Escherichia coli S2P family intramembrane protease RseP is engaged in the regulated sequential cleavages of FecR in the ferric citrate signaling

Escherichia coli RseP, a member of the S2P family of intramembrane proteases, is involved in the activation of the σE extracytoplasmic stress response and elimination of remnant signal peptides. However, whether RseP has additional cellular functions is unclear. In this study, we attempted to identify new RseP substrates to explore still unknown physiological roles of this protease. Our mass spectrometry-based quantitative proteomic analysis revealed that the levels of several Fec system proteins encoded by the fecABCDE operon (fec operon) were significantly decreased in an RseP-deficient strain. The Fec system is responsible for the uptake of ferric citrate, and the transcription of the fec operon is controlled by FecI, an alternative sigma factor, and its regulator FecR, a single-pass transmembrane protein. Assays with the fec operon expression reporter demonstrated that the proteolytic activity of RseP is essential for the ferric citrate-dependent upregulation of the fec operon. Analysis using the FecR protein and FecR-derived model proteins showed that FecR undergoes sequential processing at the membrane and that RseP participates in the last step of this sequential processing to generate the N-terminal cytoplasmic fragment of FecR that participates in the transcription of the fec operon with FecI. Ferric citrate signal-dependent generation of this cleavage product is the essential and sufficient role of RseP in the transcriptional activation of the fec operon. Our study unveiled that E. coli RseP performs the intramembrane proteolysis of FecR, a novel physiological role that is essential for regulating iron uptake by the ferric citrate transport system.


Introduction
While bacterial cellular membranes act as a barrier to protect a cell from extrinsic damages caused by various xenobiotics and hazardous changes in environmental conditions, they must mediate not only selective import of nutrients and other small molecules but also transduction of signals from the external milieu to adapt to environmental changes. A variety of mechanisms exist to transmit information across the membrane. Among them, regulated intramembrane proteolysis (RIP) is a crucial mechanism conserved among all kingdoms (1,2). In RIP, a class of membrane proteases called intramembrane proteases (IMPs) mediate transmembrane signaling through the cleavage of target membrane proteins. IMPs are unique membrane-integrated proteases in that they have their proteolytic active site located within the lipid bilayer and catalyze proteolysis in the membrane. IMPs are classified into four families: site-2 protease (S2P; zinc metallopeptidase), rhomboid protease (serine protease), presenilin/signal peptide peptidase (SPP; aspartyl protease), and Rce1 (glutamyl protease) (3,4). IMPs cleave various substrates and thereby play diverse cellular roles including stress responses, development of Alzheimer's disease, induction of apoptosis, maintenance of mitochondrial homeostasis, invasion of apicomplexan parasites, and quality control of membrane proteins and bacterial pathogenicity (5)(6)(7)(8).
Escherichia coli RseP is one of the most well-studied members of the S2P family proteases.
RseP was first identified as a key factor that regulates the σ E extracytoplasmic stress response through the cleavage of a single membrane-spanning anti-σ E protein, RseA (9)(10)(11). In the σ E extracytoplasmic stress response, the accumulation of misfolded outer membrane proteins and lipopolysaccharide biosynthesis intermediates in the cell envelope acts as stress cues to induce the cleavage of RseA by DegS, a membrane-anchored serine protease, on the periplasmic side (site-1 cleavage). This first cleavage triggers the following RseP-catalyzed second cleavage of RseA inside the membrane (site-2 cleavage), leading to liberation from the membrane of the RseA cytoplasmic domain fragment complexed with σ E . Finally, degradation of the RseA cytoplasmic domain fragment by cytoplasmic proteases such as ClpXP activates σ E to induce the transcription of stress genes (12,13). The first cleavage of RseA by DegS is a prerequisite for the subsequent site-2 cleavage by RseP. Similarly, most of the other S2P family proteases are known to catalyze intramembrane proteolysis of a target protein only after preceding trimming of the substrate on the extracytoplasmic side by other protease(s). E. coli RseP has also been shown to eliminate remnant signal peptides generated during the membrane translocation of presecretory proteins that would contribute to the quality control of the cytoplasmic membrane (14). RseP homologs (or S2P family proteases) of many gram-negative and gram-positive bacteria are involved in various cellular processes (7). They include the production of sex pheromones in Enterococcus faecalis (15), sporulation in Bacillus subtilis (16), acid response in Salmonella enterica (17), mucoid conversion and alginate overproduction in Pseudomonas aeruginosa (18), production of cholera toxin in Vibrio cholerae (19), and iron acquisition in Bordetella bronchiseptica and P. aeruginosa (20, 21), RseP spans the membrane 4 times with both of its N-and C-termini facing the periplasmic space. The residues of the first and third transmembrane segments constitute the intramembrane proteolytic active site (22, 23), and the central periplasmic region between the second and the third transmembrane segments contains tandemly-arranged two PDZ domains (PDZ tandem) (24,25) and an amphiphilic helix that is presumably involved in the proper positioning of the PDZ tandem and a substrate (26). RseP and other S2P proteases generally cleave a single-spanning membrane protein with type II (NIN-COUT) topology. We previously proposed a model that the bulky PDZ tandem acts as a size-exclusion filter to prevent the access of a substrate with a large periplasmic domain to the protease active site in the membrane (25,27). According to this model, the DegScleaved form of RseA that has lost most of its periplasmic domain (leaving a ~30 a. a. C-terminal periplasmic tail), but not the full-length RseA, can pass through the PDZ filter and access the active site of RseP. Cleavage of other known substrates (remnant signal peptides (14) and a small membrane protein, YqfG (28)) can also be explained by this model, because the formers are generated by the preceding cleavage of precursor secretory proteins around the periplasmic surface of the cytoplasmic membrane by a leader peptidase and this processing is required for the cleavage of signal peptides by RseP, whereas the latter intrinsically has a very small (~12 a. a.) periplasmic domain. The transmembrane segments of the known substrates of RseP share no detectable homology in their primary sequences, suggesting that RseP does not recognize a specific sequence motif(s) in the transmembrane segment of a substrate for its cleavage. Mutational analysis of the transmembrane segments of model substrates showed that the stability of their helical structures is an important determinant for their susceptibility to RseP (28, 29). Consistently, RseP homologs in other bacteria cleave a variety of membrane proteins with no apparent sequence homology. The sequence diversity of the RseP/S2P substrates makes it difficult to predict potential substrates from a simple sequence analysis of transmembrane sequences of membrane proteins. It would thus be reasonable to assume that RseP still has unidentified substrates and plays important roles in some cellular processes by cleaving these substrates and a more comprehensive approach is needed to search for novel substrates and cellular functions of RseP.
In this study, we employed a proteomic approach to achieve the above objective.
Proteomic analysis is a powerful technique to identify novel substrates of proteases as reported previously (30-33). We found that the accumulation of several proteins encoded in the fecABCDE operon (fec operon) decreased considerably in the RseP-deficient strain. The Fec proteins constitute the Fec system that mediates the uptake of ferric citrate by a cell (34,35). We demonstrated that FecR receives sequential processing at the membrane and that RseP plays an essential role in the activation of FecI, an alternative sigma factor dedicated to the transcription of the fec operon (36-39), through the regulated intramembrane cleavage of a periplasmically processed form of FecR, the regulator of the fec operon expression (40-42), which produces the N-terminal cytoplasmic tail fragment of FecR having a "pro-sigma" activity (40, 43) as the last step in the sequential processing of FecR at the membrane. These results uncover the details of the FecR processing in the ferric citrate-induced transmembrane signaling and a new physiological role of RseP in the iron uptake in response to the environmental conditions in E. coli.

Search for new physiological substrates of RseP by quantitative proteomic analysis
For comprehensive understanding of the cellular functions of RseP in E. coli, we attempted to identify new physiological substrates of RseP by a quantitative proteomic approach. We expected that this analysis would enable us to find not only direct proteolytic substrates of RseP but also proteins whose expression is regulated as a result of substrate proteolysis by RseP. We compared the protein levels in the membrane fractions prepared from the ΔrseA ΔrseP mutant strain expressing the wild type or a proteolytically inactive form (a mutant having a Glu-23 to Gln alteration (E23Q) in the conserved zinc metallopeptidase active site motif, HE 23 xxH) of RseP from a plasmid. Although the rseP gene is essential for growth, it can be deleted in the absence of the functional rseA gene encoding anti-σ E protein RseA, a physiological substrate of RseP. In both the strains, σ E is constitutively activated because of the absence of RseA and thus we can exclude the possible changes in protein expression resulting from RseP-dependent induction of the σ E stress response. The membrane fractions were prepared from these strains grown to mid-log phase in LB broth containing 1 mM IPTG, an inducer of the plasmid-encoded RseP derivatives, and subject to nanoLC/MS/MS analysis. As a result, we identified 13,815 unique peptides derived from 1,419 proteins (E. coli strain K-12). Among the identified proteins, 17 exhibited a significant (P <0.05) increase whereas 41 exhibited a significant decrease in the samples prepared from the RseP(E23Q)expressing strain compared with the samples prepared from RseP(WT)-expressing strain (Table 1).
Gene ontology enrichment analysis using the DAVID (44,45) showed that the former included groups related to TonB box receptor and transmembrane beta strand, and the latter included proteins related to iron transport (Fold Enrichment >10, Table S1). Among the former proteins, three [CyoE (heme O synthase (46,47)), AmtB (ammonium transporter (48)), and ZupT (heavy metal divalent cation transporter (49,50))], exhibited a fold change greater than 2 ( Fig. 1). Although these proteins accumulated under the RseP-deficient condition, they are unlikely to be the direct substrates of RseP, as S2P proteases generally cleave a single membrane-spanning protein with type II membrane topology. The top three of the latter group were FecA, FecD, and FecE, which also exhibited a fold change greater than 2. These proteins constitute the Fec system, one of the iron uptake systems in E. coli. Expression of these proteins is known to be induced in response to the availability of environmental iron. The decrease in their levels in the RseP-deficient strain indicates that they should also not be the direct substrates of RseP. However, because it has been shown that many bacterial S2P proteases are involved in cellular responses to environmental changes including the availability of iron (20, 21, 51), we further examined the possible involvement of RseP in the regulation of the Fec system gene expression.

RseP function is required for the transcriptional activation of the fecABCDE operon.
Since iron is generally an essential micronutrient for living organisms, cells now have many evolved iron-uptake systems (52). An E. coli cell is also equipped with multiple iron uptake systems. Among them, the Fec system acts in the uptake of ferric ion (Fe 3+ ) in the form of ferric citrate ( Fig. 2A). Ferric citrate in the extracellular milieu is first transported into the periplasmic space by FecA, an outer membrane transporter (53)(54)(55), and then imported into the cytoplasm via the ABC transporter FecCDE with the assistance of the periplasmic protein FecB (56). These Fec proteins are expressed from the single fecABCDE operon (fec operon) whose transcription is under the control of an alternative sigma factor, FecI (36-39). Because the genes for FecA, FecD, and FecE that exhibited decreased accumulation in the RseP-deficient strain belong to the same fec operon (38), we supposed that RseP is involved in the upregulation of this operon, and examined this possibility.
To easily monitor the transcription of the fec operon, we constructed a reporter (fec reporter) plasmid in which the lacZ reporter gene was placed under the FecI-dependent promoter of the fec operon (PfecA). The ΔompA ΔompC cells (used as the rseP + strain) carrying the reporter plasmid exhibited about 10-fold higher LacZ activity when 1 mM Na3 citrate was added to the medium containing 0.1 μM FeCl3 (Fig. 2B, rseP + ). Note that the addition of increasing concentrations of FeCl3 to the medium led to a drastic decrease in LacZ activity in both ΔompA ΔompC (Fig. S1A) and wild-type (ompA + ompC + ) cells (Fig. S1B). This effect was presumably caused by the negative regulation of PfecA by Fur, a global transcriptional regulator for iron homeostasis (38, 39, 41, 57). These results showed that our reporter plasmid can be used to evaluate the transcription from PfecA in response to ferric citrate.
We then introduced the reporter plasmid into the ΔompA ΔompC ΔrseP cells (rseP can be deleted in a strain lacking the two outer membrane proteins OmpA and OmpC (58)) and examined the effect of rseP disruption on the transcriptional activation of the fec operon in response to ferric citrate. In sharp contrast to rseP + cells, the expression of the fec operon as revealed by LacZ activity was not increased by the addition of citrate (Fig. 2B, ΔrseP / vec). Expression of the wild-type RseP with a C-terminal His6-Myc tag (hereafter RseP-HM), but not its E23Q derivative, from another plasmid restored a citrate-dependent reporter expression (Fig. 2B, WT and E23Q), while the anti-RseP immunoblotting showed that the accumulation levels of the expressed RseP proteins were comparable (Fig. 2C). These results strongly suggest that the proteolytic activity of RseP is essential for the ferric citrate-dependent transcriptional activation of the fec operon.

Proteolytic function of RseP is involved in the processing of the FecR protein.
The transcription of the fec operon is known to be controlled by FecI, an alternative sigma factor whose activity is regulated by FecR (40-42) ( Fig. 2A). FecR is a single-spanning cytoplasmic membrane protein of type II topology, similar to other RseP substrates, and has been reported to be processed into several fragments in vivo (59,60). In addition, its N-terminal cytoplasmic region has been suggested to exhibit an activity ("pro-sigma" activity) that is required for the function of FecI (40, 43). These facts led us to examine the possibility that FecR is cleaved by RseP to activate FecI.
We first investigated whether RseP could affect the in vivo processing of FecR. To this end, we constructed a derivative of FecR with an N-terminal 3xFLAG tag (F-FecR) (Fig. 3A). We also constructed a new fecR (fecR1) strain in which a part of the chromosomal fecR gene was deleted so that the deletion totally disrupts the FecR function but negligibly affects the expression of the upstream and downstream genes (see Supplementary Experimental Procedures). Assays with the fec reporter using the fecR cells showed that the expression of F-FecR restored the ferric citrate-dependent induction of transcription from PfecA that was abolished in the fecR cells, indicating that F-FecR is functional (Fig. 3B). We examined the effects of rseP deletion on the processing of the FecR protein (Fig. 3C). F-FecR was expressed in a ΔrseA rseP + and a ΔrseA ΔrseP strain by growing the cells in M9 medium supplemented with 1 mM Na3-citrate and 10 μM FeCl3 (Fig. 3C, lanes 9 and 10). Note that 10 μM FeCl3 was added to the medium during the analyses of the FecR processing to obtain clear and reproducible results (see Supplementary   Results). Anti-FLAG immunoblotting with total cellular proteins showed that in addition to the full-length F-FecR band (~37 kDa; FL), several smaller FecR-derived fragments were accumulated; FL and an ~25 kDa band (labeled as CL(a)) were detected in both rseP + and ΔrseP strains, whereas a ~17 kDa band (CL(b)) and a ~15 kDa band (CL(c)) were detected only in rseP + and ΔrseP strains, respectively. Also, CL(b) and CL(c) were generated in a citrate-dependent manner (Fig. 3C, compare lanes 5 and 6 with 9 and 10). While CL(a) was generated irrespective of the addition of citrate, the accumulation levels of CL(a) were considerably higher in the absence of citrate than in its presence (Fig. 3C, compare lanes 5 and 6 with 9 and 10). Taken together, these results suggest that ferric citrate affects the generation and/or accumulation of FecR species in the cells. All these species (CL(a), CL(b) and CL(c)) should be N-terminal fragments of F-FecR as they retained the N-terminal FLAG tag, and most likely are the degradation products of FecR.
Citrate did not affect the accumulation levels of the RseP proteins in these strains. When wild-type RseP-HM was ectopically expressed in the ΔrseP stain, CL(b) disappeared and instead CL(c) was detected, while no such effect was observed with the expression of the protease active-site mutant RseP(E23Q) (Fig. 3C, lanes 11 and 12). These results suggest that CL(c) is generated by the RsePdependent proteolytic cleavage of the FecR-related proteins.

RseP cleaves the transmembrane region of FecR to yield the cytoplasmic tail fragment.
In the above immunoblotting experiments, we observed significant variations in the relative amounts of the FecR-related species (data not shown). We suspected that these variations could be caused by the instability of these species in a cell. In addition, the positive feedback regulation of the Fec system (34, 35) could complicate the results as it could enhance the ferric citrate-dependent signal and as a result the signal-induced processing of FecR. To circumvent these problems, we constructed a FecR-derived model substrate, F-MBP-FecR, in which the entire cytoplasmic domain of FecR was replaced with a tightly folded MBP (maltose binding protein) domain (Fig. 4A). This approach has proved to be useful to analyze the RseP-mediated or periplasmic proteolysis in our past studies (14,29). The expression of F-MBP-FecR in the fecR strain did not activate the expression of the fec reporter in the presence or absence of citrate, suggesting that it is not functional, as expected from its lack of the cytoplasmic domain required for the interaction with and activation of FecI (Fig. 3B). Next, we examined the in vivo processing of the F-MBP-FecR protein by immunoblotting ( Fig. 4B and S2). In the presence of citrate, we obtained essentially the same results as those obtained with F-FecR; we detected four F-MBP- Taken together, these results strongly suggest that FecR receives a cleavage in its periplasmic region to generate CL(b). As the "intact" band of F-MBP-FecR121 was slightly larger than the CL(b) band, the possible cleavage site would be located just upstream of reside 121.
The mobility of CL(c), whose production is dependent on the proteolytic function of RseP, was almost the same as that of F-MBP-FecR85 (compare Fig. 4C, lanes 8-10 and 11), suggesting that CL(c) would be generated by the RseP-mediated cleavage around residue 85. This cleavage site is highly likely to be located within the transmembrane region of FecR, given that RseP cleaves a transmembrane segment of a substrate. Accordingly, our results support the prediction that the FecR spans the membrane in the region between residues 80 and 100.

RseP cleaves the FecR CL(b) fragment and converts it to the CL(c) fragment
The were elevated considerably in a citrate-independent manner both in the presence or absence of chromosomal rseP. Consistent with the above result, the accumulation level of F-FecR85 in the rseP + and rseP cells was not affected by the addition of citrate (Fig. S5). In this experiment, the reporter activities in the rseP cells were apparently a little higher than those in the rseP + cells, possibly reflecting the increased accumulation of the F-FecR85 protein in the former cells (Fig.   S5), although the exact reason for this differential accumulation is unclear. Taken together, these results demonstrated that the essential role of RseP in the activation of the fec operon is to produce the CL(c) fragment that co-functions with FecI from CL(b).

Discussion
In this study, we performed proteomic analysis to identify the substrates of RseP, the S2P family intramembrane protease of E. coli, to explore novel functions of this protease, and unveiled that RseP is involved in the regulation of the Fec system (Ferric citrate uptake system) genes through the intramembrane cleavage of a novel physiological substrate, FecR. Our proteomic analysis identified multiple Fec system components (FecA, FecD and FecE) encoded by the fec operon as proteins whose levels were significantly decreased in the RseP-deficient strain, suggesting that RseP is required for the expression of the fec operon. We thus examined the transcriptional regulation of this operon by using a lacZ-reporter and demonstrated that proteolytic activity of RseP is essential for its activation. Transcription of this operon is known to be controlled by FecI, an alternative sigma factor (36-39), whose activity is regulated by a cytoplasmic membrane protein, FecR (40-42). Since FecR is a single-pass transmembrane protein of type II topology, which is a shared feature of substrates of the bacterial S2P proteases including RseP, we examined the possibility that RseP cleaves this protein to induce expression of the fec operon. The experiments using the FecR-derived model substrates revealed that FecR receives sequential processing at the membrane and that RseP participates in the last step in the processing that generates the cytoplasmic tail fragment of FecR required for the transcriptional activation of the fec operon.
FecR was processed to yield the fragments that we named here CL(a), CL(b), and CL(c) (Fig. 5). We demonstrated that RseP cleaves the transmembrane region of CL(b), and converts it to CL(c) (Fig. 6). The production of CL(c) is required and sufficient for the FecI-mediated activation of the fec operon transcription (Fig. 2 and 7). These observations coincide with and further support the results of previous studies, which showed that the expression of the cytoplasmic region of FecR  (51)). Our study added another example to this list.
Our results suggest that CL(b) is produced by the cleavage of CL(a) and that this process is promoted by the ferric citrate signal. Thus, CL(a) apparently receives two successive cleavages to yield CL(c), which is reminiscent of the two-step cleavage of E. coli RseA triggered by the extracytoplasmic stresses. In case of RseA, it is recognized and cleaved by RseP ("site-2 cleavage") only after it received prior "site-1 cleavage" by the membrane serine protease DegS on the periplasmic side. While the two successive cleavages are common among many bacterial S2P substrates, the protease catalyzing the site-1 cleavage is not necessarily a DegS homolog. For example, B. subtilis PrsW, a multipass metalloprotease that cleaves RsiW in response to antimicrobial peptides and envelope stresses is unrelated to E. coli DegS (67). Also, in case of P.
Although we have no information on the putative protease(s) responsible for the production of CL(b) form CL(a) in the FecR-mediated signal transduction, the ferric citrate signal might induce some conformational change in CL(a) to make it susceptible to the site-1 cleavage. Alternatively, the ferric citrate signal might directly activate the putative "site-1" protease. Further analysis, especially the identification of the putative "site-1" protease, will be needed to reveal the detailed molecular mechanism of this process.
The production of CL(a) occurred immediately after the synthesis of FecR and independently of RseP. A previous study suggested that CL(a) is generated by the cleavage between Gly-181 and Thr-182 (59), which is consistent with our result that CL(a) migrated on an SDS-PAGE gel to almost the same position as the FecR-derivative that had been truncated at Gly-181.
This GT motif is conserved among several anti-sigma factors of P. aeruginosa and P. putida (63) and E. coli FecR. Furthermore,  showed that one of them, P. aeruginosa FoxR, underwent self-cleavage at this site and suggested that the other GT motif containing antisigma proteins are also processed auto-catalytically at this motif in the non-enzymatic fashion, i.e., N-O acyl rearrangement (62). It would be very likely that E. coli FecR CL(a) is generated by the self-cleavage at the GT motif, too. Whether prior autoproteolysis is required for the production of CL(b) is unclear. Previous studies showed that some mutations in the GT motif of FoxR, which receives a cleavage by RseP during the signal transduction like FecR, blocks the self-cleavage but still allow iron-dependent signal transduction, suggesting that the self-cleavage is not essential in this system, although it cannot be excluded that the FoxR mutants with the GT motif mutations receive degradation by some proteases to generate a small amount of a CL(a)-like fragment, which is sufficient to drive the down-stream reactions. If the self-cleavage of FecR is not required for the signal transduction in the E. coli Fec system as suggested for FoxR, the possible protease responsible for generation of CL(b) might be able to cleave both of the FL and CL(a) forms of  We previously proposed that the periplasmic PDZ tandem of RseP serves as a sizeexclusion filter to avoid proteolysis of membrane proteins with a large periplasmic domain by RseP (27). In the  E extracytoplasmic stress response, the site-1 protease DegS is activated by the stress signals and cleaves full-length RseA, the ant- E protein, in its periplasmic region, which triggers the following site-2 cleavage by RseP as the RseA degradation intermediate (RseA148) generated by the site-1 cleavage has a small periplasmic region (~30 a. a.) and can get access to the intramembrane active site passing through the PDZ filter. Thus, the PDZ filter ensures the stress dependent activation of  E (Fig. S6) (68, 69). Similarly, in the case of FecR we expect that the  Table. 1). Both AmtB and ZupT are transporters of the cytoplasmic membrane; the former facilitates the uptake of ammonia (48,78) whereas the latter imports divalent metal cations (49,50). Since these proteins are multi-spanning transmembrane proteins, it is unlikely that RseP directly proteolyze these proteins. Their expression or stability Recently, the Fec system has been reported to be crucial in bovine mastitis caused by a pathogenic E. coli strain (81). Bovine mastitis is a disease in dairy cows that causes economic loss to the global dairy industry. This study and further analysis of the regulation of the Fec system may hopefully contribute to unveiling the mechanism and development of the treatment of this disease. Our mass analysis results suggest that the levels of the proteins involved in a variety of cellular activities such as ion transport, rRNA processing, and acetylation are affected by the impairment of the RseP function (Table S1). It raises the possibility that RseP may have additional substrates and act directly or indirectly in still unknown cellular processes. Further study by using a variety of approaches discussed above will lead to the comprehensive understanding of the significance of RseP in cellular activities.

Strains, plasmids and oligonucleotides
Escherichia coli K-12 strains, plasmids and oligonucleotides used in this work are listed in Supplementary Tables S2, S3  Enzyme was set as trypsin/P (cleaves after lysine and arginine also if a proline follows) and semispecific search was performed. Cysteine carbamidomethylation was set as a fixed modification.
Methionine oxidation and acetylation on protein N-termini were set as variable modifications. The search results were filtered with FDR < 1% at the peptide spectrum match (PSM) and protein levels.

β-galactosidase (LacZ) activity assay
LacZ activity of the cells carrying the reporter plasmid pYK149 (PfecA-lacZ) was measured basically according to the procedure described previously (89). The cells were grown at 30°C in M9-based medium with 20 μg/mL each of the 20 amino acids, 2 mg/mL thiamine and 0.4% glucose until mid-log phase. The cells were mixed with Reporter 5xLysis buffer (Promega), frozen at -80°C for more than 1 h, and thawed by incubation at 37°C for 30 min in a clear 96-well plate. Then, the equal volume of Z-buffer (60 mM Na2HPO4·7H2O, 40 mM NaH2PO4·H2O, 10 mM KCl, 1 mM MgSO4·7H2O, 40 mM β-mercaptoethanol) containing 1.32 mg/mL 2-nitrophenyl β-Dgalactopyranoside (ONPG, MilliporeSigma) was added to this lysate and incubated at room temperature, followed by detection of the absorbance at 420 and 550 nm at every 2 min using the Viento Nano microplate reader (BioTek Instruments). The relative LacZ activity was calculated as follows. First, the "raw LacZ activity" was calculated according to the following equation; "raw LacZ activity (arbitrary units)" = (A420 -1.75 x A550) / (incubation time (min)). The value of the raw LacZ activity of each sample was divided by that of a standard cell sample (CU141 cells cultured in M9-based medium), and then by A600 of the bacterial culture at the time of collection, giving the "corrected LacZ activity". Finally, the relative LacZ activity of each sample was obtained by dividing the value of the corrected LacZ activity by that of the corresponding control (see the legends for the control in each experiment).

Immunoblotting
Cells were grown at 30°C in M9-based medium with 20 μg/mL each of the 20 amino acids, 2 μg/mL thiamine and 0.4% glucose until mid-log phase. Total cellular proteins were precipitated with 5% trichloroacetic acid (TCA), washed with acetone and dissolved in SDS sample buffer. Immunoblotting was carried out essentially as described previously (90,91). Proteins were separated by SDS-PAGE and electroblotted onto an Immobilon-P membrane filter (MilliporeSigma). Only when 15% bis-Tris gel was used for SDS-PAGE, a transferred membrane filter was dried at 37°C for 30 min and then hydrophilized with methanol. After blocking with BLOTTO (90), the filter was incubated with an appropriate antibody. For anti-RseP immunoblotting, anti-RseP antibodies were pre-incubated with whole-cell lysates of AD1840 (the ΔrseA ΔrseP ΔdegS strain) at 4°C for 1 h to reduce a background as described previously (83). The filter was then washed, and incubated with goat anti-mouse or anti-rabbit IgG conjugated with horseradish peroxide (Bio-Rad). After washing of the filter, proteins that reacted with secondary antibodies were visualized using ECL or ECL Prime Western Blotting Detection Reagents (Cytiva) and Bio image analyzer LAS4000mini (Cytiva).

Pulse-chase experiment
Cells were grown at 30°C in M9-based medium with 20 μg/mL each of the 18 amino acids (other than methionine and cysteine), 2 μg/mL thiamine and 0.4% glucose until early-log phase. In the experiments shown in Fig. 5