SsrA tagging of Escherichia coli SecM at its translation arrest sequence.

SecM is expressed from the secM-secA operon and activates the expression of secA in response to secretion defects. The 3'-end of secM encodes an "arrest sequence," which can interact with the ribosomal exit tunnel, preventing complete secM translation under secretion-defective conditions. In a cis-acting manner, ribosome stalling enhances secA translation. Pro(166) is the last residue incorporated when SecM elongation is arrested. We report that secretion deficiencies lead to SsrA tagging of SecM after Pro(166), Gly(165), and likely Arg(163). Northern blot analysis revealed the presence of a truncated secM transcript, likely issued from a secM-secA cleavage. The level of secM transcripts was decreased either when secM translation was totally prevented or when Pro(166) was mutated. However, the accumulation of a truncated secM transcript required secM translation and was prevented when the SecM arrest sequence was inactivated by a point mutation changing Pro(166) to Ala. We suggest that ribosome pausing at the site encoding the arrest sequence is required for formation of the truncated secM mRNA. SsrA tagging affected neither the presence of the secM mRNA nor secA expression, even under translocation-defective conditions. It is therefore likely that SsrA tagging of SecM occurs only after cleavage of secM-secA mRNA within the secM open reading frame encoding the SecM arrest sequence. Accumulation of transcripts expressing arrested SecM generated growth inhibition that was alleviated by the SsrA tagging system. Therefore, SsrA tagging of SecM would rescue ribosomes to avoid excessive jamming of the translation apparatus on stop-less secM mRNA.

SecM is expressed from the secM-secA operon and activates the expression of secA in response to secretion defects. The 3-end of secM encodes an "arrest sequence," which can interact with the ribosomal exit tunnel, preventing complete secM translation under secretion-defective conditions. In a cis-acting manner, ribosome stalling enhances secA translation. Pro 166 is the last residue incorporated when SecM elongation is arrested. We report that secretion deficiencies lead to SsrA tagging of SecM after Pro 166 , Gly 165 , and likely Arg 163 . Northern blot analysis revealed the presence of a truncated secM transcript, likely issued from a secM-secA cleavage. The level of secM transcripts was decreased either when secM translation was totally prevented or when Pro 166 was mutated. However, the accumulation of a truncated secM transcript required secM translation and was prevented when the SecM arrest sequence was inactivated by a point mutation changing Pro 166 to Ala. We suggest that ribosome pausing at the site encoding the arrest sequence is required for formation of the truncated secM mRNA. SsrA tagging affected neither the presence of the secM mRNA nor secA expression, even under translocation-defective conditions. It is therefore likely that SsrA tagging of SecM occurs only after cleavage of secM-secA mRNA within the secM open reading frame encoding the SecM arrest sequence. Accumulation of transcripts expressing arrested SecM generated growth inhibition that was alleviated by the SsrA tagging system. Therefore, SsrA tagging of SecM would rescue ribosomes to avoid excessive jamming of the translation apparatus on stop-less secM mRNA. Nascent proteins emerge from the 50 S ribosomal subunit through an exit tunnel with a length of ϳ100 Å and containing 35-40 amino acids of the elongating polypeptide. Its narrowness prevents peptide folding, and certain sequences from nascent peptides can interact with ribosomal elements of the tunnel to affect protein biosynthesis. The ribosome is therefore sensitive to the composition of the amino acid chain it is syn-thesizing, and the exit tunnel contributes to the dynamics of protein elongation (1). In Escherichia coli, this is exemplified by a sequence of the periplasmic protein SecM (secretion monitor) at a position close to its C terminus called the arrest sequence, which interacts with the ribosomal exit tunnel and causes elongation arrest within the ribosome (Refs. 2 and 3; for review, see Refs. 4 and 5). Ribosomal arrest is a key element in the regulation of SecA, an ATPase that targets protein precursors to the SecYEG core translocon for secretion. SecA is expressed from the secM-secA-mutT operon; its translation is up-regulated by a block in SecM secretion in a cis-specific manner (2,6). A secondary structure encompassing the secA Shine-Dalgarno sequence of the secM-secA-mutT RNA can conditionally inhibit secA translation. Due to its arrest sequence, secM translation stalls when its N-terminal signal sequence is not "pulled" by the protein export machinery (7,8). Paused ribosomes are thought to mediate unfolding of the RNA secondary structure, thereby exposing the initiation signal for SecA biosynthesis (9). By this mechanism, the amount of SecA is adjusted to the intracellular demand for protein export (6).
Pausing of translating ribosomes can lead to protein tagging by the transfer-messenger RNA (tmRNA 1 ; also called SsrA in E. coli) (for review, see Ref. 10). When charged with an alanine, tmRNA is recruited to A-sites of stalled ribosomes, and growing polypeptide chains are transferred to alanyl-tmRNA by a transpeptidation reaction. tmRNA then becomes a template for translation of an oligopeptide tag, which in E. coli corresponds to ANDENYALAA (11). The change of matrix from mRNA to tmRNA is referred to as trans-translation. This mechanism allows release of stalled ribosomes (11) and elimination of tagged products by specific proteases (12)(13)(14)(15). Ribosome stalling and trans-translation occur during translation of truncated mRNAs lacking a stop codon because release factors are not recruited (11, 16 -18). Such truncated messengers could result from mRNA damage, premature transcription termination, or degradation or cleavage by ribonucleases. The bacterial toxins RelE, MazF, and ChpB were recently shown to cleave mRNAs and induce to SsrA tagging (19 -22). Cognate tRNA scarcity (23) and inefficient stop codons and/or specific sequences (24 -28) generate ribosome pausing, which can also lead to SsrA tagging.
Here, we show that ribosomes stalled on the secM transcript are rescued by the SsrA tagging system. We show the presence of a truncated secM mRNA transcript likely issued from endonucleolytic cleavage of a secM-secA mRNA. It requires translation of secM mRNA and the integrity of the SecM arrest sequence. SsrA tagging does not influence secA expression but prevents the toxicity of accumulated non-exported SecM.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Growth Conditions, and Enzyme Assay-The E. coli strains used in this work are presented in Table I. P1 virmediated transduction was carried out as described (29). Cells were grown at the temperatures indicated in Terrific broth, LB broth, or M63 broth supplemented with 0.4% glucose. Solid medium contained 15 mg/ml agar (29). Antibiotics were added as necessary at the following concentrations: ampicillin, 100 g/ml; chloramphenicol, 10 g/ml; and kanamycin, 20 g/ml. Expression from the P LtetO-1 promoter was obtained by addition of 1 M anhydrotetracycline. Expression of cloned inserts under the control of the T7 promoter in pET-3b derivatives (in BL21(DE3) pLys derivatives) was achieved by addition of isopropyl ␤-D-thiogalactopyranoside (1 mM) to the growth medium. Sodium azide was added where indicated to a final concentration of 0.02%. Bacterial growth was monitored at A 600 . ␤-Galactosidase activity was determined as described (29).
Plasmids and Molecular Biology Techniques-Plasmid preparation, DNA cloning and ligation, PCR amplification, and DNA transformation were carried out according to standard protocols (30) or manufacturers' instructions. DNA sequencing was performed by Eurogentec (Angers, France) or at the CNRS genomic platform (Gif-sur-Yvette, France). Oligonucleotides used for PCR amplification and for priming standard sequencing reactions were purchased from Sigma). E. coli strain C600 was the source of chromosomal DNA for PCR amplifications. Short descriptions of the plasmids used in this work are presented in Table II.
The pssM2SecM plasmid expresses a fusion protein with internal tandem FLAG epitopes between Val 57 and Asn 58 of the SecM protein (with its signal sequence). It was obtained by cloning the linker resulting from the nucleotide annealing of 5Ј-GCTAGCTTTA TCATCATCAT CTTTATAATC TTTATCATCA TCATCTTTAT AATCCAT-3Ј and 5Ј-AT-GGATTATA AAGATGATGA TGATAAAGAT TATAAAGATG ATGAT-GATAA AGCTAGC-3Ј into the HpaI site of pSecM.
ssrA/ssrAH7 Gene Replacement-The ssrA gene encoding the SsrA tag ANDENYALAA was previously modified to encode the peptide tag ANDENYHHHHHHH (ssrAH7, expressed in pSsrA-H 7 ) (25). The chromosomal ssrA gene was replaced with the modified ssrAH7 gene at its own locus using pKO3SsrA-H 7 introduced into strain PhB1805 (MG1655 ssrA::Km) as described (33). The resulting ssrAH7 strain supported hybrid immP22 growth with the same efficiency as the wild-type strain, whereas the ssrA::Km strain showed a 10 4 -fold reduction in plaque formation (34,35).
Northern Blot Analyses-Overnight bacterial cultures grown in LB medium at 37°C were diluted 1000-fold into fresh LB medium. Isopropyl ␤-D-thiogalactopyranoside was added to the culture medium at A 600 ϭ 0.15. At A 600 ϭ 0.3, the culture (10 ml) was mixed with an equal volume of cold (Ϫ80°C) absolute ethanol and immediately spun down. The pellet was suspended in 300 l of lysis buffer (0.02 M sodium acetate (pH 5.5), 0.5% (w/v) SDS, and 1 mM EDTA), and 600 l of hot acidic phenol was added. Phenol extraction was repeated three times, and RNA was precipitated with 0.1 volume of 3 M sodium acetate and 2 volumes of absolute ethanol. RNA samples were separated by electrophoresis on agarose gel containing formaldehyde and capillary-transferred onto nylon membrane (Hybond-XL, Amersham Biosciences, Buckinghamshire, England) according to published protocols (30) and the manufacturer's instructions. The secM (5Ј-GCTGAATTTT GCTT-GTGGGG TAAAATGCGC A-3Ј), intergenic (5Ј-GCGTTGCGAG TTA-ATAAAAT GAAGTAAAGG TTTATTGTTG-3Ј), and secA (5Ј-TAC-CGAAAAC TTTAGTTAAC AATTTGATTA GCATA-3Ј) probes were 5Ј-end-labeled using T4 polynucleotide kinase (New England Biolabs Inc.) according to the manufacturer's instructions with [␥-32 P]ATP (Amersham Biosciences) and purified using the QIAquick nucleotide removal kit (QIAGEN Inc., Courtaboeuf, France). Prehybridization (3 h at 52°C) and hybridization (20 h at 52°C) of the membrane with labeled oligonucleotides were performed in SSC/Denhardt's buffer as described (30). Radioactivity was detected on storage phosphor screens using a PhosphorImager (Amersham Biosciences).
Purification of SsrA-H 7 -tagged Proteins-Cultures of 0.75 liters were grown overnight in Terrific broth at 37°C. After centrifugation, cell pellets were suspended in lysis buffer A (6 M guanidine, 50 mM NaH 2 PO 4 , 10 mM Tris-HCl, and 100 mM NaCl (pH 8)) at 5 ml/g of cells and lysed by stirring for 1 h at room temperature. Cell debris was removed by centrifugation for 15 min at 10,000 ϫ g. The supernatant was mixed with 1 ml of a nickel-nitrilotriacetic acid slurry (QIAGEN Inc.) equilibrated in lysis buffer A and allowed to bind by rocking for 1 h at room temperature. The resin was collected by centrifugation and washed (incubation for 10 min followed by 7 min of centrifugation at 700 ϫ g twice with 15 ml of lysis buffer A) and three times with lysis buffer B (8 M urea, 20 mM Tris-HCl, 100 mM NaCl, and 10 mM imidazole (pH 8)). An additional wash with 1 ml of lysis buffer B was performed. Proteins were eluted with 4 ϫ 1 ml of elution buffer (lysis buffer B ϩ 100 mM imidazole (pH 8)). SsrA-H 7 -tagged proteins were usually recovered in the first 1-ml fraction. SDS-15% PAGE and Coomassie Blue staining were performed according to manual (36) and manufacturers' protocols.
Identification of SsrA-H 7 -tagged ⌬ssSecM by Mass Spectrometry-Tryptic peptides from SsrA-H 7 -tagged ⌬ssSecM were identified by matrix-assisted laser desorption ionization time-of-flight mass spectrometry according to a described protocol (25). Strain PhB2431 (with a slyD mutation to avoid sample contamination by the histidine-rich peptidyl cis,trans-isomerase SlyD) containing p⌬ssSecM was used for identification of SsrA-H 7 -tagged ⌬ssSecM. The percentage of protein length covered by peptide matches was ϳ40% for SsrA-H 7 -tagged ⌬ssSecM.
Cetyltrimethylammonium Bromide (CTABr) Fractionation for Identification of Peptidyl-tRNA Proteins-Cultures were grown overnight at 37°C and normalized to the same absorbance. Samples were treated with 5% trichloroacetic acid, and denatured proteins were collected, washed, and dissolved in SDS buffer (37). RNase A treatment was performed where indicated using 100 g/ml RNase A for 10 min at 37°C. CTABr precipitation was then performed as described (7). Precipitates were washed twice with 80% acetone and dissolved in electrophoresis sample buffer (7) before SDS-PAGE and Western blotting. Supernatants were treated with 5% trichloroacetic acid to recover protein contents and analyzed similarly.
Pulse-Chase Analysis-Cells were grown at 37°C in minimal M63 medium supplemented with 16 amino acids (20 g/ml each, except Met, Cys, Ala, and Gly), 0.4% glucose, chloramphenicol, and anhydrotetracycline to induce ssM2SecM expression. Exponential phase cells were pulse-labeled with [ 35 S]methionine for 1 min. The chase was initiated by addition of unlabeled L-methionine (200 g/ml). At each time point indicated, samples were treated with 5% trichloroacetic acid, washed twice with 80% acetone, and dissolved in SDS buffer (37). Samples were processed for immunoprecipitation as described (38) except that protein A-agarose beads (Roche, Mannheim, Germany) were used with anti-FLAG polyclonal antibody M2. Immunoprecipitates were subjected to SDS-PAGE, and labeled proteins were visualized on storage phosphor screens.

RESULTS
Elongation-arrested SecM Is SsrA-tagged-Elongation of secM translation is arrested when SecM is not translocated to the periplasm (7,39). Since SsrA tagging of proteins can be induced by ribosome stalling, we hypothesized that elongationarrested SecM protein is subject to SsrA tagging. We constructed SecM derivatives containing a tandem FLAG epitope to allow immunodetection or immunoprecipitation of SecM. Accumulation of elongation-arrested SecM was achieved by preventing SecM export either (i) by using a signal sequence-less SecM derivative (⌬ssM2SecM encoded by p⌬ssM2SecM) (Fig. 1A) or (ii) by impairing secretion via addition of sodium azide (7,40) to the growth medium (Fig. 1B). For this approach, we used a full-length SecM protein containing an internal fusion of a tandem FLAG epitope named ssM2SecM and encoded by pssM2secM (Fig. 1B).
Expression of cytoplasmic ⌬ssM2SecM gave rise to a protein of ϳ17 kDa, detected by Western blotting using anti-FLAG antibody M2 (Fig. 1A, lane 2). Bands often appeared as closely space doublets (see "Discussion"). SecM elongation-arrested peptidyl-tRNA was shown previously to transiently accumulate in wild-type cells, whereas the complete SecM protein is unstable (7), suggesting that the protein we detected (Fig. 1A, lane 2) corresponds to elongation-arrested ⌬ssM2SecM. CTABr is used to precipitate peptidyl-tRNA molecules because of their RNA moiety (41). CTABr precipitation of total proteins from strains expressing ⌬ssM2SecM led to the detection of a protein cross-reacting with antibody M2 (Fig. 1C, lanes 2 and 5). This band disappeared if the samples were treated by RNase prior to CTABr precipitation (Fig. 1C, lanes 3 and 6). These experi-ments demonstrate that the only detectable form of ⌬ssM2SecM in the wild-type strain is an elongation-arrested peptidyl-tRNA (Fig. 1C, compared lanes 4 -6).
In a wild-type strain expressing periplasmic ssM2SecM, no translation product was detected by Western blotting using anti-FLAG antibody M2 (Fig. 1B, lane 2). It is likely that ssM2SecM is efficiently synthesized and rapidly translocated to the periplasm, where the resulting mature M2SecM protein is rapidly degraded by Tsp protease, as was reported for SecM (7). To confirm that ssM2SecM behaves like native SecM, it was analyzed by immunoprecipitation using M2 antiserum after Tsp protease (Fig. 1D). Two bands were detected: elongationarrested ssM2SecM (slow migrating) and mature M2SecM (fast migrating). When secretion was efficient, elongation-arrested ssM2SecM was very unstable, probably because it matured rapidly and was exported to the periplasm. When secretion was impaired by addition of sodium azide to the growth medium, elongation-arrested ssM2SecM was stabilized, confirming that ssM2SecM was produced. The elongation-arrested ssM2SecM derivative was also detected when secretion was impaired by Western blotting using anti-FLAG antibody M2 (Fig. 1B, lane  6). In the presence of modified SsrA-H 7 (stabilizing the tagged product), sodium azide addition to the growth medium resulted in the appearance of a second translation product of ϳ22 kDa (Fig. 1B, lane 8), corresponding to SsrA-H 7 -tagged ssM2SecM. Elongation-arrested ssM2SecM intensity was increased in the ⌬ssrA strain compared with the ssrA ϩ strain (Fig. 1B, lanes 6  and 7), suggesting that the ⌬ssrA mutation stabilizes elongation-arrested ssM2SecM.
Approximately half of the detected elongation-arrested M2SecM was tagged by the SsrA system (Fig. 1, A and B). We conclude that secM translation arrest allows its efficient SsrA tagging and subsequent degradation by ClpP-dependent proteases.
Trans-Translation of secM Occurs at Codons Expressing the Arrest Sequence-SsrA-H 7 allows affinity purification of SsrA-H 7 -tagged proteins (25). To identify the SsrA tagging sites of SecM, we purified SsrA-H 7 -tagged proteins under conditions maximizing SecM elongation arrest. The non-exported ⌬ssSecM protein was overproduced (p⌬ssSecM) in an ssrAH7 strain. SsrA-H 7 -tagged proteins were affinity-puri-fied and analyzed by gel electrophoresis. Two closely spaced bands at ϳ19 kDa were observed (Fig. 2A, lane 3) only in the ssrAH7 strain and in the presence of p⌬ssSecM. Proteins extracted from these bands were digested by trypsin, and the resulting peptides were analyzed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry. Both bands gave the same results. Several peptide masses matched the mass of the SecM protein primary sequence, confirming that the observed bands correspond to SsrA-H 7tagged ⌬ssSecM protein. Two peptides had masses corresponding to those expected for junction peptides containing SecM residues 164 -166 (AGP) or SecM residues 164 and 165 (AG) with the SsrA-H 7 tag AANDENYH 7 (Fig. 2B). A third peptide had a mass corresponding to that of just the SsrA-H 7 tag, AANDENYH 7 (Fig. 2B). Tryptic digestion cleaved after lysine or arginine, indicating that the SsrA tagging of SecM occurs after one of these amino acids. These results show that SsrA tagging occurs after Gly 165 and Pro 166 and likely after Arg 163 of nascent SecM since all of the SsrA-tagged products analyzed have about the same total molecular masses.
The ⌬ssrA Mutation Does Not Affect secA Expression-When SecM secretion is blocked, ribosomes pause on the secM coding sequence due to the SecM arrest sequence; this should cause unfolding of the secM-secA RNA secondary structure involved in secA repression, allowing secA translation initiation (9). Considering this model, SsrA tagging of elongation-arrested SecM could allow reformation of the inhibitory RNA secondary structure and thus reduce secA translation since the SsrA system can rescue stalled ribosomes. We therefore tested the effect of SsrA on SecA expression by means of a secA-lacZ translational fusion integrated at the secA locus (42). ␤-Galactosidase activities expressed from a secA-lacZ fusion in the ssrA ϩ or ⌬ssrA strain were measured under growth conditions without sodium azide; no difference was observed (Fig. 3). This result was expected since secretion then occurs efficiently, and secM translation is subject only to a transient pause. When secretion was impaired by treating cells with sodium azide, a condition under which arrested SecM was subject to strong SsrA tagging (Fig. 1), SecA expression was greatly increased. (␤-Galactosidase activity increased up to 8-fold after sodium azide was added to the medium.) However, induction was identical in the ssrA ϩ and ⌬ssrA strains (Fig. 3). The presence or absence of ClpP-dependent proteases also had no effect on induction of secA-lacZ fusion expression in response to sodium azide addition to the culture medium (data not shown). We conclude that SsrA tagging and the subsequent degradation of elongation-arrested SecM do not influence SecA expression, even under very defective translocation conditions in which SecM elongation arrest is prolonged and induces SsrA tagging of SecM.
Detection of secM and secM-secA mRNAs-It was recently demonstrated that ribosome stalling can cause mRNA cleavage at or next to arrest sites (43)(44)(45)(46)(47). We therefore considered that secM mRNA could be subject to such endoribonucleolytic cleavage. If so, it would result in a stop-less secM mRNA, which is the usual substrate for SsrA tagging. In that case, secM mRNA would be dissociated from secA-mutT RNA when tagging occurs, which is in keeping with the above result showing that tagging does not affect SecA expression.
To test this hypothesis, we performed Northern blot analyses on total RNAs from strains containing pET-3b derivatives carrying the secM gene and part of the secA gene. The pET-3b vector was chosen to obtain a high level of mRNA upon induction. We used several oligonucleotides probing for the secM, secM-secA intergenic, and secA RNA regions (Fig. 4A). With all plasmids and all probes, two large transcripts were detected (Fig. 4B): one of the size expected (ϳ1.6 kilonucleotides) for the full-length secM-secA mRNA transcribed from the cloned sequence and the other likely generated by a transcription termination read-through since it is longer (ϳ2 kilonucleotides).
With pETSecM (expressing wild-type SecM) or pET⌬ssSecM (expressing SecM devoid of its signal sequence), an additional shorter product (ϳ0.6 and ϳ0.5 kilonucleotides, respectively) was detected using the secM probe, but not the intergenic probe (Fig. 4B, lanes 1 and 3) or a probe hybridizing immediately after the SecM Pro 166 -encoding codon (data not shown). These results demonstrate that a truncated secM mRNA lacking the end of its coding sequence is produced together with the full-length secM-secA mRNA. No product of the expected size for secA mRNA only was detected using the intergenic or secA probe, suggesting that secA mRNA may be unstable if it is produced from a cleavage of secM-secA mRNA.
With pETSecM* (in which the secM initiation codon was mutagenized to a stop codon), we observed that the absence of secM translation drastically reduced the quantity of the fulllength secM-secA mRNA (Fig. 4B, lane 2), probably because the transcript is not protected by translating ribosomes. We also detected a shorter product (Ͼ0.6 kilonucleotides) that interacted as well with the secM probe as with the intergenic probe, but not with the secA probe. These results demonstrate that a transcript including the complete secM coding sequence and all or part of the intergenic sequence between secM and secA is produced together with the full-length secM-secA mRNA. This transcript may be issued from a premature transcription termination. Interestingly, a mutation affecting nusB was previously isolated as a suppressor of a temperature-sensitive secA mutation (48), suggesting that altering anti-termination can stimulate secretion. However, no shorter transcript interacting only with the secM probe but not with the intergenic or secA probe could be detected. This result shows that preventing translation of the secM coding sequence prevents formation of the truncated secM transcript.
A mutation changing SecM Pro 166 to Ala was shown previously to prevent translation arrest (2,3). pETSecM, pETSecM*, and pET⌬ssSecM derivatives in which the secM sequence corresponding to the Pro 166 -encoding codon was mutated to express Ala were constructed, yielding pETSecM-P166A, pET-SecM*-P166A, and pET⌬ssSecM-P166A, respectively. Upon Northern blot analysis using the same probes as described above, a similar transcript pattern as that obtained with the pETSecM* construct was observed at comparable quantities (Fig. 4B, lanes 5-7). This result shows that not only translation, but also the integrity of the SecM arrest sequence is needed for production of secM mRNA. The overall level of secM transcripts was decreased when the translation arrest-essential Pro 166 was mutated to Ala 166 , as was observed when secM translation was prevented (Fig. 4B). In both cases, secA translation was decreased, and as a result, the absence of translating ribosomes could leave the mRNAs more susceptible to degradation.
All of Northern blot analyses described above were performed in either an ssrA ϩ or ssrA Ϫ background. The overall Shown is the induction of secA-lacZ fusion expression in response to a secretion defect caused by sodium azide addition. PR478 (‚ and ϫ) and its ⌬ssrA derivative (PhB2469; छ and Ⅺ) were grown at 37°C in LB medium. Where indicated, sodium azide (0.02% final concentration) was added to cultures at A 600 ϳ 0.03, and the ␤-galactosidase activity (Miller units quantities of plasmid-encoded mRNAs were much lower in ssrA ϩ (data not shown) compared with ssrA Ϫ (Fig. 4B). Despite weak signals in the ssrA ϩ background, the transcript patterns were essentially the same as those in the ssrA Ϫ strain, showing that SsrA does not influence the production of the truncated secM transcript.
RelE, ChpB, and ChpAK and the SsrA System Are Not Required for the Production of Truncated secM mRNA-The bacterial toxins RelE, ChpB, and possibly ChpAK (MazF) cleave mRNAs in the ribosomal A-site (19 -21) (although it was recently shown that ChpAK can cleave mRNAs independently of ribosomes (22)). We therefore considered that one of these bacterial toxins could be responsible for secM transcript cleavage in response to ribosome stalling. If so, toxin-defective mutants could prevent the accumulation of SsrA-tagged SecM. However, SecM was still observed in strains deleted for the relBEF, chpB, or chpAK gene (Fig. 5). As cleavage activities might be redundant, we also constructed a ⌬relBEF ⌬chpB ⌬chpAK strain. Again, no change in the level of tagging relative to the wild-type control was observed (Fig. 5). We therefore conclude that RelE, MazF, and ChpBK do not mediate secM mRNA cleavage. Our Northern blot experiments also showed that the SsrA tagging system is not needed for secM mRNA cleavage (Fig. 4). In addition, the ⌬relBEF ⌬chpB ⌬chpAK (data not shown) and ⌬ssrA (Fig. 3) mutations do not affect the regulation of secA since no differences in secA-lacZ fusion expression were observed whether the secretion was blocked by sodium azide or not. We conclude that secM mRNA cleavage is probably not mediated by the bacterial toxins RelE, MazF, and ChpBK or by SsrA.
SsrA Counteracts Truncated secM mRNA Toxicity-We established above that SsrA is not required for secM mRNA cleavage. SsrA tagging nevertheless appears to have a biological role. Induction of ⌬sssecM transcription in the ssrA ϩ or ssrAH7 strain resulted in partial inhibition of cell growth, leading to the formation of small colonies compared with those obtained under non-induced conditions (Fig. 6) (7). When expression of ⌬ssSecM was induced, growth was more severely defective in the ⌬ssrA strain than in the isogenic ssrA ϩ strain. However, overproduction of either SsrA or SsrA-H 7 (each of which mediates release of blocked ribosomes) alleviated the growth defect due to ⌬ssSecM expression (Fig. 6). ⌬ssSecM simulates conditions in which SecM secretion is blocked; the growth defect was not observed when wild-type (secreted) SecM was expressed (Fig. 6).
We initially considered that toxicity of ⌬sssecM expression could be due to overproduction of either the ⌬ssSecM protein or its coding mRNA. Our results show that the poor growth phenotype occurred regardless of whether the tagged ⌬ssSecM protein accumulated (in the presence of ssrAH7) or was degraded (in the ssrA ϩ strain). We conclude that ribosome stall-ing on the mRNA encoding ⌬ssSecM, more than the protein itself, is responsible for toxicity. These results indicate that the SsrA tagging system alleviates toxicity by rescuing ribosomes that are stalled on truncated secM transcripts when SecM protein elongation is arrested and also by preventing accumulation of cleaved secM mRNA. DISCUSSION The sequence motif AR17 (FXXXXWIXXXXGIRAGP, where X stands for any amino acid) present in SecM was shown to cause elongation arrest within the ribosome (2). The secretiondependent SecA regulation is mediated by translation of secM and the presence of an mRNA stem-loop structure between secM and secA. All of the work presented here was performed on strains expressing constructs with secM and the complete secM-secA intergenic region. Work was performed simultaneously by Sunohara et al. (46) using a short portion of the secM gene encoding only the arrest sequence cloned between two reporter genes (crp and ccr), resulting in the expression of CRP-GIRAGP-IIA Glc .
reported previously (46). In the case of SsrA-H 7 -tagged SecM (Fig. 2), the two closely spaced bands separately analyzed by mass spectrometry showed the presence of similar peptide species, probably due to cross-contamination. One band should correspond to SsrA tagging of ⌬ssSecM at Pro 166 and Gly 165 , whereas the other band may correspond to tagging at Arg 163 .
Our Northern blot results show that a truncated secM mRNA was produced together with the full-length secM-secA mRNA. In principle, it could arise by premature transcription termination, by cleavage of the secM mRNA before translation, or by cleavage when the ribosome pauses when synthesizing the arrest sequence. Yet, we have shown that formation of the truncated secM mRNA depends on secM translation and possibly on the integrity of the ribosome pausing site in secM mRNA (Fig. 4), supporting a cotranslational cleavage mechanism. Using the construct expressing CRP-GIRAGP-IIA Glc , Sunohara et al. (46) observed five clusters of mRNA cleavage sites within their hybrid open reading frame, only one which is within the sequence encoding GIRAGP. We also observed cleavage sites at codons expressing the arrest sequence of natural SecM. 2 We therefore propose that SsrA tagging of natural SecM occurs due to secM mRNA cleavage at or around the Pro 166 -encoding sequence, leading to the appearance of a stop-less secM mRNA, which is the usual substrate for trans-translation.
Truncated secM mRNA production occurs independently of SsrA tagging. Accumulation of secM transcripts encoding nonexported SecM generates a growth defect, which is alleviated by SsrA tagging activity (Fig. 6). We propose that one role of the SsrA tagging system is to clear ribosomes stalled on truncated secM messengers. This activity may serve to avoid excessive jamming of the translation apparatus when translocation is impaired. Furthermore, the SsrA tagging system facilitates degradation of messengers lacking a stop codon (49), so it is likely that SsrA tagging of SecM may also prevent accumulation of truncated secM transcripts.
SsrA Tagging Rescues Ribosomes Stalled on Stop-less secM mRNA-The SsrA tagging system can operate on idle ribosomes at the 3Ј-end of the stop-less truncated secM mRNA and perhaps on ribosomes stalled on the uncleaved secM-secA-mutT RNA. In the latter case, ribosomes released by the SsrA system would stabilize the secA regulatory structure and thereby limit secA expression. If SsrA tagging occurs on truncated secM mRNA, which is dissociated from secA mRNA, secA translation should not be affected by the presence of SsrA. Despite its involvement in rescuing SecM stalled ribosomes, the integrity of the SsrA tagging system is not required for regulated expression of SecA (⌬ssrA (Fig. 3) or ⌬smpB (data not shown)). SsrA tagging of SecM is likely a consequence of secM-secA-mutT RNA cleavage at the end of the secM coding sequence. In E. coli, ribosome stalling seems to be a prerequisite for SsrA recruitment to the ribosome. It is therefore likely that the ribosomal A-site is always empty prior to SsrA tagging either because the mRNA was initially stop-less or because it became so after degradation or after ribosome-associated endoribonucleolytic cleavage (44,46).
Like the SecM arrest sequence, erythromycin causes elongation arrest by interacting with the exit tunnel (50 -52). Two mutations (rrlB2058 and rplV281) confer erythromycin resistance; these mutations also bypass elongation arrest during secM translation (2), thus revealing mechanistic similarities between the effects of erythromycin and SecM elongation arrest. Furthermore, SsrA-defective cells are more sensitive to erythromycin compared with wild-type E. coli cells (17) 2 as well as to accumulation of elongation-arrested only on a shorter transcript that does not include secA but includes the secM-secA intergenic sequence. In that case, secM mRNA processing and subsequent SsrA tagging of SecM would not be associated with the secretion-responsive regulation of secM. Identification of the ribosome-dependent endonucleolytic activity involved in this process will be instrumental in determining the role of secM mRNA cleavage in the secretion-responsive regulation of secA.