Prolyl-tRNAPro in the A-site of SecM-arrested Ribosomes Inhibits the Recruitment of Transfer-messenger RNA*

Translational pausing can lead to cleavage of the A-site codon and facilitate recruitment of the transfer-messenger RNA (tmRNA) (SsrA) quality control system to distressed ribosomes. We asked whether aminoacyl-tRNA binding site (A-site) mRNA cleavage occurs during regulatory translational pausing using the Escherichia coli SecM-mediated ribosome arrest as a model. We find that SecM ribosome arrest does not elicit efficient A-site cleavage, but instead allows degradation of downstream mRNA to the 3′-edge of the arrested ribosome. Characterization of SecM-arrested ribosomes shows the nascent peptide is covalently linked via glycine 165 to \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{tRNA}_{3}^{\mathrm{Gly}}\) \end{document} in the peptidyl-tRNA binding site, and \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{prolyl}\mathrm{-}\mathrm{tRNA}_{2}^{\mathrm{Pro}}\) \end{document} is bound to the A-site. Although A-site-cleaved mRNAs were not detected, tmRNA-mediated ssrA tagging after SecM glycine 165 was observed. This tmRNA activity results from sequestration of \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{prolyl}\mathrm{-}\mathrm{tRNA}_{2}^{\mathrm{Pro}}\) \end{document} on overexpressed SecM-arrested ribosomes, which produces a second population of stalled ribosomes with unoccupied A-sites. Indeed, compensatory overexpression of \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{tRNA}_{2}^{\mathrm{Pro}}\) \end{document} readily inhibits ssrA tagging after glycine 165, but has no effect on the duration of SecM ribosome arrest. We conclude that, under physiological conditions, the architecture of SecM-arrested ribosomes allows regulated translational pausing without interference from A-site cleavage or tmRNA activities. Moreover, it seems likely that A-site mRNA cleavage is generally avoided or inhibited during regulated ribosome pauses.

A-site 3 mRNA cleavage is a novel RNase activity that acts on A-site codons within paused ribosomes. Ehrenberg, Gerdes and their colleagues (1) first demonstrated that Escherichia coli RelE protein causes cleavage of A-site mRNA in vitro. Subsequently, A-site cleavage was also shown to occur at stop codons during inefficient translation termination in cells that lack RelE and related proteins (2,3). The latter finding indicates that another unknown A-site nuclease also exists in E. coli. Indeed, it is possible the ribosome itself catalyzes A-site cleavage. The molecular requirements for A-site cleavage are incompletely understood, but an unoccupied ribosome A site appears to be important for both RelE-dependent and RelE-independent nuclease activity (1,2).
A-site nuclease activity truncates mRNAs and produces stalled ribosomes that are unable to continue standard translation. In bacteria, ribosomes stalled at the 3Ј termini of such truncated messages are "rescued" by the tmRNA quality control system. tmRNA is a specialized RNA that acts first as a tRNA to bind the A-site of stalled ribosomes, and then as an mRNA to direct the addition of the ssrA peptide degradation tag to the C terminus of the nascent polypeptide (4,5). As a result of tmRNA activity, incompletely synthesized proteins are targeted for proteolysis and stalled ribosomes undergo normal translation termination and recycling (5). In this manner, A-site mRNA cleavage and tmRNA work together as a translational quality control system that responds to paused and stalled ribosomes.
Although a paused ribosome can be a manifestation of translational difficulty, translational pausing is also used to control and regulate protein synthesis. In many instances, the newly synthesized nascent peptide inhibits either translation elongation or termination (6,7). A recently described example is the SecM-mediated ribosome arrest, which controls expression of SecA protein from the secM-secA mRNA in E. coli (8). The SecM nascent peptide interacts with the ribosome exit channel to elicit a site-specific ribosome arrest (9). The SecM-stalled ribosome is postulated to disrupt a downstream mRNA secondary structure that sequesters the secA ribosome binding site (9,10). Thus, efficient initiation of secA translation depends upon ribosome pausing at the upstream secM open reading frame (11). SecM-mediated ribosome pausing is regulated in turn by the activity of SecA protein. SecM is secreted co-translationally by the general Sec machinery, which is powered in part by the SecA ATPase (12). It is thought that the mechanical pulling force exerted by SecA on the SecM nascent chain during secretion alleviates the ribosome arrest and allows translation to continue (13,14). This intriguing regulatory circuit allows the cell to monitor protein secretion activity via ribosome pausing and adjust SecA synthesis accordingly.
One outstanding question is how A-site cleavage and tmRNA activities affect regulatory translational pauses such as the SecM-mediated ribosome arrest. If all paused ribosomes are subject to A-site cleavage, then this nuclease activity would be expected to interfere with SecA regulation. The experiments presented in this paper demonstrate that A-site mRNA cleavage and the tmRNA quality control system do not significantly affect SecM-mediated ribosome arrest. Two recent reports have demonstrated that the A-site of SecM-arrested ribosomes is filled with tRNA (15,16). The cryo-EM structure from Frank and colleagues (15) shows that ϳ40% of SecM-arrested ribosomes contain a fully accommodated A-site tRNA. Ito and colleagues (16) have recently analyzed SecM-arrested ribosomes prepared by in vitro translation and concluded that the A-site tRNA is a prolyl-tRNA Pro . In our analysis of SecM-arrested ribosomes in vivo, we also find that the P-and A-sites of the SecM-arrested ribosome are occupied with peptidyl-and aminoacyl-tRNAs, respectively. Additionally, we show that the occupied A-site prevents tmRNA recruitment during ribosome arrest and may also inhibit A-site mRNA cleavage. Thus, regulation by SecM ribosome arrest is able to operate efficiently in the presence of quality control systems that alleviate ribosome stalling.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Plasmids- Table 1 lists the bacterial strains and plasmids used in this study. All bacterial strains were derivatives of E. coli strain X90 (17). Strain CH12 (X90(DE3)) was generated using the Novagen (DE3) lysogen kit according to the manufacturer's instructions. Strain CH2198 (X90 ssrA(his6)(DE3)) was obtained by introducing the ssrA(His6) allele (18) of tmRNA into the ssrA chromosomal locus using the phage Red recombination method with minor modifications (17,19). The same method was used to delete the rna (encoding RNase I), rnb (encoding RNase II), and pnp (encoding PNPase) genes. The rnr::kan disruption and the strain expressing truncated RNase E have been described previously (2,20). All gene disruptions and deletions were introduced into strain CH113 by phage P1-mediated transduction. The kanamycin resistance cassette was removed from strain CH113 ⌬rnb::kan using FLP recombinase as described (19), allowing construction of the ⌬rnb rnr::kan double mutant.
Lac Ϫ strains of X90 and X90 ssrA::cat were obtained by curing the strain of the FЈ episome as described (17). The details of all strain constructions are available upon request.
mRNA Expression and RNA Analysis-E. coli strains were grown overnight at 37°C in LB medium supplemented with the appropriate antibiotics (150 g/ml of ampicillin, 25 g/ml of tetracycline, or 50 g/ml of kanamycin). The next day, cells were resuspended at an optical density at 600 nm (A 600 ) of 0.05 in 15 ml of fresh medium and grown at 37°C with aeration. Once cultures reached an A 600 of ϳ0.3, mRNA expression was induced with isopropyl ␤-D-thiogalactopyranoside (1.5 mM). After further incubation for 30 min, 15 ml of ice-cold methanol was added to the cultures, the cells collected by centrifugation, and the cell pellets frozen at Ϫ80°C. Total RNA was extracted from cell pellets using 1.0 ml of a solution containing 0.6 M ammonium isothiocyanate, 2 M guanidinium isothiocyanate, 0.1 M sodium acetate (pH 4.0), 5% glycerol, 40% phenol. The disrupted cell suspension was extracted with 0.2 ml of chloroform, the aqueous phase removed and added to an equal volume of isopropyl alcohol to precipitate total RNA. RNA pellets were washed once with ice-cold 75% ethanol and dissolved in either 10 mM Tris-HCl (pH 7.5), 1 mM EDTA or 10 mM sodium acetate (pH 5.2), 1 mM EDTA.
SecM Expression and Protein Analysis-Strains were cultured as described above for RNA analysis. Protein extraction and Western blot analyses were conducted as described (17). Anti-His 6 polyclonal antibodies were obtained from Santa Cruz Biochemical. Monoclonal antibodies specific for E. coli ␤-galactosidase and the FLAG M2 epitope were obtained from Sigma. SsrA(His 6 )-tagged SecM proteins were purified by Ni 2ϩ -NTA agarose (Qiagen) affinity chromatography as described (17,18). Ni 2ϩ -NTA purified protein was further purified by reverse phase HPLC. N-terminal gas-phase sequencing was performed on a Porton 2020 protein sequencer (Beckman-Coulter) with a dedicated in-line HPLC (model 2090) for separation of phenylthiohydantoin derivatives. Molecular masses were determined by liquid chromatography mass spectrometry. Samples were applied to a Zorbax 300SB-C18 reverse phase column in aqueous 0.1% formic acid and proteins eluted using a linear gradient of acetonitrile using an Agilent 1100 LC nano-system. Eluted proteins were infused into a Waters Q-Tof II TM mass spectrometer for ionization.
␤-Galactosidase assays were conducted essentially as described (23). Strains expressing secAЈ::lacZ translational fusions were inoculated at A 600 of 0.05 in LB medium and grown at 37°C with aeration to A 600 of 0.3-0.6. ␤-Galactosidase activity for each construct was measured from 5 to 8 independent cultures and reported as mean Ϯ S.D.
Cell Extract Fractionation-Strains CH12 ⌬rna::kan and CH113 ⌬rna::kan containing plasmid pSecMЈ or pSecMЈ-(P166A) were grown in 1 liter of LB media at 37°C with aeration in Fernbach flasks. At A 600 ϳ 0.6, SecMЈ expression was induced by the addition of isopropyl ␤-D-thiogalactopyranoside to 1.5 mM and cultures incubated for 1 h at 37°C with aeration. Cultures were harvested over ice, the cells were collected by centrifugation and washed once with cold, high-Mg 2ϩ S30 buffer (60 mM potassium acetate, 30 mM magnesium acetate, 0.2 mM EDTA, 10 mM Tris acetate (pH 7.0)). Washed bacterial pellets were resuspended in 10 ml of cold high-Mg 2ϩ S30 buffer and the cells were broken by one passage through a French press at 12,000 p.s.i. Cell lysates were cleared by centrifugation at 30,000 ϫ g for 15 min at 4°C, and the supernatants layered onto cushions of cold high-Mg 2ϩ S30 buffer containing 1.1 M sucrose in ultracentrifuge tubes (Beckman number 344057). Samples were centrifuged in a Beckman-Coulter Optima TM ultracentrifuge at 45,000 ϫ g for 1 h at 4°C using an MLS-50 rotor. Total RNA was extracted from the high-speed supernatants and pellets for analysis as described above.

RESULTS
SecM Ribosome Arrest Leads to mRNA Cleavage-To determine whether SecM-mediated ribosome arrest leads to A-site cleavage, we generated plasmids to express mRNA encoding SecM and the first 62 residues of SecA (Fig. 1A). Three SecM variants were used throughout this work: (i) FLAG-SecM, which is the wild-type protein fused to an N-terminal FLAG epitope tag; (ii) FLAG-(⌬ss)SecM, which lacks the secretion signal sequence (⌬ss ϭ deleted signal sequence; residues 1-37); and (iii) FLAG-(⌬ss-P166A)SecM, which lacks the signal sequence and has alanine in place of proline 166. Deletion of the SecM signal sequence prevents its secretion and leads to a profound ribosome arrest, whereas the P166A variant completely abrogates arrest (9,14). The FLAG sequence was added to facilitate analysis of SecM proteins by Western blot. However, secretion of FLAG-SecM protein resulted in the removal of the FLAG epitope along with the signal sequence (see below).
Each SecM protein was expressed in wild-type cells (tmRNA ϩ ) and cells that lack tmRNA (⌬tmRNA), and the corresponding messages examined by Northern blot analysis using a probe specific for the ribosome binding site upstream of secM. In addition to the full-length mRNAs, truncated flag-secMAЈ and flag-(⌬ss)secMAЈ messages were also detected (Fig. 1B). The truncated mRNAs did not hybridize to a probe specific for the downstream secA sequence (data not shown). No truncated flag-(⌬ss-P166A)secMAЈ mRNA was apparent, suggesting that ribosome arrest was required for mRNA cleavage. Interestingly, steady state levels of truncated flag-secMAЈ and flag-(⌬ss)secMAЈ mRNAs were similar in wild-type and ⌬tmRNA cells (Fig. 1B). This finding was noteworthy because tmRNA activity usually promotes rapid degradation of truncated mRNAs, including those produced by A-site mRNA cleavage (2,3,24). Moreover, the truncated mRNAs appeared to be somewhat larger than in vitro transcripts that terminate in the codon for glutamine 167, a position that is adjacent to the A-site of the arrested ribosome (Fig. 1, A and B) (16).
The 3Ј ends of the truncated messages were mapped more precisely using S1 nuclease protection analysis. The termini were somewhat heterogeneous but strong cleavage was detected inside and adjacent to the secM stop codon (Fig. 1C). A, secMAЈ mRNA variants are shown with the FLAG epitope, signal sequence, and oligonucleotide probe binding sites indicated. SecM residues glycine 165 to threonine 170 and the encoding mRNA sequence are shown, as is the complementary sequence of the S1 nuclease probe and the 3Ј terminus of the truncated in vitro transcripts used. The position of the P166A alteration is indicated by (Ala). Arrows indicate the positions of HphI and Sau96I restriction endonuclease cleavages in the S1 probe used to generate gel migration standards. B, Northern blot of secMAЈ mRNAs purified from tmRNA ϩ and ⌬tmRNA cells. The positions of full-length and truncated flag-(⌬ss)secMAЈ mRNA are indicated. Both in vitro transcripts were truncated after the second nucleotide of the glutamine 167 codon. C, S1 nuclease protection map of truncated secMAЈ mRNAs. Cleavages were detected in the secM stop codon and at positions 1-4 nucleotides downstream. No S1 protection was detected with RNA purified from a strain that had not been induced with isopropyl ␤-D-thiogalactopyranoside (IPTG). Truncated and full-length transcripts were produced by in vitro transcription and analyzed by S1 nuclease protection. The HphI and Sau96I oligonucleotide standards were generated by annealing the 3Ј-labeled S1 probe to a complementary DNA oligonucleotide followed by digestion with the appropriate endonucleases.
No cleavages were detected in the codon for glutamine 167, which would have produced an S1 protection pattern similar to that observed with the truncated control in vitro transcript (Fig.  1C, truncated lane). As suggested by the Northern analysis described above, mRNA cleavage occurred ϳ13 to 19 nucleotides downstream of the predicted A-site codon during SecM ribosome arrest.
3Ј 3 5Ј Exonucleases Generate Truncated secM mRNA during Ribosome Arrest-Two models account for ribosome arrestdependent cleavage at the secM stop codon: (i) A-site cleavage due to inefficient translation termination as originally described in Refs. 2 and 3, or (ii) exonucleolytic trimming of downstream mRNA to the 3Ј margin of the arrested ribosome. To differentiate between these possibilities, we fused secM codons 150 -166 in-frame between two thioredoxin genes (trxA). The encoded FLAG-TrxA-SecMЈ-TrxA fusion protein contained the minimal SecM peptide motif ( 150 FST-PVWISQAQGIRAGP 166 ) sufficient for ribosome arrest (9). However, in contrast to the wild-type secM gene, the flag-trxA-secMЈ-trxA stop codon is positioned several hundred nucleotides downstream of the predicted ribosome arrest site (21).
Northern analysis of flag-trxA-secMЈ-trxA mRNA also showed ribosome arrest-dependent truncated messages (data not shown), and S1 nuclease protection analysis detected two prominent cleavage sites at 13 and 19 nucleotides downstream of the proline 166 codon (wild-type SecM numbering) (Fig. 2B,  wild-type lane). The cleavages were the same distance from the proline 166 codon as was observed with flag-secMAЈ and flag-(⌬ss)secMAЈ mRNAs (Figs. 1C and  2A). Although the cleavage patterns were not strictly identical between truncated messages, the secM stop codon was clearly not required for mRNA cleavage.
We reasoned that if ribosome arrest-dependent mRNA cleavage was due to exonuclease activity, then cleavage could be modulated by deletion of known 3Ј 3 5Ј exoribonucleases. Fig. 2B shows the effects of specific exoribonuclease deletions on mRNA cleavage using the flag-trxA-secMЈ-trxA message. Deletion of RNase R leads to an increase in the ϩ19 cleavage product and a decrease in the ϩ13 cleavage product compared with wild-type (Fig. 2B). Similarly, removing polynucleotide phosphorylase (PNPase) activity also lead to increased levels of the ϩ19 product (Fig. 2B). In contrast, there was a slight decrease in the ϩ19 product in ⌬RNase II cells (Fig. 2B). The RNase R/RNase II double deletion strain exhibited less cleavage at both sites, whereas deletion of the C-terminal domain of RNase E had little effect on cleavage (Fig. 2B). Although RNase E is an endoribonuclease, the C-terminal domain is required for the organization of the degradosome, a multienzyme complex that contains PNPase and is important for the degradation of many mRNAs in E. coli (25,26). In general, the accumulation of specific cleavage products was dependent upon exoribonuclease activities.
The SecM Nascent Peptide Is Linked to tRNA Gly during Ribosome Arrest-The accumulation of truncated secM messages in tmRNA ϩ cells and the involvement of exoribonucleases in mRNA cleavage are inconsistent with what is known about A-site cleavage. Moreover, the SecM-induced ribosome arrest occurs at the codon for proline 166 (9, 16), a position that is 13-15 nucleotides upstream of the stop codon (Fig. 1A). We sought to confirm the position of SecM-stalled ribosomes using a mini-gene that encodes SecM residues glutamine 149glutamine 167 directly downstream of the FLAG epitope. Additionally, the flag-secMЈ mini-gene was synonymously recoded to change the codon for proline 153 from CCC to CCG, and the codon for glycine 161 from GGC to GGA.
Northern analysis using a probe specific for the ribosome binding site of flag-secMЈ detected truncated mRNA, and this cleavage appeared to depend upon ribosome stalling because truncated mRNA was not observed with the P166A variant (Fig.  3, RBS probe blot). The position of the arrested ribosome was determined by identifying the nascent peptidyl-tRNA by Northern blot analysis. Induction of FLAG-SecMЈ synthesis led  Fig. 1C. Numerical position is reported with respect to the codon corresponding to SecM proline 166, where position ϩ1 is the first nucleotide of the codon corresponding to SecM glutamine 167. Downward arrows labeled DdeI, EcoRI, and Sau96I indicate mRNA cleavage sites corresponding to the migration positions of S1 oligonucleotide probe standards. B, S1 nuclease protection analysis of flag-trxA-secMЈ-trxA mRNA purified from cells lacking 3Ј 3 5Ј exoribonucleases. Gene deletions and disruptions were constructed as described under "Experimental Procedures." Positions ϩ13 and ϩ19 downstream of the codon corresponding to SecM proline 166 are indicated. The DdeI, EcoRI, and Sau96I oligonucleotide standards were generated by annealing the labeled S1 probe to a complementary DNA oligonucleotide followed by digestion with the appropriate endonucleases. IPTG, isopropyl ␤-D-thiogalactopyranoside.
to a shift in the electrophoretic mobility of tRNA 3 Gly but not that of tRNA 2 Pro (Fig. 3, glyV and proL probe blots). The tRNA 3 Gly mobility shift was not observed when the FLAG-(P166A)SecMЈ variant was expressed (Fig. 3, glyV probe blot). The tRNA 3 Gly mobility shift was not seen when RNA samples were incubated at pH 8.9 for 1 h at 37°C to deacylate tRNAs (data not shown) (22). The arrested ribosome could be positioned unambiguously because the recoded mini-gene contained only one codon (GGC of glycine 165) that is decoded by tRNA 3 Gly . Therefore, during SecM-mediated ribosome arrest, the nascent peptide is covalently linked to tRNA 3 Gly via glycine 165 and the codon for proline 166 is positioned in the A-site.
SecM-arrested Ribosomes Contain Prolyl-tRNA Pro in the A-site-Elegant studies have shown that SecM ribosome arrest is prevented if proline residues are replaced with the imino acid analog, azetidine-2-carboxylic acid (14). Based on this finding, it has been reasonably assumed that proline 166 is incorporated into the SecM nascent peptide during ribosome arrest (8,9,14). However, recent work from Ito and colleagues (16), as well as our analysis, indicates that ribosome arrest occurs prior to proline 166 addition. One model that is consistent with all available data postulates that prolyl-tRNA Pro occupies the A-site of the SecM-arrested ribosome. If this model is correct, tRNA Pro should be stably associated with arrested ribosomes.
Extracts from cells expressing FLAG-(⌬ss)SecM were separated into high-speed pellet and supernatant fractions by ultracentrifugation through sucrose cushions. Polyacrylamide gel analysis of RNA extracted from these fractions showed that the rRNA (i.e. ribosomes) was present in the pellet fraction, whereas the majority of tRNA was in the supernatant fraction (data not shown). Partitioning of tRNA to the supernatant fraction was confirmed by Northern analysis for tRNA 2 Arg (Fig. 4, argQ probe blot), which was not predicted to associate with SecM-arrested ribosomes. In contrast, a higher proportion of tRNA 2 Pro was found in the pellet fractions from cells expressing FLAG-(⌬ss)SecM, but not FLAG-(⌬ss-P166A)SecM (Fig. 4, proL probe blot). Enrichment of tRNA Pro in pellet fractions was dependent upon cognate tRNA/codon interactions. tRNA 2 Pro , the cognate tRNA for CCU and CCC codons, was not enriched in high-speed pellets if SecM proline 166 was encoded by CCG (Fig. 4, proL probe blot), even though the CCG codon fully supports ribosome arrest (9). Moreover, although tRNA 1 Pro partitioned to the pellet fractions when the CCU construct was expressed, significantly more tRNA 1 Pro was found in the pellet fraction when its cognate CCG codon was used to code for proline 166 (Fig. 4, proK probe blot). The partitioning of tRNA 1 Pro to the ribosome fraction with the CCU construct may be due to association with trailing ribosomes within the SecM-stalled polysome, because tRNA 1 Pro is not known to decode CCU and is found in the high-speed supernatant in the absence of ribosome arrest (Fig. 4, (⌬ss)P166A lanes). Finally, the association of tRNA Pro with pellet fractions was not inhibited by the tmRNA quality control system (Fig. 4, ⌬tmRNA versus tmRNA ϩ ).
tmRNA Activity at SecM-arrested Ribosomes-The data presented thus far indicate that tmRNA does not play a significant role in rescuing SecM-arrested ribosomes. However, published reports show SecM and SecM variants are ssrA-tagged by tmRNA as a consequence of ribosome arrest (27,28). We examined tmRNA-mediated peptide tagging of SecM proteins in cells that express tmRNA(His 6 ), which encodes a hexahistidine-containing ssrA peptide that is resistant to proteolysis (18). Western blot analysis using antibodies specific for His 6 detected two ssrA(His 6 )-tagged species of (⌬ss)SecM (Fig. 5A, (⌬ss)SecM His6 lane). A similar ssrA(His 6 )-tagged doublet was observed with signal sequence-containing FLAG-SecM (data not shown), but not with the FLAG-(⌬ss-P166A)SecM protein, which does not cause ribosome arrest (Fig. 5A, (⌬ss)P166A). All ssrA(His 6 )-tagged species were also detected by Western analysis using antibody specific for the N-terminal FLAG epitope (Fig. 5A, anti-FLAG panel). Gly . RNA from strains expressing flag-secMЈ (pSecMЈ) and flag-(P166A)secMЈ (pSecMЈ(P166A)) mini-genes was analyzed by Northern blot to identify ribosome arrest-dependent peptidyl-tRNA. The RBS, glyV, and proL oligonucleotide probes were specific for the ribosome binding site of mRNA, tRNA 3 Gly , and tRNA 2 Pro , respectively. The migration positions of full-length mRNA, truncated mRNA, tRNA 3 Gly , peptidyl-tRNA 3 Gly , and tRNA 2 Pro are indicated. Samples containing overexpressed tRNA 2 Pro are indicated by 1[tRNA 2 Pro ]. The slower migrating species detected in the proL probe blot is probably incompletely processed tRNA 2 Pro , as it is present in all overexpressed tRNA 2 Pro samples, regardless of SecM expression. IPTG, isopropyl ␤-D-thiogalactopyranoside. To determine the sites of tagging, we purified ssrA(His 6 )-tagged FLAG-SecM and FLAG-(⌬ss)SecM by Ni 2ϩ -NTA affinity chromatography and subjected the purified proteins to mass spectrometry and N-terminal sequence analysis. Although FLAG-SecM was initially expressed as an N-terminal FLAG fusion, the N-terminal amino acid sequence (AEPNA) of the purified protein indicated that the epitope tag had been removed along with the signal sequence peptide during secretion (data not shown). The masses of tagged SecM species were consistent with the addition of ssrA-(His 6 ) tags after glycine 165 ( 5B, (⌬ss)SecM spectrum). We suspected that the tagged proteins detected by Western blot analysis corresponded to the two species observed by mass spectrometry. These assignments were confirmed through analysis of the FLAG-(⌬ss-Q167UAA)SecM protein, which was synthesized from a construct containing a mutation that changes glutamine 167 codon to a stop codon (UAA) (Fig. 1A). The FLAG-(⌬ss-Q167UAA)SecM protein lacks four C-terminal amino acid residues, but still causes ribosome arrest (9). FLAG-(⌬ss-Q167UAA)SecM protein was tagged after glycine 165, but not after threonine 170 (Fig. 5A, (⌬ss)Q167UAA, and data not shown). Presumably, the premature stop codon prevented ribosomes from translating to the 3Ј end of truncated mRNA.
The effect of tmRNA activity on total SecM protein production was examined by Western blot analysis using a monoclonal antibody specific for the N-terminal FLAG epitope present on all FLAG-(⌬ss)SecM variants. Two species of FLAG-(⌬ss)SecM accumulated in ⌬tmRNA   (Fig. 5A, anti-FLAG panel, ⌬tmRNA lane). The higher molecular weight protein represented full-length polypeptide and this species co-migrated with FLAG-(⌬ss-P166A)SecM (which does not cause ribosome arrest) on SDS-polyacrylamide gels (Fig. 5A,  anti-FLAG panel). The lower molecular weight species seen in ⌬tmRNA cells corresponded to incompletely synthesized FLAG-(⌬ss)SecM protein (to residue glycine 165) produced during ribosome arrest (Fig. 5A, and data not shown). However, analyses of cetyl trimethylammonium bromide precipitates and isolated ribosomes indicated that most of the incompletely synthesized FLAG-(⌬ss)SecM protein was not covalently linked to tRNA and therefore did not represent ribosome-bound nascent chains (data not shown). Therefore, incompletely synthesized FLAG-(⌬ss)SecM polypeptide chains were released from the arrested ribosome in a tmRNA-independent manner. In contrast to ⌬tmRNA cells, fulllength FLAG-(⌬ss)SecM protein was not detected in tmRNA ϩ cells (Fig. 5A, anti-FLAG panel, tmRNA ϩ lane). Presumably, the full-length FLAG-(⌬ss)SecM protein was ssrA-tagged and degraded rapidly in wild-type cells. Similarly, full-length FLAG-(⌬ss)SecM did not accumulate to very high levels in tmRNA(His 6 )-expressing cells, although the two ssrA(His 6 )tagged species were readily detected (Fig. 5A, anti-FLAG panel).
Prolyl-tRNA Pro in the A-site Inhibits tmRNA Activity-SsrA tagging after glycine 165 appears to contradict the other data indicating that tmRNA plays no significant role in resolving SecM-arrested ribosomes. However, this work and previous studies relied upon SecM overexpression (27,28), which is predicted to deplete limiting tRNA Pro species. tRNA 3 Gly , which holds the SecM nascent chain during ribosome arrest, is found at ϳ4,400 molecules per E. coli cell, whereas tRNA 2 Pro and tRNA 3 Pro , which occupy the arrested ribosome A-site, are present at only ϳ1,300 copies per cell (29). Therefore, if the number of SecM-arrested ribosomes exceeds 1,300 per cell, a second population of stalled ribosomes with unoccupied A-sites will accumulate due to prolyl-tRNA Pro sequestration, potentially allowing for adventitious ssrA tagging after glycine 165.
To test this model, we overexpressed tRNA 2 Pro and examined the effects on ssrA peptide tagging and three other properties of SecM ribosome arrest: (i) nascent peptidyl-tRNA stability, (ii) cleavage of flag-secMЈ mRNA, and (iii) regulation of secA translation. Overexpression of tRNA 2 Pro significantly suppressed ssrA(His 6 ) tagging after glycine 165, but increased tagging after threonine 170 (Fig. 5A, (⌬ss)SecM-ptRNA 2 Pro lane). Although tmRNA activity was significantly altered, tRNA 2 Pro overexpression had no effect on nascent peptidyl-tRNA 3 Gly accumulation (Fig. 3, glyV probe blot), and actually appeared to increase flag-secMЈ mRNA cleavage (Fig. 3, RBS probe blot). Finally, tRNA 2 Pro overexpression had no effect on the regulation of secA translation. We made secAЈ::lacZ translational fusions and confirmed that deletion of the SecM signal sequence increased SecAЈ-LacZ expression, whereas further introduction of the P166A mutation reduced fusion protein synthesis (Fig. 6). Overexpression of tRNA 2 Pro had no significant effect on the ribosome arrestdependent increase in ␤-galactosidase activity (Fig. 6). Moreover, deletion of tmRNA had no effect on SecAЈ-LacZ expression, as determined by Western blot and ␤-galactosidase activity analyses (Fig. 6). We also attempted to examine the effects tRNA 1 Pro overexpression on ribosome arrest from con-structs that encoded proline 166 as CCG. Unfortunately, all plasmid clones carrying the proK gene under its own promoter also contained mutations in the tRNA 1 Pro -encoding sequence (data not shown). Seven distinct mutations were found mapping to the D-arm, T-arm, anticodon loop, and the promoter (data not shown). These results suggest that high-level overexpression of tRNA 1 Pro is deleterious to the cell.

DISCUSSION
Several lines of evidence indicate that the primary SecM-mediated ribosome arrest is resistant to A-site mRNA cleavage and subsequent tmRNA recruitment. First, although the secM mRNA was truncated in a ribosome arrest-dependent manner, the cleavage sites were 13 to 19 nucleotides downstream of the A-site codon. Second, the steady-state number of SecM-arrested ribosomes (as determined by Northern analysis of nascent peptidyl-tRNA) was not significantly affected by tmRNA. Third, incompletely synthesized SecM protein (to residue glycine 165) accumulated in tmRNA ϩ and tmRNA(His 6 )-expressing cells. Fourth, SecM-dependent regulation of secA translation was essentially identical in ⌬tmRNA and tmRNA ϩ cells (27). Finally, A-sitebound tRNA 2 Pro inhibits ssrA tagging after SecM glycine 165. The surprising discovery of A-site-bound prolyl-tRNA Pro has also been recently reported by Ito and colleagues (16). That study used entirely different methods than ours to characterize arrested ribosomes produced in vitro (16), and is completely congruent with our analysis of the in vivo SecM ribosome arrest. Altogether, our data strongly suggest that tmRNA recruitment during the primary ribosome arrest is an artifact of SecM overexpression, and that A-site mRNA cleavage and ssrA tagging at this site do not occur under physiological conditions. We feel this conclusion makes biological sense because A-site cleavage is predicted to interfere with cis-acting SecM regulation of secA translation initiation. Moreover, co-translational secretion of the SecM nascent peptide ensures that SecA is synthesized in close proximity to the inner membrane (30), a phenomenon that presumably requires synthesis of SecM and SecA from the same mRNA molecule. Deletion of the SecM signal sequence prevents co-translational secretion and thereby precludes the mechanism that normally alleviates ribosome arrest (14). Secreted SecM also elicited ribosome arrest in our study, presumably because the overexpressed protein saturated the secretion machinery. The (⌬ss)SecM-mediated ribosome arrest exhibits a t1 ⁄ 2 Ͼ 4 -5 min in vivo (14), which exceed the half-life of bulk E. coli mRNA turnover (ϳ2.4 min at 37°C) (31). Thus, prolonged translational pausing allows degradation of the downstream mRNA to the 3Ј edge of the arrested ribosome (Fig. 7). Presumably, the 5Ј portion of the message is protected by ribosomes queued behind the primary SecM-arrested ribosome. The influence of SecM ribosome arrest on mRNA degradation appears to differ from other reported ribosome pauses, which tend to stabilize mRNA downstream of the arrest site (32)(33)(34). Although specific endonuclease cleavage between cistrons has been observed in E. coli operons (35), we find the same cleavages in mRNAs that lack the secM-secA intergenic region. Moreover, the cleavage appears to require ribosome arrest, arguing against a sequence-specific endonuclease activity. Our observations suggest that 3Ј 3 5Ј exoribonucleases generate the 3Ј termini of truncated secM messages. First, cleavage occurred downstream of the A-site codon, at sites consistent with the 3Ј border of a stalled E. coli ribosome (36,37). Second, the proportion of ϩ13 and ϩ19 cleavage products was dependent upon exoribonuclease activities present in the cell. Longer cleavage products accumulated in the absence of either RNase R or PNPase, both of which degrade secondary structure-containing RNAs more efficiently than RNase II (38,39). Although RNase R and PNPase are not known to work together, our data suggests that these enzymes may cooperate to convert the ϩ19 cleavage product to the ϩ13 product. Finally, RNase II can indirectly inhibit the degradation of structured mRNAs by removing 3Ј single-stranded regions required by PNPase to bind substrate (39,40). These biochemical properties are consistent with the accumulation of cleavage products in our exoribonuclease knock-out strains.
The details of mRNA cleavage notwithstanding, it is interest- ing that the secM stop codon is in position to be cleaved during prolonged ribosome arrest. SecM-arrested ribosomes clearly resumed translation, and upon reaching the 3Ј end of the truncated mRNA, they stalled for a second time (Fig. 7). However, tmRNA is readily recruited to ribosomes stalled at the extreme 3Ј termini of mRNAs, and SecM was ssrA-tagged after the C-terminal residue threonine 170 (Fig. 7, ribosome fate II).
Because little full-length (⌬ss)SecM protein accumulated in tmRNA ϩ or tmRNA(His 6 ) cells, it appears that degradation of mRNA to the ribosome edge preceded the resumption of translation. It is unclear whether exoribonuclease cleavage also leads to ssrA-dependent degradation of SecM under physiological conditions. Secreted SecM was shown to be rapidly degraded in tmRNA ϩ cells (14), and we find the non-degradable ssrA(His 6 ) tag stabilizes SecM in the periplasm. However, both of these studies employed SecM overexpression. At lower expression levels, co-translational secretion of SecM is expected to prevent ribosome arrest, and thereby inhibit mRNA cleavage and subsequent tmRNA recruitment/ssrA-tagging (Fig. 7, ribosome fate I). In any event, prolonged ribosome arrest stimulates SecA expression, so significant protein synthesis must occur prior to degradation of the downstream secA cistron. A-site mRNA cleavage and tmRNA activities were clearly not able to resolve the majority of primary SecM-arrested ribosomes. However, ssrA tagging after glycine 165 indicates limited tmRNA recruitment during primary ribosome arrest, at least when SecM is overexpressed. Ivanova et al. (41) showed tmRNA is recruited to ribosomes stalled on mRNAs where the 3Ј terminus is 12 nucleotides downstream of the A-site codon, albeit at a ϳ20-fold lower rate than maximum. Therefore, cleavage of mRNA to the 3Ј-edge of the arrested ribosome could allow relatively inefficient tmRNA recruitment, provided the A-site is not occupied with prolyl-tRNA Pro (Fig. 7, ribosome fate III). Alternatively, limited A-site mRNA cleavage may have occurred under SecM overexpression conditions (Fig. 7, ribosome fate IV). It appears that A-site nuclease activity is restricted to codons within unoccupied A-sites (1, 2), so presumably A-site cleavage in this instance would be an artifact of SecM overexpression. Based on Northern blot analysis, ϳ60% of cellular tRNA 3 Gly is sequestered as SecM nascent peptidyl-tRNA during SecM overexpression. This corresponds to roughly 2,600 SecM-arrested ribosomes per cell, of which only ϳ1,300 can simultaneously contain A-site prolyl-tRNA Pro (29). Therefore, we estimate ϳ50% of the SecM-arrested ribosomes have unoccupied A-sites in the absence of compensatory tRNA 2 Pro overexpression, in accord with recent studies of SecM-arrested ribosomes (15,16). Given incomplete A-site occupancy, perhaps the lack of A-site mRNA cleavage reflects sequence specificity of the A-site nuclease. The RelE protein shows marked preference for A-site codons, cleaving CAG and UAG at the highest rate (1). However, we have observed RelEindependent A-site mRNA cleavage at several different codons, suggesting that many sequences are potential substrates (2). 4 Alternatively, the low rate of A-site mRNA cleavage may be due to the substantial structural rearrangements that occur in the ribosome during SecM-mediated arrest (15). SecM-induced structural rearrangements originate in the 50 S exit channel and are propagated to the 30 S subunit via inter-subunit bridges and ribosome-bound tRNAs (9,15). Although structural changes are transmitted by tRNA, SecM-arrested ribosomes adopt the same conformation independent of A-site bound prolyl-tRNA Pro (15). Several elements of 16 S rRNA are rearranged during ribosome arrest, including helix 44, which forms part of the 30 S A-site and makes contact with A-site mRNA (15). Clearly, alteration of A-site structure could significantly affect A-site nuclease activity, whether catalyzed by the ribosome or a trans-acting factor. Interestingly, Aiba and colleagues (28) showed that expression of a fusion protein containing the SecM-derived residues 161 GIRAGP 166 resulted in significant mRNA cleavage at sites immediately adjacent to the A-site codon, although the majority of cleavages still occurred at other positions corresponding to the 3Ј-and 5Ј-edges of the paused ribosome. We also observed similar cleavages near the A-site codon when expressing a fusion protein containing the longer SecM-derived 149 QFSTPVWISQAQGIRAGP 166 sequence (Fig. 2B), but failed to detect these mRNA cleavages when expressing full-length SecM sequences. Perhaps the fulllength SecM nascent peptide is required for complete structural rearrangement and inhibition of A-site mRNA cleavage.
Gene regulation by translational pausing has long been recognized in prokaryotes, although its importance is still often underestimated. Indeed, the SecM ribosome arrest is a newly characterized example of translational attenuation, which was shown to control inducible expression of erythromycin and chloramphenicol resistance genes over 20 years ago (42)(43)(44)(45). The role of translational pausing in transcriptional attenuation of the E. coli trp operon was recognized even earlier (46,47). In each case, A-site mRNA cleavage and tmRNA activities have the potential to interfere with regulation by "rescuing" paused ribosomes. However, in our view, it makes little sense to employ regulatory strategies that are undermined by translational quality control systems, and we predict that regulatory ribosome pauses are generally immune to A-site cleavage and tmRNA activities. The mechanisms involved are likely varied, and characterization of other ribosome pauses will hopefully increase our understanding of the molecular requirements for A-site mRNA cleavage.