Multiple Exoribonucleases Catalyze Maturation of the 3′ Terminus of 16S Ribosomal RNA (rRNA)*

Background: The RNases involved in 3′ maturation of E. coli 16S rRNA were not known. Results: E. coli mutants lacking RNase II, RNase R, PNPase, and RNase PH accumulate 17S rRNA precursor. Conclusion: Four known exoribonucleases are responsible for 3′ processing of 16S rRNA. Significance: The maturation pathway for E. coli 16S rRNA has been determined. Processing of ribosomal RNA (rRNA) precursors is an important component of RNA metabolism in all cells. However, in no system have we yet identified all the RNases involved in this process. Here, we show that four 3′→5′-exoribonucleases, RNases II, R, and PH, and polynucleotide phosphorylase (PNPase), participate in maturation of the 3′ end of 16S rRNA. In their absence, 16S precursor molecules with 33 extra 3′-nt accumulate; however, the presence of any one of the four RNases is sufficient to allow processing to occur, although with different efficiencies. Additionally, we find that in the absence of 3′ maturation, 5′ processing proceeds much less efficiently. Moreover, mutant 30S particles, containing immature 16S rRNA, form 70S ribosomes very poorly. These findings, together with the earlier discovery that RNases E and G are the 5′-processing enzymes, completes the catalogue of RNases involved in maturation of Escherichia coli 16S rRNA.

In the case of 16S rRNA, the extra 115 nt 2 at the 5Ј end of the molecule are removed in a two-step process that involves cleavage by the endoribonuclease RNase E, at a position 66 nt upstream of the 5Ј end, followed by the action of a second endoribonuclease RNase G, which cleaves at the mature 5Ј terminus (5,6). Thus, 5Ј processing of 16S rRNA involves an ordered sequence of events, and the RNases involved have been identified.
In contrast, very little is understood about maturation of the 3Ј terminus of 16S rRNA. It is known that 3Ј maturation can proceed in the absence of 5Ј maturation (5). It is also known that the 33 extra 3Ј-residues are removed rapidly such that no intermediates with less than 33 nt are observed. This finding suggested that 3Ј maturation might be an endonucleolytic process, although the action of a 3Ј-5Ј processive exoribonuclease is also possible. In fact, in Pseudomonas syringae, the processive exoribonuclease RNase R is required for 3Ј maturation of 16S rRNA in the cold, although RNase R is not essential at room temperature (7). An exoribonuclease has also been implicated in the 3Ј maturation of rRNAs in chloroplasts of Arabidopsis thaliana (8). However, it is not clear whether these are unusual examples or whether exoribonucleolytic processing of the 3Ј terminus of 16S rRNA is a widespread phenomenon.
In this study, we examine 3Ј processing of 16S rRNA in E. coli. We show that any one of several, known exoribonucleases can carry out removal of the 33 extra 3Ј-nucleotides. Most important are the three 3Ј-5Ј processive exoribonucleases, RNase II, RNase R, and polynucleotide phosphorylase (PNPase). However, in their absence, 3Ј processing continues at a slow rate, largely due to RNase PH. Mutant strains lacking all four exoribonucleases are essentially unable to generate mature 16S rRNA. Such cells accumulate precursors that retain not only the 33-nt 3Ј trailer sequence, but also their 5Ј leader. These data demonstrate that, as with maturation of tRNAs in E. coli (9), 3Ј maturation of 16S rRNA makes use of multiple exoribonucleases.

MATERIALS AND METHODS
Bacterial Strains-E. coli MG1655*(seq) I Ϫ was used as wild type for this study (10). MG1655 I Ϫ was used as the RNase PH Ϫ strain into which RNase R, RNase II, and PNPase mutations were introduced as described previously (10).
Site-directed RNase H Cleavage-Total RNA was isolated using a hot phenol method (11) and subjected to RNase H assay followed by Northern blot analysis (5,12). The chimera C16S3 (5Ј-CCdCdGdAdAGGUUAAGCUACCU-3Ј) was used for 3Ј end cleavage of 16S rRNA, and cleavage products were visualized by using a probe complementary to residues 1502-1523. Chimera C16S5 (5Ј-CAdTdGdTdGUUAGGCCUGCCG-3Ј) was used for 5Ј end cleavage and visualized by a probe complementary to residues 18 -36.
Sucrose Gradient Analysis of Ribosomes-Cells were grown at 37°C until the A 600 ϭ ϳ0.4 and then shifted to 44°C for 1 h. Cells were collected by centrifugation and resuspended in buffer A (10 mM Tris-Cl, pH 7.5, 10 mM MgCl 2 , 60 mM KCl, and 1 mM DTT) containing 2 mg/ml lysozyme, and the lysate was clarified by sonication followed by centrifugation. The supernatant fraction was spun at 55,000 rpm for 2 h. Ribosome pellets were resuspended and layered on a 10 -30% sucrose gradient in buffer B (10 mM Tris-Cl, pH 7.5, 60 mM NH 4 Cl, and 1 mM DTT) containing 0.1 mM MgCl 2 for analysis of 30S and 50S subunits. For analysis of 70S particles, ribosomes were layered on a 14 -40% sucrose gradient made in buffer B containing 10 mM MgCl 2 . The gradient samples were centrifuged for 19 h at 21,000 rpm. Fractions were collected and quantified by A 260 measurement. The ribosomal fractions were subjected to Northern blot analysis using probe P16S3 (5Ј-TGTGAG-CACTGCAAAGAACGC-3Ј) for detecting 3Ј precursor 16S rRNA and probe 16S (5Ј-CCATGGTGTGACGGGCGGTG-3Ј) for detecting mature 16S rRNA.

RESULTS
Exoribonucleases Participate in 3Ј Processing of 16S rRNA-To assess whether exoribonucleases might play a role in 16S rRNA processing, we made use of mutant strains lacking one or more known exoribonucleases. Total RNA was isolated from each of 10 strains and examined by Northern blot analysis using one probe directed against the mature portion of 16S rRNA and a second probe directed against the 33-nt precursor region. From the ratio of precursor to mature signal, we could determine whether any mutant strain accumulated more precursor than wild type cells. Based on this preliminary screening (data not shown), we observed that cells deficient in certain combinations of exoribonucleases showed an elevated level of precursor. These data focused our attention on the processive exoribonucleases, RNase II, RNase R, and PNPase, as well as on RNase PH. No evidence was obtained for the involvement of RNases D, BN, or T. Nevertheless, these data suggested that 3Ј processing of 16S rRNA might require certain exoribonucleases and that more detailed examination was warranted.
Four Known Exoribonucleases Contribute to 3Ј Processing of 16S rRNA-To confirm the involvement of the exoribonucleases identified by the screening assay, we constructed a quadruple mutant strain lacking RNases II, R, and PH and containing a temperature-sensitive PNPase. This was necessary because double mutant strains lacking RNase II and PNPase (13) or RNase R and PNPase (14,15) are inviable, and cells lacking RNase PH and PNPase are slowed in growth (16). The mutant strain was grown at 37°C and transferred to 44°C for 1 h to inactivate PNPase such that the maturation of 16S RNA could be examined under conditions in which the activities of four exoribonucleases were lacking. Using RNase H digestion to more easily distinguish precursor from mature 16S rRNA, we compared the amount of precursor in the quadruple RNase mutant strain with that in wild type using Northern blot analysis. The data in Fig. 1A show that there is essentially no precursor (2%) in wild type cells (lane 1), whereas in cells lacking RNases II, R, and PH and PNPase, close to 90% of 16S rRNA contains a 33-nt 3Ј extension (lane 2). These data demonstrate that exoribonucleases catalyze 3Ј maturation of 16S rRNA, and because no intermediate between the ϩ33 and mature species is evident, demonstrate that they act processively on the 33-nt extension.
To assess the role of each enzyme in the process, we utilized four triple mutant strains in which the activity of one of the four missing RNases was restored. These strains were compared with wild type and with the quadruple RNase-deficient mutant strain (Fig. 1A). The data show that reintroduction of any one of the four exoribonucleases increased the amount of 3Ј mature 16S rRNA as compared with the quadruple mutant strain. RNase II and RNase R were most effective, each restoring 3Ј maturation almost completely. PNPase was only slightly less effective. RNase PH, on the other hand, led to only ϳ50% of the normal amount of 16S rRNA 3Ј processing. These findings indicate that four different exoribonucleases each can contribute to 3Ј maturation of 16S rRNA and that only in the absence of all four RNases is 3Ј maturation largely eliminated.
Essentially identical results were obtained by Northern blot analysis using probes directed against the 3Ј precursor region and mature 16S rRNA (Fig. 1B). However, the Northern analysis also revealed that in the most defective mutant strains, those lacking all four exoribonucleases (lane 2) or containing only RNase PH (lane 4), extensive degradation of 16S rRNA occurs. Thus, if maturation is blocked, a previously described quality control process on rRNA becomes prevalent (15), although in the absence of RNase R and PNPase, degradation is not complete, and specific fragments accumulate.
Pulse-Chase Analysis of 3Ј Maturation of 16S rRNA-To expand on the steady-state analysis presented in Fig. 1, we carried out pulse-chase experiments to compare the rate of 3Ј processing in the quadruple RNase-deficient strain with that in wild type cells. After incubation at 44°C for 1 h to inactivate PNPase, cells were labeled for 5 min with 32 Pi followed by the addition of rifampicin to prevent synthesis of additional rRNA molecules. Samples were taken over a period of 20 min to measure the rate of conversion of precursor to mature 16S rRNA (Fig. 2). Precursor RNA in wild type cells is so rapidly converted to the mature species that even at the zero time point, 40% of the labeled RNA is already mature, representing RNA molecules that had been transcribed early during the pulse period. In contrast, conversion of precursor to mature 16S rRNA is dramatically slowed in the RNase-deficient strain. Even after 20 min of chase, Ͼ70% of the RNA is still present in precursor form. These data confirm that removal of the four exoribonucleases, RNase II, RNase R, PNPase, and RNase PH, dramatically decreases 3Ј processing of 16S rRNA. It is not yet clear whether the very slow rate of processing that remains is due to incomplete inactivation of the temperature-sensitive PNPase or to yet another RNase with low processing activity.
Lack of 3Ј Processing of 16S rRNA Affects 5Ј Processing-To determine whether 5Ј processing of 16S rRNA is affected by the absence of exoribonucleases that mature the 3Ј end, we simultaneously examined both termini of the RNA molecule in wild type and RNase-deficient strains. For this analysis, we used the same RNase H cleavage method as in Fig. 1A, but included as well a 5Ј-specific chimeric oligonucleotide complementary to residues 40 -57. This allowed clear separation of RNA populations containing 115 or 66 extra 5Ј-residues or mature 5Ј ends (Fig. 3). The experiment shown in panel A was carried out as in Fig. 1A, with very similar results for 3Ј processing in the various mutant strains. Analysis of the 5Ј ends of the 16S rRNA present in each of the strains (Fig. 3B) revealed that the percentage of molecules with unprocessed 5Ј termini was essentially the same as those with unprocessed 3Ј ends. Moreover, most of the unprocessed 5Ј termini contained 115 rather than 66 extra 5Ј-residues. These data indicate that 5Ј maturation is dependent on prior 3Ј processing and that in its absence, the initial RNase E cleavage at position ϩ66 (5) is severely inhibited.
Mutant 30S Subunits Are Not Assembled into 70S Ribosomes-To determine the fate of unprocessed 16S rRNA precursor, we compared ribosomes isolated from the quadruple RNase-deficient cells with those isolated from wild type (Fig. 4). Both ribo-some subunits and intact ribosomes were examined using sucrose gradient sedimentation. At low Mg 2ϩ , in which ribosomes are present as subunits, the profiles from the wild type and mutant strains are similar, although the 30S peak in the mutant is slightly broader (Fig. 4A). However, Northern blot analysis revealed that although 30S particles from the wild type are devoid of 16S precursor RNA, those derived from the mutant strain contain a majority of precursor RNA molecules. Note also that the ratio of precursor to mature 16S RNA differs in the two mutant fractions analyzed, indicating that the mutant 30S peak is heterogeneous. Thus, 30S particles can be assembled even when 16S RNA is not matured, although as shown below, these 30S particles are abnormal.
In contrast to low Mg 2ϩ , at 10 mM Mg 2ϩ , wild type ribosomes are found primarily as 70S particles (85%) (Fig. 4B). In contrast, Ͻ60% of mutant ribosomes are present as 70S particles, and there is a deficit of 30S subunits relative to 50S subunits. This finding is consistent with the considerable degradation of 16S rRNA seen in the RNase-deficient strain. Moreover, Northern blot analysis indicates that the precursor 16S rRNA present in the mutant strain is found largely in the 30S peak. These data support the conclusion that the 30S subunits assembled in the absence of the four exoribonucleases are abnormal, and presumably because they are degraded so rapidly, participate poorly in the formation of 70S particles. The small amount of 16S precursor in the 70S peak likely is due to particles that associated with 50S subunits and thereby escaped degradation.

DISCUSSION
The findings presented here indicate that maturation of the 3Ј end of 16S rRNA in E. coli is carried out by four known 3Ј35Ј-exoribonucleases, and the presence of any one of them is sufficient to enable cell growth. The three processive exoribonucleases, RNase II, RNase R, and PNPase, are most effective, but cells grow even when only RNase PH remains, although considerably more slowly. The fact that each of the four RNases can completely mature the 3Ј end of the RNA explains why it has been so difficult to identify the enzyme responsible for 3Ј   processing. Mutants lacking exoribonucleases had previously been examined for defects in 3Ј maturation of 16S RNA (1), but in no case were all removed simultaneously. Because elimination of the four RNase activities decreases 16S 3Ј maturation by close to 90%, it is clear that they are the main participants in the process.
Based on the information presented here, we now have a fairly complete picture of the maturation pathway for 16S rRNA in E. coli (1). Initially, the double strand-specific endoribonuclease, RNase III, cleaves the rRNA transcript in a doublestranded stem generated by complementarity between precursor regions flanking both sides of 16S RNA to generate a 17S precursor molecule containing 115 extra 5Ј-nucleotides and 33 extra 3Ј-nucleotides. Then, the 33 extra 3Ј-nucleotides are removed by any one of four exoribonucleases, RNase II, R, PH, or PNPase. Removal of the extra 3Ј-residues results in the extra 5Ј-residues becoming single-stranded, which enables the single-strand-specific endoribonuclease, RNase E, to cleave at residue ϩ66 to generate the 16.3S rRNA precursor (5,6). Cleavage at ϩ66 facilitates the subsequent cleavage by RNase G at the mature 5Ј terminus. This sequence of RNA processing events occurs within the context of a preribosomal particle, and the next major challenge is to elucidate how they are coordinated with ribosome assembly.
One interesting question that remains to be answered is how the exoribonucleases identified here digest through the extensive secondary structure that results from the complementarity of the extra 5Ј-and 3Ј-residues in the 16S rRNA precursor. Although RNase R and PNPase as part of the degradosome have such capability, RNase II does not. Consequently, at least with this exoribonuclease, it is extremely likely that an RNA helicase will be directly involved in 3Ј processing of 16S rRNA. Further analysis will determine whether this prediction is borne out.
With the identification of the RNases responsible for 3Ј maturation of 16S RNA, we now know how the 3Ј termini of essentially all stable RNAs are generated in E. coli. This includes 16S (this work), 23S (12,17), and 5S (18) rRNAs, tRNAs (9), and other small stable RNAs (19). In all cases, removal of the last precursor residues involves the action of an exoribonuclease. This contrasts with other organisms, including other eubacteria, in which endoribonuclease action may occur (e.g. Refs. 20 -22). The exoribonucleolytic mode of 3Ј processing in E. coli raises not only interesting evolutionary questions, but also questions of how the exoribonucleases accurately stop at the mature 3Ј position, particularly when some of the RNases are processive enzymes. Undoubtedly, the context of the preribosomal particle is involved, as was previously shown for RNase T action on 5S RNA (18) and 23S RNA (12) precursors, but further study will be required to clarify this for 16S RNA.
Although identification of the RNases involved in 16S RNA processing provides a major advance, it is clear that multiple other factors also participate in ribosome biogenesis. These include ribosomal proteins, rRNA-modifying enzymes, RNA helicases, GTPases, and other proteins of unknown function. Exactly what role, if any, each of these proteins plays in removal of the 33 extra residues at the 3Ј end of the 16S rRNA precursor is not yet known, but it is clear that they are necessary to facil-itate the ordered sequence of events that make up rRNA processing and ribosome assembly.
In a very recent study (23), one protein of previously unknown function, YbeY, known to participate in 3Ј processing of 16S rRNA (24), was shown to be an endoribonuclease. As a consequence, it was proposed that YbeY is responsible for removal of most of the 33 extra 3Ј-residues, although no direct evidence for this assertion was presented (23). Based on our data, we believe it is unlikely that YbeY directly removes extra 3Ј-nucleotides inasmuch as all 33 residues remain when the exoribonucleases identified here are removed. Thus, although it is clear that YbeY is needed for 3Ј maturation of 16S rRNA (24), its actual role remains to be determined.