Salmonella enterica Serovar Kentucky Flagella Are Required for Broiler Skin Adhesion and Caco-2 Cell Invasion

ABSTRACT

Nontyphoidal Salmonella strains are the main source of pathogenic bacterial contamination in the poultry industry. Recently, Salmonella enterica serovar Kentucky has been recognized as the most prominent serovar on carcasses in poultry-processing plants. Previous studies showed that flagella are one of the main factors that contribute to bacterial attachment to broiler skin. However, the precise role of flagella and the mechanism of attachment are unknown. There are two different flagellar subunits (fliC and fljB) expressed alternatively in Salmonella enterica serovars using phase variation. Here, by making deletions in genes encoding flagellar structural subunits (flgK, fliC, and fljB), and flagellar motor (motA), we were able to differentiate the role of flagella and their rotary motion in the colonization of broiler skin and cellular attachment. Utilizing a broiler skin assay, we demonstrated that the presence of FliC is necessary for attachment to broiler skin. Expression of the alternative flagellar subunit FljB enables Salmonella motility, but this subunit is unable to mediate tight attachment. Deletion of the flgK gene prevents proper flagellar assembly, making Salmonella significantly less adherent to broiler skin than the wild type. S. Kentucky with deletions in all three structural genes, fliC, fljB, and flgK, as well as a flagellar motor mutant (motA), exhibited less adhesion and invasion of Caco-2 cells, while an fljB mutant was as adherent and invasive as the wild-type strain.

IMPORTANCE In this work, we answered clearly the role of flagella in S. Kentucky attachment to the chicken skin and Caco-2 cells. We demonstrated that the presence of FliC is necessary for attachment to broiler skin. Expression of the alternative flagellar subunit FljB enables Salmonella motility, but this subunit is unable to mediate strong attachment. Deletion of the flgK gene prevents proper flagellar assembly, making Salmonella significantly less adherent to broiler skin than the wild type. S. Kentucky with deletions in all three structural genes, fliC, fljB, and flgK, as well as a flagellar motor mutant (motA), exhibited less adhesion and invasion of Caco-2 cells, while an fljB mutant was as adherent and invasive as the wild-type strain. We expect these results will contribute to the understanding of the mechanisms of Salmonella attachment to food products.

INTRODUCTION

Salmonella enterica serovar Kentucky has been recognized as the most prevalent Salmonella serotype on broilers in poultry-processing plants in the United States and some European countries, and some strains of this serovar contain the ciprofloxacin resistance gene (1). Interestingly, this serovar has rarely been reported in nontyphoidal salmonellosis cases in humans in the United States (2). However, according to recent studies, S. Kentucky exhibits the potential to emerge as the most prominent Salmonella serotype in human disease (1, 3).Therefore, this particular serovar is potentially an emerging risk for foodborne illness from a public health standpoint.

Salmonella colonizes on broiler surfaces and persists in all stages of chicken processing, regardless of the hygienic steps that are taken. Early bacterium-epithelial cell interaction on the broiler skin, and possibly the epithelial layer of the intestinal tract, is the primary route for contamination leading to possible pathogenesis.

Cell surface structures are a determining factor for bacterial attachment to surfaces (4). Specifically, nonflagellated bacteria rarely attach to broiler skin (5). Nonflagellated and flagellar motor mutants of Salmonella enterica serotype Enteritidis were less adherent to chick gut explant than the wild-type strain (6). However, conflicting data have suggested that motility has a negligible role in the bacterial attachment compared to bacterial density (7, 8). Furthermore, a nonflagellated mutant strain of Salmonella enterica serovar Typhimurium was able to attach to cultured intestinal epithelial cells but was impaired in its ability to invade the cells (9). Similarly, a nonflagellated mutant of S. Enteritidis was 50-fold less invasive to Caco-2 cell lines, even if the adhesion rate was similar to that of a flagellated strain (10). However, comparison of flagellar motor mutants and nonflagellated mutants of S. Enteritidis to the wild-type strain suggested that flagellin-mediated motility accelerates the invasion process rather than being essential (11).

Previous work in our laboratory (13) indicated that flagella might play a significant role in colonizing S. Kentucky on broiler skin. The focus of this study was to investigate the role of flagellar subunits and flagellar motor on attachment to broiler skin and invasion of Caco-2 cells. To this goal, several individual deletions in structural subunits and the flagellar motor gene were generated in a bioluminescent strain of S. Kentucky. The attachment of nonflagellated mutants, flagellar motor mutants, and mutants with only one of the flagellin subunits was compared with the wild-type strain in broiler skin attachment using bioluminescence. Understanding the molecular basis for Salmonella adhesion to broiler skin and Caco-2 cells may enable the development of new strategies to reduce contamination of poultry during processing.

RESULTS

Construction of S.KΔfliC, S.KΔfljB, S.KΔfliCΔfljB, S.KΔflgK, and S.KΔmotA mutants.

To determine the role of flagellar structural and motor genes in the attachment of Salmonella to broiler skin, mutants with one flagellin subunit (S.KΔfliC and S.KΔfljB), nonflagellated mutants (S.KΔfliCΔfljB and S.KΔflgK), and nonmotile strains (S.KΔmotA) were generated. PCR with chloramphenicol and site-specific primers for each mutant revealed that each mutant had the intended mutation, and the chloramphenicol gene replaced the target gene (Fig. 1 and 2). Upon elimination of the antibiotic resistance gene, PCR with site-specific primers produced the expected size product, confirming the deletion of the target gene. The chloramphenicol resistance gene-specific primers did not yield any product, which confirms the removal of the chloramphenicol gene. Sequencing of the amplified fragments also confirmed the deletion of the target genes. All of the deletions are considered nonpolar mutations because elimination of the antibiotic resistance gene created a scar containing a stop codon and a ribosomal binding site and a start codon for downstream gene expression (12).

FIG 1.

The locations of designed primers for junction verification of the mutants.

FIG 2.

PCR confirmation for the correct mutant junction. Bands from left to right: 1-kb plus ladder; flgK 1, upstream junction of ΔflgK with chloramphenicol and flgL; flgK 2, downstream junction of ΔflgK with chloramphenicol and flgJ; flgK 3, a fragment containing upstream and downstream of ΔflgK; fliC 1, upstream junction of ΔfliC with chloramphenicol and sekA-A1494; fliC 2, downstream junction of ΔfliC with chloramphenicol and sekA-A1497; fliC 3, a fragment containing upstream and downstream of ΔfliC; motA 1, upstream junction of ΔmotA with chloramphenicol and motB; motA 2, downstream junction of ΔmotA with chloramphenicol and flhC; motA 3, a fragment containing upstream and downstream of ΔmotA.

Motility.

The goal of the motility experiment was to determine the role of flagellar structural and motor genes in Salmonella motility. The test revealed that all the mutants, except for the S.KΔfliC and S.KΔfljB (one of the flagellin subunits was disrupted in each mutant) were nonmotile (Fig. 3). Complementation of nonmotile S.KΔfliCΔfljB, S.KΔflgK, and S.KΔmotA mutants using pBBR1MCS-4 containing the wild-type genes restored motility. In the case of the S.KΔfliCΔfljB mutant, expression of one of the flagellar subunits (fliC gene) was sufficient to restore motility. The S.KΔmotA mutant was complemented with a parental copy of both the motA and motB genes.

FIG 3.

Motility test confirmed S.KΔfliC and S.KΔfljB mutants are still motile.

SEM.

The goal of the scanning electron microscopy (SEM) experiment was to determine the presence of flagella in Salmonella mutants. Flagella were not detectable in S.KΔfliCΔfljB and S.KΔflgK mutants (Fig. 4D and F). Similarly, flagella were not detected in the S.KΔflgK-t, S.KΔflgA-t, S.KΔflgC-t, S.KΔflhB-t, and S.KΔflgJ-t mutants (Fig. 4G to K), which are the transposon insertion mutants constructed by our group previously (Table 1) (13). However, flagella were detected on the S.KΔmotA mutant (Fig. 4E) and the S.KΔfliC and S.KΔfljB mutants (Fig. 4B and C).

FIG 4.

SEM images of the mutants made with lambda Red system and transposon insertion. (A) S. Kentucky wild-type (wild type) (S.KTn7lux). (B) S.KΔfliC. (C) S.KΔfljB. (D) S.KΔfliCΔfljB. (E) S.KΔmotA. (F) S.KΔflgK. (G) S.KΔflgK-t. (H) S.KΔflgA-t. (I) S.KΔflgC-t. (J) S.KΔflhB-t. (K) S.KΔflgJ-t.

TABLE 1.

Bacterial strains and plasmids used in this study

Bacterial strain or plasmid	Characteristic(s) or genotypea	Reference or source
Strains
S. enterica serovar Kentucky (S.KTn7lux)	Wild type::luxCDAB	26
E. coli DH5α	Contains lux gene	26
S.KΔfliC	Mutation in alternative flagellin subunit, Aps, Cms	This study
S.KΔfljB	Mutation in alternative flagellin subunit, Aps, Cms	This study
S.KΔmotA	Mutation in motion gene, Aps, Gms	This study
S.KΔflgK	Mutation in hook-associated protein, Aps, Gms	This study
S.KΔfliCΔfljB	Double mutation in both flagellin subunits, Aps, Gms	This study
S.KΔflhB-t	Transposon mutation in flagellar export apparatus, Gmr	13
S.KΔflgA-t	Transposon mutation in basal body p-ring protein, Gmr	13
S.KΔflgK-t	Transposon mutation in hook-associated protein, Gmr	13
S.KΔflgJ-t	Transposon mutation in rod assembly protein, Gmr	13
S.KΔflgC-t	Transposon mutation in basal body rod protein, Gmr	13
S.KΔmotAC	Complemented strain of S.KΔmotA mutant, Apr	This study
S.KΔflgKC	Complemented strain of S.KΔflgK mutant, Apr	This study
S.KΔfliCΔfljBC	Complemented strain of S.KΔfliCΔfljB mutant, Apr	This study
E. coli K-12 (BW25141/pKD3)	lacIq rrnBT14 ΔlacZWJ16 ΔphoBR580 hsdR514 ΔaraBADAH33 ΔrhaBADLD78 galU95 endABT333 uidA(ΔMluI) pir+ recA1	12
E. coli K-12 (BW25141/pKD4)	lacIq rrnBT14 ΔlacZWJ16 ΔphoBR580 hsdR514 ΔaraBADAH33 ΔrhaBADLD78 galU95 endABT333 uidA(ΔMluI) pir+ recA1	12
E. coli DH5α/pCP20	Δ(argF-lac)169 ϕ80dlacZ58 glnV44 λ−rfbC1 gyrA96 recA1 spoT1 thi-1 hsdR17	12
E. coli K-12 (BW25113)/pKD46	lacIq rrnBT14 ΔlacZWJ16 hsdR514 ΔaraBADAH33 ΔrhaBADLD78	12
Plasmids
pCP20	Ts-rep [c857] (lambda)ts bla(apr) cat (FLP)	12
pKD46	repA101(ts) araBp-gam-bet-exo oriR101 bla(apr)	12
pKD3	oriR6Kγ bla(apr) rgnB(Ter) cat	12
pKD4	oriR6Kγ bla(apr) rgnB(Ter) kan	12
pBEN276	mTn7::MC apr luxCDABE	26
pBBR1MCS-4	Expression vector; apr	28
pBBR1MCS-4 flgK	pBBR1MCS-4 expressing wild-type flgK	This study
pBBR1MCS-4 fliC	pBBR1MCS-4 expressing wild-type fliC	This study
pBBR1MCS-4 motA	pBBR1MCS-4 expressing wild-type motA	This study

^aAp^s, ampicillin susceptibility; Cm^s, chloramphenicol susceptibility; Gm^s, gentamicin susceptibility; Gm^r, gentamicin resistance; Ap^r, ampicillin resistance.

Attachment to chicken skin.

The goal of attachment experiment was to determine the role of flagellar structural and motor genes in Salmonella attachment. All bioluminescent S. Kentucky flagellar mutants, except S.KΔfljB, showed significantly reduced adhesion to broiler skin compared to that of the wild-type S. Kentucky (S.KTn7lux) (P < 0.05) (Fig. 5). The data indicate clearly that the flagellar subunit FliC, the flagellar hook protein FlgK, and flagellar motor protein MotA contributed to chicken skin attachment, while the flagella subunit FliB did not have a role in skin attachment. Transposon mutants with transposon insertions in genes encoding hook-associated protein, basal body p-ring, flagellar export apparatus, rod assembly protein, and in the basal body rod protein (13) also had significantly reduced broiler skin attachment (P < 0.001). In summary, these results confirm the essential role of S. Kentucky flagella in attachment to the chicken skin.

FIG 5.

Average bioluminescence (photons/s/cm²/steradian) of each mutant on the broiler skin measured by IVIS imaging system 100 series and quantified by the Living Image software version 2.5. A circular region of interest was drawn around each well on a 96-well plate, and bioluminescence values were calculated and used to derive the average bioluminescence in each group. P values of <0.05 and <0.001 were used for the deletion and transposon mutants, respectively. N.C., negative control.

Complementation of the motA gene in the S.KΔmotA mutant restored skin attachment (Fig. 6). No difference was observed between the S.KΔmotAC mutant and the parent strain (P > 0.05). Additionally, complementation of fliC in the S.KΔfliCΔfljB mutant and complementation of flgK in the S.KΔflgK mutant restored attachment to broiler skin in these mutants to levels similar to those of the wild-type strain (P > 0.05).

FIG 6.

Average bioluminescence (photons/s/cm²/steradian) of mutants and their complemented strains on the broiler skin measured by IVIS imaging system 100 series and quantified by the Living Image software version 2.5. A circular region of interest was drawn around each well on a 96-well plate, and bioluminescence values were calculated and used to derive the average bioluminescence in each group. A P value of <0.05 was used. N.C., negative control; P.C., positive control.

Adhesion and invasion of Caco-2 cells.

The goal of the adhesion and invasion experiment was to determine the role of flagellar structural and motor genes in Salmonella attachment to Caco-2 cells. The S. Kentucky mutant with deletions in both flagellin subunits (S.KΔfliCΔfljB) was less adherent and invasive in Caco-2 cells than the wild-type strain (P < 0.001). However, the deletion of each subunit had a different outcome (Fig. 7a). Deletion of the fliC gene reduced cell adhesion and invasion significantly (P < 0.001), but the S.KΔfljB mutant was as adherent and invasive as the parent strain. Mutation of the flgK gene encoding hook protein also significantly decreased adherence and invasion (P < 0.001). The flagellar motor mutant (S.KmotA) also had a reduction in cell adhesion and invasion (P < 0.001).

FIG 7.

Mean percentage invasion (a) and adhesion (b) of the mutants and their complement strains. P values of <0.001 and <0.05 were used for panels a and b, respectively. W.T., wild type.

The S.KΔfliCΔfljB, S.KΔflgK, and S.KΔmotA mutants were complemented with wild-type fliC, flgK, and motA and motB genes, respectively. Complementation restored invasion into intestinal epithelial cells compared to the wild type (S.KTn7lux). The expression of fliC in the S.KΔfliCΔfljB mutant, flgK in the S.KΔflgK mutant, and motA and motB in the S.KΔmotA mutant made these mutants as invasive as the parental strain (P > 0.05) (Fig. 7a). Complementation also restored adherence to Caco-2 cells in comparison to the wild type (P > 0.05) (Fig. 7b).

DISCUSSION

Attachment to broiler surfaces helps pathogenic bacteria survive broiler production and processing procedures (14). It has been suggested that bacteria attached to broiler skin are less susceptible to detergents and other antibacterial treatments (15), as well as to oxidative stress (16). Sessile and planktonic bacteria have some physical changes in protein expression; in particular, increased expression of stress genes in sessile bacteria influences the persistence of attached bacteria (17).

However, the mechanism of bacterial attachment to broiler skin is not well understood. There have been several conflicting reports on bacterial colonization on broiler surfaces, and different bacterial properties have been suggested to contribute to colonization. In our previous research, we concluded that attachment is a multifactorial process (13). According to our results, several bacterial structures and metabolites, such as flagella, lipopolysaccharide core biosynthesis protein, amino acid metabolites, as well as different secretion and signaling systems, contribute to the bacterial attachment on broiler skin. Of these factors, flagella demonstrated the highest potential for broiler skin colonization, but it was not clear whether flagellar structure or rotation is more important in mediating attachment (13).

Accordingly, several in-frame deletions were constructed in the current study, in particular, flagellar subunits and motor genes to determine their relative roles in the attachment. Due to the high frequency of S. Kentucky contamination in broiler processing plants, this serovar was selected as the representative serotype for these experiments. The high prevalence of S. Kentucky can be correlated with its ability to attach to broiler surfaces (18). Our current study reveals that the expression of flagellar subunit FliC and an active flagellar motor are necessary for successful broiler skin colonization of S. Kentucky.

Salmonella flagella consist of two antigenically distinct proteins, FliC and FljB. Expression of these flagellin subunits is controlled by phase variation (19). Our study shows that deletion of the fliC gene does not eliminate motility or expression of flagella, but the mutant has significantly reduced attachment. Scanning electron microscopy images and the positive motility test of the ΔfliC mutant confirmed the existence of short active flagella. However, mutants expressing fliC with an fljB deletion were as adherent as the wild type. Furthermore, the presence of a defective flagellar motor significantly reduced S. Kentucky colonization of broiler skin, which indicates that FliC expression is not enough for attachment; active motility is also required. This explains our previous findings that nonflagellated transposon mutants are defective in chicken skin attachment.

Although bacterial attachment to the chicken skin is not long enough to be considered a biofilm formation process, it is comparable to the early stage of biofilm formation and bacterial attachment to surfaces. In several biofilm studies, the role of flagella in bacterial attachment has been investigated. In a comparison of the ability of several different Salmonella serovar strains to form biofilm in poultry-processing plants, it was concluded that the attachment of bacteria to different surfaces is strain dependent, and bacterial surface components, such as cell wall proteins, flagella, and lipopolysaccharide, all contribute to biofilm formation (20). Furthermore, transposon mutagenesis of S. Typhimurium in flagellar hook-associated protein and lipopolysaccharide synthesis severely altered attachment to meat and poultry and biofilm formation on different surfaces (21). In a more recent study with S. Typhimurium motA, fliC, fljB, and fliC fljB mutants, it was concluded that flagellar motility does not affect biofilm formation on cholesterol surfaces, but fliC expression does (22).

Our results with Caco-2 cells confirmed our findings on attachment to the chicken skin. In both adhesion and invasion assays, flagellar ΔfliCΔfljB, ΔflgK, and ΔmotA mutants were less adherent and invasive than the wild-type strain. The deletion of individual flagellar subunits resulted in two motile mutants with different adhesion characteristics. The fliC deletion caused significantly reduced adherence and invasion; however, the ΔfljB mutant was as adherent and invasive as the wild type. There are some conflicting data on the role of flagella in cellular adhesion and invasion in other Salmonella serotypes. An ΔfliC mutant of S. Enteritidis had decreased adhesion to human (HEp-2) and avian (Div-1) cell lines with microscopic methods. However, biological counting did not show a difference in adhesion between mutants and the wild type. Invasion and membrane ruffling in both cell lines were reduced in flagellar mutants (23). Adherence and invasion of S. enterica serovar Typhimurium phase-locked mutants (ΔfliC and ΔfljB) were not affected in murine intestinal epithelial cell lines, but the ΔfliC mutant was more efficient in virulence (19). Nonflagellated fliC fljB mutants of S. enterica serovar Typhimurium did not exhibit a difference in pathogenicity but were less invasive to Caco-2 cells (9).

The majority of Salmonella studies on cell adhesion and invasion were more focused on the comparison of nonflagellated mutants and flagellated nonmotile mutants without comparing the flagellar subunits separately. Mutation of the fliC gene together with motA mutation in S. Enteritidis impaired bacterial invasion severely but not adhesion (11). A comparison of several naturally occurring motA mutants of S. Enteritidis with nonflagellated mutant strains (fliC mutants) determined that they exhibit less invasiveness to Caco-2 cells (24). It was concluded that a strain with paralyzed flagella induced a higher proinflammatory response than a nonflagellated strain, suggesting that presence of paralyzed flagella is more effective than the absence of flagella (24).

It is important to note that deletion of the flgK gene, which encodes a hook-associated protein, results in a mutant unable to assemble flagella and that is nonmotile, but the strain can still secrete flagellin subunits. In the current study, an flgK mutant did not show a difference in invasiveness compared to an fliC fljB mutant that does not produce flagella. Both of these mutants were less invasive than the wild type. Thus, proper assembly of flagella in S. Kentucky is required for attachment and invasion, not just the production of flagellin. These results are consistent with findings of previous studies on several poultry-associated isolates of S. enterica with mutations in flgK, flgL, and fljB genes, which exhibited invasion reduction in Caco-2 cell line compared to the parental strain (25).

In conclusion, the expression of flagellin subunit fliC, flagellar assembly, and a functional flagellar motor are required for optimal attachment of S. Kentucky on chicken skin and for cellular attachment and invasion. Future studies on the interaction of Salmonella flagella and host epithelial cells may allow the development of improved methods to reduce contamination of chicken during processing.

MATERIALS AND METHODS

Bacteria, plasmids, and media.

Table 1 lists the bacterial strains and plasmids that were used in the current study. Bioluminescent S. Kentucky (S.KTn7lux) was constructed in our laboratory by inserting the bacterial lux operon in the S. Kentucky chromosome using plasmid pBEN276 (26). In addition to the flagellar deletion mutants constructed in the current study, some flagellar gene mutants constructed by transposon mutagenesis from a previous study were included in some assays. Escherichia coli DH5α carrying the lux operon was used as a negative control for the chicken skin attachment assay. Plasmids pCP20, pKD46, pKD3, and pKD4 were obtained from Yale University and used to generate gene deletions (12). Transformants were selected on LB agar containing ampicillin and chloramphenicol (100 μg/ml and 10 μg/ml, respectively).

PCR and primers.

To delete the fliC, flgK, motA, and fljB genes, 50- to 70-nucleotide (nt) primers were designed. Each primer contained a 30- to 50-nt sequence of homology extension that was complementary to the flanking sequence of the genes of interest and a 20-nt sequence that was complementary to template plasmid pKD3. The primer sequences are listed in Table 2. Using an Applied Biosystems 2720 thermal cycler, 50-μl PCRs were conducted with the following cycling conditions: initial denaturation for 5 min at 94°C, 30 cycles of denaturation for 30 s at 94°C, annealing for 30 s at 56°C, and elongation for 1 min at 72°C, and a final extension for 10 min at 72°C.

TABLE 2.

List of primers used to generate and verify the deletions and complementation of the mutants

Primer	Sequence (5′–3′)a
flhC	TGTGGATGCGGTGATTAAAG
motB	GCGCATATACGGCTCAACTT
sekA_A1494	CAGGCAAGACTCAGGGAGTT
sekA_A1497	TGGACGCCACGGTAGACTTA
flgL	TGCTGGGTAACACTTTTCTCG
flgJ	ACTGGCCAGCGAACAAAG
sekA_A2274	TACCGTATGCGTTATTCAGCA
sekA_A2276	AAGGGTTCGCTCGACAATTA
c1	TTATACGCAAGGCGACAAGG
c2	GATCTTCCGTCACAGGTAGG
motAH1P1F2	TGTCATAGTCAACAGCGGAAGGATGATGTCGTGCTTATCTTATTAGGTTAGTGTAGGCTGGAGCTGCTTC
motAH2P2 R2	TTACGACGACAATGGGATGAGCCTGATTTTTCATGCTTCCTCAGTCGTCTCATATGAATATCCTCCTTA
fliCH2p2	ATTGTGTACCACGTGTCGGTGAATCAATCGCCGGACATATGAATATCCTCCTTA
fliCH1p1	AATAACATCAAGTTGTAATTGATAAGGAAAAGATCGTGTAGGCTGGAGCTGCTTC
flgKH2p2	CATCTGGGTACTGATACGCATGTCATCCTTCTCCTCATATGAATATCCTCCTTA
flgKH1p1	TAACAACGAGTATTGAAGGATTAAAAGGAACCATCGTGTAGGCTGGAGCTGCTTC
fljBH1p1	TTGCTTTATCAAAAACCTTCCAAAAGGAAAATTTTGTGTAGGCTGGAGCTGCTTC
fljBH2p2	CCCCGGATTCACGGGGCTGAATAAAACAAAATAAACATATGAATATCCTCCTTA
flgKFc	AAGAGCTCGCGAATCTCGACAATCT
flgKRc	AACCCGGGATACGCATGTCATCCTT
fliCFc	AAGAGCTCCCTTGATTGTGTACCAC
fliCRc	AACCCGGGGAAATTCAGGTGCCGA
motAFc	AAGAGCTCCTCACGCTATCACCTCG
motBRc	AACCCGGGGAAACGGTGTGGACAA

^aItalic letters represent SacI restriction site added to 5′ end of primer sequence. Bold letters represent SmaI restriction site added to the 5′ end of the primer sequence. AA nucleotides were added to the end of each restriction site to increase the enzyme efficiency. Underlined letters stand for homology extensions added to the 3′ end of the primer sequence.

Deletion of flagellin genes.

Deletion of flagellin genes (fliC and fljB), hook-associated protein gene (flgK), and flagellar motor gene (motA) of S.KTn7lux was achieved using the λ Red recombinase system (12) and the PCR products obtained with the primer sets listed in Table 3. Briefly, S.KTn7lux competent bacteria were transformed with λ Red helper plasmid (pKD46), an ampicillin-resistant and temperature-sensitive plasmid that contains an arabinose-induced λ Red recombinase system. Competent bacteria were made by washing cells with 10% ice-cold glycerol water. Electroporation, using the PCR products described above, was performed with a 1-mm cuvette under the parameters 1.8 kV, 25 μF, and 400 Ω. Bacteria were recovered in super optimal broth with catabolite repression (SOC) medium at 30°C for 1 h at 200 rpm. Transformants were grown on LB agar with ampicillin for 24 h at 30°C.

TABLE 3.

Primers that have been applied for junction confirmation of constructed mutants

Gene	Upstream chloramphenicol-gene junction	Downstream chloramphenicol-gene junction	Upstream-downstream genes
flgK (A)	C2, flgL (A1)	A2: C1, flgJ (A2)	flgL, flgJ (A3)
fliC (B)	C2, sekA-A1494 (B1)	C1, sekA-A1497 (B2)	sekA-A1494, sekA-A1497 (B3)
motA (C)	C2, motB (C1)	C1, flhC (C2)	motB, flhC (C3)
fljB (D)	C2, sekA-A2274 (D1)	C1, sekA-A2276 (D2)	sekA-A2274, sekA-A2276 (D3)

Bacteria carrying pKD46 were grown overnight in LB broth with ampicillin at 30°C, and 1 ml was transferred to 100 ml of LB broth with ampicillin and l-arabinose (10 mM) at 30°C for 3 to 4 h to optical density at 600 nm (OD₆₀₀) of 0.6. Bacteria were made competent and transformed with PCR amplicons (described above). Each amplicon consisted of the chloramphenicol resistance gene (cm^r) flanked by flippase recognition target (FRT) sites, which were amplified from pKD3 using primers with homology to the 5′ and 3′ ends of the target gene. PCR products were digested with 1 μl of DpnI (Promega, Madison, WI) added to a 50-μl PCR product at 37°C for 1 h and gel purified using a QIAquick gel extraction kit (Qiagen, Valencia, CA). In preparation for electroporation, 40 μl of competent bacteria was mixed with 2 to 5 μl of PCR amplicon (final concentration, 20 to 50 ng/μl) and transferred to a 1-mm electroporation cuvette. Transformed bacteria were recovered in SOC medium at 30°C for 3 to 4 h at 200 rpm. A portion of the bacterial suspension (100 to 200 μl) was spread on LB agar with chloramphenicol and incubated at 37°C to select successful transformants. After 24 h, colonies were picked and subcultured at 42°C to cure plasmid pKD46.

To verify gene deletions, primers were designed approximately 400 to 500 bp upstream and downstream of the gene of interest (Table 2). Two primers for the chloramphenicol resistance gene (c1 and c2) were used in combination with site-specific primers in PCRs to verify that new junctions were created after recombination. PCR products were sequenced for confirmation, and the primer sets used to obtain the PCR products are listed in Table 3. To remove the antibiotic resistance cassette, mutants were transformed with pCP20 by electroporation. Ampicillin-resistant (Ap^r) colonies were selected after 48 h and passed on LB agar at 43°C. Colonies were tested for loss of both chloramphenicol and ampicillin resistance, and loss of the Cm^r gene was confirmed by PCR.

Construction of complement strains.

Primers were designed 50 bp upstream and downstream of the flgK, fliC, and motA genes to amplify the whole genes with their promoters. SmaI and SacI restriction site sequences were incorporated to facilitate cloning. Genomic DNA was isolated from S.KTn7lux (wild type) using Wizard genomic DNA kit (Promega, Madison, WI). Genes were amplified from S. Kentucky genomic DNA in a 50-μl PCR mixture containing 0.2 mM dinucleoside triphosphate (dNTP) mixture, 1.5 mM MgCl₂, 0.2 mM primers, and 1.25 U of Taq DNA polymerase (Promega). In the case of motA, a long fragment consisting of both the motA and motB genes was amplified because motA and motB overlap; therefore, a mutation in motA could interfere with motB expression. The amplified fragment lengths for motA-motB, fliC, and flgK were 1,813 bp, 1,487 bp, and 1,661 bp, respectively. Amplicons were purified using the PCR Clean-Up kit (Promega, Madison, WI), digested with SmaI and SacI (Promega), gel purified using a QIAquick gel extraction kit (Qiagen, Valencia, CA), and ligated into SacI and SmaI-digested pBBR1MCS-4 with T4 DNA ligase (Promega). Plasmids were sequenced for confirmation and transformed into each corresponding mutant.

Motility.

Motility was determined by stabbing motility agar with a bacterial strain and assessing migration from the stab after incubation at 37°C for 18 h. Motility agar consisted of 10 g of tryptone and 5 g of NaCl per liter plus 0.35% (wt/vol) agar (pH 7.4) (9).

SEM.

After the motility tests, the presence or absence of flagella was determined by scanning electron microscopy (SEM). Briefly, poly-l-lysine-coated glass coverslips were placed in 6-well cell culture plates containing LB broth. Each well was inoculated with a mutant strain, and plates were incubated at 37°C with shaking at 250 rpm overnight. Coverslips were removed from each well, air-dried, and prepared according to Merritt et al. (27), and several random frames were viewed with a field emission scanning electron microscope (JEOL JSM-6500F).

Chicken skin attachment.

Attachment of all the flagellar mutants and the complement strains (Table 1) to chicken skin were compared. Clear-bottomed 96-well black cell culture plates containing 100 μl of LB broth were inoculated with 5 μl of flagellar mutant, their complemented strains, and transposon-inserted flagellar mutants (8 wells per strain). Each 96-well plate contained eight replicates of positive and negative controls (wild-type bioluminescent S. Kentucky and bioluminescent E. coli DH5α). Each plate was covered with Breathe-Easy film (Diversified Biotech, Boston, MA) and incubated overnight at 37°C in a shaker-incubator at 250 rpm. The second 96-well plate was prepared by inoculating fresh LB broth (100 μl/well) with the overnight cultures and incubating for 2 h to reach the log phase. The optical densities of the cultures were measured and equalized (OD₆₀₀, 0.35). Fresh chicken carcasses were obtained from a poultry plant, and the skin tissue was removed. Uniform circular skin sections were made using a 6-mm skin biopsy punch and placed in each well of the log-phase-culture plate. The plates were incubated at room temperature (25°C) for 1 h to allow bacterial attachment. Following incubation, bacterial suspensions were removed, and wells were washed with 200 μl of distilled water by pipetting twice to remove unattached bacteria.

Bioluminescent imaging of attached bioluminescent bacteria was conducted according to the procedures in our earlier work (26). Briefly, bioluminescence was recorded for 15 s at 37°C using an IVIS imaging system 100 series. To determine the effect of washing on attachment properties, wells were filled with 200 μl of sterilized water and shaken on a plate shaker at 700 rpm for 1 h. After removal of the water, bioluminescence on skin sections was measured as described above. The mean bioluminescence for each treatment following the washing step was compared to positive and negative controls to determine chicken skin attachment for each strain. Mutants were considered deficient in attachment if their attachment percentage fell out of the lower 95% confidence limit calculated from attachment rate of wild-type strain S.KTn7lux replicates.

Adhesion and invasion assay.

We did a search to get the poultry epithelial cell lines but were not successful, so we had to do the mammalian epithelial cells instead. Caco-2 cells (ATCC HTB-37) were maintained in Eagle's modified essential medium with l-glutamine (EMEM, ATCC 30-2003) supplemented with 10% fetal bovine serum in 200-ml cell culture flasks. Cells were incubated at 37°C with 5% CO₂.

Adhesion and invasion assays were performed as described previously (29). Briefly, a confluent layer of Caco-2 cells was prepared by seeding 5 × 10⁵ cells per well in 24-well cell culture plates. For the adhesion assay, an overnight culture of each Salmonella strain was adjusted to an OD₆₀₀ of 1.0 (approximately 10⁷ CFU/ml), and sufficient bacteria were added to yield a multiplicity of infection (MOI) of 100:1. Cells were washed twice with phosphate-buffered saline (PBS) and once with fresh prewarmed medium before inoculation. Plates were centrifuged for 1 min at 800 rpm and incubated at 37°C in 5% CO₂ for 45 min to allow for adhesion. Inoculums were removed, and plates were washed with PBS five times to remove unattached bacteria. Cells were then lysed with cold 0.1% Triton X-100, and bacterial numbers were determined by serial dilution and plate counts on LB agar. The results were calculated as percent adherence ([no. of bacteria after wash/no. of inoculated bacteria] × 100).

For the invasion assay, plates were incubated for 1 h to allow invasion after adding the bacteria to the cells. The medium was replaced with fresh medium containing 100 μg/ml gentamicin. Following 2 h of incubation, the suspension was aspirated and plated to confirm the removal of all extracellular bacteria. Cells were washed twice with PBS and lysed with 0.1% cold Triton X-100, and bacteria were quantified by serial dilution and plate counts. Percent invasion was calculated for each treatment ([no. of intracellular bacteria/no. of inoculated bacteria] × 100). Four replicates were conducted for each treatment on a plate (24).

Statistical analysis.

Adhesion of mutants to broiler skin was analyzed with a completely randomized design in eight replications. The mean value of bioluminescence (P/S) for each strain was compared with the analysis of variance (ANOVA) (SPSS 21.0 software program) and Tukey's multiple comparison posttest to investigate the difference in adhesion of mutants. A P value of <0.05 was considered statistically significant.

The attachment and invasion assay with broiler skin was performed in two sets in quadruplicate. The percentage of intracellular bacteria was used in the ANOVA (SPSS 21.0 software program) with Tukey's multiple comparisons posttest to analyze the difference in the invasion for mutants. The attachment assay was performed on two different days, and a generalized randomized block design with four replicates was used for comparison of the percentage of associated bacterial count between the mutants and wild-type strain. A P value of ≤0.05 was considered statistically significant.