgRNA yeast expression plasmid pML104 (Addgene entry 67638) (Laughery et al. 2015). Yeast strains were simultaneously transformed with pML104-HAC1-ct and FLAG donor DNA with homology to the HAC1 gene for homology-directed repair. Cotransformed yeast were selected on uracil drop-out media (US Biologicals) and confirmed to have an in-frame FLAG epitope tag by Sanger sequencing. Cell lysates were analyzed by immunoblotting using an anti-FLAG M2 antibody (Sigma) and secondary anti-mouse HRP-conjugated antibodies (ThermoFisher Scientific) with Enhanced Chemiluminescence detection (Promega). Loading controls included Ponceau staining of the membrane and anti-GAPDH antibody (UBP-Bio).

SUPPLEMENTAL MATERIAL

Supplemental material is available for this article.