Transcriptional and Translational Control of the Salmonella fliC Gene

Abstract

The flagellin gene fliC encodes the major component of the flagellum in Salmonella enterica serovar Typhimurium. This study reports the identification of a signal within the 5′ untranslated region (5′UTR) of the fliC transcript required for the efficient expression and assembly of FliC into the growing flagellar structure. Primer extension mapping determined the transcription start site of the fliC flagellin gene to be 62 bases upstream of the AUG start codon. Using tetA-fliC operon fusions, we show that the entire 62-base 5′UTR region of fliC was required for sufficient fliC mRNA translation to allow normal FliC flagellin assembly, suggesting that translation might be coupled to assembly. To identify sequence that might couple fliC mRNA translation to FliC secretion, the 5′ end of the chromosomal fliC gene was mutagenized by PCR-directed mutagenesis. Single base sequences important for fliC-dependent transcription, translation, and motility were identified by using fliC-lacZ transcriptional and translational reporter constructs. Transcription-specific mutants identified the −10 and −35 regions of the consensus flagellar class 3 gene promoter. Single base changes defective in translation were located in three regions: the AUG start codon, the presumed ribosomal binding site region, and a region near the very 5′ end of the fliC mRNA that corresponded to a potential stem-loop structure in the 5′UTR. Motility-specific mutants resulted from base substitutions only in the fliC-coding region. The results suggest that fliC mRNA translation is not coupled to FliC secretion by the flagellar type III secretion system.

Type III secretion (T3S) is one of several mechanisms designed to secrete proteins from the cytoplasm of bacterial cells. T3S is utilized in flagellar construction and in the delivery of virulence determinants to eukaryotic cells by the injectisomes of gram-negative pathogens (13). The flagellum of Salmonella enterica serovar Typhimurium is composed of a rigid, helical filament of about 20,000 identical flagellin subunits that extends up to 10 microns from the cell (26). The filament is attached to an ion-powered rotary motor by the hook and rod structures of the flagellar hook-basal body. The hook is a flexible structure acting as a universal joint to transmit rotational energy from the motor to the filament (7). The rod acts as the drive shaft, which penetrates through the peptidoglycan and lipopolysaccharide layers (20). For the flagellum, the T3S apparatus functions to secrete components including the rod, hook, and filament subunits for extracellular assembly. The core of the flagellum is hollow, and secreted subunits polymerize at the tip of the growing flagellum (32, 35, 36). A cap at the tip of the flagellum ensures efficient polymerization of secreted subunit proteins (34).

There are three components that contribute to efficient secretion by T3S systems. An N-terminal peptide secretion signal, disordered in structure, is absolutely required (27). A T3S-chaperone is sometimes required for efficient secretion of specific substrates. T3S-chaperones bind directly to their cognate substrates and act either to protect secretion substrates from proteolytic degradation in the cytoplasm, to facilitate the secretion of specific substrates, or both to protect from degradation and to facilitate secretion (6). The flagellar FlgN chaperone has been shown to bind directly to the secretion apparatus (33). The third component of substrate recognition by T3S systems is an mRNA secretion signal and has been the subject of controversy for many years (15). The type II Sec-translocase had been shown to target periplasmic and outer-membrane proteins for export by a well-defined N-terminal amino acid secretion signal (29). The N terminus of T3S substrates was also required for secretion, but the lack of a conserved amino acid sequence in the N-terminal secretion signal led to the proposal of an elegant mechanism that the mRNA within the N-terminal sequence could be the secretion signal and that T3S occurred by a cotranslation secretion mechanism (4, 5).

Our recent studies of the flagellar assembly pathway of Salmonella enterica serovar Typhimurium demonstrate that the efficiency of protein secretion through the flagellum can be affected by the 5′ untranslated region (5′UTR) sequence of the mRNA (17). This led to the model that translation of secreted flagellar proteins might be localized to the cytoplasmic base of the flagellum-specific secretion apparatus (1). Since the flagellin gene fliC encodes the major component of the flagellum (36), we predict that efficient FliC secretion is a strong candidate to be influenced by signals encoded in the fliC mRNA. Therefore, we explored the possibility that the fliC gene might contain regulatory signals in its 5′UTR region that couples at least transcription to translation. In the present study, we found that signals in the 5′UTR of the fliC transcript were required for the efficient assembly of flagellin on the external surface of the cell. This suggested that sequences in the 5′UTR of fliC might be required for targeting of the fliC transcript to the flagellum for localized translation. We set up a screen following targeted mutagenesis of the 5′ end of the fliC gene of motility-defective mutants. The motility-defective mutants were categorized as resulting in defects specific to transcription, translation, or motility. The goal was to identify any potential mRNA targeting signals as sequences in the 5′UTR of the fliC transcript that were specific to motility independent of transcription or translation. After an exhaustive screen and sequence analysis of hundreds of motility-defective mutants in the 5′ end of the fliC gene, no motility-specific mutants were found in the 5′UTR region. All motility-specific mutants were in the fliC-coding region. A novel result that came from this study was the discovery of a potential stem-loop region in the 5′UTR that appeared to be required for fliC translation. However, in a separate study, we show that this stem-loop structure in the fliC 5′UTR functions to inhibit fliC mRNA translation and that the mutants isolated in the present study enhance this translation-inhibitory activity (31).

MATERIALS AND METHODS

Media and standard genetic manipulations.

Media, growth conditions, transductional methods, and motility assays were as described previously (10, 11). Strains containing the pKD46 plasmid were grown at 30°C in Luria-Bertani medium with aeration in the presence of 100 μg of ampicillin/ml (Sigma, St. Louis, MO) (9) (Table 1). The λ-Red genes were induced approximately 1 h prior to preparation of electrocompetent cells by the addition of arabinose to 0.2% (wt/vol). The expression of the fliC::MudJ and fliC::MudK constructs was determined by using X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside; 0.01%) MacConkey-lactose and tetrazolium-lactose indicator plates (18).

TABLE 1.

List of S. enterica serovar Typhimurium strains

Strain	Genotype	Source or referencea
LT2	Wild type	John Roth
TH714	fljB5001::MudJ	11
TH4753	fliC5532::tetRA fljB5001::MudJ
TH4754	fliC5533::tetRA fljB5001::MudJ
TH6232	Δhin-5717::FRT (fljBAON)	8
TH6299	Δhin-5717::FRT fliC5050::MudJ ΔfliC5533::tetRA	8
TH6301	Δhin-5717::FRT fliC5469::MudK ΔfliC5533::tetRA	8
TH6303	pKD46/Δhin-5717::FRT fliC5532::tetRA
TH6305	pKD46/Δhin-5717::FRT fliC5050::MudJ fliC5532::tetRA
TH6307	pKD46/Δhin-5717::FRT fliC5469::MudK fliC5532::tetRA
TH6699	fliC5747::Tn10dTc fliC5050::MudJ
TH6700	fliC5747::Tn10dTc fliC5469::MudK
TH6705	fliC5747::Tn10dTc

^aUnless indicated otherwise, all strains were constructed during the course of this study.

β-Galactosidase assays.

β-Galactosidase assays were performed in triplicate on mid-log-phase cells as described previously (23). β-Galactosidase activities are expressed as nanomoles/minute/(optical density at 650 nm.

Immunoblot analysis.

Immunoblot analysis of FliC levels has been previously described (8). All assays were performed in triplicate on protein samples taken from mid-log-phase cultures.

DNA sequence analysis.

DNA or PCR products were sequenced using ABI Prism BigDye Terminator Cycle Sequencing Ready Reaction Kits as described previously (PE Applied Biosystems, California). The ABI BigDye Terminator reactions were run at the Biochemistry DNA Sequencing Facility, University of Washington, Seattle.

Predicted RNA secondary structures.

RNA secondary structures were predicted by using the M-FOLD program available online (M-FOLD, version 3.1 [25]; http://www.bioinfo.rpi.edu/applications/mfold/). We analyzed two different partial RNA secondary structures. One structure included the fliC +1 translational start site and, 12 nucleotides (nt) downstream, the AUG start codon. The second structure included the fliC +1 translational start site, the full length of the 5′UTR (62 nt), and 42 nt downstream the AUG initiation codon. In the wild-type strain TH6232 the examined RNA total length was 77 nt for the first structure and 108 nt for the second one. M-FOLD analyses were conducted at a fixed temperature of 37°C.

Transcriptional start site mapping.

The FliC primer extension oligonucleotide (5′-CAGGTTATTCTGGGTCAACAGCGACAGGC-3′), which is complementary to the promoter proximal end of the fliC gene, was used to map the transcriptional start site. RNA was purified as previously described (12) from strains TH2592 (fljB5001::MudJ), TH2151 [fla-2039(Δtar-flhD) fljB5001::MudJ], and TH3770 (ΔflgM5301 fljB5001::MudJ). Sequencing reactions were performed by using pJK124 as the template and the FliC primer extension oligonucleotide.

Isolation of tetRA insertion/replacements of the fliC 5′UTR sequences.

The divergently transcribed tetR and tetA genes (referred to as tetRA) were amplified from a Tn10dTc insertion in strain TH2788 (Tn10dTc inserted 89 bases upstream of the fliY start codon and has no effect on motility) with the primers FliCtetR (5′-CGGGGGAAGTGAAAAATTTTCTAAAGTTCGAAATTCAGGTTTAAGACCCACTTTCA-3′) and FliC(start)tetA (5′-AACAGCGACAGGCTGTTTGTATTAATGACTTGTGCCATGACTAAGCACTTGTCTCC-3′), with FliCtetR and FliC(SD)tetA (5′-GGCTGTTTGTATTAATGACTTGTGCCATGATCTTTTCCTTCTAAGCACTTGTCTCCTG-3′) and with FliCtetR and FliC+1/tetA (5′-CTTGATGTTATTGGGCTGTTGCCCACGGTTTCTCACCGTCTAAGCACTTGTCTCCTG-3′). These amplified fragments were introduced into strain TH714 expressing the λ-Red recombination functions from plasmid pKD46 by electroporation, selecting for tetracycline resistance. The primers are flanked by fliC sequence such that selection of the fliC-flanked tetRA sequences resulted in the replacement of the −10 region of the fliC promoter to sites 2 (fliC5533::tetRA), 12 (fliC5532::tetRA), or 62 (fliC5569::tetRA) bases upstream of the ATG start codon, resulting in tetA-fliC operon fusions with 2, 12, or 62 bases of the fliC 5′UTR between tetA and fliC (Fig. 1). For simplicity, these three constructs are described from here on as ΔP_fliC::tetRA-2, ΔP_fliC::tetRA-12, and ΔP_fliC::tetRA-62.

FIG. 1.

Explanation of the strategy used to create tetA-fliC operon fusions at the fliC chromosomal locus. The tetR-tetA region from transposon Tn10dTc was amplified with primers whose 5′ sequences were homologous to the sites of insertion in fliC and whose 3′ sequences were homologous to the ends of either tetR or tetA. After amplification, tetRA fragments flanked by 40 bases of fliC sequences were recombined into the chromosome by λ-Red recombination as described previously (9). For all three constructs the 5′ recombination site is identical, replacing the 5′ end of the −10 sequence of the fliC promoter (5′-GCCGATAC-3′), whereas the 3′ recombination site varied (recombinations A, B, or C) depending on how much of the fliC transcript was still present. The ΔP_fliC::tetRA-2 construct resulting from recombination A, also known as fliC5533::tetRA, deletes the fliC promoter and 5′UTR through to base −3 before the fliC AUG codon with the tetRA cassette. For ΔP_fliC::tetRA-12, the allele, fliC5532::tetRA, required recombination B and is a tetRA replacement starting at the same point as for fliC5533::tetRA, but this time through to base −13 before the fliC AUG codon. For ΔP_fliC::tetRA-62, the 3′ end point of the allele, fliC5569::tetRA, resulting from recombination C is at base −62 before the fliC AUG codon. For visualization of the resulting transcripts from these tetA-fliC operon fusions, the tetA transcript is shown in light gray and the fliC transcript is dark gray.

Localized mutagenesis of the fliC promoter-UTR region.

Three primers were designed for the doped mutagenesis of the fliC 5′UTR region (MWG Biotech): primer fliC1 (5′-ACCAGGGTTAcggtgagaaaccgtgggcaaCAGCC-3′, primer fliC2 (5′-GGGCAAcagcccaataacataacatcaagttGTAATT-3′), and primer fliC3 (5′-CAAGTTgtaattgataaggaaaagatcATGGCA-3′), in which the lowercase bases were doped with 1% of each of the other three bases (Fig. 2A). These were used to amplify sequences 3′ to the 5′UTR with the primer fliC/fljB13 (5′-CGCTGCAGGTTGTTGTTG-3′), located about 250 bases into the fliC structural gene. The amplified fragments were cleaned up with a QIAGEN PCR purification kit and used as the second primer to amplify sequences 5′ to the 5′UTR with the primer fliC13 (5′-GTTCTTTGTCAGGTCTGTC-3′), located about 250 bases upstream of fliC (∼60 bases within the fliD structural gene) (Fig. 2A). The resulting fragments (FliC-1, FliC-2, and FliC-3) were cleaned up with a QIAGEN PCR purification kit and contained mutagenized fliC 5′-UTR sequences flanked by about 272 (3′) and 280 (5′) bases of homology to allow for recombination. In the first set of mutagenesis, these donor fragments were electroporated separately into strains TH6303, TH6305, and TH6307, selecting for tetracycline sensitivity (24) (Fig. 2B), whereas for three further rounds of mutagenesis only TH6303 was used. The tetracycline-sensitive (Tc^s) electroporants were screened for Mot⁻. These λ-Red-mediated recombination events resulted in the replacement of the ΔfliC5532::tetRA insertion with the mutagenized fliC 5′UTR (Fig. 2B).

FIG. 2.

Strategies used for the isolation and characterization of mutants in the 5′ region of the fliC gene. (A) PCR strategies used during the isolation of random and targeted fliC 5′UTR mutants. The outside primers fliC/fljB13 and fliC13 stayed consistent throughout all mutagenesis. Internal primers were either doped (fliC1, fliC2, and fliC3) or possessed specifically designed mutations leading to the double −45G:C −38C:G mutant. (B) The generation of mutants in the 5′ region of the fliC gene was carried out by replacement of the ΔP_fliC::tetRA-12 (ΔfliC5532::tetRA) allele with a PCR-mutagenized DNA fragment covering this region and selecting Tc^s recombinants in the presence of the λ-Red recombinase. (C) Motility-defective mutants in the 5′ region of the fliC gene were screened for effects on expression of a lac transcriptional or translational reporter by recombining in either a fliC::MudJ (transcriptional reporter) or a fliC::MudK (translational reporter) located downstream in the fliC gene. Because the donor fliC::Mud strains also have an upstream fliC::Tn10dTc insertion, selection for Mud-encoded Km^r and screening for retention of the Tc^s recipient genotype ensures that the motility-defective mutation is retained in the recombinants.

To isolate motile revertants of the nonmotile fliC::tetRA insertions, 20 independent colonies from TH4754 (ΔP_fliC::tetRA-2) and 22 from TH4753 (ΔP_fliC::tetRA-12) were stabbed into motility agar and incubated at 37°C. These strains possess an insertion in the alternative flagellin fljB to ensure that a motile phenotype is due to fliC-dependent motility. After overnight incubation, motile revertants were picked and purified by two successive single colony isolations on nonselective LB plates and retested for motility. The ΔP_fliC::tetRA-2 and ΔP_fliC::tetRA-12 deletion-insertion mutations were then transduced into a wild-type background.

Introduction of fliC.

MudJ and fliC::MudK insertions downstream of fliC 5′UTR mutant alleles. The fliC5747::Tn10dTc insertion is located near the 5′ end of the fliC coding region after bp 79 from the translational start site of fliC (Fig. 2C). fliC::MudJ transcriptional (11) and fliC::MudK (8) translational reporters were introduced into strain TH6705 (fliC5747::Tn10dTc) by P22 generalized transduction, selecting Mud-encoded Km^r and screening Tc^r to generate TH6699 (fliC5747::Tn10dTc fliC5050::MudJ) and TH6700 (fliC5747::Tn10dTc fliC5469::MudK), respectively. TH6699 and TH6700 were then used as donors in P22-mediated transductions into the Mot⁻ mutants generated by directed mutagenesis of the fliC 5′UTR region selecting for kanamycin-resistant (Km^r) and screening for Tc^s by replica printing the l-Km selection plates onto Tc^s selection plates containing Km (to hold selection) (Fig. 2C). Because the Mud insertions are located in the 3′ half of the fliC structural gene and the Tn10dTc is located near the 5′ end of the fliC structural gene, Km^r Tc^s transductants must occur by recombination (see Fig. 2C) between the points of insertion of the Mud and Tn10dTc transposons so that the fliC 5′UTR mutant region of the recipient remains in the recombinant. Two Km^r Tc^s transductants were kept from each cross and examined on indicator media for lac expression.

Targeted mutagenesis of a stem-loop region in the fliC 5′UTR region.

The double mutant −45G:C −38C:G was constructed by the PCR-based technique of GenSOEing (2). Primers fliC/fljB13 and fliC13 were used as “outside” primers (Fig. 2A) with the internal primers FliCUTRSL2GCF (5′-AGAAACCGTGGCCAACAGGCCAATAACATC-3′) and FliCUTRSL2GCR (5′-GATGTTATTGGCCTGTTGGCCACGGTTTCT-3′), respectively, for a first round of PCR with LT2 DNA as a template. The resulting PCR products were purified with the QIAGEN PCR clean-up kit and used in a second round of PCR with templates mixed at approximately a 1:1 ratio. The primers for the second round were only fliC/fljB13 and fliC13. This generated a 622-bp PCR product that was isolated from a 1% agarose gel and used to replace the fliC5532::tetRA allele in TH6303 by λ-Red recombination (Fig. 2B). As internal controls, the single mutations −38C:G and −45G:C were created by using the internal primers FliCUTRSL2GF (5′-AGAAACCGTGGGCAACAGGCCAATAACATC-3′), FliCUTRSL2GR (5′-GATGTTATTGGCCTGTTGCCCACGGTTTCT-3′), FliCUTRSL2CF (5′-AGAAACCGTGGCCAACAGCCCAATAACATC-3′), and FliCUTRSL2CR (5′-GATGTTATTGGGCTGTTGGCCACGGTTTCT-3′). All mutations were confirmed by phenotypic and DNA sequence analysis.

RESULTS

The fliC transcriptional start site. As a first step in the characterization of the 5′ end of the fliC gene, the transcriptional start site was determined by primer extension analysis (see Materials and Methods). RNA samples were isolated from wild-type and ΔflhDC and flgM mutant strains and used to determine the start of transcription. The flhDC operon is required for all flagellar gene expression and thus serves as a negative control (19). FlgM is an inhibitor of σ²⁸-dependent fliC transcription (28). A null mutation in flgM results in increased σ²⁸-dependent transcription and represents a positive control.

The results of the primer extension analysis are shown in Fig. 3. No primer extension product was observed in the ΔflhDC control strain, and products were increased in the flgM mutant strain. We observed adjacent transcriptional start sites 63 (T) and 62 (A) bases upstream of the initiator AUG codon, with the majority of the transcription starting at −62 (A) relative to the AUG (Fig. 3). These start sites correspond to 11 and 12 bases downstream from the center of the −10 region of a deduced fliC promoter based on the consensus σ²⁸-dependent flagellar promoter sequence (14). This verifies that the fliC promoter identified by its strong consensus sequence for σ²⁸ promoters (14) and position upstream of the fliC coding region is, in fact, the fliC promoter.

FIG. 3.

Transcriptional start site mapping for the serovar Typhimurium fliC gene. (A) Lanes C, T, A, and G are the sequencing ladder. Primer extensions were measured in isogenic flgM (TH3770 = ΔflgM5301 fljB5001::MudJ), wild-type (WT; TH2592 = fljB5001::MudJ), and ΔflhDC [TH2151 = fla-2039(Δtar-flhD) fljB5001::MudJ] strains. The bases shown in italics are the transcriptional start sites. (B) The sequence upstream of fliC is shown, with the predicted flagellar promoter consensus sequences underlined. The transcription and start codons are also shown in italics.

Effect of fliC 5′UTR sequence on fliC expression from tetA-fliC operon fusions.

Previous data show that fliC expression is regulated at both the transcriptional level and the translational level (8). The identification of a 62-bp 5′UTR in the fliC transcript by promoter mapping leads to the prediction that the fliC 5′UTR may be required not only for the transcription and translation of fliC but for the proper assembly of FliC subunits into the external filament structure. To test this possibility, the fliC promoter and 5′UTR region were replaced with the tetA promoter at the fliC chromosomal locus (Fig. 1). Recent advances using the bacteriophage λ recombination system (λ-Red) has allowed for the construction of precise, targeted changes in the bacterial chromosome using homologies of as little as 30 bp (9, 37). A segment of DNA containing the divergently transcribed tetR and tetA genes from transposon Tn10, hereafter referred to as tetRA, was recombined into the fliC promoter region, resulting in three insertion-deletion constructs (ΔP_fliC::tetRA-2, ΔP_fliC::tetRA-12, and ΔP_fliC::tetRA-62) (Fig. 1). The deletion constructs all have the same upstream endpoint and remove the −10 sequence of the fliC promoter, while the downstream deletion endpoints vary and are either 2 bases prior to the AUG (ΔP_fliC::tetRA-2), 12 bases prior to the AUG (ΔP_fliC::tetRA-12), or just prior to the +1 position of the fliC transcript (ΔP_fliC::tetRA-62) (Fig. 1). The tetA gene product confers resistance to tetracycline and is repressed by the TetR repressor. In the presence of tetracycline, TetR no longer represses. Thus, these constructs place the fliC gene under the control of the tetA promoter, which is induced by the addition of tetracycline. We have effectively created tetA-fliC operons at the fliC locus such that fliC transcription is induced by the addition of tetracycline to the growth medium.

The three ΔP_fliC::tetRA constructs were placed upstream of either a fliC-lac transcriptional (operon) fusion construct, a fliC-lacZ translational (gene) fusion construct, or the wild-type fliC structural gene. Transcription, translation, FliC protein levels, and motility were then analyzed after induction of the ΔP_fliC::tetRA constructs by tetracycline (Fig. 4). No construct produced fliC expression comparable to the wild type, although ΔP_fliC::tetRA-62 did have a motile phenotype. The differences in fliC expression can in part be explained by the fact that fliC is now in a multicistronic operon with tetA. For fliC transcription, ΔP_fliC::tetRA-62 had a 1.4-fold reduction in transcription compared to the wild type, while ΔP_fliC::tetRA-2 and ΔP_fliC::tetRA-12 were reduced 4.3- and 3.1-fold, respectively (Fig. 4: fliC-lac). A much stronger effect was observed for fliC translation. For ΔP_fliC::tetRA-2, a 167-fold reduction in fliC translation was observed, presumably resulting from the lack of a defined ribosomal binding site in the tetA-fliC intergenic region (Fig. 4, fliC-lacZ). The construct ΔP_fliC::tetRA-12 also showed a much stronger decrease in translation (55-fold) compared to transcription, but here the native ribosomal binding site region was included. These results are consistent with previous findings that showed regulation of fliC translation is stronger than that exhibited at the transcriptional level (8).

FIG. 4.

Effect of fliC 5′UTR sequences in tetA-fliC operon fusions on transcription, translation, and assembly. The three fliC 5′UTR tetRA replacements described in Fig. 1 were used to determine the extent to which the 5′UTR is required for fliC expression and assembly into the flagellar filament. The four columns represent the fliC transcription and translation levels, the intracellular concentration of FliC, and the motility of all three constructs compared to wild-type. No tetA-fliC operon fusion possessed a fliC expression profile comparable to the wild type. ΔP_fliC::tetRA-2 had very little translational activity leading to low levels of the FliC protein, and ΔP_fliC::tetRA-12 had just a bit more. The ΔP_fliC::tetRA-62 construct was translated at high levels and possessed significant levels of FliC. The ΔP_fliC::tetRA-12 possessed a nonmotile phenotype compared to the wild type and ΔP_fliC::tetRA-62. Strains used during this analysis included the following. For fliC-lac transcription: wild type, TH5947 [hin-5717::FRT-Cm-FRT (fliC^ON) fliC5050::MudJ]; ΔP_fliC::tetRA-2, TH6299 [hin-5717::FRT-Cm-FRT (fliC^ON) fliC5533::tetRA fliC5050::MudJ]; ΔP_fliC::tetRA-12, TH6293 [hin-5717::FRT-Cm-FRT (fliC^ON) fliC5532::tetRA fliC5050::MudJ]; and ΔP_fliC::tetRA-62, TH6781 [hin-5717::FRT-Cm-FRT (fliC^ON) fliC5569::tetRA fliC5050::MudJ]. For fliC-lacZ translation: wild type, TH5951 [hin-5717::FRT-Cm-FRT (fliC^ON) fliC5469::MudK]; ΔP_fliC::tetRA-2, TH6301 [hin-5717::FRT-Cm-FRT (fliC^ON) fliC5533::tetRA fliC5469::MudK]; ΔP_fliC::tetRA-12, TH6295 [hin-5717::FRT-Cm-FRT (fliC^ON) fliC5532::tetRA fliC5469::MudK]; and ΔP_fliC::tetRA-62, TH6782 [hin-5717::FRT-Cm-FRT (fliC^ON) fliC5569::tetRA fliC5469::MudK]. For FliC^IN and motility: wild type, TH5947 [hin-5717::FRT-Cm-FRT (fliC^ON)]; ΔP_fliC::tetRA-2, TH6297 [hin-5717::FRT-Cm-FRT (fliC^ON) fliC5533::tetRA]; ΔP_fliC::tetRA-12, TH6291 [hin-5717::FRT-Cm-FRT (fliC^ON) fliC5532::tetRA]; and ΔP_fliC::tetRA-62, TH6111 [hin-5717::FRT-Cm-FRT (fliC^ON) fliC5569::tetRA].

Upon comparison of intracellular FliC protein levels, ΔP_fliC::tetRA-12 produced significantly less FliC than ΔP_fliC::tetRA-62 and ΔP_fliC::tetRA-2 did not produce FliC (data not shown), which was consistent with the translation results (Fig. 4). Only ΔP_fliC::tetRA-62 containing the full-length 5′UTR of fliC was able to assemble enough FliC subunits to produce motility. These data suggested that FliC could be produced independently of the fliC promoter and 5′UTR, but sequences in the 5′UTR are important for fliC translation and possibly for secretion and assembly into external filaments.

Analysis of motile revertants of ΔP_fliC::tetRA-2 and ΔP_fliC::tetRA-12 constructs.

The low level of fliC-lacZ translation observed for both the ΔP_fliC::tetRA-2 and the ΔP_fliC::tetRA-12 constructs suggests that ribosomes do not bind this region effectively. It is possible that the entire ribosome binding site is not included in the 12 bases immediately preceding the fliC AUG start codon. The recommended strong ribosome binding site for E. coli is UAAGGAAG (30). Our construct included the sequence AAGGAAAAAGAUC and lacked the U that is present normally at position −13 relative to the AUG start codon, which may have been important for fliC translation. We isolated and characterized motile revertants of the ΔP_fliC::tetRA-2 and ΔP_fliC::tetRA-12 constructs to see what changes would restore fliC mRNA translation and motility. Of 42 motile revertants analyzed, 12 of 20 from ΔP_fliC::tetRA-2 parent strain and 13 of 22 from the ΔP_fliC::tetRA-12 parent strain showed 100% linkage of Tc^r to motility. The 25 revertants showing 100% linkage were sequenced for the entire tetRA region through the first 30 bases of fliC. The results are presented in Table 2 and Fig. 5. For ΔP_fliC::tetRA-2, only 8 of the 12 mutants had sequence changes in the region sequenced, resulting in five different mutations (Fig. 5A). In this construct, the AUG start codon is within the stem of a potential stem-loop structure, and six of the eight sequenced changes would result in a disruption of the stem. Another mutation, isolated twice, was due to a C-to-A (C:A) substitution 46 bases upstream of the tetA stop codon and was significantly more motile than the other revertants, but it is not clear how this substitution affects fliC translation. For ΔP_fliC::tetRA-12, of the 13 revertants, sequence analysis identified 7 individual mutations (Fig. 5B). Four deletions allow readthrough into fliC, resulting in a stop codon starting at base 15 of the fliC coding sequence. One insertion of 26 bases allows readthrough ending at a stop codon starting at base 38 of the fliC coding sequence. The other two resulted in G:U substitutions, one of which was 10 bases upstream of the tetA stop codon and the other was the base for fliC codon 2, resulting in an A2S amino acid substitution. All increased fliC translation and presumably increase ribosome binding upstream of fliC.

TABLE 2.

Summary of mutations isolated as tetA-fliC motile revertants

tetA-fliC construct	Gene	Mutation		Motility
		No. of isolates	Base (amino acid)
LT2	Wild type			+++
DPfliC::tetRA-2	tetA	2	C:A (F387L)	+++
	tetA	1	G:T (A401S)	+
	tetA	3	C:T (A401V)	+
	tetA	1	Δ(bp 1197 to 1200 from ATG)	+
	fliC	1	C:A (A2E)	++
DPfliC::tetRA-12	tetA	7	G:T (E398D)	+
	tetA (Tcs)	1	Δ(bp 427 to 1205 from ATG)	++
	tetA (Tcs)	1	Δ(bp 473 to 1200 from ATG)	++
	tetA (Tcs)	1	Δ(bp 1088-1206 from ATG)	+++
	tetA (Tcs)	1	Δ(bp 1193-4 from ATG)	+
	fliC 5′UTR	1	INS(26 bp −12 from ATG) (identical to bp 126 to 151 of fliD)	++
	fliC	1	G:T + four from ATG (A2S)	+++

FIG. 5.

Motile revertants from ΔP_fliC::tetRA-2 and ΔP_fliC::tetRA-12 constructs. DNA sequence analysis of the tetRA sequence and fliC translation initiation regions of isolated motile revertants. The fliC AUG start codon is boxed and underlined, and the tetA stop codon is underlined. (A) Mutations isolated for ΔP_fliC::tetRA-2. (B) Isolated mutations from the ΔP_fliC::tetRA-12 construct. The mutations from mutants selected for increased motility starting with the G:U substitution mutant 10 bases upstream of the tetA stop codon are in parentheses.

One of the motile revertants from the ΔP_fliC::tetRA-12 construct, resulting in a G:U substitution 10 bases upstream of the tetA stop codon, showed weak motility and was used to select for secondary mutations that increased its motility. From 20 independent colonies stabbed into motility-Tc plates, 12 independent motile revertants were isolated, and 9 were characterized by DNA sequencing. One had no change in the region sequenced; an A:G change 7 bases upstream of the fliC AUG start codon was isolated twice; four base substitutions of either G:U (twice) or G:A (twice) at position +4 in the fliC coding region resulting in A2S and A2T amino acid changes, respectively; and an A:U base substitution at position +6 in the fliC coding region resulting in no amino acid change was isolated twice (Fig. 5B [in parentheses]).

Isolation and characterization of mutants in the 5′ end of the fliC flagellin gene.

The sequence analysis of motile revertants of the ΔP_fliC::tetRA-2 and ΔP_fliC::tetRA-12 constructs were consistent with the mechanism that ribosomes do not bind upstream of the fliC gene effectively in these constructs. However, it was recently reported that the 5′- and 3′UTRs of the fliC gene in combination would target foreign proteins for secretion through the flagellum independent of the FliC amino acid secretion signal (21). We decided to target the 5′ end of the fliC gene and determine by mutant analysis sequences critical for transcription, translation, and assembly. For this analysis it was necessary to develop a two-step genetic screen (Fig. 2). First, mutants were isolated and screened for motility defects (Fig. 2B). Second, nonmotile mutants were further screened with Mud-lac transcriptional and translational reporter constructs in fliC (Fig. 2C) in order to categorize the mutants according to those affecting transcription, translation, or motility. Our mutagenesis strategy for the 5′UTR region used doped oligonucleotides and is outlined in Fig. 2A (see Materials and Methods). The resulting 622-bp mutagenized DNA fragment including the doped fliC 5′UTR and flanking regions was recombined onto the chromosome under λ-Red induction conditions using the plasmid pKD46 (9) (Fig. 2B). Our recipient strain for the PCR-mutagenized sequences was TH6303(pKD46/Δhin-5717::FRT fliC5532::tetRA), selecting for Tc^s recombinants that replaced the fliC5532::tetRA (ΔP_fliC::tetRA-12) allele. Tc^s recombinants were then screened for a nonmotile phenotype (Fig. 2B). The Δhin-5717::FRT allele of TH6303 prevents flagellar phase variation and locks this strain into the fliC^ON orientation (8). As a result, any nonmotile Tc^s recombinants isolated must alter fliC expression or secretion. Mutagenesis performed throughout the present study allowed for the isolation of approximately 200 nonmotile mutants that were analyzed for fliC transcription, translation, and secretion.

To classify the nonmotile mutants in the 5′ end of the fliC gene according to those affecting transcription, translation, or motility, insertions of the MudJ and MudK transcriptional and translational reporters, respectively, were recombined into the fliC gene (Fig. 2C). Transcription and translation of fliC was determined by measuring the β-galactosidase activities of the resulting strains. Examples of mutants defective in translation are shown in Fig. 6. These mutants exhibit wild-type expression of a lac transcriptional reporter (fliC-lac) but are defective in expression of a fliC-lacZ translational reporter (FliC-LacZ). In addition, the nonmotile phenotype of the original mutants was confirmed after two passages on nonselective agar plates in comparison to the wild-type strain LT2 on soft agar motility plates (data not shown). Figure 7 summarizes the results of the mutagenesis performed, highlighting 44 of the mutants obtained. Mutants classified as motility specific were those showing levels of β-galactosidase comparable to the wild-type sequence upstream of the fliC::MudJ and fliC::MudK lac transcriptional and translational reporters but reduced motility on soft agar motility medium. Mutants classified as translation specific were those showing both reduced motility and reduced levels of β-galactosidase when combined with the fliC::MudK lac translational reporter but wild-type levels of β-galactosidase when combined with the fliC::MudJ lac transcriptional reporter. Mutants classified as transcription specific showed reduced motility and reduced β-galactosidase when combined with either the fliC::MudJ or the fliC::MudK reporter.

FIG. 6.

fliC expression profiles using fliC-lac and fliC-lacZ fusions of selected mutations isolated during the present study. The transcription and translation activities of selected mutations, described in Fig. 7, are shown. The −9 and −10 substitutions are located within the native fliC ribosome binding site, whereas the −36 to −45 substitutions all lie within a predicted stem-loop structure. All of the mutants belong to a class of mutants that are nonmotile have reduced translation but normal transcription.

FIG. 7.

Mutations affecting fliC transcription, translation and FliC-dependent motility. (A) Single DNA base substitutions in the fliC coding region led to amino acid substitutions that were defective in motility. The effect of the D43G, A64T, and N83S mutants on motility in a strain expressing fljB⁺ and deleted for fljA is also shown. ✽, Not determined. (B) Single DNA base substitutions in the fliC promoter region that were defective in fliC-lac transcription. (C) Single DNA base substitutions in the fliC 5′UTR region that were defective in transcription (red) and translation (green).

The majority of the 200 nonmotile mutants were analyzed further by DNA sequence analysis of the fliC 5′UTR region and flanking sequences subject to mutagenesis. Those that had more than one base substitution were not analyzed further. Ten mutants resulting from single base substitutions were classified as motility specific. All 10 were located in the fliC-coding region (Fig. 7A). All but one of the transcription-specific mutants was in the fliC promoter, including base substitutions in the −10 and −35 sequences and a single base deletion in the spacer region (Fig. 7B). The unlinked transcription-specific mutant resulted in a C:U transition mutation changing fliC codon 3 to a ochre stop codon. The effect of this mutation on the expression of the fliC::MudJ transcriptional fusion is most likely due to transcriptional polarity. The translation-specific mutants included expected single base substitutions in the AUG start codon and consensus ribosomal binding site region (Fig. 6, −9G:A, −9G:C and −10G:C). However, other translation-specific mutations appeared in a predicted stem-loop structure (Fig. 7C). These were single base substitutions that inhibited fliC mRNA translation (Fig. 7C and Fig. 6, bases −36 to −45). The single mutants −45G:C and −38C:G in the predicted stem-loop possess a motility defect and are defective in fliC-lacZ translation (Fig. 8 and 9). The mutations −45G:C and −38C:G are predicted to interact with each other in the formation of an RNA stem structure. To confirm this prediction, the double mutant −45G:C −38C:G was constructed by PCR and introduced into TH6303 by λ-Red recombination. Analysis of the fliC expression profile of Tc^s recombinants showed that the −45G:C −38C:G double mutant recovered motility and 80% of the wild-type fliC expression (Fig. 8 and 9). This suggests that the stem-loop structure and possibly the entire predicted structure is genuine and that mutations altering the predicted 5′UTR secondary structure prevent efficient fliC expression.

FIG. 8.

fliC expression profiles using fliC-lac and fliC-lacZ fusions of mutations in a predicted stem structure in the 5′UTR. The transcriptional and translational activities of −45G:C, −38C:G, and the double mutant −45G:C −38 C:G are compared to the wild type. Both single mutants that disrupt the predicted stem-loop show reduced translation but comparable transcriptional levels. In contrast, the double mutant retains a fliC expression profile similar to the wild type even though both transcription and translation are reduced twofold. Upon comparison with Fig. 9, the level of translation in −45G:C −38C:G is sufficient to obtain a motile phenotype.

FIG. 9.

Motility phenotypes of selected single base substitution mutants and the double mutant −45G:C −38 C:G. Motility agar plates showing the phenotypes of selected point mutations compared to wild type. Mutation or wild type: 1, −37C:A; 2, −38C:U; 3, −38C:A; 4, −45G:U; 5, wild type; 6, −45G:C; 7, −38C:G; 8, −45G:C −38C:G.

DISCUSSION

The original goal of this study was to isolate mutants in the fliC 5′UTR important for fliC gene expression, which were transcribed and translated normally and specific to filament assembly. Using a set of tetA-fliC operon constructs with different amounts of the 5′UTR fliC region between the tetA stop codon and the fliC start codon, we found that the entire 5′UTR region is needed for a motile phenotype under inducing conditions. These results suggested a requirement for the fliC 5′UTR in filament assembly when fliC is expressed from the chromosome. However, numerous labs, including ours, have expressed the fliC gene from multicopy plasmids without the fliC 5′UTR sequence, and these plasmids complement a fliC-null allele for motility (21, 22). This suggests that any requirement of sequences in the fliC 5′UTR for FliC assembly independent of transcription would likely be involved in efficiency of assembly and not absolutely required. Such a phenotype has been observed before in Salmonella flagellar assembly. When the FlgN secretion chaperone of the hook-filament junction proteins (FlgK and FlgL) is absent, FlgK was found to become unstable and resulted in a defect in motility. Motility in an flgN-null mutant was restored when the flgKL operon was expressed from a high-copy-number plasmid, suggesting that FlgN contributed to the efficiency of assembly and was not essential (3). Because the flagellum is a complex structure and we believe that efficiency of assembly has been under strong evolutionary pressure, we expect to find many stages in flagellar assembly where the temporal efficiency of assembly, rather than energy efficiency, has been maximized (the first to assemble flagella and get to the food source wins the evolution race). The exhaustive search for single base changes in the 5′UTR that affected assembly independent of transcription and translation suggests, in light of the results with the tetA-fliC operon constructions, that multiple changes are going to be required to see an effect.

Finally, we want to stress that the significant findings reported here were a result of targeted mutagenesis to the chromosomal fliC region. The use of multicopy plasmids introduces the artifact of excess gene dosage, which in the case of the fliC structural gene masks the effects of the 5′UTR on assembly. In the type III secretion field, there are three types of reported secretion signals: N-terminal amino acid structural, 5′UTR, and type III chaperone assisted (4). How these individual or a combination of signals are utilized for a given secretion substrate could influence the timing and amount of secretion and therefore the efficiency of flagellum assembly or pathogenesis. For FliC, overexpression from a multicopy vector would presumably allow secretion to be directed by amino acid and chaperone binding signals only. The simple technology of the bacteriophage λ recombination system to efficiently direct mutagenesis to a specific region of the chromosome such as the fliC 5′UTR is a powerful tool not only for the selection and screening of specific mutants from the natural chromosomal location but also for mutational changes resulting in no phenotypic differences from the wild type (16). This allows for a thorough analysis of any specific segment of a gene to be analyzed. The present study took advantage of the tetracycline resistance genes from transposon Tn10, which allowed for replacement of the region to be targeted by the tetRA sequences followed by replacement of the tetRA sequences by the mutagenized segment. Any construct that can be selected for and against can be used for such studies. For example, a sacB-npt cassette can be inserted selecting for npt-encoded neomycin resistance and selected against because sacB-expressing cells die on media containing sucrose. Such tools will allow geneticists to focus on the isolation and characterization of mutants in the single-copy setting of the chromosome and avoid the complications of plasmid artifacts.