Fatty Acid Homeostasis Tunes Flagellar Motility by Activating Phase 2 Flagellin Expression, Contributing to Salmonella Gut Colonization

ABSTRACT

Long-chain-fatty-acid (LCFA) metabolism is a fundamental cellular process in bacteria that is involved in lipid homeostasis, energy production, and infection. However, the role of LCFA metabolism in Salmonella enterica serovar Typhimurium (S. Tm) gut infection remains unclear. Here, using a murine gastroenteritis infection model, we demonstrate involvement of LCFA metabolism in S. Tm gut colonization. The LCFA metabolism-associated transcriptional regulator FadR contributes to S. Tm gut colonization. fadR deletion alters the gene expression profile and leads to aberrant flagellar motility of S. Tm. Colonization defects in the fadR mutant are attributable to altered swimming behavior characterized by less frequently smooth swimming, resulting from reduced expression of the phase 2 flagellin FljB. Notably, changes in lipid LCFA composition by fadR deletion lead to reduced expression of fljB, which is restored by exogenous LCFA. Therefore, LCFA homeostasis may maintain proper flagellar motility by activating fljB expression, contributing to S. Tm gut colonization. Our findings improve the understanding of the effect of luminal LCFA on the virulence of enteric pathogens.

INTRODUCTION

Long-chain fatty acid (LCFA) is a constituent of the phospholipid membranes in bacteria that are required for membrane homeostasis. Additionally, LCFA plays a role in bacterial metabolism by acting as energy sources. Bacteria can synthesize LCFA, whereas external LCFA is taken up by bacteria. Transported LCFA is degraded by β-oxidation, leading to the generation of acetyl coenzyme A (acetyl CoA) that is further metabolized, and thus bacteria obtain energy and precursors for cellular biosynthesis. In Gram-negative bacteria, an array of proteins known as fatty acid degradation (Fad) proteins are involved in LCFA metabolism (see Fig. S1A in the supplemental material). LCFA metabolism in bacteria is regulated through transcriptional regulator FadR-dependent expression of fad genes (1, 2).

Enteric pathogens sense intestinal metabolites as environmental signals and adapt to the environment by altering their metabolic profile and virulence strategies, and thus, they can colonize the intestinal tract and cause disease accompanied by gut inflammation and diarrhea (3, 4). LCFA is a critical environmental signal that regulates the expression of virulence genes in some bacterial pathogens (5–9). Furthermore, a high-fat diet such as a Western-type diet without fiber significantly influences the gut microbiota and infection with enteric pathogens (10, 11). These facts highlight the importance of LCFA metabolism in gut microbes, including commensals and bacterial pathogens.

Salmonella enterica serovar Typhimurium (S. Tm) is a common enteric pathogen that causes foodborne diarrhea globally (12). For initial gut colonization, S. Tm employs flagellar motility to accumulate at the mucosal surface, which is also required for efficient induction of inflammation (13). Subsequently, S. Tm invades mucosal epithelial cells and elicits gut inflammation using the SPI-1-encoded type III secretion system (T3SS-1) through the promotion of actin cytoskeletal rearrangement and activation of the NAIP/NLRC4 inflammasome accompanied by secretion of proinflammatory cytokines (14–16). Furthermore, flagellar motility in the inflamed mucosa fosters S. Tm growth and leads to continued infection (17). Unfortunately, the inflammatory responses fuel luminal growth of S. Tm, allowing for further pathogen colonization in the gut lumen and transmission to a new host (18). Thereafter, S. Tm is outcompeted by the restored microbiota and microbiota-produced metabolites, leading to pathogen clearance from the gut lumen (19–21).

The critical role of LCFA metabolism in infection by pathogens, including Salmonella, was demonstrated previously (22–30). Most of these studies focused on the link between LCFA metabolism and systemic infection using a murine typhoid infection model. However, the intestinal levels of LCFA were shown to act as a signal that allows S. Tm to be located at the infection site (8). Therefore, we investigated the impact of LCFA metabolism on Salmonella gut infection using a murine gastroenteritis infection model (14, 31).

RESULTS

FadR contributes to S. Tm gut colonization.

To reveal the involvement of LCFA metabolism in Salmonella gut infection, we performed a competitive-infection (CI) experiment by using S. Tm fad mutants which were verified by examining LCFA requirement for in vitro growth and gene expression (Fig. S1A to F). Streptomycin-pretreated C57BL/6 mice were infected with a 1:1 mixture of the wild-type (wt) strain and Δfad mutants (ΔfadR, ΔfadL, ΔfadD, ΔfadE, and ΔfadBA, respectively) via oral gavage. In this model, S. Tm colonized the cecum and colon but not the small intestine (31). On day 4 postinfection, the mice were dissected, and bacterial loads were counted in the feces and cecal contents. The ΔfadR mutant displayed competitive colonization defects in the feces and cecal contents (Fig. 1A and B). Similarly, a slight decrease in colonization levels of the ΔfadL and ΔfadE mutants was observed (Fig. 1A and B). In contrast, the ΔfadD and ΔfadBA mutants showed no colonization defects (Fig. 1A and B). Finally, we observed that introduction of a plasmid encoding fadR into the ΔfadR mutant restored competitive colonization, which was not a plasmid effect (Fig. 1A and B). Collectively, these results indicate that FadR (and FadL and FadE, to a lesser extent) contributes to gut colonization by S. Tm.

FIG 1.

Competitive colonization of S. Tm Δfad mutants on coinfection with the wild-type (wt) strain. (A to D) Streptomycin-pretreated C57BL/6 mice (n = 6 to 8 per group) were infected by gavage via the oral route with a 1:1 mixture (total, 5  × 10⁷ to 10 × 10⁷ CFU) of S. Tm wild-type strain (wt) and fad mutants. Competitive index (CI) of S. Tm loads in the feces (A and C) and cecal contents (B and D) on day 4 postinfection is shown. (E) Streptomycin-pretreated C57BL/6 mice (n = 8 per group) were infected by gavage via the oral route with a 1:1 mixture (total, 5 × 10⁷ to 10 × 10⁷ CFU) of the ΔfadD and ΔfadD ΔfadR mutants. CI of S. Tm loads in the feces and cecal contents on day 4 postinfection is shown. Bar shows medians. ns, not significant. *, P < 0.05; **, P < 0.01; Wilcoxon signed-rank test.

To investigate the involvement of secondary Fad proteins (YdiD, FadI, and FadJ) in S. Tm gut colonization, which are required for LCFA degradation under anaerobic conditions (32), we constructed ΔfadD ΔydiD and ΔfadBA ΔfadIJ mutants, and CI experiments were performed as for the wt. The ΔfadD ΔydiD and ΔfadBA ΔfadIJ mutants displayed no colonization defects in the feces and cecal contents compared to the wt (Fig. 1C and D). These data suggest that YdiD, FadI, and FadJ are not involved in gut colonization.

The ΔfadD mutant leads to a buildup of LCFA in the cytoplasm and thereby represses the T3SS-1 activities (8), as seen in our results (Fig. S1E and S1F). We thus investigated the link between FadR-dependent colonization and LCFA-repressing T3SS-1 by using a strain with an fadD deletion genetic background. Similar CI experiments with the ΔfadD ΔfadR and ΔfadD mutants were performed. The ΔfadD ΔfadR mutant was outcompeted by the ΔfadD mutant in the feces and cecal contents (Fig. 1E). The results indicate that repression of T3SS-1 by LCFA does not affect FadR-dependent gut colonization.

Absence of fadR affects the gene expression profile of S. Tm.

To delineate the mechanism by which the ΔfadR mutant is impaired in gut colonization, we compared the gene expression profiles of the wt and the ΔfadR mutant. RNA was isolated from the cultures of the wt and ΔfadR mutant grown to middle logarithmic growth phase in LB medium, followed by RNA sequencing (RNA-seq). Principal-component analysis of the RNA-seq data revealed two clusters (Fig. 2A), suggesting that the gene expression profile of the ΔfadR mutant was altered compared to that of the wt. The transcription of 109 genes was altered in the ΔfadR mutant relative to that in the wt (Tables 1 and 2). Of these, 76 genes were significantly downregulated (Table 1) and 33 genes were upregulated (Table 2). The genes showing altered expression were classified into several broad functional classes (Fig. 2B), which suggested that FadR participates in regulation of the expression of genes unrelated to LCFA metabolism.

FIG 2.

Transcriptome analysis of the S. Tm wt strain and ΔfadR mutant. (A) Principal-component analysis plot of the RNA-seq data of triplicate samples of the S. Tm wt and ΔfadR mutant. (B) Classification of transcripts significantly up- or downregulated in the ΔfadR mutant.

TABLE 1.

Genes with decreased expression in the ΔfadR mutant relative to that in the wild-type strain^a

Function and gene	Product	Fold change	P adj
Lipid metabolism
fadL	Long-chain fatty acid transport protein	−5.13	5.2.E−26
fabA	3-Hydroxyacyl-[acyl-carrier protein] dehydratase	−2.82	8.4.E−30
Peptidoglycan biosynthesis and degradation
lpp	Major outer membrane lipoprotein	−3.80	6.6.E−03
Unknown
SL1344_0698	Hypothetical protein	−3.77	NA
ypfM	Conserved sORF	−3.60	1.5.E−04
pspE	Bacteriophage shock protein E precursor	−2.94	2.3.E−09
yoaG	Conserved hypothetical protein	−2.89	NA
ibsC	Conserved sORF	−2.75	2.6.E−06
ytfK	Conserved hypothetical protein	−2.73	8.0.E−08
ybhT	Hypothetical membrane protein	−2.70	1.0.E−06
SL1344_4431	Uncharacterized protein	−2.68	5.2.E−03
yhjH	Conserved hypothetical protein	−2.67	2.6.E−09
SL1344_1032	Hypothetical protein	−2.62	1.7.E−04
yfcZ	Conserved hypothetical protein	−2.60	1.7.E−11
ypdK	Conserved sORF	−2.59	2.8.E−04
orfX	Conserved hypothetical protein	−2.54	1.2.E−03
SL1344_4239	Hypothetical protein	−2.51	1.2.E−02
dsrB	Uncharacterized protein	−2.30	1.2.E−11
SL1344_1737	Conserved hypothetical protein	−2.28	9.7.E−08
yjfO	Hypothetical exported protein	−2.26	6.3.E−04
SL1344_2732	Hypothetical transcriptional regulator	−2.25	6.3.E−09
ydfZ	Conserved hypothetical protein	−2.22	1.0.E−07
ydcX	Hypothetical inner membrane protein	−2.22	2.8.E−08
yecF	Conserved hypothetical protein	−2.17	2.4.E−12
SL1344_1235	Hypothetical exported protein	−2.12	1.7.E−13
SL1344_P1_0040	Predicted protein	−2.12	3.8.E−03
b2145	Hypothetical membrane protein	−2.09	1.8.E−08
ybdQ	Conserved hypothetical protein	−2.08	7.1.E−06
yciN	Conserved hypothetical protein	−2.07	2.4.E−08
ycjE	Conserved hypothetical protein	−2.07	NA
SL1344_0322	Conserved hypothetical protein	−2.07	2.6.E−02
SL1344_0323	Hypothetical permease/MSF transporter	−2.05	1.0.E−02
bssS	Conserved hypothetical protein (biofilm regulator)	−2.03	8.1.E−08
yaiZ	Hypothetical membrane protein	−2.02	9.5.E−05
yjbE	Hypothetical exported protein	−2.01	2.8.E−02
hybG	Hydrogenase 2 component protein	−2.01	1.4.E−03
Motility and chemotaxis
fljA	Repressor of phase 1 flagellin gene	−3.76	9.7.E−12
fljB	Flagellin	−3.63	2.0.E−57
cheW	Purine binding chemotaxis protein	−2.36	1.1.E−26
trg	Methyl-accepting chemotaxis protein III	−2.23	1.2.E−06
SL1344_3189 (mcpC)	Methyl-accepting chemotaxis protein II	−2.08	3.0.E−05
Noncoding RNA
STnc1460	Small RNA (240 nt)	−3.40	5.2.E−07
RyfA	Small RNA (294 nt)	−3.31	1.0.E−05
STnc880	Small RNA (79 nt)	−2.70	2.8.E−04
STnc1110	Small RNA (194 nt)	−2.58	1.2.E−08
GcvB	Small RNA (201 nt)	−2.52	1.2.E−06
STnc840	Small RNA (73 nt)	−2.49	NA
RygD	Small RNA (145 nt)	−2.49	5.8.E−04
RyeA	Small RNA (311 nt)	−2.41	4.7.E−05
SdsR	Small RNA (103 nt)	−2.41	5.3.E−05
RyeC	Small RNA (138 nt)	−2.28	2.1.E−03
finP	Antisense RNA-determining region (168 nt)	−2.28	NA
STnc1060	Small RNA (66 nt)	−2.25	9.1.E−04
STnc710	Small RNA (67 nt)	−2.24	9.0.E−06
STnc1080	Small RNA (62 nt)	−2.20	7.7.E−04
OmrA	Small RNA (87 nt)	−2.16	NA
Transcription factor
tdcA	LysR family transcriptional regulator	−2.73	1.8.E−02
cspE	Cold shock-like protein	−2.55	5.0.E−05
Ribosome
rpsO	30S ribosomal subunit protein S15	−2.63	3.9.E−06
rpsU	30S ribosomal subunit protein S21	−2.45	3.7.E−05
rpsV	30S ribosomal protein S22	−2.31	1.0.E−07
rpmH	50S ribosomal protein L34	−2.20	1.1.E−04
rpsQ	30S ribosomal subunit protein S17	−2.13	4.4.E−08
rpmG	50S ribosomal subunit protein L33	−2.12	4.6.E−06
rplY	50S ribosomal protein L25	−2.12	4.9.E−05
rpmA	50S ribosomal subunit protein L27	−2.09	1.7.E−04
Chaperones and folding catalysts
grxA	Glutaredoxin 1	−2.58	2.4.E−08
ppiC	Peptidyl-prolyl cis-trans isomerase C	−2.17	2.8.E−03
pndC	Counter protein for PndA	−2.14	1.0.E−05
Translation factor
infA	Initiation factor IF-1	−2.32	6.6.E−05
Carbohydrate metabolism
tdcD	Propionate kinase	−2.31	1.7.E−01
SL1344_3749	Hypothetical IIB component of PTS system	−2.23	7.0.E−04
Pseudogene
SL1344_3673A	Pseudogene (120 nt)	−2.21	3.9.E−02
SL1344_1874A	Pseudogene (132 nt)	−2.18	1.4.E−02
Secretion
tatE	Sec-independent protein translocase	−2.09	2.9.E−07
Bacteriophage related
pagK	Bacteriophage-encoded PagK	−2.02	1.6.E−02

^aP_adj, adjusted P value; sORF, small open reading frame; MSF, major facilitator superfamily; NA, not available; nt, nucleotides; PTS, phosphotransferase system.

TABLE 2.

Genes with increased expression in the ΔfadR mutant relative to that in the wild-type strain^a

Function and gene	Product	Fold change	P adj
Lipid metabolism
fadA	Fatty acid oxidation complex beta subunit	12.51	4.5.E−42
fadB	Fatty acid oxidation complex alpha subunit	11.81	1.5.E−57
fadI	3-Ketoacyl-CoA thiolase	5.95	3.7.E−14
fadE	Acyl-CoA dehydrogenase	4.60	4.9.E−19
fadJ	Fatty acid oxidation complex alpha subunit	3.28	6.6.E−11
fadH	2, 4-Dienoyl-CoA reductase	3.25	5.7.E−12
Carbohydrate metabolism
aceB	Malate synthase A	8.34	2.6.E−61
aceA	Isocitrate lyase	5.37	2.4.E−20
sgaH	Hexulose-6-phosphate synthase	2.49	NA
iolA	Malonate-semialdehydrogenase	2.21	1.6.E−02
sucC	Succinyl-CoA synthetase beta chain	2.09	1.3.E−04
Unknown
ybfA	Hypothetical exported protein	4.25	7.0.E−09
hycH	Formate hydrogen lyase maturation protein	2.89	NA
SL1344_1780	Conserved hypothetical protein	2.29	2.5.E−10
ytfJ	Hypothetical exported protein	2.15	2.6.E−14
SL1344_1344	Hypothetical pathogenicity island protein	2.03	NA
yrbN	Conserved sORF	2.03	2.2.E−02
Bacteriophage related
SL1344_1942	Predicted bacteriophage protein	3.90	NA
SL1344_1951	Predicted bacteriophage protein	2.28	NA
SL1344_1947	Capsid-related protein	2.00	NA
DNA repair and recombination
phrB	Deoxyribodipyrimidine photolyase	3.45	5.6.E−27
Secretion
ssaK	Type III secretion system apparatus	2.97	NA
ssaJ	Type III secretion system apparatus	2.15	NA
Noncoding RNA
STnc1220	Small RNA (73 nt)	2.95	NA
Amino acid metabolism
SL1344_2154	Fumarylpyruvate hydrolase	2.68	NA
Fimbriae
fimA	Type 1 fimbrial protein	2.54	7.6.E−17
fimI	Major pilin protein	2.33	3.9.E−05
fimD	Outer membrane usher protein	2.06	6.5.E−02
Chaperones and folding catalysts
cpxP	Extracytoplasmic stress protein for protein-mediated toxicities	2.49	4.0.E−04
Metabolism of terpenoids and polyketides
entA	2,3-Dihydro-2,3-dihydroxybenzoate dehydrogenase	2.17	3.4.E−06
entB	Isochorismatase	2.01	4.6.E−04
Bacterial toxin
yqfA	Hemolysin III	2.12	1.9.E−25
Transporter
sufC	Hypothetical ABC transport ATP-binding subunit	2.11	5.4.E−03

^aNA, not applicable.

Altered flagellar motility in the ΔfadR mutant influences colonization defects.

Given that several genes involved in bacterial motility and chemotaxis were downregulated following deletion of fadR (Fig. 2B and Table 1), we hypothesized that FadR controls flagellar motility. Thus, we investigated the motility of the ΔfadR mutant by using soft agar containing LB and 0.3% (wt/vol) agar. The ΔfadR mutant showed reduced motility compared to that of the wt, whereas the ΔfliGHI mutant lacking flagella was nonmotile (Fig. S2A and B). Furthermore, the complemented strain spread on soft agar like the wt. Next, swarming was investigated using soft agar containing 0.5% (wt/vol) agar and 0.5% (wt/vol) glucose. The wt spread on the swarming agar more than the ΔfliGHI mutant (Fig. 3A). The ΔfadR mutant exhibited faster spreading than did the wt, which was restored in the complementation strain. These results indicate that FadR is involved in proper flagellar motility.

FIG 3.

The ΔfadR mutant displays aberrant flagellar motility, which is involved in colonization defects. (A and B) Swarming on soft agar. S. Tm strains as indicated were spotted onto an LB plate containing 0.5% agar and 0.5% glucose and incubated at 37°C for 8 h. (C) Streptomycin-pretreated C57BL/6 mice (n = 10) were infected by gavage via the oral route with a 1:1 mixture (total, 5 × 10⁷ to 10 × 10⁷ CFU) of S. Tm ΔfliGHI and ΔfliGHI ΔfadR mutants. CI of S. Tm loads in the feces and cecal content on day 4 postinfection is shown. Bar shows median. *, P < 0.05; Wilcoxon signed-rank test and Mann-Whitney U test.

Next, to reveal a causal link between flagellar motility and gut colonization by the ΔfadR mutant, we performed CI experiments to compare the colonization levels of the ΔfliGHI and ΔfliGHI ΔfadR mutants. The nonmotile phenotype of the ΔfliGHI ΔfadR mutant was confirmed in a soft agar assay for swarming (Fig. 3B). In CI experiments, fecal colonization levels of the ΔfliGHI ΔfadR mutant were slightly reduced in comparison with that of the ΔfliGHI mutant (Fig. 3C). In contrast, no colonization defects were observed in the cecal contents. Furthermore, the CI values were lower than those of the wt versus the ΔfadR mutant from Fig. 1A and B. These results indicate that the nonmotile phenotype cancels out the colonization defects of the ΔfadR mutant and that flagellar motility plays a critical role in FadR-dependent colonization.

FadR-regulated fljBA genes influence S. Tm gut colonization.

To identify the FadR-dependent gut luminal colonization factors of S. Tm, we thus focused on the following FadR-regulated genes involved in motility: fljBA (Fig. 4A). fljB encodes the phase 2 flagellin (33), and FljA posttranscriptionally represses fliC, encoding the phase 1 flagellin (34). FadR-dependent expression of the fljBA was verified using quantitative PCR (qPCR) analysis (Fig. S3A), confirming that the expression of fljB depends on FadR. We further evaluated FljB protein levels, which indicated that the regulation of fljB expression by FadR affects protein expression levels (Fig. S3B and C).

FIG 4.

Competitive colonization of S. Tm lacking FadR-regulated genes. (A) Global changes in gene expression between S. Tm wt and ΔfadR strains as inferred from RNA-seq. (B) Schematic of FadR-regulated fljBA gene loci. Identified FadR-regulated genes, fljBA, are highlighted in yellow. (C and D) Streptomycin-pretreated C57BL/6 mice (n = 6 or 10 per group) were infected by gavage via the oral route with 1:1 mixture (total, 5 × 10⁷ to 10 × 10⁷ CFU) of S. Tm wt and Δ fljBA strains (C) or ΔfljBA and ΔfljBA ΔfadR mutants (D). CI of S. Tm loads in the feces and cecal contents on day 4 postinfection is shown. Bar shows median. *, P < 0.05; Wilcoxon signed-rank test and Mann-Whitney U test.

Flagellar motility is required for efficient induction of murine S. Tm-induced colitis (13). Since fljA is located adjacent to fljB and transcribed together with fljB (Fig. 4B), we constructed an additional S. Tm mutant lacking fljBA to reflect the expression pattern of the ΔfadR mutant. Subsequently, the fljBA mutant strain, along with the wt, was subjected to the CI experiments. The ΔfljBA mutant showed reduced gut colonization in the feces and cecal content compared to that of the wt (Fig. 4C). These results raise the possibility that colonization defects of the ΔfadR mutant are attributable to reduced expression of the FadR-regulated genes, fljBA.

To demonstrate the requirement of expression of fljBA in FadR-dependent colonization, we performed CI experiments with the ΔfljBA ΔfadR mutant. Streptomycin-pretreated mice were infected with a 1:1 mixture of the ΔfljBA and ΔfljBA ΔfadR mutants by oral gavage. The colonization levels of the ΔfljBA ΔfadR mutant in feces and cecal contents were equivalent to those of the ΔfljBA mutant (Fig. 4D). The CI values were not statistically significant in comparison with those of the ΔfliGHI mutant versus the ΔfliGHI ΔfadR mutant from Fig. 3C.

To determine whether fljB, fljA, or both contribute to the FadR-dependent colonization, we first performed CI experiments with the wt and the ΔfljB mutant and with the wt and the ΔfljA mutant. The colonization levels of the ΔfljB mutant in feces were reduced compared to those of the wt, whereas the ΔfljA mutant colonized the gut lumen at levels similar to those of the wt (Fig. S4A). Thus, we next performed CI experiments with the ΔfljB and ΔfljB ΔfadR mutants. The colonization levels in feces and cecal contents between the ΔfljB and ΔfljB ΔfadR mutants were not statistically significant (Fig. S4B). Furthermore, the CI values were lower than those of the wt versus the ΔfadR mutant from Fig. 1A and B. These results indicate that fljBA expression is required for FadR-dependent colonization.

Reduced smooth swimming of the ΔfadR mutant plays a role in the gut colonization defects.

Flagellar phase variation alters behavior of flagellar motility, and thereby influences virulence during gastrointestinal infection with S. Tm (35). Thus, we hypothesized that reduced expression levels of fljBA in the ΔfadR mutant alter the behavior of flagellar motility, resulting in less efficient interactions with epithelial cells and gut colonization. Tracing experiments with individual S. Tm cells swimming near the microscope coverslip in liquid culture showed that S. Tm wt cells moved with smooth swimming and tumbling, or they stopped moving (Fig. 5A and B). As controls, enhanced smooth swimming of the ΔcheY mutant, incapable of promoting flagellar motor reversals, was confirmed. In contrast, the ΔcheB mutant, overproducing phosphorylated CheY, displayed minimal smooth swimming but showed enhanced tumbly swimming (Fig. 5A and B). The proportion of smooth swimming was reduced in the ΔfadR mutant compared to that in the wt; this reduction was restored by introducing a plasmid bearing fadR (Fig. 5B). These results indicate that deletion of fadR leads to reduced levels of smoothly swimming S. Tm cells.

FIG 5.

Reduced smooth swimming confers colonization defects in the ΔfadR mutant. (A) Representative microscopy images of green fluorescent protein (GFP)-expressing S. Tm wt, ΔcheY, and ΔcheB strains. GFP-expressing S. Tm strains were grown in LB containing 0.5% glucose to the middle logarithmic growth phase, and the cells were immediately placed on a glass slide, sealed under a glass coverslip, and observed by fluorescence microscopy. Arrows indicate smooth swimming, and arrowheads show tumbly swimming. Scale bar, 10 μm. (B) Microscopy quantification of S. Tm swimming behavior. Percentages of smooth swimming (smooth), tumbly swimming (tumbly), and stopping cells (stop) were calculated. Data are shown as the means ± standard deviations of the results from three independent experiments (n = 2,947 [wt], 1,831 [ΔfadR mutant], 1,448 [ΔfadR pfadR mutant], 2,247 [ΔcheY mutant], and 1,407 [ΔcheB mutant]). *, P < 0.05; **, P < 0.01; ***, P < 0.001; unpaired Student’s t test. (C and D) Streptomycin-pretreated C57BL/6 mice (n = 7 or 8 per group) were infected by gavage via the oral route with a 1:1 mixture (total 5 × 10⁷ to 10 × 10⁷ CFU) of S. Tm ΔcheY and ΔcheY ΔfadR or ΔcheB and ΔcheB ΔfadR mutants. CI of S. Tm loads in the feces (C) and cecal content (D) on day 4 postinfection is shown. Bar shows median. **, P < 0.01; Wilcoxon signed-rank test.

To determine the relationship between colonization defects and reduced smooth swimming in the ΔfadR mutant, we conducted CI experiments with ΔcheY ΔfadR and ΔcheB ΔfadR mutants. The ΔcheY ΔfadR mutant displayed a competitive disadvantage in colonization compared to the ΔcheY mutant (Fig. 5C and D). In contrast, gut colonization levels of the ΔcheB ΔfadR mutant were identical to those of the ΔcheB mutant (Fig. 5C and D). The results of the CI experiments suggest that swimming behavior is of particular importance in the FadR-dependent gut colonization. Collectively, these results indicate that the gut colonization defects in the ΔfadR mutant are attributable to altered swimming behavior, specifically by reduced smooth swimming.

Reduced expression of FljB, but not FliC, influences gut colonization defects in the ΔfadR mutant.

To evaluate the importance of FljB in aberrant flagellar motility and altered swimming behavior, we created flagellar-phase-locked S. Tm strains by deleting hin, encoding the recombinase responsible for DNA inversion. The phase lock was confirmed by Salmonella flagellin-specific antiserum-mediated inhibition of flagellar-based movement on soft agar (Fig. S5A). Furthermore, monophasic expression of the flagellin in individual phase-locked strains was confirmed (Fig. S5B). We thus investigated which flagellar phase is involved in faster spreading on swarming agar. The FliC-locked (FliC^ON) and FljB-locked (FljB^ON) strains moved on the swarming agar at similar levels (Fig. 6A). The FljB^ON ΔfadR mutant displayed faster spreading identical to that of the ΔfadR mutant, whereas the spreading levels of the FliC^ON ΔfadR mutant were similar to those of the FliC^ON mutant (Fig. 6A), indicating that FljB, but not FliC, participates in aberrant flagellar motility of the ΔfadR mutant.

FIG 6.

Aberrant flagellar motility of ΔfadR depends upon FljB but not FliC. (A to C) Swarming on a soft agar. S. Tm strains as indicated were spotted onto an LB plate containing 0.5% agar, 0.5% glucose, and 0.2% l-arabinose (Ara) and then incubated at 37°C for 8 h. (D) Relative expression of fljB in S. Tm strains grown in LB containing 0.5% glucose and 0.2% l-arabinose by targeted RT-qPCR. Data are the means from at least three independent experiments, and error bars indicate standard deviations. ***, P < 0.001; unpaired Student’s t test. (E) GFP-expressing S. Tm strains as the indicated were grown in LB containing 0.5% glucose to the middle logarithmic growth phase, and the cells were immediately placed on glass slide, sealed under a glass coverslip, and observed by fluorescence microscopy. Percentages of smooth swimming (smooth), tumbly swimming (tumbly), and stopping cells (stop) were calculated. Data are shown as the means ± standard deviations of the results from three independent experiments (n = 1,704 [FliC^ON mutant], 1,047 [ΔfadR FliC^ON mutant], 1,849 [FljB^ON mutant], and 1,572 [ΔfadR FljB^ON mutant]). **, P < 0.01; ***, P < 0.001; unpaired Student’s t test. (F and G) Streptomycin-pretreated C57BL/6 mice (n = 8 or 11 per group) were infected by gavage via the oral route with a 1:1 mixture (total, 5 × 10⁷ to 10 × 10⁷ CFU) of S. Tm FliC^ON and FliC^ON ΔfadR mutants or FljB^ON and FljB^ON ΔfadR mutants. CI of S. Tm loads in the feces (F) and cecal contents (G) on day 4 postinfection is shown. Bar shows the median. **, P < 0.01; Wilcoxon signed-rank test.

We next wished to determine whether reduced expression of fljBA in the ΔfadR mutant was responsible for the aberrant flagellar motility. To this end, we investigated whether inducible expression of fljBA via a plasmid restored normal motility. The ΔfadR mutant harboring pFLAG-CTC exhibited aberrant motility, whereas introduction of pFLAG-CTC carrying fljBA into the ΔfadR mutant partially restored flagellar motility (Fig. 6B).

Expression of fljBA is activated by the flagellum-specific sigma factor FliA (σ²⁸). Hence, we examined whether introduction of a plasmid expressing FliA restores the aberrant flagellar motility, in which fliA expression was induced by the presence of l-arabinose. The ΔfadR mutant moved faster than the wt on soft agar both with and without l-arabinose, whereas the spreading of the ΔfadR mutant harboring the FliA-expressing plasmid on soft agar with l-arabinose was suppressed compared with that of the ΔfadR mutant harboring the same plasmid on soft agar without l-arabinose (Fig. 6C). Further, we confirmed fljB expression in these strains. Supplementation with l-arabinose in the culture media restored fljB expression in the ΔfadR mutant harboring the FliA-expressing plasmid (Fig. 6D). These data suggest that the aberrant flagellar motility characterized by faster movement on swarming agar is attributable to reduced expression of fljBA.

We next examined the swimming behavior of flagellar-phase-locked strains. The FliC^ON and FliC^ON ΔfadR mutants showed no change in swimming behavior, whereas the proportion of smooth swimming in the FljB^ON ΔfadR mutant was decreased compared to that in the FljB^ON mutant (Fig. 6E). The FljB^ON ΔfadR mutant showed more tumbly swimming and stopping. These results indicate that FljB facilitates smooth swimming of S. Tm cells and that the reduced smooth swimming of the ΔfadR mutant depends on the presence of FljB but not on FliC.

Next, we performed CI experiments with the flagellin-locked S. Tm strains. The gut colonization levels of the FliC^ON ΔfadR mutant were identical to those of the FliC^ON mutant (Fig. 6F and G). However, the FljB^ON ΔfadR mutant showed impaired gut colonization compared to that by the FljB^ON mutant (Fig. 6F and G). These results suggest that the gut colonization defects of the ΔfadR mutant depend upon FljB but not on FliC.

Luminal LCFA restores attenuated gut colonization by the ΔfadR mutant by maintaining S. Tm lipid LCFA composition.

An electrophoretic mobility shift assay showed that FadR did not bind upstream region of fljB (Fig. S6A to C), suggesting that FadR indirectly activates fljB expression. We thus hypothesized that alterations in LCFA metabolism in the ΔfadR mutant could lead to reduced fljB expression. We examined the lipid and fatty acid profiles of the ΔfadR mutant. We first determined the proportions of three major lipids (phosphatidylethanolamine [PE], phosphatidylglycerol, and cardiolipin). The proportions of each lipid were not significantly different among the wt and the ΔfadR and ΔfadR mutants incubated with oleate (Table S1). Next, the fatty acid profile was determined for the major lipid PE. In liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-ESI-MS/MS) analysis, we found that the peak corresponding to palmitic/oleic acid differed from that of the wt with the ΔfadR mutant (Fig. S7A and B). Thus, we compared the relative ratio of palmitic/oleic acid to palmitic/palmitoleic acid in PE. The ratio of LCFAs in the ΔfadR mutant was significantly reduced compared with that of the wt (Table 3). The reduced proportion of ΔfadR was restored by addition of oleate to the culture media (Table 3 and Fig. S7C). These results indicate that the LCFA composition in the lipid is altered in the ΔfadR mutant compared to that in the wt. The unbalanced LCFA composition was restored by LCFA supplementation.

TABLE 3.

Ratio of LCFA in PE of S. Tm wild-type, ΔfadR, and ΔfadR strains with oleate^a

Ratio of PE 16:0/18:1 to PE 16:0/16:1 in strain (mean ± SD)			P (wt vs fadR)	P (wt vs fadR + oleate)
wt	fadR	fadR + oleate
1.25 ± 0.2	0.76 ± 0.1	1.36 ± 0.03	0.0043	0.38

^aRatios are means ± standard deviations of three independent experiments. P values were calculated by unpaired Student’s t tests comparing the wild-type strain with the ΔfadR mutant or the ΔfadR mutant with oleate.

Supplementing the ΔfadR mutant culture with oleate increased the proportion of smooth swimming to levels similar to those of the wt (Fig. 7A). Additionally, the levels of smooth swimming in the wt supplemented with oleate showed an increasing trend, although the increase was not significant. These data show that supplemented external LCFA can restore the reduced smooth swimming of the ΔfadR mutant.

FIG 7.

External LCFA rescues flagellar motility- and gut colonization-defective phenotypes of the ΔfadR mutant. (A and C) GFP-expressing S. Tm wt and ΔfadR strains were grown in LB containing oleate and cerulenin to the middle logarithmic growth phase, and the cells were immediately placed on a glass slide, sealed under a glass coverslip, and observed by fluorescence microscopy. Percentage of smooth swimming cells (smooth) was calculated. Data are shown as the means ± standard deviations of the results of three independent experiments. (A) n = 1,318 (wt LB-Glc), 1,460 (wt LB-Glc with oleate), 1,264 (ΔfadR LB-Glc), and 847 (ΔfadR LB-Glc with oleate). (C) n = 1,284 (wt), 1,454 (ΔfadR), 1,288 (wt cerulenin), and 1,449 (wt cerulenin and oleate). *, P < 0.05; ***, P < 0.001; unpaired Student’s t test. (B and D) Relative expression of fljB in ΔfadR in response to oleate and cerulenin by targeted RT-qPCR. Data are represented as the means from at least three independent experiments, and error bars indicate standard deviations. *, P < 0.05; ***, P < 0.001; unpaired Student’s t test. (E and F) C57BL/6 mice (n = 6 or 8 per group) on normal chow diet or high-fat diet were administered streptomycin and infected with wt and ΔfadR strains. Fecal loads of S. Tm (E) and CI of S. Tm loads in the feces (F) on days 1 and 4 postinfection are shown. dpi, days postinfection. Bar shows the median. *, P < 0.05; Mann-Whitney U test.

To demonstrate the link between oleate supplementation and fljB expression, we examined if oleate supplementation in a culture of the ΔfadR mutant restores fljB expression. The reduced expression of fljB in the ΔfadR mutant was partly restored by supplementation with oleate (Fig. 7B), which indicated that external LCFA facilitates fljB expression by the ΔfadR mutant.

To examine the effect of external LCFA on motility of the ΔfadR mutant, we next investigated expression levels of genes involved in flagellar motility upon supplementation with oleate by reverse transcription (RT)-qPCR. A number of genes are responsible for construction of the bacterial flagellum and chemotaxis. The flagellar regulon is a hierarchy of coordinately transcribed genes, which are assigned to three classes, class 1, class 2, and class 3 (36). Expression levels of all of the tested genes were not affected by supplementation with oleate (Fig. S8). The results show that in the ΔfadR mutant grown in LB supplemented with oleate, the restoration of gene expression by external LCFA is specific to fljB.

We further investigated whether inhibition of LCFA biosynthesis affects the swimming behavior of S. Tm. Addition of cerulenin, an inhibitor of bacterial LCFA biosynthesis, to the culture media reduced the proportion of smooth swimming for the wt (Fig. 7C). This reduced proportion was restored by supplementation with oleate. Similarly, expression levels of fljB were reduced following addition of cerulenin to the culture media (Fig. 7D). Supplementation with oleate partially restored the reduced expression levels. These results suggest that suppression of LCFA biosynthesis results in reduced fljB expression, which alters the swimming behavior of S. Tm.

We next investigated if external administration of LCFA in the mouse diet restored the fitness defect of the fadR mutant. CI experiments with the wt and ΔfadR mutant were performed, in which mice were fed either a normal diet or a high-fat diet. The wt colonized the gut lumen at high levels in both groups of mice (Fig. 7E). The colonization levels of the ΔfadR mutant on day 4 postinfection were decreased in both group of mice, whereas the colonization levels of the ΔfadR mutant in mice fed a high-fat diet were significantly increased compared to those in mice fed a normal chow diet (Fig. 7E). Furthermore, CI values in mice fed a high-fat diet on day 4 postinfection were lower than those in mice fed a normal chow diet (Fig. 7F). Additionally, we performed similar CI experiments by daily gavage of mice with oleate; similarly, the colonization levels of the ΔfadR mutant in mice supplemented with oleate were significantly increased compared to those in mice administered phosphate-buffered saline (PBS) orally (Fig. S9).

To further our understanding of the role of luminal LCFA, we performed CI experiments with the wt and the ΔfadD or ΔfadBA mutant, in which mice were fed a high-fat diet, and compared the CI values with those in mice fed a normal chow diet from Fig. 1A and B. CI values of the wt versus the ΔfadD mutant in feces of mice fed a high-fat diet were elevated, whereas those in the cecal contents were not statistically significant in comparison with those of mice fed a normal chow diet from Fig. 1A and B (Fig. S10A). Similarly, CI values of the wt versus the ΔfadBA mutant in feces and cecal contents of mice fed a high-fat diet were identical to those of mice fed a normal chow diet from Fig. 1A and B (Fig. S10B). These results suggest that degradation of luminal LCFA has no role in Salmonella gut colonization. Notably, given that the fecal fitness advantage of the ΔfadD mutant found in mice fed a normal chow diet is abrogated in mice fed a high-fat diet, abundant luminal LCFA in mice fed a high-fat diet represses T3SS-1 (8), which might cancel out the T3SS-1 expression-dependent retarded growth (37).

Finally, we repeated the CI experiments with the ΔcheY mutant versus the ΔcheY ΔfadR mutant or the FljB^ON mutant versus the FljB^ON ΔfadR mutant using mice fed a high-fat diet. The mutant with the cheY deletion genetic background and the FljB-locked ΔfadR mutant displayed colonization defects (Fig. 5C and D and Fig. 6C and D). In contrast, CI values of the ΔcheY mutant versus the ΔcheY ΔfadR mutant in mice fed a high-fat diet were significantly reduced compared to those of mice fed a normal chow diet (Fig. S11A). Similar results for feces, but not cecal contents, were obtained with the FljB^ON mutant versus the FljB^ON ΔfadR mutant in mice a high-fat diet (Fig. S11B). These results demonstrate the causal link between fljB expression-dependent smooth swimming and FadR-dependent gut colonization.

In summary, these results indicate that external LCFA restores the reduced smooth swimming and colonization defects of the ΔfadR mutant by maintaining the LCFA composition homeostasis in the lipid membrane.

DISCUSSION

LCFA metabolism by β-oxidation has been shown to be involved in microbial pathogenesis (22, 24–26, 29). In the case of Salmonella, β-oxidation contributes to sustained colonization of tissues in a mouse typhoid model (30). In contrast, we showed that both aerobic β-oxidation and anaerobic β-oxidation are dispensable for acute infection in the intestinal lumen, as the ΔfadBA and ΔfadBA ΔfadIJ mutants colonized the murine gut at levels identical to those in the wt. Our data are consistent with those of previous reports, which showed that LCFA exerts an environmental cue to proper intestinal localization (8), suggesting that luminal LCFA is used by Salmonella as a significant signal rather than an energy source.

In this study, we identified the FadR regulon in S. Tm. In addition to genes associated with LCFA metabolism, FadR activates the expression of genes involved in flagellar motility. Flagellar motility is an important virulence factor contributing to Salmonella gut infection (13, 17, 38). Our data showing that the colonization levels of the ΔfliGHI ΔfadR mutant in the cecum lumen are identical to those of the ΔfliGHI mutant indicate that flagellar motility is mainly involved in FadR-dependent gut colonization. Furthermore, deletion of fljBA did not induce colonization defects in the ΔfadR mutant, demonstrating that fljBA expression is required for FadR-dependent gut colonization. Additionally, in analyzing the swimming behavior of individual bacterial cells, we observed that ΔfadR cells showed less frequent smooth swimming and stopped more often. Therefore, the altered swimming behavior in ΔfadR may be due to reduced expression of the phase 2 flagellin FljB. Our data showing that the FljB^ON ΔfadR mutant, but not the FliC^ON ΔfadR mutant, exhibited aberrant flagellar motility support our hypothesis. The results of CI experiments using S. Tm strains with cheY and cheB mutated genetic backgrounds, and the flagellar phase-locked background, demonstrate that the altered swimming behavior induced by reduced expression of fljBA results in competitive colonization defects on S. Tm ΔfadR.

Environmental cues in the intestinal tract profoundly influence virulence during gastrointestinal infection with S. Tm. Short-chain fatty acids (SCFAs) such as acetate, propionate, and butyrate play critical roles by positively or negatively affecting the expression of T3SS-1 genes (39–41). Thus, the distribution and abundance of SCFAs in the intestinal tract may define the proper sites of infection for S. Tm. Luminal LCFA likely has the same effect, although through a different mechanism. LCFA imported to the bacterial cytoplasm may repress FadR and T3SS-1 activities (8). In contrast, our findings indicate that luminal LCFA contributes to maintenance of the lipid LCFA composition. LCFA composition homeostasis leads to activation of phase 2 flagellin fljBA expression, which contributes to S. Tm gut colonization. FljB-expressing S. Tm cells likely prefer luminal expansion rather than invasion into the host epithelial cells and trigger inflammation (35). Based on our findings combined with earlier reports, the presence of LCFA in the intestinal lumen may limit S. Tm invasion by suppressing T3SS-1 activities, whereas LCFA facilitates S. Tm luminal expansion by activating phase 2 flagellin.

A major limitation of this study was the absence of in vivo complementation experiments by introducing a plasmid bearing fljBA into the ΔfadR mutant. In general, abundant expression of flagellin from a high-copy-number plasmid is required for flagellin complementation, as observed in this study. However, the high-copy-number plasmid is not stable in vivo, and protein expression from the plasmid confers adverse effects on the complemented strain in vivo, leading to rapid elimination from the host. Therefore, it is challenging to use a high-copy-number plasmid in in vivo complementation experiments.

In conclusion, LCFA metabolism is involved in S. Tm gut colonization. LCFA homeostasis is required for the expression of the phase 2 flagellin and proper flagellar motility. The fljBA-hin locus was acquired during Salmonella evolution, enabling diphasic flagellin expression. Therefore, it is speculated that luminal LCFA may facilitate the gut colonization in the diphasic S. Tm.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

Bacterial strains and plasmids used in this study are listed in Table S2. S. Tm SL1344 mutants harboring chromosomal in-frame deletions were constructed using the lambda/red homologous recombination system (42) and P22 phage-mediated transduction. In the flagellar-phase-locked strain, the orientation of the invertible hin region was determined by PCR with the primers HIXL and BMID (Table S3). Bacteria were routinely grown overnight at 37°C in LB broth with agitation or on an agar plate supplemented with appropriate antibiotics, as required. Bacteria were grown in M9 minimal medium broth supplemented with 0.01% histidine as indicated.

Construction of chromosomal transcriptional lacZ reporter strains.

DNA fragments including the sicA promoter region were amplified by PCR using the primer set Pro-sicA-SalI and sicA-Rev-BamHI (Table S3), and ligated into the same sites of pLD-lacZΩ containing promoterless lacZ (43), yielding pLD-sicAZ. The sicA::lacZ allele was transduced into SL1344 via the P22 phage. Similarly, the hilA::lacZ allele (43) was introduced into SL1344 by P22 phage transduction.

β-Galactosidase assay.

The β-galactosidase assay was performed as described previously (44). Briefly, bacterial reporter strains were grown overnight in LB medium, diluted 1:100, and subcultured for 6 h in the same medium. The bacterial culture (100 μL) was added to 900 μL of Z buffer (60 mM Na₂HPO₄·7H₂O, 40 mM NaH₂PO₄·H₂O, 10 mM KCl, 1 mM MgSO₄·7H₂O, 50 mM β-mercaptoethanol), 20 μL of 0.1% (wt/vol) SDS, and 40 μL of chloroform and mixed well. After incubation for 5 min at 28°C, the reaction was started by adding 200 μL of o-nitrophenyl-β-d-galactopyranoside in 0.1 M potassium phosphate buffer (pH 7) (4 mg/mL) and developed at 28°C. The reactions were stopped by adding 500 μL of 1 M Na₂CO₃, and the optical densities at 420 nm were measured. β-Galactosidase activity units (Miller units) were calculated as previously described (45).

Mouse infection experiments.

C57BL/6 mice were housed under specific-pathogen-free conditions at the institute of experiments of animals at the School of Pharmacy, Kitasato University. All mouse experiments were reviewed and approved by the Kitasato University Institutional Animal Care and Use Committee (permit numbers 20-34 and 20-36). Mice were maintained on the mouse maintenance diet (normal chow diet, CE-2 [4.51% fat, 25.1% protein, 5.05% fiber]; CLEA Japan). Mice were shifted to a high-fat diet (high-fat diet 32 [HFD32; 32.0% fat, 25.5% protein, 2.9% fiber]; CLEA Japan) as indicated. Additional experimental groups were treated daily with phosphate-buffered saline (PBS) or 100 μL of oleic acid (Nacalai Tesque). Mouse infection experiments were performed in 6- to 12-week-old male and female mice (age and sex matched) as described previously (20). Briefly, the mice were pretreated with 25 mg of streptomycin by gavage 24 h prior to infection. For infection, bacteria grown for 12 h in LB broth supplemented with 0.3 M NaCl and appropriate antibiotics were diluted 1:20 and subcultured for 4 h in the same medium without antibiotics. The bacteria were washed twice with PBS, and mice were infected by gavage with 5 × 10⁷ to 10 × 10⁷ CFU bacteria. Mice were kept in cages with a mesh floor to avoid transmission between mice. To determine the S. Tm population sizes, the fecal pellets and cecal contents were freshly collected and homogenized in sterile PBS using a TissueLyser device (Qiagen) for differential plating onto MacConkey agar plates (Nissui Pharmaceutical) supplemented with appropriate antibiotics. The competitive index was calculated as the output ratio of the S. Tm parent strain and the derivative mutant normalized against the input ratio of the parent strain and the mutant.

Construction of complementary plasmid.

The complementary plasmid pfadR was constructed by PCR with primer set fadR-KpnI-FW and fadR-SalI-RV (Table S3) and ligation into pMW118 (Nippon Gene). The complementary plasmid carrying fljBA was created by PCR using the primers SL fljB-XhoI-FW and SL fljA-SalI-R (Table S3) and ligation into pFLAG-CTC (Sigma).

Generation of RNA sequencing data.

Triplicate cultures of S. Tm SL1344 and the fadR mutant were grown until the middle-logarithmic growth phase independently at 37°C in LB. Bacteria were isolated from the medium by centrifugation, and RNA was isolated using a Direct-zol RNA miniprep kit (Zymo Research) according to the manufacturer’s instructions. rRNA was removed using the Ribo-Zero magnetic kit (Epicentre) following the manufacturer’s protocol. Ribodepleted RNA was prepared and then inserted into strand-specific RNA libraries using the NEBNext Ultra directional RNA library prep kit for Illumina (New England BioLabs). The sequencing cDNA libraries were sequenced using an Illumina NovaSeq 6000 platform.

Analysis of RNA sequencing data.

Sequencing reads were mapped to the NCBI reference sequence with GenBank accession number NC_016810.1 using HISAT2 (46). Differential expression analysis was conducted with DESeq2 package (47). Transcripts with an adjusted P value of <0.05 and fold change using default cutoff greater than or equal to ±2.0 were considered significantly differentially expressed.

RT-qPCR.

Reverse transcription using total RNA from S. Tm SL1344 and the ΔfadR mutant was performed using TaqMan reverse transcription reagents (Invitrogen). The target genes were amplified with specific primer pairs listed Table S3, using a CFX96 real-time PCR detection system (Bio-Rad) under standard cycle conditions for SYBR Fast qPCR master mix (Kapa Biosystems). Relative transcript levels were normalized to rpoD and calculated by using the threshold cycle (2^−ΔCT) method (48).

Bacterial motility assay with soft agar plate.

Motility assay was performed by using a 0.3% agar LB plate (for swimming) or 0.5% agar LB plate supplemented with 0.5% glucose (for swarming) as previously described (49). If necessary, anti-FliC (Salmonella H-i antiserum, 1/500; Denka) and anti-FljB (Salmonella H-2 antiserum, 1/250; Denka) antisera were added to an 0.3% agar LB plate.

Analysis of swimming behavior with microscopy.

S. Tm strains were grown in LB broth supplemented with 0.5% glucose, placed on a glass slide, and sealed under a glass coverslip. The samples were observed and imaged with an exposure time of 2.6 s with a Zeiss Axio Vert.A1 microscope using an objective EC Plan-Neofluar 40×/0.75 Ph2. Swimming behavior (smooth swimming, tumbly swimming, and stop) was evaluated by visual monitoring. Smooth swimming or tumbly swimming was referred to as the movement of the ΔcheY (smooth swimming locked) or ΔcheB (tumbly swimming locked) mutant, respectively (Fig. 5A).

Flagellin secretion analysis.

Secreted proteins were isolated as previously described (49). Samples were loaded and separated on denaturing polyacrylamide gels. Western blotting was performed using primary anti-FliC (H-i antiserum), anti-FljB (H-2 antiserum), or anti-FljB (provided by Tohru Minamino [50]) and secondary alkaline phosphatase-conjugated anti-rabbit (Sigma) antibody.

Construction of FliA-expressing plasmid.

The DNA fragments including fliA were amplified by PCR using the primer set fliA-XhoI-FW and fliA-HindIII-RV (Table S3) and cloned into pBAD/His (Invitrogen), yielding pBAD-fliA expressing N-terminally His-tagged FilA proteins.

FadR protein purification.

S. Tm fadR was cloned into pBAD/His to create an N-terminally His-tagged construct, which was transformed into Escherichia coli strain TOP10. His-tagged FadR protein was purified through cobalt columns (HisTALON gravity column purification kit; Clontech Laboratories) according to the manufacturer’s instructions. Samples were dialyzed against 20 mM Na₂HPO₄-KH₂PO₄ buffer.

Electrophoretic mobility shift assay.

DNA probes were prepared by PCR from S. Tm SL1344 genomic templates using the following primer sets (Table S3): fadE fragment (PfadE) −165 to +50, SL fadE-F1 and SL fadE-R1; fljB fragment 1 (PfljB-1) −130 to +70, SL fljB-F3 and SL fljB-R1; and fljB fragment 2 (PfljB-2) −530 to +70, SL fljB-F1 and SL fljB-R1. For the electrophoretic mobility shift assay, 200 ng of PCR product was incubated with increasing concentrations of purified His-tagged FadR protein in reaction buffer (50 mM Na₂HPO₄-KH₂PO₄ at pH 7.7, 100 mM NaCl, 1.5 mM dithiothreitol [DTT], 100 μg/mL of bovine serum albumin) for 60 min at room temperature and resolved on 5% nondenaturing polyacrylamide gels in 0.5× Tris-borate-EDTA buffer. Gels were stained with Midori Green advance (Nippon Genetics) and visualized with an FAS-Digi PRO LED transilluminator (Nippon Genetics).

Lipid analysis.

Phospholipids were extracted as described by Bligh and Dyer (51). Phospholipids were separated by one-dimensional thin-layer chromatography (TLC) on Silica Gel 60 plates (Merck) in chloroform/methanol/acetic acid (65:25:13, vol/vol/vol). Phospholipids were identified by comigration with known standards. The position of each spot was visualized by staining with 0.001% primuline in acetone/H₂O (4:1, vol/vol) under UV light. The area of silica gel corresponding to each phospholipid was scraped off and subjected to phosphorus quantification (52).

LC-ESI-MS/MS analysis.

LC-ESI-MS/MS analysis was performed using a QTRAP 4500 quadrupole linear ion trap hybrid mass spectrometer (AB Sciex) coupled to a Nexera XR high-performance liquid chromatography system (Shimadzu Co.). The sample was analyzed by flow injection mode of MS/MS experiments. Phosphatidylethanolamine was detected by a neutral loss scan of 141 Da in positive ion mode, and the fatty acid composition was confirmed by product ion scanning of deprotonated ions ([M-H]) in positive ion mode.

Statistical analysis.

All experiments were performed at least three times except for mouse infection experiments, and the data were analyzed by the statistical tests, which are indicated in each figure legend. The precise numbers of animals and samples and experimental replication are described in the figure legends. Statistical significance was determined by Wilcoxon signed-rank test or Mann-Whitney U test or Student’s t test using GraphPad Prism software version 5 for Mac (GraphPad Software, Inc.). P values of less than 0.05 (two-tailed) were considered statistically significant.

Data availability.

RNA-seq raw data have been deposited at the DNA Data Bank of Japan (DDBJ) under accession number DRA010004 and is publicly available as of the date of publication.