The hdeD Gene Represses the Expression of Flagellum Biosynthesis via LrhA in Escherichia coli K-12

ABSTRACT

Escherichia coli survives under acid stress conditions by the glutamic acid-dependent acid resistance (GAD) system, which enzymatically decreases intracellular protons. We found a linkage between GAD and flagellar systems in E. coli. The hdeD gene, one of the GAD cluster genes, encodes an uncharacterized membrane protein. A reporter assay showed that the hdeD promoter was induced in a GadE-dependent manner when grown in the M9 glycerol medium. Transcriptome analysis revealed that most of the transcripts were from genes involved in flagellum synthesis, and cell motility increased not only in the hdeD-deficient mutant but also in the gadE-deficient mutant. Defects in both the hdeD and gadE increased the intracellular level of FliA, an alternative sigma factor for flagellum synthesis, activated by the master regulator FlhDC. The promoter activity of the lrhA gene, which encodes repressor for the flhDC operon, was found to decrease in both the hdeD- and gadE-deficient mutants. Transmission electron microscopy showed that the number of flagellar filaments on the hdeD-, gadE-, and lrhA-deficient cells increased, and all three mutants showed higher motility than the parent strain. Thus, HdeD in the GAD system activates the lrhA promoter, resulting in a decrease in flagellar filaments in E. coli cells. We speculated that the synthesis of HdeD, stimulated in E. coli exposed to acid stress, could control the flagellum biosynthesis by sensing slight changes in pH at the cytoplasmic membrane. This could help in saving energy through termination of flagellum biosynthesis and improve bacterial survival efficiency within the animal digestive system.

IMPORTANCE E. coli cells encounter various environments from the mouth down to the intestines within the host animals. The pH of gastric juice is lower than 2.0, and the bacterial must quickly respond and adapt to the following environmental changes before reaching the intestines. The quick response plays a role in cellular survival in the population, whereas adaptation may contribute to species survival. The GAD and flagellar systems are important for response to low pH in E. coli. Here, we identified the novel inner membrane regulator HdeD, encoding in the GAD cluster, to repress the synthesis of flagella. These insights provide a deeper understanding of how the bacteria enter the animal digestive system, survive, and form colonies in the intestines.

INTRODUCTION

Unicellular bacterial cells are directly exposed to the environment. Escherichia coli must survive in various environments from the mouth down to the intestines for colonization in host animals. The regulatory network forming the expression pattern of the E. coli genome makes them respond to continuous harsh stresses, from acidic conditions in the stomach to anaerobic conditions in the intestines (1). In response to pH changes caused by gastric juice, E. coli cells exhibit the consumption of intracellular protons by glutamic acid-dependent acid resistance (GAD) systems and pH taxis by negative chemotaxis.

The GAD system is known to be most effective for consumption of intracellular protons in E. coli cells for the maintenance of intracellular neutral pH in acidic environments (2). The GAD system is composed of two glutamic acid decarboxylases (GadA and GadB) and one gamma-aminobutyric acid exporter (GadC), which decreases intracellular protons by enzymatic conversion of glutamic acid to gamma-aminobutyric acid (3–7). The GAD cluster was located at 79 min of the E. coli K-12 genome and included 12 genes slp, dctR, yhiD, hdeB, hdeA, hdeD, gadE, mdtE, mdtF, gadW, gadX, and gadA (Fig. 1). Among the 12 genes, five genes encoded cytoplasmic proteins (dctR, gadE, gadW, and gadX as transcriptional factors and gadA as a glutamic acid decarboxylase), whereas seven genes (slp, yhiD, hdeB, hdeA, hdeD, mdtE, and mdtF) encoded bacterial envelope proteins (Fig. 1). Slp is a starvation lipoprotein located on the outer membrane (8). YhiD was predicted to be as a Mg²⁺ transporter, based on the amino acid sequence homology of MgtC in Salmonella enterica (9). HdeA and HdeB are periplasmic chaperones for DegP and SurA, a periplasmic serine endoprotease, and a periplasmic chaperone, respectively (10–13). MdtE and MdtF form a putative tripartite efflux pump complex with the outer membrane protein TolC (14). The expression of these components is under the control of a sequential transcriptional regulation system by three transcription factors, EvgA, YdeO, and GadE, of which GadE is the master activator of the GAD cluster genes (3, 15–18) (Fig. 1). Among the GAD cluster genes, only the hdeD gene is uncharacterized, except for the identity of an acid resistance membrane protein (19, 20).

FIG 1.

Schematic representation of Gad proteins distributed in the cell envelope. The Gad cluster retains 12 genes at 78.8 min of E. coli K-12 MG1655 genome. Direction of transcription is illustrated by thick arrows. Among 12 genes, each of 4 genes (dctR, gadE, gadW, and gadX) encodes helix-turn-helix transcription regulator. The gadA gene encodes a glutamate decarboxylase. All the remaining 7 genes encode membrane proteins as the following: a starvation lipoprotein (Slp), a putative Mg²⁺ transporter (YhiD), an inner membrane protein (HdeD), periplasmic chaperons (HdeA/HdeB), and putative multidrug efflux pumps (MdtEF). The gadBC genes are located at 33.8 min of the genome, far from the GAD cluster. The gadB gene encodes another glutamate decarboxylase as an isozyme of GadA. A gamma-aminobutyric acid exporter (GadC) is distributed in the inner membrane.

Bacterial cells migrate away from an acidic pH using flagellar motility and chemotaxis by sensing changes in intracellular pH through the methyl-accepting protein MCP I (21–23). In chemotaxis, chemicals are sensed through their interactions with transmembrane chemoreceptor proteins as attractants or repellents (for a review, see reference 24). In contrast to receptor sensing, excess of intracellular protons interferes with the release of protons from the torque-generating units, resulting in slowing or stopping of the motors and impairing the motility of E. coli and Salmonella enterica (25). The bacterial flagellum is composed of more than 60 gene products and ∼2% of the biosynthetic energy expenditure in E. coli is used by the flagellar synthesis (26). Therefore, the flagellar synthesis genes are tightly regulated by an ordered transcriptional cascade, and these genes are accordingly divided into three groups: class I, class II, and class III (26). FlhDC belongs to class I and is known to be a master regulator at the top of the cascade of flagellum biosynthesis in E. coli and Salmonella enterica. The flhDC operon is regulated by both transcription and posttranscriptional pathways (26). The transcription of the σ⁷⁰-dependent flhDC operon is influenced by numerous environmental signals, such as temperature, osmolarity, and pH as well as by many kinds of transcriptional factors (for review, see reference 27) with H-NS (28, 29), CRP (29, 30), QseB (31), and AtoC (32) as positive regulators, and Fur (33), OmpR (34), IHF (34), HdfR (35), LrhA (36), FliZ (37), MatA (38), RcsB (39, 40), and AcrR (41) as negative regulators. In addition, flhDC expression is posttranscriptionally stimulated by CsrA (42).

In this study, we demonstrated that HdeD plays a role in the regulatory linkage between GAD and the flagellar systems in E. coli. We also found that the hdeD gene is required for lrhA promoter activation, which represses the mRNA level of flhDC, following the repression of flagellum biosynthesis and motility. This regulatory pathway is under the cascade regulation of HdeD induced by GadE. Our study indicates a novel pathway for the transcriptional regulation of acid resistance and flagellum biosynthesis in E. coli, which allows the adaptive growth of cells under acid stress.

RESULTS

Genome expression profile of the hdeD-deficient mutant.

The GAD cluster consists of 12 genes working together for acid resistance. The roles of the 11 genes are partially known, but that of the hdeD gene is unclear (Fig. 1). The gene product HdeD is predicted to encode an inner membrane protein with six transmembrane domains (see Discussion for details). First, we examined the activity of the hdeD promoter in E. coli grown under several conditions using a luciferase reporter assay. The hdeD promoter activity was induced only when grown in M9 glycerol medium but not in LB broth (Fig. 2a), in agreement with previous reports that the GAD cluster genes were hardly expressed in LB broth due to the tight gene silencing by H-NS (43, 44). The induction of the hdeD promoter requires the master regulator GadE of the GAD cluster (Fig. 2b) (3).

FIG 2.

Activation of the hdeD promoter in E. coli grown in M9 glycerol. The activity of the hdeD promoter was measured by luciferase reporter system as described in Materials and Methods. pLUXhdeD, harboring the hdeD-lux transcription fusion gene, is transferred into BW25113 (the parent strain) and JW3480ΔKm (ΔgadE). (a) BW25113 harboring pLUXhdeD was grown in M9 glycerol or LB broth until log phase, the luciferase activity in each culture was measured with the plate reader, and then the ratio of chemiluminescence to OD₆₀₀ values was evaluated as the promoter activity. (b) BW25113 and JW3480ΔKm harboring pLUXhdeD were grown in M9 glycerol and the hdeD promoter activity was measured in each culture as described above. Bars show the average from three independent experiments with standard deviation.

Next, we performed transcriptome analysis to elucidate the physiological functions of hdeD. E. coli BW25113, the parent strain, and JW3479, the hdeD-deficient strain (ΔhdeD), were grown in M9 glycerol medium until the log phase, and total RNA extracted from the cultures was subjected to DNA microarray analysis. As shown in Table 1, transcripts of 70 genes in ΔhdeD were significantly increased (>0.45 of log₁₀ ratio), but those of eight genes were significantly decreased (less than −0.45 of log₁₀ ratio) as measured with duplicated DNA microarrays. As expected, the transcript of hdeD was negligibly low in ΔhdeD (Table 1). Among the 70 upregulated genes in ΔhdeD, 53 genes were involved in flagellum biosynthesis, motility, and chemotaxis, suggesting that HdeD might repress the transcription of a set of flagellar genes. The remaining 17 upregulated genes encode the following proteins: a curli assembly protein (csgC), a putative O-antigen capsule (gfcD), a c-di-GMP phosphodiesterase (pdeH), a transcriptional regulator of colanic acid capsular biosynthesis (rcsA), a regulator of diguanylate cyclase (rdcB), a cold- and stress-inducible protein (ves), three prophage proteins (flxA, yfjS, and ykfB), seven uncharacterized proteins (ydjE, yecR, yedN, yiaW, yjdA, ymdA, and ynjH), and a pseudogene (intG). In contrast, only seven genes, other than hdeD, were downregulated in ΔhdeD. The functions of these genes are follows: a lipoprotein on DLP12 prophage (borR), a transcriptional regulator of the GAD cluster (gadX), a ketol-acid reductoisomerase (ilvC), a LysR-type transcriptional factor (lrhA), an uncharacterized protein (yddB), an inner membrane protein of CP4-44 prophage (yeeR), and a putative 4Fe-4S cluster-containing protein (ygcO).

TABLE 1.

Expression profile of DNA array in an hdeD-deficient strain

Gene	Fold change (log10)a
	Expt 1	Expt 2	p valueb
Upregulated in ΔhdeD strain
aer	1.25	1.23	8.94E−04
bdm	0.59	0.65	1.66E−02
cheA	1.75	1.70	3.08E−04
cheB	1.53	1.35	2.90E−03
cheR	1.66	1.65	2.73E−04
cheW	1.67	1.56	8.78E−04
cheY	1.69	1.53	2.91E−03
cheZ	0.86	0.57	2.42E−02
csgC	1.18	1.11	4.96E−04
flgA	0.85	0.84	1.97E−03
flgB	1.00	1.08	1.51E−03
flgC	0.94	1.04	2.02E−03
flgD	0.94	1.06	2.93E−03
flgE	0.96	1.05	1.27E−03
flgF	0.90	0.98	3.14E−03
flgG	0.88	1.01	2.64E−03
flgH	0.85	0.98	7.35E−03
flgI	0.77	0.88	4.20E−03
flgJ	0.65	0.80	8.29E−03
flgK	1.24	1.32	5.80E−04
flgL	1.21	1.23	5.23E−03
flgM	1.03	1.18	4.56E−03
flgN	0.95	1.11	4.76E−03
flhA	0.80	0.91	4.65E−03
flhB	0.74	0.84	5.78E−03
flhC	0.63	0.56	2.20E−02
flhE	0.80	0.89	4.26E−03
fliA	1.05	1.04	1.57E−04
fliC	1.81	1.78	1.16E−04
fliD	1.39	1.38	1.59E−03
fliE	0.94	0.74	7.95E−03
fliF	0.71	0.90	8.36E−03
fliG	0.70	0.85	1.30E−02
fliH	0.64	0.84	3.32E−02
fliJ	0.67	0.81	9.52E−03
fliK	0.57	0.61	1.33E−03
fliL	0.85	0.99	5.44E−03
fliM	0.86	1.00	1.00E−02
fliN	0.81	0.90	1.09E−02
fliO	0.81	0.93	7.17E−03
fliP	0.73	0.75	1.25E−02
fliQ	0.81	0.92	9.67E−03
fliR	0.82	0.79	5.34E−03
fliS	1.41	1.38	7.54E−04
fliT	1.17	1.17	2.45E−04
fliY	0.59	0.52	2.77E−02
fliZ	1.08	1.12	7.42E−04
flxA	1.35	1.38	8.04E−04
gfcD	0.55	0.63	2.56E−03
intGc	0.56	0.58	5.41E−04
motA	1.53	1.51	1.20E−04
motB	1.57	1.56	4.83E−04
pdeH	1.57	1.58	8.32E−04
rcsA	0.48	0.61	1.81E−02
rdcB	0.72	0.65	1.77E−03
tap	1.59	1.48	7.85E−04
tar	1.40	1.37	2.00E−03
trg	1.08	0.87	7.04E−03
tsr	1.63	1.65	3.19E−04
ves	0.83	0.99	9.71E−03
ycgR	1.70	1.61	6.69E−04
ydjE	0.48	0.48	1.19E−01
yecR	1.06	1.06	1.27E−03
yedN	0.69	0.50	1.79E−02
yfjS	0.55	0.62	5.70E−03
yiaW	0.66	0.58	1.41E−01
yjdA	1.13	1.17	6.59E−03
ykfB	0.68	0.78	3.33E−03
ymdA	1.09	1.06	3.46E−04
ynjH	0.61	0.60	5.02E−03
Downregulated genes in ΔhdeD strain
borD	−0.75	−1.03	3.56E−02
gadX	−0.49	−0.55	5.61E−02
hdeD	−0.72	−0.87	2.89E−02
ilvC	−0.58	−0.45	3.95E−02
lrhA	−0.65	−0.62	2.78E−02
yddB	−0.62	−0.63	8.08E−03
yeeR	−1.08	−0.51	1.02E−01
ygcO	−0.69	−0.47	8.59E−02

^aTwo values of the fold change independently examined.

^bSignificance analysis between two values of fold change was performed using Student's t test with Benjamini-Horchberg false discovery rate correction.

^cPseudogene.

Increase in the intracellular level of FliA, an alternative sigma factor for flagellar biosynthesis genes, in the hdeD-deficient mutant.

The expression of flagellar genes is mainly controlled by two regulatory genes, flhDC and fliA. The flhDC genes encode a master regulator that regulates a set of class II genes. The fliA gene, belonging to class II, encodes an alternative sigma factor for promoter recognition of a set of class III genes (26). Our microarray data showed an increase in the expression of flhC, flhD, and fliA genes in ΔhdeD. RT-qPCR results confirmed that the mRNA levels of flhD, fliA, fliC, and flhA increased in ΔhdeD compared with the parent strain (Fig. 3A). In agreement with the increase in transcripts in ΔhdeD, quantitative immunodetection showed that intracellular levels of FliA were 3-fold higher in ΔhdeD than in the wild type (Fig. 3B). These results suggest that the hdeD gene is required for the negative regulation of the flhDC operon and, subsequently, class II and III genes involved in flagellum synthesis.

FIG 3.

Expression of flagellar genes are regulated by GAD system. (A) The mRNA levels of flhD, fliA, fliC, flhA, and lrhA were measured by RT-qPCR. The mRNA level was normalized relative to 16S rRNA. Two experiments were independently performed. Bars show the average with standard error. Student's t test calculated P values: P < 0.01 for fliA and lrhA and P < 0.15 for flhD, fliC, and flhA. (B) The intracellular level of FliA was measured by Western blotting using FliA serum. BW25113 (parent strain), JW3480 (ΔgadE), JW3478 (ΔhdeA), and JW3479 (ΔhdeD) were grown in M9 glycerol until log phase. Cell lysates were analyzed by SDS-PAGE and the protein bands were blotted onto the membrane. (a) FliA and RpoA were detected on the membrane as described in Materials and Methods. (b) For quantitative analysis of FliA and RNA polymerase α subunit (RpoA), Image J software was used. The levels of FliA and RpoA were measured and the ratio of FliA against RpoA, was represented. (C) The activity of lrhA promoter was measured by luciferase reporter. The pLUXlrhA is transferred into BW25113 (parent strain), JW3480ΔKm (ΔgadE), JW3478ΔKm (ΔhdeA) and JW3479ΔKm (ΔhdeD), transformants were grown in M9 glycerol until log phase, and the lrhA promoter activity was measured in each culture as described for Fig. 2. Bars show the average from three independent experiments with standard deviation. Student's t test shows P < 0.01 in comparison of ΔgadE or ΔhdeD to a parent strain. (D) Data set of genes regulated by HdeD (Table 1) were compared with those of JW3480 (ΔgadE) or JW3478 (ΔhdeA) by a cluster analysis using Cluster 3.0 software. The dendrogram is represented as a hierarchical cluster with heat map. Red and green boxes show upregulated genes and downregulated genes, respectively, in each mutant compared with parent strain.

Decrease in the promoter activation of the lrhA gene, encoding a transcriptional repressor of the flhDC operon, in the hdeD-deficient mutant.

Among the 14 known regulators of the flhDC operon, only the lrhA gene was detected as a downregulated gene in the hdeD-deficient mutant (Table 1). The lrhA gene encodes a transcriptional regulator that represses the flhDC operon in E. coli (36). RT-qPCR showed that the mRNA level of lrhA significantly decreased in ΔhdeD cells (Fig. 3A). Promoter assays using a luciferase reporter was carried out to confirm the requirement of the hdeD gene for expression of the lrhA gene. We constructed a lrhA-lux transcriptional fusion gene in pLUXlrhA and measured the luciferase activities of E. coli parent and ΔhdeD strains harboring a lrhA-lux gene grown in M9 glycerol medium. The promoter activity of lrhA decreased by more than 2-fold in ΔhdeD, indicating that HdeD is required for lrhA promoter activation (Fig. 3C). We concluded that hdeD gene negatively regulates the expression of flhDC via LrhA expression.

Transcriptional regulation of flagellum biosynthesis by GAD cluster genes.

GadE directly activates the hdeD promoter (19, 45). To gain deeper insight into the mechanism by which GadE is involved in the repression of flagellar genes by HdeD, genome expression profiles in ΔgadE were obtained, and a comparative study was performed using Cluster 3 (Fig. 3D). The results indicated that the genome expression profiles of ΔhdeD were extremely similar to those in ΔgadE (Fig. 3D). In addition, FliA levels increased by 3-fold in ΔgadE (Fig. 3B) and the lrhA promoter decreased by 3-fold (Fig. 3C), which was also similar to ΔhdeD. Thus, flagellar repression is controlled by the GadE-HdeD cascade regulation. Because GadE binds to an intergenic region between hdeD and hdeA and then activates both promoters (19, 45), it is possible that hdeA also affects the regulation of the lrhA promoter, following repression of flagellar genes. However, the hdeA-deficient (ΔhdeA) strain did not affect the lrhA promoter activity (Fig. 3C). Similarly, FliA levels and mRNA levels of the flagellar genes did not change between the parent and ΔhdeA strains (Fig. 3). Hence, hdeD, but not hdeA, is required for repression of flagellar biosynthesis and motility genes, even though both genes are activated by GadE.

Flagella and motility of hdeD-deficient mutant.

The hdeD gene repressed the expression of flagellar genes, as shown above. We observed flagellar formation in the hdeD-deficient cells using transmission electron microscopy (Fig. 4A). Most BW25113 cells had no flagella or had one flagellum on a cell (panel a in Fig. 4A). Thus, it was reported that the parent strain BW25113 had fewer flagella and is poorly motile (46). In contrast to the parent strain, ∼90% of the ΔhdeD mutant cells retained two to five flagella (panel b in Fig. 4A). The number of flagella on the ΔgadE and ΔlrhA cells was similiar to that of ΔhdeD (panels c and d in Fig. 4A). Next, we examined the motility of the parent and mutant cells on soft agar plates containing M9 glycerol medium. As expected, the parent strain BW25113 was hardly motile, whereas the ΔhdeD strain formed a swarm ring with a significantly increased diameter (panel a in Fig. 4B). Both ΔgadE and ΔlrhA strains formed swarm rings that were as large as those of ΔhdeD (panel a in Fig. 4B). Expression of hdeD genetically suppressed the motility of the ΔhdeD strain (panel b in Fig. 4B). The number of flagella on the ΔhdeD cells harboring the hdeD-expressing plasmid recovered to that of the parent strain (panel e in Fig. 4A). Thus, the hdeD gene repressed the formation of flagella and motility via the activation of lrhA gene (Fig. 5) (see Discussion for details).

FIG 4.

Flagella and motility in ΔhdeD, ΔgadE and ΔlrhA strains. (A) Representative TEM images of ΔhdeD, ΔgadE, and ΔlrhA strain cells are shown. BW25113 (parent strain) (a), JW3479 (ΔhdeD) (b), JW3480 (ΔgadE) (c), JW2284 (ΔlrhA) (d), and JW3479 harboring pBAD33hdeD (ΔhdeD/hdeD) (e) were grown in M9 glycerol until log phase, and then each cell was observed with TEM (left panels). Histograms showing the number of flagellar filaments per cell for each strain counted from more than 50 cells (right panels). Significance analysis of the number of flagella between parent and each deficient strain was performed using Welch’s t test. The P value is the following: 7.45 × 10⁻⁷ for ΔhdeD; 1.18 × 10⁻⁸ for ΔgadE; 1.11 × 10⁻³ for ΔlrhA. (B) A representative image of swarm assay. (a) A single colony of parent, ΔhdeD, ΔgadE, and ΔlrhA strains picked up using a toothpick was spotted onto M9 glycerol swarm agar and incubated for 4 h at 37°C and then overnight at room temperature. (b) A single colony of parent and ΔhdeD strains harboring pBAD33 vector and ΔhdeD harboring pBAD33hdeD picked up using a toothpick was spotted onto M9 glycerol swarm agar and incubated.

FIG 5.

A model of transcriptional repression of the flagellar biosynthesis by HdeD in GAD cascade regulation. In EvgSA → YdeO → GadE cascade regulation, referred in previous works (3, 15–20), HdeD positively regulates the lrhA promoter, which downregulates the flagellar genes.

DISCUSSION

E. coli HdeD is a small protein with 190 amino acids. The protein secondary structure prediction software SOSUI (https://harrier.nagahama-i-bio.ac.jp/sosui/) (47) showed that HdeD has six transmembrane regions (TM): TM1 (Ala²³ to Ser⁴⁵), TM2 (Ser⁵⁰ to Arg⁷²), TM3 (Phe⁷⁶ to Arg⁹⁸), TM4 (Gly¹⁰³ to Ser¹²⁵), TM5 (Gly¹³⁴ to Thr¹⁵⁶), and TM6 (Leu¹⁶³ to Leu¹⁸⁵). Moreover, computational prediction of the three-dimensional structure of HdeD with AlphaFold structure prediction (https://alphafold.ebi.ac.uk) (48) and SWISS-MODEL (https://swissmodel.expasy.org) revealed that HdeD is partially similar to the YetJ of Bacillus subtilis, the ubiquinol oxidase of E. coli, and the helirhodopsin of Thermoplasmatales archaeon. These proteins have several transmembrane helices. Among these, YetJ is a smaller protein of 217 amino acid residues with seven transmembrane regions, and its monomer functions as a pH-sensing calcium channel (49). Changes in pH in the range of pH 6 to 8 reversibly switches between closed and open conformations of YetJ by intramolecular ion bonds between different TMs via Asp and Arg residues. When YetJ is in a closed conformation at pH 8, the deprotonated Asp forms a hydrogen-bonded salt bridge with Arg, and the pK_a value of Asp is 3.1 (49). Its pK_a value is changed to 6.2 at pH 6 (49). Arg¹²² and Asp¹⁴⁵ are found in the predicted TM4 and TM5 of HdeD, respectively, implying that the pH-sensing of HdeD might occur via these amino acid residues in a similar way to B. subtilis YetJ. This presumption was validated by pK_a calculations using the PROPKA program (https://www.ddl.unimi.it/vegaol/propka.htm) (50). The pK_a values of Asp¹⁴⁵ is 3.86 at pH 7, which may be able to form the side chain hydrogen-bonded salt bridge between Asp¹⁴⁵ and Arg¹²² as well as YetJ. Thus, one possible explanation is that HfvdeD could be a pH sensor that changes structural conformations between pH 6 to 8.

The lrhA promoter is activated by the hdeD gene and represses of flagellar gene expression (Fig. 3). However, the mechanism by which the membrane-bound HdeD regulates lrhA promoter activity needs to be understood. A previous study indicated that the transcription of the lrhA promoter is regulated by positive feedback by LrhA itself (36). Therefore, HdeD may function as an enhancer of LrhA-positive feedback. LrhA belongs to the LysR-type transcriptional factor and is expected to require an effector(s) to function change for recognition of target promoters, although none of the effectors of LrhA have been identified yet. One possibility is that LrhA anchored to HdeD is released into the cytoplasm by pH-sensing HdeD. If HdeD is a similar pH-sensing channel such as YetJ, it is also possible that pH-sensing HdeD stimulates the transport of effectors for LrhA activation.

It has been shown that transcription of flagellar genes is repressed in E. coli growing at pH 8.7 rather than at pH 7.0 or pH 5.0 (51). In this study, we showed that HdeD represses the set of genes involved in flagellum formation in E. coli grown in M9-glycerol medium at pH 7.0 (Fig. 3 and 4). It is probable that HdeD may sense a slight change in extracellular pH, which is continuously made up of proton transition across the membrane by proton-driven proteins. Even under growth conditions at pH 7.0, E. coli cells partially and slightly changed the level of extracellular protons due to aerobic respiration. When E. coli cells utilize glycerol as a carbon source, the activated glycerol-3-phosphate dehydrogenase GlpD forms an aerobic respiratory chain with cytochrome bo3 ubiquinol oxidase CyoABCD, resulting in the movement of a total of four protons from the cytosol to the periplasm per reaction (52). These protons could be used as proton motive forces for proton-driven macromolecules such as F-type ATPase and/or flagellum. Therefore, E. coli must have the proper number of macromolecules of F-type ATPase and flagella per cell to avoid the overdose of protons when low respiratory quotient nutrients such as glycerol are utilized. HdeD could sense a slight increase in pH in the range between pH 7 and 8, and regulate the flagellar number on the cell to maintain an available level of proton motive force for growth.

Here, we show the linkage of transcriptional regulation between the GAD system and flagellar synthesis via HdeD and LrhA. In pathogenic enterohemorrhagic E. coli (EHEC), LrhA also regulates genes for type III secretion system and flagellar synthesis (53). HdeD is conserved among pathogenic bacteria such as enteropathogenic E. coli, EHEC, enterotoxigenic E. coli, and Shigella. Solving the HdeD structure in various pH conditions at a higher resolution will help to better understand the pH-sensitive virulence expression of pathogenic bacteria in the future.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The E. coli strains and plasmids used in this study are listed in Table 2. The gene knockout strains, Keio strains, were obtained from the NBRP E. coli strain. Removal of the kanamycin cassette in Keio strain was performed according to the original method (54), resulting in the isolation of JW3478ΔKm, JW3479ΔKm, and JW3480ΔKm (Table 2). E. coli cells were grown at 37°C in M9-Glycerol medium (40 mM Na₂HPO₄, 20 mM KH₂PO₄, 20 mM NH₄Cl, 1 mM MgCl₂, 0.25 mM K₂SO₄, 0.1 mM CaCl₂, 8 mM NaCl, 10 mM glycerol, pH 7.0) containing 0.2% Casamino Acids. The standard procedure of cell cultivation was as follows: a single colony was isolated from an overnight culture on an LB agar plate and inoculated into fresh M9-glycerol-0.2% Casamino Acid medium. This liquid culture was grown overnight at 37°C, and the overnight culture was diluted 100-fold into fresh M9-glycerol-0.2% Casamino Acid medium. The culture was incubated at 37°C with reciprocal shaking (160 rpm).

TABLE 2.

Bacterial strain, plasmids, and oligonucleotides used in this study

Name	Characterization	Reference
E. coli strains
BW25113	F− laclq rrnB3 lacZ4787 hsdR514(araBAD)567DE(rhaBAD)568 rph-1	57
JW2284	BW25113 lrhA::Kmr	54
JW3478	BW25113 hdeA::Kmr	54
JW3479	BW25113 hdeD::Kmr	54
JW3480	BW25113 gadE::Kmr	54
JW3478ΔKm	BW25113 ΔhdeA (the derivative of JW3478)	This study
JW3479ΔKm	BW25113 ΔhdeD (the derivative of JW3479)	This study
JW3480ΔKm	BW25113 ΔgadE (the derivative of JW3480)	This study
Plasmids
pLUX	Promoterless luxCDABE	44
pBAD33	Arabinose-inducible expression vector	55
pLUXhdeD	pLUX hdeD-lux	This study
pLUXlrhA	pLUX lrhA-lux	This study
pBAD33hdeD	pBAD33 hdeD	This study
Oligonucleotides
FLHA-qPCR-F	5'-CACTGCCGGATTGCTCGGGC-3'	This study
FLHA-qPCR-R	5'-CGACCATCGGGATCAGTCGA-3'	This study
FLHD-qPCR-F	5'-CGCAAATGGTTAAGCTGGCA-3'	This study
FLHD-qPCR-R	5'-CAGCGCTTCTTCAGGCTGAT-3'	This study
FLIA-qPCR-F	5'-CGTTTAGGGATCGATATTGC-3'	This study
FLIA-qPCR-R	5'-ACCCGCTGGCGCAGATTACT-3'	This study
FLIC-qPCR-F	5'-CTTTCTACGGAAGCAGCCAC-3'	This study
FLIC-qPCR-R	5'-GCCGTTGCTCCAGTCGCCAT-3'	This study
LRHA-qPCR-F	5'-CCGCGTTAAACGTAATGCCT-3'	This study
LRHA-qPCR-R	5'-GCAATACAAGAGGGATCGGC-3'	This study
16s-982	5'-CGATGCAACGCGAAGAACCT-3'	18
16s-1143	5'-GCCGGACCGCTGGCAACAAA-3'	18
HDED-LUX-F	5'-TCGTCTTCACCTCGAAGCTGGGGTTACGGTTGCAATACCC-3'	This study
HDED-LUX-R	5'-ACTAACTAGAGGATCAGAACCACCCTATAAAATTAAGAAG-3'	This study
LRHA-LUX-F	5'-TCGTCTTCACCTCGAGGGGGCAAGGTGTGAAACTATTGTT-3'	This study
LRHA-LUX-R	5'-ACTAACTAGAGGATCTTATCGGACGATTTGCACTTATCAT-3'	This study
Lux-R	5'-GGCAGGTAAACACTATTATCACC-3'	44
HDED-SacI-F	5'-GGGCTAGCGAATTCGAGGAGGAATTCACCATGTTATATATAGATAAGG-3'	This study
HDED-HindIII-R	5'-CAAAACAGCCAAGCTTTATTGCTGCTTAACGAACA-3'	This study
pBAD5’-3	5'-ATTAGCGGATCCTACCTGAC-3'	This study

Construction of the hdeD-expressing plasmid.

An arabinose-inducible hdeD gene was cloned into the pBAD33 plasmid (55). In brief, the hdeD DNA fragment was amplified by PCR using the genome of E. coli MG1655 as a template and a pair of primers, HDED-SacI-F and HDED-HindIII-R (Table 2). The PCR-amplified fragments were cloned into the pBAD33 vector, linearized by digestion with SacI and HindIII, using the In-Fusion HD cloning kit (Clontech). The resulting plasmid was confirmed by DNA sequencing using the pBAD5’-3 primer complementary to a vector.

Total RNA preparation.

Total RNA was prepared using the hot phenol method as described previously (18). In brief, E. coli was grown in M9-glycerol-0.2% Casamino Acid medium to an OD₆₀₀ of 0.3. Cells were harvested and total RNA was prepared. The concentration of total RNA was determined by measuring the absorbance at 260 nm. The purity of the total RNA was determined by agarose gel electrophoresis.

Measurement of promoter activity by the luciferase reporter system in E. coli.

The luciferase reporter assay was performed as previously described (56). In brief, the promoter DNA fragment was amplified by PCR using the genome of E. coli BW25113 as a template. The primers used were HDED-LUX-F and HDED-LUX-R for pLUXhdeD and LRHA-LUX-F and LRHA-LUX-R for pLUXlrhA (Table 2). The PCR-amplified fragments were cloned into the pLUX vector using the In-Fusion HD cloning kit (Clontech) and then confirmed by DNA sequencing using the Lux-R primer complementary to luxC in a vector. A single colony of a strain freshly transformed with one of the luciferase reporter plasmids was grown in M9-glycerol-0.2% Casamino Acid medium supplemented with 50 μg · ml⁻¹ kanamycin to OD₆₀₀  =  0.3. Luciferase activity in E. coli was measured using a plate reader (Corona). An average of the Lux/OD₆₀₀ values of the three technical replicates was taken.

Transcriptome analysis.

Transcriptome analysis was performed as described previously (18). The Cy3- or Cy5- labeled cDNAs were prepared using total RNA and the FairPlay III Microarray Labeling kit (Agilent). Cy3- and Cy5-labeled cDNA were mixed and applied to the E. coli Gene Expression Microarray 8 × 15 K (Agilent). After hybridization, the DNA chip was scanned using an Agilent G2565CA microarray scanner (version 8.1). The intensities of both Cy3 and Cy5 were quantified using Feature Extraction version 8.1, and then the Cy5/Cy3 ratios were calculated from the normalized values.

RT-qPCR.

Preparation of cDNA and quantitative PCR (qPCR) were performed as previously described (18). The primer pairs used were as follows: FLHA-qPCR-F and FLHA-qPCR-R for flhA; FLHD-qPCR-F and FLHD-qPCR-R for flhD; FLIA-qPCR-F and FLIA-qPCR-R for fliA; FLIC-qPCR-F and FLIC-qPCR-R for fliC; and LRHA-qPCR-F and LRHA -qPCR-R for lrhA (Table 2). The mRNA levels of the 16S rRNA gene, detected by qPCR using the primers 16s-982 and 16s-1143, were used for normalization of data, and the relative mRNA levels were quantified using the delta-delta method.

Western blot analysis.

E. coli cells grown in M9-glycerol-0.2% Casamino Acid medium were harvested by centrifugation, resuspended in lysis buffer containing 8 M urea, and sonicated. After centrifugation, 15 μg of the protein was subjected to 15% SDS-PAGE and blotted onto polyvinylidene difluoride membranes using an iBlot semidry transfer apparatus (Invitrogen). Membranes were stained with the following antibodies: a mixture of FliA serum and anti-RpoA (Neoclone) as the primary antibody and a mixture of horseradish peroxidase (HRP)-conjugated anti-rabbit IgG (Nacalai Tesque) and HRP-conjugated anti-rabbit IgG (Nacalai Tesque) as the secondary antibody. The FilA and RpoA bands were developed with a chemiluminescence kit (Nacalai Tesque) and detected using a LAS-4000 IR multi-color imager (Fuji Film).

Swarm assay.

Swarm plates containing M9-glycerol-0.2% Casamino Acid medium with 0.3% agar were used for the swarm assay. A single colony picked from the LB plate using a toothpick was spotted. Swarm plates were incubated for 4 h at 37°C, followed by overnight incubation at room temperature (25°C). For strains harboring a pBAD33-based plasmid, 20 μg · ml⁻¹ chloramphenicol and 0.1 mM arabinose were supplemented in both swarm and LB plates.

Transmission Electron Microscopy.

Cells were grown in M9-glycerol-0.2% Casamino Acid medium for overnight by shaking at 37°C. For strains containing a pBAD33 plasmid, 20 μg · ml⁻¹ chloramphenicol and 0.1 mM arabinose were supplemented. One aliquot of culture medium was diluted with 9 ml of distilled water (DW) and centrifuged at 5000 rpm for 5 min. The cells resuspended in 100 μl DW were negatively stained with 1% (wt/vol) phosphotungstic acid (pH 7.0) and observed using a JEM-1200EXII electron microscope (JEOL). Micrographs were obtained at an accelerating voltage of 60 kV.

Data availability.

The microarray data available have been deposited with the GEO (accession code: GSE178954).