TusA Is a Versatile Protein That Links Translation Efficiency to Cell Division in Escherichia coli

More than 100 different modifications are found in RNAs. One of these modifications is the mnm⁵s²U modification at the wobble position 34 of tRNAs for Lys, Gln, and Glu.

ABSTRACT

To enable accurate and efficient translation, sulfur modifications are introduced posttranscriptionally into nucleosides in tRNAs. The biosynthesis of tRNA sulfur modifications involves unique sulfur trafficking systems for the incorporation of sulfur atoms in different nucleosides of tRNA. One of the proteins that is involved in inserting the sulfur for 5-methylaminomethyl-2-thiouridine (mnm⁵s²U34) modifications in tRNAs is the TusA protein. TusA, however, is a versatile protein that is also involved in numerous other cellular pathways. Despite its role as a sulfur transfer protein for the 2-thiouridine formation in tRNA, a fundamental role of TusA in the general physiology of Escherichia coli has also been discovered. Poor viability, a defect in cell division, and a filamentous cell morphology have been described previously for tusA-deficient cells. In this report, we aimed to dissect the role of TusA for cell viability. We were able to show that the lack of the thiolation status of wobble uridine (U₃₄) nucleotides present on Lys, Gln, or Glu in tRNAs has a major consequence on the translation efficiency of proteins; among the affected targets are the proteins RpoS and Fis. Both proteins are major regulatory factors, and the deregulation of their abundance consequently has a major effect on the cellular regulatory network, with one consequence being a defect in cell division by regulating the FtsZ ring formation.

IMPORTANCE More than 100 different modifications are found in RNAs. One of these modifications is the mnm⁵s²U modification at the wobble position 34 of tRNAs for Lys, Gln, and Glu. The functional significance of U34 modifications is substantial since it restricts the conformational flexibility of the anticodon, thus providing translational fidelity. We show that in an Escherichia coli TusA mutant strain, involved in sulfur transfer for the mnm⁵s²U34 thio modifications, the translation efficiency of RpoS and Fis, two major cellular regulatory proteins, is altered. Therefore, in addition to the transcriptional regulation and the factors that influence protein stability, tRNA modifications that ensure the translational efficiency provide an additional crucial regulatory factor for protein synthesis.

INTRODUCTION

Sulfur is an essential element to all living organisms. In bacteria, sulfur is present not only in the form of cysteine and methionine but also in a variety of cofactors and biomolecules such as thiamine, iron-sulfur (Fe-S) clusters, biotin, lipoic acid, the molybdenum cofactor (Moco), and thio-nucleosides in tRNA (1). In the biosynthetic pathways for these biomolecules, protein persulfide groups (R-SSH) serve as the dominant source of reduced sulfur (2–6). For the initial mobilization of sulfur, these pathways share several highly conserved protein components; among the best known are l-cysteine desulfurases which use l-cysteine as their sulfur source (6–8). These enzymes catalyze the formation of a persulfide group on specific conserved cysteine residues, which, in turn, serve as sulfur donor for the biosynthesis of numerous und diverse sulfur-containing biomolecules. One of these sulfur acceptor proteins common to many different sulfur mobilization routes is the small and highly conserved sulfur-binding protein TusA.

A first functional description of the TusA protein was published in 1998 (9), initially named YhhP. This report described that a mutation in the Escherichia coli yhhP gene affected the stability of σ^S in the logarithmic growth phase (9). Several publications followed on the structure of YhhP, and a possible function in cell division was assigned with a connection to FtsZ ring formation (10, 11). Later, yhhP was identified as essential for 2-thiouridine formation of nucleosides at position 34 in tRNAs for Lys, Gln, and Glu. Accordingly, the gene was renamed tusA, for tRNA 2-thiouridine synthesizing protein.

In recent years, the TusA protein has been identified as a central element supplying and transferring sulfur as persulfide for several pathways, not only including the transfer for 5-methylaminomethyl-2-thiouridine (mnm⁵s²U34) modifications of tRNAs but also the synthesis of the dithiolene group in Moco. The highly conserved cysteine present in the Cys-Pro-X-Pro motif at the N terminus of the protein has been identified as the catalytically active sulfane sulfur-binding residue, which can accept sulfur from the l-cysteine desulfurase donor and transfer it further to different acceptor proteins (11–16).

For its role in inserting thio modifications into certain tRNAs, E. coli TusA was shown to directly interact with the l-cysteine desulfurase IscS. The role of the thio modification at the wobble position U34 of nucleotides present in tRNAs for lysine, glutamine, or glutamate was suggested to be for enhanced translation efficiency by enhancing aminoacylation kinetics, assisting proper codon-anticodon pairing, and preventing frameshifting during translation (17, 18). In a sulfur relay system, TusA transfers the sulfur further to TusD in the TusBCD complex, and then TusE receives the sulfur and finally transfers it to MnmA (15). MnmA is a member of the ATP-pyrophosphatase family, which bears a PP-loop as a signature motif (19). MnmA binds tRNA and ATP and activates the bound tRNA by forming an activated acyl-adenylated intermediate on U34. Subsequently, a nucleophilic attack by the persulfide sulfur of MnmA generates a tRNA thiocarbonyl group and releases AMP (19–21). This thio modification results in a conformation in which mnm⁵s²U is trapped in the C3′ endo form of the ribose, since the large van der Waals radius of the 2-thio group causes a steric clash with its 2′ OH group (22, 23). The conformational rigidity causes preferential pairing of the mnm⁵s²U modified bases with purines and prevents misreading of codons ending in pyrimidines (23–25). Furthermore, the 2-thio group ensures a higher stability of tRNA binding to the ribosomal A site and prevents frameshifting during translation (26). In this context, Maynard et al. (27) identified the 2-thiouridine modification of tRNA^Lys as also responsible for enhanced susceptibility of viral infection by the inhibition of programmed ribosomal frameshifting. The three other thio modifications present in tRNA in E. coli are 4-thiouridine at position 8 (s⁴U8) involving the ThiI protein, 2-thiocytidine at position 32 (s²C32) inserted by TtcA, and 2-methylthio-N⁶ isopentenyladenosine at position 37 (ms²i⁶A37), which depends on the activity of MiaB (28). Both MiaB and TtcA are Fe-S cluster-containing proteins; thus, the pathways for ms²i⁶A37 and s²C32 tRNA modifications are considered Fe-S cluster-dependent pathways, while the s⁴U8 and mnm⁵s²U34 tRNA modifications are described as Fe-S cluster-independent pathways (5). Sulfur modification on the uridine at position 8 of tRNA (s⁴U8) serves as a photosensor in prokaryotes for cells mediating the stringent response under environmental UV stress (29). The 37 position of tRNA, neighboring the anticodon, seems to stabilize and secure the first base pair interaction between codon and anticodon of A·U/U·A, thereby preventing frameshifting. ms²i⁶A37 sulfur modifications further help to structure the anticodon loop in an open conformation for proper decoding by preventing intraloop base pairing (30). Thiolation of cytosine at the C-2 position (s²C32) has been suggested to play a role in improving translational accuracy by reducing the speed of translation, particularly during suboptimal growth conditions (28). However, the role of this tRNA modification is not as well understood as the other ones.

Despite its role as a sulfur transfer protein for 2-thiouridine formation, a fundamental role for TusA in the general physiology of E. coli has been discovered (9). Originally, tusA was identified in a screen by identifying genes that negatively influenced the σ^S-dependent activation of the katE gene in a Δhns background (9). H-NS is a major constituent of the E. coli nucleoid, whereas σ^S is a stress-induced sigma factor. In this initial report, it was concluded that TusA influences the stability of RpoS. As a multisuppressor gene of the tusA-deficient phenotype, the dksA gene was identified, coding for a transcriptional regulator that is involved in the regulation of rpoS (10). A following report analyzing a tusA-deficient strain concluded that the absence of TusA impairs FtsZ ring formation, resulting in the formation of nondivided filamentous cells. This filamentous cell morphology and poor viability of tusA-deficient cells appears mainly in rich media but not in minimal media (9). The effect of TusA on the stability of RpoS and the connection to FtsZ and the filamentous growth phenotype have since remained enigmatic.

In this report, we wanted to shed light on the connection of TusA and RpoS on FtsZ ring formation. We were able to show that the absence of TusA results in an altered translation efficiency of RpoS and not in a reduced stability of RpoS. The changes in cellular RpoS abundance result in a different FtsZ regulation, which results in filament formation at later growth phases. We further show that the abundance of Fis, the “factor of inversion stimulation,” is also altered in ΔtusA cells. Overall, changes in the translation efficiency based on the absence of TusA misbalance the complex cellular regulatory network composed by RpoS and Fis, with one of the consequences being a major defect in cell division.

RESULTS

Growth curves of selected E. coli mutant strains involved in tRNA thio modifications.

It was reported previously that a ΔtusA strain displays a growth-deficient phenotype, especially in 0% NaCl LB medium (9, 10). TusA was shown to be involved in mnm⁵s²U34 tRNA modifications of Lys, Gln, and Glu, with a role in the transfer of sulfur from IscS to TusD in the TusBCD complex (15). To analyze whether the growth-deficient phenotype is based on the lack of mnm⁵s²U tRNA modifications, we recorded the growth curve of strain BW25113 wild type and ΔtusA, ΔtusD, ΔtusE, ΔmnmA, and ΔiscS strains in 0% NaCl LB medium for comparison. Figure 1 shows the growth curves recorded over 10 h, revealing that the ΔtusA mutant strain and the ΔtusD, ΔtusE, and ΔmnmA mutant strains showed the same growth phenotype, with a retarded entry into the exponential phase and the stationary phase of approximately 3 h in comparison to that of the corresponding wild-type strain. All ΔtusA, ΔtusD, ΔtusE, and ΔmnmA mutant stains reached the stationary phase after approximately 8 h, with a comparable optical density at 600 nm (OD₆₀₀) as for the wild-type strain. Only the strain with a mutation of iscS, involved in numerous cellular sulfur transfer pathways, showed a more retarded growth and reached only about half of the OD₆₀₀ after 10 h of growth in comparison to that of the other mutant strains. However, the growth defect of the ΔiscS strain can partially be repaired by introduction of a plasmid containing the sufABCDSE operon (31). The introduction of this operon complements the Fe-S cluster assembly in this strain, but not the mnm⁵s²U34 tRNA modifications, as reported previously (32). Consequently, the growth of the BW25113(DE3) ΔiscS/p(T7)sufABCDSE strain was comparable to that of the ΔtusA, ΔtusD, ΔtusE, and ΔmnmA single mutant strains.

FIG 1.

Growth curves of E. coli BW25113 and strains with deletions for genes involved in mnm⁵s²U34 tRNA modifications. BW25113, ΔtusD, ΔtusE, ΔmnmA, ΔiscS, and DE3 ΔiscS/p(T7)sufABCDSE strains were grown in 50 ml LB medium (0% NaCl) at 37°C with 180 rpm for 10 h. The increase in OD₆₀₀ was recorded every 30 min in the first 6 h and then every hour. The starting OD₆₀₀ was 0.05. At certain time points, 2 ml of cells was collected for microscopy studies (see Fig. 3). Growth curves represent the means from 3 biological replicates (±standard deviations [SDs]).

To test whether the growth retardation of the ΔtusA strain is based mainly on the lack of mnm⁵s²U34 tRNA modifications and does not additionally depend on other tRNA modifications, we also analyzed the growth phenotype of ΔiscU, ΔsufS, ΔttcA, ΔmiaB, and ΔthiI strains. These strains are either impaired in Fe-S cluster biosynthesis (IscU), Fe-S cluster biosynthesis/repair (SufS), s²C32 tRNA modifications (TtcA), ms²i⁶A37 tRNA modifications (MiaB), or s⁴U8 tRNA modifications/thiamine biosynthesis (ThiI). The growth curves in Fig. 2 show that only the ΔthiI and ΔiscU strains showed retarded growth, while the growth behavior of the ΔmiaB and ΔttcA strains was not altered in comparison to that of the BW25113 wild-type strain.

FIG 2.

Growth curves of BW25113 and ΔsufS, ΔiscU, ΔthiI, ΔttcA, and ΔmiaB deletion strains. BW25113, ΔsufS, ΔiscU, ΔthiI, ΔttcA, and ΔmiaB strains were grown in 50 ml LB medium (0% NaCl) at 37°C with 180 rpm for 10 h. The increase in OD₆₀₀ was recorded every 30 min in the first 6 h and then every hour. The starting OD₆₀₀ was 0.05. At certain time points, 2 ml of cells was collected for fluorescence microscopy studies (see Fig. 4). Growth curves represent the means from 3 biological replicates (±SDs).

An elongated cell morphology is a common growth phenotype of strains with mutations in mnm⁵s²U tRNA thio modifications.

It was reported previously that ΔtusA mutant strains show a filamentous growth phenotype similar to that observed for strains deficient in ftsZ (10), which encodes a protein involved in the cell division apparatus by formation of the FtsZ ring (33). Since the ΔtusD, ΔtusE, and ΔmnmA mutant strains showed the same growth phenotype as the ΔtusA strain, we investigated the cell morphology of BW25113, ΔtusA, ΔtusD, ΔtusE, ΔmnmA, and ΔiscS strains and the DE3 ΔiscS/p(T7)sufABCDSE strain in which the sufABCDSE operon is introduced containing an isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible promoter. Cell samples were taken from the exponential growth phase after 3 h and 5 h and the early/late stationary growth phase after 7 h and 10 h, and cells were fixed and stained for fluorescence microscopy analysis (Fig. 3). While the wild-type cells were rod shaped at these time points, with an average cell length of 2 μm at 3 h and 1 μm at 5 h, 7 h, and 10 h, the ΔtusA, ΔtusD, ΔtusE, and ΔmnmA mutant strains showed an elongated cell morphology, with an average cell length of 3 to 4 μm in the exponential phase at 3 h and 5 h. These longer cells disappeared when reaching the stationary phase (7 h and 10 h), where the same rod shape morphology was observed as for the wild-type cells. In contrast, the ΔiscS mutant strain showed an elongated cell morphology during all time points, and the longer cells did not disappear after entering the stationary phase. However, when the sufABCDSE plasmid was introduced into the DE3 ΔiscS strain, the elongated cells disappeared in the stationary phase (7 h and 10 h) and the cell morphology was similar to that for the ΔtusA, ΔtusD, ΔtusE, and ΔmnmA mutant strains.

FIG 3.

Cell morphology analysis of BW25113 and strains with gene deletions in mnm⁵s²U34 tRNA modifications. BW25113, ΔtusA, ΔtusD, ΔtusE, ΔmnmA, ΔiscS, and DES ΔiscS/p(T7)sufABCDSE cells were grown in LB medium (0% NaCl) at 37°C with 180 rpm for 10 h, and 2 ml of cells was analyzed by fluorescence microscopy at time points 3 h, 5 h, 7 h, and 10 h after inoculation. DNA was stained with DAPI (1:1,000). Morphology of the cells was visualized by fluorescence microscopy (×100 magnification). Scale bar, 1 μm.

For comparison, we also analyzed the growth phenotype and the cell morphology of ΔiscU, ΔsufS, ΔttcA, ΔmiaB, and ΔthiI strains. As shown in Fig. 4, no elongated cell morphology was observed for ΔiscU, ΔsufS, ΔttcA, ΔmiaB, and ΔthiI mutant strains, revealing that this cell morphology in the exponential phase is likely caused by a defect in mnm⁵s²U34 tRNA modifications and not by a defect in Fe-S cluster biosynthesis or tRNA thio modifications at positions s⁴U8, s²C32, or ms²i⁶A37. The data also reveal that IscU and ThiI are not involved in this specific pathway.

FIG 4.

Cell morphology analysis of BW25113 and ΔsufS, ΔiscU, ΔthiI, ΔttcA, and ΔmiaB deletion strains. BW25113, ΔsufS, ΔiscU, ΔthiI, ΔttcA, and ΔmiaB cells were grown in LB medium (0% NaCl) at 37°C with 180 rpm for 10 h, and 2 ml of cells was analyzed by fluorescence microscopy at time points 3 h, 5 h, 7 h, and 10 h after inoculation. DNA was stained with DAPI (1:1,000). Morphology of the cells was visualized by fluorescence microscopy (×100 magnification). Scale bar, 1 μm.

FtsZ levels are mainly deregulated in mutant strains impaired in mnm⁵s²U34 tRNA thio modifications.

A major factor involved in the assembly of the cell division apparatus is the FtsZ protein (33, 34). Since ΔtusA, ΔtusD, ΔtusE, ΔmnmA, ΔiscS, and ΔiscU mutant strains showed a defect in cell division during exponential growth, we investigated the growth-dependent levels of FtsZ in these strains by using antisera raised against FtsZ. Since FtsZ was shown to be particularly expressed at the initial growth phase (35, 36), cells were harvested from the different mutant strains at the early stages of the exponential growth phase at 90 min, 3 h, 4 h, and 5 h. The cell lysate was subjected to immunodetection using an antiserum derived against FtsZ. As shown in Fig. 5, FtsZ is mainly abundant at 90 min of growth in the BW25113 wild-type strain, with a decrease in its concentration after 3 and 4 h of growth, finally being almost undetectable after 5 h of growth. In contrast, in the ΔtusA mutant strain, the highest concentration of FtsZ was detectable after 5 h of growth; the concentration of FtsZ gradually increased during the earlier growth points from 1.5 to 4 h. Also, a higher concentration of FtsZ was observed in the ΔtusD, ΔtusE, ΔmnmA, ΔiscS, and ΔiscU mutant strains in the later growth points, with the highest concentration after 5 h (Fig. 5).

FIG 5.

Detection of cellular FtsZ levels by immunodetection in different E. coli mutant strains. Immunodetection of FtsZ in BW25113 wild type and ΔtusA, ΔtusD, ΔtusE, ΔmnmA, ΔiscS, and ΔiscU mutant strains. The strains were grown aerobically in LB medium (0% NaCl) at 37°C for 1.5 h, 3 h, 4 h, and 5 h. Cells were lysed, and 50 μg of each total protein fraction was separated using 12% SDS-PAGE and transferred onto a PVDF membrane. A FtsZ-specific antiserum (1:20,000 dilution) was used, and proteins were visualized by enhanced chemiluminescence. The prestained protein molecular weight marker (Fermentas; 20 kDa to 120 kDa) was used as a reference.

Detection of the levels of Fis in different E. coli mutant strains.

TusA was initially identified in a screen to identify genes in a Δhns background that are differently regulated by RpoS during logarithmic growth (9). The expression of hns was shown to be regulated by Fis, the factor of inversion stimulation, a transcriptional regulator that regulates the transcription of 21% of genes in E. coli (37–39). Fis, together with HNS, HU, IHF, and DPS, is one of the largest components of the nucleoid (40). Under optimal growth conditions in the logarithmic phase, up to 60,000 copies of Fis can be found in the cell, while the protein is nearly undetectable in the stationary phase (<100 molecules per cell) (41). To determine whether an altered regulation of Fis can be observed in the mutant strains impaired in mnm⁵s²U34 tRNA thio modification, cells from ΔtusA, ΔtusD, ΔtusE, ΔmnmA, ΔiscS, and ΔiscU mutant strains were harvested in the logarithmic and stationary growth phases after 90 min, 4 h, 5 h, 7 h, and 10 h of growth and subjected to immunodetection using a Fis antiserum after SDS-PAGE and transfer to polyvinylidene difluoride (PVDF) membranes. Figure 6 shows that in the wild-type cells, Fis was most abundant in the early logarithmic phase, between 90 min and 4 h of growth, while at later growth phases, the protein was almost undetectable with the Fis antiserum. In contrast, in the ΔtusA, ΔtusD, ΔtusE, and ΔmnmA mutant strains, Fis was almost undetectable at 90 min of growth but most abundant between 4 h and 7 h of growth. When entering the stationary growth phase, Fis was almost undetectable in these mutant strains, as also observed in the wild-type. The ΔiscU and ΔiscS strains showed the most severely retarded growth phenotype, and neither of these strains reached the OD₆₀₀ in the stationary phase of the wild-type strain; only approximately half of the OD₆₀₀ value was reached. Both strains are impaired in the formation of Fe-S clusters. In the ΔiscS mutant strain, the amount of Fis was most abundant after 5 h, 7 h, and 10 h of growth, while being almost undetectable after 90 min and 4 h of growth (Fig. 6). In the ΔiscU mutant strain, Fis was not detectable after 90 min of growth but most abundant after 4 h of growth, with gradually decreasing concentrations after growth at 5 h, 7 h, and 10 h.

FIG 6.

Detection of cellular Fis levels by immunodetection in different E. coli mutant strains. Immunodetection of Fis in BW25113 wild type and ΔtusA, ΔtusD, ΔtusE, ΔmnmA, ΔiscS, and ΔiscU mutant strains. The strains were grown aerobically in LB medium (0% NaCl) at 37°C for 1.5 h, 4 h, 5 h, 7 h and 10 h. Cells were lysed, and 50 μg of each total protein fraction was separated using 15% SDS-PAGE and transferred onto a PVDF membrane. A Fis-specific antiserum (1:10,000 dilution) was used, and proteins were visualized by enhanced chemiluminescence. The prestained protein molecular weight marker (Fermentas; 20 kDa to 120 kDa) was used as a reference.

Immunodetection of RpoS in mutant strains impaired in mnm⁵s²U34 tRNA thio modifications.

Earlier studies have reported that TusA influences the stability and translation efficiency of RpoS (10, 42). The rpoS gene encodes the alternative sigma factor σ^S that efficiently transcribes genes involved in general stress response and that also has important roles in nucleic acid synthesis, modification, and turnover (43).

The regulation of RpoS is growth phase regulated and in an inverse relationship to Fis (44–50). A strong Fis-binding site in the rpoS promoter region has been shown to be responsible for this regulation (48). While the Fis protein in wild-type cells is undetectable in the stationary growth phase, its concentration is the highest during early growth phases (41, 51). During exponential growth when Fis is present, the protein binds to the rpoS promoter and represses rpoS transcription (52). When the cells enter the stationary growth phase, however, Fis disappears and rpoS transcription increases nearly 10-fold (47, 48).

Since our results show that mainly the mnm⁵s²U34 tRNA modifications are involved in the filamentous growth phenotype, resulting in lower Fis abundance during the early exponential growth phase and higher Fis abundance during the late exponential and stationary growth phases, we first determined the amounts of RpoS in the logarithmic growth phase (5 h) and the stationary growth phase (7 h and 10 h) in the BW25113 wild type and in ΔtusA, ΔtusD, ΔtusE, ΔmnmA, ΔiscS, ΔiscU, ΔmiaB, ΔthiI, ΔttcA, ΔsufS, and ΔrpoS mutant strains by immunodetection using an RpoS-derived antiserum. The ΔrpoS mutant strain served as a negative control. Cells were lysed and pelleted, and the proteins of the soluble fractions were separated by SDS-PAGE, transferred onto PVDF membranes, and immunodetected using a RpoS antiserum. Figure 7 shows that, as expected, the amount of RpoS correlated in an inverse manner with the amount of Fis in the different growth stages. In the logarithmic growth phase after 3 h and 5 h, RpoS was undetectable in ΔtusA, ΔtusD, ΔtusE, ΔmnmA, and ΔiscS mutant strains, strains that were shown to contain larger amounts of Fis at this growth time (Fig. 6). In the early stationary growth phase at 7 h, the amount of RpoS was only slightly increased in ΔtusA, ΔtusD, ΔtusE, ΔmnmA, and ΔiscS mutant strains (Fig. 7), a growth point at which Fis was also still detectable in these mutant strains (Fig. 6). After 10 h of growth, only the IscS mutant strain showed a slightly reduced RpoS abundance (Fig. 7), a growth point at which Fis was still detectable in this mutant strain (Fig. 6). In comparison, in BW25113 wild-type, ΔmiaB, ΔthiI, ΔttcA, and ΔsufS strains, the abundance of RpoS remained the same at all growth times. Only the ΔiscU strain showed a slightly smaller RpoS amount during all growth times.

FIG 7.

Detection of cellular RpoS levels by immunodetection in different E. coli mutant strains. Immunodetection of RpoS in BW25113 wild type and ΔtusA, ΔtusD, ΔtusE, ΔmnmA, ΔiscS, ΔiscU, ΔmiaB, ΔthiI, ΔttcA, ΔsufS, and ΔrpoS mutant strains. The strains were grown aerobically in LB medium (0% NaCl) at 37°C for 3 h, 5 h, 7 h, and 10 h. Cells were lysed, and 100 μg of each total protein fraction was separated using 12% SDS-PAGE and transferred onto a PVDF membrane. An RpoS-specific antiserum (1:1,000 dilution) was used, and proteins were visualized by enhanced chemiluminescence. The prestained protein molecular weight marker (Fermentas; 20 kDa to 120 kDa) was used as a reference.

Analysis of the translation efficiency of RpoS in mutant strains impaired in mnm⁵s²U34 tRNA thio modifications.

Since tusA, tusD, tusE, mnmA, and iscS are involved in mnm⁵s²U34 tRNA thio modifications, we further investigated the effect of mutations in these genes on the translation efficiency of RpoS and Fis. The translation efficiency was recorded in the BW25113 wild type and in ΔtusA, ΔmnmA, ΔiscU, and ΔiscS mutant strains after 5 h of growth by flow cytometry, since at this growth point, the effect on RpoS protein abundance was most pronounced.

We used translational fusions containing the IPTG-inducible T7 promoter and the rpoS or fis coding sequence without the stop codon fused in frame to the coding sequence of enhanced green fluorescent protein (EGFP) as a readout. BW25113 wild-type, ΔtusA, ΔmnmA, ΔiscU, and ΔiscS cells were transformed with the EGFP fusion reporter constructs in addition to plasmids containing mCherry under the control of the T7 promoter without a gene fusion as control. The mCherry fluorescence thereby serves as a control to assay for changes in translation efficiency caused by the gene mutation of the respective strain on the reporter itself.

The EGFP and mCherry fluorescence levels were measured by flow cytometry and compared in the different mutant strains after 5 h of growth when cells were in the logarithmic growth phase. In the BW25113 wild type and each mutant strain, the EGFP-fusion and mCherry fluorescence was determined and compared to the fluorescence determined in the same mutant strains that contained mCherry and EGFP without a gene fusion. Figure 8 shows the flow cytometry fluorescence obtained from the rpoS-EGFP fusion measured in the BW25113 wild type and ΔtusA, ΔmnmA, ΔiscU, and ΔiscS strains. In Fig. 9 the differences in fluorescence peaks measured by the distance to the respective control peak are depicted from three independent measurements. The fluorescence obtained for the mCherry control in the different mutant strains did not differ when the strains containing the rpoS-EGFP reporter fusion were compared to the ones containing only the EGFP reporter plasmid (Fig. 8B and 9). This shows that, generally, the translation efficiency of the fluorescence reporter plasmid is not influenced by the mutant strains. In comparison, the rpoS-EGFP fusion showed enhanced fluorescence in the BW25113 wild type compared to that of EGFP alone, correlating with higher levels of EGFP expression in the rpoS-fusion protein (Fig. 8A and 9). These higher fluorescence levels of the rpoS-EGFP fusion were also observed in the iscU mutant strain but not in the ΔtusA, ΔmnmA, or ΔiscS mutant strains. In the strains with an impairment of mnm⁵s²U34 tRNA modifications, the fluorescence levels of EGFP and rpoS-EGFP were comparable. This shows that the mnm⁵s²U34 tRNA modifications have a positive effect on the translation efficiency of RpoS. Consequently, the loss of mnm⁵s²U34 tRNA modifications resulted in lower cellular levels of RpoS. The complementation of the ΔtusA mutant strain with a plasmid expressing rpoS from an IPTG-inducible promoter did not rescue the ΔtusA phenotype and consequently did not revert the elongated cell morphology (see Fig. S1 and S2 in the supplemental material). This shows that the effect on cell division observed in the ΔtusA cells is not solely based on a reduced level of RpoS but rather is the consequence of reduced concentrations of several proteins, based on a reduced translation efficiency in these strains.

FIG 8.

Fluorescence quantification of rpoS-EGFP translational gene fusions expressed in strains with deletion of genes responsible for mnm⁵s²U34 tRNA modifications. Translation efficiencies of rpoS were analyzed in BW25113(DE3) wild type and strains DES ΔtusA, DE3 ΔmnmA, DE3 ΔiscU, and DE3 ΔiscS that were transformed with plasmids rpoS-EGFP-pACYCDuet-1/mCherry-pCDFDuet-1 or the empty EGFP-pACYCDuet-1/mCherry-pCDFDuet-1 plasmids as a reference. Cells were grown in 50-ml cultures of LB (0% NaCl) at 37°C with 180 rpm for 5 h. Protein expression was induced with 100 μM IPTG at time point 0. The cell count for each sample was set to 10⁸ cells/ml for flow cytometry. Blue peaks show the fluorescence of the empty EGFP-pACYCDuet-1/mCherry-pCDFDuet-1 plasmid reference, and red peaks show the fluorescence from rpoS-EGFP-pACYCDuet-1/mCherry-pCDFDuet-1 fusions. (A) The fluorescence of EGFP was recorded at an excitation of 488 nm and an emission of 507 nm. (B) The fluorescence of mCherry was recorded at an excitation of 587 nm and an emission of 610 nm. The distributions shown are taken from single experiments and are representative of three independent experiments.

FIG 9.

rpoS-EGFP translation differences shown in distance measurements to the control fluorescence signal. The differences in peak distance represent the differences in EGFP translation caused by the rpoS-EGFP fusion. Each bar represents the peak distance in centimeters from the flow cytometry histograms from the rpoS-EGFP peak to the EGFP control peak (in blue) or the mCherry peaks (in red). Each bar represents the mean value from three independent measurements ± the standard deviation.

Analysis of the translation efficiency Fis in mutant strains impaired in mnm⁵s²U34 tRNA thio modifications.

In addition to rpoS, we investigated the translation efficiency of fis in the same mutant strains. The fis-EGFP fusion also had a positive effect on the translation efficiency, since a higher fluorescence level was obtained in the BW25113 wild type strain in comparison to that for EGFP alone (Fig. 10). In the ΔiscU deletion strain, an even slightly higher fluorescence level was obtained for the fis-EGFP fusion compared to the fluorescence obtained in the BW25113 strain. In the ΔtusA mutant strain, the fluorescence of the fis-EGFP fusion in comparison to the EGFP fluorescence alone was reduced (Fig. 10). Even lower levels in the fis-EGFP fluorescence were obtained in the ΔmnmA and ΔiscS mutant strains.

FIG 10.

Fluorescence quantification of fis-EGFP translational gene fusions expressed in strains with deletions of genes responsible for mnm⁵s²U34 tRNA modifications. Translation efficiencies of fis were analyzed in BW25113(DE3) wild type and strains DE3 ΔtusA, DE3 ΔmnmA, DE3 ΔiscU, and DE3 ΔiscS that were transformed with plasmids fis-EGFP-pACYCDuet-1/mCherry-pCDFDuet-1 or the empty EGFP-pACYCDuet-1/mCherry-pCDFDuet-1 plasmids as a reference. Cells were grown in 50-ml cultures of LB (0% NaCl) at 37°C with 180 rpm for 5 h. Protein expression was induced with 100 μM IPTG at time point 0. The cell count for each sample was set to 10⁸ cells/ml for flow cytometry. Blue peaks show the fluorescence of the empty EGFP-pACYCDuet-1/mCherry-pCDFDuet-1 plasmid reference, and red peaks show the fluorescence from fis-EGFP-pACYCDuet-1/mCherry-pCDFDuet-1 fusions. (A) The fluorescence of EGFP was recorded at an excitation of 488 nm and an emission of 507 nm. (B) The fluorescence of mCherry was recorded at an excitation of 587 nm and an emission of 610 nm. The distributions shown are taken from single experiments and are representative of three independent experiments.

To analyze whether the translation efficiency of Fis is only influenced by mnm⁵s²U34 thio modifications in tRNA for Glu, Gln, or Lys, we exchanged all codons in fis coding for Lys, Gln, and Glu to codons for Ala. This fis^mut gene variant was also fused to the reading frame of EGFP, resulting in a fis^mut-EGFP fusion. The results show that the translation efficiency was reduced in this fis^mut-EGFP fusion (Fig. 11). Furthermore, the results also show that when fis translation does not involve any mnm⁵s²U34-modified tRNAs, the overall abundance of Fis is reduced during the exponential growth phase to levels obtained in the ΔtusA, ΔmnmA, or ΔiscS strains (Fig. 10A). This clearly shows that the mnm⁵s²U34 tRNA modifications in tRNAs for Glu, Gln, and Lys positively influence the translation of fis, an effect that is reduced either in cells that do not harbor mnm5s2U34 tRNA modifications or when using a construct that does not depend on thio-modified tRNAs. In comparison, almost the same fluorescence levels of the fis^mut-EGFP fusion and EGFP were obtained in ΔtusA, ΔmnmA, and ΔiscS strains, showing that the positive effect on the translation efficiency was completely diminished in these strains. This might indicate that TusA, MnmA, and IscS have additional roles that influence the translation of the fis mRNA, e.g., the absence of the proteins involved in mnm⁵s²U34 tRNA modifications might slightly negatively influence the Fe-S-dependent tRNA modifications, since it has been shown that the absence of IscS and TusA reduces the cellular Fe-S cluster levels (32). In Fig. 12A and B, the differences in fluorescence peaks measured by the distance to the respective control peak are depicted from three independent measurements.

FIG 11.

Fluorescence quantification of fis^mut-EGFP translational gene fusions expressed in strains with deletions of genes responsible for mnm⁵s²U34 tRNA modifications. Translation efficiencies of a mutated version (fis^mut) in which all codons for Lys, Gln, and Glu were replaced by codons for Ala were analyzed in BW25113(DE3) wild type and strains DES ΔtusA, DE3 ΔmnmA, DE3 ΔiscU, and DE3 ΔiscS strains that were transformed with plasmids fis^mut-EGFP-pACYCDuet-1/mCherry-pCDFDuet-1 or the empty EGFP-pACYCDuet-1/mCherry-pCDFDuet-1 plasmids as a reference. Cells were grown in 50-ml cultures of LB (0% NaCl) at 37°C with 180 rpm for 5 h. Protein expression was induced with 100 μM IPTG at time point 0. The cell count for each sample was set to 10⁸ cells/ml for flow cytometry. Blue peaks show the fluorescence of the empty EGFP-pACYCDuet-1/mCherry-pCDFDuet-1 plasmid reference, and red peaks show the fluorescence from fis^mut-EGFP-pACYCDuet-1/mCherry-pCDFDuet-1 fusions. (A) The fluorescence of EGFP was recorded at an excitation of 488 nm and an emission of 507 nm. (B) The fluorescence of mCherry was recorded at an excitation of 587 nm and an emission of 610 nm. The distributions shown are taken from single experiments and are representative of three independent experiments.

FIG 12.

fis-EGFP and fis^mut-EGFP translation differences shown in distance measurements relative to the control fluorescence signal. The differences in peak distance represent the differences in EGFP translation caused by the fis-EGFP fusion (A) and the fis^mut-EGFP fusion (B). Each bar represents the peak distance from the flow cytometry histograms from the fis-EGFP peak or the fis^mut-EGFP to the EGFP control peak (in blue) or the mCherry peaks (in red). Each bar represents the mean value from three independent measurements ± the standard deviation.

DISCUSSION

A growth-deficient phenotype has been reported for a deletion strain in tusA (9, 10). Especially during the early exponential growth phase, the formation of filaments has been observed, which then disappear in the later exponential growth phase (10). Therefore, earlier studies suggested that TusA influences FtsZ ring formation. This conclusion was also based on the fact that the ftsZ gene was able to serve as multicopy suppressor of the ΔtusA mutant phenotype (10). In the study by Ishii et al. (10), the dksA gene was identified to also suppress the ΔtusA phenotype. In the same study, it was reported that a mutation in tusA reduced the stability of σ^S in the logarithmic growth phase (10). However, the exact link of FtsZ, DksA, RpoS, and TusA remained elusive. In our study, we aimed to dissect the effect of TusA on cell division and filamentous growth in detail (Fig. 13). We were able to reproduce the growth defect of a ΔtusA mutant strain in low-salt LB medium. Since the main cellular role of TusA is its involvement in transferring sulfur for the formation of mnm⁵s²U34 thio modifications of tRNA for Lys, Gln, or Glu (15), we analyzed additional strains with mutation in genes involved in mnm⁵s²U34 thio modifications for their growth phenotype. We were able to show that the ΔtusD, ΔtusE, and ΔmnmA mutant strains have a similarly retarded growth and form filaments as the ΔtusA mutant strain. For comparison, growth of the ΔiscS mutant strain was even more reduced, not reaching the OD₆₀₀ of the wild-type strain in the stationary phase. Since IscS is involved in sulfur transfer to numerous important sulfur-containing biomolecules in E. coli (6), the most important of which are Fe-S clusters (53), we suppressed the defect in Fe-S cluster biosynthesis by introducing the sufABCDSE operon into the ΔiscS mutant strain (54). This complemented DE3 ΔiscS/p(T7)sufABCDSE strain showed the same growth curve as strains with an impairment in mnm⁵s²U34 formation, showing that the growth defect is based on the absence of mnm⁵s²U34-modified tRNAs and not on the absence of Fe-S clusters. The same effect was obtained for the formation of elongated cells. Our results show that the elongated cell morphology is common to all mutant strains with an impairment in mnm⁵s²U34 tRNA modifications, since ΔtusD, ΔtusE, ΔmnmA, and ΔiscS mutant strains showed the same elongated cell morphology. For comparison, cell strains with a defect in tRNA modifications such as s²C32 (ΔttcA) or ms²i⁶A37 (ΔmiaB) or in Fe-S cluster biosynthesis (ΔiscU) did not show this elongated cell morphology.

FIG 13.

Model for the regulatory network connecting tRNA thio modifications to RpoS regulation, FtsZ ring formation, and Fis regulation. The IscS-derived persulfide-sulfur serves as a sulfur source for the biosynthesis of numerous and diverse sulfur-containing biomolecules, among which are Fe-S clusters and thio modifications in tRNA. The synthesis of thio-nucleosides is divided into those which are Fe-S-cluster dependent and those that are Fe-S cluster independent. The Fe-S cluster-independent pathways form mnm⁵s²U34 and s⁴U8 thio-modified nucleosides in tRNA. The Fe-S cluster-dependent pathways form s²U32 and ms²i⁶A37-modified nucleosides in tRNA. Sulfur-containing tRNA modifications are essential to ensure an accurate translation efficiency. One of the major cellular targets for an altered translation efficiency by the absence of mnm⁵s²U34 tRNA modifications is the RpoS protein. RpoS is a major regulatory protein of the cell which was shown to positively regulate the expression of numerous genes. Two of the proteins regulated by RpoS are Fis and FtsZ. Fis is the “factor of inversion stimulation” that, together with H-NS and other proteins, forms the nucleoid. Both proteins bind to the DNA and influence its bending. RpoS and Fis are regulated by the DksA protein. FtsZ is involved in cell division. For cell division, the FtsZ protein assembles in the center of the cell and forms a ring structure, the FtsZ ring. The deregulation of both Fis and RpoS decreases the FtsZ concentration in the early exponential growth phase and results in an elongated cell morphology. +, activation; −, repression; black arrows, regulation at the level of transcription; red arrows, regulation at the level of translation.

Originally, TusA was identified in a Δhns background for the identification of genes that influence the cellular content of RpoS during logarithmic growth (9). Since the expression of hns was shown to be regulated by Fis, we additionally investigated the concentration of Fis in ΔtusA cells. Additionally, longer cells have been described not only for ΔftsZ mutants but also for cells with a defect in the global regulator Fis (10, 33, 55).

Overall, we analyzed the concentrations of FtsZ and Fis in strains impaired in mnm⁵s²U34 tRNA modifications. The results showed that the concentrations of FtsZ are higher in later growth phases. This behavior is consistent with a retardation of the exponential growth phase, showing the highest concentration of FtsZ during this growth phase. A similar behavior was observed for Fis in these mutant strains. While the concentration of Fis is the highest in the early exponential growth phase in the wild type, the highest concentration was shifted to a growth time of 4 to 5 h in ΔtusA, ΔtusD, ΔtusE, and ΔmnmA mutant strains. Therefore, the formation of filaments and the defect in cell division are likely based on changes in the concentrations of Fis and FtsZ in the cell (Fig. 13).

Since Fis inversely regulates rpoS expression by acting as a repressor for rpoS transcription, we showed that in strains in which the Fis abundance is shifted to later growth phases, the protein amount of RpoS decreased. Furthermore, since the results described above show that the major cellular effect on cell division and the formation of filaments is based on the absence of mnm⁵s²U34 tRNA modifications, we also analyzed whether the translation efficiency of RpoS and Fis in strains deficient in mnm⁵s²U34 tRNA modifications was altered. The results show a major effect of tRNA modifications in mnm⁵s²U34 on RpoS translation, while Fis translation was also reduced but to a smaller extent. We consequently believe that the translation of rpoS and fis is regulated by mnm⁵s²U34 tRNA modifications, which is growth phase dependent.

In addition to the repression of RpoS transcription by Fis in the stationary growth phase, the translation efficiency of rpoS was shown to be largely reduced in the ΔtusA mutant strain in the exponential growth phase, consequently resulting in smaller amounts of RpoS. In the report by Ishii et al. (10), the filamentous phenotype was rescued by introduction of the dksA gene into the ΔtusA mutant strain. The expression of both fis and rpoS is regulated by DksA (45, 46). While DksA represses the transcription of fis (56), the protein acts as an activator for rpoS transcription (45). Since the filamentous ΔtusA growth phenotype was suppressed by the introduction of larger amounts of DksA (10), we conclude that a higher transcription of rpoS was obtained in this strain. The repression of fis transcription and the activation of rpoS transcription consequently resulted in elevated levels of RpoS and rescued the RpoS deficiency of the ΔtusA strain. This might have overcome the negative influence on RpoS translation caused by the absence of TusA. RpoS itself has also been reported to further regulate ftsZ transcription in the stationary phase (57, 58). FtsZ synthesis is normally dependent upon the housekeeping sigma 70 factor; although, as cells enter stationary phase, RpoS is responsible for increased ftsZ expression from a second promoter site (58, 59).

Consequently, the deletion of tusA deregulated a complex regulatory network in the cell, which is mainly based on the reduced translation of RpoS and Fis due to the lack of mnm⁵s²U34 tRNA modifications (Fig. 13). RpoS is a major regulatory protein of the cell which was shown to positively regulate the expression of 268 genes; thus, major cellular changes are expected when the translation of RpoS is reduced (60, 61). Fis, which together with HNS, HU, IHF, and DPS, is one of the largest components of the nucleoid and is a transcriptional regulator that regulates the transcription of 21% of genes in E. coli (37–39). Thus, in addition to the effect on cell division, the absence of mnm⁵s²U34 tRNA modifications and the reduced levels of RpoS and Fis are expected to have multiple cellular effects. Conclusively, as revealed in this report, alterations of the translation efficiency of major regulatory proteins, which is controlled by tRNA thio modifications, can cause major changes to the regulatory cellular network.

MATERIALS AND METHODS

Bacterial strains, plasmids, media, and growth conditions.

BW25113 (referred to as the wild-type strain) and the isogenic mutant ΔtusA, ΔtusD, ΔtusE, ΔmnmA, ΔiscS, ΔiscU, ΔttcA, ΔmiaB, ΔthiI, ΔsufS, and ΔrpoS strains were obtained from the Keio collection from the National BioResource Project (National Institute of Genomics, Japan) (62, 63). Successful deletion of the respective genes was validated by PCR amplification. E. coli cultures were generally grown in LB medium (0% NaCl) under aerobic conditions at 37°C.

Growth curves.

Precultures of BW25113 wild-type and deletion strains were inoculated in M9 minimal medium and grown overnight at 37°C with 200 rpm. The next day, cells were transferred to 50 ml of LB (0% NaCl) at an optical density at 600 nm (OD₆₀₀) of 0.05, and growth was continued for 10 h with a cultivation at 37°C and 180 rpm. Within the first 6 h, OD₆₀₀ was measured every 30 min. After 6 h of growth, the measurements were performed every hour. Cell samples (2 ml) of each strain were taken in the exponential (3 h and 5 h) and stationary (7 h and 10 h) growth phases for morphological analyses by fluorescence microscopy. BW25113 and ΔiscS cells were additionally transformed with plasmid pPH151 for expression of the sufABCDSE genes (31), BW25113 and ΔtusA cells were transformed with a pET15-based and pACYC-based plasmids expressing ftsZ and rpoS, respectively.

Cloning of genes for EGFP and mCherry fusions.

Coding regions for rpoS, fis, and a fis variant in which all codons within the fis gene for Lys, Gln, and Glu were exchanged to the codon for Ala were synthesized as C-terminal fusions to EGFP vector pACYCDuet-1 under an IPTG-inducible promoter (Thermo Fisher). The mCherry gene was cloned under the control of an IPTG-inducible promoter into the vector pCDFDuet-1. Both plasmids were transformed into the respective BW25113 wild-type and mutant strains.

Cell morphology analysis.

Two milliliters of cell culture was taken at the respective growth times and harvested at 11,000 × g for 1 min. The pellet was resuspended in 1× phosphate-buffered saline (PBS; pH 7.4) and centrifuged at 14,000 × g for 1 min. This washing step was repeated 2 times. The pellet was resuspended in 1 ml 1× PBS, and 5 μl of the cells was transferred to a cleaned glass slide. Cell drops were dried at 30°C. To fix the cells on the glass slide, 1 drop of methanol (MeOH) was applied to the cells and incubated for 5 min at room temperature. The remaining MeOH was washed by dipping the slide 6 times in H₂O. After drying the slides for 5 min, 20 μl poly-l-lysine (5 μg/ml in 1× PBS) was pipetted on the cells to enhance the adhesion of cells to the glass slide. The DNA in the cells was labeled with 20 μl 4′,6-diamidino-2-phenylindole (DAPI; 1:1,000) (5 μg/ml in 1× PBS) and incubated 10 min at 8°C in the dark. The blue fluorescence was recorded with excitation at 358 nm and emission at 461 nm. Afterwards, the DAPI solution was washed with 200 μl 1× PBS and the cells were embedded in 20 μl Mowiol. Finally, the cells were covered with a cleaned coverslip and subjected to fluorescence microscopy (Axiovert 200M; Zeiss). A PlanApo oil lens objective with ×100 magnification and 1.4 numeric aperture (NA) was used for microscopy. The imaging of the cells was performed using an AxioCam MRm Rev. 3 camera and Axiovision 4.7 Rel software. ImageJ 1.8 (National Institutes of Health) was used as software for figure production.

Immunodetection.

Immunoblot analyses were performed to detect cellular levels of FtsZ, Fis, and RpoS in E. coli wild-type (WT) and deletion strains. Precultures of the E. coli strains were inoculated in M9 minimal medium with the appropriate antibiotics, and they were grown overnight at 37°C with 200 rpm. The next day, the cells were transferred to 50 ml of LB (0% NaCl) at a starting OD₆₀₀ of 0.05. For cell growth, the cells were cultivated at 37°C with 180 rpm for 5 h, 7 h, and 10 h (for RpoS), 5 h (for FtsZ), or 10 h (for Fis). After 5 h, 7 h, or 10 h of growth, the cells were harvested in 50-ml Falcon tubes at 8,000 × g for 5 min at 4°C. Cell pellets were washed 3 times with 25 ml 50 mM Tris-HCl (pH 7.5) and lysed in 3.5 ml 50 mM Tris-HCl, 150 mM NaCl, 0.5% (wt/vol) NP-40 (pH 8.0) by sonification. Protein concentration of the supernatant was determined by Bradford assay. Fifty micrograms of protein was loaded on SDS polyacrylamide gels (12% or 15%) and separated. Proteins were then transferred to a PVDF membrane (for FtsZ, Fis, and RpoS). After the transfer of proteins to the respective membranes, the membranes were washed in 1× TBST (50 mM Tris-HCl, 150 mM NaCl, 0,1% [wt/vol] Tween 20, pH 7.4) and blocked with 5% bovine serum albumin (BSA) overnight at 4°C. The primary antiserum was incubated with the membrane for 1 h at room temperature in 1× TBST. The incubation with FtsZ antiserum (rabbit, obtained from Miguel Vicente, Centro Nacional de Biotecnología, CSIC, Madrid) was performed at a dilution 1:20,000 for 1 h at room temperature. The incubation with Fis antiserum (rabbit, obtained from Reid C. Johnson, University of California) was performed at a 1:10,000 dilution at 4°C overnight. The incubation with RpoS antiserum (rabbit, obtained from BioLegend) was performed at a dilution 1:1,000 overnight at 4°C. The detection of FtsZ, Fis, and RpoS antibodies was performed by anti-rabbit IgG horseradish peroxidase (POD)-conjugated secondary antibody (goat; Sigma) at 1:10,000 dilution at room temperature for 1 h. The respective protein bands were detected by chemiluminescence using a 2-ml mix of solution A (100 μl 250 mM luminol, 44 μl 90 mM p-coumaric acid, 8.85 ml H₂O, 1 ml 1 M Tris-HCl, pH 8.5) and solution B (6 μl 30% H₂O₂, 9 ml H₂O, 1 ml 1 M Tris-HCl, pH 8.5). The bands were detected by using a FUSION FX7 instrument (Vilber). For determination of the size, a prestained protein molecular weight marker, 20 to 120 kDa (Fermentas), was used.

Flow cytometry of EGFP fusion proteins.

The translation levels of the EGFP fusion proteins were measured by flow cytometry. The E. coli BW25113 wild type and the respective mutant strains were DE3 lysogenized and transformed with rpoS-EGFP-pACYCDuet-1, fis-EGFP-pACYCDuet-1, fis^mut-EGFP-pACYCDuet-1, and mCherry-pCDFDuet-1 vectors. Expression of mCherry thereby served as an internal control for translation. In addition, the strains were also transformed with corresponding empty EGFP-pACYCDuet-1 and mCherry-pCDFDuet-1 vectors as an additional internal control. Precultures of the E. coli strains were grown in M9 minimal medium overnight at 37°C with 200 rpm. The next day, the cells were transferred to 50 ml of LB (0% NaCl) at a starting OD₆₀₀ of 0.05, and the cells were grown at 37°C with 180 rpm for 5 h. Expression of fusion proteins was induced with 100 μM IPTG at time point 0 h. After 5 h of growth, cell cultures were transferred to 50-ml Falcon tubes, and the OD₆₀₀ was determined. The cell count for each sample was set to 10⁸ cells/ml for flow cytometry. Five hundred microliters of cells in 1× PBS was subjected to flow cytometry. Each sample was detected for EGFP and mCherry fluorescence signal using a fluorescence-activated cell sorting (FACS) Melody system (Bioscience). In total, 10,000 cells were measured for each sample. EGFP was excited at 488 nm and mCherry at 587 nm. The fluorescence signal of EGFP was detected at 507 nm in the GFP channel, and fluorescence of mCherry was detected at 610 nm in the mCherry channel.