RflM Functions as a Transcriptional Repressor in the Autogenous Control of the Salmonella Flagellar Master Operon flhDC

Abstract

Motility of bacteria like Salmonella enterica is a highly regulated process that responds to a variety of internal and external stimuli. A hierarchy of three promoter classes characterizes the Salmonella flagellar system, and the onset of flagellar gene expression depends on the oligomeric regulatory complex and class 1 gene product FlhD₄C₂. The flhDC promoter is a target for a broad range of transcriptional regulators that bind within the flhDC promoter region and either negatively or positively regulate flhDC operon transcription. In this work, we demonstrate that the RflM protein is a key component of flhDC regulation. Transposon mutagenesis was performed to investigate a previously described autoinhibitory effect of the flagellar master regulatory complex FlhD₄C₂. RflM is a LuxR homolog that functions as a flagellar class 1 transcriptional repressor. RflM was found to be the negative regulator of flhDC expression that is responsible for the formerly described autoinhibitory effect of the FlhD₄C₂ complex on flhDC operon transcription (K. Kutsukake, Mol. Gen. Genet. 254:440–448, 1997). We conclude that upon commencement of flagellar gene expression, the FlhD₄C₂ complex initiates a regulatory feedback loop by activating rflM gene expression. rflM encodes a transcriptional repressor, RflM, which fine-tunes flhDC expression levels.

INTRODUCTION

The biosynthesis, assembly, and rotation of flagella require a significant amount of biosynthetic resources and energy (1, 2). The expression of flagellar genes is regulated in response to various environmental conditions, which can determine the onset of flagellar biosynthesis and the overall degree of flagellation (3, 4). One difference between flagellar gene expression in Escherichia coli and Salmonella is that under low-nutrient conditions, the flagellar regulon can be either induced (E. coli) or repressed (Salmonella) (5–8). In all systems examined to date, flagellar synthesis is repressed during biofilm formation and upon entry into host cells (9–12).

The more than 60 genes of the flagellar regulon are organized into a transcriptional hierarchy of three promoter classes. The flagellar master operon, flhDC (under the control of the flagellar class 1 promoter), is transcribed to produce the FlhD₄C₂ transcriptional activator complex that is needed for the activation of genes downstream in the flagellar transcriptional hierarchy, which are transcribed from flagellar class 2 promoters. FlhD₄C₂-dependent genes are required for the structure and assembly of the flagellar hook-basal body (HBB), which functions as the motor of the flagellum. Also transcribed from a class 2 flagellar promoter is the fliA gene, which encodes the flagellum-specific transcription factor σ²⁸, required to transcribe flagellar class 3 promoters. Class 3 flagellar genes encode proteins required after HBB completion, including filament (fliC and fljB), motor force generators (motAB), and components of the chemosensory system (che).

A broad range of factors positively and negatively regulate expression of the flhDC operon. Transcription of flhDC from the class 1 promoter is dependent on the binding of the cyclic AMP-catabolite gene activator protein complex and can also be activated by the iron-regulatory protein Fur and by the nucleoid proteins Fis and H-NS (13–17). Those regulatory factors bind directly to the flhDC promoter region, as shown for Fis in Salmonella and for Fur and H-NS in E. coli (15, 17, 18). There are numerous negative regulators of flagellar biosynthesis, including SlyA, a transcriptional regulator of flhDC (19) that is also required for Salmonella virulence (20), RtsB, a Salmonella pathogenicity island 1 (SPI-1)-encoded repressor of flagellar class 1 transcription (21), LrhA (22), and RcsB. RcsB is reported to positively regulate SPI-2 expression and favor Salmonella growth in macrophages (23). RcsB also binds an RcsB box in the flhDC promoter region to repress the flagellar master operon (23, 24). Finally, flhDC transcription is reported to be under autogenous control (13). FlhD₄C₂ is also under posttranscriptional regulation. The flagellar protein FliZ is a product of the fliAZ operon, which is transcribed from flagellar class 2 and 3 promoters. FliZ positively regulates additional flagellar class 2 gene expression (25). According to Saini et al., FliZ posttranslationally alters the concentration of FlhD₄C₂ (26). Data provided by Wada et al. suggest that FliZ has a repressing effect on YdiV, another posttranscriptional anti-FlhD₄C₂ factor (27). YdiV binds to FlhD and prevents the FlhD₄C₂ complex from binding to class 2 promoters, either by physically keeping the FlhDC complex away from free promoter DNA or by releasing FlhDC from the DNA-bound state (7, 28).

In an earlier study, we identified RflM, previously known as EcnR (see Materials and Methods), as a repressor of flhDC transcription. RflM is encoded in close proximity to the antidote/toxin gene pair ecnAB and exhibits the typical characteristics of an OmpR-like response regulator (29). It was therefore originally named EcnR (entericidin gene R), but experimental analyses showed ecnAB to be positively regulated by σ^S and negatively regulated by EnvZ/OmpR (29). There is no evidence for regulation of ecnAB by RflM (EcnR). In our previous work, we isolated flhDC promoter mutants that suppressed RflM inhibition (30). The inhibitory effect of RflM on motility has been shown to be dependent on the RcsCDB system (30). An additional deletion of rcsDBC or an insertion in rcsB prevented the reported loss of motility of P_ararflM. Here, we characterized the inhibitory effect of RflM on flhDC transcription. FlhD₄C₂ activates transcription of rflM, and in turn, RflM represses flhDC transcription. The RflM-FlhD₄C₂ feedback loop thereby accounts for the formerly described autoregulatory effect of FlhD₄C₂ on flhDC operon transcription.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media.

All bacterial strains used in this study are listed in Table 1. Cells were cultured in lysogeny broth (LB) medium, supplemented with tetracycline (15 μg ml⁻¹) or anhydrotetracycline (AnTc) (1 μg ml⁻¹) if needed. Gene expression from the arabinose promoter was achieved by addition of 0.2% l-arabinose to the medium. The generalized transducing phage of Salmonella enterica serovar Typhimurium P22 HT105/1 int-201 was used in all crosses (31). Experiments using the virulent Salmonella enterica serovar Typhimurium ATCC 14028s background were performed under SPI-1-inducing conditions (24, 32, 33). Cultures were grown under high-osmolarity (1% NaCl, final) and low-oxygen (without agitation) conditions and if necessary supplemented with anhydrotetracycline and arabinose as described above. Phenotypic Lac activity was observed using MacConkey lactose (MacLac) agar supplemented with 0.2% arabinose if required.

Table 1.

Salmonella enterica serovar Typhimurium LT2 and ATCC 14028s (ST14028) strains used in this study

Strain or plasmid	Relevant characteristics	Reference or source
EM57	LT2 ΔaraBAD1007::flhD+C+ ΔrflM32 flhC5213::MudJ	This study
EM59	LT2 ΔaraBAD1007::flhD+C+ flhC5213::MudJ	This study
EM71	LT2 ΔaraBAD925::tetRA flhC5213::MudJ ΔrflM32	This study
EM153	LT2 ΔaraBAD921::rflM+ rflM3::MudJ P(flhDC)5451::Tn10dTc[del-25]	This study
EM154	LT2 ΔaraBAD1005::FCF P(flhDC)5451::Tn10dTc[del-25] rflM3::MudJ	This study
EM635	14028s ΔrcsB::tetRA luxCDABE (Kmr)::ybaJ	This study
EM636	14028s ΔrflM5::FCF luxCDABE (Kmr)::ybaJ	This study
EM642	14028s ΔaraBAD925::rcsB+ luxCDABE (Kmr)::ybaJ	This study
EM643	14028s ΔaraBAD921::rflM+ luxCDABE (Kmr)::ybaJ	This study
EM666	14028s ΔaraBAD1005::FRT luxCDABE (Kmr)::ybaJ	This study
EM672	14028s ΔaraBAD1007::flhD+C+ rflM3::MudJ	This study
EM673	14028s ΔaraBAD1005::FRT rflM3::MudJ	This study
EM700	14028s ΔaraBAD1007::flhD+C+ rflM3::MudJ ΔinvH-sprB::FCF	This study
EM701	14028s ΔaraBAD1005::FRT rflM3::MudJ ΔinvH-sprB::FCF	This study
TH437	LT2	J. Roth
TH3923	pJS28 (Apr P22-9+) F′114(Ts) Lac+ zzf-20::Tn10[tetA::MudP] (Tcs) zzf-3823::Tn10dTc[del-25] leuA414 hsdSB Fels2−	Lab collection
TH5971	Δhin-5717::FCF
TH6701	LT2 ΔaraBAD925::tetRA	Lab collection
TH8972	LT2 ΔaraBAD925::tetRA flhC5213::MudJ	P. Aldridge
TH9386	LT2 ΔaraBAD921::rflM+	J. Karlinsey
TH10068	LT2 rflM3::MudJ	Wozniak et al., 2009 (30)
TH13067	LT2 ΔaraBAD996::rcsB+ flhC5213::MudJ	Lab collection
TH13069	LT2 flhC5213::MudJ ΔaraBAD921::rflM+	Lab collection
TH14156	LT2 ΔaraBAD1007::flhD+C+	Erhardt and Hughes, 2010 (19)
TH15941	LT2 ΔaraBAD1007::flhD+C+ flhC5213::MudJ fliA5886 (R91C L207P)	This study
TH16205	LT2 ΔaraBAD1007::flhD+C+ rflM3::MudJ	This study
TH16952	ΔrflM4::FKF Δhin-5717::FCF	This study
TH16964	ΔrcsB::tetRA Δhin-5717::FCF	This study

Nomenclature.

The gene rflM was originally named ecnR since it was a putative regulatory protein linked to the toxin-antitoxin genes ecnAB (29). The change in nomenclature thereby displays the newly characterized function of RflM as a modulator of the flagellar master operon flhDC (regulator of flagellar master operon).

Isolation of random T-POP insertions.

Strain TH15941 [ΔaraBAD1007::flhD⁺C⁺ flhC5213::MudJ fliA5886(R91C L207P)] carries the flhD⁺C⁺ operon expressed from the chromosomal araBAD promoter (P_araBAD) (19). It also carries a lac operon (via insertion of the MudJ transposon into the flhC gene) transcriptional reporter fusion to the chromosomal flhDC promoter (flhC5213::MudJ) and a fliA null allele that is defective in binding DNA (34). The introduction of the fliA null allele was to prevent any potential effects of σ²⁸ activity on the flhDC autoregulatory control. Strain TH15941 is Lac⁺ but becomes Lac⁻ in the presence of arabinose (Ara-Lac⁻) due to induction of flhD⁺C⁺ transcription from P_araBAD, resulting in autorepression of flhC-lac reporter transcription by FlhD₄C₂. P22 phage prepared from T-POP donor strain TH3923 {pJS28 (Ap^r P22-9⁺) F′114(Ts) Lac⁺ zzf-20::Tn10[tetA::MudP] (Tc^s) zzf-3823::Tn10dTc[del-25] leuA414 hsdSB Fels2⁻} was used to introduce T-POP into strain TH15941 carrying plasmid pNK2881, which constitutively expresses the Tn10 transposase gene carrying the ats-1 and ats-2 (altered target specificity mutations) alleles, which result in the random insertion of Tn10 derivatives into the chromosome (35). About 30,000 T-POP insertions in TH15941 were screened for loss of FlhD₄C₂ autorepression, an Ara-Lac⁺ phenotype, in the presence of tetracycline (Tc-Ara-Lac⁺). Thirty T-POP insertions that had an initial Ara-Lac⁺ phenotype in the presence of Tc were isolated and were further characterized as described in Results.

β-Galactosidase assays.

β-Galactosidase activity was measured according to method of Zhang and Bremer (36) with minor modifications as described in the following. For each strain, a minimum of three independent biological replicates were picked from individual colonies. For experiments performed with LT2, cells were grown overnight in general LB medium. Strain 14028s was grown under SPI-1 inducing conditions in LB medium supplemented with 1% NaCl (final) and without shaking. LT2 samples were diluted 1:100 in LB medium supplemented with 0.2% l-arabinose and 1 μg ml⁻¹ anhydrotetracycline if needed. Cultures were grown until mid-log phase at 37°C before permeabilization of the cells using 100 mM Na₂HPO₄, 20 mM KCl, 2 mM MgSO₄, 0.08% CTAB (hexadecyl-trimethyl ammonium bromide), 0.04% sodium deoxycholate, and 5.4 μl ml⁻¹ β-mercaptoethanol.

To start the reaction, substrate solution containing 60 mM Na₂HPO₄, 40 mM NaH₂PO₄, 1 mg ml⁻¹ o-nitrophenyl-β-d-galactoside (ONPG) and 2.7 μg ml⁻¹ β-mercaptoethanol was added, and the time was measured until sufficient color had been developed. The enzyme reaction was then stopped by addition of 1 M sodium carbonate (Na₂CO₃), and remaining cell fragments were pelleted by centrifugation at full speed. The optical density of the supernatant at 405 nm was measured, and Miller units were calculated according to the method of Miller (37).

All samples were normalized against their corresponding arabinose-induced wild-type control in the presence or absence of anhydrotetracycline.

For statistical analysis, an unpaired t test was performed using the software program GraphPad Prism version 5.0d for Mac (GraphPad Software, San Diego, CA, USA). The P value summaries are displayed above the respective bars.

RNA isolation and quantitative real-time PCR.

RNA isolation was performed for three independent biological replicates using the RNeasy minikit (Qiagen). For removal of genomic DNA, RNA was treated with DNase I for 30 min at 37°C using a DNA-free RNA kit (Zymo Research) or on-column treatment was performed using kit 79254 (Qiagen). Subsequently, RNA samples were reverse transcribed according to the RevertAid first-strand cDNA synthesis kit (Fermentas). Quantitative real-time PCRs were carried out using the EvaGreen quantitative real-time PCR (qPCR) master mix (Bio-Rad) and primers 5′-TCTCAACGATGCCTTACCCGAACA plus 5′-GCAAGCTCATGTAAAGGCGTGTGT (rflM), 5′-CTGCTCAAAGAGCTGGTGTATCA plus 5′-AGCGCGTTACAGTCTGCTCAT (gyrB), 5′-CAACCTGTTCGTACGTATCGAC plus 5′-CAGCTCCATCTGCAGTTTGTTG (rpoB), 5′-CAACAGTATGCGCGTGATGAT plus 5′-CGACGCAGAGCTTCATGATC (rpoD), 5′-TTGCAGAAATGAGCCATTACGCCG plus 5′-GACGTTCAGC GCGAATGATGGTTT (gmk), 5′-GTAGGCAGCTTTGCGTGTAG plus 5′-TCCAGCAGTTGTGGAATAATATCG (flhDC), 5′-AACGTCTATTTTGTGAAAACCAAAG plus 5′-AGACTCCAGAATCCCGTTTTC (flgE), and 5′-AACGACGGTATCTCCATTGC plus 5′-ATTTCAGCCTGGATGGAGTC (flagellin, conserved region of fliC and fljB). Experiments were performed on a CFX96 real-time PCR instrument (Bio-Rad), with the exception of the experiment leading to Fig. 5. Here we used a Rotor-Gene Q 2plex real-time PCR system (Qiagen). Relative changes in mRNA levels were analyzed according to the Pfaffl method (38) and normalized against the transcript levels of the reference genes rpoB, rpoD, gyrB, and gmk.

Fig 5.

Flagellar class 1, 2, and 3 gene expression of rflM and rcsB mutants. Relative expression of the different flagellar promoter classes under wild-type, ΔrflM, and ΔrcsB conditions is shown. Quantitative real-time analysis was performed with two independent mRNA purifications. Expression levels are shown for class 1 (flhDC), class 2 (hook subunit flgE), and class 3 (flagellin subunits fliC and fljB) gene products. gyrB, rpoB, and gmk served as reference genes for normalization.

Motility plates.

Motility plates were prepared as described previously (30, 39). Strains were picked from fresh individual colonies grown on LB and pocked into 0.2% arabinose-containing (+Ara) and arabinose-free (−Ara) motility agar. Plates were kept at 30°C and grown for 5.5 h (−Ara) and 6.5 h (+Ara), respectively. The diameter of the motility swarm was measured using the software program NIH ImageJ 1.44g. In order to increase the contrast between motility and background, contrast levels were equally adjusted using iWork Pages software.

Flagellar staining.

For immunostaining of flagellar filaments, fliC-ON phase-locked (Δhin-5717) Salmonella bacteria containing a deletion of either rflM or rcsB were grown to mid-log phase and immobilized on poly-l-lysine-treated coverslips. The bacteria were fixed by addition of 5% formaldehyde and 0.5% glutaraldehyde. Membrane staining was performed using FM-64 (0.5 mg ml⁻¹). Flagella were stained using polyclonal anti-FliC antibodies (rabbit) and anti-rabbit-Alexa 488 secondary antibodies (Invitrogen). Images were collected using an inverted Axio Observer.Z1 fluorescence microscope (Zeiss).

RESULTS

FlhD₄C₂ activates its own repressor, RflM.

The master regulatory operon of flagellar gene expression, flhDC, is under autogenous repression (13). It was reported that increased levels of active FlhD₄C₂ complex resulted in decreased flhDC transcription in cases where σ²⁸ (FliA) was absent or where σ²⁸ was in the presence of its cognate anti-σ²⁸ factor FlgM (13). We confirmed the autoregulatory effect of FlhD₄C₂ on flhDC operon transcription in the presence of both σ²⁸ and FlgM as shown in Fig. 1. When FlhD₄C₂ was overexpressed from the arabinose promoter upon addition of 0.2% arabinose, a chromosomal flhC-lac transcriptional fusion under the control of the flhDC promoter reported a drop in flhC-lac expression of about 2-fold (Fig. 1A). It was shown previously that tuning expression of FlhD₄C₂ from the arabinose promoter by addition of 0.05% to 0.6% arabinose resulted in a steady 60-fold increase in flhDC mRNA levels compared to flhDC expressed from the native P_flhDC promoter (19). In addition, the FlhD₄C₂ levels needed for maximal activation of downstream flagellar genes appear to be rapidly saturated, since class 2 and class 3 gene expression increased only 3- to 5-fold in the tested range of P_araflhDC overexpression (19).

Fig 1.

RflM inhibits its own activator (flhDC). (A) Expression levels of an flhC-lac transcriptional fusion under P_araBAD::flhD⁺C⁺ inducing conditions (0.2% arabinose), measured by β-galactosidase activity as described in Materials and Methods. Three independent biological replicates of the P_araBAD::flhD⁺C⁺ flhC::MudJ (EM59; lane 2) and P_araBAD::flhD⁺C⁺ ΔrflM flhC::MudJ (EM57; lane 4) strains were measured and compared to ΔaraBAD::tetRA flhC::MudJ (TH8972; lane 1) and ΔaraBAD::tetRA ΔrflM flhC::MudJ (EM71; lane 3) control strains. (B) Relative flhDC expression compared to wild-type mRNA levels of strain TH6701 ΔaraBAD::tetRA or TH437 LT2, respectively. Transcript levels were compared by quantitative real-time PCR as described in the text. The effect of arabinose-induced overexpression of RflM (TH9386; P_araBAD::rflM⁺) on flhDC expression was monitored in independent experiments using total mRNA of three independent biological samples grown in arabinose-containing medium. (C) β-Galactosidase activity showing transcription levels in strain expressing RflM from the arabinose promoter (TH13069 P_araBAD::rflM⁺ flhC::MudJ, labeled ↑), compared to results for the wild-type control (TH8972; ΔaraBAD::tetRA flhC::MudJ). TH8972 contains chromosomal rflM but lacks rflM expression due to the absence of functional flhDC (labeled as +*). (D) rflM expression levels in a β-galactosidase assay showing repression of rflM transcription in strains lacking flhDC. For each strain, four independent replicates were analyzed. EM154 (ΔaraBAD::FCF P_flhDC::Tn10dTc[del-25] rflM::MudJ; lanes 1 and 3) and EM153 (P_araBAD::rflM⁺ P_flhDC::Tn10dTc[del-25] rflM::MudJ; lanes 2 and 4) were grown until mid-log phase in arabinose-supplemented medium in the presence and absence of 1 μg ml⁻¹ anhydrotetracycline (AnTc). The relevant genotype is labeled as followed: a chromosomal wild-type copy of the gene is present (+), the gene is chromosomally overexpressed from an arabinose-inducible promoter (↑), or the strain is deleted for the respective gene (−). It is important to note that rflM requires the presence of FlhDC in order to be expressed. The asterisk (∗) indicates conditions where the wild-type rflM gene is present but not transcribed due to the absence of flhDC. Error bars represent the standard deviations of the means. Data were analyzed by the Student t test. Gene expression levels that differed significantly are shown (∗∗, P < 0.01; ∗∗∗, P < 0.001).

To address whether FlhD₄C₂ could directly repress its own transcription or if FlhD₄C₂ mediated autorepression indirectly by activating the transcription of an unknown flhDC operon repressor, we performed T-POP transposon mutagenesis and screened for potential FlhD₄C₂-dependent repressors of flhDC gene transcription. T-POP transposons are derivatives of transposon Tn10 that can transcribe chromosomal genes adjacent to the site of insertion from tetracycline-inducible promoters within the T-POP element (40). T-POP lacks transposition functions, which can be provided in trans by a Tn10 transposase expression plasmid present in recipient cells (40). After transposon insertions have been isolated and moved into a strain lacking Tn10 transposase, the T-POP transposon is no longer capable of further transposition. Strain TH15941 (ΔaraBAD1007::flhD⁺C⁺ flhC5213::MudJ fliA5886) carries an flhC-lac transcriptional fusion and the flhD⁺C⁺ operon expressed from an arabinose-inducible promoter (P_araflhD⁺C⁺). This strain is Lac⁻ in the presence of arabinose due to FlhD₄C₂ autorepression. The phenotype on lactose MacConkey (MacLac) agar is shown in Fig. 2. MacLac indicator medium shows the optimal sensitivity in the intermediate lac operon expression range (41). It thus represents the indicator medium of choice to visualize repression of flhDC, such as flhDC autorepression (P_araflhD⁺C⁺) or repression of flhDC by the known transcriptional regulator RcsB (P_ararcsB⁺). Using MacLac indicator medium, a transition from white (Lac⁻) to dark-pink (Lac⁺) colonies can be readily observed (Fig. 2). Strain TH15941 was mutagenized with the T-POP transposon Tn10dTc[del-25] and screened for insertions that were Lac⁺ (dark pink) in the presence of arabinose due to loss of FlhD₄C₂ autorepression. The T-POP transposon Tn10dTc[del-25] is deleted for the terminator of the tetA tetracycline resistance gene transcript, allowing transcription from the tetA promoter (P_tetA) within the T-POP transposon into adjacent chromosomal DNA (40). Four groups of T-POP transposon insertion mutants that resulted in the apparent loss of FlhD₄C₂ autorepression (i.e., Lac⁺ in the presence of arabinose [Ara-Lac⁺]) were obtained. One group included T-POP insertions linked to the arabinose region and are presumed to have disrupted the flhDC operon. A second group included T-POP insertions linked to the flhC-lac operon fusion and are presumed to be polar insertions between the flhDC promoter and the lac operon inserted in flhC, thereby putting lac directly under P_tetA control. These two groups of Ara-Lac⁺ T-POP insertions in strain TH15941 were not further characterized. The remaining two groups of Ara-Lac⁺ T-POP insertions were unlinked to both ara and flhC-lac fusion chromosomal positions and represented candidates for insertions in a gene or genes responsible for FlhD₄C₂-dependent repression of flhDC operon transcription. One group included two insertions that exhibited an Ara-Lac⁺ phenotype only in the presence of tetracycline (Tc). These were presumed to turn on an activator of flhDC gene expression from P_tetA that is currently being subjected to further characterization. The final group of insertion mutants was Ara-Lac⁺ in either the presence or absence of added Tc, and their Ara-Lac⁺ phenotype was presumed to be due to the loss of function of the gene into which the T-POP was inserted. T-POP insertions of this group that were linked to known negative regulators of flhDC expression (e.g., lrhA, rtsB, and rcsB) were not characterized further. Of the remaining unlinked T-POP insertions, we sequenced 10 Tc-independent Ara-Lac⁺ insertion mutants, and DNA sequence analysis revealed transposition of the T-POP element into the rflM locus (Table 2). This indicated that rflM encodes an FlhD₄C₂-induced repressor of flhDC transcription (the “autorepressor”).

Fig 2.

Phenotypes of different P_ara constructs on MacLac indicator plates in the presence (+ara) and absence (−ara) of arabinose. Plates were incubated at 37°C overnight. TH8972 P_araBAD::tetRA, EM71 P_araBAD::tetRA ΔrflM, EM59 P_araBAD::flhDC, EM57 P_araBAD::flhDC ΔrflM, TH13069 P_araBAD::rflM⁺, and TH13067 P_araBAD::rcsB are shown.

Table 2.

Locations of tetracycline-independent Ara-Lac⁺ T-POP insertions

T-POP no.	Location of insertion (bp downstream of rflM start site)	DNA strand
4	157	Minus strand
5 and 9	39	Minus strand
6 and 24	200	Minus strand
10	157	Minus strand
11	191	Reference (plus) strand
17	542	Reference (plus) strand
22	473	Minus strand
25	149	Reference (plus) strand

RflM inhibits flhC gene transcription.

The results of the random T-POP transposon mutagenesis suggested that RflM functions as an FlhD₄C₂-dependent inhibitor of flhDC operon transcription. Accordingly, we tested autoinhibition of flhDC transcription upon induction of P_araflhD⁺C⁺ in the absence of the putative regulator RflM (Fig. 1A). The inhibitory effect of P_araflhD⁺C⁺ induction on flhDC gene transcription was lost in the rflM deletion strain, confirming that RflM was responsible for the FlhD₄C₂ autoregulation effect observed by Kutsukake (13). In order to confirm that RflM represses flhDC operon transcription, we analyzed flhDC mRNA levels in a strain that expresses excess rflM from an arabinose-inducible promoter (P_ararflM⁺). As shown in Fig. 1B, under P_ararflM⁺-inducing conditions, flhDC expression decreased about 4-fold in a reverse transcriptase qPCR (RT-qPCR) assay. Reduction in flhDC transcription was also observed under P_ararflM⁺-inducing conditions using an flhC-lac operon fusion reporter shown in Fig. 1C.

Effect of RflM on rflM gene transcription.

We investigated whether RflM was under autogenous control. To test if RflM autoregulates its own transcription, we utilized a previously isolated rflM-lac operon transcriptional fusion (30). A strain overexpressing RflM from the arabinose promoter (P_ararflM⁺) was analyzed in comparison to the wild-type control, where the arabinose genes araBAD were deleted by inserting an Flp recombination target (FRT)-chloramphenicol acetyltransferase-FRT (FCF) resistance cassette (ΔaraBAD::FCF). Both strains carried the above-mentioned rflM-lac reporter and a T-POP insertion in the promoter region of flhDC. The latter allows the activation of flhDC transcription upon addition of anhydrotetracycline (AnTc), which activates T-POP-encoded tetA and tetR gene transcription, and prevents flhDC autorepression. As shown in Fig. 1D, the absence of FlhD₄C₂ (e.g., in the absence of the inducer AnTc, the flhDC operon in these strains is not transcribed) resulted in a reduction of rflM-lac transcription, presumably because the rflM gene required FlhD₄C₂ for its transcription. However, induction of rflM from the arabinose locus had no effect on the remaining rflM-lac transcription, suggesting that RflM does not play an additional role in its own transcriptional regulation except through regulation of flhDC.

FlhD₄C₂ activates its own repressor, RflM.

The results described above indicate that FlhD₄C₂ activates its own repressor, RflM. To analyze the effects of P_araflhD⁺C⁺ induction on rflM gene expression, we analyzed β-galactosidase activity of a transcriptional rflM-lac reporter in Salmonella enterica LT2 and the virulent strain Salmonella enterica ATCC 14028s to exclude strain-specific effects and in addition analyze a potential cross talk between the interconnected virulence and flagellar gene regulation networks. The attenuated strain, LT2, was grown under regular LB conditions, whereas the virulent strain, 14028s, was grown under high-salt and low-oxygen conditions, known to induce the virulence-associated Salmonella pathogenicity island 1 (SPI-1) (24, 32, 33). In both strain backgrounds, a 3-fold induction of rflM was observed under P_araflhD⁺C⁺-inducing conditions (Fig. 3A). We also performed complementary quantitative real-time PCR assays. Upon P_araflhD⁺C⁺-inducing conditions, the rflM mRNA level was significantly increased, as shown in Fig. 3B.

Fig 3.

FlhD₄C₂ activates its own repressor, RflM. (A) Expression of an rflM-lac transcriptional fusion under flhDC overexpression conditions compared to wild-type flhDC expression conditions. Transcription of rflM-lac as determined by β-galactosidase assays for Salmonella enterica serovar Typhimurium LT2 and ATCC 14028s shows activation of rflM gene expression in a strain expressing excess FlhD₄C₂. Three independent biological replicates were grown in LB medium containing 0.2% arabinose. TH10068 (LT2 rflM::MudJ; lane 1) and TH16205 (LT2 P_araBAD::flhD⁺C⁺ rflM::MudJ; lane 2) were grown shaking in normal LB medium. For ST14028s, strains EM673 (ΔaraBAD::FRT rflM::MudJ; lane 3), EM701 (ΔaraBAD::FRT rflM::MudJ ΔinvH-sprB::FCF; lane 4), EM672 (P_araBAD::flhD⁺C⁺ rflM::MudJ; lane 5), and EM700 (P_araBAD::flhD⁺C⁺ rflM::MudJ ΔinvH-sprB::FCF; lane 6) were grown under SPI-1-inducing conditions as further described in Materials and Methods. (B) Effects of excess FlhD₄C₂ on rflM gene expression. Strain TH14156 P_araBAD::flhD⁺C⁺ (↑) was grown to mid-log phase in LB medium containing 0.2% arabinose. Total RNA of three biological replicates was isolated, and rflM transcript levels were analyzed in independent experiments by real-time qPCR as described in the text. Relative gene expression was determined using the Pfaffl method. Individual mRNA levels were normalized against those for multiple reference genes (gmk, gyrB, rpoB, and rpoD) and are presented as fold change relative to those for the wild-type control, TH6701 ΔaraBAD::tetRA or TH437 LT2. Data were analyzed by the Student t test. Gene expression levels that differed significantly are indicated (∗∗∗, P < 0.001). Error bars represent the standard deviations of the means. The relevant genotype is labeled as followed: a chromosomal wild-type copy of the gene is present (+), the gene is chromosomally overexpressed from an arabinose-inducible promoter (↑), or the strain is deleted for the respective gene (−).

Effect of RflM on motility and flagellation.

We have shown above that overexpression of rflM resulted in a substantial repression of the flagellar master regulatory operon, flhDC. We therefore analyzed the effect of a deletion of the rflM gene and rflM overexpression on motility and flagellar assembly. A deletion in rflM resulted in an increase in motility compared to the wild type (Fig. 4A and B) and on average one additional flagellum per cell body (Fig. 4C). Under rflM-overexpressing conditions (labeled rflM↑), a substantial loss of motility was observed and motility was significantly impaired compared to that with the known repressor RcsB (Fig. 4B).

Fig 4.

Motility and flagellation of rflM and rcsB mutants. (A) Motility of rflM and rcsB mutants with or without induction by arabinose. Fresh colonies were poked into motility plates and grown at 30°C for 5.5 h (−Ara) or 6.5 h (+Ara). (B) Motility diameter was measured using ImageJ, and values for EM636 (ΔrflM::FCF), EM635 (ΔrcsB::tetRA), EM643 (ΔaraBAD::rflM⁺), and EM642 (ΔaraBAD::rcsB⁺) are shown relative to those for the wild type (WT) EM666 (ΔaraBAD::FRT). Relevant genotypes are listed. For details, see the list of strains (Table 1). (C) Numbers of flagella per cell in a fliC-ON (Δhin-5717) phase-locked background. Cell bodies were stained using FM-64, and flagella were labeled using antibodies against the expressed filament subunit fliC as described in Materials and Methods. The number of flagella per cell body was manually counted using ImageJ. TH5971 (Δhin-5717), TH16952 (ΔecnR Δhin-5717), and TH16964 (ΔrcsB Δhin-5717) were analyzed.

Additionally, we analyzed gene expression of the three flagellar promoter classes by quantitative real-time PCR. Expression of all the classes was upregulated in the ΔrflM background, with a decreasing effect along the transcriptional flagellar hierarchy (Fig. 5). While no significant difference could be observed between the two regulators RflM and RcsB in terms of class 1 (flhDC) expression, the RflM effect is clearly visible at later class 2 (flgE) and class 3 (fliC and fljB) transcription. This effect supports the observation that the RflM effect on motility and flagellation was more pronounced than that of RcsB.

In summary, our results support a regulatory feedback loop for the autogenous control of flhDC operon transcription. Expression of flhDC leads to production of a functional FlhD₄C₂ activator complex, which, in addition to activating flagellar class 2 promoter transcription for hook-basal body assembly, activates rflM gene transcription. RflM, when produced, acts to inhibit flhDC transcription. We have thereby demonstrated that this feedback loop is responsible for the formerly observed autoregulatory effect of FlhD₄C₂.

DISCUSSION

In this study, we show that the formerly observed autoregulatory effect of the flagellar master regulator FlhD₄C₂ is caused by FlhD₄C₂-dependent production of a negative regulator, RflM. FlhD₄C₂ and RflM thereby form a regulatory feedback loop, which regulates the rate of flagellar gene expression with respect to the level of already-initiated flagellar biosynthesis. Expression of rflM is directly coupled to the presence of functional FlhD₄C₂, which could be concluded from the fact that the amount of rflM mRNA was significantly increased under FlhD₄C₂ overexpression conditions. Conclusively, flhDC expression decreased in the presence of excess RflM, and FlhD₄C₂ autoregulation was completely abolished in an rflM deletion background.

Flagellar synthesis is dependent on a hierarchy of three transcriptional promoter classes, with the master regulator, FlhD₄C₂, at the top. FlhD₄C₂ is a known DNA binding complex that has been shown to bind to the −40 to −80 regions of multiple flagellar operons (42, 43) and together with σ⁷⁰ targets RNA polymerase to transcribe class 2 flagellar promoters (44). FlhD₄C₂ allows expression of the subsequent flagellar genes (under the control of the FlhD₄C₂-dependent class 2 promoters) needed for hook-basal-body assembly. The presence of functional FlhD₄C₂ thereby plays the main role in determining whether or not flagella are synthesized. Accordingly, negative feedback between the FlhD₄C₂ protein and flhDC operon expression would prevent the cell from synthesizing an excess of flagella. Overexpression of flhDC has previously been described to turn off flhDC operon transcription and was attributed to a direct inhibition of the flhDC operon by the FlhD₄C₂ complex (13). It was unclear, however, whether this inhibitory function was direct or indirect. Thus, we designed a genetic selection and screen to differentiate between two possibilities: in addition to the known role of FlhD₄C₂ as a transcriptional activator of flagellar class 2 promoters, FlhD₄C₂ could either act as a direct repressor of its own (flhDC operon) transcription or as an activator of an unknown repressor of flhDC operon transcription. Our results support the latter possibility.

The identification of an FlhD₄C₂-dependent flhDC regulator allows us to postulate a novel regulatory feedback loop that accounts for the previously described autoregulation of FlhD₄C₂. The loop consists of two mechanisms: (i) FlhD₄C₂ activates rflM and (ii) RflM represses flhDC transcription. With this study, we added a new component to the complex regulatory mechanisms that controls expression and activity of the flagellar master regulator FlhD₄C₂. Many environmental and regulatory stimuli are integrated at the level of σ⁷⁰-dependent transcription of the flhDC operon. Expression of FlhD₄C₂ therefore forms the prerequisite for the initiation of flagellar synthesis by activation of flagellar class 2 and 3 gene expression. Regulation at the level of flhDC operon transcription is thus the main target by which flagellar synthesis is regulated. In this study, we further characterized the flhDC master operon regulation that is placed on top of the flagellar gene expression cascade. In addition to environmental (external) stimuli that control expression of flhDC, an endogenous (internal) regulation mechanism that directly provides feedback of FlhD₄C₂ protein levels exists. An increase in flhDC expression and the subsequent flagellar synthesis requires a mechanism of counterbalance. Such an effect was previously referred to as FlhD₄C₂ autorepression (Fig. 6A), which our results show to be a regulatory feedback loop via RflM. A previous study indicated an additional role of the RcsCDB system in the RflM-mediated repression of flhDC (30). The response regulator RcsB has been shown to directly bind to the flhDC promoter region (23, 24). Accordingly, different scenarios of RflM-dependent repression of flhDC can be envisioned, and follow-up experiments are currently in progress to test the role RcsB in this regulation network. RflM could directly repress flhDC, while RcsB takes over a supporting role (Fig. 6B). Alternatively, RcsB could directly repress flhDC independently of RflM, and both proteins might compete for binding to the flhDC promoter region (Fig. 6C). Finally, RflM could act as a corepressor of RcsB (Fig. 6D).

Fig 6.

Possible models of the regulatory flhDC feedback loop. Panel A shows the formerly described autoregulatory effect of flhDC, where FlhD₄C₂ directly represses flhDC operon transcription. In contrast to this previous model, FlhD₄C₂ induces expression of rflM, and the RflM protein acts as a repressor of flhDC transcription. Repression of RflM might be direct but facilitated by RcsB (B). Alternatively, RflM might repress flhDC independently of RcsB (C) or act as a corepressor of RcsB (D).

The complexity and the hierarchical structure of the flagellar system explain the frequency by which endogenous regulation mechanisms occur. Feedback regulation in the flagellar system targets multiple levels throughout biosynthesis of the flagellum: (i) on the level of flhDC gene expression via FlhD₄C₂-dependent activation of the LuxR-type repressor RflM, (ii) on the level of FlhDC protein complex via posttranscriptional repression (FliT and YdiV) and activation (FliZ) of FlhDC-dependent class 2 gene expression, and (iii) on the level of flagellar class 3 gene expression by secretion of the anti-σ²⁸ factor FlgM after HBB completion and positive regulation of flgM translation by FlgN (45–47). In summary, the following scenario could be envisioned to couple the assembly process of the flagellum to flagellar gene expression. σ⁷⁰-dependent class 1 transcription of flhDC leads to formation of the flagellar master regulatory complex FlhD₄C₂. FlhD₄C₂ activates class 2 gene transcription, as well as that of its own repressor, rflM. While class 2 products build up the HBB, FliZ activates FlhDC complex and RflM acts as a repressor of flhDC transcription. Upon completion of the HBB, the secretion specificity switch results in secretion of FlgM and FliD, and class 3 gene expression takes place. Secretion of FliD upon HBB completion allows FliT to bind FlhDC complex and inhibit FlhD₄C₂ activity. Reduced activity of FlhDC decreases expression of rflM and allows a new cycle of flagellar gene expression and assembly starting with derepression of the flhDC promoter.