Polyadenylation promotes degradation of 3'-structured RNA by the Escherichia coli mRNA degradosome in vitro.

Polyadenylation contributes to the destabilization of bacterial mRNA. We have investigated the role of polyadenylation in the degradation of RNA by the purified Escherichia coli degradosome in vitro. RNA molecules with 3'-ends incorporated into a stable stem-loop structure could not readily be degraded by purified polynucleotide phosphorylase or by the degradosome, even though the degradosome contains active RhlB helicase which normally facilitates degradation of structured RNA. The exoribonucleolytic activity of the degradosome was due to polynucleotide phosphorylase, rather than the recently reported exonucleolytic activity exhibited by a purified fragment of RNase E (Huang, H., Liao, J., and Cohen, S. N. (1998) Nature 391, 99-102). Addition of a 3'-poly(A) tail stimulated degradation by the degradosome. As few as 5 adenosine residues were sufficient to achieve this stimulation, and generic sequences were equally effective. The data show that the degradosome requires a single-stranded "toehold" 3' to a secondary structure to recognize and degrade the RNA molecule efficiently; polyadenylation can provide this single-stranded 3'-end. Significantly, oligo(G) and oligo(U) tails were unable to stimulate degradation; for oligo(G), at least, this is probably due to the formation of a G quartet structure which makes the 3'-end inaccessible. The inaccessibility of 3'-oligo(U) sequences is likely to have a role in stabilization of RNA molecules generated by Rho-independent terminators.

Polyadenylation of mRNA contributes to its stability and maturation and to the initiation of translation in eukaryotic cells (2)(3)(4). In prokaryotes, poly(A) polymerase activity was first identified over 3 decades ago (for a review of polyadenylation in bacteria, see Ref. 5). Nevertheless, only recently was polyadenylation considered to have a role in determining mRNA stability in bacteria. Two poly(A) polymerases have been cloned and characterized from Escherichia coli (6,7), and several mRNAs have been shown to possess poly(A) tails (6, 8 -12). Post-transcriptional addition of a poly(A) tail at the 3Ј-end of mRNA has been shown to destabilize certain RNA molecules (13,14). Furthermore, disruption of the pcnB gene, encoding poly(A) polymerase I, results in increased stability of RNA I (12,15) as well as lpp, ompA, rpsO, and trxA mRNAs in a pnp rnb rne background (9,11). This is in contrast to the situation in eukaryotes, where polyadenylation stabilizes RNA.
In E. coli, degradation of mRNA is mediated by the concerted action of endo-and exoribonucleases. A large multienzyme complex (the degradosome) includes two of the most important ribonucleases, the endoribonuclease RNase E and the 3Ј 3 5Ј exoribonuclease PNPase 1 (16 -19), together with the DEAD box helicase RhlB, enolase, and a non-stoichiometric amount of polyphosphate kinase (19,20). The roles of enolase and polyphosphate kinase in mRNA degradation are not yet clear. However, RhlB has a central role in the degradation of mRNAs with stable stem-loop structures (19). Purified PNPase is impeded by stem-loop structures in RNA, whereas PNPase in the degradosome can degrade structured RNA: in the degradosome, the RhlB helicase unwinds such structures in an ATPdependent fashion to permit the passage of PNPase. Thus, the degradosome is generally efficient at degrading structured RNA. However, in this study, we show that the degradosome is inefficient in degrading RNA molecules with their 3Ј-ends incorporated into a stem-loop structure. Addition of a singlestranded 3Ј-poly(A) tail facilitated degradation, providing a "toehold" permitting the degradosome to initiate its attack on the 3Ј-end of the RNA.
Oligonucleotides and DNA Fragments for in Vitro Transcription-For in vitro RNA degradation assays, RNA was transcribed from DNA templates generated by the polymerase chain reaction (PCR). DNA fragments from the intergenic region of the malE-malF operon of E. coli were amplified using pCH77 as template (22) with the same 5Ј-primer A (5Ј-AAATTAATACGACTCACTATAGGG-3Ј) and each of the three alternative 3Ј-primers: 3Ј-primer i, 5Ј-TGCCGGATGCGACGCTGACG-3Ј; 3Ј-primer ii, 5Ј-(T 15 )TGCCGGATGCGACGCTGACG-3Ј; and 3Јprimer iii, 5Ј-TTATCCGACAACAACTGCCGGATGCGACGCTGACG-3Ј. The PCR fragments generated (320, 335, and 335 bp, respectively) contained a T7 promoter at the 5Ј-end and could therefore be used directly for in vitro transcription (see below) to synthesize RNA molecules of 297, 312, and 312 nucleotides, respectively. All these RNAs also contained the A 3 C change indicated in Fig. 1. The 3Ј termini of these RNAs ended precisely at the 3Ј-base of the malE-malF stem-loop structure ("stem") (3Ј-primer i), with a 3Ј-A 15 tail ("stemA 15 ") (3Ј-primer ii), or with 15 nucleotides of the original malE-malF sequence downstream of the stem-loop ("stemN 15 ") (3Ј-primer iii). To transcribe RNA molecules with 5 or 10 adenosine residues or 15 guanosine or 15 uridine residues downstream of the stem-loop, derivatives of 3Ј-primer ii were used in which the T 15 sequence (boldface) was exchanged for the appropriate C or A nucleotides.
Two DNA fragments in which the malE-malF intergenic region was shortened at the 5Ј-end by 86 nucleotides, compared with the fragments described above, were amplified using 5Ј-primer B (5Ј-AAATAATAC-GACTCACTATAGGGCCCGCAGATGTCCGCTTTC-3Ј) together with each of 3Ј-primers ii and 3Ј-iii to generate 3Ј-A 15 or 3Ј-N 15 tails downstream of the stem-loop, respectively. These PCR products (250 bp) also contained a T7 promoter at the 5Ј-end such that in vitro transcription generated RNA molecules of 228 bp (named "sstemA 15 " and "sstemN 15 ").
A DNA fragment in which the stem-loop structure of the malE-malF region was replaced by the terminator stem-loop of the lpp gene (for the mRNA sequence, see Ref. 22) was amplified using 5Ј-primer A (see above) and the 3Ј-primer 5Ј-AAAAAAATGGCGCACAATGTGCGC-CATTTTTCATTTCACAGCATTACTTGGTGAT-3Ј. The PCR product (266 bp) used as a template for in vitro transcription gave a 243nucleotide RNA molecule (named "lpp-stem").
In Vitro Transcription-Radiolabeled RNA was synthesized in vitro from linearized, gel-purified DNA templates using bacteriophage T7 RNA polymerase (Pharmacia) and [␣-32 P]CTP (400 Ci/mmol; Amersham) as described previously (22). The transcription reactions were diluted to 70 l in water, extracted with phenol/chloroform, and eluted through a Push column (Stratagene) to separate unincorporated nucleotides. The RNA was precipitated with ethanol and resuspended in water (MilliQ). The RNA products stemA 15 , stemU 15 , and sstemA 15 migrated as broad bands (especially sstemA 15 in Fig. 8, band e), probably because runoff transcription does not stop precisely at the end of the DNA template due to the weak deoxyribo(T) n -ribo(A) n or deoxyribo(A) n -ribo(U) n interactions.
Polyadenylation of mRNA-In vitro transcribed, radiolabeled RNA was incubated with E. coli poly(A) polymerase and 5 mM ATP in the buffer recommended by the manufacturer (Cambio) for 20 min at 37°C. The reaction mixture was extracted with phenol/chloroform and further treated as described above (see "In Vitro Transcription").
In Vitro RNA Degradation Assays-In vitro RNA degradation assays were performed as described (19). The reaction (100 l) contained exonuclease assay buffer (20 mM Tris-HCl (pH 7.5), 1 mM MgCl 2 , 20 mM KCl, 5 mM dithiothreitol, and 10 mM K 2 HPO 4 ), 1 mM ATP, and 80 units of RNasin (Promega) and was incubated with enzyme at 37°C. Unless otherwise stated, 0.18 g of His-tagged PNPase or 0.66 g of the degradosome was added to each reaction. Aliquots were taken at appropriate time points, loaded onto a 4 or 6% sequencing gel, and electrophoresed at 60 watts. The gel was dried on Whatman No. 3MM paper, and radioactivity in each band was quantitated with a PhosphorImager.
Cloning of the Gene Encoding Polynucleotide Phosphorylase-The gene encoding PNPase was amplified by PCR from E. coli HB101 genomic DNA using Vent polymerase (New England Biolabs Inc.). Oligonucleotide primers 5Ј-ATATTAATCATATGCTTAATCCGATCGT-T-3Ј and 5Ј-CCCCAAGCTTCTCGCCCTGTTCAGC-3Ј were based on the pnp sequence (GenBank TM accession number U18996) and amplified the pnp gene flanked by NdeI and HindIII restriction sites (boldface) to facilitate subsequent cloning. After amplification, gel purification, and cleavage with HindIII, the 2254-bp PCR product was cloned into pBluescript II KS(ϩ) digested with SmaI and HindIII. The pnp gene was excised from the resulting plasmid with NdeI and HindIII and recloned into pET-20b (NdeI/HindIII). The resulting plasmid, pPNPHIS, contained the pnp gene under the control of the T7 promoter, allowing inducible overexpression of a PNPase derivative with a C-terminal hexahistidine tag in E. coli strain BL21.
Purification of His-tagged PNPase-His-tagged PNPase was purified by metal-chelate affinity chromatography (24) as described previously for His-tagged polyphosphate kinase (20) with the following modifications. Buffer A for resuspending the cell pellet and for washing steps consisted of 0.1 M Na 2 HPO 4 /NaH 2 PO 4 (pH 7.8), 10% glycerol, 1% Tween, 1 M NaCl, and 5 mM imidazole. The protein was eluted in steps using 5 ml of buffer B (0.1 M Na 2 HPO 4 /NaH 2 PO 4 (pH 7.8), 10% glycerol, and 1% Tween) containing 30, 50, 80, 120, and 500 mM imidazole, respectively. His-tagged PNPase was purified to near homogeneity as judged by Coomassie Blue staining (see Fig. 4A). The fractions containing the protein peak were pooled and stored in 50% glycerol (final concentration) at Ϫ20°C. Protein concentrations were determined as described (25) with bovine serum albumin as standard.

Effects of 3Ј-Adenylation on the Stability of Structured RNA
in Vitro-To examine the influence of polyadenylation on RNA degradation in vitro, various polyadenylated and non-polyadenylated mRNAs were incubated with the degradosome in the presence or absence of 1 mM ATP. As the primary substrate a segment of malE-malF mRNA (19,20) was used (named stem). The 3Ј-end of this RNA corresponded to the last nucleotide of the stable stem-loop structure of the malE-malF intergenic region (Fig. 1A), such that the 3Ј-end was fully incorporated into the secondary structure. Derivatives of this mRNA were generated with an additional 15 adenosine residues (stemA 15 ) or a poly(A) tail ("stem-poly(A)" containing 200 -300 adenosine residues), providing a single-stranded 3Ј-tail. These substrates were incubated with the degradosome, samples were taken at defined time points, and the products were separated on a denaturing gel. Fig. 2A shows an example of such a gel. The rate of disappearance of the mRNA species was quantified using a PhosphorImager (Fig. 2B).
In all reactions, the oligo(A) tail (when present) was rapidly removed to generate a stable intermediate with its 3Ј-end close to the base of the stem-loop structure (Fig. 2) (22). The curves are biphasic (Fig. 2B) because, for a small proportion of the FIG. 1. RNA secondary structures. A, sequence of the malE-malF stem-loop structure. In all experiments described here, a derivative was used with an A 3 C change, as indicated, to correct the mismatch at the base of the stem and to ensure that no single-stranded "tail" is exposed. B, terminator of the lpp mRNA (23).
RNA starting material, it appears the degradosome falls off the RNA at the base of the stem-loop (having removed the poly(A) tail), and this species is then unable to be degraded further (see below). In the absence of ATP this intermediate RNA species was only poorly degraded by the degradosome, irrespective of the starting RNA from which it was generated, because ATP is required for the RhlB helicase to unwind the secondary structure (19). In the presence of ATP, however, this intermediate was rapidly degraded, but only for the RNA molecules that initially had a single-stranded A 15 or poly(A) tail at their 3Јends (stemA 15 and stem-poly(A)); the RNA lacking any nucleotides added to the 3Ј-base of the stem-loop (stem) was still resistant to degradation (Fig. 2). The degradation of RNAs containing 5 or 10 adenosine residues did not differ significantly from those containing 15 adenosine residues or a poly(A) tail, demonstrating that the length of the oligo(A) tail is relatively unimportant (Fig. 3).
As a control, we purified His-tagged PNPase (Fig. 4A). This C-terminally His-tagged PNPase was active as a 3Ј-exoribonuclease (Fig. 4B), in contrast to N-terminally His-tagged PN-Pase which is inactive (17). The stem and stemA 15 RNA molecules were incubated with the purified His-tagged PNPase. The purified exoribonuclease did not degrade either stem RNA or stemA 15 RNA, apart from removing the oligo(A) tail (Fig. 4C), confirming that, under these conditions, the RhlB helicase of the degradosome is required for the stem-loop structure to be degraded (19,22).
Effect of Alternative Tails on the Degradation of Structured RNA-Since addition of a 3Ј-poly(A) tail stimulated degradation of the stem-loop structure by the degradosome, we assessed whether there was any specificity for the composition of the 3Ј-tail. The rate of degradation of RNAs with a 3Ј-tail of 15 guanosines ("stemG 15 ") or 15 uridines (stemU 15 ) was compared with that of the stem, stemA 15 , and stemN 15 substrates. The stemN 15 substrate incorporated a 3Ј-tail consisting of the 15 nucleotides normally 3Ј of the stem-loop in the malE-malF mRNA. Assays were in the presence of 1 mM ATP. The stemN 15 RNA was degraded as rapidly as the stemA 15 RNA (Fig. 5). In contrast, the stemG 15 and stemU 15 RNAs were degraded slowly, at a rate similar to that of the stem RNA. Furthermore, and in contrast to the stemA 15 , stemN 15 , and stemU 15 RNA substrates, the oligo(G) tail of the stemG 15 RNA was removed poorly by the degradosome (Fig. 6). Incubation of RNA substrates with purified His-tagged PNPase showed that the purified enzyme was also unable to degrade the oligo(G) tail (Fig.  6), demonstrating that the oligo(G) tail is not accessible to either enzyme.
Degradation of Native malE-malF RNA Intermediates in Vitro-In vivo, transcription of the malE-malF mRNA terminates several hundred base pairs downstream of the large intergenic stem-loop. A stable RNA molecule with the stemloop near its 3Ј-end is then generated by processive exoribonucleolytic degradation from the 3Ј-end (22,26). To mimic the in vivo situation in vitro, stemN 15 RNA was incubated with Histagged PNPase for 15 min; the reaction was stopped; and the resulting RNA was purified. Fig. 7B shows that all the initial substrate was converted to the stable intermediate, designated "stem(N 15 )*." Aliquots of stem(N 15 )* RNA were then polyadenylated and used as substrate for the degradosome in the presence of ATP. Although the residual 3Ј-tail on stem(N 15 )*, compared with stem RNA, permitted slightly faster degradation of the stem-loop structure, the degradosome degraded the polyadenylated RNA significantly faster than the non-polyadenylated precursor (Fig. 7, A and C). Similar results were obtained when the stem(N 15 )* RNA was generated using the degradosome (in the absence of ATP) rather than His-tagged PNPase (data not shown). Thus, polyadenylation enhances the rate of degradation of intermediates equivalent to those generated in vivo.

E. coli PNPase Does Not Have a Preference for 3Ј-Poly(A) Tails Compared with Other
Single-stranded 3Ј-Tails-To determine whether PNPase has higher affinity for poly(A) tails than for generic single-stranded 3Ј-ends, the rate of degradation of the N 15 tail of stemN 15 RNA was compared in the presence of different competing RNA substrates. 5Ј-Truncated malE-malF competitor RNAs were constructed to distinguish

FIG. 4. Purified C-terminally His-tagged PNPase is active.
A, Coomassie blue-stained protein gel loaded with 1.2 g of purified His-tagged PNPase (Pnp-His 6 ; arrow). B and C, autoradiographs of 6% gels of RNA degradation assays as a function of time (in minutes). B, 8 ng of purified His-tagged PNPase was incubated with malE-malF mRNA (19) to demonstrate that the enzyme is active. Because this study was designed to show that the His-tagged PNPase was active, only a small amount of enzyme was used. Under these conditions, the enzyme paused at small stem-loop structures (as does native PNPase) (22) and finally stalled at the large stem-loop, generating a stable degradation intermediate (arrow). C, His-tagged PNPase did not degrade the stem RNA (left panel). Only the A tail of the stemA 15 RNA substrate (right panel) was degraded by His-tagged PNPase, again generating a stable intermediate (arrow). The intermediate migrated slightly slower than the stem RNA, indicating that PNPase does not remove the entire 3Ј-tail, but leaves a few nucleotides (nt) of the base of the stem-loop structure (as shown previously (22)). them from the substrate RNA following gel electrophoresis. These shorter RNAs (sstemA 15 and sstemN 15 ) ended with an A 15 or N 15 tail 3Ј to the stem-loop structures, respectively. In this assay, the concentration of enzyme was reduced to amounts equimolar to the substrate RNA. The competitor RNAs were also added in equimolar amounts. The rate of degradation of the N 15 tail of stemN 15 in the presence of either competitor (band a going to band b) calculated from the data shown in Fig. 8 give the time, in which 50% of the full-length stemN 15 substrate is shortened to the base of the stem-loop structure (td 50 ), as 0.5 min for the sstemN 15 substrate alone. Addition of either of the competing substrates (sstemN 15 or sstemA 15 ) resulted in a similar increase in td 50 to ϳ2 min. The removal of the 3Ј-tails from sstemN 15 and sstemA 15 can also be seen to occur at a similar rate (generated band d and f, respec-tively). These data demonstrate that, as long as the 3Ј-tail is single-stranded and accessible to the enzyme (unlike the G 15 tail; see above), it does not matter whether the 3Ј-tail is poly(A) or is a generic sequence.
Degradation of RNA Containing the lpp Terminator Stemloop-To demonstrate the general relevance of polyadenylation, the degradation of a primary transcript ending in a native terminator stem-loop was characterized. The lpp terminator (Fig. 1B) was selected because the lpp mRNA is relatively stable and polyadenylation has been reported in vivo (6,27). The native lpp terminator was used to replace the large stemloop of the malE-malF RNA, and the resulting mRNA (lppstem) and its polyadenylated derivative were used as substrates for the degradosome and for His-tagged PNPase. Equimolar amounts of PNPase were used, whether His-tagged or in the degradosome (0.18 g of purified His-tagged PNPase is equivalent to 0.66 g of degradosome containing 28% PN-Pase (19)). In contrast to the malE-malF stem-loop, ATP was not required for degradation of the lpp stem-loop by the degradosome, presumably because the lower thermodynamic stability of the stem-loop is such that the RhlB helicase is not essential for unwinding (Fig. 9). Therefore, addition of ATP promoted the degradation only slightly. Nevertheless, the presence of a 3Ј-poly(A) tail still markedly stimulated degradation of the RNA (whether or not ATP was present) both by the degradosome and by purified PNPase (Fig. 9).

DISCUSSION
Polyadenylation of bacterial mRNA is increasingly recognized as a potentially important destabilizing signal for the RNA degradation machinery. RNA molecules or fragments generated by endoribonuclease cleavage are degraded by one of  15 was used (*) has not been identified. One explanation might be that the structure of the RNA is changed by the oligo(U) sequence and that an RNase E site is exposed. B, graph showing rates of RNA degradation, derived by phosphoimaging of the gel in A. Derivatives of malE-malF RNA containing an A 15 (Ⅺ) or an N 15 (‚) tail downstream of the stem-loop structure were degraded by the degradosome faster than RNAs with a G 15 (f) or an U 15 (OE) tail or the 3Ј-structured stem RNA itself (q).
FIG. 6. Accessibility of RNAs with different homopolymeric tails. Shown is autoradiograph of a denaturing gel. The indicated malE-malF RNA derivatives were incubated with the degradosome (D), with His-tagged PNPase (P), or with no added enzyme (Ϫ) for 10 min (stemA 15 ) or for 30 min (stemG 15 and stemU 15 ) in the presence of ATP. the 3Ј 3 5Ј exoribonucleases: PNPase or RNase II. These processive exoribonucleases are stalled by stem-loop structures; temporarily at smaller stem-loop structures, but more stable stem-loops (e.g. the malE-malF intergenic stem-loop structure) (19,22) or complex secondary structures (3Ј-end of the S20 mRNA) (14) act as a very effective block to these enzymes. In the cell, however, a proportion of PNPase is part of a multiprotein complex, the degradosome (16 -19). The ATPdependent helicase activity of RhlB in the degradosome strongly promotes the degradation of structured RNA by unwinding sequences and offering them as single-stranded substrates to PNPase (19). Thus, the degradosome appears to be fully equipped to degrade highly structured RNA molecules (19). Since polyadenylation is important for destabilizing structured RNA molecules (9,11,13), we set out to address the apparent paradox of how polyadenylation might stimulate the activity of the degradosome. In particular, it has previously been suggested that polyadenylation may provide a point of access for ribonucleases (11); this idea was tested experimentally for the degradosome.
The malE-malF intergenic region forms a very stable stemloop structure, and its degradation does not require RNase E cleavage (19). 2 The degradosome and purified PNPase are stalled by the stem-loop structure to give a very stable intermediate in vivo and in vitro (19, 22, 26 -29). A derivative of the malE-malF message with its 3Ј terminus incorporated into the stem-loop structure was degraded very poorly by the degradosome in vitro, even when RhlB was active. In contrast, 3Јadenylated derivatives of the RNA were degraded rapidly. This demonstrates that the ability of the degradosome can be limited by the absence of a single-stranded 3Ј-RNA end. Addition of an oligo(A) tail efficiently destabilized the message in vitro, even when as few as 5 nucleotides were added. Fewer than 5 adenosine residues did not seem to be recognized (data not shown). Although it has been suggested that as few as 2 adenosine residues might target RNA for degradation (12), these studies were carried out in vivo, and nucleases other than the degradosome might be involved. Similarly, Coburn and Mackie (14) have indicated that RNA with Ͻ6 -10 unpaired 3Ј-residues is stable; these were not adenosines, and the experiments were carried out with purified nucleases, not the degradosome. Although the model malE-malF transcript studied initially is not normally generated in vivo, we also showed that polyadenylation stimulated degradation of a malE-malF intermediate equivalent to that generated in vivo and of the lpp 3Ј-stem-loop generated in vivo by transcription termination. Thus, polyadenylation appears to stimulate degradation by the degradosome of any structured RNA generated as a termination product or degradation intermediate where the 3Ј-end has a limited single-stranded tail. Since an N 15 generic single-stranded sequence of the stem-loop was as effective as the A 15 sequence and since A 15  degradation, the destabilizing effect appears be due to the single-stranded nature of the 3Ј-end rather than to a high affinity for poly(A) per se.
Interestingly, a 3Ј-oligo(G) tail did not stimulate degradation of RNA in vitro (Fig. 5). Furthermore, neither the degradosome nor purified PNPase could even shorten the G 15 tail, suggesting that the sequence is inaccessible to these enzymes. G-rich sequences are able to form intramolecular or intermolecular G quartets (30,31), which are very resistant to nucleases (32) and which presumably render the 3Ј-end inaccessible to enzymatic attack.
A 3Ј-U 15 tail also failed to destabilize the RNA. There is no known structure an oligo(U) tail might adopt, although in contrast to the G 15 tail, the U 15 tail was shortened by the degradosome and by PNPase in vitro (Fig. 6), showing that it remains accessible. More important, every RNA generated from a gene with a Rho-independent terminator ends in a 3Ј-oligo(U) tail. Thus, apart from the terminator stem-loop itself, this oligo(U) sequence of the terminator may contribute significantly to the stability of such mRNAs, protecting against 3Ј 3 5Ј exoribonucleolytic attack.
Recently, a purified N-terminal fragment of RNase E was reported to remove 3Ј-poly(A) and 3Ј-poly(U) tails (1). However, several lines of evidence argue against the possibility that RNase E, rather than PNPase, is responsible for the exoribonucleolytic activities of the degradosome observed here. First, the exoribonucleolytic activity of the degradosome containing a temperature-sensitive RNase E is still efficient at the nonpermissive temperature (19). Since both the endoribonucleolytic and poly(A) tail-shortening activities of RNase E are in the same N-terminal region of the polypeptide (1), the poly(A) shortening should also be inactive at the nonpermissive temperature. Second, purified PNPase gave the same pattern of 3Ј-poly(A) tail shortening as the degradosome. Third, characterization of the released nucleotides by thin-layer chromatography allowed us to distinguish which enzyme is involved. 3 Thus, the reported exoribonucleolytic activity reported for a fragment of RNase E is unlikely be significant in RNA degradation by the degradosome.
In conclusion, we have shown that polyadenylation can play an important role in the degradation of mRNA molecules that have 3Ј-terminal secondary structures, providing a singlestranded toehold recognized by the degradosome. 3Ј-Structured RNAs are normally generated by Rho-independent termination, by endonuclease cleavage, or by stalling of 3Ј 3 5Ј exonucleases at an internal secondary structure. Poly(A) polymerase can potentially pin a label "to be degraded" to structured 3 C. Burns and C. F. Higgins, unpublished data. RNA generated by any of these processes that is otherwise inaccessible to ribonucleases. Thus, "initiation of degradation" of structured RNA by polyadenylation may play an important role not only in the degradation of functional RNA molecules with inaccessible (structured) 3Ј-ends, but also in recycling the small structured RNA degradation intermediates generated during cellular RNA turnover.