Splicing of Intron-containing tRNATrp by the ArchaeonHaloferax volcanii Occurs Independent of Mature tRNA Structure*

We have investigated the requirements for mature tRNA structure in the in vivo splicing of theHaloferax volcanii, intron-containing tRNATrp RNA. A partial tRNATrp gene, which contained only the anticodon stem-loop region of the mature tRNA, was fused to a carrier yeast tRNA gene for expression in H. volcanii. Transcripts from this hybrid gene were found to be processed by endonuclease and ligase at the tRNATrp exon-intron boundaries. These results verify that the substrate recognition properties of the halobacterial endonuclease observed in vitro reflect the properties of this enzymein vivo, namely that mature tRNA structure is not essential for recognition by the endonuclease. The independence of these reactions on mature tRNA provides further support for a relationship between archaeal tRNA and rRNA intron-processing systems and highlight a difference in the substrate recognition properties between the archaeal and eucaryal processing systems. The significance of these differences is discussed in light of the observation that the tRNA endonucleases of these organisms are related.

While intron-containing tRNAs are present in the Archaea (formerly the Archaebacteria), Eucarya, and Bacteria, these sequences do not represent a single homogeneous class of introns. Bacterial and chloroplast tRNA introns are either group I or group II introns, whereas the eucaryal nuclear and archaeal tRNA introns lack any identifiable sequence or structural relationship to the group I, group II, group III, or mRNA introns (1). In the absence of defining sequence or structural characteristics in the archaeal and eucaryal tRNA introns, speculation on the relatedness of these introns has been based primarily on comparisons of their splicing systems. Until recently, it was thought that the archaeal and eucaryal splicing enzymes were distinct systems that were related in function only, and consequently that eucaryal and archaeal introns potentially represented two separate classes of introns. This argument was based on the observations that the archaeal and eucaryal tRNA intron endonucleases differed in subunit composition and substrate recognition mechanisms. The eucaryal endonuclease was observed to be a heterotrimer (2), which has recently been shown to be a tetramer (3), whereas the archaeal enzyme is a homodimer (4). Recognition of the exon-intron boundaries by the eucaryal enzyme involves a complex mechanism that is dependent on the presence of mature tRNA structure. All eucaryal tRNA introns are located in the anticodon loop between positions 37 and 38 of the mature tRNA, extending the anticodon helix, while maintaining the overall mature tRNA structure (1). The eucaryal enzyme senses the distance from the top of the anticodon stem to the 5Ј and 3Ј cleavage sites (5,6) and requires the formation of a threenucleotide bulge loop at the intron-exon 2 cleavage site, the A-I interaction (7). This mechanism is well suited for the identification of eucaryal tRNA introns where all introns are located in the same relative position. In contrast, in vitro studies with the Haloferax volcanii intron endonuclease showed that this enzyme does not require complete mature tRNA structure in its substrate (8,9). This enzyme requires a defined structural element at the exon-intron boundaries, the bulge-helix-bulge motif (9). In this structure each cleavage site is located in a three-nucleotide bulge loop, and the two loops are separated by 4 base pairs. The enzyme senses the distance between the bulge loops, rather than the length of the anticodon stem (9). This mechanism is well suited for the archaeal tRNA intron since these introns are not restricted to a single location in the mature domain of the tRNA. In the Archaea, tRNA introns have been observed in the anticodon loop, the anticodon stem, and the extra arm (10 -16). Despite their variability in location, all archaeal intron-containing tRNAs can assume the bulgehelix-bulge structure at their intron-exon boundaries.
With such fundamental differences in subunit composition and recognition mechanisms, the proposal that the archaeal and eucaryal tRNA processing systems were different appeared justified. However, the recent characterization of the genes encoding the H. volcanii and Saccharomyces cerevisiae tRNA intron endonucleases has unexpectedly revealed that these two enzymes are related (3,4). A comparison of the amino acid sequences of the halobacterial endonuclease monomer and the yeast endonuclease complex revealed that the halobacterial protein shared sequence similarity with two subunits of the yeast tetramer. This similarity extended over an approximately 115-amino acid region, and in each case, this sequence was located in the carboxyl terminus of the protein (4). Knowing that the archaeal and eucaryal endonucleases are related underscored the need to verify that the substrate recognition properties of the archaeal endonuclease defined in vitro are the same as those used in vivo. In this report we describe experiments to test the proposal that the halobacterial tRNA intron endonuclease can process a tRNA intron from a RNA molecule that lacks full tRNA structure. As an in vivo test for this model, we have constructed an H. volcanii expression module that is capable of producing a hybrid RNA that encodes a partial H. volcanii tRNATrp RNA fused to the 5Ј leader region of the non-processing yeast tRNAProM RNA (17). The tRNAT-rp⌬13115Ј-tRNAProM hybrid RNA encodes the tRNATrp anticodon stem-loop region and intact intron fused to the carrier RNA. This represents the minimum exon sequences required for in vitro cleavage. Analysis of RNA from cells carrying this hybrid gene demonstrate that this partial tRNATrp RNA is processed by both endonuclease and ligase enzymes in the absence of a complete mature tRNA structure. We discuss the implications of this observation in defining the relationships between archaeal and eucaryal tRNA processing systems and the roles of these archaeal enzymes in cellular RNA processing.

EXPERIMENTAL PROCEDURES
Culture Conditions and Materials-H. volcanii strain WFD11 (18) was grown aerobically at 37°C in complex medium (19), and when necessary to ensure maintenance of pWL-based expression plasmids, this medium was supplemented with 20 M mevinolin (a gift from Merck). Escherichia coli strains DH5␣-FЈ and JM110 were cultured in Luria Broth (LB) medium or LB medium supplemented with 100 g/ml ampicillin when cells carried pUC-or pWL-based plasmids.
T4 polynucleotide kinase, T4 DNA ligase, Klenow DNA polymerase, SuperScript TM II, Moloney murine leukemia virus reverse transcriptase, and all restriction enzymes were purchased from Life Technologies, Inc.; Sequenase TM , 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside and isopropyl-1-thio-␤-D-galactopyranoside were obtained from U. S. Biochemical Corp.; AmpliTaq TM DNA polymerase and Gene-Amp TM core reagents were purchased from Perkin-Elmer, and Zeta-Probe nylon membrane was obtained from Bio-Rad Laboratories. Oligonucleotides used in this study were synthesized by The Ohio State University Biochemical Instrument Center or Ransome Hill Biosciences, Inc.
Cloning of the H. volcanii tRNATrp⌬13115Ј Derivatives into the H. volcanii Expression Vector-The intron-containing H. volcanii tRNATrp⌬13115Ј gene, which contains a complete intron and only the anticodon stem and loop regions of the mature tRNATrp RNA, was isolated from plasmid pVT22-⌬13115Ј (20) as a 150-base pair EcoRI-HindIII restriction fragment. Protruding 5Ј and 3Ј ends of the tRNATrp⌬13115Ј fragment were filled in using Klenow DNA polymerase, and this fragment was cloned into the HincII restriction site of the vector pUC1318 (21). The tRNATrp⌬13115Ј gene was recovered from the pUC1318 vector as a XbaI-XbaI restriction fragment, which was then subcloned into the XbaI site of the H. volcanii expression vector pWL302A1 (22) to yield the plasmid pWL302A1-⌬13115Ј. The A3 to T3 mutation of the tRNATrp⌬13115Ј gene ( Fig. 1B) was prepared using the polymerase chain reaction (PCR). 1 The PCR reaction contained 30 mM Tricine, pH 8.4, 2 mM MgCl 2 , 5 mM ␤-mercaptoethanol, 0.01% gelatin, 0.1% Thesit, 200 M each dNTP, 1 M of the mutagenic primers T7O168A3-T (5Ј-AGCTCTAGATAATACGACTCACTATAGGCATGGC-GACTGACTCCAGTGGCT-3Ј), 1 M of the primer O168short (5Ј-AGG-GATCTAGACCCGATCGACTG-3Ј), and 2.5 units of AmpliTaq TM DNA polymerase. The reaction mixture was incubated at 95°C for 5 min, and polymerization was carried out for 30 cycles. Each cycle consisted of incubation at 94°C for 1 min, 55°C for 1 min, and 72°C for 2 min. The resulting fragment was cloned into pUC1318 and subcloned into pWL302A1 as described for the wild-type tRNATrp⌬13115Ј gene, yielding the plasmid pWL302A1-⌬13115ЈT3. These plasmids were introduced into E. coli strain JM110, and the plasmids isolated from these strains were then used to transform H. volcanii strain WFD11 (22). Passage through strain E. coli JM110 (dam Ϫ ) reduces restriction during transformation of H. volcanii (23).
RNA Isolation and Northern Analysis-For analysis of tRNAT-rp⌬13115Ј-tRNAProM and tRNATrp⌬13115ЈT3-tRNAProM RNAs, H. volcanii cells were grown to an A 560 of 1.0 at 37°C in complex medium (22) containing 20 M mevinolin. Cells from a 25-ml culture volume were harvested by centrifugation at 5000 ϫ g and lysed by adding 1.25 ml of lysis buffer (10 mM Tris-HCl, pH 8.0, 10 mM NaCl, 1 mM trisodium citrate, and 1.5% SDS) to the cell pellet. The lysate was incubated for 10 min at 37°C, followed by an additional 10-min incubation on ice. DEPC (25 l) and NaCl-saturated double distilled H 2 O (625 l) were added to the mixture, and the incubation was continued for an additional 15 min. DEPC was omitted from preparations to be used in reverse transcription reactions. The mixture was then transferred to a RNase-free Corex tube, and the contents were centrifuged at 12,000 ϫ g for 10 min at 4°C. The aqueous phase was recovered, and the RNA was ethanol-precipitated. The resulting RNA pellet was washed once with 2 ml of 95% ethanol and resuspended in RNase-free double distilled H 2 O or RNA loading buffer (7 M urea, 10% glycerol, 0.05% bromphenol blue, and 0.05% xylene cyanol). A typical yield was 120 g of total RNA per 25 ml of original culture volume.
For Northern analysis, total RNA isolated from H. volcanii cells, approximately 50 g in 50 l of RNA loading buffer, was separated by electrophoresis in a 6% denaturing (7 M urea) polyacrylamide gel. After separation, the RNA was electrophoretically transferred to a Zeta-Probe membrane as described by the manufacturer (Idea Scientific Co., Minneapolis, MN). Northern blots were prehybridized at 50°C for 10 min in a solution containing 5 ϫ SSC (1 ϫ SSC contains 150 mM NaCl and 15 mM trisodium citrate), 20 mM NaH 2 PO 4 , pH 7.0, 7% SDS, 10 ϫ Denhardt's solution, and 100 g/ml single-stranded salmon sperm DNA. The prehybridization mixture was discarded and replaced with the same buffer at 150 l/cm 2 Zeta-Probe membrane. Transcripts were detected by probing with one of either two 5Ј end-labeled oligonucleotides: ProExI (5Ј-CCCAAAGCGAGAATCATACCAC-3Ј) specific for the 3Ј reporter gene tRNAProM or IntDel (5Ј-GGACTCTAGAATTCGAG-3Ј) specific for the splice junction formed by exons 1 and 2. Oligonucleotides were labeled with [␥-32 P]ATP and T4 polynucleotide kinase. Hybridizations were conducted overnight at 50°C for ProExI and 45°C for IntDel. Following hybridization the membranes were washed three times for 15 min in 400 ml of wash buffer (2 ϫ SSC, 0.5% SDS) at 22°C. Hybrids were detected by autoradiography. For sequential hybridizations, blots were stripped between hybridizations by washing the membrane twice in a solution containing 0.1 ϫ SSC, 0.5% SDS at 95°C.
Analysis of the 180-Nucleotide RNA-A reverse transcription-PCR approach was used to verify that the 180-nucleotide RNA species resulted from accurate exon-intron cleavage and exon ligation at the tRNATrp⌬13115Ј processing site of the hybrid RNA. ProExI (20 pmol) was annealed with 25 g of total RNA in a 20-l reaction volume of 1 ϫ Moloney murine leukemia virus reverse transcriptase buffer (50.0 mM Tris-HCl, pH 8.3, 40 mM KCl, 6.0 mM MgCl 2 , 1.0 mM dithiothreitol) at 80°C for 4 min. This mixture was then allowed to cool to room temperature over a course of 5 min. First strand cDNA synthesis was initiated with the addition of 30 l of cDNA extension mixture (50.0 mM Tris-HCl, pH 8.3, 40 mM KCl, 6.0 mM MgCl 2 , 1.0 mM dithiothreitol, 0.1 mg/ml bovine serum albumin, and 1.25 mM of each dNTP) and 200 units of Moloney murine leukemia virus reverse transcriptase buffer; reactions were incubated for 1 h at 50°C. Reverse transcriptase was then inactivated by raising the reaction temperature to 95°C for 2 min. RNA was degraded by adding 1 l of RNase A (10 mg/ml) and incubation of the mixture for 10 min at 42°C. Proteins were removed from the reaction by extracting the mixture once with an equal volume of phenol/chloroform and once with chloroform. cDNAs were ethanol-precipitated, and the precipitate was washed once with 400 l of 75% ethanol, dried, and redissolved in 20 l of double distilled H 2 O. cDNAs were amplified by PCR as described above using the primers Prol5Ј (5Ј-GCAAGGGGACTCTAGAGT-3Ј) and ProExI. Reverse transcriptase DNA products from this reaction were cloned into the SmaI site of pUC19 and sequenced using the Sequenase TM system. Five individual pUC19 clones were examined, and each was determined to have the identical insert.

RESULTS
In Vivo Expression of the H. volcanii tRNATrp⌬13115Ј-tRNAProM Hybrid RNA-Based on in vitro processing studies, the tRNATrp⌬13115Ј variant of the H. volcanii tRNATrp gene was chosen as a model RNA to investigate the requirements for mature tRNA structure in in vivo tRNA splicing. This gene encodes an RNA having the complete tRNATrp intron and only the anticodon stem-loop of the mature tRNA. This RNA is accurately and efficiently cleaved by a partially purified H. volcanii tRNA intron endonuclease (8). To test whether these sequences and structures were sufficient for cleavage in vivo we needed a carrier RNA to express this potentially unstable form of the tRNATrp RNA. Previous in vivo expression studies showed that a modified version of the yeast tRNAPro(UGG) RNA, tRNAProM, could be expressed in H. volcanii on the expression plasmid pWL302A1 (22). This gene encoded a single stable transcript that represented the primary transcript from this gene. The production of a single RNA species was the result of two processing defects in this RNA, a U 6 -U 72 pair preventing 5Ј and 3Ј termini processing and an intron that is not recognized by the H. volcanii endonuclease (24). The yeast tRNAProM DNA fragment also carried a RNA polymerase III termination element that functioned as a strong terminator in H. volcanii. 2 We reasoned that introduction of the tRNATrp⌬13115Ј gene into the 5Ј leader region of the yeast tRNAProM construct would lead to the production of a stable RNA hybrid. This hybrid gene and its expected transcript are show in Fig. 1.
Analysis of tRNATrp⌬13115Ј-tRNAProM Transcripts-Processing of the tRNATrp⌬13115Ј-tRNAProM hybrid transcript was followed by Northern analysis. When an oligonucleotide specific to the yeast tRNAProM RNA exon 2 was used as a probe, three RNA species were detected (Fig. 2A). The approximated sizes of these RNAs were 285, 180, and 140 nucleotides. The largest RNA species corresponded in size to the expected primary transcript, and the smallest RNA species, 140 nucleotides, corresponded in size to the predicted RNA resulting from cleavage at the tRNATrp⌬13115Ј RNA intron-3Ј exon boundary. The intermediate species, 180 nucleotides, was similar in size to the product predicted for cleavage at both exonintron boundaries, followed by exon ligation (Fig. 2A, left panel). The other expected intermediate of the reaction, exon 1-intron RNA, was also detected when an exon 1-specific probe was used (data not shown).
Since correct cleavage and ligation of tRNATrp⌬1315Ј-tRNAProM RNA would generate a structure having sequences identical to the mature tRNATrp anticodon loop, these RNAs were also probed with an oligonucleotide that corresponds to the mature anticodon stem and loop sequence ( Fig. 2A, right  panel). This probe hybridized to both the mature, chromosomeencoded tRNATrp RNA and the 180-nucleotide species, suggesting that correct cleavage and ligation had occurred with the hybrid RNA. As an independent test that the cleavage reactions observed in vivo were the result of the tRNA intron endonuclease activity, and not a general ribonuclease, the processing pattern of a cleavage-defective form of the tRNATrp 2 Y-P. Kuo and C. J. Daniels, unpublished results. RNA, tRNATrp⌬13115ЈT3-ProM, was also examined. This RNA contains a single point mutation at the exon 1-intron boundary, A3 to T3 at position 41 (see Fig. 1B), which leads to an 80% decrease in in vitro cleavage when compared with the wild-type RNA (Fig. 2B, left panel). Northern analysis of RNAs from cells carrying the tRNATrp⌬13115ЈT3-ProM gene show that this RNA remains predominately as the primary transcript (Fig. 2B, right panel). A minor species (Ͻ10% of the total), which corresponded in size to an RNA resulting from cleavage at the exon 1-intron boundary, was also detected. The inability of this RNA to process in vivo is consistent with its cleavage properties in vitro and suggests that the in vivo processing of the tRNATrp⌬13115Ј-ProM RNA was the result of endonuclease activity.
To further verify that the 180-nucleotide RNA species produced from the tRNATrp⌬13115Ј-ProM gene was the product of both cleavage and ligation, cDNAs were synthesized from this RNA and used as template for PCR amplification. The cDNA was synthesized using a primer specific for the yeast tRNAProM RNA, and PCR amplification was carried out with oligonucleotides specific for sequences 5Ј of the tRNATrp⌬13115Ј encoding region and a sequence internal to the yeast tRNAProM RNA. This prevented cDNA synthesis and amplification of chromosome-encoded tRNATrp RNA. Five DNA products were sequenced, and all had the predicted sequence for accurate cleavage and ligation of the tRNATrp⌬13115Ј RNA (Fig. 3). DISCUSSION Based on in vitro processing studies of the H. volcanii tRNATrp RNA we predicted that in vivo cleavage of the intron from this pretRNA would be dependent on exon-intron boundary sequences and structures and independent of mature tRNA structure and sequences beyond those of the anticodon stem and loop (8,9). To determine if the requirements observed in vitro reflected the requirements for in vivo processing we introduced the H. volcanii tRNATrp⌬13115Ј gene into the 5Ј unprocessed leader region of the yeast tRNAProM RNA. As anticipated, Northern analysis confirmed that the predicted hybrid RNA was produced and that this RNA underwent cleavage in the absence of a complete mature tRNA structure ( Fig.  2A). RNA species consistent with cleavage at both 5Ј and 3Ј exon-intron boundaries were detected indicating that both cleavage sites were recognized. The inability of the in vitro processing-defective tRNATrp⌬13115ЈT3-ProM RNA to undergo processing in vivo supported the proposal that the hybrid RNA was cleaved by endonuclease rather than a general ribonuclease. Unexpectedly, an additional 180 nucleotide species was detected in cells that carried the tRNATrp⌬13115Ј-ProM RNA gene. This RNA corresponded in size to an RNA that had undergone cleavage at both sites and exon ligation (Fig. 2B). Sequence analysis of cDNAs derived from this 180-nucleotide RNA by reverse transcription and PCR amplification verified that this RNA resulted from both accurate cleavage and exon ligation (Fig. 3). These data confirm earlier in vitro observations that the halobacterial tRNA processing system is capable of acting on non-tRNA substrates (8) and show for the first time that the tRNA ligation reaction is independent of mature tRNA structure. These results also provide an explanation for how a single endonuclease could act on a population of intron-containing pretRNAs where all introns are not located in the same relative position in the mature tRNA. In this case the primary criteria for recognition would be the presence of the bulgehelix-bulge motif. Indeed, most archaeal intron-containing tRNAs possess this or a closely related structure at their exonintron boundaries (8,25,26).
A processing system that is directed toward sequences and structures at the exon-intron boundaries could in theory cleave any RNA, regardless of its origin or final structure. Structural analysis of archaeal intron-containing 16 S and 23 S rRNA precursors has shown that these RNAs have the tRNA bulgehelix-bulge motif or closely related structures at their exonintron boundaries (27). Some rRNA introns also encode homing endonucleases (28 -31) similar to those found in some group I introns; however, the characteristic core group I RNA struc- tures and sequences are absent in these introns. This has led to the proposal that these intron-containing rRNAs are processed by the same enzyme system as the pretRNAs (8,15,25,26). In support of this proposal a partially purified Desulfurococcus mobilis 23 S rRNA intron endonuclease was found to cleave an intron-containing tRNA from this same organism (15). The ability of this endonuclease to act on tRNA and rRNA substrates raises the question of whether this enzyme could act as a general RNA endonuclease. One possible candidate is the primary transcript from the rRNA operon. Sequence analysis of archaeal rRNA operons has shown that the 16 S and 23 S rRNA coding regions are flanked by large inverted repeats. As in bacterial cells, these helices are though to be sites for RNaseIII cleavage (32). However, we and others (25,33) have noted that the archaeal helices possess the characteristic bulge-helixbulge motif of the tRNA exon-intron boundaries. It is possible that these structures are recognized by the tRNA endonuclease rather than a RNaseIII-like enzyme. Interestingly, a survey of the Methanococcus jannaschii genome (34) did not reveal the presence of an RNaseIII-like gene in this organism.
The results of this study also show that ligation can take place in the absence of mature tRNA structure. A ligase capable of joining exon ends held in close juxtaposition would be expected to act on those tRNAs where the intron was excised from a position other than the anticodon loop. This ligase could also act on the rRNA exons produced by endonuclease cleavage of introns from these RNAs. The lack of an in vitro assay for ligase has prevented us from determining if the archaeal ligase is similar to the ATP-GTP-requiring ligase of yeast and HeLa cells (35) or the ATP-independent ligase of vertebrates (36). Intron circularization has been observed during D. mobilis rRNA cleavage; however, the exons remain unligated in this reaction (37). Halophile extracts also lack ligase activity (8). A survey of the M. jannaschii genome (34) did not reveal genes related to the eucaryal tRNA ligase.
Similarities in the exon-intron boundaries of archaeal rRNAs and tRNAs and the finding that the halobacterial tRNA intron endonuclease and ligase enzymes act on non-tRNA RNAs provide further support for the proposal that the archaeal tRNA and rRNA introns are processed by the same enzyme system. In addition, the recently discovered similarities between the archaeal and eucaryal tRNA intron endonucleases (3,4) strongly suggests that the tRNA introns of these two domains and the rRNA introns of the Archaea represent a single class of introns. The molecular mechanisms that have led to changes in the substrate recognition properties of the archaeal and eucaryal endonucleases is not yet understood. It is likely that the these differences are in part due to the divergence in the two yeast endonuclease subunits that are related to the halobacterial endonuclease protein. These two subunits are not identical as they are in the archaeal enzyme, and as a consequence they may have different RNA binding characteristics. The presence of other subunits in the yeast endonuclease may also influence the interaction of this enzyme with its substrate. Finally, no intron-containing mRNAs have been detected in the Archaea to date; however, the properties of the archaeal intron processing system described in this report indicate that a tRNA-like intron could exist in a mRNA.