Reduced Protein Synthesis Fidelity Inhibits Flagellar Biosynthesis and Motility

Accurate translation of the genetic information from DNA to protein is maintained by multiple quality control steps from bacteria to mammals. Genetic and environmental alterations have been shown to compromise translational quality control and reduce fidelity during protein synthesis. The physiological impact of increased translational errors is not fully understood. While generally considered harmful, translational errors have recently been shown to benefit cells under certain stress conditions. In this work, we describe a novel regulatory pathway in which reduced translational fidelity downregulates expression of flagellar genes and suppresses bacterial motility. Electron microscopy imaging shows that the error-prone Escherichia coli strain lacks mature flagella. Further genetic analyses reveal that translational errors upregulate expression of a small RNA DsrA through enhancing its transcription, and deleting DsrA from the error-prone strain restores motility. DsrA regulates expression of H-NS and RpoS, both of which regulate flagellar genes. We demonstrate that an increased level of DsrA in the error-prone strain suppresses motility through the H-NS pathway. Our work suggests that bacteria are capable of switching on and off the flagellar system by altering translational fidelity, which may serve as a previously unknown mechanism to improve fitness in response to environmental cues.


Results
Mistranslation suppresses motility and flagellar assembly. To investigate the physiological impact of mistranslation, we previously engineered an E. coli error-prone strain by introducing a point mutation (I199N) into the chromosomal rpsD gene, which encodes a protein component of the ribosomal small subunit 29 . The resulting rpsD* strain (Table S1) displays 5-fold increased readthrough of the UAG stop codon compared to the parent strain MG1655, but does not show a decreased protein synthesis rate 29 . Mutations in the rpsD gene decrease accuracy during codon-anticodon pairing to cause global mistranslation of all mRNAs, and may decrease fidelity of initiation, elongation, and termination during protein synthesis 10 . RNA sequencing of rpsD* cells grown at 37 °C 29 reveals that flagellar assembly is the most significantly downregulated pathway compared to wild-type (WT) MG1655 (P = 1.9 × 10 −25 ). Because even WT MG1655 shows low expression of flagellar genes and slow motility at 37 °C, we tested the motility of WT and rpsD* strains at room temperature (25 °C). Our results showed that the rpsD* strain was defective in motility on soft agar plates (Fig. 1). The motility defect was rescued by either reverting the chromosomal rpsD* mutation or introducing a second mutation (K42N) in the rpsL gene to reduce translational errors (Fig. 1). The K42N mutation is located near the ribosomal A site and restricts pairing between codon and anticodon, and has been shown to increase decoding fidelity 37 . In addition to mistranslation caused by the rpsD* mutation, codon-specific mistranslation caused by addition of canavanine (an arginine analogue recognized by arginyl-tRNA synthetase and mistranslates arginine codons) also decreases motility (Fig. S1). Next, we used negative-staining electron microscopy to visualize the flagella of WT and rpsD* strains. Whereas WT cells contained multiple flagella per cell, most rpsD* displayed no mature flagella at all (Fig. 2). These results suggest that the motility defect caused by mistranslation is due to impaired flagellar assembly. Mistranslation decreases expression of flagellar genes. We next tested the expression levels of flagellar genes in the WT and rpsD* strains at 25 °C using quantitative reverse transcription polymerase chain reaction (qRT-PCR). The rpsD* mutation significantly decreased the mRNA levels of all tested flagellar genes, including flgB (encoding a flagellar basal-body rod protein), flgK (encoding a hook-filament junction protein), fliA (encoding Sigma 28 involved in synthesis of later-stage flagellar genes), fliF (encoding an MS-ring structural protein), and flhDC (encoding the master regulator of flagellar genes FlhD and FlhC) (Figs 3 and 4). Among these genes, transcription of flgB and fliA is dependent on the FlhDC complex, and flgK and fliF are controlled by both FlhDC and FliA 34,35 .
To determine how translational errors affect the protein level of FlhD, we inserted a Flag tag at the 3′ -end of the chromosomal flhD gene at the native locus. Western blot using an anti-Flag antibody revealed that the FlhD protein level decreased 60% in the rpsD* strain compared to the WT (Fig. 4B). We further showed that such decrease was not due to accelerated degradation (Fig. 4C), suggesting that mistranslation downregulates FlhD at the transcriptional and/or translational level.
Small RNA DsrA inhibits motility in error-prone strain. We have previously shown that translational errors activate the general stress response, which is controlled by RpoS 29 . The increase of RpoS level under error-prone conditions at 37 °C depends on a small RNA DsrA 29 . It has been suggested that RpoS negatively regulates expression of FliA and cell motility in E. coli 38 . We thus tested whether mistranslation suppresses motility through upregulation of RpoS. Deleting rpoS in the rpsD* strain was not able to restore motility (Fig. 5), suggesting that RpoS does not play a major role in flagellar synthesis under error-prone conditions. However, deleting dsrA fully rescued the motility defect of the rpsD* strain (Fig. 5). Consistently, overexpressing DsrA from a plasmid in the WT strain suppressed motility (Fig. 5). To test the role of DsrA in regulating expression of flagellar genes, we constructed a lacZ reporter under the control of flgB promoter. In line with the qRT-PCR results (Fig. 3), the activity of flgB promoter (controlled by FlhDC) decreased 60% in the rpsD* strain compared to the WT (Fig. 6). Deleting DsrA enhanced transcription of flgB to almost the same level as the WT. Addition of canavanine also decreased the activity of flgB promoter (Fig. S1B).
DsrA is induced at low temperatures (e.g., at 25 °C) through enhanced transcription and improved stabilization 39 . Using qRT-PCR, we found that the RNA level of DsrA was increased 3-fold by the rpsD* mutation at 25 °C (Fig. 7A). To further investigate how mistranslation enhances DsrA level, we tested transcription of dsrA using a yellow fluorescent protein reporter under the control of dsrA promoter. Transcription of dsrA promoter increased 2.5-fold in the rpsD* strain compared to the WT (Fig. 7B). Next, we determined the stability of DsrA by inhibiting transcription with rifampicin (Rif) and following the RNA level over time. The rpsD* mutation did not enhance the stability of DsrA (Fig. 7C), suggesting that the increase in DsrA RNA occurred at the transcriptional  level. Collectively, our data suggest that mistranslation elevates the DsrA RNA level, which in turn downregulates expression of flagellar genes and suppresses motility.

DsrA-mediated motility suppression depends on H-NS.
In addition to RpoS, another major target regulated by DsrA is H-NS 40 . We showed that deleting dsrA in the rpsD* strain restored motility (Figs 5 and 8A). In the absence of hns, deleting dsrA no longer increased motility of rpsD* cells (Fig. 8A). In contrast, deleting rpoS did not completely prevent the rescuing effect of dsrA deletion (Fig. 8A). To dissect the roles of the RpoS and H-NS pathways in regulation of motility by DsrA, we further took advantage of previously reported DsrA mutants that specifically impair regulation of rpoS (dsrA * R) or hns (dsrA * H) 41 . In the complementation assay, overexpressing WT DsrA or DsrA * R, both of which are able to inhibit H-NS activity, substantially reduced motility of the WT Δ dsrA strain (Fig. 8B). In contrast, overexpressing DsrA * H, which does not directly affect the H-NS pathway, showed only a minor decrease in motility (Fig. 8B).
DsrA regulates H-NS at the translational level 42 . In line with this, we found that the mRNA level of hns was unchanged by the rpsD* mutation (Fig. S2A). However, the activity of an H-NS repressed promoter (hdeA) increased significantly in the rpsD* strain (Fig. S2B), suggesting that the overall H-NS activity is lowered by the rpsD* mutation. In addition, the mRNA levels of flhDC were downregulated in the rpsD* strain (Fig. 4A), which is consistent with previous reports that H-NS stimulates transcription of flhDC 43 . Our results therefore suggest that DsrA regulates flagellar synthesis and motility mainly through the H-NS pathway.

Discussion
Bacteria utilize flagella for movement in the environment. Flagella are also used as bacterial mechanosensors to initiate biofilm formation 44 , and are important for virulence in many bacterial pathogens 32 . On the other hand, biosynthesis and functioning of flagella consume substantial cellular resources 45 , and flagella also activate the host immune response that inhibits and kills invading bacteria [46][47][48] . Flexible modulation of flagellar synthesis is thus important for bacterial adaptation to frequently changing natural environments. In this study, we demonstrate that reducing translational fidelity leads to reduced flagellar synthesis and loss of motility in E. coli. We have previously shown that reduced translational fidelity activates the general stress response, promoting bacterial survival under stress conditions 29 . Suppressing flagellar synthesis would allow cellular resources to be conserved for essential activities to maintain cell viability, e.g., synthesis of stress response effector proteins.
We show that mistranslation suppresses flagellar synthesis and motility through enhanced transcription of DsrA. DsrA is a small RNA found in multiple Gram negative bacteria, including Escherichia, Salmonella and Shigella. DsrA RNA level is significantly increased at low temperatures due to both increased transcription and decreased degradation 39 , and temperature regulation of dsrA transcription depends on complex promoter architecture 49 . Our results show that transcription driven by dsrA promoter is enhanced in the error-prone strain (Fig. 7). To date, the only known transcriptional regulator of DsrA is LeuO, which represses DsrA transcription 50 .     In our previous RNA sequence results 29 , LeuO mRNA level is increased in the rpsD* strain compared to the WT. Exactly how DsrA transcription is regulated by mistranslation remains to be clarified in the future. It is likely that another unknown transcriptional regulator of DsrA is affected by global protein mistranslation, e.g., through stabilization of a transcriptional activator due to titration of available proteases by an increased level of mistranslated proteins. It is also possible that mistranslation causes LeuO to misfold and lose its activity.
Small RNAs have been shown to regulate motility via diverse mechanisms 51 . Our data suggest that the effect of DsrA on bacterial motility requires H-NS instead of RpoS. DsrA blocks synthesis of H-NS protein by base pairing with the translational start site of its mRNA 42 . Increased expression of DsrA in the error-prone strain is thus expected to lower H-NS activity. We have used an H-NS repressed promoter hdeA as a reporter to test the activity of H-NS and show that the H-NS activity is suppressed in the rpsD* strain compared to the WT (Fig. S2). H-NS regulates a large number of genes, including activation of flhDC transcription 40,43 . A recent study also suggests that H-NS influences bacterial motility via FlhDC-independent pathways 52 . We show that overexpression of FlhC is sufficient to restore motility of the error-prone strain (Fig. S3), suggesting that mistranslation suppresses motility mainly through downregulation of flhDC in a process that requires DsrA and H-NS. Collectively, our results have revealed a previously unknown linkage between translational fidelity and flagellar synthesis, which may play an important role in bacterial adaptation to ever changing environmental conditions.

Materials and Methods
Strains, plasmids, growth conditions and reagents. Strains and plasmids used in this study are listed in Table S1, and the oligos are listed in Table S2. E. coli was grown in Lennox broth (LB) at 37 °C with agitation unless otherwise indicated. Antibiotics were used at the following concentrations: ampicillin (Amp), 100 μ g/ml; chloramphenicol (Chl), 25 μ g/ml. Antibiotics and other chemicals were purchased from Sigma-Aldrich (St. Louis, MO), and RNase-free DNase I was from Thermo Scientific (Rockford, IL).
Genome engineering of bacterial strains. All strains used in this study are derivatives of E. coli K-12 strain MG1655 (WT), which was obtained from The E. coli Genetic Stock Center at Yale University. All in-frame gene deletion mutants were constructed as described using chloramphenicol as the resistance marker 53 . All mutants were verified by PCR, and the antibiotic resistance marker was subsequently removed from the deletion strains using plasmid pCP20, which was cured at 42 °C afterwards. The marker-free deletion mutants were verified by both loss of resistance and PCR.
The fusion of 3 × FLAG tag at the 3′ end of flhD was conducted as follows. A cassette containing the toxin encoding gene ccdB under control of araBAD promoter and a kanamycin resistance gene (Ranquet et al., submitted, deposit patent number: FR11/60169, 08/11/2011, UJF/BGene) was amplified from template genomic DNA of CR201 strain (obtained from N. De Lay) using the primers FlhD-KN1 and FlhD-CCDB1, and introduced into chromosome by λ red recombinase-mediated gene replacement. The kan-ccdB cassette fused with flhD in the chromosome was then replaced with the gBlock fragment of 3 × FLAG tag (FlhD-FLAG, synthesized from Integrated DNA Technology). The successful recombinants (YF56 and YF57) were obtained by selection for growth in the presence of arabinose (1%) and verified by PCR.
Electron microscopy. Overnight culture of bacteria were diluted 1:100 into fresh LB and grown to OD 600 ~ 0.8 at 25 °C with agitation. Cells were collected and washed in 0.1 M NaCl, and resuspended in phosphate-buffered saline. To examine cells by electron microscopy, 7 μ l of culture was placed onto carbon-coated nickel grids (Electron Microscopy Sciences) for 1 minute, washed three times with sterile water and then negatively stained with 0.2% uranyl acetate for 30 seconds. The samples were visualized using a JEOL JEM-1400 electron microscope. Cells were randomly selected to count the number of flagella.
Swimming motility assay. Overnight culture of bacteria were diluted 1:100 into fresh LB and grown to OD 600 ~ 0.8 at 25 °C with agitation. All cultures were normalized to the same OD 600 before being spotted on freshly made tryptone broth (10 g/L of tryptone and 5 g/L of NaCl) plates containing 0.25% agar. For strains harboring plasmids, appropriate antibiotics were added into the tryptone broth motility plates. Plates were incubated at 25 °C overnight before taking pictures and measuring diameters of spots. The quantitative results represent the percentage of the diameter compared to that of the WT strain on the same plate.
Quantitative reverse transcription-PCR. Mid-log phase cells grown in LB medium at 25 °C was normalized to the same OD 600 and harvested. Total RNA was extracted using hot phenol and residual chromosomal DNA was removed as previously described 54 , except that glycogen was used to precipitate RNA samples. To test RNA degradation, freshly made rifampicin (250 μ g/ml final concentration) was added into normalized bacterial cultures to fully stop transcription at time zero.
Reverse transcription and quantitative PCR were performed using the iScript cDNA Synthesis Kit and the SsoAdvanced Universal SYBR Green Supermix Kit (Bio-Rad, Hercules, CA, USA) according to the manufacturer's instructions. 16S rRNA was used as an internal reference for normalization. The Δ Δ C t method was used to obtain the fold change of target genes in the mutant strains compared to those in the WT strains.

Determination of FlhD protein expression.
To determine expression of the FlhD, a 3 × FLAG tag was fused at the C terminal of flhD right before the stop codon in both WT and rpsD* strains. Mid-log phase cells grown in LB medium at 25 °C was normalized to the same OD 600 and harvested. Same volume of bacterial cultures was used to prepare total protein using the standard trichloroacetic acid/acetone protein precipitation protocol. For sample preparation to test FlhD degradation, freshly made chloramphenicol (100 μ g/ml final concentration) was added into normalized bacterial cultures to fully stop translation at time zero, and same volume of cultures Scientific RepoRts | 6:30960 | DOI: 10.1038/srep30960 was collected for protein preparation at specific time point. Western blot was performed according to standard procedures 55 using a primary anti-FLAG antibody.
Bacterial fluorescence protein and lacZ reporter. To measure the fluorescence intensity of reporter strains, overnight culture of bacteria was diluted to 0.01 OD 600 in LB. Cells were further grown in 96-well plates incubated at 25 °C in the plate reader (BioTek) with shaking. Both OD 600 and fluorescence were measured at 15 minute intervals for a total of 20 hours. Strains carrying pZS* 11 were used as positive controls to eliminate the differences of protein synthesis rate between different strains. For lacZ reporter measurement, β -galactosidase assay was conducted as described 24 .