Escherichia coli small heat shock protein IbpA is an aggregation‐sensor that self‐regulates its own expression at posttranscriptional levels

Aggregation is an inherent characteristic of proteins. Risk management strategies to reduce aggregation are critical for cells to survive upon stresses that induce aggregation. Cells cope with protein aggregation by utilizing a variety of chaperones, as exemplified by heat‐shock proteins (Hsps). The heat stress‐induced expression of IbpA and IbpB, small Hsps in Escherichia coli, is regulated by the σ32 heat‐shock transcriptional regulator and the temperature‐dependent translational regulation via mRNA heat fluctuation. We found that, even without heat stress, either the expression of aggregation‐prone proteins or the ibpA gene deletion profoundly increases the expression of IbpA. Combined with other evidence, we propose novel mechanisms for the regulation of the small Hsps expression. Oligomeric IbpA self‐represses the ibpA/ibpB translation, and mediates its own mRNA degradation, but the self‐repression is relieved by sequestration of IbpA into the protein aggregates. Thus, the function of IbpA as a chaperone to form co‐aggregates is harnessed as an aggregation sensor to tightly regulate the IbpA level. Since the excessive preemptive supply of IbpA in advance of stress is harmful, the prodigious and rapid expression of IbpA/IbpB on demand is necessary for IbpA to function as a first line of defense against acute protein aggregation.


| INTRODUC TI ON
Since proteins tend to form aggregates, cellular maintenance by keeping proteins in their native states and removing denatured proteins is crucial for all organisms. Multilayered quality control systems are essential to maintain such cellular protein homeostasis (proteostasis) (Hipp et al., 2019;Mogk et al., 2018). Refolding and degradation of denatured proteins caused by stresses are two primary strategies to prevent the accumulation of protein aggregates. Sequestration of denatured proteins is a third strategy, to keep misfolded proteins in a state that is easy to restore or degrade after stresses (Hipp et al., 2019;Mogk et al., 2018). Small heat shock proteins (sHsps) participate in the third strategy as "sequestrases," constituting a first line of stress defense against irreversible protein aggregation (Haslbeck and Vierling, 2015;Haslbeck et al., 2019;Liberek et al., 2008;Mogk et al., 2019).
However, σ 32 is stabilized to allow the transcription of many Hsp genes upon heat shock, since DnaK/J is sequestered to rescue the emerging heat-denatured proteins (Guisbert et al., 2008;Guo and Gross, 2014). The other is the thermoresponsive mRNA structures in the 5′ untranslated region (UTR), called RNA thermometers (RNATs), which mask the Shine-Dalgarno (SD) sequence in their stem-loop structures at normal or low temperatures (Kortmann and Narberhaus, 2012;Krajewski and Narberhaus, 2014). The heat fluctuation by a temperature up-shift melts the stem loops in RNATs and allows the ribosome to initiate translation, using the exposed SD sequence (Kortmann and Narberhaus, 2012;Krajewski and Narberhaus, 2014). The thermo-responsivity of many bacterial RNATs has been established, and the RNATs of sHsps have conserved shapes harboring two to four stem loops (Kortmann and Narberhaus, 2012;Krajewski and Narberhaus, 2014). Thus, the expression of sHsps is controlled at both the transcriptional level using heat-shock transcriptional factors, and the translational level using RNATs, in contrast with other Hsps, which are only controlled at transcriptional levels (Kortmann and Narberhaus, 2012;Krajewski and Narberhaus, 2014).
The mRNA encoding the ibpA-ibpB operon has RNATs in the 5′ UTRs of both the ibpA and ibpB ORFs, as revealed by RNA structure probing and reporter assays (Gaubig et al., 2011;Kortmann and Narberhaus, 2012;Waldminghaus et al., 2009). Previous analyses of the RNAT in ibpB using a reporter revealed the possible influence of the IbpA protein on ibpB expression (Gaubig et al., 2011;Kortmann and Narberhaus, 2012). In addition to the heat stress, the expression level of IbpA/IbpB was profoundly upregulated, by 10 ~ 50-fold, under non-heat stressed conditions such as 30 ~ 37°C in the dnaK-dnaJ deleted strain (Calloni et al., 2012;Zhao et al., 2019) or upon oxidative stress induced by copper (Matuszewska et al., 2008). Since the RNAT regulation would not be effective at normal growth temperatures, the mechanism for the massive upregulation under the non-heat stressed conditions remains to be elucidated.
Here, we addressed why IbpA is upregulated in non-heat stressed cells. We found that the accumulation of protein aggregates was sufficient for the upregulation. Intriguingly, a reporter assay using the 5′ UTR of the ibpA mRNA revealed that the deletion of the ibpA gene increased the reporter translation, which was repressed by the overexpression of oligomeric IbpA. Combined with other evidence including the requirement of oligomerization of IbpA on the repression and the specific interaction of IbpA with 5′ UTR of ibpA mRNA, we propose that the IbpA oligomers self-repress the ibpA translation and are involved in its own mRNA degradation, which are relieved by the sequestration of IbpA by co-aggregation with protein aggregates. The role of the aggregation sensor is specific to IbpA, since the homologous IbpB lacks this self-repression function. The significance of the self-repression by IbpA at the translational level is discussed in relation to the unique role of IbpA in protecting cells from acute heat stress.

| ibpA translation is upregulated in response to protein aggregation
Although the thermometer in the mRNA (RNAT) and the transcriptional control by σ 32 are known mechanisms to upregulate the expression of IbpA, previous studies have reported that IbpA expression is also upregulated under non-heat stressed conditions, such as in the dnaKJ deletion strain or upon copper stress (Calloni et al., 2012;Matuszewska et al., 2008;Zhao et al., 2019). The absence of DnaK/DnaJ leads to the production of protein aggregates (Calloni et al., 2012;Mogk et al., 1999). The addition of copper disturbs protein homeostasis in cells with oxidative stress (Matuszewska et al., 2008;Yang et al., 2015). A common consequence in E. coli cells would be the accumulation of protein aggregates (Calloni et al., 2012;Matuszewska et al., 2008;Mogk et al., 1999;Yang et al., 2015). Therefore, we hypothesized that protein aggregation might somehow be involved in the upregulation of IbpA under the nonheat stressed conditions.
To investigate whether the expression of IbpA is upregulated not only by heat shock but also by protein aggregation under non-heat stress conditions, we expressed aggregation-prone proteins in a wild-type E. coli (BW25113). To do so, we overexpressed rhodanese, a bovine mitochondrial protein, which is known to aggregate in E. coli at 37°C (Ewalt et al., 1997) ( Figure S1A). Strikingly, the expression of IbpA increased upon the rhodanese expression (agg ++ , Figure 1A).
Overexpression of another aggregation-prone protein, SerA of E. coli (Mogk et al., 1999), ( Figure S1A), also massively induced the IbpA expression in the wild-type E. coli strain ( Figure S1B). The rhodanese expression did not increase the levels of DnaK and GroEL, two of the representative Hsps in E. coli ( Figure S1C). The rhodanese expression did not affect the IbpA degradation rate in cells ( Figure S1D), suggesting that the increased amount of IbpA is not due to an inhibition of IbpA degradation, at least partly mediated by Lon protease. These results support the idea that the accumulation of aggregated proteins in cells increases the IbpA expression.
We suspected that the aggregates might induce IbpA via upregulated σ 32 -mediated transcriptional control at 37°C, since the aggregation-prone proteins could sequester DnaK/J, thus, protecting σ 32 from DnaK/J-mediated degradation and eventually stabilizing σ 32 to promote the expression of Hsps. If so, then, the overexpression of σ 32 would increase the mRNAs to upregulate IbpA as well as other Hsps. After we confirmed the ~10-fold induction of ibpA mRNAs in the σ 32 -overexpressing cells, using quantitative real-time PCR (qRT-PCR) ( Figure 1B), we compared the protein expression levels. Upon the σ 32 overexpression, the expression level of IbpA did not obviously increase ( Figure 1A), even though the ibpA mRNA increased.
In contrast, the DnaK and GroEL expression levels increased ( Figure S1C), confirming that the excess σ 32 is effective in increasing the Hsps under the non-heat stress. The results indicate that the IbpA induction in the presence of aggregates is not explained by the σ 32 -mediated transcriptional upregulation.
To demonstrate that the ibpA upregulation by the accumulation of protein aggregates does not occur at the transcriptional level, we investigated the efficiency of the ibpA translational initiation in a reporter assay. Since the 5′ UTR in the ibpA mRNA is critical for the translational control using the stem loops in the mRNA, we constructed a plasmid harboring the 5′ UTR of ibpA fused with the gfp gene, under the control of an arabinose-inducible promoter ( Figure 1C). We observed a substantial induction of GFP upon the rhodanese overexpression ( Figure 1C). The effect is specific for the 5′ UTR of ibpA, since there was no upregulation of GFP upon the rhodanese overexpression when the 5′ UTR was substituted with 5′ UTRs derived from the parent plasmid ( Figure 1C) or dnaK ( Figure S1E). In addition, the overexpression of another aggregation-prone protein, SerA, also induced the GFP reporter with the 5′ UTR of ibpA ( Figure S1F). Combined with the results that the F I G U R E 1 Accumulation of protein aggregates upregulates ibpA expression. (a) Western blotting to evaluate the endogenous ibpA expression in E. coli wild-type strain (BW25113) under various conditions. E. coli cells were grown at 37°C or shifted to 42°C for 10 min. agg ++ , E. coli wild-type BW25113 expressing rhodanese; σ 32++ , E. coli wild-type BW25113 expressing σ 32 . Unless otherwise indicated, E. coli cells were grown at 37°C. Expression of FtsZ was also examined as a control for a constitutively expressed protein.
(b) Ratios of the ibpA mRNA amounts in cells under conditions corresponding to (a). Error bars represent SD; n = 3 biological replicates. Student's t test was used to assess the statistical significance of differences (*p < .01). (c) Evaluation of ibpA translation by GFP reporters. The reporters harboring the 5′ UTR of ibpA or the 5′ UTR from a plasmid without the ibpA sequence were expressed under various conditions. Western blotting was performed using anti-GFP and anti-FtsZ antibodies [Colour figure can be viewed at wileyonlinelibrary.com]

| IbpA self-represses IbpA expression level
Next, we investigated the connection between protein aggregation and IbpA expression induction. The well-known physiological function of IbpA as a chaperone is the co-aggregation with denatured  (Govers et al., 2018;Lindner et al., 2008;Pu et al., 2019). Taking this into consideration, we hypothesized that the entrapment of free IbpA from the cytosol, due to the sequestration of denatured proteins, would induce the ibpA translation ( Figure 2B). If this model is correct, then, the deletion of IbpA, which is nonessential for E. coli growth, would F I G U R E 2 IbpA self-represses ibpA expression. (a) Localization of IbpA-GFP in E. coli cells. Top. The IbpA-GFP reporter construct is schematically shown. Bottom. Bright-field and fluorescence images of IbpA-GFP in the absence (-) or presence (agg ++ ) of induced rhodanese in wild-type strain (BW25113). (b) Titration model of the IbpA-mediated negative feedback mechanism. IbpA (circles) suppresses its own mRNA translation. Upon aggregate (dark blobs) formation, the IbpA co-aggregated with aggregation-prone proteins is sequestered at the cell poles, relieving the translation suppression. (c) Evaluation of the ibpA expression by the GFP reporter used in Figure 1C. E. coli lysates from wild-type BW25113 strain (WT) or the ibpAB operon-deleted strain (∆AB) with (+) or without (-) the co-expression of rhodanese (agg ++ ) were analyzed. Western blotting using anti-GFP and anti-FtsZ antibodies is shown. (D) Evaluation of the ibpA expression by the reporters used in Figure 1C. IbpA ++ , IbpA was induced by 0.1 mM of IPTG for 1 h. Western blotting with anti-GFP or anti-FtsZ antibodies is shown [Colour figure can be viewed at wileyonlinelibrary.com] upregulate the translation of the reporter harboring the 5′ UTR of ibpA, used in Figure 1C.
We deleted the operon including ibpA-ibpB. After we confirmed that the growth of the ∆ibpAB cells was similar to that of the wildtype E. coli BW25113 ( Figure

| mRNA degradation is partially involved in the IbpA expression regulation
qRT-PCR analysis revealed that the mRNA level of the reporter harboring the ibpA 5′ UTR in the ∆ibpAB cells was ~6-fold higher than that in wild-type cells ( Figure 3A). Kinetic analysis of the mRNA degradation showed that the higher abundance was attributed to the stabilization of the mRNA in the ∆ibpAB cells ( Figure 3B). The stabilization was reverted by the overexpression of IbpA ( Figure 3B), suggesting that IbpA might be involved in the mRNA turnover to regulate its own expression. We also found that the aggregation in- suggesting that the effect of PNPase is not a major mechanism to regulate the IbpA expression.

| IbpA directly suppresses its own translation
In addition to the involvement of IbpA to its own mRNA degradation, we tested whether IbpA affects its own translation in a direct approach. We translated the GFP reporter harboring the 5′ UTR of ibpA using a reconstituted cell-free translation system of E. coli (the PURE system), which only contains essential factors for the translation, where there is no chaperone including IbpA (Shimizu et al., 2001). Strikingly, the ibpA translation was repressed by almost half in the presence of recombinant IbpA ( Figure 4A). The repression by the recombinant IbpA was not observed in the control reporter without 5′ UTR of ibpA ( Figure 4A). The effect was specific to IbpA since the addition of purified GFP instead of IbpA did not repress the reporter translation ( Figure S4).
Since IbpA specifically repressed the translation of the reporter gene with the 5′ UTR of ibpA in the reconstituted cell-free translation, we anticipated that the self-repression of the ibpA might be through a direct recognition of IbpA with the 5′ UTR of ibpA mRNA.
We next performed an electrophoretic mobility shift assay (

| Oligomeric IbpA is critical for the selfrepression of translation
What region of IbpA is critical for the self-regulation? At first, we deleted the N-or C-terminal domain of IbpA and examined the effect on expression repression ( Figure 5A). The GFP reporter assay revealed that the C-terminal truncation eliminated the ability to suppress the expression in the ∆ibpAB cells ( Figure 5A), although the expression level of the C-terminal truncated mutant was comparable to that of wild-type IbpA ( Figure S5A). In contrast, the suppression by the N-terminal truncation was almost the same as that by wild-type IbpA ( Figure 5A), showing that the C-terminal domain is responsible for the self-repression.
Although the expression level of the IbpA (AEA) mutant was almost the same in the ∆ibpAB cells, co-expression of the AEA mutant tion. Therefore, we tested the effect of IbpB on the ibpB translation, after we replaced the 5′ UTR of ibpA with that of ibpB in the GFP reporter system. Although the induction level of IbpA and IbpB were almost the same ( Figure S6), IbpA, but not IbpB, suppressed the reporter for the ibpB expression in the ∆ibpAB strain ( Figure 6A). The translation of the ibpB reporter was also repressed by the recombinant IbpA in the PURE system ( Figure 6B).
These results show the specific function of IbpA as a repressor of the small Hsps, IbpA and IbpB, in E. coli.
In contrast, the overexpression of IbpB did not change the ibpA reporter upregulation in the ∆ibpAB strain ( Figure 6C, lane 4)

(b)
increased the amount of the GFP reporter ( Figure 6C, lane 3), probably reflecting the hetero-oligomerization of IbpB with endogenous IbpA to reduce the amount of free IbpA for the self-repression.

| Stem loops in the 5′ UTR of ibpA mRNA mediate the self-repression of ibpA expression
Previous studies revealed that the secondary mRNA structures of the ibpA 5′ UTR regulate the ibpA translation (Waldminghaus et al., 2009).
The ibpA 5′ UTR contains three stem loops, and the two upstream stem The expression patterns of the reporters using the strong SL1 and SL2 stem loops were almost the same as those using the weak variants ( Figure 7A,B). In contrast, the expression of the reporter using the strong SL3 was not observed under all conditions examined ( Figure 7C), probably because the strong stem loop containing the SD motif is too tight to expose the SD motif for the translation initiation. Translation repression by a direct binding of IbpA is unexpected, since so far there is no known RNA-binding motif in IbpA. In addition to the translation repression, we found that the increased IbpA expression is caused by a relief of ibpA mRNA degradation mediated by PNPase and IbpA. Although the involvement of PNPase in the ibpA mRNA degradation was previously reported (Carpousis, 2007), our finding adds IbpA as a critical player in this mRNA degradation pathway for the self-regulation. It has been known that PNPase usually functions in a second step in the RNA degradosome pathway after the first step mediated by the RNaseE. However, it is also reported that PNPase participates in post-transcriptional regulations depending on 5´ UTR (Chen et al., 2019;Jarrige et al., 2001). In this manner, it would be plausible that PNPase would participate in the repression of the ibpA translation. How the interaction between IbpA and PNPase, as reported previously (Applied et al., 2005), regulates the ibpA mRNA degradation via the 5´ UTR of ibpA is not known, but would be worth pursuing in the future.
Thus, we revealed two mechanisms for the IbpA-mediated self-regulation. Then, what is the relation between the translation suppression and stimulated mRNA degradation? One immediate explanation would be a synergistic effect to complement each other.
Alternatively, these two mechanisms might be not necessarily mutually exclusive. Translation suppression by direct binding of IbpA to its own mRNA might be just an intermediate step to deliver the mRNA to the degradation pathway.
This model of IbpA expression regulation resembles the σ 32 -mediated transcriptional regulation of Hsps (Guisbert et al., 2008;Guo and Gross, 2014), since the chaperones are titrated away by denatured F I G U R E 6 IbpB cannot substitute for IbpA as the sHsp expression suppressor. (a) Evaluation of the ibpB expression by IbpA or IbpB. The 5′ UTR of ibpA in the GFP reporter used in Figure 1C was replaced with the 5′ UTR of ibpB. The modified GFP reporter was expressed in E. coli BW25113 wild-type strain (WT) or the ibpAB-deleted strain (∆AB) co-expressing either IbpA or IbpB. (b) Cell-free translation in the absence (−) or the presence of purified IbpA. Upper, translation was evaluated by fluorescence detection. Lower, relative translation ratio quantified by fluorescence intensity of the bands. Error bars correspond to the SD. The Student's t test was used to assess the statistical significance of differences (*p < .01). (c) Effect of IbpB overexpression on the ibpA expression evaluated by the GFP reporter used in Figure 1C. Previous studies revealed the layered regulation of IbpA expression: σ 32mediated transcriptional control and RNA thermometer (RNAT) translational control (Gaubig et al., 2011;Kortmann and Narberhaus, 2012). Since the stem loops in the RNAT system influence the IbpA-mediated expression repression (Figure 7), RNAT and the F I G U R E 7 Effect of stem loops in the ibpA mRNA on the IbpA-mediated suppression. The stem loops in the ibpA 5′ UTR of the GFP reporter used in Figure 1C were mutated and evaluated in the wild-type BW25113 (WT), and ∆ibpAB (∆AB) strains. Mutations to weaken or strengthen the stem-loop structures were introduced in SL1 (a), SL2 (b), and SL3 (c). Schematic representations of stem-loop mutations are shown (see also Figure S7). Where indicated, IbpA was overexpressed (+). The ibpA 5′ UTRs with no (WT), weak and strong mutations in the stem loops were tested using the GFP reporter.
self-repression control are not independent. In the RNAT mechanism, the stem loops fluctuate and melt to expose the SD region, depending on the temperature. As reported previously, higher temperatures cause more melting of the stem loops, like a "thermometer." In other words, the temperature responsiveness is not an all-or-none fashion (Kortmann and Narberhaus, 2012;Krajewski and Narberhaus, 2014).
This inherent property would allow the stem loops to partly open even under mild conditions such as 37°C (Waldminghaus et al., 2009), leading to a potential leaky expression of certain amounts of IbpA. Thus, the IbpA-mediated self-repression would function to tightly shut off the IbpA expression as a "safety catch" in the leaky RNAT system.
Taken together, the stringent repression mechanism, combining RNAT and the IbpA-mediated negative feedback control, has evolved to fulfill the following requirements: tight repression under unstressed conditions, and acute upregulation upon aggregation-stress. This mechanism enables IbpA to be one of the most upregulated chaperones upon aggregation inducing stresses.
IbpA serves as a first line of defense against protein aggregation, where oligomerized IbpA co-aggregates with aggregation-prone proteins for sequestration. Why does IbpA employ such a feedback control mechanism in addition to the known regulation controls including σ 32 and RNATs? Considering that IbpA is an ATP-independent oligomeric chaperone, a greater than stoichiometric amount of IbpA would be necessary to sequester the aggregation-prone proteins. One strategy for risk management is to prepare an abundance of IbpA protein even under unstressed conditions. However, this might not be appropriate, since IbpA overexpression had detrimental effects under the normal conditions ( Figure S2B), probably due to the self-formation of fibril aggregates (Ratajczak et al., 2010), which could perturb proteostasis and compromise the sequestration activity. Thus, the expression of IbpA should be tightly repressed under normal conditions, since IbpA can be regarded as a "double-edged sword." The self-regulation mechanism proposed here can overcome the dilemma that the high abundance of IbpA is necessary in cases of aggregation stress, but an excessive preemptive supply could be detrimental to the cell.
Our analysis of the stem loops in the 5′ UTR of the ibpA mRNA revealed that the secondary structures of the mRNA are critical to regulate the expression. How does IbpA couple to the mRNA structure in RNAT for the suppression? One possibility is that the oligomeric states of IbpA bind to their own mRNA to suppress the expression. E. coli has two small Hsps, IbpA and its highly homologous paralog IbpB, which are encoded in the ibpAB operon (Allen et al., 1992).
Several lines of evidence have shown the distinct roles of IbpA and IbpB, where IbpA and IbpB function as a canonical binder and its noncanonical paralog that enhances the dissociation of sHsps from the co-aggregates, respectively (Obuchowski et al., 2019;Ratajczak et al., 2009). In addition to this distinction, our findings demonstrated another aspect of the difference between the two sHsps in the expression regulation. IbpA suppressed the ibpB reporter translation in the ∆ibpAB strain ( Figure 6) (Gaubig et al., 2011). In contrast, IbpB could not suppress the ibpA translation. Thus, IbpA plays a pivotal role as a master regulator of the expression of sHsps at the posttranscriptional levels, ensuring that IbpA and IbpB cooperate to cope with protein aggregation.
The overexpression of IbpB in the wild-type strain increased the IbpA level ( Figure 6). This IbpA induction is interpreted to be due to the IbpA deprivation by hetero-oligomer formation between IbpA and IbpB, implying that, in addition to the aggregation-prone proteins, the factors that can associate with IbpA could trigger its upregulation. Therefore, we suggest the possibility that IbpA plays a pivotal role as a trans-regulator for the expression of other proteins. Indeed, the translation level of ibpB is decreased upon IbpA co-expression (Gaubig et al., 2011). The fact that the ibpB mRNA also has an RNAT in the 5′ UTR (Gaubig et al., 2011;Waldminghaus et al., 2009) implies that IbpA recognizes a particular structure of stem loops in 5′ UTRs, such as RNAT.

| Quantitative RT-PCR
Primers used for PCR are listed in Table S2.

| Microscopy
To observe the IbpA localization in cells, we used the E. coli BW25113 wild-type strain carrying pCA24N-rhodanese and the dnaKJ deletion strain. Each strain carrying pBAD30-ibpA 5′ UTR-ibpA-gfp was grown to an OD 660 of ~0.4 at 37°C in LB medium. Cells were observed with an inverted microscope IX71 (Olympus) and a mercury lamp with a GFP filter. Fluorescent images were recorded with an iXon DV897 electron multiplying CCD camera (Andor) and the Andor SOLIS software (Andor).

| Protein purification
To purify IbpA, we used the BL21 (DE3) strain carrying pCA24N-ibpA or pCA24N-ibpA_AEA. Cells were grown in LB media at 37°C to an OD 660 of 1.0, and IbpA production was induced with 1 mM IPTG for 2 h. The cells overexpressing IbpA were lysed by sonication (Branson) in buffer A (50 mM of Tris-HCl, pH 7.4, 10% of glycerol, 1 mM of dithiothreitol, and 100 mM of KCl). After the sonication, we followed the established methods using anion exchange chromatography for the purification of the wild-type IbpA (Matuszewska et al., 2005) and the AEA mutant (Strózecka et al., 2012).

| Sucrose density gradient assay
The purified IbpA (10 µM) was incubated for 30 min at 48°C in buffer B (50 mM of Tris-HCl, pH 7.4, 150 of mM KCl, 20 mM of magnesium acetate, and 5 mM of dithiothreitol). After the incubation, the samples were applied onto an 11 ml gradient of 10%-50% (w/v) sucrose in buffer B and centrifuged, using a Beckman SW41Ti rotor at 150,000 × g at 4°C for 80 min. The samples were collected as 20 separate fractions, using a fractionator (BioComp). The fractions were separated by SDS-PAGE and visualized by Coomassie Brilliant Blue staining.

| Cell-free translation
The transcription-translation-coupled PURE system (PUREfrex®, GeneFrontier) reaction, including Cy5-labeled tRNA fMet , was performed at 37°C for 2 h in the presence or the absence of 1 µM IbpA.
After the protein synthesis, the SDS-sample buffer (

| Electrophoretic mobility shift assay
The reporter mRNA harboring 5´ UTR of ibpA vector was synthesized by in vitro transcription with CUGA7 in vitro transcription kit (Nippon Gene), and purified with RNeasy Mini kit (Qiagen). The mRNA at 100 nM and purified IbpA were incubated for 30 min at 37°C. The samples were loaded on 1 × TBE 5% of acrylamide gel.
The samples were electrophoresed for 80 min at 200 V. After the electrophoresis, the gel was immersed in 1 × TBE with 1/50000 SYBR Gold (Thermo Fisher) for 5 min. Fluorescence was detected with a fluorescence imager IN-6W-CAM (Natural Immunity).

| Growth assay
The E. coli wild-type strain (BW25113), the ibpAB deleted strain, and the wild-type strain harboring pCA24N-ibpA or pCA24N-gfp were precultured at 30°C for 16 h in LB medium. The precultured cells were incubated at 37°C with shaking at 70 rpm, using a TVS062CA incubator (Advantec).

| Statistical analysis
Student's t test was used for calculating statistical significance, with a two-tailed distribution with unequal variance. All experiments represent a minimum of three independent experiments, with the bars showing the mean values ± SD.

ACK N OWLED G M ENTS
We thank Tatsuya Niwa for valuable discussions, Eri Uemura for