Portal de Periódicos da CAPES

A ribosome–nascent chain sensor of membrane protein biogenesis in Bacillus subtilis

2009; Springer Nature; Volume: 28; Issue: 22 Linguagem: Inglês

10.1038/emboj.2009.280

1460-2075

Shinobu Chiba, Anne Lamsa, Kit Pogliano,

Bacteriophages and microbial interactions

Article24 September 2009free access A ribosome–nascent chain sensor of membrane protein biogenesis in Bacillus subtilis Shinobu Chiba Shinobu Chiba Division of Biological Sciences, University of California, San Diego, CA, USAPresent address: Institute for Virus Research, Kyoto University, Sakyo-ku, Kyoto 606-8507, Japan Search for more papers by this author Anne Lamsa Anne Lamsa Division of Biological Sciences, University of California, San Diego, CA, USA Search for more papers by this author Kit Pogliano Corresponding Author Kit Pogliano Division of Biological Sciences, University of California, San Diego, CA, USA Search for more papers by this author Shinobu Chiba Shinobu Chiba Division of Biological Sciences, University of California, San Diego, CA, USAPresent address: Institute for Virus Research, Kyoto University, Sakyo-ku, Kyoto 606-8507, Japan Search for more papers by this author Anne Lamsa Anne Lamsa Division of Biological Sciences, University of California, San Diego, CA, USA Search for more papers by this author Kit Pogliano Corresponding Author Kit Pogliano Division of Biological Sciences, University of California, San Diego, CA, USA Search for more papers by this author Author Information Shinobu Chiba1, Anne Lamsa1 and Kit Pogliano 1 1Division of Biological Sciences, University of California, San Diego, CA, USA *Corresponding author. Division of Biological Sciences, University of California, San Diego, 9500 Gilman Drive La Jolla, San Diego, CA 92093-0377, USA. Tel.: +1 858 822 1314; Fax: +1 858 822 5740; E-mail: [email protected] The EMBO Journal (2009)28:3461-3475https://doi.org/10.1038/emboj.2009.280 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Proteins in the YidC/Oxa1/Alb3 family have essential functions in membrane protein insertion and folding. Bacillus subtilis encodes two YidC homologs, one that is constitutively expressed (spoIIIJ/yidC1) and a second (yqjG/yidC2) that is induced in spoIIIJ mutants. Regulated induction of yidC2 allows B. subtilis to maintain capacity of the membrane protein insertion pathway. We here show that a gene located upstream of yidC2 (mifM/yqzJ) serves as a sensor of SpoIIIJ activity that regulates yidC2 translation. Decreased SpoIIIJ levels or deletion of the MifM transmembrane domain arrests mifM translation and unfolds an mRNA hairpin that otherwise blocks initiation of yidC2 translation. This regulated translational arrest and yidC2 induction require a specific interaction between the MifM C-terminus and the ribosomal polypeptide exit tunnel. MifM therefore acts as a ribosome–nascent chain complex rather than as a fully synthesized protein. B. subtilis MifM and the previously described secretion monitor SecM in Escherichia coli thereby provide examples of the parallel evolution of two regulatory nascent chains that monitor different protein export pathways by a shared molecular mechanism. Introduction Approximately 30% of cellular proteins are translocated across or inserted into the membrane using a conserved and essential membrane translocation apparatus. In the Eubacteria, movement of proteins across the membrane depends on three integral membrane proteins that assemble a protein-conducting channel, SecY and SecE, which are present in all cells, and SecG, which is replaced by Sec61β in eukaryotes and archaea (reviewed by Pohlschroder et al, 2005; Papanikou et al, 2007; Rapoport, 2007; Driessen and Nouwen, 2008). Bacterial protein secretion depends on the SecA ATPase, which functions as a motor to translocate secreted proteins through the SecYEG channel. The insertion of integral membrane proteins into the membrane is also essential and the machinery for this process partially overlaps with the secretory apparatus: most substrates seem to require the SecYEG protein-conducting channel, but only substrates with large extracellular domains require the SecA ATPase for membrane insertion (Kol et al, 2008; Xie and Dalbey, 2008). The membrane insertion of many proteins also requires YidC, a conserved membrane protein in the YidC/Oxa1/Alb3 family that is required for membrane protein insertion and assembly of membrane protein complexes in bacteria, mitochondria and chloroplasts (Yi and Dalbey, 2005; van der Laan et al, 2005; Kiefer and Kuhn, 2007). Biochemical experiments suggest that some membrane proteins first interact with the SecYEG channel and are later transferred to YidC for integration into the lipid bilayer or for folding into functional state (Urbanus et al, 2001; van der Laan et al, 2001). However, certain membrane proteins that lack large extracellular domains are inserted into the membrane in a manner that depends on YidC but not on SecYEG, suggesting flexibility in the pathway for membrane protein biogenesis (Kiefer and Kuhn, 2007; Xie and Dalbey, 2008). The cryoelectron microscopy images of YidC (Lotz et al, 2008; Kohler et al, 2009) and the ability of purified YidC to insert certain proteins into the membrane in the absence of the Sec proteins (Serek et al, 2004; van der Laan et al, 2004), suggest that YidC provides a second protein-conducting channel. The essentiality of protein secretion and membrane protein biogenesis makes it likely that all cells can respond to decreased secretion capacity by either prioritizing certain substrates, as in eukaryotes (Kang et al, 2006; Oyadomari et al, 2006; Hegde and Kang, 2008), or by increasing the secretion capacity of the cell. Indeed, Escherichia coli and closely related bacteria have a feedback mechanism to maintain cellular protein secretion capacity by increasing SecA translation when the SecYEG pathway is compromised (Oliver and Beckwith, 1982; Schmidt and Oliver, 1989). This feedback mechanism depends on a gene located upstream of secA, secM, that acts as a ribosome–nascent chain complex to monitor protein secretion capacity (McNicholas et al, 1997; Schmidt et al, 1988; Nakatogawa and Ito, 2001, 2002). Prior data suggested that Bacillus subtilis cells might have the ability to monitor the YidC-dependent membrane protein insertion pathway (Rubio et al, 2005). Specifically, B. subtilis, like many other Gram-positive bacteria, encodes two YidC homologs (SpoIIIJ and YqjG) either of which can support viability (Murakami et al, 2002; Tjalsma et al, 2003). However, the role of these proteins in B. subtilis remains unclear (Murakami et al, 2002; Tjalsma et al, 2003), in contrast those encoded by Streptococcus mutans, which clearly has a role in membrane protein insertion (Hasona et al, 2005; Dong et al, 2008; Funes et al, 2009). The B. subtilis spoIIIJ gene is likely the primary yidC homolog (YidC1), because it is constitutively expressed and located in a similar chromosomal context as E. coli yidC (Errington et al, 1992). The second yidC homolog (yqjG; here renamed yidC2) is likely the backup system for spoIIIJ (yidC1), because it is induced in spoIIIJ mutants (Rubio et al, 2005). This latter observation suggests that B. subtilis has a mechanism by which it can sense changing SpoIIIJ activity and increase YidC2 levels as needed to maintain appropriate cellular levels of YidC activity. We here describe the mechanism by which B. subtilis monitors SpoIIIJ (YidC1) activity. Specificially, we show that YidC2 induction involves the product of the upstream gene, yqzJ, which is in an operon with yidC2, and an mRNA hairpin that occludes the yidC2 Shine–Dalgarno (SD) site. The yqzJ gene encodes a small membrane protein with an N-terminal transmembrane (TM) segment that is a likely substrate of SpoIIIJ (Xie and Dalbey, 2008). YqzJ also has a C-terminal translational arrest domain that interacts with the ribosomal polypeptide exit tunnel to induce a stable translational arrest when SpoIIIJ is absent or limiting. This translational arrest positions the stalled ribosome over the 5′ region of the mRNA hairpin, thereby exposing the yidC2 SD site and allowing increased translational initiation. Membrane insertion relieves the translational arrest, so when membrane protein insertion is inhibited, the hairpin remains unfolded for longer periods of time, allowing additional YidC2 to be synthesized until cellular membrane protein insertion capacity is restored. Thus, YqzJ acts as a cellular monitor of SpoIIIJ-dependent membrane protein insertion while it is a ribosome–nascent chain complex, rather than as a fully synthesized protein. The regulatory mechanism we describe bears striking similarities to the E. coli protein secretion monitor SecM. We therefore propose renaming yqzJ as mifM, for membrane protein insertion and folding monitor. However, genomic and sequence comparisons between MifM and SecM reveal that the translational arrest motifs of these two regulatory proteins share no detectable sequence similarity and that they are found in different bacterial kingdoms. This suggests that SecM and MifM are independently evolved ribosome–nascent chain sensors that use a common molecular mechanism to monitor different protein export pathways. Results Disruption of spoIIIJ upregulates yidC2 expression at the translational level To understand the mechanism of yidC2 (yqjG) induction in spoIIIJ mutants, we first investigated whether regulation occurs at the level of transcription, translation or protein stability. The yidC2 gene is in an operon with the ORF yqzJ (here renamed mifM, as described below), which encodes a 95 amino acid protein (Figure 1A). As shown earlier (Rubio et al, 2005), an in-frame fusion of lacZ to the stop codon of yidC2 (yidC2275–lacZ; Figure 1Aa) exhibits elevated β-galactosidase activity in spoIIIJ strains (Figure 1Ba). This was also the case for an in-frame fusion of lacZ after the 6th codon of yidC2 (yidC26–lacZ; Figure 1Ab and Bb), so it is unlikely that YidC2 induction is due to increased proteolytic stability of YidC2 in the absence of SpoIIIJ. If yidC2 induction occurs by modulating mifM–yidC2 transcription or mifM translational initiation, then an in-frame fusion of lacZ after the 5th codon of mifM (mifM5–lacZ: Figure 1Ac) should also show elevated β-galactosidase activity in spoIIIJ strains. However, identical levels of β-galactosidase activity were observed in the wild-type and spoIIIJ strains with the mifM5–lacZ fusion (Figure 1Bc). Thus, the absence of SpoIIIJ is likely to induce yidC2 translation in a manner independent of mifM transcription and translational initiation. Figure 1.mifM regulates yidC2 expression. (A) Genetic context of yqjG (yidC2) and in frame translational lacZ fusions to the 3′-end of yidC2 (a) or mifM (e), producing YidC2275–LacZ and MifM95–LacZ, respectively, and after yidC2 codon 6 (b, d) and mifM codon 5 (c). Construct (d) also contains an in-frame deletion of the mifM region encoding the transmembrane (TM) segment. (B) β-galactosidase (β-gal) activity in wild type (white) or the spoIIIJ mutant (black). The fusions (a–e) correspond to those shown in (A). (C) Predicted secondary structure of mifM–yidC2 mRNA. SD, Shine–Dalgarno site recognized by the ribosome to initiate translation. The class II mutations of mifM are indicated by arrows with residue's numbers. (D) Diagram of mifM mutants in the yidC26–lacZ fusion. (f) Silent mutations to disrupt the hairpin without changing MifM amino acids (yidC26–lacZ(Δstem)). (g) Stop codon at codon 86. (h) Duplication of codons 91–96 (including the stop codon) to allow translation of full-length mifM without unfolding the hairpin. (E) β-gal activity with the stem-loop mutation shown in (D) (f). (F) Translation through the hairpin is required for yidC2 expression. β-gal activity of strain with mutations (g) and (h) from (D). wt, wild type. (G–H) mifM functions in cis. Expression of yidC2-GFP monitored by fluorescence microscopy with identical GFP (green) exposures and image adjustments for each strain; membranes were stained with FM 4-64 (red). (G) Expression of yidC2-GFP downstream of wild-type mifM in wild type (H) Expression of yidC2-GFP downstream of wild-type mifM in spoIIIJ (I) MifMΔTM expressed in a wild-type strain with yidC2-gfp downstream of wild-type mifM. Download figure Download PowerPoint mifM translation disrupts a stem loop that blocks yidC2 translation To further understand the mechanism by which yidC2 is regulated, we performed a genetic screen to isolate mutants that induce yidC2. A strain containing the yidC26–lacZ fusion was mutagenized with NTG (N-methyl-N′-nitro-N-nitrosoguanidine) and plated on DSM plates containing the β-galactosidase indicator XG. Blue, Lac+ colonies were picked and purified. Of the 150 Lac+ mutants, 37 contained mutations in spoIIIJ and 17 were 100% linked to the yidC26–lacZ reported gene. DNA sequence analysis revealed that eight of the spoIIIJ mutations altered the SD sequence (Supplementary data), suggesting that reducing the level SpoIIIJ protein is sufficient to induce yidC2. All of the mutations linked to yidC6–lacZ were within the upstream ORF, mifM. Ten mifM mutants had the same point mutation, a G-A transition at nucleotide 16, causing a glutamate to lysine substitution at the 6th amino acid of MifM; hereafter referred to as the class I mutation. The remaining seven mutations were at three sites near the 3′ end of mifM, C275T (Ser92Thr), C277T (Leu93Phe) or C280T (Leu94Phe); hereafter referred to as the class II mutations. These results suggested that mifM has an important function in regulation of yidC2 expression. Examination of the 5′ end of yidC2 revealed a potential stem loop that could mask the yidC2 SD site (Figure 1C). The 5′ half of the stem loop is located within the 3′ end of mifM, where the class II mutations are located (Figure 1C). We therefore hypothesized that the stem loop repressed the initiation of yidC2 translation. In keeping with this idea, all class II mutations disrupt G-C pairing of the stem loop, suggesting that the mutations destabilize the structure, thereby leading to high-level constitutive expression of yidC2. To test this model, we introduced silent mutations that disrupt the structure without changing the MifM amino acid sequence (yidC26–lacZ(Δstem); Figure 1Df). Disrupting the stem loop indeed caused high-level constitutive expression of yidC2 in wild-type and spoIIIJ strains (Figure 1E), suggesting that the stem loop inhibits yidC2 translation. To test whether mifM translation is necessary for yidC2 translation, we first terminated translation before the ribosome reached the stem loop by introducing a TAA stop codon at codon 86 (Figure 1Dg). This abolished yidC2 induction in a spoIIIJ strain and slightly reduced yidC2 expression in wild type (Figure 1Fg). We next duplicated six mifM codons (91–96) including the stop codon, so that wild-type MifM was synthesized but translation stopped before the ribosome moved into the stem loop (Figure 1Dh). This also resulted in low and constitutive expression of yidC2 (Figure 1Fh) that was increased when the stem loop was disrupted (not shown). We conclude that translation of mifM into the 5′ half of the stem loop is necessary for high-level yidC2 expression. MifM senses membrane protein insertion SpoIIIJ and YidC2 are likely involved in membrane protein biogenesis or protein secretion (Tjalsma et al, 2003; Camp and Losick, 2008). Interestingly, MifM is predicted to be a small membrane protein with a single N-terminal TM segment (amino acids 12–34) and a C-terminal cytoplasmic domain (amino acids 35–95). This topology is similar to that of E. coli proteins that are inserted into the membrane in a manner dependent on the SpoIIIJ homologue YidC (Xie and Dalbey, 2008), suggesting that MifM is inserted into the membrane by SpoIIIJ. Interestingly, the class I MifM mutation changes the charge of the N-terminal domain from negative to positive, which might produce a membrane insertion defect that induces yidC2, thereby mimicking SpoIIIJ limitation. To test this model, we deleted the region encoding the MifM TM segment and found that it caused elevated yidC2 expression in wild-type cells (Figure 1Ad and Bd). These results suggest that MifM serves as a sensor of SpoIIIJ activity that upregulates yidC2 when SpoIIIJ activity is limiting and membrane insertion is blocked. MifM is a cis-acting regulator of yidC2 translation If mifM translation unfolds the stem loop in a manner regulated by SpoIIIJ-dependent membrane insertion of MifM, then it should regulate YidC2 expression only in cis. To test this prediction, we expressed MifM(ΔTM) in trans to yidC2-gfp with wild-type mifM upstream. The strain carrying only yidC2-gfp showed weak GFP fluorescence in wild-type cells (Figure 1G), and stronger membrane-associated GFP fluorescence in the spoIIIJ background (Figure 1H) as expected (Rubio et al, 2005). Expression of MifM(ΔTM) did not elevate the YidC2-GFP signal (Figure 1I) unless it was encoded upstream of YidC2-GFP, indicating that mifM functions only in cis. Together these results suggest that MifM is a cis-acting regulator of yidC2 translation that senses SpoIIIJ activity, likely when it is inserted into the membrane as a ribosome-associated nascent chain. We therefore renamed its gene (formerly yqzJ) mifM for membrane insertion and folding monitor. MifM is incompletely translated in the absence of SpoIIIJ The above results suggest that in the absence of SpoIIIJ, the mifM–yidC2 stem loop is more efficiently unfolded than in the presence of SpoIIIJ. Unfolding could be accomplished either by allowing more ribosomes to translate mifM–yidC2 mRNA or by having ribosomes move more slowly across the stem loop so it remains unfolded for longer times. The former model is unlikely because the initiation of mifM translation occurs at similar levels in the presence or absence of SpoIIIJ (Figure 1Ac and Bc). We therefore hypothesized that the absence of SpoIIIJ might stall the ribosome in a location that occluded the 5′ side of the stem loop thereby exposing the yidC2 SD. To test this idea, we fused lacZ at the 3′-end of the mifM coding sequence (producing mifM95–lacZ; Figure 1Ae). This fusion showed lower β-galactosidase activity when spoIIIJ was disrupted, as expected if mifM translation was attenuated under these conditions (Figure 1Be). This suggests that mifM translation is arrested when membrane insertion of MifM is blocked by the spoIIIJ mutation or deletion of the MifM TM domain. Detection of a MifM translational intermediate If mifM translation is arrested in the absence of membrane insertion, it might be possible to observe a translational intermediate in mifM(ΔTM) strains. We therefore constructed strains expressing an N-terminal fusion of GFP to MifM lacking its N-terminal TM domain with or without a C-terminal FLAG-tag for detection of the full-length protein (Figure 2A; GFP–MifM35−95–FLAG and GFP–MifM35−95, respectively). Immunoblotting using GFP-specific antibodies detected two major products in the strain expressing GFP–MifM35−95 (Figure 2B, lane 2) and an additional larger minor product in the strain expressing GFP–MifM35−95–FLAG (Figure 2B, lane 3). This minor product (band a) was detected with both GFP- and FLAG-specific antibodies (Figure 2B, lanes 3 and 6), so it must be full-length GFP–MifM35−95–FLAG. However, the shorter products (bands b and c) were not detected by FLAG-specific antibodies, so they must be N-terminal fragments (Figure 2A). To test whether these shorter products were translational intermediates or degradation products, we used CTABr precipitation (Gilmore et al, 1991; Nakatogawa and Ito, 2001), which precipitates RNA and RNA–protein complexes such as translational intermediates that retain a covalently attached tRNA at the C-terminus. Fragment b but not fragment c was precipitated by CTABr (Figure 2C) and this CTABr precipitation was abolished by RNase treatment. Thus, fragment b is an MifM translational intermediate whereas fragment c is likely a degradation product. These results show that mifM(ΔTM) undergoes a strong translational arrest such that the arrest species appears more abundant than full-length protein. The apparent size of fragment b suggests that the elongation arrest occurs within the 3′-region of mifM, a position at which the arrested ribosome could prevent formation of the mRNA hairpin that blocks yidC2 translation. Figure 2.Visualization of MifM translational arrest. (A) The C-terminus of GFP was fused to amino acids 35–95 of MifM with or without the C-terminal FLAG-tag. (a–c) Products visualized in immunoblots. (B) Immunoblotting of GFP–MifM35−95 derivatives. Proteins from strains expressing GFP–MifM35−95 (lanes 2, 5), GFP–MifM35−95–FLAG (3, 6) or no GFP fusion (1, 4) analysed by immunoblotting with anti-GFP (left panel) or anti-FLAG (right panel). (C) CTABr fractionation of GFP–MifM35−95. Proteins from strains expressing GFP–MifM35−95 (lanes 2, 5, 8, 11, 14), GFP–MifM35−95–FLAG (3, 6, 9, 12, 15) or no GFP fusion (1, 4, 7, 10, 13) analysed by immunoblotting with anti-GFP after CTABr precipitation. Samples 10–15 were treated with RNaseA before CTABr. WCL, whole cell lysate; ppt, precipitated fraction; sup, supernatant. Download figure Download PowerPoint A specific amino acid sequence is required for MifM translational arrest To further elucidate the mechanism by which mifM regulates yidC2, we tested the relative importance of the mifM mRNA and amino acid sequences in the region of the hairpin. To do so, we designed a frame-shift mutation within mifM that would produce an mRNA sequence almost identical to native mifM but encode a completely different amino acid sequence. We first introduced an extra stop codon and a silent mutation (Figure 3Aa) to prevent out of frame translation from producing a product much longer than native MifM, and to prevent premature translational termination, respectively. The yidC26–lacZ fusion with these two mutations showed low-level expression in the wild-type strain and induction in the spoIIIJ strain to levels similar to the wild-type fusion (Figure 3Ca). When then introduced a frame-shift mutation at codon 78 (Figure 3Ab), which resulted in low and constitutive yidC26–lacZ expression (Figure 3Cb), showing that the translational arrest was abolished. We conclude that a specific amino acid sequence at the MifM C-terminus is required for translational arrest. Figure 3.Identification of MifM amino acids required for translational arrest. (A) The mifM frame-shift mutant; white characters in black box indicate the mutations. Arrows show site of stem loop. (B) Introduction of stop codons in mifM (*). The number at left represents the codon substituted by a TAA stop codon. (C) β-gal activity of mutants in A in wild type (white) or spoIIIJ (black). (D) β-gal activity of mutants in B in wild type (white) or spoIIIJ (black). (E) Ala-replacement of GFP–MifM35−95–FLAG at the mifM codons indicated by numbers. Immunoblotting with anti-GFP and anti-FLAG. Arrest-defective substitutions accumulate full-length protein. (F) β-gal activity of yidC26–lacZ containing the indicated MifM Ala-replacements, in wild type (white) or spoIIIJ (black). Download figure Download PowerPoint Identification of the site of translational arrest Stop codons before the translational arrest site should prematurely release the ribosome from the mRNA, allowing the stem loop to refold and inhibiting YidC2 expression, whereas those after the arrest site should have no effect. We therefore introduced stop codons immediately to the 5′ of the stem loop to determine whether the mutation prevented yidC2 induction in the spoIIIJ strain (Figure 3B). Stop codons before codon 88 blocked yidC2 induction (Figure 3D), whereas a stop codon at codon 90 showed wild-type induction (Figure 3D). These results indicate that mifM translation must continue to codon 88 to allow yidC2 expression and that codon 90 is dispensable for translational arrest. Codon 89 was not fully essential for the induction of yidC2, though it might participate in stabilizing the translational arrest, because substitution of codon 89 by a stop codon showed intermediate yidC2 expression. These data indicate that the translational arrest occurs near the MifM C-terminus, most likely after codon 88 is translated, consistent with the estimated molecular weight of the translational arrest product of GFP–MifM35−95 (∼35 kDa, Figure 2B, band b). Translating ribosomes cover approximately 6–9 nucleotides of mRNA to the 3′ of the A site (Culver, 2001; Yusupova et al, 2001, 2006; Takyar et al, 2005), which would allow a ribosome arrested at codon 88 to partially cover the 5′ side of the stem loop (see Figure 1C), thereby exposing the yidC2 SD for the duration of the arrest. Arrest requires amino acids within the ribosomal polypeptide exit tunnel To identify amino acids required to arrest MifM translation, we mutagenized the C-terminal region of MifM. Codons 59–89 were targeted because the above results suggested that arrest occurred near codon 89, which would place this codon near the peptidyl transfer site of the ribosome, and because the C-terminal 30–40 amino acids of nascent polypeptides lie within the polypeptide exit tunnel of the ribosome and are therefore well positioned to mediate a translational arrest (Krieg et al, 1989; Mothes et al, 1994; Frank et al, 1995; Matlack and Walter, 1995; Ban et al, 2000; Daniel et al, 2008). We constructed a series of alanine replacements in GFP–MifM35−95–FLAG and monitored accumulation of the arrest species by immunoblotting with GFP- and FLAG-specific antibodies. Changing I65, H68, R69, I70, W73, I74, M80, N81 to alanine clearly increased accumulation of full-length MifM (Figure 3E), suggesting they might be required for the arrest, whereas other substitutions had more subtle effects. These substitutions could affect protein or mRNA stability as well as translation, so we introduced identical substitutions in wild-type mifM in the context of yidC2–lacZ to assess the ability of these proteins to induce YidC2 in the spoIIIJ strain. Substitutions at amino acids Y67, R69, I70, T71, W73, I74, M80 or N81 reduced YidC2 induction in the spoIIIJ strain as expected for mutations that reduce the translational arrest (Figure 3F). Amino acids that both increased production of full-length MifMΔTM and reduced yidC2 induction in the spoIIIJ strain allowed us to identify a minimal set of amino acids that are clearly critical for the arrest (R69, I70, W73, I74, M80, N81); these are upstream of the translational arrest that occurs at E88 or D89. Interestingly, substitutions at the site of the arrest (E88A, D89A) had only minor effects in either assay, indicating that the specific amino acid or peptidyl–tRNA at the site at which translation arrests is not critical (although it is critical that translation continues beyond these amino acids, Figure 3D). The amino acids necessary for translational arrest lie within the polypeptide exit tunnel of the ribosome and distal to the peptidyl transferace centre (PTC) at which elongation occurs. A mutation that alters the polypeptide exit tunnel of the ribosome compromises elongation arrest The above results suggest that an interaction between the MifM nascent polypeptide and the ribosomal polypeptide exit tunnel arrests translation. If so, then mutations that alter the polypeptide exit tunnel could impair the interaction and prevent the translational elongation arrest. To isolate arrest-defective ribosomes, we took advantage of the similarity in the mode of action of MifM and erythromycin, which binds within the exit tunnel to inhibit polypeptide elongation (Schlunzen et al, 2003; Tu et al, 2005; Vazquez-Laslop et al, 2008). This effect can be prevented by mutations that change the amino acids and rRNA that line the exit tunnel, which convey erythromycin resistance (ErmR; (Gaynor and Mankin, 2003; Zaman et al, 2007; Diner and Hayes, 2009). Certain ErmR mutants in E. coli are defective in translational arrests mediated by regulatory nascent chains, such as the SecAYEG monitor SecM (Nakatogawa and Ito, 2002; Lawrence et al, 2008), the tryptophan sensor TnaC (Cruz-Vera et al, 2005), the erythromycin sensor ErmCL (Vazquez-Laslop et al, 2008) and the chloramphenicol sensor CrbcmlA (Lawrence et al, 2008). We therefore isolated spontaneous ErmR mutants from independent B. subtilis cultures. These mutants all had an identical duplication of seven amino acids in the L22 protein (94SQINKRT100; Figure 4A). On the basis of crystal structure of the ribosome (Ban et al, 2000), it is likely that this duplication causes a seven amino acid insertion within the interior of the ribosome exit tunnel (Figure 4A and E), in a region that can be mutated to ErmR in other bacteria (Franceschi et al, 2004; Zaman et al, 2007). Figure 4.A mutation affecting the ribosomal protein L22 compromises elongation arrest (A) The spontaneous ErmR mutant has a seven amino acid duplication in L22 (white letters on black). Alignment of this region of B. subtilis (Bs) and E. coli (Ec) L22; underlined amino acids affect elongation arrest of E. coli tnaC (K90; Cruz-Vera et al, 2005) or secM (G91 and A93; Nakatogawa and Ito, 2002). (B) β-gal activity of mifM(ΔTM)-yidC26–lacZ (left) or the stem-loop mutant (right) in ErmR (black) and wild-type (white) strains. (C) Immunoblotting of GFP–MifM35−95–FLAG in the ErmR mutant with GFP and FLAG-specific antibodies. Strains with (2, 3, 5, 6) or without GFP fusion (1, 4) in wild type (lanes 1, 2, 4 and 5) or ErmR (3, 6). *, non-specific bands (D) Synthetic cold sensitive growth is observed in the erm spoIIIJ double mutant. Strains were streaked on LB plates and incubated at the indicated temperature (RT, room temperature ∼24°C). (E) View of the ribosome polypeptide exi