Purification, cloning and characterization of XendoU, a novel endoribonuclease involved in processing of intron-encoded small nucleolar RNAs in Xenopus laevis

Here we report the purification, from Xenopus laevis oocyte nuclear extracts, of a new endoribonuclease, XendoU, that is involved in the processing of the intron-encoded box C/D U16 small nucleolar RNA (snoRNA) from its host pre-mRNA. Such an activity has never been reported before and has several uncommon features that make it quite a novel enzyme: it is poly(U)-specific, it requires Mn(2+) ions, and it produces molecules with 2'-3'-cyclic phosphate termini. Even if XendoU cleaves U-stretches, it displays some preferential cleavage on snoRNA precursor molecules. XendoU also participates in the biosynthesis of another intron-encoded snoRNA, U86, which is contained in the NOP56 gene of Xenopus laevis. A common feature of these snoRNAs is that their production is alternative to that of the mRNA, suggesting an important regulatory role for all the factors involved in the processing reaction.


Introduction
Endoribonucleases play essential roles in RNA metabolism participating both in "degradative" pathways, such as mRNA decay, and in "maturative" pathways, to generate functional RNA molecules (1,2). Despite the plethora of functions played by processing enzymes in RNA metabolism, in eukaryotes only a few endoribonucleases have been isolated to date. Most of these activities are involved in the biosynthesis of translation components. In particular, RNase P and RNase MRP are ribonucleoprotein enzymes, functioning as site-specific endoribonucleases (3,4). Other well characterized endonucleolytic activities, such as the 3'-tRNase, the tRNA splicing endonuclease and members of the RNase III-like family are protein-only enzymes (5)(6)(7). While the majority of these activities participate in the biosynthesis of a specific class of RNA molecules, RNase III was shown to be required for a large number of different maturative pathways. S.cerevisiae RNase III (Rnt1p) was shown to be involved in pre-rRNA, snRNA and snoRNA processing (8)(9)(10)(11)(12). Recently, Rnt1p was also shown to participate in processing the intron-encoded snoRNAs U18 and snR38 from their host pre-mRNA (13). Furthermore, a new member of the metazoan RNase III family has been identified to be involved in the RNA interference process (14).
Another process in which the participation of endoribonucleases was expected to play an important role is the biosynthesis of snoRNAs. These RNAs are part of a complex class of molecules which are localized in the nucleolus where participate, as small ribonucleoprotein particles (snoRNPs), in different rRNA maturative events: processing and nucleotide modifications (15,16). Most snoRNAs in vertebrates are encoded in introns of proteincoding genes and are released from the host primary transcript either by debranching and exo-trimming of the spliced lariat (splicing-dependent pathway) or by endonucleolytic cleavage of the pre-mRNA (splicing-independent pathway) (15,16). There are only a few cases of intron-encoded snoRNAs in vertebrates which are released through the intervention of endoribonucleases (17,18), but so far these activities have not been purified and characterized. We previously showed, by microinjection experiments in X.laevis oocytes, that precursors containing U16 and U86 snoRNAs undergo very little splicing, while they efficiently produce snoRNAs through a processing pathway, involving specific endonucleolytic cleavages inside the intron (17,18). The common feature of U16 and U86 snoRNAs is their localization in introns which are poor splicing substrates, due to the presence of non-canonical consensus sequences.
We previously reported the identification in oocyte nuclear extracts (ONE) of an endoribonucleolytic activity, named XendoU (19), that produced the release of U16 snoRNA from its host intron, by cleaving at the same sites identified in vivo (17). The same activity was described to operate also for the processing of U86 snoRNA (18 and this paper).
From massive preparation of X.laevis ONE, we purified to homogeneity the XendoU endoribonuclease and characterized its activity. Partial protein sequencing enabled us to clone a XendoU cDNA, to express it and to perform a functional characterization of this enzyme.
Several aspects of XendoU make this protein a novel enzyme, different from all known endoribonucleases characterized so far: i) it is poly-U specific, ii) its activity depends on Mn 2+ ions and iii) it releases cleavage products with 2'-3'cyclic phosphate termini.

Experimental Procedures
Purification of XendoU activity. The X.laevis oocyte nuclear extracts were prepared as already described (17). The pellet obtained after two sequential ammonium sulphate precipitations (45% and 70% saturation) was dissolved in buffer A (25 mM Hepes, 50 mM NaCl, 0.1 mM EDTA, 1 mM DTT, 10% glycerol, pH 7.5) and applied onto an hydroxyapatite column (CHT-II Econocolumn, Biorad). Elution was carried out with 100 mM Naphosphate, pH 7, in buffer A; the active fractions were diluted with 3 volumes of buffer A and applied on a Blue Sepharose column (Blue Sepharose Fast Flow Pharmacia). Elution was performed with 0.2 M NaCl in buffer A and fractions containing XendoU activity were pooled and dialyzed against buffer A. The protein mixture was subjected to a second fractionation on hydroxyapatite column. The elution was performed with 10 column volumes of a linear gradient 0-100 mM Na-phosphate, pH 7, in buffer A. Selected fractions were then applied on a gel-filtration column (Pharmacia) previously equilibrated in buffer A.
By this procedure, starting from 15 ml of ONE (7 mg/ml) we obtained sufficient amount for protein sequencing analysis (approximately 70 µg).
Preparation and isolation of tryptic peptides. The protein band from SDS-PAGE (5 µg) stained with Coomassie Blue R250 was excised, reduced with DTT and carboxamidomethylated. The gel piece was equilibrated in 25 mM NH 4 HC0 3 , pH 8, and finally digested in situ with trypsin at 37 °C for 18 h. Peptides were extracted by sonication with 100 µl of 25 mM NH 4 HC0 3 /acetonitrile 1:1 v/v, pH 8 (twice). Peptide mixture was fractionated by reverse-phase HPLC on a Vydac C 18 column 218TP52 (250 x 1 mm), 5 µm, 300 Å pore size (The Separation Group, USA) by using a linear gradient from 5% to 60% of acetonitrile in 0.1% TFA over 60 min, at flow rate of 90 µl/min. Individual components were manually collected and lyophilized.
Peptide Sequencing and mass spectrometry analysis. Sequence analysis was performed using a Procise 491 protein sequencer (Applied Biosystems, USA) equipped with a 140C microgradient apparatus and a 785A UV detector (Applied Biosystems, USA) for the automated identification of PTH-amino acids, as already described (20).
Matrix assisted laser desorption ionization mass spectra were recorded using a Voyager DE-PRO mass spectrometer (Applied Biosystems, USA), as previously reported (20); a mixture of analyte solution, α-cyano-4-hydroxy-cinnamic acid was applied to the sample plate and dried. Mass calibration was performed using the molecular ions from peptides produced by trypsin auto-proteolysis and the matrix as internal standards.
Plasmids and templates for RNA transcription. U16 and U86-containing precursors were already described (17,18). The following U16-containing mutant derivatives were obtained by inverse PCR on plasmid 003 (17) with the oligonucleotides indicated in parentheses: In vitro processing reactions of wild type and mutant derivative RNAs. U16 and U86containing precursors (17,18) were in vitro transcribed in the presence of [α-32 P]UTP and pre-mRNAs were injected into nuclei of stage VI oocytes as already described (22).
Alternatively, 3 X 10 4 cpm of 32 P-labeled pre-mRNAs were incubated with 1 µg of ONE (17) or with 1 ng of purified XendoU, in 5 mM MnCl 2 , 50 mM NaCl, 25 mM Hepes, pH 7.5, Analysis of 3' termini of cleavage products. This analysis was performed by two different approaches. In the first one, 32 P-labeled gel-purified I-1b molecules, generated by incubation of U16-containing precursor with ONE, with purified XendoU or in vivo, were treated with 10 µl of 10 mM HCl, at 25 °C, for 2 hr, to hydrolyse the cyclic phosphate moiety as described by Forster (23). The phosphate was then removed by incubation of the RNA in 50 mM Tris-HCl, 0.1 mM EDTA, pH 8.5, in the presence of 1 U of calf intestine alkaline phosphatase at 50 °C, for 60 min. RNA was subjected to two subsequent purification steps with phenol:chloroform, ethanol precipitated and analyzed on 10% polyacrylamide-7 M urea gel. In the second approach, the I-1b molecule, obtained by incubation with purified XendoU, was gel-eluted and redissolved in 30 mM Tris, pH 8.0, 15 mM MgCl 2 and 1.5 units/µl T4 polynucleotide Kinase, and the mixture was incubated for 45 min at 37 °C to remove the 2-3 cyclic phosphate (24). RNA was then extracted and labeled using [5-32 P] pCp and T4 RNA ligase for 5 h at 16°C. RNA was then analyzed on 6% polyacrylamide-7 M urea gel. For the I-4 products derived from U86 processing, oligonucleotide UHindIII (5-AAGCTTCTTCATGGCGGCTCGGCCAAT-3) was utilized.

Results
Purification of XendoU from X.laevis oocyte nuclear extracts. We previously developed an in vitro system able to reproduce the in vivo processing of U16 snoRNA from its host intron (17). When 32 P-labeled U16-containing precursor was incubated with X.laevis oocyte nuclear extracts (ONE), specific endonucleolytic products were obtained (Fig. 1B): the I-1 and I-2 molecules derive from cleavage upstream to the U16 coding region, while the I-3 and I-4 molecules are produced by cleavage downstream to U16. When double cleavage occurs on the same precursor molecule, pre-U16 products accumulate; these intermediates are eventually converted by exo-trimming to the mature snoRNA (see Fig.1A). The cleavage sites were previously mapped in correspondence of short U-stretches: four of them are clustered upstream to U16 and one is located downstream. This cleaving activity was named XendoU (19).
In this work we carried out the biochemical purification of XendoU (Fig. 1D). The enzymatic activity was followed, throughout the different steps, by incubating 32 P-labeled U16-containing precursor with aliquots of the different fractions and by analyzing the cleavage products on polyacrylamide gels. Since we previously observed the dependence of XendoU activity on Mn 2+ ions (19), this cofactor was always added to the reaction mixture.
The protein content of the active fractions is shown in Fig. 1E. After several chromatographic steps, a single component of 37 kDa was identified in those fractions displaying specific activity (Fig. 1E, lane 6). The elution profile on the gel filtration chromatography was consistent with XendoU being a monomeric protein (not shown). Fig. 1B shows the comparison of processing activity of 1 µg of ONE (lanes ONE) with that of 1 ng of the purified 37 kDa polypeptide (lanes XendoU). In both cases, the same primary cleavage products (I-2 and I-3), their complementary cut-off molecules (I-1 and I-4) and pre-U16 molecules were generated.
Characterization of XendoU cleavage. In order to analyze the specificity of cleavage of the purified enzyme, primer extension analysis was performed on gel-purified I-2 and I-4 products (Fig. 1C). The results indicate that, at short incubation times, XendoU cleaves intronic sequences at the same U-rich regions previously identified in vivo and in extracts (17): four I-2 molecules are generated by cleavages at the a, b, c and d sites, while two I-4 molecules are produced by cleavage at two adjacent U residues, 14 nucleotides downstream to U16 (see representation of Fig. 1A). From the reverse transcriptase experiment it appears that the a and b sites are preferentially utilized in the upstream cleavage. As a consequence two major I-1 molecules (a and b) are identified (see gels of Fig.1 and 4).
Efficient cleavage with the purified enzyme was obtained only when Mn 2+ ions were present in the reaction (Fig. 1B, lane 7). It is remarkable that the addition of Mn 2+ ions in the in vitro assay is required for the purified protein and for all the fractions obtained after the blue- indicate that the sequence specificity of XendoU is limited to U-stretches and that only two U residues are sufficient for cleavage ( Fig. 2A). Incubation of XendoU with the P2 oligo in a double stranded configuration, did not produce any cleavage (lanes dsP2), indicating that the enzyme is unable to cleave U-stretches present in double stranded structures.
From these data it appears that, despite the presence of several U-stretches in the U16-host intron upstream and downstream to the snoRNA, specific sites are preferentially cleaved by XendoU. This is similar to what described for the bacterial RNase E endonuclease which, even if has a low primary sequence specificity, cleaves the substrate only at a limited number of sites. In this case, it was shown that the overall secondary structure of the substrate modulated cleavage activity and that the role of stem-loop structures was to limit rather than promote RNase E cleavages (25). Since snoRNAs are folded in specific secondary structures characterized by the conserved "terminal core motif" (see wild type RNA in lower part of Fig.2B) (16,26) and, in many cases, by an apical stem-loop region (27,28), we asked whether these structural elements could influence XendoU activity. Mutants affecting either structures were raised on precursor molecules and tested for XendoU cleavage. The "terminal core motif" was destroyed in the "pre-open stem" and "pre-∆C/bD mutants, while the apical stem-loop structure was deleted in the pre-Ms1 derivative (see schematic representation of Fig. 2B). The pattern of XendoU cleavage on such mutants (Fig. 2B), was not altered suggesting that these structures do not represent entry sites or positioning elements for XendoU. This conclusion was definitely confirmed by the analysis of pre-∆U16, a mutant completely lacking the snoRNA sequence: XendoU cleaves this RNA at the same sites as the wild type RNA substrate, as indicated by the size of the cleavage products (refer to Fig.1A for schematic representation).

Characterization of the reaction products. The chemistry of XendoU cleavage was assessed
by determining the chemical nature of the termini in the cleaved products. The ends of 32 Plabelled I-1b molecules produced with ONE or with the purified XendoU were analyzed; these molecules were gel purified and treated either with HCl or alkaline phosphatase, or with both. Fig. 2C shows that a slight decrease in migration, due to the loss of a negative charge, is obtained only when the alkaline phosphatase follows the HCl treatment (lanes 3). These data allowed to conclude that the 3' end of the cleavage products, obtained with ONE (lanes ONE) and with the purified XendoU (lanes XendoU), carry a 2'-3' cyclic phosphate (23,29). In fact, only after the acid treatment the phosphate group can be removed by phosphatase such as to confer slight decrease in gel mobility. As previously reported (21), the products of primary cleavage such as I-1 molecules, are quite unstable in vivo because, after cleavage, they are rapidly trimmed out. Nevertheless, at very short incubation times, we were able to purify little amounts of I-1b molecules and to subject them to the same treatment described above. Fig. 2C (lanes in vivo) shows that a slight reduction in migration was obtained, demonstrating that also the products of the in vivo reaction have 2'-3' cyclic phosphate ends.
The nature of the 3' ends was also tested by a different approach (24): the I-1b molecule, generated by XendoU cleavage, was ligated to [5'-32 P]pCp directly or after kinase treatment, which removes the 2'-3' cyclic phosphate. The appearance of radioactive band only after kinase treatment (lane 3 of Fig. 2D) confirms that this molecule has 2'-3' cyclic phosphate.

Isolation of a XendoU cDNA.
After elution from the gel, the 37 kDa polypeptide was reduced, alkylated and digested with trypsin as reported in the experimental procedures. The resulting peptide mixture was resolved by reversed-phase HPLC and selected peptide fractions were submitted to automated Edman degradation. Three tryptic peptides (indicated as #1, #2 and #3 in Fig. 3) were utilized to derive degenerate oligonucleotides. These were employed, in different combinations, in PCR amplification reactions on cDNA from polyA + RNA, extracted from X.laevis oocytes. Only the reaction performed with sequence #1 (forward) and sequence #3 (reverse) gave a specific amplification product of 500 nucleotides.
Sequencing of this product indicated the presence of an Open Reading Frame containing peptide #2. This cDNA probe was then utilized for screening a X.laevis stage 28 embryo cDNA library, allowing the isolation of a full-length cDNA (Fig. 3). 65% of the amino acid sequence determined was confirmed by MALDI-MS spectra of the tryptic peptides.

Activity of in vitro translated and recombinant XendoU.
The XendoU ORF, 876 bp long, was cloned into the Blue Script vector and the protein was produced by in vitro transcription and translation, using a reticulocyte lysate. The translation product was analyzed on SDS-PAGE and resulted in a 37 kDa protein (Fig. 4A). In order to assess the activity of this polypeptide, the reticulocyte lysate expressing the XendoU ORF was incubated with 32 P-labeled U16containing precursor. Figure 4B shows that the cleavage pattern produced by the in vitro XendoU also participates in the biosynthesis of another intron-encoded snoRNA. We previously identified a novel box C/D snoRNA, named U86, which is encoded by an intron of the NOP56 gene of X.laevis (18). Similarly to U16 snoRNA, also U86 is contained in a poorly spliceable intron and its biosynthesis appears to be alternative to that of the cotranscribed mRNA. Injection of 32 P-labeled U86-containing precursor into X.laevis oocytes generates the truncated products I-2 and I-3 and their 5' and 3' cut-off molecules, I-1 and I-4 (Fig. 5A, lanes in vivo).
Processing of U86-containing precursor with purified XendoU (Fig. 5A, lanes XendoU) or with the reticulocyte lysate expressing XendoU ORF (Fig. 5C, lane 2) demonstrates that the enzyme is responsible for the cleavage occurring downstream to the U86 coding region. The activity responsible for the cleavage upstream to U86, that produces I-2 and I-1 molecules, is still unidentified and it is not reproduced also in oocyte nuclear 14 by guest on March 24, 2020 http://www.jbc.org/ Downloaded from extracts (Fig. 5A, lanes ONE). The XendoU cleavage sites, downstream to U86, were mapped by primer extension on I-4 molecules and found to localize in correspondence of three U-rich sequences (Fig. 5B).

DISCUSSION
In this paper we report the purification to homogeneity of XendoU, an activity previously shown to be involved in the release of the intron-encoded U16 and U86 snoRNAs from their host primary transcripts in X.laevis oocytes.
XendoU is a novel endoribonuclease in that it requires Mn 2+ ions and produces 2'-3' cyclic phosphate termini. Such ends have been previously associated solely with metalindependent (30,2) or Mg 2+ -dependent endoribonucleases (31,32,33). On the contrary, Mn 2+ -requiring endonucleases usually produce 5'-P and 3'-OH ends (34,35). The chemistry of cleavage of XendoU strongly resembles that of ribozymes, where the metal, positioned near the attacking 2' oxygen, increases its nucleophilicity and allows the transesterification reaction with the production of cyclic 3' ends (36). Until structural data are available, it will not be possible to assess the role of Mn 2+ in the catalytic activity of XendoU. A clear example that metal ions can have a direct role in phosphoryl-transfer reactions in the context of metallo-proteins was derived from the crystal structure of the DNA polymerase I, 3'-5'-exonuclease domain, complexed with single stranded DNA. In this case, two metal ions form complexes with the scissible phosphate and water, facilitating formation of the attacking hydroxide ion and stabilizing the transition state (37,38). The role of the protein component would be exclusively to correctly orient the metal ions, the substrate and the attacking water molecule. By analogy, it could be possible that in the case of XendoU, Mn 2+ might participate directly in the catalytic step, while the protein component may assist the reaction by orienting the ions to specific sites on the substrate. The availability of the active recombinant protein will allow us to answer this question in the near future.
Characterization of XendoU activity indicated that it does not have a stringent sequence specificity in that only two U-residues are sufficient for cleavage; nevertheless, preferential cleavages occur at specific U-stretches localized upstream and downstream to U16 snoRNA.
At prolonged incubations, also the other U-rich regions are cleaved, converting the primary products into small sized RNA species. Since snoRNAs, and in particular U16, have been described to be folded in specific secondary structures characterized by the conserved "terminal core motif" (16,26,39) and, in many cases, by an apical stem-loop region (27,28), we asked whether these elements could provide some structural information for directing the preferential activity of the enzyme. Instead, we have observed that these structural motifs of U16 are not required for positioning XendoU cleavage. This is analogous to the case of RNase E in which McDowall et al. (25) have described that stem-loops do not serve as entry sites for the enzyme, but instead they limit cleavage at potentially susceptible sites, more accessible than others to the nuclease.