RtcB, a Novel RNA Ligase, Can Catalyze tRNA Splicing and HAC1 mRNA Splicing in Vivo*

RtcB enzymes are novel RNA ligases that join 2′,3′-cyclic phosphate and 5′-OH ends. The phylogenetic distribution of RtcB points to its candidacy as a tRNA splicing/repair enzyme. Here we show that Escherichia coli RtcB is competent and sufficient for tRNA splicing in vivo by virtue of its ability to complement growth of yeast cells that lack the endogenous “healing/sealing-type” tRNA ligase Trl1. RtcB also protects yeast trl1Δ cells against a fungal ribotoxin that incises the anticodon loop of cellular tRNAs. Moreover, RtcB can replace Trl1 as the catalyst of HAC1 mRNA splicing during the unfolded protein response. Thus, RtcB is a bona fide RNA repair enzyme with broad physiological actions. Biochemical analysis of RtcB highlights the uniqueness of its active site and catalytic mechanism. Our findings draw attention to tRNA ligase as a promising drug target.

RtcB enzymes are novel RNA ligases that join 2,3-cyclic phosphate and 5-OH ends. The phylogenetic distribution of RtcB points to its candidacy as a tRNA splicing/repair enzyme. Here we show that Escherichia coli RtcB is competent and sufficient for tRNA splicing in vivo by virtue of its ability to complement growth of yeast cells that lack the endogenous "healing/ sealing-type" tRNA ligase Trl1. RtcB also protects yeast trl1⌬ cells against a fungal ribotoxin that incises the anticodon loop of cellular tRNAs. Moreover, RtcB can replace Trl1 as the catalyst of HAC1 mRNA splicing during the unfolded protein response. Thus, RtcB is a bona fide RNA repair enzyme with broad physiological actions. Biochemical analysis of RtcB highlights the uniqueness of its active site and catalytic mechanism. Our findings draw attention to tRNA ligase as a promising drug target.
Escherichia coli RtcB exemplifies a new family of RNA ligases that directly seal 2Ј,3Ј-cyclic phosphate and 5Ј-OH ends (1)(2)(3). Direct ligation is thought to be the main pathway of tRNA splicing in animals and archaea (4 -6). By contrast, yeast and plants rely on a different mechanism of tRNA splicing in which the broken 3Ј and 5Ј ends are healed (converted to a 3Ј-OH/2Ј-PO 4 and 5Ј-PO 4 , respectively) and then sealed by a classical ATPdependent RNA ligase (7) (see Fig. 1). RNA ligases of the RtcB family are present in metazoa and protozoa, but not in fungi and plants. RtcB homologs purified from archaeal and mammalian cells can seal broken tRNA halves and are thereby imputed to be the catalysts of archaeal and mammalian tRNA splicing (2,3). However, this scenario is complicated by the existence of a yeast/plant-like tRNA splicing pathway in mammalian cells (8 -12) and of analogous yeast-like RNA repair enzymes in many archaeal taxa (13)(14)(15)(16). Definitive genetic evidence that RtcB is the sole essential agent of the repair phase of mammalian or archaeal tRNA splicing is lacking, and the available genetic evidence concerning the healing-sealing pathway in animals is equivocal. Genetic ablation of a murine homolog of a yeast-like pathway component Tpt1 (the enzyme that removes the 2Ј-phosphate at the splice junction; see Fig. 1) has no discernible phenotype (17), suggesting that the mammalian yeast-like pathway either is functionally redundant with direct ligation or is non-contributory to mammalian tRNA splicing. By contrast, siRNA-directed depletion of the mammalian RNA 5Ј-kinase (an ortholog of the kinase domain of yeast/plant tRNA ligase) elicited a defect in tRNA splicing in vitro (10). However, siRNA-directed depletion of mammalian RtcB was also reported to inhibit ligation of tRNA halves in cell extracts and to delay tRNA splicing in living cells (3). These findings leave unresolved the following key issues: (i) whether RtcB can suffice for tRNA splicing as the only source of tRNA ligase activity in a eukaryal cell and (ii) whether RtcB can perform other RNA repair functions in vivo. Here we address these questions by using budding yeast as a surrogate genetic model for tRNA splicing and RNA repair by heterologous enzymes (7).

EXPERIMENTAL PROCEDURES
Expression Plasmids-The bacterial plasmid pET28b-His 10 Smt3-RtcB encodes wild-type E. coli RtcB fused to an N-terminal His 10 Smt3 tag (1). Missense mutations were introduced into the RtcB ORF by two-stage overlap extension PCR. The RtcB inserts were sequenced to confirm the desired mutations and exclude the acquisition of unwanted coding changes. The wild-type and mutated RtcB ORFs were excised from their respective pET plasmids and inserted into the yeast vector pRS423 (2 HIS3) wherein RtcB expression is under the control of the yeast TPI1 promoter. The yeast p(CEN LEU2 GAL1 PaT) plasmid for galactose-inducible expression of intracellular Pichia acaciae toxin (PaT) 2 lacking the N-terminal 12-amino acid signal peptide was constructed by excising an NdeI/SalI fragment containing the PaT ORF from plasmid pPACBX (18) (a gift of Roland Klassen and Friedhelm Meinhardt) and inserting it into pRS415 (CEN LEU2) between GAL1 promoter and terminator elements.
RtcB Purification-The wild-type and mutant pET28b-His 10 Smt3-RtcB plasmids were transformed into E. coli BL21-CodonPlus(DE3). Induction of RtcB expression, preparation of soluble lysates, recovery of the His 10 Smt3-RtcB proteins by nickel-agarose chromatography, excision of the tags by treatment with the Smt3-specifc protease Ulp1, and separation of the tag-free RtcB proteins from His 10 Smt3 by a second round of nickel-agarose chromatography were performed as described (1). Protein concentrations were determined by using the Bio-Rad dye reagent with BSA as the standard. The polypeptide compositions of the RtcB preparations were analyzed by SDS-PAGE (supplemental Figs. S1 and S2).
RNA Repair Substrates-A synthetic RNA oligonucleotide R30 containing the anticodon stem-loop of yeast tRNA Glu(UUC) was 5Ј 32 P-labeled by reaction with T4 Pnkp and [␥-32 P]ATP. R30 was then cleaved 3Ј of the wobble uridine by reaction with Kluyveromyces lactis ␥-toxin (1). The 32 P-labeled R19Ͼp strand with a 2Ј,3Ј cyclic phosphate end (5Ј-pUGGCUCCGA-UAUCACGCUUϾp) was purified by preparative PAGE. Oligo-* This work was supported, in whole or in part, by National Institutes of Health Grant GM46330 (to S. S.). □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Figs. S1 and S2. 1 To whom correspondence should be addressed. E-mail: s-shuman@ski. mskcc.org.
nucleotide HO R20 (5Ј-HO UCACCGUGGUAUCGGAGCGC) was employed to form a broken tRNA-like stem-loop (see Fig.  4) by mixture with the 5Ј 32 P-labeled R19 single strand at an R20:R19 ratio of 10:1. To test the RNA requirements for the RtcB ligation reaction, the 32 P-labeled R19Ͼp strand was also annealed to a synonymous deoxyuridine-containing HO D20 oligonucleotide to form a broken RNA-DNA hybrid stem-loop (see Fig. 5).
Alternative stem-loop RNA repair substrates with a 3Ј-OH end at the break site (see Fig. 5) were prepared using a synthetic R19 OH oligonucleotide that had been 5Ј 32 P-labeled and then purified by PAGE. The labeled R19 OH oligonucleotide was annealed to a 10-fold excess of unlabeled D20 and R20 strands.
RNA repair substrates with a 5Ј-PO 4 at the break site (see Fig. 5) were formed by annealing radiolabeled R19Ͼp or R19 OH strands to a 10-fold excess of cold pR20 strand. The pRNA strand was generated by enzymatic phosphorylation of HO R20 using T4 Pnkp and cold ATP and then purified by PAGE.

RESULTS AND DISCUSSION
Here we interrogated the tRNA repair function of E. coli RtcB in vivo by complementation of a lethal deletion of the TRL1 gene encoding the tRNA ligase of Saccharomyces cerevisiae (7). Yeast Trl1 is a trifunctional tRNA repair enzyme (with 5Ј-OH kinase, 2Ј,3Ј cyclic phosphodiesterase, and ATP-dependent RNA ligase activities) that heals and seals the broken tRNA halves with 2Ј,3Ј cyclic phosphate and 5Ј-OH ends generated by incision of pre-tRNAs at the exon-intron junctions (Fig. 1). E. coli RtcB is a monofunctional ligase that directly joins 2Ј,3Ј cyclic phosphate and 5Ј-OH ends (1) (Fig. 1). We expressed the 408-amino acid E. coli RtcB protein in S. cerevisiae under the control of a constitutive promoter on a 2 plasmid and demonstrated by plasmid shuffle that RtcB could indeed sustain growth of a trl1⌬ strain (Fig. 1). These results prove that RtcB is able and sufficient to perform the essential repair steps of eukaryal tRNA splicing in vivo.  To gauge HAC1 mRNA splicing, cultures (10 ml) of S. cerevisiae trl1⌬ p(2 HIS3 TRL1) and trl1⌬ p(2 HIS3 RtcB) cells were grown at 30°C in YPD media until the A 600 reached 0.6. The cultures were split into 5-ml aliquots, one of which was adjusted to 1.5 g/ml tunicamycin to induce endoplasmic reticulum stress. One hour later, the cells were harvested by centrifugation. Total RNA was isolated by using a yeast RNA purification kit (Epicentre Biotechnologies). HAC1 transcripts were detected by RT-PCR using primers flanking the intron. The cDNAs were synthesized by using the SuperScript III system (Invitrogen) with 1.0 g of total RNA as template and 0.1 M of HAC1-specific antisense primer (5Ј-dCAT-GAAGTGATGAAGAAATCATTCAATTC; complementary to nucleotides 940 -968). The cDNAs were amplified by 22 cycles of PCR with Herculase (Stratagene) using the antisense primer and a sense-strand primer (5Ј-dCCAAGGAAAAGAGCCAAGACAAAAGAGG; corresponding to nucleotides 82-109). The RT-PCR products were analyzed by 1.4% agarose gel electrophoresis. A negative image of the ethidium bromide-stained gel is shown. The discovery that the healing and sealing activities of yeast Trl1 are also responsible for unconventional splicing of the HAC1 mRNA in the yeast unfolded protein response (UPR) pathway (19,20) extended the RNA repair paradigm to mRNA metabolism. Endoplasmic reticulum stress induces the Ire1 endonuclease to cleave HAC1 mRNA at two sites, which liberates a 252-nucleotide intron (Fig. 2) and leaves 2Ј,3Ј cyclic phosphate and 5Ј-OH termini on the proximal and distal exons, respectively. Healing and sealing of the exons by Trl1 creates a new open reading frame encoding an active Hac1 transcription factor. The mammalian UPR entails stress-induced Ire1 cleavage and unconventional splicing of the XBP1 mRNA (21), but the enzyme responsible for sealing the broken XBP1 transcript is not known. Here we tested the capacity of RtcB to execute unconventional mRNA splicing during the yeast UPR. Yeast TRL1 cells exposed to tunicamycin shifted their HAC1 mRNA profile, as assayed by RT-PCR with primers flanking the intron, whereby the intron-containing long form was replaced by a spliced short form (Fig. 2). Sequencing of the isolated RT-PCR products confirmed their identity as unspliced and spliced HAC1 cDNAs, respectively. Yeast trl1⌬ cells reliant on RtcB as their source of tRNA ligase were proficient in generating the spliced HAC1 RNA in the presence of tunicamycin (Fig. 2), signifying that RtcB is competent for mRNA splicing in the eukaryal UPR. Sequencing of the RT-PCR product verified that the Ire1 endonuclease cleavage sites were ligated faithfully.
Another manifestation of tRNA repair is the capacity to protect cells from the cytotoxicity inflicted by tRNA-specific ribotoxins that incise the anticodon loop by a transesterification mechanism that leaves 2Ј,3Ј cyclic phosphate and 5Ј-OH termini at the broken tRNA ends (22,23). PaT is a secreted fungal defense molecule that penetrates S. cerevisiae cells and, upon accessing the cytoplasm, breaks the anticodon loop of tRNA Gln(UUG) . Consequent depletion of the tRNA Gln(UUG) pool arrests yeast growth (18). Galactose-induced expression in S. cerevisiae of an intracellular form of PaT recapitulates its toxicity (18). The salient finding here was that replacing yeast Trl1 with RtcB as the source of tRNA ligase protected S. cerevisiae from PaT-mediated growth inhibition (Fig. 3). Taken together, these experiments establish that RtcB is a tRNA/mRNA repair enzyme with broad physiological actions.
The RtcB sealing reaction is posited to entail nucleophilic attack by the O5Ј nucleophile on the cyclic phosphate, with expulsion of the ribose O2Ј. Our initial findings that RtcB is manganese-dependent (1) indicated that sealing is not merely reversal of the metal-independent cleavage transesterification mechanism used by tRNA splicing endoribonuclease and tRNA-damaging anticodon nucleases. Manganese might pro-  mote RtcB catalysis by coordinating the O5Ј nucleophile to lower its pK a and/or engaging the cyclic phosphate oxygens to stabilize the transition state. The mechanistic novelty of RtcB is underscored by the crystal structure of the RtcB homolog from Pyrococcus horikoshii (24) (Fig. 4). RtcB has a distinctive tertiary structure with no similarity to any known ligases or phosphotransferases. RtcB has a deep and wide hydrophilic pocket lined by conserved histidines and a cysteine, suggestive of a metalbinding site that, being dominated by "soft" metal contacts to histidine nitrogens and cysteine sulfur, could account for the fact that E. coli RtcB requires manganese and is virtually inactive with magnesium (1). A water coordinated by the equivalents of E. coli RtcB side chains Asp-75, Cys-78, Asn-167, and His-168 is a potential mimic of the enzyme-bound metal (Fig.  4). Flanking the putative metal site in RtcB are two sulfate anions (potential mimetics of RNA phosphates) coordinated by basic amino acid side chains equivalent to E. coli RtcB residues Lys-298, His-168, and Arg-189 (Fig. 4).
Structure-activity relations at seven of the RtcB amino acids defined as essential by the alanine scan were probed by introducing conservative substitutions (supplemental Fig. S2).
Replacing Asp-75 by either asparagine or glutamate inactivated RtcB, attesting to the requirement for a carboxylate at this position and the steric constraints on the main chain to carboxylate distance (supplemental Fig. S2). Changing Cys-78 to serine abolished ligase activity, signifying that the S␥ atom is critical. His-168 and His-337 were replaced by glutamine and asparagine; the H168N, H168Q, H337N, and H337Q mutants were catalytically inactive (supplemental Fig. S2). The mutational effects are consistent with roles for Asp-75, Cys-78, His-168, Asn-167, His-185, and His-337 in metal binding and/or RNA transesterification. Although replacing the sulfate-binding Lys-298 with glutamine phenocopied K298A, the arginine substitution restored ligation to one-fourth of the wild-type RtcB (supplemental Fig. S2), highlighting positive charge as the key property of this residue. By contrast, neither Arg-189 nor Arg-341 could be functionally substituted by lysine or glutamine, implying that the multivalent ionic and hydrogen-bonding contacts of these arginines seen in the RtcB crystal structure (Fig. 4) are indeed pertinent to enzyme activity, either via binding to the RNA phosphates (a putative function of Arg-189) or in stabilizing the active site conformation (a likely role of Arg-341). None of the conservative mutants complemented trl1⌬ (not shown).
Further insights to the substrate specificity of RtcB were gained by varying the 3Ј and 5Ј termini of the broken stem-loop substrate and testing the capacity of a 5Ј-OH DNA strand to serve as the nucleophile for the sealing reaction. A 5Ј 32 P-labeled 19-mer RNA with either a 2Ј,3Ј cyclic phosphate end (R19Ͼp) or a 3Ј-OH end (R19 OH ) was annealed to an unlabeled 20-mer strand composed of all ribonucleotides (R20) or all deoxynucleotides (D20) to form the broken stem-loops depicted in Fig. 5. The unlabeled 20-mer strands had either a 5Ј-OH terminus ( HO R20 or HO D20) or a 5Ј-PO 4 terminus (pR20). The results of the ligation assays established that: (i) 2Ј,3Ј cyclic phosphate and 5Ј-OH ends are the only suitable reactants for E. coli RtcB among the combinations tested and (ii) RtcB is adept at joining the R19Ͼp "donor" strand to either RNA or DNA "acceptor" strands so long as they have a 5Ј-OH end. The versatility of RtcB with respect to the 5Ј-OH acceptor suggests practical applications for this enzyme in tagging and/or cloning mature RNAs or RNA processing intermediates that have 2Ј,3Ј cyclic phosphate ends.
In summary, we have provided convincing genetic and biochemical evidence that RtcB can serve as a genuine tRNA splicing/repair enzyme in a eukaryal cell. Our findings here and previously (7) attest to the portability of viral, fungal, plant, and bacterial tRNA repair systems. The salient theme is that sealing by any of several distinct enzymatic/chemical pathways suffices for tRNA splicing and unconventional mRNA splicing in vivo. The candidacy of mammalian RtcB homolog as an agent of tRNA splicing by direct ligation is supported by our genetic results. However, the jury is still out as to whether RtcB is functionally redundant with a mammalian tRNA repair pathway via sequential end-healing and end-sealing steps (8,10).
Our findings also fortify the case for the Trl1-associated ligase as a promising antifungal drug target for the following reasons: (i) Trl1 is present in all fungal proteomes; (ii) ligaseinactivating mutations in the active site of Trl1 are lethal in vivo (25,26); and (iii) mammalian taxa encode no homolog of the Trl1 ligase domain. Thus, any ATP-dependent ligase involved in the putative mammalian healing-sealing pathway (8) must either belong to a different enzyme family or have diverged so far from a fungal-type ancestor as to be unrecognizable. In either event, one predicts that a mechanismbased inhibitor of the Trl1 ligase should selectively block fungal growth without affecting mammalian cells. By the same token, an antagonist of RtcB might be useful in transiently impeding the UPR for therapeutic benefit in human diseases involving endoplasmic reticulum stress (21). The availability of isogenic yeast strains with orthogonal tRNA splicing systems should enable differential cell-based screening for bioactive molecules that selectively inhibit growth by targeting Trl1 versus RtcB and vice versa.