Background

Campylobacter jejuni is the most common cause of foodborne gastroenteritis in humans worldwide, and infections caused by this microbe often leads to diarrhea and sometimes to a severe inflammatory response with clinical symptoms of fever, abdominal cramping, bloody stools and other symptoms [1, 2]. C. jejuni is prevalent in the environment and can infect humans in a variety of ways [3, 4]. Although the incidence of C. jejuni infection is very high at present, causing it to receive increased attention [5], a complete understanding of the mechanisms of C. jejuni-associated disease is still an ongoing effort [3]. An analysis of the genome sequence of C. jejuni revealed that the cytolethal distending toxin (CDT) is the only toxin produced by this bacterium. Many other classical virulence factors that are common to other pathogens are missing in the C. jejuni genome [6]. It has now been shown that colonization in the host intestinal tract and persistence to a sufficient load is crucial for C. jejuni pathogenesis [3, 7, 8].

Intestinal colonization by C. jejuni is influenced by multiple factors, including the capsule, glycosylation system, catabolism of L-serine, cytochrome c peroxidase, transport systems, putative adhesins, chemotaxis and flagellar motility [9,10,11,12,13,14,15,16]. The flagellum is particularly important among these factors [17]. C. jejuni produces a single flagellum at one or both poles, which not only provides chemotactic motility for migration to replicative niches but also helps the bacteria penetrate the mucus that covers the epithelial cells. In addition, the flagellum is involved in interactions with and invasion of the host epithelium, the secretion of invasion antigens, and the evasion of the innate immune system, is the essential virulence and colonization factor of C. jejuni [6, 15, 18,19,20]. The flagellum is a multicomponent organelle composed of a basal body, motor, rod, hook and flagellin filament. It appears that the C. jejuni flagellar substructures, flagellin glycosylation, motility and secreted invasion antigens are all required for host colonization but are not the only determinants [9, 16, 19, 21,22,23]. There are over 50 flagellum-related genes in the C. jejuni genome, and their corresponding contributions to colonization are also diverse [15, 16, 18, 22,23,24,25]. Considering the multifunctional role played by flagella during C. jejuni pathogenesis [26], a comprehensive understanding of their virulence potential is urgently needed.

The importance of FlhF for flagellar biosynthesis has been observed for several pathogens (primarily for the polarly flagellated bacteria) in recent years [27,28,29,30,31,32,33]. In C. jejuni, the inactivation of flhF has been observed to inhibit the synthesis of flagella and motility in previous studies [34, 35], but its contribution to C. jejuni colonization has not yet been characterized. In this study, the influence of flhF on C. jejuni colonization was systematically evaluated both in the reference strain and in an isolate. The results showed the significant effect this gene has in the host colonization process. The possible factors that may contribute to the colonization defect were assayed, including the flagella-associated abilities such as adherence and invasion to the host cells, autoagglutination (AAG) and biofilm formation. The mechanism of by which FlhF influences host colonization was also explored at the gene expression level. We observed that flhF is continuously upregulated during the cell infection process, and in addition to its major role in the flagellar system, some other infection-related genes were observed to be downregulated in the flhF mutant strain. The results of this study provide a reference for the further analysis of this virulence factor and investigations of the mechanism of flagella mediated co-regulation of pathogenesis in this bacterium.

Methods

Bacteria and cell culture

C. jejuni strains were grown on Campylobacter blood-free selective agar containing charcoal cefoperazone deoxycholate (CCDA) (Oxoid, Basingstoke, UK) or in Mueller-Hinton (MH) broth (BD Biosciences, Sparks, MD, USA) in a microaerophilic environment (5% O2, 10% CO2, and 85% N2) and incubated at 42 °C. The C. jejuni isolate used in this study was isolated from the anal swabs of a monkey with diarrhea and was further identified using multiplex PCR (polymerase chain reaction) (Additional file 1: Figure S1) [36]. Appropriate antibiotics (50 μg/ml kanamycin, 20 μg/ml chloramphenicol, 10 μg/ml rifampicin, 6 μg/ml polymyxin B, 10 μg/ml trimethoprim, 100 μg/ml cycloheximide, 40 μg/ml vancomycin, 20 μg/ml amphotericin B or 60 μg/ml cefoperazone) were added to the medium as required. Inactivation of flhF and complementation of the mutant in C. jejuni NCTC 11168 and in the isolate were performed as previously described [34]. Human colon cancer cell line (Caco-2) was acquired from American Type Cell Culture (ATCC) and routinely cultured in Dulbecco’s minimal essential medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and maintained in a humidified 37 °C atmosphere containing 5% CO2.

Colonization assay

Chick and mice colonization assays were largely performed as described in previous studies [16, 24, 37]. Briefly, C. jejuni strains were harvested in phosphate buffered saline (PBS), and 1 × 108 colony-forming units (cfu) of bacteria were subsequently orally inoculated into 1-day-old Campylobacter-free Leghorn chicks obtained from the Jiangsu Institute of Poultry Science. Chicks were randomly allocated into three groups and sacrificed by CO2 asphyxiation at 1 and 7 days post inoculation. For each chick, the intestinal tract was collected, separated following gentle extrusion of the lumenal contents, weighed, homogenized, serially diluted and plated on selective CCDA agar (containing rifampicin, polymyxin B, trimethoprim, cycloheximide, vancomycin, amphotericin B and cefoperazone) to count the bacteria load. Six-week-old C57/BL6 mice were acquired from the Comparative Medicine Center of Yangzhou University. The mice initially received 200 μl of a sodium bicarbonate solution (50 mg/ml), after which 1 × 108 cfu of bacteria were used to infect mice by orogastric inoculation 15–30 min later. The mice were sacrificed by cervical dislocation at the designated time point, and the colonization levels were evaluated using selective media containing amphotericin B and cycloheximide as described above.

Growth and biochemical tests

Overnight cultures of C. jejuni were harvested and diluted in MH broth to a starting optical density at 600 nm (OD600) of 0.08. Two-milliliter aliquots of the suspension were transferred into the wells of a 12-well plate and incubated at 42 °C and 120 rpm under a microaerophilic atmosphere. Samples were collected at indicated intervals and the OD600 values were determined to compare the growth of the different strains. Biochemical testing was performed as described by Valenza et al. [38]. Bacteria were harvested in a 0.45% NaCl solution and the turbidities were adjusted to an equivalent of a McFarland standard of 3. The suspensions were inoculated into a Neisseria-Haemophilus (NH) identification card, which was automatically analyzed by a VITEK 2 microbiology system (bioMérieux, France).

Cell adhesion and invasion assay

Adhesion and invasion assays were performed as described previously [39]. Caco-2 cells were first seeded into a 24-well plate at a density of 1 × 105. After the cells incubated for 18 h, C. jejuni that had grown for 20 h were harvested and resuspended in DMEM, and 1 ml suspensions containing 1 × 107 cfu were then inoculated into each well. The plates were centrifuged at 600×g for 5 min to enhance the contact between bacteria and eukaryotic cells. For bacterial adhesion assays, cells were incubated for 2 h, washed three times with PBS to remove the unattached bacteria, and then were lysed in 1 ml of PBS containing 0.1% TritonX-100, with serial dilutions of the suspensions subsequently spread onto CCDA plates to quantify the number of bacteria. To assess the number of invaded bacteria, the inoculated cells were first incubated for 2 h, washed once with DMEM and then were incubated in DMEM containing gentamicin (100 μg/ml) for another 1 h. The extracellular bacteria were killed during this period, and the number of invaded bacteria was determined as described above.

Autoagglutination assay

The autoagglutination assay was performed as previously described [23]. C. jejuni strains were grown for 20 h prior to use, after which each strain was suspended in PBS and adjusted to an OD600 of 1.0. Five-milliliter aliquots of the suspension were transferred into a series of sterile glass tubes (13 × 100 mm), which were incubated under microaerophilic conditions for a predetermined time. After incubation, 200 μl of the bacterial suspension was carefully removed from the top of each tube, and the OD600 was measured to evaluate the degree of AAG. Bacterial cells that strongly agglutinate will fall to the bottom of the tube, leading to a decreased OD600 value.

Biofilm-formation assay

The ability of C. jejuni strains to form biofilms was measured by crystal violet staining, using a method described by Naito et al. with some modifications [40]. Briefly, bacteria cultured for 20 h were harvested and diluted in MH broth to an OD600 of 0.1, after which 200 μl of each suspension was inoculated into 96-well polypropylene plates and incubated at 42 °C under microaerobic conditions. After incubating for 48, 60, 72, 84 and 90 h, the plates were washed with PBS and dried at 60 °C for 30 min. Next, 200 μl of a 1% crystal violet solution was added to each well to stain the biofilms that had formed. After staining at room temperature for 30 min, the plates were rinsed thoroughly with PBS three times to remove the unbound crystal violet. The plates were subsequently dried at 60 °C for 15 min, and the biofilms were quantified by determining the OD570 values after the remaining crystal violet was solubilized in a solution composed of 30% methanol and 10% acetic acid.

RNA isolation and quantitative real-time PCR

Quantitative real-time PCR (qRT-PCR) was performed as described previously [34]. C. jejuni was cultured for designated time, after which the RNA was extracted using an RNeasy plus mini kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. The isolated RNA was then used for cDNA synthesis using an RT reagent Kit (TaKaRa, Dalian, China). qRT-PCR was performed with a ABI PRISM 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) using the FastStart Universal SYBR Green Master (ROX) (Roche Diagnostics, GmbH, Germany). The primers used for qRT-PCR are listed in the Additional file 1: Table S1. The cj0402 gene (glyA) was used as an endogenous control to evaluate the relative expression of the target genes.

Gene collection and microarray analysis

The genes required for C. jejuni adhesion/invasion and colonization that were not directly involved in flagellar assembly were identified and are listed in Additional file 1: Table S2, and this information was primarily obtained from previously published reports. The relative mRNA expression of the selected genes in the C. jejuni flhF mutant strain versus the wild-type strain were acquired and analyzed using our previous whole transcriptome microarray data [34], which has been submitted to the NCBI Gene Expression Omnibus (GEO) database with the accession number GSE 87852. The fold change in gene expression was graphically presented in a heatmap that was generated using the Hemi software, with different colors representing the degree of variation in expression [41].

Statistical analysis

The data are presented as the means ± standard deviation (SD) from three independent experiments. Student’s t-test was used to assess the significance of the observed differences (P < 0.05 *, and P < 0.01 **).

Results

Inactivation of flhF in C. jejuni dramatically influences its colonization in both chicks and mice

The importance of flhF on C. jejuni commensal colonization in chicks was systematically analyzed for both the reference strain NCTC 11168 and for a C. jejuni isolate. For NCTC 11168 (Fig. 1a), the C. jejuni bacterial load in the intestinal tract was relatively low 1 day after inoculation. However, we could not isolate the flhF mutant strain at this time point and a significant difference in colonization was observed compared to the wild-type strain in the large intestine (caeca and colon) (P < 0.05). On day 7, the flhF mutant strain still could not colonize the small intestines of chicks (duodenum, jejunum and ileum), and although it was present in the large intestines of some chicks, the bacterial load was significantly lower than that observed for NCTC 11168-infected group (P < 0.01).

Fig. 1
figure 1

Colonization capacity of C. jejuni strains in chicks. The influence of flhF on C. jejuni colonization in chicks was evaluated for both the reference strain NCTC 11168 (a) and a C. jejuni isolate (b). The wild-type strain (hollow circle), the flhF mutant strain (hollow triangle) and the flhF complemented strain (cross) are represented by each sample. Five chicks were used in each group, the mean values for each group are depicted by the horizontal bar and asterisks indicate significant differences (P < 0.05 *, and P < 0.01 **)

Full size image

A hyper-colonization phenotype was observed for the C. jejuni isolate, as the bacterial loads in the intestines of chicks infected with the same dose were notably higher than in those of NCTC 11168-infected chicks, thus the effect of flhF was further determined in this C. jejuni isolate. As shown in Fig. 1b, the flhF gene was observed to be important in the initial establishment of colonization, since at 1 day post infection, the number of recovered isogenetic flhF mutant was significantly lower than that observed for chicks infected with the wild-type strain for both the small intestine (P < 0.01) and the caeca (P < 0.05). On day 7, the flhF mutant strain was almost completely cleared from the chicks and could be found only in the caeca of one chick, while the wild-type C. jejuni isolate consistently colonized the large intestines of chicks with a high load (approximately 109 cfu/g in the caeca). Complementation of flhF recovered this deficiency somewhat, both in C. jejuni NCTC 11168 and in the isolate, indicating that flhF significantly influences the commensal colonization of chicks.

The mice were less susceptible to C. jejuni colonization than the chicks. Although the C. jejuni isolate was observed to stably colonize both animals, the reference strain NCTC 11168 failed to colonize the mice in our study. For the C. jejuni isolate (Fig. 2), the bacteria recovered from the intestinal tract of mice were notably less abundant than those recovered from the chicks. However, the results showed that flhF does not greatly influence the initial colonization of mice. Although the average amount of the isolated flhF mutant was less than that of the wild-type strain, no significant differences were observed after 1 day post inoculation. On day 7, the flhF mutant strain could not be recovered from any regions of the intestines, while the wild-type strain continuously colonized the mice, with significant differences observed in the jejunum, ileum, caecum and colon (P < 0.01). These results suggested that inactivation of flhF resulted in the eventual elimination of the bacteria in mice.

Fig. 2
figure 2

Colonization capacity of C. jejuni strains in mice. The influence of flhF on C. jejuni colonization in mice was evaluated in the C. jejuni isolate. The wild-type strain (hollow circle), the flhF mutant strain (hollow triangle) and the flhF complemented strain (cross) are represented by each sample. Five mice were used in each group, the mean values for each group are depicted by the horizontal bar and asterisks indicate significant differences (P < 0.05 *)

Full size image

flhF does not influence the growth or primary biochemical characteristics of C. jejuni

The observed growth curves indicated that the inactivation of flhF does not influence the normal growth of C. jejuni. When cultured in MH broth, the OD600 values for C. jejuni NCTC 11168, the flhF mutant and the complemented strains were nearly same at different time points and no significant differences were observed, suggesting that these strains have a similar growth rate (Fig. 3). The biochemical characteristics of the C. jejuni strains were analyzed using an NH card with a VITEK 2 compact instrument. The NH card contains 30 biochemical tests that includes five alkalization tests, 10 acidification tests, 11 glycosidase and peptidase tests and four miscellaneous tests. For the C. jejuni flhF mutant and the complemented strain, all the indexes tested were the same as the NCTC 11168 strain except for OPS (phenylphosphonate) (Additional file 1: Table S3), indicating that the primary biochemical characteristics of C. jejuni were unaffected in the flhF mutant.

Fig. 3
figure 3

Growth curves of C. jejuni strains. C. jejuni NCTC 11168 and its derivatives were initially adjusted to an optical density at 600 nm (OD600) of 0.08, after which they were grown in MH broth under microaerophilic conditions with shaking at 42 °C, with 200 μl of samples were withdrawn at designated time point to examine growth. The wild-type strain (red circle), the flhF mutant strain (blue triangle) and the flhF complemented strain (green cross) are represented by each sample. The data are presented as the means ± standard deviation (SD) from three independent experiments

Full size image

C. jejuni flhF mutant shows a decreased ability to infect cultured epithelial cells

To characterize the infectivity of C. jejuni after inactivation of flhF, the ability of the strains to invade cultured epithelial cells was assessed. As shown in Fig. 4, inactivation of flhF significantly influenced the ability of C. jejuni to infect intestinal epithelial cells. Despite the effort made to minimize motility-dependent effects via centrifuging cells, the relative percentage of the flhF mutant strain to adhere and invade Caco-2 cells was 52 and 27%, respectively, compared to the NCTC 11168 strain. Complementation with flhF restored this defect in the mutant, and the infection ability of the complemented strain was nearly same as that observed for the wild-type strain (129 and 97% for adherence and invasion, respectively). These results suggest that the adherence to the epithelium and the subsequent invasion during C. jejuni infection is hampered by the inactivation of flhF.

Fig. 4
figure 4

The ability of C. jejuni strains to adhere to and invade Caco-2 cells. Cells were infected with C. jejuni NCTC 11168 and its derivatives for 2 h. A comparison of the levels of adherence is presented as the values of C. jejuni mutant and complemented strains relative to that of the wild-type strain, which was set at 100% (a). Gentamicin was added to another plate of cells and incubated for another 1 h to enumerate the invaded bacteria (b). The data are presented as the means ± standard deviation (SD) from three independent experiments. Significant differences (P < 0.05 *, and P < 0.01 **) are indicated by asterisks

Full size image

The autoagglutination and biofilm-formation abilities of C. jejuni are decreased by the inactivation of flhF

The AAG ability of C. jejuni strains was compared after they were statically incubated for a short (3, 6, and 9 h) and long (21, 24, and 27 h) time spans (Fig. 5a). The AAG phenotype of each tested strain became more prominent over time. During the first 6 h, no significant difference in the AAG ability was observed between the C. jejuni NCTC 11168 and flhF mutant strains. The influence of flhF on C. jejuni AAG became noticeable after the strains were incubated for 9 h. The measured OD600 of C. jejuni NCTC 11168 was significantly lower than that of the flhF mutant strain (P < 0.05), and this effect increased over time, as the observed difference was more significant at 21–27 h (P < 0.01).

Fig. 5
figure 5

Autoagglutination and biofilm formation of C. jejuni strains. C. jejuni NCTC 11168 and its derivatives were cultured for the designated times. Strains that strongly agglutinate will fall to the bottom of the tube, and a 200 μl samples from the top of the suspensions were taken to determine the OD600 values. The autoagglutination abilities of the C. jejuni wild-type strain (hollow circle), the flhF mutant strain (hollow triangle) and the flhF complemented strain (cross) were compared (a). The formed biofilms were stained with crystal violet, and biofilm formation was quantitated by determining the OD570 value (b). The data are presented as the means ± standard deviation (SD) from three independent experiments. Significant differences (P < 0.05 *, and P < 0.01 **) are indicated by asterisks

Full size image

The biofilms formed by the strains were quantified after being statically cultured for 48–96 h. As shown in Fig. 5b, the biofilms formed for each strain continuously increased over time, but the biofilm-forming potential was different between the strains. The flhF mutant strain showed a defect in biofilm formation, as the OD570 values of the tested strains across all assayed time points were significantly decreased compared to those observed for the NCTC 11168 strain (P < 0.01).

The biofilm-formation defect was partially restored after complementation of flhF, the AAG ability was also increased somewhat in the complemented strain. The detected values for the biofilms formed and AAG ability of the complemented strain were all higher than the mutant at all the assayed time points, although these values did not reach the level of the wild-type strain.

flhF is continuously upregulated during C. jejuni infection and influences the transcription of other infection-related genes

To analyze whether flhF is associated with C. jejuni infection in vivo, the NCTC 11168 strain was cultured in vitro and used to infect Caco-2 cell for 4, 8, 16 or 24 h, respectively. The RNA was extracted from samples at each time point, and a qRT-PCR assay was performed to compare the relative expression of flhF at the different time points. Compared to the in vitro groups, flhF was observed to be upregulated during the infection process at all the assayed time points. The observed fold-changes in expression were all over 3, and specifically for the 4, 8, 16 and 24 h time points were increased by 3.19-, 5.10-, 3.98- and 4.74-fold, respectively (Fig. 6a). This result suggests that flhF is a virulence-associated gene that may have a continuous role during C. jejuni infection.

Fig. 6
figure 6

Analysis of relative mRNA expression. C. jejuni NCTC 11168 was grown in vitro and in Caco-2 cells. The flhF gene was observed to be continuously upregulated during the Caco-2 cell infection process. Each column represents the relative mRNA expression of flhF during the cell infection process to the in vitro group. The data are presented as the means ± standard deviation (SD) from three independent experiments (a). Heatmap presentation of relative mRNA expression of other infection-related genes not directly involved in flagellar assembly. Red and green in the heatmap represent gene upregulation and downregulation, respectively, in the flhF mutant relative to the wild-type strain, with more saturated colors representing a greater difference as indicated in the bar (b). Some representatives were selected for further qRT-PCR verification. Each column represents the relative mRNA expression of the flhF mutant strain relative to the C. jejuni NCTC 11168 strain. The data are presented as the means ± standard deviation (SD) from three independent experiments (c)

Full size image

The genes not directly involved in flagellar assembly but that were reported to influence C. jejuni adhesion, invasion and colonization were summarized (Additional file 1: Table S2) and their relative expression after inactivation of flhF was analyzed using our previous microarray data. As shown in Fig. 6b, both the colonization- (40%, 20/50) and adhesion/invasion- (50%, 25/50) related genes were downregulated after the inactivation of flhF. The average influence on adhesion/invasion was slightly higher than that observed for colonization. Some genes that could influence both the adhesion/invasion and colonization ability of C. jejuni were identified (the overlapping genes are presented in the middle of the heatmap), and their average transcription was observed to be decreased more than that of the other two groups. Some genes were selected for further qRT-PCR verification. In addition to cadF and docC, which were slightly upregulated in the qRT-PCR assay (Fig. 6c), all the other assayed genes showed the same trends as the microarray, where the transcription of fedA, fedB, fedC and fedD (Feds) were apparently affected. Thus, besides the major role flhF plays in the flagellar system, it may also affect C. jejuni infection in other ways, but the underlying mechanism still requires further investigation.

Discussion

The efficient establishment of colonization is a prerequisite for C. jejuni pathogenesis. It was originally hypothesized that motility plays an essential role in C. jejuni colonization, as only motile strains could be recovered from human volunteers [42]. Subsequent studies revealed that flagella are an important colonization factor, as intact and motile flagella are required for efficient colonization [43]. The flagellar filament accounts for the largest proportion of the synthesized flagella and was the earliest studied flagellar substructure. The C. jejuni flagellar filament is composed of two almost identical flagellin proteins. Flagellin A is predominantly responsible for the intact and fully motile flagellation phenotype and has a specific contribution to C. jejuni colonization regardless of the effects observed by a defect in motility or inactivation of flaB [44]. Although Wassenaar et al. suggested that flagellin A, rather than motility, is essential for optimal colonization [44]. However, the role of motility cannot be ignored, as the mutation of pflA, which results in nonmotile bacteria with paralyzed flagella, also influences C. jejuni colonization [16].

Subsequent studies focused on the role of other flagellar elements. For the flagellar biosynthesis regulators RpoN (σ54) and FliA (σ28), mutation of either of these regulators resulted in an apparent attenuation for caecal colonization in chicks, but the loss of fliA had a more obvious effect than that of rpoN, despite its lesser influence on C. jejuni motility [18]. FlgR is a σ54-dependent response regulator that was observed to decrease the C. jejuni colonization ability by 10- to 500-fold in chicks [45]. Although inactivation of the flagellar motor protein MotAB caused C. jejuni to be nonmotile, these cells expressed a full-length flagellum yet completely lost the ability to colonize chicks [24, 46]. FlgK is a flagellar hook-associated protein, and the colonization of an flgK mutant was five orders of magnitude lower than that observed for wild-type strain in chicks [18]. The flagella secreted Campylobacter invasion antigen (Cia) proteins also play a role in colonization. Mutations in CiaI and CiaB lead to a defect in chick colonization compared to the wild-type strain [15, 22, 25]. Furthermore, genes associated with the flagellin O-linked glycosylation system also affect C. jejuni colonization and may be mediated by structural and surface changes in the glycan when interacting with the host [16, 23].

Although evaluation of bacterial colonization ability between studies may be difficult due to the diverse methods used by various laboratories, when assayed using the same method the flagellar elements were shown to truly have a various contributions to C. jejuni colonization [16]. Thus, the role of FlhF on C. jejuni colonization was systematically analyzed in this study. C. jejuni is more commonly considered a commensal intestinal microbiota in chickens, where it has a persistent presence in the gut but does not lead to pathology [7]. The chick colonization assay was first performed using both the reference strain NCTC 11168 and an isolate from monkey diarrhea. Different C. jejuni strains shows diverse infectivity, and the colonization level of the NCTC 11168 strain was not ideal compared to that of the fecal isolate, which may be due to it being a poor colonizer [9]. In our study, inactivation of flhF in NCTC 11168 and the isolate both resulted in significantly decreased C. jejuni colonization at 1 and 7 days post inoculation, indicating that the commensal colonization of C. jejuni was severely hampered by the loss of flhF.

In contrast to chicks, C. jejuni infection in humans often leads to a series of illnesses, but the underlying mechanism is not fully understood [47]. To understand whether the importance of flhF on commensal colonization was consistent in mammals, mice were used for further assays. The reference strain NCTC 11168 failed to colonize the mice in our assay. Similar findings were also reported in a previous study, where NCTC 11168 was observed to colonize mice at very low efficiency, although after this strain was reisolated from mice it could then colonize mice with a high efficiency [48]. Thus, this phenomenon may be caused by the subculture of this genome-sequenced strain in the laboratory or the need for genetic adaptations in the intestinal tracts to promote its growth [48, 49]. The C. jejuni isolate used in our study could stably colonize the assayed mice. Compared to the chicks, the initial establishment of colonization in mice was not severely influenced by the inactivation of flhF, which may be due to the diverse characteristics of C. jejuni with respect to infecting different hosts. At 7-days post inoculation, all of the isogeneic flhF mutants were cleared from the mice, indicating its important role in promoting mice colonization.

The possible factors that may contribute to the colonization defect were investigated. C. jejuni growth and metabolism were observed to not be influenced by the mutation of flhF, eliminating the possible effects of the flhF mutation on bacterial multiplication. Previous studies showed that the motility and flagellar biosynthesis of C. jejuni were severely affected by the inactivation of flhF [34, 35]. The flagella of C. jejuni are involved in not just motility [26], so some of the flagella-associated phenotypes which might play a role in the bacterial colonization were subsequently analyzed. Adherence to the epithelium helps bacteria resist intestinal peristalsis and expulsion [17], and the subsequent invasion is also important. These factors are key determinants for C. jejuni colonization [50]. The potential infectivity of the assayed C. jejuni strains on Caco-2 cells was assessed. Despite efforts to minimize the influence caused by motility, both the adhered and invaded bacteria were observed to be significantly decreased after inactivation of flhF, indicating that the colonization was hampered at this step. Autoagglutination is the basis of bacterial aggregation, which is involved in biofilm formation [51]. Biofilm formation is essential for both environmental survival and successful infections [52, 53], and this mode of bacterial growth is also required for C. jejuni colonization [53]. The autoagglutination and biofilm-formation abilities of the C. jejuni flhF mutant strain were observed to be decreased in our study, contributing to the observed colonization defect of this strain. The complementation of flhF completely or partially restored the defects of adhesion, invasion and biofilm-formation abilities of the flhF mutant strain. However, limited restoration of the AAG ability was observed. Previous studies showed that the presence of the flagellin and its glycosylation modification is crucial for C. jejuni AAG [51]. We has observed the intact flagellin in the flhF complemented strain in our previous study [34], so flagellin glycosylation may not be fully restored after complementation of flhF.

In addition to the above mentioned phenotypic changes, the mechanism by which flhF affects C. jejuni colonization was also explored at the gene expression level. Although flhF is a class I flagellar gene that is stably expressed both in vitro and in vivo [34], when the two sets of expression data were compared, we observed that flhF is continuously upregulated during the infection process. This observation suggests a close relationship between this gene and C. jejuni infectivity, maybe an intact flagellum is required to maintain the secretion of virulence proteins during the infection process [6], or FlhF may function continuously to regulate other virulence factors in an unclarified-method. Inactivation of flhF globally affects flagellar gene transcription [34]. Unlike many other flagellated species whose flagellar genes are clustered into operons, the flagellar genes in C. jejuni are scattered throughout the genome [34, 54]. Many of these genes are grouped into operons that are involved in other pathways [54], and some genes are also co-expressed with flagellar genes [15]. Thus, we were interested in determining whether flhF could affect other infective factors not directly involved in flagellar assembly in this bacterium. Some representative genes were identified based on the literature, and using our previously obtained microarray data [34]. We observed that 45% (39/86) of the colonization and adhesion/invasion related genes were downregulated, among which the expression of flagellar coexpressed determinants (Feds) was apparently affected. This phenomenon may be caused by the Feds genes that belong to the flagellar σ28 regulon [15], which are significantly affected by the inactivation of flhF [34].

In this study, we systematically analyzed the influence of FlhF on C. jejuni colonization and the possible mechanisms that may contribute to this defect, including the associated virulence phenotypes and a preliminary exploration of changes at the gene expression level. The pathogenesis of C. jejuni is not yet well elucidated compared to other analogous species that can also cause diarrheal diseases. C. jejuni lacks many of the classic virulence factors [1, 3, 6]. It is generally believed that C. jejuni has a regulatory mechanism that co-regulates its virulence. For example, the polar flagella are not only required for motility but also functions as a type III secretion system (T3SS) [26, 54]. The flhF gene is continuously upregulated during the infection process and influences the expression of other infection-related factors, which indicates a close relationship between this gene and C. jejuni infectivity. It is unknown whether this is an isolated case or a common phenomenon for other flagellar genes or whether other aspects related to C. jejuni pathogenesis are affected. These questions still need to be further investigated, which will provide insights into the role of this gene as well as the mechanism of the co-regulation of virulence of C. jejuni flagella.

Conclusion

In this study, the influence of flhF on C. jejuni colonization was systematically evaluated. Inactivation of flhF resulted in severe defects in the commensal colonization of chicks and resulted in a complete loss in the ability of this bacterium to continuously colonize mice. A combination of phenotypic and genetic characterizations revealed that there is a decreased ability of flhF mutant strain to adhere and invade epithelial cells, and autoagglutination and biofilm formation, which may contribute to the colonization defect. The flhF gene was observed to be continuously upregulated during the infection process and affects the transcription of other infection-related genes that are not directly involved in flagellar assembly, such as the flagellar coexpressed determinants (Feds). This result makes sense, but their definite relationship with flhF and the underlying mechanisms that influence C. jejuni pathogenesis still requires further investigation.