Exoribonuclease and Endoribonuclease Activities of RNase BN/RNase Z both Function in Vivo*

Background: In contrast to other RNase Z endoribonucleases, RNase BN is also an exoribonuclease in vitro. Results: Cells dependent on RNase BN for growth are affected by a single amino acid change that eliminates exoribonuclease activity. Conclusion: In addition to its endoribonuclease activity, the exoribonuclease of RNase BN also can participate in tRNA maturation. Significance: RNase BN can serve as a dual function nuclease in vivo. Escherichia coli RNase BN, a member of the RNase Z family of endoribonucleases, differs from other family members in that it also can act as an exoribonuclease in vitro. Here, we examine whether this activity of RNase BN also functions in vivo. Comparison of the x-ray structure of RNase BN with that of Bacillus subtilis RNase Z, which lacks exoribonuclease activity, revealed that RNase BN has a narrower and more rigid channel downstream of the catalytic site. We hypothesized that this difference in the putative RNA exit channel might be responsible for the acquisition of exoribonuclease activity by RNase BN. Accordingly, we generated several mutant RNase BN proteins in which residues within a loop in this channel were converted to the corresponding residues present in B. subtilis RNase Z, thus widening the channel and increasing its flexibility. The resulting mutant RNase BN proteins had reduced or were essentially devoid of exoribonuclease activity in vitro. Substitution of one mutant rbn gene (P142G) for wild type rbn in the E. coli chromosome revealed that the exoribonuclease activity of RNase BN is not required for maturation of phage T4 tRNA precursors, a known specific function of this RNase. On the other hand, removal of the exoribonuclease activity of RNase BN in a cell lacking other processing RNases leads to slower growth and affects maturation of multiple tRNA precursors. These findings help explain how RNase BN can act as both an exo- and an endoribonuclease and also demonstrate that its exoribonuclease activity is capable of functioning in vivo, thus widening the potential role of this enzyme in E. coli.

The biosynthesis of a mature, functional tRNA requires a series of processing steps in which both 5Ј-leader and 3Ј-trailer sequences are removed. The 5Ј-leader sequence is removed by the universal endoribonuclease, RNase P. However, processing of the 3Ј-trailer sequence differs among organisms (1)(2)(3)(4)(5)(6)(7)(8)(9). In those organisms in which tRNA precursors are devoid of a 3Ј-CCA sequence, such as eukaryotes, a single endonucleolytic cleavage by RNase Z at the discriminator nucleotide generates a substrate for CCA addition (5,9). In contrast, in those organisms, such as Escherichia coli, in which the CCA sequence is already present in the tRNA precursors, exonucleolytic trimming generates the mature tRNA (3,10). Thus, the function of RNase Z would appear to be unnecessary in E. coli. Surprisingly, however, an RNase Z homologue, termed RNase BN, is present (11)(12)(13). As a consequence, it is of considerable interest to understand what the function of RNase BN might be.
RNase BN was originally found to be required for the maturation of those T4 tRNA precursors that lacked a CCA sequence at their 3Ј terminus (14). Subsequent studies showed that it is also able to mature CCA-encoded E. coli tRNA precursors when all other processing exoribonucleases are absent, although it is very inefficient in this regard (3,10,15). Moreover, deletion of the gene encoding RNase BN (rbn) in E. coli has no effect on cell growth, and it has even been reported that cells lacking RNase BN do not show any alteration in their transcriptome or proteome profile (16), although this remains controversial (17). Nevertheless, under normal circumstances, RNase BN is unlikely to be involved in tRNA maturation in vivo except in bacteriophage T4-infected cells (13).
RNase BN was initially reported to be an exoribonuclease based on its ability to remove a mononucleotide residue from the 3Ј terminus of certain synthetic tRNA precursors (13), in contrast to other members of the RNase Z family, which act as endoribonucleases. However, in a recent study, we showed that RNase BN is both a distributive exoribonuclease and an endoribonuclease on model RNA substrates (18). It is strongly inhibited by the presence of a CCA sequence or a phosphoryl group at the 3Ј-end of RNA (18). The mode of action of RNase BN on tRNA precursors differs based on whether or not the universal 3Ј-terminal CCA sequence is present. It acts as an endoribonuclease on tRNA precursors that lack a CCA sequence, cleaving after the discriminator nucleotide. In contrast, maturation of tRNA precursors containing a CCA sequence is dependent on the metal ion present; exoribonuclease activity is favored in the presence of Co 2ϩ , whereas endoribonuclease activity is stimulated by Mg 2ϩ (19). RNase BN cleaves after the CCA sequence in its endoribonucleolytic mode, and it trims off extra 3Ј residues up to the CCA sequence when acting as an exoribonuclease. In neither case does it remove the CCA sequence from tRNA (19). Interestingly, no other member of the RNase Z family has been reported to have an exoribonuclease activity in addition to the known endoribonuclease activity. However, the mechanism responsible for the dual activities of RNase BN has not been elucidated. Most importantly, it is not known whether the exoribonuclease activity actually functions in vivo or whether it is simply an in vitro manifestation.
In this study, we identified a structural feature in RNase BN that differs from its RNase Z homologue in Bacillus subtilis. Based on this analysis, we were able to generate a mutant form of RNase BN that lacks almost all exoribonuclease activity but retains its endoribonuclease activity. Expression of this mutant protein in place of wild type RNase BN in E. coli had no obvious effect on wild type cells. However, in cells lacking certain other exoribonucleases and dependent on RNase BN for growth, the presence of the mutant form of RNase BN resulted in even slower growth and affected processing of tRNA precursors. On the other hand, maturation of phage T4 tRNA precursors was unaffected. Biochemical analysis revealed that maturation of the phage T4 tRNA precursors depended solely on the endonucleolytic activity of RNase BN, whereas maturation of host tRNA precursors utilized both activities. Moreover, we found that the endoribonuclease activity of RNase BN cleaved tRNA precursors after the CCA sequence, confirming earlier in vitro findings (19). These data indicate that both the exo-and endoribonuclease activities of RNase BN can function in vivo and thereby expand the possible roles for this enzyme in E. coli.

EXPERIMENTAL PROCEDURES
Materials-T4 RNA ligase, calf intestine alkaline phosphatase, Nucway TM spin columns, and the MEGAshortscript TM kit were purchased from Ambion Inc. T4 polynucleotide kinase, DNase I, and RNase A were obtained from New England Biolabs. ExpressHyb hybridization solution was purchased from Clontech. DpnI was from Fermentas. [␥-32 P]ATP, 5Ј-[ 32 P]pCp, and GeneScreen Plus hybridization transfer membrane were obtained from PerkinElmer Life Sciences. The GeneElute TM PCR clean-up kit and bis(p-nitrophenyl) phosphate were from Sigma. The KOD Hot Start DNA polymerase was purchased from Novagen. Sequagel for denaturing urea-polyacrylamide gels was obtained from National Diagnostics. The His-Trap HP column was obtained from GE Healthcare. All other chemicals were reagent grade.
Site-directed Mutagenesis, Overexpression, and Purification of the Mutant Protein-Cloning of the wild type rbn gene into the pET15b plasmid and its overexpression and purification were described previously (12). The mutation P142G was introduced into a plasmid-encoded rbn gene by site-directed mutagenesis using primers D1 and D2 (supplemental Table S1). KOD hot start DNA polymerase was used in the PCR. A template plasmid containing the wild type rbn gene was then digested by DpnI treatment at 37°C for 2 h. Mutant rbn-containing plasmids were purified by gel extraction using a gel extraction kit (Qiagen) and transformed into E. coli BL21(DE3)I Ϫ II Ϫ . Mutations of P142Q and L143V were also introduced into a plasmid-encoded rbn gene in the same manner using primers P1 and P2 (for P142Q) and R1 and R2 (for L143V) (supplemental Table S1). Mutant proteins were overexpressed in E. coli strain BL21(DE3)I Ϫ II Ϫ /pLys and purified using the same procedure used previously for purification of wild type His-tagged RNase BN (12,18). The purity of wild type and mutant RNase BN proteins was determined on an overloaded SDS-polyacrylamide gel (ϳ3.0 g of the purified protein). For all of the proteins, a single band at ϳ35 kDa was observed without any detectable minor contaminating bands.
Bacterial and Phage Strains and Growth Conditions-Wild type E. coli K-12 strain MG1655(Seq)rph ϩ was obtained from laboratory stock, and its derivative MG1655(Seq)rph ϩ mutS Ϫ was a gift from Dr. Richard Myers (University of Miami). Deletions of the genes encoding RNase I, II, D, or T were introduced into strain MG1655, which is RNase PH Ϫ , by transduction with phage P1vir. Mutagenesis of proline 142 to glycine of RNase BN in strain MG1655(Seq)rph ϩ mutS Ϫ was performed by recombineering using oligonucleotide B1 (supplemental Table S1) (20). Recombinants were selected by PCR using primers T1 and T2 (supplemental Table S1). A kanamycin resistance gene was also introduced by recombineering after the terminator sequence of the chromosomal rbn gene (20). The mutant rbn gene along with the adjacent kanamycin resistance marker from strain MG1655(seq)rph ϩ mutS Ϫ was introduced into wild type MG1655(seq)rph ϩ or into MG1655 I Ϫ II Ϫ D Ϫ T Ϫ PH Ϫ . Each of these strains was then transformed with the temperature-sensitive plasmid pCP20. The kanamycin resistance gene was then flipped out by activation of the flippase encoded in plasmid, pCP20. The P142G mutation in each strain was confirmed by DNA sequencing. Wild type phage T4 and its mutant derivative BU33 were obtained from laboratory stock. BU33 contains an amber mutation in its head protein and a suppressor function for tRNA Ser as described (14,(21)(22)(23). A cca mutation containing the chloramphenicol resistance marker was introduced into strains MG1655 I Ϫ II Ϫ D Ϫ and MG1655 I Ϫ II Ϫ D Ϫ P142G by transduction with phage P1vir. The rnt mutant gene was subsequently introduced into these strains by P1 transduction.
Synthesis and 3Ј-End Labeling of tRNA Precursors-E. coli and phage T4 tRNA precursors were synthesized from DNA templates of tRNA genes in in vitro transcription reactions using the MEGAshortscript TM transcription kit as described previously (19). All templates were synthesized from genomic DNAs by PCR with the forward primer containing the T7 RNA polymerase promoter sequence. PCR products were purified using the GeneElute TM PCR clean-up kit. Precursor tRNAs were purified by phenol/chloroform/isoamylalcohol (25:24:1) extraction followed by ethanol precipitation as described.
5Ј-[ 32 P]pCp and T4 RNA ligase were used to label the 3Ј-end of tRNA precursors in the presence of unlabeled ATP at 4°C for 16 h as described previously. Unincorporated 5Ј-[ 32 P]pCp was removed using a Nucway TM spin column, and calf intestine alkaline phosphatase was used to remove the 3Ј-terminal phos-phate from 3Ј-[ 32 P]pCp-labeled tRNAs. Upon dephosphorylation, 3Ј-end-labeled tRNAs were purified as described (19).
RNase BN Assay-3Ј-[ 32 P]pC-labeled tRNAs (ϳ0.05 M) or 5Ј-32 P-labeled model RNA substrates (10 M) and 0.14 M purified wild type or mutant RNase BN, except as otherwise stated in the figure legends, were incubated at 37°C in a 30-l reaction mixture that contained either 20 mM HEPES, pH 6.5, 200 mM potassium acetate, and 0.2 mM CoCl 2 or 10 mM Tris-HCl, pH 7.5, 200 mM potassium acetate, and 5 mM MgCl 2 (19). Portions were taken at the indicated times, and the reaction was terminated by the addition of 2 volumes of gel loading buffer (90% formamide, 20 mM EDTA, 0.05% SDS, 0.025% bromphenol blue, and 0.025% xylene cyanol). Reaction products were resolved on 20% denaturing 7.5 M urea polyacrylamide gels and visualized using a STORM 840 phosphorimaging device (GE Healthcare). ImageQuant (GE Healthcare) was used to quantitate the bands.
Phosphodiesterase Assay-Bis(p-nitrophenyl) phosphate was used as substrate to determine the phosphodiesterase activity of RNase BN. Standard reaction conditions were 20 mM Tris-HCl (pH 7.4), 2 mM substrate, 2.0 g of purified His-tagged wild type or mutant RNase BN, and 5 mM MgCl 2 . Release of p-nitrophenol (⑀ ϭ 11,500 M Ϫ1 cm Ϫ1 at pH 7.4) was continuously monitored for 3 min at 405 nm. One unit of activity corresponds to 1 mol of p-nitrophenol liberated/min at 37°C.
Phage Assay-Fifty microliters of overnight culture of wild type or mutant E. coli cells were incubated with 10 l of either bacteriophage T4 or T4 mutant strain BU33 (10 8 plaque-forming units/l) at 37°C for 5 min. The suspension was then overlaid onto LB plates with 2.5 ml of top agar. Plates were incubated overnight at 37°C prior to counting plaques.
RNA Preparation and Northern Blotting-Cells were grown in YT medium to an A 600 of ϳ1.0. Total cellular RNA was isolated by phenol extraction as described (24). RNA samples from wild type and mutant strains, containing the same amount of RNA, were dissolved in gel loading buffer and loaded on a 6% denaturing 7.5 M urea-polyacrylamide sequencing gel. The gel was run at 1000 V until the xylene cyanol dye had migrated ϳ30 cm. The RNA was transferred to a GeneScreen plus membrane by horizontal transfer for 1.5 h at 150 mA using 0.5ϫ Tris borate/EDTA as the transfer solution. T4 polynucleotide kinase was used to prepare a 5Ј-32 P-labeled DNA oligonucleotide probe complementary to the 5Ј-end of the tRNA. Hybridization of the probe to the transferred RNA was carried out by overnight incubation in ExpressHyb hybridization solution, and the detected bands were visualized by PhosphorImager analysis (GE Healthcare).

RESULTS
In earlier work (18,19), we showed that RNase BN can mature E. coli tRNA precursors in vitro using either an exo-or endoribonucleolytic mode of action. However, it has been unclear how the enzyme carries out this dual mode of action. It is also not known whether both activities actually function in vivo. Because the ability of RNase BN to function as both an exo-and an endoribonuclease differs from other members of the RNase Z family, we reasoned that comparison of RNase BN with another RNase Z member might shed light on the explanation for their different catalytic properties.
Structural Comparison of RNase BN and an RNase Z-The crystal structures of E. coli RNase BN and its B. subtilis homologue (RNase Z) have been solved (25)(26)(27). Each enzyme is a dimer containing a core zinc-dependent ␤-lactamase domain with a HXHXDH metal binding motif as well as additional His and Asp residues that contribute coordination sites to the metal ions. Each of the subunits possesses a protruding flexible arm that is believed to have a role in tRNA binding (25)(26)(27)(28). Both subunits of RNase BN from E. coli appear to be identical, whereas those of B. subtilis RNase Z maintain a different conformation despite having identical sequences (25,27).
To identify structural differences between the E. coli and B. subtilis proteins that might account for their catalytic differences, we superimposed their crystal structures as shown in Fig.  1. The region encompassing the two metal binding sites, the putative site of catalysis, also contains an RNA-binding channel (9,26). RNA bound in this channel is presumed to be cleaved at the site of metal binding. The cleavage product would then diffuse away through the putative exit channel, which is the portion of the RNA-binding channel that lies downstream of the catalytic site and through which the cleavage product will leave the enzyme. For endonucleolytic cleavage, we propose that the RNA molecule would lie in the RNA binding channel in a manner such that its 3Ј-end extends well into the exit channel. As a consequence, the catalytic site would access and cleave an internal phosphodiester bond. On the other hand, if the 3Ј-end of the RNA cannot extend beyond the catalytic site, exonucleolytic cleavage of the terminal 3Ј-nucleotide would be favored. Thus, we suggest that the position of the RNA substrate, deter- mined by the extent of its entry into the putative exit channel, would result in either endonucleolytic or exonucleolytic action on the RNA chain. Structural features that impede or promote RNA access into the exit channel would thereby favor one or the other catalytic activity.
Comparison of the putative exit channel regions of E. coli RNase BN and B. subtilis RNase Z (9) indicates that in the E. coli protein, a loop extending into the exit channel (shown in yellow in Fig. 1) just downstream of the catalytic site narrows the exit channel more than the corresponding loop in the B. subtilis protein (red). This loop in E. coli RNase BN also contains a proline (Pro-142) which should make it much more rigid than the B. subtilis loop, which has a glycine at the corresponding position. Other regions around the catalytic site of the E. coli and B. subtilis proteins are otherwise very similar. Thus, we hypothesized that the loop and especially proline 142, which could impede an RNA substrate from entering into the exit channel, might be responsible for promoting exonucleolytic activity in E. coli RNase BN.
Isolation and Activity of RNase BN Mutant Proteins-To examine the importance of the putative exit channel loop and its flexibility on the activity of RNase BN, we generated several mutations within the loop and purified the resulting proteins. These included conversion of Pro-142 to either glycine or glutamine and a change of Leu-143 to Val. The mutations of Pro-142 would be expected to increase the flexibility of the loop and to also substitute this residue either with the small glycine residue present in the B. subtilis protein or with a relatively large residue, glutamine. The Leu-143 to Val change was expected to slightly widen the exit channel and to convert this residue to that present in the B. subtilis enzyme. As an initial screen, the mutant proteins were compared with wild type RNase BN for its three known activities (i.e. endoribonuclease, exoribonuclease, and phosphodiesterase) ( Table 1) using tRNA SelC precursor as substrate for the nuclease activities and bis-p-nitrophenyl phosphate as substrate for the diesterase activity.
The data presented in Table 1 show that all of the loop mutants retain or even increase their endoribonuclease activity. In contrast, exoribonuclease activity is essentially eliminated in the two Pro-142 mutant proteins and is reduced about 3-fold in the L143V mutant protein. Interestingly, although exoribonuclease activity is dramatically reduced, phosphodiesterase activity is unaffected in all of the mutants. These initial assays support a model in which mutation of Pro-142 increases the flexibility of the loop, thereby effectively widening the exit channel and enabling the 3Ј-end of the tRNA precursor substrate to bind mainly within the exit channel, which promotes the endoribonuclease activity and inhibits exoribonuclease activity. A similar effect was seen with the L143V mutant protein, which also would be expected to widen the exit channel, but the effect is less pronounced. Because the small substrate used in the phosphodiesterase assay is probably confined to the vicinity of the catalytic site, this activity is unaffected by mutations in the putative exit channel. Because both Pro-142 mutant proteins were essentially the same with regard to loss of exoribonuclease activity, all subsequent studies were carried out with the P142G mutation.
Effect of P142G Mutation in RNase BN on Its Catalytic Activity-To further analyze the apparent loss of exoribonuclease activity in the P142G mutant protein, it was compared with the wild type enzyme using a variety of known RNase BN substrates. We first tested two synthetic RNA substrates, A 17 and G 5 A 12 CCA-A 5 , previously shown to be acted on by RNase BN in an exonucleolytic manner (18). However, as shown in Fig.  2, these substrates were resistant to the P142G mutant protein, whereas they were active substrates of the wild type enzyme, confirming that the P142G mutant enzyme is essentially devoid of exoribonuclease activity.
The wild type and mutant RNase BN proteins were further analyzed using 3Ј-[ 32 P]pC-labeled tRNA SelC and tRNA PheV precursors from E. coli as substrates (prepared as described under "Experimental Procedures"). With pre-tRNA SelC as substrate (containing 5 nt 2 following the CCA sequence), wild type RNase BN generates approximately equal amounts of exo-and endoribonucleolytic cleavage products in the presence of Mg 2ϩ . In contrast, essentially only the endonucleolytic cleavage product is produced by mutant RNase BN under the same conditions (Fig. 3A). In the presence of Co 2ϩ , which strongly favors the exoribonucleolytic activity of RNase BN (18), both the wild type and mutant enzymes were able to remove the 3Ј-terminal mononucleotide from the tRNA SelC precursor (Fig. 3A). However, even in this sensitive assay, the mutant RNase BN exoribonuclease activity was reduced ϳ20-fold compared with the wild type enzyme.
The second E. coli tRNA precursor examined, tRNA PheV , which contains an extra 7 residues following the CCA sequence, is matured primarily by the endoribonuclease activity of RNase BN (19). As shown in Fig. 3B, mutant RNase BN also acts primarily endoribonucleolytically on this substrate in the presence of Mg 2ϩ . However, when the assay is carried out in the presence of Co 2ϩ , major differences between wild type and mutant enzymes are observed (Fig. 3B). Whereas the wild type protein now produces essentially only the exoribonucleasegenerated CMP product, mutant RNase BN continues to act primarily as an endoribonuclease with only a very low level of exoribonuclease activity, even under conditions that strongly promote this mode of action. Thus, for all substrates tested, and even in the presence of Co 2ϩ , conversion of Pro-142 to Gly in RNase BN strongly inhibits its exoribonuclease activity. Based on these observations, we anticipate that the P142G mutant 2 The abbreviation used is: nt, nucleotide(s). RNase BN would function as an endoribonuclease on tRNA precursors in vivo.

Construction of Strains Lacking the Exoribonuclease Activity of RNase BN-Although
RNase BN displays both exo-and endoribonucleolytic activities in vitro, it is not known whether both activities can function in vivo. The availability of a mutant RNase BN that lacks the exoribonuclease activity enabled us to address this question. Accordingly, the P142G mutation was first introduced into strain MG1655(seq)rph ϩ by recombineering (20). The mutant rbn gene was subsequently transferred into a strain lacking RNases I, II, D, T, and PH. To ensure that the mutant gene was expressed normally, we compared the level of phosphodiesterase activity in the rbn ϩ and P142G derivatives of the multiple RNase-deficient strain. Because the phosphodiesterase activity is unaffected by the P142G mutation, the amount of this activity is a direct measure of the amount of RNase BN protein present in each strain.
As can be seen in supplemental Table S2, there is no significant difference in the level of phosphodiesterase activity between extracts of the wild type and the P142G mutant strain. Moreover, based on the small amount of phosphodiesterase activity remaining in a rbn null strain, ϳ90% of the activity in the extract is due to RNase BN. Thus, based on these data, we conclude that P142G RNase BN is expressed normally and is active.
Effect of Pro-142 to Gly Mutation on the Maturation of Phage T4 tRNA Precursors in Vivo-One known function of RNase BN is the maturation of those phage T4 tRNA precursors that lack a 3Ј-CCA sequence (14,22). Thus, an E. coli strain deficient in RNase BN is unable to support growth of mutant T4 phage strain BU33 due to its inability to process the 3Ј-end of a required phage-encoded suppressor tRNA Ser (14). However, it is not known whether the exo-or endoribonucleolytic activity of RNase BN is important for this action on phage T4 tRNA  P]pC at its 3Ј-end was treated with wild type or mutant (P142G) RNase BN (0.14 M) in the presence of either Mg 2ϩ at pH 7.5 or Co 2ϩ at pH 6.5. Cleavage products were analyzed by 20% denaturing PAGE. The positions of the 5-nt endoribonucleolytic cleavage product (Endo-5Ј nt) and the mononucleotide generated as a result of exoribonucleolytic trimming are indicated. B, the structure of the E. coli tRNA PheV precursor is shown with the 3Ј-CCA sequence in boldface type. 3Ј-[ 32 P]pC-labeled tRNA PheV was treated with wild type or mutant RNase BN and analyzed as described in A. The 7-nt endoribonucleolytic cleavage product (Endo-7Ј nt) and the mononucleotide generated as a result of exoribonucleolytic trimming are indicated. Identification of the reaction products was determined with RNA oligonucleotides and with CMP as standards. The mononucleotide exonuclease product was also confirmed to be CMP by paper chromatography, as described (18,19). OCTOBER 12, 2012 • VOLUME 287 • NUMBER 42 precursors in vivo. To examine this question, wild type phage T4 and T4 phage BU33 were each plated on wild type E. coli, on an RNase BN-deficient strain, or on the strain that is deficient in the exoribonuclease activity of RNase BN. The data presented in Table 2 show that wild type phage T4 is able to form plaques on all three strains with equal efficiency. Likewise, T4 phage BU33 generates nearly the same number of plaques on wild type and on the P142G mutant strain. In contrast, a strain with a deletion of the rbn gene cannot support BU33 growth. These data show that elimination of the exoribonuclease activity of RNase BN does not affect the plating efficiency of either wild type T4 phage or phage BU33, suggesting that the exoribonuclease activity of RNase BN is not required for the maturation of phage T4 tRNA Ser .

Activities of RNase BN in Vivo
To obtain further support for this conclusion, we prepared by in vitro transcription the T4 phage dimeric tRNA Pro-Ser precursor (Fig. 4), known to be a substrate for 3Ј-processing by RNase BN in vivo (14,22,23). The substrate was labeled by ligation of [ 32 P]pCp to its 3Ј-terminus, followed by removal of the 3Ј-phosphoryl group by phosphatase. The resulting labeled precursor was then treated with either purified wild type RNase BN or the P142G exoribonuclease-deficient mutant enzyme, and the products were analyzed on a 20% denaturing acrylamide gel. The major product produced with both enzymes is a tetranucleotide, with no [ 32 P]CMP evident (Fig. 4). These data confirm that RNase BN removes the extra 3Ј-terminal nucleotides from the dimeric precursor in an endoribonucleolytic manner. Note also that endoribonuclease activity is elevated in the mutant enzyme, amounting to ϳ5-fold with this substrate.
Effect of Pro-142 to Gly Mutation on the Maturation of E. coli tRNA Precursors in Vivo-It is unlikely that RNase BN is involved in the maturation of E. coli tRNA precursors under normal conditions because it is less efficient in removing extra residues from the 3Ј-end of pre-tRNAs than other tRNA processing enzymes present in the cell. However, in the absence of the other tRNA 3Ј-processing enzymes, tRNA maturation becomes dependent on RNase BN (15), providing a system to determine the effect of removing its exoribonuclease activity. Thus, in a genetic background in which RNases I, II, D, T, and PH are absent, we performed Northern blot analysis on multiple tRNAs in strains containing or lacking the exoribonuclease activity of RNase BN. Fig. 5 shows the results obtained with one of these tRNAs, tRNA PheV . tRNA precursors do not accumulate to a detectable level in the wild type strain in which all of the tRNA processing RNases are present. The amount of tRNA precursors increases to ϳ5% when the exoribonucleases, RNase II, D, T, and PH, are removed. Additional removal of the exoribonuclease activity of RNase BN leads to increased accumulation of the PheV precursor, amounting to ϳ30% of the total tRNA PheV species present. These observations indicate that the exoribonuclease activity of RNase BN participates in maturation of the tRNA PheV precursor in vivo.  A similar analysis was carried out for nine additional tRNAs ( Table 3). As can be seen, five of these nine tRNAs display a significant increase in the amount of tRNA precursor present when RNase BN lacks its exoribonuclease activity. These data show that the exoribonuclease activity of RNase BN can function in vivo and can contribute to the maturation of tRNA precursors. However, because mature tRNA is present for all of the tRNAs studied, it is likely that the intact endoribonuclease activity of RNase BN provides a significant portion of overall tRNA processing activity under these special conditions. Moreover, these data also indicate that the relative importance of the endo-and exoribonuclease activities of RNase BN varies depending on the identity of the tRNA precursor examined. The reasons for these differences among precursors are not understood, but it is well known that tRNA precursors respond very differently to removal of individual processing exoribonucleases (3).
Effect of the Pro-142 to Gly Mutation in RNase BN on Cell Growth-Deletion of the gene encoding RNase BN (rbn gene) has no effect on growth of E. coli. However, its presence is required in a cell lacking other tRNA 3Ј-end processing exoribonucleases (II, D, T, and PH) because additional deletion of the rbn gene makes the cell inviable. Although the presence of RNase BN maintains viability, cells grow poorly, with a doubling time of 75-120 min, depending on the genetic background (15). Nevertheless, the requirement for RNase BN provided an opportunity to investigate the effect of the P142G mutation on RNase BN function under these conditions. Substitution of the P142G mutation for the chromosomal copy of the rbn gene in wild type MG1655(seq)rph ϩ revealed no difference between the growth rates of wild type and mutant strains, indicating that the P142G mutation has no effect on growth of wild type cells (data not shown). In contrast, introduction of the P142G mutation into a cell devoid of the other tRNA 3Ј-end processing exoribonucleases (MG1655 I Ϫ II Ϫ T Ϫ PH Ϫ D Ϫ ) results in an increased doubling time of the already slow growing cells (Fig. 6).
As noted in Table 1, in addition to loss of its exoribonuclease activity, the P142G mutant also displays increased endoribonuclease activity on some substrates. Thus, although unlikely, it was possible that stimulation of the endoribonuclease activity somehow was responsible for the slower growth. To test this possibility, we overexpressed the P142G mutant RNase BN in the 3Ј-exoribonuclease-deficient strain and found that the cell's growth rate actually increased by ϳ10 min (data not shown). Based on this observation, it is very unlikely that any increase in the endoribonuclease activity of mutant RNase BN would cause slower growth of the P142G mutant cell. Taken together, these results indicate that removal of the exoribonuclease activity of RNase BN slows cell growth when it is dependent on the presence of RNase BN, providing additional evidence that the exoribonuclease activity can function in vivo.
RNase BN Cleaves tRNA Precursors after the CCA Sequence in Vivo-As noted above, the endoribonuclease activity of RNase BN processes tRNA precursors in vivo, but its site of cleavage was not known. RNase BN does not remove the CCA sequence from either precursor or mature tRNAs in vitro (18), and it was of interest to determine whether the same mode of action is operative in vivo. To answer this question, we introduced a cca mutation by P1 transduction eliminating the enzyme, tRNA nucleotidyltransferase. We hypothesized that if RNase BN cleaves tRNA precursors after the discriminator nucleotide, tRNA nucleotidyltransferase would be required to add back the CCA sequence to generate a mature tRNA, whereas this enzyme would not be needed if the cleavage were after the CCA sequence. We found that introduction of a cca mutation into cells containing wild type RNase BN or containing the P142G mutant RNase BN and lacking other 3Ј-processing exoribonucleases has no effect on the growth rate (data not shown), strongly suggesting that cleavage is after the CCA sequence in each case. To confirm these data, Northern blot analysis was carried out on a series of tRNAs. As can be seen, the removal of processing nucleases leads to accumulation of tRNA precursors. However, none of the tRNAs tested show any shorter species devoid of the CCA sequence in the tRNA nucleotidyltransferase mutants (Fig. 7). Moreover, the removal of tRNA nucleotidyltransferase from the RNase-deficient strain has no effect on the amount of mature tRNA generated. These results indicate that RNase BN does not remove the CCA sequence from tRNA in vivo and, therefore, must cut after the CCA sequence.

DISCUSSION
RNase BN, unlike other RNase Z homologues, has dual endoand exoribonuclease activities (18). In this study, we examined the role of each of these activities in vivo. We showed that (a) conversion of a single proline residue to glycine or glutamine in the putative exit channel essentially eliminates the exoribonuclease activity of RNase BN in vitro; (b) the exoribonuclease activity of RNase BN is not required for maturation of phage T4 tRNA precursors in vivo; (c) the absence of RNase BN exoribonuclease activity results in slower growth of a cell that lacks other processing exoribonucleases; (d) the exoribonuclease activity of RNase BN participates in E. coli tRNA processing; and (e) the endoribonuclease activity of RNase BN cleaves after the CCA sequence in E. coli tRNA precursors.
Based on the known crystal structures of E. coli RNase BN (27), which displays both endo-and exoribonuclease activities and of B. subtilis RNase Z (25), which is only an endoribonuclease (5), we suspected that the observed differences in the width and flexibility of their putative exit channels might play a role in their different catalytic properties. Accordingly, we individually mutated two residues within a loop protruding into this channel of RNase BN. As predicted, the resulting mutant RNase BN proteins had reduced or essentially no exoribonuclease activity, although endoribonuclease activity remained intact or even increased slightly. These findings provide important clues about the structural basis for the distinct catalytic properties of exo-and endoribonucleases and warrant more detailed investigation. However, for the purpose of the present study, the generation of a mutant RNase BN, essentially devoid of exoribonuclease activity, enabled us to examine the functional role of each of the activities of RNase BN in vivo.
Elimination of the exoribonuclease activity of RNase BN did not affect the growth of a wild type strain. Undoubtedly, the other exoribonucleases present are able to take over any exoribonuclease function of RNase BN. However, in a strain that lacks the other 3Ј-processing exoribonucleases, the absence of the exoribonuclease activity of RNase BN causes growth to slow even further and also affects maturation of certain tRNA precursors. This is the only example in which the exoribonuclease activity of an RNase Z homologue has been shown to have a function in vivo. In contrast, the exoribonuclease activity of RNase BN is not required for maturation of phage T4 tRNA precursors, the major known function of this RNase, even when the other tRNA-processing exoribonucleases are absent. This finding is consistent with our previous in vitro observations that RNase BN acts primarily as an endoribonuclease on tRNA precursors that lack a CCA sequence at their 3Ј-end (19). In fact, the data presented here indicate that RNase BN can act as an endoribonuclease even on CCA-containing tRNA precursors inasmuch as an RNase-deficient cell is able to survive in the absence of the RNase BN exoribonuclease activity. Thus, of the tRNA precursors tested, the endoribonuclease activity of RNase BN is able to generate as much as 40 -90% of the normal levels of mature tRNAs when the exoribonuclease activity is absent. Moreover, complete removal of RNase BN in a strain lacking the other processing exoribonucleases leads to inviability, indicating that no other tRNA 3Ј-processing activity is present (15).
Interestingly, although all RNase Z homologues exhibit endoribonuclease activity, their site of action on tRNA precursors may differ. Eukaryotic members of the family all cleave after the discriminator nucleotide (9), as does B. subtilis RNase Z (5). In eukaryotic cells and in those tRNA precursors that are the substrates of RNase Z in B. subtilis, no CCA sequence is present. In contrast, both the Thermotoga maritima and E. coli enzymes cleave after the CCA sequence (7,19) (present results), and in these organisms, all but one (in T. maritima) tRNA precursor contain an encoded CCA sequence. Based on this information, it appears that the catalytic properties of the organism's RNase Z have co-evolved with whether or not the CCA sequence is encoded, because it would be energetically wasteful to first remove and then resynthesize the CCA sequence. In fact, a CCA sequence strongly inhibits both the exo-and endoribonuclease activities of the E. coli enzyme (18,19). However, the mechanism by which a CCA sequence inhibits catalysis by RNase BN remains to be determined.
Although RNase BN is the only known member of the RNase Z family that acts as both an exo-and an endoribonuclease, two other RNases belonging to the ␤-lactamase family were reported to have such a dual function (29,30). These are RNase J from B. subtilis (29) and CPSF 73 (30) from mammalian cells, an RNase that functions in histone mRNA processing. How- FIGURE 7. Northern blot analysis of tRNA Cys , tRNA Tyr , and tRNA Ala . RNA samples were prepared as described under "Experimental Procedures." Eight micrograms of total RNA were subjected to electrophoresis and then transferred to a GeneScreen Plus membrane. Transfer RNAs were detected using a 32 P-labeled 5Ј-end-specific probe. M, mature RNA substrate. ever, in contrast to RNase BN, which acts 3Ј to 5Ј, each of these enzymes functions as a 5Ј to 3Ј exoribonuclease. Moreover, both RNase J1 and CPSF 73, unlike RNase BN, are essential for cell survival under normal conditions.
The results presented here add considerably to our knowledge of RNase BN and of its capabilities and substrate specificity in vivo. However, its primary function in E. coli remains unknown. The fact that it can function both as an exo-and endoribonuclease opens up additional possibilities for its role in vivo.