Recruitment of a species-specific translational arrest module to monitor different cellular processes

Abstract

Nascent chain-mediated translation arrest serves as a mechanism of gene regulation. A class of regulatory nascent polypeptides undergoes elongation arrest in manners controlled by the dynamic behavior of the growing chain; Escherichia coli SecM monitors the Sec protein export pathway and Bacillus subtilis MifM monitors the YidC membrane protein integration/folding pathway. We show that MifM and SecM interact with the ribosome in a species-specific manner to stall only the ribosome from the homologous species. Despite this specificity, MifM is not exclusively designed to monitor membrane protein integration because it can be converted into a secretion monitor by replacing the N-terminal transmembrane sequence with a secretion signal sequence. These results show that a regulatory nascent chain is composed of two modular elements, one devoted to elongation arrest and another devoted to subcellular targeting, and they imply that physical pulling force generated by the latter triggers release of the arrest executed by the former. The combinatorial nature may assure common occurrence of nascent chain-mediated regulation.

All proteins are conducted through the polypeptide exit tunnel of the ribosome during their synthesis. An increasing number of examples are now known for nascent chains that interact with the exit tunnel and induce arrest in the elongation or the termination process of translation (1, 2). These studies revealed that the ribosome cannot necessarily complete translation of messages for any given amino acid sequences without any delay. Remarkably, programmed elongation arrest can serve as a novel mechanism of cellular regulation, for instance to monitor protein localization, metabolite concentration, and reception of an antibiotic (1–6). Although regulatory nascent peptides having divergent stall-inducing amino acid sequences have been identified from different living organisms, their species-specificity and exchangeability among different systems have not been explored. Thus, significance of the sequence variation and molecular designs of regulation is not fully understood.

To address the above problems, we here highlight two regulatory systems, MifM of Bacillus subtilis (5) and SecM of Escherichia coli (6, 7), that use programmed ribosomal stalling in a manner feedback-regulated by targeting of the nascent chain; MifM monitors membrane insertion/folding pathway whereas SecM monitors protein secretion across the membrane. E. coli SecM is encoded by the 170-codon open reading frame upstream of secA, encoding the protein export-driving ATPase, in the same transcription unit (8–10). It is designed as a periplasmic protein but contains an arrest sequence, F₁₅₀XXXXWIXXXXGIRAGP₁₆₆, which was suggested to interact with the exit tunnel of the ribosome (6). Thus, its translation is subject to elongation arrest (7), generating the nascent chain-ribosome complex having SecM_1–165-tRNA at the P site and Pro166-tRNA at the A site (11). The stalled ribosome uncovers the SD sequence required for translation of secA (12). Therefore, the elongation arrest, which is transient in normal cells but prolonged in secretion-compromised cells, assures the basal level expression of secA as well as its up-regulation in response to a secretion defect (13). MifM monitors YidC-dependent membrane biogenesis pathway in B. subtilis, which has two YidC homologs, the primary SpoIIIJ (YidC1) and the secondary YidC2 (5, 14). It is encoded by the open reading frame upstream of yidC2 and contains a hydrophobic sequence at the N-terminal region, which is guided by SpoIIIJ for insertion into the membrane and/or subsequent folding (15). Like SecM, it also contains an arrest-inducing sequence, which is suggested to interact with the ribosomal exit tunnel, and undergoes elongation arrest. The stalled ribosome then uncovers the SD sequence of yidC2. Because the SpoIIIJ-mediated insertion of MifM leads to the release of elongation arrest, yidC2 is induced when the SpoIIIJ function is impaired (5). SecM and MifM share no appreciable sequence similarity and monitor different protein localization pathways in different organisms, yet their modes of actions are strikingly similar.

To study MifM’s ability to stall the ribosome, we modified the in vitro translation system known as protein synthesis using recombinant elements (PURE) (16) to use either the B. subtilis ribosome or the E. coli ribosome, and were able to reproduce the MifM elongation arrest in vitro and to demonstrate that the arrest-inducing abilities of nascent MifM and nascent SecM are only executed effectively when translated by the homologous, native ribosome. The species specificity we observed substantiates the divergent evolution of the ribosome-stalling sequences (2, 5, 17), which are likely to have been “tailor-made” to fit the native ribosome of the organism. Nevertheless, we were able to engineer the MifM polypeptide, which originally monitors protein insertion into the membrane, to monitor protein secretion by replacing its membrane-anchoring sequence by an export signal sequence. Thus, even an independently evolved arrest element can be placed in a common platform of regulation to sense different categories of cellular physiology.

Results

Hybrid PURE System Active with the B. subtilis Ribosome.

Previous genetic evidence suggests that the elongation arrest of MifM translation is mediated by molecular interactions between the intraribosomal region of the MifM nascent chain and the exit tunnel of the ribosome (5). To gain insights into the molecular mechanisms of the elongation arrest, we set up an in vitro translation system that uses the B. subtilis (Bs) ribosome. One of the basic questions to be addressed would be whether the ribosome-nascent chain interaction is sufficient for the elongation arrest without involving any cytoplasmic component. We tested this using the PURE system, a coupled in vitro transcription–translation system entirely composed of purified components, IF1, IF2, IF3, EF-G, EF-Tu, EF-Ts, RF1, RF2, RF3, RRF, 20 aminoacyl-tRNA synthetases (ARSs), methionyl-tRNA transformylase (MTF), T7 RNA polymerase (RNAP), and the ribosome, all of which have been purified from E. coli cells and originally encoded by the E. coli genome, except for the RNAP (16). This system minimizes the effects of cytosolic factors such as molecular chaperones (18) and targeting factors. We envisioned that if the PURE system worked with ribosomes prepared from B. subtilis, it would prove to be an ideal assay system to determine if MifM translation arrest occurred in the absence of any effective amounts of nonribosomal components from this bacterium.

We thus constructed “Bs hybrid PURE system,” in which the E. coli ribosome in the original PURE system was replaced by that of B. subtilis (Fig. 1A). We purified the ribosomes from B. subtilis (Bs ribosome; lane 1 in Fig. 1B) and from E. coli (Ec ribosome; lane 2 in Fig. 1B) and combined them with the other components of the PURE system. A DNA fragment encoding GFP under the T7 promoter was used as the template to examine whether the Bs hybrid PURE system was active in synthesizing protein. Indeed, GFP was synthesized not only with the Ec ribosomes but also with the Bs ribosomes, as shown by in-gel renaturation of the synthesized GFP molecules and their fluorescence emission. Judging from the fluorescence intensities, it was estimated that the Bs ribosome was typically 10–20% as active as the Ec ribosome (Fig. 1B, Lower). We next examined whether the activity of the Bs ribosome to synthesize protein was dependent on the E. coli translation factors by removing each one of them from the reaction mixture. Omission of any one of the three initiation factors reduced GFP synthesis by both Ec and Bs ribosomes (Fig. 1C). In addition, the Ec elongation factors proved essential for protein synthesis even by the Bs ribosome because the absence of either EF-G or EF-Tu alone abolished GFP production almost completely, whereas the absence of Ts strongly reduced it, irrespective of the sources of the ribosome. These results indicate that the Bs ribosomes can functionally interact with the E. coli-derived initiation and elongation factors in the PURE system reaction of protein synthesis. The lower overall translation activity observed with the Bs ribosome may have been due to some species specificity in the actions of EF-G and EF-Tu.

Fig. 1.

Construction of Bs hybrid PURE system for in vitro translation. (A) Components included in Ec PURE system and Bs hybrid PURE system. (B) Purified B. subtilis and E. coli ribosomes and their translation activities. Ribosomes were prepared from B. subtilis (Bs; lane 1) and E. coli (Ec; lane 2) and their protein components were stained with coomassie brilliant blue after SDS/PAGE separation (upper panel). GFP was synthesized with Bs hybrid PURE system (lane 1) or with Ec PURE system (lane 2) and visualized by its fluorescence after in-gel renaturation. (C) Requirements for the translation factors. The indicated components were removed from the Ec and Bs PURE systems (white and black bars, respectively) and the synthesis of GFP was examined. Intensities of GFP fluorescence (upper panels) were quantified from gel images (upper panels) obtained with voltages of photomultiplier tube set at 850 V for Bs and 450 V for Ec and shown as values relative to that obtained with the complete reaction mixture (lower panel).

In Vitro Reproduction of MifM Elongation Arrest.

We used the Bs hybrid PURE system to translate MifM in vitro to determine if translational arrest required any other cellular components. The elongation-arrested peptidyl-tRNA form of MifM was visualized by separating the translation products by SDS/PAGE under neutral pH conditions (NuPAGE), which preserves the alkali-labile aminoacyl-tRNA ester bonds (11). The translation time course was followed by subjecting samples at intervals to NuPAGE and subsequent anti-MifM Western blotting. Initially (at 5 min), a 30 kDa product appeared (Fig. 2A, lane 2; MifM’-tRNA) that increased in intensity and plateaued after 30 min. This band likely represents an elongation-arrested MifM-tRNA product, because it disappeared when samples were treated with RNaseA before electrophoresis (lanes 7–10). A 10 kDa product (MifM/MifM′) became visible at 10 min (lane 3), and its intensity kept increasing up to the 60-min time point examined (lanes 3–5).

Fig. 2.

In vitro reproduction of translation arrest of MifM with the Bs ribosome. (A) In vitro translation of mifM with the wild-type ribosome. In vitro translation with Bs hybrid PURE system was directed by mifM template (lanes 1–10) or the mifM-myc template (lanes 11–20) at 37 °C. Samples were withdrawn at the indicated time points and then analyzed by SDS/PAGE using 12% NuPAGE followed by anti-MifM immunoblotting. To remove tRNA moiety from polypeptidyl-tRNA, samples before electrophoresis were treated with RNaseA (lanes 6–10 and 16–20). (B) Effects of an arrest-compromising mifM mutation on the translational arrest in vitro. Wild-type (lane 1) or Ile70Ala derivative (lane 2) of mifM-myc was translated in Bs hybrid PURE system for 30 minutes at 37 °C. SDS/PAGE samples were treated with RNaseA and analyzed by immunoblotting as described above. (C) Effects of a ribosome mutation on the MifM translational arrest in vitro. The rplV94 mutant form of the ribosome was purified and used as the ribosomal source in Bs hybrid PURE system. Translation products of mifM-myc were analyzed as described above. MifM’-tRNA and MifM’ indicate the nascent polypeptidyl-tRNA and its polypeptide moiety, respectively, produced when translation of MifM was arrested. MifM and MifM-myc indicate the full-length, translation-completed products. Asterisks indicate products likely produced by the failure in termination or by some nonspecific elongation pause at a C-terminal region.

The RNaseA-treated sample at 5 min gave a band of virtually identical mobility as the 10 kDa band at the expense of the 30 kDa band. Because translation of MifM is arrested in vivo at or after the 88th codon (5), it appeared difficult to distinguish between the polypeptide portion of the arrested product (likely having 88 amino acids) and the full-length, translation-completed species (95 amino acids) from their electrophoretic mobilities. We therefore added a myc-tag to the C terminus of MifM to extend the full-length product to an extent separable from the arrested polypeptide moiety by SDS/PAGE. When mifM-myc was translated with Bs hybrid PURE system, the 30 kDa band again appeared earlier (at 5 min; lane 12) than the other major band, which now migrated at the position of 12 kDa and became detectable at 10 min (lanes 13–15). This 12 kDa band must represent the full-length MifM-myc polypeptide because it is resistant to RNaseA (lanes 18–20) and uniquely observed after the myc-tag addition (compare lanes 2–5 and lanes 12–15). RNaseA treatment of the 5-min sample led to the disappearance of the 30 kDa band with concomitant appearance of the 10 kDa band (lane 17), indicating that the former (MifM′-tRNA) was converted to the latter (polypeptide moiety, MifM′).

The myc-tagged constructs allowed us to specifically detect the hydrolysis product of MifM-tRNA (10 kDa) as well as the translation-completed species (12 kDa). Because at least approximately 90% of the translation products of MifM-myc were in the form of MifM′-tRNA 5 min after the start of the reaction, we conclude that most, if not all, of the ribosomes that translate MifM-myc had undergone translational pause. However, the arrest seems to be released spontaneously at a certain frequency, leading to the production of the full-length MifM-myc protein at the later time points of the incubation (lanes 18–20). The ratio of arrest/full-length went down markedly during the early part of the time course, suggesting that the arrest species was not a dead-end product but an intermediate in the synthesis of the full-length protein. It should be noted that spontaneous arrest release allows the ribosomes to initiate translation again and to produce the arrested product even at the later time points, explaining why the arrest product continued to be observable through the entire time course of this experiment.

The results presented above show that in vitro translation of MifM is arrested at a point very close to the C terminus as previously shown to be the case in vivo (5). To substantiate that the arrest we observed in vitro was physiologically relevant, we first examined an arrest-compromised mutant of MifM with Ile70Ala alteration (5). This mutant form of MifM indeed exhibited significantly lowered efficiency of elongation arrest in vitro (Fig. 2B, lane 2). Secondly, we used the rplV94 mutant form of the Bs ribosome with an altered L22 protein, which contained duplicated seven codons for rplV^94–100 and conferred both erythromycin resistance and a lower efficiency of MifM arrest in vivo (5). We purified the mutant ribosome, which was combined with the other PURE system components (without the ribosome). In comparison to the reaction with the wild-type Bs ribosome, significantly lower proportions of the translation products were in the form of either MifM′-tRNA (Fig. 2C, lanes 2–5) or MifM′ (Fig. 2C, lanes 7–10). The difference was especially striking at earlier time points of translation. Thus, the ability to arrest MifM translation efficiently was an attribute of the wild-type ribosome, showing that the ribosome has an active role in the arrest.

The in vitro reaction products included some minor bands (asterisks in Fig. 2), which disappeared after RNaseA treatment. These bands are likely to represent products of some nonspecific elongation arrest or a delay in the termination of translation; importantly, the rplV94 mutation of the ribosome did not reduce the level of these products (Fig. 2C). The data presented so far revealed that MifM can efficiently arrest its translation under the reaction conditions, which involve only the minimal essential set of translation components. Together with the previous data demonstrating that nascent polypeptide of MifM rather than its mRNA triggers the elongation arrest (5), we conclude that the translational arrest of MifM is mediated by direct nascent chain-ribosome interactions.

Species Specificity of the Stalling Sequence Action.

In spite of the importance of ribosome-nascent chain interaction and consequent ribosomal stall, little is known about the actual mechanisms of elongation arrest. An important unanswered question is the species specificity of the arrest sequences, whose known examples are quite divergent in amino acid sequences. Because the ribosome is thought to be one of the most universal enzymes, it is conceivable that different arrest sequences still interact with conserved sets of ribosomal components, as observed with a number of antibiotics (19, 20). Alternatively, they might interact with the ribosome in disparate ways, at least in the initial recognition. In the former case, an arrest sequence from a particular organism is likely to work with the ribosomes from different species. Our in vitro translation system offers an excellent opportunity to examine whether the MifM and the SecM stalling sequences can function even when they are translated by heterologous ribosomes. These are the only reported bacterial regulatory nascent chains having an intrinsic arrest sequence that does not depend on a metabolite or an antibiotic.

To test whether the MifM arrest sequence can interact with the E. coli ribosome to stall it, mifM-myc was translated with the regular PURE system having the Ec ribosome. In striking contrast to what we observed with the Bs ribosome, neither the elongation-arrested MifM′-tRNA nor its hydrolysis product MifM' was detectable evidently at any time of translation by the Ec ribosome (Fig. 3A). Because the source of the ribosome was the only difference between the Ec PURE system and the Bs hybrid PURE system, the deficiency in translational arrest of MifM in the Ec PURE system suggests that the MifM arrest sequence does not functionally interact with the E. coli ribosome.

Fig. 3.

Species specificity of the ribosome-stalling reactions. (A) In vitro translation of mifM with Ec PURE system. Ec PURE system was used to translate mifM-myc and translation products were analyzed as described in Fig. 2. (B) In vitro translation of secM by the E. coli and the B. subtilis ribosomes. secM-FLAG (lanes 1 and 2) or secM-P166A-FLAG (lane 3) was translated with either Bs hybrid (lane 1) or Ec (lanes 2 and 3) PURE systems at 37 °C for 30 min. Translation products were treated with RNaseA and then analyzed by NuPAGE and anti-SecM immunoblotting.

We next examined whether translation arrest of SecM can be reproduced using Bs hybrid PURE system. It was shown previously that in vitro SecM translation with Ec PURE system resulted in the marked elongation arrest, which was abolished completely by the Pro166Ala substitution in SecM (11). We confirmed the published nature of SecM using a C-terminally FLAG-tagged secM construct. The predominant translation product of the wild-type SecM-FLAG migrated faster than that of its Pro166Ala variant, after RNaseA treatment, in SDS/PAGE (Fig. 3B, lanes 2 and 3). In contrast, in vitro translation of secM-FLAG with Bs hybrid PURE system produced principally the full-length species, which migrated at the same position as the arrest-defective Pro166Ala variant (Fig. 3B, lane 1). Taken together with the failure of MifM to stall the Ec ribosome, it is suggested that the specific amino acids that mediate the nascent chain-ribosome interaction is not conserved even within bacteria. This contrasts with our finding that the B. subtilis ribosome interacts functionally with the translation factors of E. coli (Fig. 1).

Arrest Sequence as a Modular Unit Responding to Different Feedback Cues.

An important question about the generality and diversity of regulatory nascent chains is whether each arrest sequence is specialized to monitor a specific cellular event or can be combined with other sensing mechanisms. To address this question, we attempted to engineer MifM to sense activity of the Sec pathway. MifM contains a transmembrane sequence at its N-terminal region, which acts as a substrate of the primary YidC1 insertion/folding machinery (SpoIIIJ). The outcome of the MifM elongation arrest is the ribosomal stalling-induced exposure of the SD sequence of the secondary YidC protein (YidC2) of B. subtilis (5). Active membrane insertion of the TM region of nascent MifM results in the release of the elongation arrest and consequent down-regulation of yidC2 translation. By contrast, the absence of SpoIIIJ up-regulates yidC2 translation by increasing the duration of the arrest (5). These previously published features of the MifM-mediated regulation are illustrated in Fig. 4B (columns 1 and 2), in which lacZ that had been fused in-frame to the sixth codon of yidC2 was used as a reporter of the expression of yidC2.

Fig. 4.

Conversion of MifM from the insertion monitor into a secretion monitor in B. subtilis. (A) Schematic diagrams of MifM and BofC-MifM. TM and SS indicate the transmembrane segment of MifM and the engineered signal sequence derived from BofC, respectively. (B) Modes of yidC2 regulation by MifM and BofC-MifM. The levels of yidC2 expression in B. subtilis cells were assessed by measuring activities of β-galactosidase encoded by the yidC2⁶-lacZ indicator gene fusion placed downstream of mifM or bofC-mifM with the intact intergenic region. Host strains used were either wild-type (WT), spoIIIJ deletion (ΔspoIIIJ), or secG deletion (ΔsecG) strains of B. subtilis. To examine effects of SecA dysfunction on yidC2 expression, cells were treated with 0.5 mM sodium azide for 20 min at 37 °C (azide+) before the enzyme assay and compared with azide untreated cells (azide−). (C and D) Different protein localization pathways are monitored by MifM and BofC-MifM. Native MifM specifically monitors the YidC-dependent membrane protein biogenesis pathway (C), whereas BofC-MifM chimeric nascent chain monitors the SecAYEG-dependent protein secretion pathway (D).

We first tested if MifM also senses the general protein translocation or Sec pathway, by determining if blocking protein secretion increased yidC2 translation. Thus, we examined effects of NaN₃, an inhibitor of SecA, on the levels of the reporter β-galactosidase and found that it was ineffective (Fig. 4B, columns 5 and 6). Also, deletion of secG encoding a component of the SecYEG translocation channel (ΔsecG) did not affect the expression of lacZ (columns 9 and 10). These results show that MifM does not monitor protein secretion. In an attempt to convert MifM to a secretion monitor, we then replaced its N-terminal transmembrane region with the signal sequence of BofC, a secreted protein of B. subtilis (bofC-mifM; Fig. 4A). This modified chimeric BofC-MifM peptide was unable to sense YidC1 (SpoIIIJ) activity, because β-galactosidase activity remained low even in the absence of spoIIIJ (Fig. 4B, columns 3 and 4). However, when the Sec pathway was blocked by NaN₃, β-galactosidase under the control of bofC-mifM was induced more than twofold (Fig. 4B, columns 5 and 6). Also, deletion of secG up-regulated the β-galactosidase level about twofold. These results revealed that the chimeric BofC-MifM lost the ability to monitor the YidC pathway of membrane protein integration but instead became a functional protein secretion monitor in B. subtilis (Fig. 4D). This indicates that either the Sec or the YidC pathways are able to release the translational arrest conveyed by the C-terminal arrest motif, depending on the nature of the N-terminal signal sequence or transmembrane domain.

We conclude that the arrest sequence of MifM is not exclusively designed to respond to the YidC-mediated membrane-insertion/folding reaction as an arrest-releasing cue. Instead, it should be regarded as an independent arrest module that can be combined with different feedback mechanisms, such as protein secretion by the Sec machinery. Molecular context may be more important than the amino acid sequence itself to constitute a functional regulatory nascent chain. The lack of strict specificity seems to suggest that the physical force rather than specific signal transmission may underlie the mechanism responsible for the initiation of nascent chain-controlled release of elongation arrest (see Discussion).

Discussion

The discoveries of programmed ribosomal stalling are leading to an emerging new concept that the polypeptide exit tunnel of the ribosome is not simply a passive conduit but can interact with the newly synthesized segments of a nascent polypeptide to decide whether elongation should be continued, halted, or slowed down. Such nascent chain-exit tunnel interactions serve as a widespread and critical mechanism of gene regulation (1–6). In this mode of regulation, translation is accompanied by ongoing negotiation between the tunnel and the nascent polypeptide, which is in turn affected by either the nascent chain’s engagement in protein localization or concentration of an antibiotic or a metabolic compound. Because translation arrest could potentially be fatal if it continues over significant time period on multiple messenger RNAs, arrest sequences should have evolved to meet appropriate risk-benefit balances, for instance by combining with a releasing mechanism that senses specific physiological state of the cell.

While several stalling sequences are known in bacteria, they are quite divergent and it is difficult to find a common motif that is potentially important for the stalling functions. In this study, we focused on the two known members of the “intrinsic” class of regulatory nascent polypeptides (2), SecM and MifM, that are regulated by nascent chain dynamism rather than concentration of low-molecular weight substances/metabolites. SecM is conserved only in a subset of Gram-negative γ-proteobacteria (17), and mifM-like genes are only present in a subset of Gram-positive Bacillales (5), suggesting that they were recruited late in evolution to fulfill their regulatory roles. A central question here is whether the divergent stalling sequences exert their stalling functions through a unified mode of interaction with the ribosome or if each uniquely interacts with the ribosome.

Using the PURE system platform for our experiments, we here found that the E. coli translation factors and the B. subtilis ribosome can productively cooperate in protein synthesis in vitro. Taking advantage of this finding, we have reproduced the elongation arrest of MifM in vitro using the B. subtilis ribosome and the E. coli translation factors (the Bs hybrid PURE system). The use of the Ile70Ala mutant form of MifM as well as the rplV94 mutant form of the ribosome provided evidence that the ribosomal stall observed in the hybrid translation system reflected the physiological reaction. Interestingly, the elongation arrest was not reproduced with the E. coli ribosome. Thus, the arrest sequence of MifM is specifically recognized by the native B. subtilis ribosome but not by the ribosome of E. coli. Our results not only reinforce the previous conclusion obtained for SecM that its elongation arrest solely requires the translation apparatus but also extend it to suggest that the ribosome rather than other translation factors has the principal role in determining the species specificity of arrest sequences. The complementary finding that SecM was only able to stall the ribosome of E. coli as opposed to that of B. subtilis supports further the concept of species-specific arrest sequences (Fig. S1).

The species-specific interaction of the B. subtilis ribosome with the MifM arrest sequence is in sharp contrast to the more general use of the initiation and the elongation factors by the same ribosome. Probably the “upstream” decoding process of protein synthesis is universal, because it must have evolved in the early time of life to enable the concerted catalysis involving limited essential participants (two codons facing the P and A sites, two decoding tRNAs, and two amino acids at the ends of the tRNAs). By contrast, the downstream, postchemistry part of the protein synthesis machine might be more divergent among species to deal with the specific nascent proteome of a given organism, which itself has an overwhelming diversity of amino acid sequences. The differentiated exit tunnel of the ribosome sees all the polypeptide sequences of the cell during their synthesis, allowing evolution of polypeptide segments that can be sampled by the tunnel to cause physiologically relevant elongation arrest (Fig. S1).

While the MifM and the SecM stalling sequences may be regarded tailor-made to fit the respective ribosomes, there could be more general stalling sequences as well. For instance, regulatory peptides encoded by leader regions of genes for resistance determinants against some antibiotics, such as erythromycin and chloramphenicol, undergo antibiotic-induced elongation arrest both in Gram-positive and Gram-negative bacteria (3, 21, 22). Clusters of positively charged amino acids can cause arrest in translation in eukaryotic ribosomes (23). Such a distinct chemical feature as well as an antibiotic action might contribute to the general arrest functions in these cases. Our preliminary experiments suggest that an unnatural, laboratory-evolved stalling sequence (FXXYXIWPP; (24) arrests its translation in both Ec and Bs PURE systems (Fig. S2).

Are MifM and SecM, which are customized to fit the native ribosome, also customized to meet the specific regulatory need? Intriguingly, B. subtilis does not contain a SecA-regulating SecM and E. coli does not contain YidC-regulating MifM. Nevertheless, we were able to convert MifM to sense SecA-dependent protein secretion across the membrane rather than YidC1-dependent protein integration into the membrane. This was possible simply by replacing the N-terminal membrane-insertion sequence of MifM with a signal sequence from a secretory protein, while the C-terminal region including the arrest sequence was kept unaltered. Because the expression of lacZ placed after the chimeric bofC-mifM leader gene was induced when protein secretion pathway was impaired, it should be possible to construct an artificial secretion feedback mechanism by placing B. subtilis secA downstream of bofC-mifM. Our results have revealed that there is no causal link between the arrest sequence and the cellular event that releases the arrest. The arrest sequence seems to be a modular unit that has simply gained the ability to stall the ribosome. It acts in a species-specific manner presumably because the ribosomal tunnel had already been diverged. It is striking that the release of this species-specific arrest can be mediated by more than one pathway. Thus the species specificity of the arrest itself contrasts with the ability of different cellular pathways to release this arrest.

The mechanism by which elongation arrest of the regulatory nascent chain is released by secretion or membrane integration activity of the cell is a major unsolved question. Two possibilities have been proposed (6). In the physical/pulling model, physical force generated by the protein localization reaction triggers the release reaction (25). In the signal transmission model, interaction of the ribosome-nascent chain complex with the protein delivery machinery induces intraribosomal signal transmission to affect the peptidyl transferase center of the ribosome. Our present results seem to disfavor a simple signal transmission model, because two different protein delivery machineries, YidC1 and SecAYEG, proved equally effective in triggering the arrest release. Although it is possible that both channels trigger similar events within the ribosome, this lack of specificity is more consistent with the physical pulling force that triggers the release. The structure of the nascent TnaC peptide within the ribosome shows that it is contacting with multiple points on the tunnel wall (26), although the sequence divergence suggests that different arrest peptides will assume different configurations in the tunnel (27). It is conceivable that a pulling force can peel off the peptide from one or more of the peptide-tunnel contact sites and this leads to an altered arrangement of the ribosomal active site relative to the growing end of the peptide.

Although SecM and MifM are the only reported examples of nascent chain sensors that monitor cotranslational cellular events, any biological event that generates a pulling force could be combined with an arrest-inducing module to affect its stalling ability and thereby constitute a regulatory loop. Such ad hoc regulatory systems might have evolved after a proteome of a group of species has been nearly established, adding a benefit to the organism under changing living conditions.

Materials and Methods

Bacterial Strains and Plasmids.

Strains of B. subtilis and plasmids used in this study are described in SI Text.

β-Galactosidase Activity Assay.

B. subtilis cells were cultured at 37 °C in LB medium for β-galactosidase assay. Culture of OD600 = 0.5–1.0 was collected for each experiment. β-galactosidase assays were performed as described earlier (14, 28).

Purification of the Ribosomes.

Purification of E. coli ribosome was done as described (29). B. subtilis ribosome was purified as follows. B. subtilis strain PY79 (wild type) or SCB625 (rplV94) was grown at 37 °C in LB medium until OD600 reached 0.5. Cells were harvested by centrifugation, washed twice with buffer I-H (10 mM HEPES-KOH pH 7.6, 15 mM Mg(AOc)₂, 1 M KCl, 5 mM EDTA, 10 mM β-mercaptoethanol), then once with buffer II-H (Buffer I-H with 50 mM KCl) and stored as a pellet at -80 °C. Cells were thawed in suspension buffer (10 mM HEPES-KOH pH7.6, 50 mM KCl, 10 mM Mg(OAc)₂, 7 mM β-mercaptoethanol) with protease inhibitors (1 mg/ml PefaBlock, 10 μg/ml Leupeptin, 0.7 μg/ml Pepstatin) and passed through a cell of French press (Aminco) at 16,000 psi. Cell lysate was mixed with an equal volume of suspension buffer with 3 M ammonium sulfate and kept on ice for 30 min. Cell debris and protein precipitates were removed by centrifugation (4 °C, 10,000 g, for 30 min). The supernatant was subjected to HiTrap Butyl FF column of 10 ml volume (GE) equilibrated with buffer A (20 mM HEPES-KOH pH 7.6, 1.5 M (NH₄)₂SO₄, 10 mM Mg(OAc)₂, 7 mM β-mercaptoethanol). The column was then washed with 40 ml of buffer B (20 mM HEPES-KOH pH 7.6, 10 mM Mg(OAc)₂, 7 mM β-mercaptoethanol) containing 1.2 M ammonium sulfate and proteins were eluted with a 1.2 M to 0 M gradient of ammonium sulfate in buffer B. Fractions containing the ribosome were combined and loaded on the top of an equal volume of 30% sucrose cushion buffer (20 mM HEPES-KOH pH 7.6, 10 mM Mg(OAc)₂, 30 mM NH₄Cl, 30% sucrose, 7 mM β-mercaptoethanol) and ultracentrifuged (4 °C, 100,000  × g for 16 h) to sediment the ribosomes. The pellet was suspended with ribosome buffer (20 mM HEPES-KOH pH 7.6, 30 mM KCl, 6 mM Mg(OAc)₂, 7 mM β-mercaptoethanol) and stored at -80 °C until use for in vitro translation assays.

In Vitro Translation.

The coupled transcription–translation system with all purified components (PURE system) was used for in vitro translation as described previously (16). While the original reaction mixture containing the E. coli ribosome is referred to as Ec PURE system, Bs hybrid PURE system contained B. subtilis ribosome in place of the E. coli ribosome. The ribosome was added to the reaction mixture at a final concentration of 1 μM. Reaction was primed with appropriate DNA fragment prepared by PCR and continued at 37 °C. Samples were withdrawn at interval and mixed with SDS/PAGE loading buffer. When indicated they were treated with 0.2 mg/ml RNaseA (Promega) at 37 °C for 20 min. To quantify GFP synthesized in vitro, the samples were heated in SDS sample buffer at 80 °C for 5 min, separated by 12.5% conventional Laemmli gel, and then incubated in the Western blotting transfer buffer (25 mM Tris, 1.92 M glycine and 20% methanol) for 30 min at room temperature for in-gel renaturation. GFP fluorescence was elicited at 473 nm, passed through a 520 nm long pass filter, and recorded using a FLA9000 image analyzer (GE healthcare) with optimized voltages (450–850 V) of the photomultiplier tube. Electrophoretic separation of MifM, SecM and their derivatives was carried out by NuPAGE gel system and MES buffer (Invitrogen) and proteins were detected by Western blotting using anti-MifM, anti-SecM (7), or other antisera. PCR products used as the templates of in vitro transcription-translation reaction are described in SI Text.