RNase E cleaves at multiple sites in bubble regions of RNA I stem loops yielding products that dissociate differentially from the enzyme.

Earlier work has shown that RNase E cleaves RNAI, the antisense repressor of replication of ColE1-type plasmids, producing pRNAI-5, whose further decay is mediated by the poly(A)-dependent activity of polynucleotide phosphorylase and other 3' to 5' exonucleases. Using a poly(A) polymerase-deficient strain to impede exonucleolytic decay, we show that RNAI is additionally cleaved by RNase E at multiple sites, generating a series of decay intermediates that are differentially retained by the RNA binding domain (RBD) of RNase E. Primer extension analysis of RNAI decay intermediates and RNase T1 mapping of the cleavage products of RNAI generated in vitro by affinity-purified RNase E showed that RNase E can cleave internucleotide bonds in the bubble regions of duplex RNA segments and in single-stranded regions. Chemical in situ probing of a complex formed between RNAI and the RBD indicates that binding to the RBD destabilizes RNAI secondary structure. Our results suggest a model in which a series of sequential RNase E-mediated cleavages occurring at multiple sites of RNAI, some of which may be made more accessible to RNase E by the destabilizing effects of its RBD, generate RNA fragments that are further degraded by poly(A)-dependent 3' to 5' exonucleases.

In the bacterium Escherichia coli, RNA decay requires the coordinated activity of endonucleases and exonucleases (for review, see Ref. 1) and is usually initiated by RNase E endonucleolytic cleavage(s) followed by additional RNase III or RNase P cuts (2)(3)(4). Resulting fragments of digested RNA are further degraded from the 3Ј-end by polynucleotide phosphorylase (PNPase), 1 RNase II, and possibly a still unidentified additional exonuclease (5)(6)(7). It has been shown that degradation is retarded in E. coli strains deficient for poly(A) polymerase (8,9). The latter accelerates mRNA decay by means of the addition of poly(A) tails to the 3Ј-end (8) and has been suggested to play an important role in regulating the stability of messenger RNA (for review, see Ref. 10).
Three endoribonucleases, RNase E, RNase III, and RNase P have been demonstrated to affect the stability of individual messengers (2-4, 11, 12); inactivation of RNase E also has a stabilizing effect on the cellular pool of E. coli mRNA (13). RNase E is encoded by the rne gene and, so far, is the only endoribonuclease known to be involved in general mRNA turnover. This enzyme has been determined to have at least two separate functional domains, catalytic and RNA binding, which are involved in mRNA processing (14). Recent experiments have shown that RNase E is associated with other proteins including PNPase and/or heat shock protein GroEL (15)(16)(17)(18). Moreover, PNPase has been shown to interact functionally with RNase E as well as physically (5). The finding that key RNA-processing enzymes (RNase E and PNPase) interact supports the notion that a multicomponent complex, a "processosome" (19) may be responsible for mRNA degradation in E. coli. Earlier work has shown that the degradation of RNA I (Fig.  1), the antisense repressor of the replication of ColE1-type plasmids, is initiated by RNase E cleavage, producing pRNA I Ϫ5 , which undergoes further degradation mainly by the poly(A)dependent 3Ј-exonucleolytic activity of PNPase (for review, see Ref. 10). However, the detailed mechanism of RNA I decay, namely, whether its decay intermediate pRNA I Ϫ5 is degraded by only PNPase or also is subjected to endonucleolytic cleavages in vivo, has not been elucidated. In the experiments reported here, by inactivating poly(A) polymerase and consequently slowing 3Ј to 5Ј exonuclease decay, we detected other rne-dependent decay intermediates of pBR322 RNA I in addition to the previously described RNase E cleavage product, pRNA I Ϫ5 . We report the characterization of these intermediates, and present a model for possible role of functional domains of RNase E in rne-dependent decay of RNA I.

MATERIALS AND METHODS
Bacterial Strains, Plasmids, and Growth Conditions-The bacterial hosts used for studying the RNA I stability were E. coli K12 isogenic strains N3433 (lacZ43, relA, spoT1, thi1) and N3431 (rne-3071 ts (20)). YHC3393 (pcnB Ϫ ) and LK01 (pcnB Ϫ, rne ts ) result from the P1 transduction of the pcnB allele of strain Mri93 (21) into N3433 and N3431, respectively. YHC3393 and LK01 were selected as Tc r colonies in LB medium at 30°C. YHC012 (pnp Ϫ ) results from the P1 transduction of pnp allele of strain CA244 (5) into strain N3433. YHC012 was selected as Km r colonies in LB medium. Plasmids pCML108 and pCML103 (22) encode RNA I and pppRNA I Ϫ5 , respectively. In these experiments, pCML108 and pCML103 (pSC101-derived plasmids) (22) were chosen, instead of ColE1-type plasmids, to exclude the effect of RNA II on RNA decay. This effect is particularly noticeable in comparisons with the decay rates of pppRNA I Ϫ5 encoded by the pSL-C101 (23) and by pCML103 plasmids (this study). Plasmid pUC18T7/T3 pBRori, containing a FnuDII DNA fragment of the pBR322 replicon, was used for the preparation in vitro of radioactively labeled RNA transcribed from the T7 promoter as described (23). Plasmid pCRB011 encodes the RNA binding domain (amino acids 597-844) of RNase E and was kindly provided by K. J. McDowall and S. N. Cohen (Stanford University). Luria-Bertani broth (24) was used for plasmid DNA preparation, and M9/glucose medium (25) was used to grow cultures for isolation of RNA.
Northern Blot Analysis-Cell were grown exponentially in M9/glucose medium at 33°C or alternatively shifted to 43°C (rne ts strain) for 30 min prior to RNA extraction. The total cellular RNA isolated as described (23) at various times after rifampicin (0.5 mg/ml) was added to logarithmic phase cultures (OD 460 ϭ 0.4) and was separated on an 8% polyacrylamide sequencing gel, transferred to Zeta-Probe blotting membrane (Bio-Rad), and hybridized with radioactively labeled riboprobe containing a sequence complementary to RNA I of pBR322 according to the vendor's instructions.
Synthesis and RNase E Cleavage of GGGRNA I in Vitro-The template for in vitro synthesis of GGGRNA I by T7 RNA polymerase was generated by polymerase chain reaction amplification of a corresponding segment of pCML108 plasmid encoding RNA I and using primers 5Ј-GGTACCTAATACGACTCACTATAGGGACAGTATTTGGTATC-TGCGC (the sequence encoding a T7 promoter is in bold) and 5Ј-ACAAAAAAACCACCGCTACC. Continuously labeled GGGRNA I was purified in an 8% polyacrylamide sequencing gel, and 1-5 pmol of it was incubated with 50 ng of full-length RNase E (affinity purified under non-denaturing conditions) or with 2-3 g of catalytic domain (affinity purified under denaturing conditions and refolded) in 50 l of 20 mM Tris-HCl (pH 8.0), 5 mM magnesium chloride, 100 mM sodium chloride, 5% glycerol, 0.1% Triton X-100, 0.1 mM dithiothreitol. Both protein preparations were kindly provided by McDowall and Cohen (Stanford University). Aliquots (10 l) were removed from the reaction after 1, 3, 10, and 30 min, phenol extracted, mixed separately with 8 l of sequencing stop buffer, denatured by incubating at 85°C for 3 min, and coelectrophoresed with molecular weight standards. Standards were prepared by incubation of 5 pmol of terminally labeled GGGRNA I in 10 mM Tris-HCl (pH 8.0), 100 mM sodium chloride, 5 mM magnesium chloride with 5 units of RNase T 1 at 37°C for 10 min (RNase T 1 digest) and in 1 mM EDTA, 50 mM sodium carbonate (pH 9.2) at 85°C for 20 min (1-nucleotide alkaline hydrolysis ladder).
Primer Extension Analysis-The isolation of total cellular RNA and preparation of the RNase E digest of GGGRNA I were done as described above except that in vitro synthesis of GGGRNA I was carried out without radioactively labeled triphosphate. A synthetic Su-29-mer (5Ј-GCTACCAGCGGTGGTTTGTTTGCCGGATC-3Ј) was used for both the DNA sequencing and primer extension, according to the procedure described (26).
Shift Mobility Assay and Footprinting Analysis-Labeled GGGRNA I or its degradative intermediates isolated from RNase E digest of uniformly labeled GGGRNA I were incubated separately with/without RBD protein for 20 min at 37°C in 20 l of 5 mM Tris-HCl (pH 8.0), 125 mM sodium chloride, and 3% glycerol. The resulting mixtures were resolved by electrophoresis on a 4% nondenaturing polyacrylamide gel (60:1 acrylamide:bisacrylamide ratio) with 5 l of glycerol (30%) added to the samples prior to loading. Electrophoresis was with TBE buffer at 3 v/cm Ϫ1 for 7 h at 4°C. Purification of RBD protein encoded by pCRB011 plasmid was performed essentially as described (27). In situ (in gel) exposure of free and bound RNAs to chemical nuclease was performed as described (28). The products of cleavage were eluted from the gel, dissolved in a sequencing buffer, and run on an 8% sequencing gel using RNase T 1 digest and partial alkaline hydrolysate of terminally labeled RNA I as a size marker.

Inactivation of Poly(A) Polymerase Results in Accumulation of Several rne-dependent Degradative
Intermediates of RNA I-Northern blot hybridization to a 32 P-labeled riboprobe complementary to RNA I was used to detect full-length RNA I and its degradative products in preparations of total cellular RNA isolated from wild-type E. coli and from strains mutated in PNPase (pnp Ϫ ), RNase E (rne ts ), or poly(A) polymerase (pcnB Ϫ ). As seen in Fig. 2A, all RNA samples except the one isolated from a pcnB Ϫ strain mainly contain full-length RNA I, pRNA I Ϫ5 , and their polyadenylated products (bands that are positioned above the bands corresponding to RNA I and pRNA I Ϫ5 species). These data are consistent with the earlier evidence that RNase E cleavage of RNA I, together with polyadenylation, initiates its further degradation by PNPase or RNase II (5). However, in the pcnB mutant (YHC3393), we also detected other low molecular weight (LMW) degradative intermediates (Fig. 2, A-C) in cells containing the pCML108 plasmid, which encodes full-length RNA I (Fig. 2B), or the pCML103 plasmid (Fig. 2C), which encodes pppRNA I Ϫ5 . The preferential accumulation of LMW products in a strain deficient for poly(A) polymerase suggests that RNA I decay in wild-type E. coli may involve multiple endonucleolytic cleavages producing short RNA I fragments that are made unstable by polyadenylation and consequently undergo rapid digestion by PNPase, RNase II, or other nucleases. To determine whether mutations in the rne gene affect formation of these degradative products, we performed Northern blot analysis of total RNA isolated from the pcnB/rne ts double mutant (LK01) at permissive and non-permissive temperatures after rifampicin treatment. As shown in Fig. 2D, after shift of cells to 43°C, which results in inactivation of RNase E, the concentration of LMW intermediates is decreased, whereas the same temperature shift in pcnB Ϫ rne ϩ strain does not affect the degradation pattern even after a longer incubation time (Fig. 2E). Some additional bands located between LMW intermediates and RNA I Ϫ5 (Fig. 2D) cannot be detected by primer extension analysis (see below) and are likely pRNA I Ϫ5 and pppRNA I species truncated from the 3Ј-end (3Ј-⌬RNAI (I Ϫ5 )) by the action of PNPase (or RNase II). This implies that the conversion of RNA I and pRNA I Ϫ5 into smaller RNA I degradative products is retarded by the rne mutation and thus that RNase E is the enzyme responsible for their formation. Fig. 3, incubation of in vitro prepared GGGRNA I, which has the same secondary structure and substrate properties as native RNA I (27) with affinity-purified RNase E, leads to the formation of 70-mer, 67-mer, 41-mer, and 29-mer in addition to pRNA I Ϫ5 . To determine the nature of these species, we subjected GGGRNA I to RNase E treatment, and the resulting products were dephosphorylated with alkaline phosphatase and 5Ј-end labeled with 32 P. Then, the terminally labeled 70-mer, 67-mer, 41-mer, and 29-mer species were individually isolated and analyzed by RNase T 1 partial digestion as described under "Materials and Methods." To facilitate assignment of the position of G nucleotides in each RNA species, an alkaline hydrolysate of individually terminally labeled intermediate was included as a size marker. According to the results of RNase T1 mapping shown in Fig. 4 and Table I, the first base of 70-mer was A 6 and the last one is A 75 . The 67-mer also had  1 and 2) and for 1, 3, 10, and 30 min (lanes 3-6) with affinity-purified RNase E. The cleavage products were loaded on two separate 8% sequencing gels and electrophoresed for 2 or 3 h (left or right panel, respectively). Lanes T and L, RNase T 1 partial digest and partial alkaline hydrolysate of GGGRNA I, respectively. The major cleavage products (pRNA I Ϫ5 , 70-mer, 67-mer, 41-mer, and 29-mer) are shown.

FIG. 2. Northern blot analyses of RNA I molecules isolated from wildtype E. coli or from pcnB, rne, pnp mutant strains containing pCML108 plasmid (panel A).
Comparison of processing of RNA I (panel B) and pppRNA I Ϫ5 (panel C) in pcnB ϩ and pcnB mutant strains carrying either plasmid pCML108 or pCML103, respectively. Degradation of RNA I encoded by plasmid pCML108 in pcnB Ϫ /rne ts strain is shown (panel D). Processing of RNA in E. coli pcnB mutant strain carrying pCML108 plasmid at 33 and 43°C is shown (panel E). Isogenic strains YHC3393 (pcnB Ϫ ), YHC012 (pnp Ϫ ), and LK01 (pcnB Ϫ /rne ts ) were constructed as described under "Material and Methods." The corresponding host strain used for the individual experiment is indicated above each lane. Cells were grown exponentially in M9/glucose medium at temperature shown beneath of each lane to an OD 460 of 0.4 followed by addition of rifampicin (0.5 mg/ml). Total cellular RNA was isolated from aliquots withdrawn at successive times, as indicated in minutes at the top of each lane, after rifampicin addition. Equal amounts of total RNA were separated on a 8% polyacrylamide sequencing gel, transferred to membrane, probed for RNA I, and autoradiographed. Negative of 5 S rRNA used as internal loading standard and taken from the ethidium bromide staining gel prior to the Northern blotting is shown beneath panels B, C, and E. RNA extraction and Northern blot analysis were carried out according to the procedures described under "Materials and Methods." RNA I, pRNA I Ϫ5 , the decay intermediates (LMW), as well as pRNA I Ϫ5 and pppRNA I species truncated from the 3Јend (3Ј-⌬ RNA I (I Ϫ5 ) indicated in panel D) by the action of 3Ј-exoribonucleases are shown. A 6 as the first base, whereas the last one is C 72 . Formation of these degradative intermediates can be explained by RNase E cleavage of pRNA I Ϫ5 in stem-loop III at two different sites: after A 75 (70-mer) and after C 72 (67-mer). A more complicated situation was observed in the case of the 41-mer and 29-mer. The RNase T 1 digestion patterns of these oligoribonucleotides do not correspond to any definite fragment of RNA I; that is, the 29-mer and 41-mer appear to be a mixture of two or more oligoribonucleotides. On the other hand, the presence of fragments of 29 and 41 nucleotides in length in an RNase E digest of the 70-mer (data not shown), plus the existence of an RNase E cleavage site in a bubble region of a duplex segment of GGGRNA I between G 34 and U 35 (Fig. 5A primer extension analysis; the RNase E cleavage sites are indicated in lanes 6 and 7), suggests that the 29-mer and the 41-mer contain RNA species 29-mer* and 41-mer*, respectively (Fig. 4A). These species correspond to the degradative products of the 70-mer. However, the nature of other components of these mixtures could not be identified unambiguously, and thus these species were not investigated further.
Primer extension analysis of total cellular RNA I isolated from a pcnB mutant showed that, in addition to the expected cleavage site at the Ϫ5 position (the 5Ј-end of pRNA I Ϫ5 ), another cleavage occurred between G 34 and U 35 (Fig. 5B, lane  8). This cleavage was reduced sharply in an isogenic strain carrying the pcnB/rne double mutations under conditions inactivating RNase E (Fig. 5B, lane 1 versus lane 2). These results indicate that RNase E also cleaves the internucleotide bond between G 34 and U 35 in vivo. Using an RNase E digest of GGGRNA I as a size marker, we determined by Northern blot analysis that two major LMW intermediates approximately 71 and 73 nucleotides in length hybridized to a 32 P-labeled riboprobe complementary to 3Ј-end of RNA I (data not shown). Our data indicate that both LMW intermediates observed in pcnB mutant are decay intermediates that correspond to the 3Ј-end of pRNAI Ϫ5 and are generated by RNase E cleavage of an internucleotide bond in the bubble region of stem-loop I (shown in Fig. 1); the 2-nucleotide difference in the length of these intermediates is due to heterogeneity at the 3Ј-end (8).   RNase T 1 mapping (panel B). GGGR-NA I was subjected to RNase E treatment, and the resulting products were dephosphorylated with alkaline phosphatase and 5Ј-end labeled with 32 P. Then, the terminally labeled products were separated in 8% sequencing gel, and the 70mer, 67-mer, 41-mer, and 29-mer species were individually isolated. Each 5Ј-endlabeled oligoribonucleotide was cleaved by RNase T 1 (lane T), and the resulting mixtures were sized in an 12% sequencing polyacrylamide gel. G residues are shown in bold, and the numbers indicate their position relative to the 5Ј-end of the corresponding oligoribonucleotide. The first nucleotide of 70-mer and 67-mer corresponds to A 6 of RNA I (see Fig. 1), and 41-mer* and 29-mer* are degradative products of 70-mer. Positions of G residues (see Table I) are determined from concurrently run partial alkaline hydrolysates of corresponding terminally labeled oligoribonucleotides (lane L) and an RNase E digest of GGGRNA I (lane E). The position of several G residues from 5Ј-end are indicated. of degradative intermediates as full-length RNase E (Fig. 6) plus additional products (bands located between pRNA I Ϫ5 and 70-mer). We do not know whether an appearance of the additional bands reflects slightly broader substrate specificity of the truncated enzyme because of absence of its C-terminal fragment or because minimal distortions in the enzyme's structure appeared during its purification.

FIG. 4. The sequences of oligoribonucleotides, the major products of GGGRNA I cleavage by affinity-purified RNase E in vitro (panel A), were deduced on the base of results of
As found by McDowall and Cohen (14), the affinity of the catalytic domain to RNA I, in comparison with the affinity of an arginine-rich RBD located near the center of the Rne protein, is negligible. To understand better the role of the RBD, we examined the effect of the RBD on RNA I structure and the accessibility of its internucleotide bonds to nuclease attack using footprinting analysis with chemical nuclease. Comparison of the footprints of unbound RNA I and RNA I complexed with the RBD showed that no region of RNA I was significantly protected by the protein after binding to RBD (Fig. 7, lane 3 versus  lane 4); instead a number of internucleotide bonds located in loop regions of RNA I became more accessible to chemical cleavage by the phenanthroline-cuprous complex. In addition, increased cleavages were observed in the 3Ј-segment of RNA I in the presence of RBD. These results indicate that the RBD of RNase E can destabilize RNA I structure, possibly facilitating accessibility of internucleotide bonds to the cleavage.
As shown above, cleavage of RNA I by RNase E results in the formation of pRNA I Ϫ5 , 70-mer, 67-mer, 41-mer, 29-mer, and several short oligoribonucleotides. We used gel retardation assay to learn whether the RBD of RNase E retains the degradative intermediates produced by the catalytic domain of the enzyme. As seen in Fig. 8, the stability of the complex formed between these degradative intermediates and RBD directly correlates with the length of the oligoribonucleotides: the shorter the RNA fragment the less stable the complex. Thus, progressive degradation of the RNA appears to reduce the stability of the enzyme-substrate complex.

DISCUSSION
While pRNA I Ϫ5 is the principal product of RNase E cleavage of RNA I (22,23,29,30), the detection of an RNase E-dependent cleavage product about 70 nucleotides in length in vivo (23,30) and in vitro (30) suggests that additional RNase E cleavages play a role in the decay of RNA I. Analysis of the decay of RNA I encoded by pCML108, a pSC101-derived non-ColE1type plasmid, in a strain deficient for poly(A) polymerase (pcnB mutant) revealed several degradative products not detected in a pcnB ϩ strain ( Fig. 2A), indicating that polyadenylation, which accelerates the decay of RNA I and other RNA species (6 -8), can prevent detection of RNA I degradative intermediates in pcnB ϩ bacteria under wild-type conditions. However, these products were detected in a pcnB ϩ strain harboring pBR322-derived plasmid (23,30), suggesting that complementary RNA II affects RNA I decay, possibly by binding to RNA I degradative products. Our results show that the accumulation of these decay intermediates depends directly on RNase E activity and decreases in a rne ts strain at a non-permissive temperature. In vitro studies confirmed that GGGRNA I, and consequently RNA I, is cleaved at multiple sites (Fig. 3). The characterization of these degradative products by RNase T 1 mapping and primer extension analysis indicates that RNase E makes cuts preferentially in bubble regions of stem-loop segments of substrate RNA (namely, after G 34 , A 75 ), as well as in single-stranded regions.
While rne-dependent in vitro cleavage in a bubble region of RNA I encoded by plasmid pACYC184 was noted previously (30), the preparation of RNase E used in that study was partially purified and therefore potentially contained other proteins now known to be associated with RNase E (15)(16)(17)(18). Our finding that multiple cleavages are produced by affinity-purified enzyme provides direct evidence that cleavage of internucleotide bonds within structured regions of substrate RNA is due to RNase E itself. Thus, the cleavage potential of this enzyme may be broader than was previously thought.
Despite RNase E cleavage of GGGRNA I in vitro in both the stem-loop I and stem-loop III regions, Northern blot analysis of RNA derivatives accumulating in vivo showed products cleaved only in stem-loop I (71-mer and 73-mer). The absence of a detectable stem-loop III cleavage product in vivo may reflect difference in lability and, consequently, in abundance of different LMW intermediates in vivo. On the other hand, preferential formation in vivo of RNA I fragments (71-mer and 73-mer) containing intact stem-loop III also can be explained by an inaccessibility to nuclease of the internucleotide bonds located in stem-loop III, possibly because of their protection by other proteins associated with RNase E in vivo. A possible model for RNA I decay reflecting variations in the products generated by RNase E in vivo and in vitro is outlined in Fig. 9.
RNase E has been demonstrated to have separate catalytic (N-terminal half of enzyme) and RNA binding (middle part of protein) domains (14). While assaying the enzymatic activity of FIG. 7. Footprinting analysis of RBD-GGGRNA I complex. Free and bound 5Ј-end-labeled GGGRNA I were separated in native 4% polyacrylamide gel and subjected in situ (in gel) exposure to the relatively unspecific activity of the 2:1 1,10-phenanthroline-cuprous complex, followed by elution of partially cleaved RNA and subsequent sizing in 8% polyacrylamide sequencing gels (lanes 3 and 4, respectively) for 2 (left panel) and 4 (right panel) h. Band assignment was facilitated by coelectrophoresing a partial RNase T 1 digest (lane 1) and a partial alkaline hydrolyzing (lane 2) of the terminally labeled RNA employed in the assay. The loop regions of GGGRNA I are shown with brackets.
FIG. 8. Binding the RNA I and its rne-dependent degradative intermediates to RBD. Free GGGRNA I or its degradative intermediate (labeled as 1) and their individual mixtures with RBD-containing protein (labeled as 2) were electrophoresed in non-denaturing 4% polyacrylamide gel. The synthesis of GGGRNA I, its cleavage in vitro by RNase E, purification of corresponding cleavage products from gel, and gel retardation assay were performed according to procedures described under "Materials and Methods." FIG. 9. Schematic drawing illustrating RNA I decay pathway in pcnB mutant strain. RNase E associated with PNPase and other proteins (processosome) cuts RNA I into several major fragments that can be classified into two groups. The first one (A) comprises pRNA I Ϫ5 , 71-mer, and 73-mer. These species are relatively stable in the pcnB mutant but are quickly degraded by PNPase or RNase II because of polyadenylation in the wild-type strain. The other group (B) includes a 70-mer, 67-mer, 41-mer, and 29-mer, which were observed in vitro but were not detected by Northern blot analysis of RNA isolated from even pcnB mutant cells.
the affinity-purified catalytic domain, we showed that it cleaves RNA I, yielding not only pRNA I Ϫ5 (27) but also yielding all the other products observed following digestion of RNA I by full-length RNase E. This observation is consistent with findings that the middle part of RNase E, which contains the enzyme's RNA binding domain, does not participate in cleavage directly. Rather, this region of RNase E, because of its high affinity to RNA I in comparison with the affinity of catalytic domain (14), may instead promote targeting of the enzyme to the substrate.
Chemical in situ probing of the effect of the RBD on the structure of RNA I (Fig. 7) indicates that regions of secondary structure within the substrate are destabilized by formation of the RBD-RNA I complex. Thus, the action of the RBD in vivo may increase accessibility of the internucleotide bonds of RNA I to the catalytic domain of RNase E and/or other RNA processing enzymes associated with RNase E in the "processosome." Moreover, the major products produced by the cleavages of RNA I by RNase E are different in their affinity to RBD (Fig.  8). The affinity reduces with the decrease in number of RNase E cleavage sites in intermediate, and is minimal in case of 29-mer and 41-mer. These observations imply that, during sequential digestion of substrate RNA, RNase E can differentially release degradative products in vivo.
Collectively, our findings suggest a model for RNase E-dependent RNA degradation. We propose that the RBD of the enzyme initiates the recognition and binding of substrate RNA, destabilizing RNA tertiary and secondary structures, and than retains the target RNA until the catalytic domain makes multiple cuts. The cleaved fragments of substrate RNA generated by RNase E, which are differentially released by the enzyme, are further degraded by poly(A)-dependent 3Ј-exonucleolytic activity of PNPase or by RNase II.