The CafA protein required for the 5'-maturation of 16 S rRNA is a 5'-end-dependent ribonuclease that has context-dependent broad sequence specificity.

The CafA protein, which was initially described as having a role in either Escherichia coli cell division or chromosomal segregation, has recently been shown to be required for the maturation of the 5'-end of 16 S rRNA. The sequence of CafA is similar to that of the N-terminal ribonucleolytic half of RNase E, an essential E. coli enzyme that has a central role in the processing of rRNA and the decay of mRNA and RNAI, the antisense regulator of ColE1-type plasmids. We show here that a highly purified preparation of CafA is sufficient in vitro for RNA cutting. We detected CafA cleavage of RNAI and a structured region from the 5'-untranslated region of ompA mRNA within segments cleavable by RNaseE, but not CafA cleavage of 9 S RNA at its "a" RNase E site. The latter is consistent with the finding that the generation of 5 S rRNA from its 9 S precursor can be blocked by inactivation of RNase E in cells that are wild type for CafA. Interestingly, however, a decanucleotide corresponding in sequence to the a site of 9 S RNA was cut efficiently indicating that cleavage by CafA is regulated by the context of sites within structured RNAs. Consistent with this notion is our finding that although 23 S rRNA is stable in vivo, a segment from this RNA is cut efficient by CafA at multiple sites in vitro. We also show that, like RNase E cleavage, the efficiency of cleavage by CafA is dependent on the presence of a monophosphate group on the 5'-end of the RNA. This finding raises the possibility that the context dependence of cleavage by CafA may be due at least in part to the separation of a cleavable sequence from the 5'-end of an RNA. Comparison of the sites surrounding points of CafA cleavage suggests that this enzyme has broad sequence specificity. Together with the knowledge that CafA can cut RNAI and ompA mRNA in vitro within segments whose cleavage in vivo initiates the decay of these RNAs, this finding suggests that CafA may contribute at some point during the decay of many RNAs in E. coli.

The overproduction of CafA (1-3) under conditions of slow growth has been shown to cause the formation of chained cells and minicells. The presence of the latter has been interpreted as evidence for CafA either enhancing the rate of cell division and/or inhibiting chromosome partitioning after replication (4). Electron microscopic examination of the chained cells revealed axial filamentous bundles, termed cytoplasmic axial filaments (hence the designation CafA), running through the center of their cytoplasms. Furthermore, the cytoplasmic axial filaments appear to be composed almost entirely of CafA (5). This finding combined with the phenotype of cells overproducing CafA has led to the proposal that in normal cells these filaments in an unbundled form may have a role as cytoskeletal-like elements in either cell division or chromosome segregation (4).
The sequence of CafA has 34% similarity with the N-terminal nucleolytic domain of RNase E (6), an essential Escherichia coli ribonuclease that is required for the generation of 5 S rRNA from a 9 S precursor (7) and has a central role in the decay and/or processing of a variety of RNAs, including many if not most mRNAs and RNAI, the antisense RNA regulator of the replication of ColE1-type plasmids (for reviews see Refs. 8 and 9). Endoribonucleolytic cleavage by RNase E occurs within single-stranded A and/or U-rich segments (10,11); however, there is no simple relationship between the order of nucleotides and the phosphodiester bond(s) that is cleaved (12,13). An oligonucleotide corresponding in sequence to the 5Ј-end of RNAI has been found to be cut efficiently by RNase E in vitro (14). Combined with the knowledge that alteration of secondary structures adjacent to RNase E sites can either increase or decrease the rate of cleavage (11,14,15), this finding has contributed to the notion that secondary structures within complex RNAs rather than serving as direct recognition motifs (14,15) affect RNase E cleavage by either limiting access of the enzyme (14,16,17) and/or determining the stability of local structures, which in turn determines the single-strandedness (and thus cleavability) of susceptible sites (15,18,19).
Recently, it has been shown that efficient cleavage by RNase E is dependent on the nature of the 5Ј-end of its substrates. Compared with a linear substrate that had a 5Ј-monophosphate group immediately followed by a single-stranded segment, circular substrates, 5Ј-triphosphorylated substrates, or 5Ј-monophosphorylated substrates that had a duplex at the extreme 5Ј-end were found to be cut inefficiently by RNase E in vitro (20). The two latter observations are consistent with the in vivo findings that pppRNAI Ϫ5 is decayed more slowly than pRNAI Ϫ5 (21) and 5Ј-stem-loops (or rather the absence of an unpaired segment) can stabilize RNA (16,22), respectively. Additionally, the preferential cleavage by RNase E of 5Ј-monophosphorylated RNAs, such as downstream fragments produced by this enzyme (e.g. pRNAI Ϫ5 ), over 5Ј-triphosphorylated intact RNAs (e.g. pppRNAI) suggests that once decay of an RNA molecule is initiated its completion will be preferred over the cutting of intact RNA (20). This notion may explain the observation that in general RNAs decay without the accumulation of significant levels of decay intermediates, the so called "all or nothing" phenomenon (20). RNase E has recently been shown to be also capable of removing 3Ј-poly(A) tails (23), which are known to facilitate 3Ј-exonucleolytic attack by polynucleotide phosphorylase (PN-Pase; for reviews see Refs. 24 -27). Furthermore, it was reported that the poly(A) nuclease activity of RNase E is blocked by the presence of a 3Ј-monophosphate group suggesting that it is 3Ј-exonucleolytic. Although, the precise role of the poly(A) nuclease activity of RNase E in RNA decay remains to be determined, it seems likely that any action it has on poly(A) tails in vivo would affect processing by PNPase, which together with RNase E, the RhlB helicase, and the glycolytic enzyme enolase form the core of the RNA degradosome (28,29). This noncovalent assembly is almost certainly the major cellular machine for the decay and processing of RNA in E. coli (for review see Ref. 30). Both the endonucleolytic and poly(A) nuclease activities of RNase E are located in its N-terminal half (23,31). The C-terminal half of RNase E contains the binding sites for the other major degradosome components (32,33) and an arginine-rich RNA binding site (31, 34 -36).
Given the extensive sequence similarity between CafA and the N-terminal domain of RNase E (6), we purified CafA and investigated whether it has ribonucleolytic activity. Here we report that CafA is indeed a ribonuclease, which is consistent with the recent finding that it is required for the 5Ј-maturation of 16 S ribosomal RNA (37,38). Furthermore, the results of our investigation into the nature of the ribonucleolytic activity of CafA raise the possibility that it may also have a role in RNA decay.

EXPERIMENTAL PROCEDURES
Chemical Synthesis of RNA Oligonucleotides-Oligoribonucleotides BR10, BR10p, and A40 were synthesized using an ABI 391 DNA synthesizer with a modified 1-mol DNA assembly cycle and standard ABI reagents, and the coupling efficiencies were determined by trityl assay as described previously (39). Deprotected RNAs were purified by anion exchange chromatography using a Dionex DNAPac PA-100 column (4 ϫ 250 mm) on a Dionex DX500 high pressure liquid chromatography unit.
In Vitro Transcription-RNAI, 9 S RNA, and the 5Ј-UTR 1 of ompA mRNA were synthesized using the T7-MEGAshortscript from Ambion. Typically 50 nM of DNA template was incubated at 37°C for 90 min in a 20-l reaction as described by the vendor. When internally labeled RNA was required, 60 Ci of [␣-32 P]UTP (ICN) was included in the reaction. Transcripts were visualized by either UV shadowing or autoradiography (40) and gel purified. 9 S RNA and a segment of the 5Ј-UTR of ompA mRNA were generated from HaeIII-cut pTH90 (41) and Hin-dIII-cut p106B-64 (a gift from J.G. Belasco, Skirball Institute), respectively, whereas the template for the synthesis of RNAI was a polymerase chain reaction product generated using primers with sequences 5Ј-GCATCCTAATACGACTCACTATAGGGACAGTATTTGGT and 5Ј-AACAAAAAACCACGCTACCACCAGC.
5Ј-Radiolabeling of RNA-Four to five pmols of RNA that had been either chemically synthesized or transcribed in vitro and then dephosphorylated was radiolabeled at the 5Ј-end by incubating with 23 pmols (160 Ci) of [␥-32 P]ATP (ICN) and 10 units of T4 polynucleotide kinase (MBI Fermentas) at 37°C for 10 min in 20 l of forward reaction buffer provided by the vendor. The reactions were quenched with urea-loading buffer (7 M urea, 0.1% (w/v) bromphenol blue, and xylene cyanol) and the 5Ј-labeled RNAs were gel purified. RNA that had been transcribed in vitro was dephosphorylated by incubating 4 -5 pmols with 1 unit of bacterial alkaline phosphatase (Life Technologies, Inc.) at 60°C for 1 h in buffer provided by the vendor. The alkaline phosphatase was removed by adding 20 g of fungal proteinase K (Life Technologies, Inc.) and incubating at 37°C for 30 min. To remove proteinase K, this reaction mixture was extracted with phenol/chloroform and chloroform and the RNA precipitated using ethanol as described previously (40).
Gel Purification of RNAs-Labeled full-length RNA was separated from truncated forms and unincorporated nucleotides by running in polyacrylamide sequencing gels. 8 and 20% (w/v) polyacrylamide gels were used to purify transcripts and oligoribonucleotides, respectively. An autoradiograph of the gel was used as a template, and a slice of gel containing the radiolabeled RNA was excised. RNA was eluted from the gel slice into 400 l of 150 mM NaCl, 50% (v/v) acidified phenol (pH 4.3) at 37°C for 2-16 h. The eluate was extracted with phenol and chloroform, precipitated using ethanol, and resuspended in water (Sigma).
Construction of pCAFA01 Encoding His-tagged CafA-The cafA gene segment was amplified from plasmid pMEL1 (42) using an Elongase kit as per the vendor's instructions (Life Technologies, Inc.). The primers used were 5Ј-GCCCGGGCATATGACGGCTGAATTGTTAGTAAACG and 5Ј-GCGGGATCCTTACATCATTACGACGTCAAACTGC, which introduced unique NdeI and BamHI sites (bold type) at the 5Ј-and 3Ј-end, respectively, of the cafA coding sequence. The resulting 1.4-kilobase polymerase chain reaction product was cut with NdeI and BamHI and cloned between the corresponding sites of pET16b (Novagen). The resulting plasmid was designated pCAFA01.
Purification of CafA and the N-terminal Catalytic Domain of RNase E-Cultures of E. coli BL21(DE3) cells harboring either pCAFA01 (this work) or pNSTOP (31), a plasmid encoding the N-terminal ribonucleolytic domain (residues 1-498) of RNase E, were grown in 2YT (40) to an A 600 of 0.6. Expression of the plasmid-cloned gene was then induced by adding isopropyl-1-thio-␤-D-galactopyranoside to 1 mM. After incubation for a further 60 min at 37°C, the cells were harvested by centrifugation at 4000 ϫ g for 10 min using a MSE 3000 centrifuge. The cell pellet was resuspended in 50 ml of binding buffer (7 M urea, 20 mM Tris-HCl (pH 7.6), 500 mM NaCl, 5 mM imidazole, 0.1% (w/v) Triton X-100), and the cells were ruptured by passage through an French Pressure Cell (Amicon). The insoluble material including intact cells was removed by high speed centrifugation (100,000 ϫ g, 30 min) using a Beckman SW28 rotor. The supernatant was loaded on to a 5-ml nickel-charged HiTrap Chelating column (Amersham Pharmacia Biotech), which was washed with 30 ml of binding buffer before bound proteins were eluted with an imidazole gradient (5 mM to 1 M over 20 ml). All of the above purification steps were done using buffers, columns, and equipment that have been cooled to below 4°C. The protein content of each fraction was assayed using a modified Bradford assay (Bio-Rad) and SDS-polyacrylamide gel electrophoresis. In initial experiments ( Fig. 1), CafA was purified from extracts of cells from a 1-liter culture; however, we found that purification from a cell extract of a 4-liter culture resulted in reduced levels of contaminating polypeptides (Fig. 2).
Electroelution of Purified CafA Protein-Recombinant CafA protein was further purified from a preparative Lamelli gel (40) using a Mini Whole Gel Eluter (Bio-Rad). Protein was eluted from a 10% (w/v) polyacrylamide gel (7.2 ϫ 10.0 ϫ 0.1 cm) as per the vendor's instructions into 50 mM Tris-HCl (pH 8.7), 25 mM boric acid. A voltage of 200 V (10 watt) was applied for 30 min. At the end, the polarity was reversed for 15 s to remove any protein that might have stuck to the membrane during elution. Residual SDS was removed from the eluate samples using Extracti-gel D as per the vendor's instructions (Pierce).
RNA Cleavage Reactions-Two hundred ng of either CafA or RNase E was incubated with 4 -15 pmols of the appropriate substrate RNA in 20 l of 20 mM Tris-HCl (pH 7.4), 100 mM NaCl, 10 mM MgCl 2 , 0.1% (v/v) Triton X-100, and 1 mM dithiothreitol containing 20 units of RNase inhibitor (Amersham Pharmacia Biotech) at 37°C for up to 60 min. Samples were taken at regular intervals and quenched using urealoading buffer before aliquots were analyzed using polyacrylamide sequencing-type gels (40).

RESULTS
CafA (RNase G) Is a "5Ј-End-dependent" Endoribonuclease-Fractions across a peak of recombinant His-tagged CafA that was purified using immobilized metal affinity chromatography (Fig. 1A) were incubated with 5Ј-labeled BR10 (5Ј-ACAGUAUUUG), a synthetic decaribonucleotide that corresponds in sequence to the single-stranded segment at the 5Јend of RNAI (14). Ribonucleolytic activity was detected in all fractions containing CafA (B); moreover, the level of activity was directly proportional to the amount of this polypeptide (C).
To confirm that CafA was the source of the ribonucleolytic activity, we further purified CafA by electroeluting, following preparative gel electrophoresis, a batch that contained reduced amounts of contaminating polypeptides (compare Fig. 1A and Fig. 2A) as a result of increasing the amount of cell extract added to the immobilized metal affinity chromatography column (see "Experimental Procedures"). The resulting CafA preparation (Fig. 2) was homogeneous as judged by staining of SDS-polyacrylamide gels using Coomassie Blue (A) and silver (data not shown) and was still able to cleave the decanucleotide substrate (B). Thus, from here on we will adopt the designation RNase G (37, 38) when referring to CafA. Comparison of the migration of the upstream products of RNase G cleavage with a 1-nt ladder generated using PNPase revealed that they were 5, 6, and 7 nt (B).
To determine whether the products of RNase G cleavage were generated endoribonucleolytically, we used as substrate 5Ј-monophosphorylated BR10 labeled at its 3Ј-end by the addition of [5Ј-32 P]pCp. As shown in Fig. 3, downstream products of 4, 5, and 6 nt were detected (A) consistent with endonucleolytic cleavage at the same positions identified using 5Ј-labeled BR10. Moreover, the relative abundance of each of the downstream products (Fig. 3A) mirrored precisely that of the corresponding upstream products (Fig. 2) indicating that RNase G is only able to cut individual BR10 decanucleotides once. Incubation of RNase G with two BR10 derivatives that had either three extra Gs at the 5Ј-end or a C at the 3Ј-end resulted in cleavage at precisely the same positions (relative to sequence) observed for BR10 (data not shown), indicating that as found for RNase E (14), the specificity of RNase G cleavage of BR10 is determined by sequence rather than a distance measured in nucleotides from either its 3Ј-or 5Ј-end.
We next investigated whether efficient RNase G cleavage is dependent on the presence of a monophosphate on the 5Ј-end of its substrates by incubating with RNase G an aliquot of 3Јlabeled BR10 that had not been 5Ј-phosphorylated (Fig. 3B). We were unable to detect cleavage of this substrate by RNase G after 60 min of incubation (B) even though 50% of 5Ј-monophosphorylated BR10 was cut within 2 min (A). This finding indicates that like RNase E, which was included as a control, RNase G is a 5Ј-end-dependent ribonuclease (20). In contrast, we found that the 3Ј-phosphorylation status did not affect the rate of cleavage of 5Ј-labeled BR10 by RNase G or RNase E; an oligonucleotide synthesized with a 3Ј-phosphate was cleaved as efficiently as one that had a hydroxyl group at its 3Ј-end (Fig. 4).
The Specificity of RNase G Cutting Overlaps That of RNase E and Is Context-dependent-Having found that RNase G can cut BR10 (Fig. 3), we decided to investigate using complex substrates whether RNase G can cut within other single-stranded segments cleavable by RNase E and/or would cut at other positions. To this end, we incubated RNase G and, as a control, RNase E with RNAI, the 5Ј-UTR of ompA mRNA, 9 S RNA, and a segment from 23 S rRNA (Fig. 5). All of the substrates used were labeled at their 5Ј-ends. Our RNase E preparation cut RNAI primarily at the Ϫ5 position within the single-stranded region at its 5Ј-end. Less efficient RNase E cleavage at the Ϫ6  (14) was incubated with 1 l of each fraction at 37°C for 30 min (see "Experimental Procedures"). S indicates the substrate, whereas P indicates the cleavage products. The reaction products were analyzed using a 20% (w/v) polyacrylamide gel. C, comparison of the relative amounts of CafA and ribonucleolytic activity in alternate fractions of the elution peak. Relative amounts of CafA were determined by densitometric scanning of the 60-kDa band in A using a system from PDI, whereas the relative activities were determined by using a Phosphoimager (Fuji) and Tina software (Raytest, Germany) to measure the proportion of the substrate that was cleaved in each fraction in B. Both sets of values were expressed in arbitrary units.  Inc.). B, assay of samples in A for ribonucleolytic activity. 5Ј-Labeled BR10 was incubated with equal amounts (50 ng) of immobilized metal affinity chromatography purified and electroeluted CafA (RNase G) under conditions used in Fig. 1. The markers in this panel (lane M) were a 1-nt ladder generated by limited PNPase decay of the substrate. The source of PNPase was a preparation of degradosome that was purified and assayed as described previously (28). The numbers on the right of this panel indicate the size in nucleotides of the substrate and cleavage products.
position and at internal sites within RNAI was also evident as observed previously (14,36). RNase G was also able to cut RNAI; moreover, the major sites of cutting overlapped those of RNase E within the single-stranded segment at the 5Ј-end (A). Relative to RNase E, however, RNase G was able to cut more efficiently at the Ϫ6 position but appeared to be unable to cleave at internal sites. In the case of the 5Ј-UTR segment of ompA mRNA, we found that RNase G was able to cut within two internal segments that contain major RNase E sites designated "c" and "d" (43,44), albeit more slowly than our RNase E preparation (B). Close examination of this gel suggests that although RNase G is able to cut within the c segment of ompA mRNA, the precise bonds that were cleaved differ from those cut by RNase E. We were unable to detect RNase G cleavage at the a RNase E site of 9 S RNA (C), which is required for the generation of 5 S rRNA (45), indicating that not all RNase E-cleavable segments within complex RNAs are substrates for RNase G. This finding is consistent with the observation that 9 S RNA processing can be blocked in cells that are wild type for RNase G (46). Taken together these results indicate that although RNase E and RNase G specificity overlap, they are not congruent.
We further examined the basis of the inability of RNase G to cleave within the a site of 9 S RNA by incubating it with a synthetic decanucleotide that corresponds in sequence to this segment of RNA. As shown in Fig. 6, RNase G was able to cleave this substrate as efficiently as RNase E indicating that the context of the a site in 9 S RNA and not its sequence blocks cleavage by RNase G. Consistent with the context being important in regulating RNase G cleavage of sites within complex RNAs is our finding that although 23 S rRNA is stable in vivo, a segment of this RNA was cleaved efficiently and at multiple sites by RNase G (Fig. 5D).
RNase G Has Only Weak Poly(A) Nuclease Activity-As RNase E has recently been shown to be able to remove 3Јpoly(A) tails (23) in addition to making decay-initiating cleavages at or near the 5Ј-end of RNAs (for reviews see Ref. 8), we also incubated RNase G with a mixture of equimolar amounts of a 5Ј-labeled A 40 -mer and BR10 (Fig. 7). The N-terminal half of RNase E cut the A 40 -mer at a rate that was reproducibly 3-4-fold slower than that of BR10; however, even though RNase G under identical reaction conditions cleaved 50% of BR10 within 6 min, it only cut 25-30% of the A 40 -mer after 90 min (A). Similar rates of decay were obtained when the substrates were incubate individually with RNase G or RNase E (B). Combined these results indicate that relative to RNase E, RNase G has weak poly(A) nuclease activity. DISCUSSION The work presented here extends recent reports that RNase G is required for the maturation of the 5Ј-end of 16 S rRNA in vivo (37,38). We have shown that a highly purified preparation of RNase G is sufficient in vitro for endonucleolytic cutting of RNA (Figs. 1-3). Moreover, we find that efficient cutting by this enzyme, like that by RNase E, is 5Ј-end dependent (20); under the experimental conditions used we were unable to detect RNase G (or RNase E) cleavage of a 5Ј-hydroxylated oligoribonucleotide that was cleaved efficiently when monophosphorylated at its 5Ј-end (Fig. 3). We also found that RNase G cleavage was blocked by the presence on substrates of a 5Јtriphosphate group (data not shown), as reported for RNase E (20). Therefore, we suggest that, as proposed for RNase E (20), RNase G will prefer to cut to completion 5Ј-monophosphorylated decay or processing intermediates rather than initiating the decay of intact 5Ј-triphosphorylated RNAs. Implicit in this model is that 5Ј-monophosphate groups only stimulate cleavage at sites present on the same RNA molecule. Support for this is provided by our finding that 5Ј-hydroxylated RNA present at a low level in our preparation of 3Ј-labeled 5Ј-monophosphorylated BR10 (as a result of incomplete enzymatic 5Ј-phosphorylation; see "Experimental Procedures") was not cleaved by RNase G or our RNase E preparation even after the bulk of the 5Ј-monophosphorylated RNA had been cut to completion (Fig. 3A).

FIG. 3. Assay of ribonucleolytic cleavage of substrates monophosphorylated or hydroxylated at their 5-ends.
A, 3Ј-Labeled BR10 that had been monophosphorylated at its 5Ј-end was incubated with RNase G (CafA) and, as a control, RNase E (NRne, 50 ng) as described in Fig. 1. The RNA species marked with an asterisk is 5Јhydroxylated BR10 resulting from incomplete 5Ј-phosphorylation. B was the same as A, except the substrate had not been 5Ј-monophosphorylated using T4 polynucleotide kinase (see "Experimental Procedures"). The double asterisk indicates a minor contaminating species generated during gel purification.
FIG. 4. Assay of ribonucleolytic cleavage of 5-labeled substrates having either a monophosphate or hydroxyl group at their 3-end. RNase G was incubated with 5Ј-labeled BR10 (5Ј-32 Plabeled ACAGUAUUUG-OH ) and BR10p, an oligonucleotide that is identical in sequence to BR10, but was synthesized with a 3Ј-monophosphate group. The reaction conditions were as in Fig. 1. The time points are indicated at the top of the panel, whereas the size of the substrate and cleavage products in nucleotides are indicate at the right. The single asterisk indicates RNA that had its 3Ј-phosphate removed by the 3Ј-phosphatase activity of T4 polynucleotide kinase during the 5Ј-labeling reaction (see "Experimental Procedures"), whereas the double asterisks show the position of a contaminating species generated during preparation of the substrate. The latter provides a convenient internal control for loading differences.
As the RNase E polypeptide we used as a control contained only the N-terminal ribonucleolytic domain and homologues of this half of RNase E have been found in the genomes of all Gram-negative and some Gram-positive bacteria that have been completely sequenced (32), we suggest that the mechanism for sensing the presence of monophosphate groups on the 5Ј-end of RNAs may be ancient and evolutionarily conserved. It is also conceivable that the distance separating a cleavable sequence(s) from a 5Ј-monophosphate group may control the efficiency of cutting by RNase G and RNase E. Indeed this notion, which is open to experimental investigation, may provide an explanation for the observation by others that RNase G is only able to generate the mature 5Ј-end of 16 S rRNA in vivo after cutting of 17 S RNA by RNase E (38), and our finding that although RNase G is unable to cut 5Ј-monophosphorylated 9 S RNA at the a RNase E site (Fig. 5); it is able to cut as efficiently as the N-terminal domain of RNase E a decanucleotide corresponding in sequence to this segment of RNA (Fig. 6). In both these examples a 5Ј-monophosphate group is brought closer (at least in terms of sequence length) to a segment that can be cleaved by RNase G. An additional possibility, which need not be mutually exclusive, is that higher order structures within 17 and 9 S RNA block access of RNase G to potentially cleavable sequences. Indeed, there is good evidence that the overall conformation of RNA can affect the efficiency of cutting by RNase E (10, 11, 16 -19). In any case, the context of a sequence within a complex RNA, such as 17 S RNA, is likely to be extremely important (and possibly the major factor) in determining whether it will be cleaved by RNase G, as examination of the sequences cleaved by this enzyme did not reveal a requirement for a particular order of nucleotides (Fig. 8). Similar conclusions have been reached regarding the specificity of RNase E cleavage (12,13).
Our finding that RNase G can cleave in vitro at multiple positions within the 5Ј-single-stranded sequence of RNAI (Figs. 2a and 5A) raises the possibility that it also contributes to the decay of this RNA in vivo. Consistent with this notion is the finding that when products of cleavage in the 5Ј-singlestranded region of RNAI were stabilized in vivo by mutational inactivation of either the gene encoding PNPase or the polymerase that adds poly(A) tails to RNAI, species generated by cleavage at the Ϫ6 and Ϫ7 positions were detected in addition to RNase E cleavage at the Ϫ5 position (47,48). Additionally, when RNase E was mutationally inactivated in vivo, RNAI (from both pACYC184 and pBR322) was still decayed, albeit more slowly, and certain cleavages within its 5Ј-end appeared FIG. 5. Assay of ribonucleolytic cutting of RNAI, the 5-UTR of ompA mRNA, 9 S RNA, and a segment from 23 S rRNA. Each 5Ј-labeled substrate was incubated with RNase G (CafA) and RNase E (NRne) as described in Fig. 1. RNAI, the 5Ј-UTR of ompA mRNA and 9 S RNA in A-C, respectively, were transcribed in vitro, whereas the segment of 23 S rRNA (residues 1052-1075; Ref. 56) in D was synthesized chemically (57). The single asterisk in A indicates a minor species generated by cleavage of an RNA transcript(s) that is 1-2 nt shorter at the 5Ј-end than full-length RNAI. In C, the minor species marked with a double asterisk in the preparation of 9 S RNA substrate result from cleavage during gel elution (see "Experimental Procedures"). In D, lanes 1, 2, 3, and 4 contain samples incubated with no enzyme, RNaseG, RNase E, and PNAase, respectively. The latter incubation generated a 1-nt ladder that to was used to determine the sizes of the endonucleolytic cleavage products.
FIG. 6. Assay of ribonucleolytic cutting of an RNA segment containing the a RNase E of 9 S RNA. 5Ј-Labeled 9SA, a decanucleotide (5Ј-ACAGAAUUUG) that corresponds in sequence to the a site of 9 S RNA, was incubated with RNase G (CafA) and RNase E (NRne)as described in Fig. 1. Samples of the reactions were ran in a 20% (w/v) polyacrylamide sequencing gel. The sizes of the products were determined by comparing their migration against that of a 1-nt ladder of the substrate generated using PNPase (data not shown).
The enzymatic activity that cleaves within the 5Ј-UTR of ompA mRNA was originally designated RNase K as several lines of evidence at that time suggested it was distinct from that of RNase E (43). Contributing to this conclusion was the finding that an activity that cleaved the 5Ј-UTR of ompA mRNA could be chromatographically separated from an activity of RNase E, i.e. the generation of pre-5 S rRNA from a 9 S precursor (for review see Ref. 49). Later, it was shown that a proteolysed preparation of RNase E could cut a site within the 5Ј-UTR of ompA, and it was proposed that RNase K activity was due to a fragment of RNase E (44,50). However, this notion was not sufficient to explain why the fractions that cleaved the 5Ј-UTR of ompA mRNA did not cleave 9 S RNA and were not inactivated by the heating of extracts from a temperature-sensitive RNase E strain of E. coli (49). In light of our results (Fig. 5), an alternative explanation is that the 5Ј-UTR of ompA mRNA was cleaved by RNase G in these experiments (38). Furthermore, a role for RNase G in the cleavage of the 5Ј-UTR of ompA mRNA in vivo would provide a mechanism to explain how the decay of ompA mRNA and the processing of 9 S RNA can be differentially regulated (51). In any case, the finding that RNase G can make cleavages that resemble those of RNase E (this work, 37, 38) combined with evidence that the functions of RNase E and RNase G overlap genetically, albeit only partially (52), suggests that the study of decay events controlled by endoribonucleolytic cleavages now requires a careful assessment of the effect of mutational inactivation of RNase G in addition to that of RNase E. For example, it is possible that RNase G contributes to the processing of the mRNA of key cell division genes, as has been found for RNase E (53, 54), thus providing an explanation for the finding that overproduction of RNase G disrupts normal cell division (4). However, given its weak poly(A) nuclease activity (Fig. 7) we think it is unlikely that RNase G will be found to affect RNA decay through an action on 3Ј-poly(A) tails.
The biological role, if any, of the cytoplasmic axial filaments formed by RNase G (4) remains to determined; however, we note with interest that there is a precedent in eukaryotes for associations between RNA and cytoskeletal filaments that have key roles in fundamental cellular processes (for review see Ref. 55). Regardless, the ability of RNase G to self-associate in vivo may explain our finding (data not shown) that this enzyme aggregates in vitro at high concentrations (Ͼ1 mg/ml) necessitating its purification and storage under denaturing conditions. A, an equimolar mixture of 5Ј-labeled A40 (a polymer of 40 As) and BR10 was incubated with RNase G and RNase E, and samples that were taken at different time intervals (given at the top of panel) were run in a 10% (w/v) polyacrylamide gel. Double asterisks are used as described in Fig. 4. B, plot of the percentage of substrate remaining with time. Squares and triangles represent the amount of BR10 remaining in RNase G and RNase E reactions, respectively. Similarly, circles and diamonds represent the amount of A40 remaining in RNase G and RNase E reactions, respectively. Closed symbols represent measurements taken from reactions contain a mixture of both substrates (A), whereas open symbols represent measurements taken from reactions (not shown) that contained only a single substrate. The values were derived from phosphoimages as described in Fig. 1.