Campylobacter jejuni Motility Is Required for Infection of the Flagellotropic Bacteriophage F341

Abstract

Previous studies have identified a specific modification of the capsular polysaccharide as receptor for phages that infect Campylobacter jejuni. Using acapsular kpsM mutants of C. jejuni strains NCTC11168 and NCTC12658, we found that bacteriophage F341 infects C. jejuni independently of the capsule. In contrast, phage F341 does not infect C. jejuni NCTC11168 mutants that either lack the flagellar filaments (ΔflaAB) or that have paralyzed, i.e., nonrotating, flagella (ΔmotA and ΔflgP). Complementing flgP confirmed that phage F341 requires rotating flagella for successful infection. Furthermore, adsorption assays demonstrated that phage F341 does not adsorb to these nonmotile C. jejuni NCTC11168 mutants. Taken together, we propose that phage F341 uses the flagellum as a receptor. Phage-host interactions were investigated using fluorescence confocal and transmission electron microscopy. These data demonstrate that F341 binds to the flagellum by perpendicular attachment with visible phage tail fibers interacting directly with the flagellum. Our data are consistent with the movement of the C. jejuni flagellum being required for F341 to travel along the filament to reach the basal body of the bacterium. The initial binding to the flagellum may cause a conformational change of the phage tail that enables DNA injection after binding to a secondary receptor.

INTRODUCTION

Campylobacter jejuni is a pathogenic bacterium that causes gastroenteritis in humans and is frequently associated with consumption or handling of undercooked Campylobacter-contaminated poultry meat (1). Early on, it was shown that the motility of Campylobacter jejuni is one of the key factors in establishment of human disease, as only motile C. jejuni could be recovered after passage in human volunteers challenged with a mixture of motile and nonmotile C. jejuni strains (2). Since then, other studies have confirmed that motility is required for C. jejuni colonization and infection using animal models such as piglets and mice that mimic disease in humans (3–5). First, flagellum-driven motility enables C. jejuni to travel through and colonize the highly viscous mucus layer covering the surface of epithelial cells (6, 7). Second, the flagellum filament is an adhesin that promotes contact to host cells and contributes to bacterial binding of the host epithelium (8–10). Likewise, C. jejuni motility is necessary for maximal invasion of human intestinal epithelial cells in vitro (11). Hence, several steps in the infection process require a functional flagellum. The flagellum of C. jejuni also functions as a type III secretion system that secretes a number of proteins called Campylobacter invasion antigens (Cia) and flagellar coexpressed determinants (Feds) required for efficient invasion of epithelial cells (12–14). In addition, wild-type C. jejuni heavily colonizes the chicken cecum with up to 10⁹ C. jejuni cells per gram cecal content, while nonmotile C. jejuni mutant cells (ΔmotAB) carrying paralyzed flagella colonized chicken cecum to only 10³ cells per gram cecal content 7 days post-inoculation of the birds (15). Thus, C. jejuni motility has also proven to be required for chicken colonization (15, 16).

The motility of C. jejuni is driven by rotation of single uni- or bipolar flagella, controlled by a chemosensory system that allows the bacteria to move toward favorable environments and away from harmful conditions (17, 18). The flagellum filament is attached to a hook protein in a basal structure embedded in the cell membrane and consists of polymeric proteins FlaA and FlaB, encoded by two almost identical adjacent genes, flaA and flaB (19–21). FlaA is the major component of the filament and is required for formation of a full-length flagellum and thus motility (11). The C. jejuni flagellar filaments are posttranslationally modified by O-linked glycans that have proven essential for assembly of the flagellum and therefore also motility (22–24). The O-linked glycans of C. jejuni flagella are highly variable, due to the genetic diversity of the O-linked glycosylation locus as well as phase-variable homopolymeric tracts promoting switching between “on” and “off” expression of glycosylation genes (25–27).

Importantly, C. jejuni flagella (16) and flagellin A per se (28) have been shown to be required for efficient colonization of chicken ceca (29–31). Also, Campylobacter bacteriophages have been isolated from the cecum of chickens (30). Most bacteriophages infecting Campylobacter belong to the family Myoviridae and are classified according to morphology and genome size, dividing them into three groups, two of which are represented by a type phage (group I, 320-kb genomes; group II, 180- to 194-kb genomes [type CP220]; group III, 130- to 140-kb genomes [type CP81]) (32, 33). The initial interaction between phage and bacteria involves a specific binding of the phage to a bacterial surface receptor such as outer membrane proteins, lipopolysaccharides, or capsular polysaccharides (CPSs) (34). The first description of possible receptors recognized by phages infecting C. jejuni came from a study using a transposon library of C. jejuni NCTC11168 to screen for phage resistance toward 16 phages (35). This screening revealed that the CPS and flagella of C. jejuni may be involved in phage recognition of this bacterium (35). Recently, we identified a specific modification of the CPS, namely, the O-methyl phosphoramidate (MeOPN) moiety of GalfNAc of C. jejuni NCTC11168, as a receptor for phage F336 (36). Furthermore, we showed that five CPS phages were all highly dependent on MeOPN attachment to GalfNAc residues for successful infection and that both absence and presence of other CPS modifications influenced phage infection (37).

In contrast, the initial phage-host interaction of phages dependent on the C. jejuni flagellum for infection is highly understudied. However, bacteriophages targeting the flagellum have been investigated in other bacteria and different infection mechanisms for bacteriophage interaction with bacterial flagella have been proposed (34, 38–42). While one study provides data indicating the flagellar rotation to drive bacteriophage χ movement along the filament of Escherichia coli and Salmonella enterica serovar Typhimurium (42), the authors of another study speculate that phage iEPS5 injects its genetic material directly into the flagellum of Salmonella Typhimurium (38). Flagella are essential for C. jejuni host colonization and infection, and therefore, it is important to investigate how flagellotropic phages interact with flagella that ultimately may affect the pathogenicity of C. jejuni. The aim of the present study was to identify a C. jejuni phage dependent on flagella for infection and to elucidate the potential mechanism for this phage-bacterium interaction. We show that phage F341 adsorbs to the flagellum of C. jejuni NCTC11168 prior to DNA injection and that both phage adsorption to flagella and the motility of the bacterium are prerequisites for infection.

MATERIALS AND METHODS

Bacteriophage, bacterial strains, media, and growth conditions.

Bacteriophage F341 was originally isolated from a broiler intestine and is propagated on C. jejuni NCTC12658 (43). C. jejuni strains used in this study are listed in Table 1 and were routinely grown on base II agar plates supplemented with 5% horse blood (BA plates) under microaerobic (MA) conditions (6% CO₂, 6% O₂, 84.5% N₂, 3.5% H₂) at 37°C. C. jejuni strains for phage work were routinely cultivated in brain heart infusion (BHI) supplemented with 10 mM MgSO₄ and 1 mM CaCl₂ (CBHI). When appropriate, kanamycin and chloramphenicol were added to plates at final concentrations of 30 μg/ml and 20 μg/ml, respectively.

TABLE 1.

Bacterial strains used in this study

C. jejuni strain	Relevant characteristic or genotype	Source or reference
NCTC12658 (SB18)	Propagating strain for bacteriophage F341	National Collection of Type Cultures
NCTC12658 kpsM (SB96)	kpsM::kan	This study
NCTC12658 ΔmotA (SB230-1)	ΔmotA::cat	This study
NCTC11168 (SB21)	Sensitive to bacteriophage F341	National Collection of Type Cultures
11168H kpsM (SB38)	kpsM::kan	48
NCTC11168 kpsM (SB100)	kpsM::kan	This study
NCTC11168 ΔflaAB (SB101)	ΔflaAB::cat	This study
NCTC11168 ΔflgP (SB102)	ΔflgP::cat	This study
NCTC11168 ΔmotA (SB103)	ΔmotA::cat	This study
NCTC11168 kpsM ΔflaAB (SB104)	kpsM::kan ΔflaAB::cat	This study
NCTC11168 kpsM ΔmotA (SB106)	kpsM::kan ΔmotA::cat	This study
NCTC11168 ΔflgP+flgP (CV1217-2)	ΔflgP::cat + flgP	This study
NCTC11168 cj1295	cj1295::kan	25
11168H cj1316c	cj1316c::kan	B. Wren
11168H cj1324	cj1324::kan	63
NCTC11168 cj1295 (SB158)	cj1295::kan	This study
NCTC11168 cj1316c (SB232)	cj1316c::kan	This study
NCTC11168 cj1324 (SB162)	cj1324::kan	This study

Natural transformation.

Natural transformation was performed as previously described (44). Transformants carrying resistance markers were isolated by plating on blood agar base II plates supplemented with appropriate antibiotics.

Construction of C. jejuni NCTC11168 and NCTC12658 mutants.

ΔflaAB, ΔmotA, and ΔflgP mutants were constructed in C. jejuni NCTC11168 by replacing the majority of the respective genes with a chloramphenicol acetyltransferase (cat) gene encoding chloramphenicol resistance as previously described (45). Primers used are listed in Table 2.

TABLE 2.

Plasmids, strains, and primer sequences used for construction of mutants

Plasmid, strain, or primer	Comment or primer sequencea	Reference or construct for which primer was used
Plasmids
pRY109	Plasmid containing chloramphenicol resistance cassette	46
pCV1210-1	pSC-A::ΔFlaAB Ampr Camr	This study
pCV1206-1	pCR2.1-TOPO::ΔMotA Ampr Kanr Camr	This study
pCV1208-1	pSC-A::ΔFlgP Ampr Camr	This study
pCV1214-3	pKfdxA::FlgP Kanr	This study
pTA-3	pTZ57R/T::kpsM Ampr Kanr	This study
Strains
C. jejuni NCTC11168		National Collection of Type Cultures
C. jejuni NCTC12658		National Collection of Type Cultures
E. coli SoloPack PCR cloning-competent cells		Stratagene
E. coli TOP10		Invitrogen
Primers
CV1106	CGCGGATCCAGATGGGCAAACACATTCTC	ΔflaAB
CV1107	GCACTGATAAATAACATCCTTGCGCAATCAGGT	ΔflaAB
CV1108	AGGATGTTATTTATCAGTGCGACAAACTGG	ΔflaAB
CV1109	TTTAGATGCTTCGGCGGTGTTCCTTTCCAAG	ΔflaAB
CV1110	ACACCGCCGAAGCATCTAAACTTTTACTATTTAAATC	ΔflaAB
CV1111	CGCGGATCCTGTATGATTGTTTTAAGAAATAAAGC	ΔflaAB
CV1094	CGCGGATCCCTTGTATCTGCTGTATCGAC	ΔmotA
CV1095	GCACTGATAAAGCTAATGGAATGGATCTTG	ΔmotA
CV1096	TCCATTAGCTTTATCAGTGCGACAAACTGG	ΔmotA
CV1097	CATCTTTCTTTCGGCGGTGTTCCTTTCCAAG	ΔmotA
CV1098	ACACCGCCGAAAGAAAGATGCACAACATGC	ΔmotA
CV1099	CGCGGATCCACACGGACAATATGCAAAAG	ΔmotA
CV1100	CGCGGATCCCAAACGAACCGCTTCTATAG	ΔflgP
CV1101	GCACTGATAAAAATCGATGCAGACAAGTGG	ΔflgP
CV1102	GCATCGATTTTTATCAGTGCGACAAACTGG	ΔflgP
CV1103	TCAAGATGTTTCGGCGGTGTTCCTTTCCAAG	ΔflgP
CV1104	ACACCGCCGAAACATCTTGAGATGGAGCAG	ΔflgP
CV1105	CGCGGATCCAGTTAGAGCGATCAACTTGG	ΔflgP
CV1141	ATCGTCTCACATGAAAAAAATTTATTTTATGC	flgP complement
CV1142	ATCGTCTCACATGTTAATAAGCAAACAATTCTT	flgP complement
kpsM 4 forw	ATCCTAGCACTCATTCCCGAAG	ΔkpsM
kpsM 4 rev	ACTAAGCATATAAGATTAGCCAGTG	ΔkpsM

^aComplementary overhangs are marked by italics.

The ΔflaAB mutant was constructed by amplifying an 859-bp DNA upstream fragment (5′ end of flaA and flanking region) and a 1,038-bp DNA downstream fragment (3′ end of flaB and flanking region) from C. jejuni NCTC11168 using the primers CV1106 plus CV1107 and CV1110 plus CV1111, respectively. The cat gene was amplified from pRY109 (46) with primers CV1108 and CV1109. The cat fragment was subsequently inserted between the upstream and downstream fragments (creating a 3,418-bp deletion of flaAB) using the SOEing (splicing by overhang extension) PCR technique (47). The ΔflaAB::cat fragment was cloned into the Strataclone vector pSC-A (Stratagene) and transformed into Escherichia coli SoloPack PCR cloning-competent cells (Stratagene). The resulting plasmid pCV1210-1 (Amp^r Cat^r) was subsequently purified from E. coli (plasmid Midiprep kit; Qiagen) and transferred to C. jejuni NCTC11168 by electroporation. Insertion of the mutated allele into the C. jejuni chromosome was verified by PCR and sequencing. ΔmotA and ΔflgP mutants were constructed essentially as described above by replacing 519 bp and 376 bp of motA and flgP, respectively, with the cat gene. The mutants were verified by PCR and sequencing, and nonmotile phenotypes were verified by lack of swarming ability in low-percentage soft motility agar (Fig. 1) as well as by microscopic inspection.

FIG 1.

Motility of C. jejuni NCTC12658 wild type (wt), C. jejuni NCTC11168 wild type, and mutants. Results are from 5 technical replicates in 3 individual experiments, shown as averages and standard deviations of growth zones after 24 h of incubation.

C. jejuni 11168H kpsM::kan, cj1316c::kan, and cj1324::kan strains were kindly donated by Brendan Wren, and C. jejuni NCTC11168 cj1295::kan was kindly provided by Dennis Linton. Mutations in kpsM, cj1316c, cj1324, and cj1295 were transferred to our background strain C. jejuni NCTC11168 by natural transformation using chromosomal DNA (DNeasy blood and tissue kit; Qiagen) isolated from each of the three mutants and selected for kanamycin resistance.

C. jejuni NCTC11168 double mutants (kpsMΔflaAB and kpsMΔmotA) were constructed by natural transformation of C. jejuni NCTC11168 kpsM::kan (SB100) using chromosomal DNA from either C. jejuni ΔflaAB::cat (SB101) or C. jejuni ΔmotA::cat (SB103) and selected for kanamycin and chloramphenicol resistance. The mutants were verified by PCR and sequencing, and nonmotile phenotypes were verified by lack of swarming ability in low-percentage soft motility agar (Fig. 1) as well as by microscopic inspection.

Complementation of the ΔflgP mutant was carried out as previously described (44) using the vector pKfdxA and the primers CV1141 and CV1142 to reintroduce the flgP gene into the ΔflgP background.

A C. jejuni NCTC12658 kpsM mutant was constructed by amplifying kpsM::kan from C. jejuni 11168H kpsM (48) using primers kpsM 4 forw and kpsM 4 rev (Table 2) and cloning the PCR fragment into the pTZ57R/T cloning vector (Thermo Scientific) according to the manufacturer's instructions. Vectors containing the amplified kpsM::kan fragment (pTA-3) were introduced into C. jejuni NCTC12658 by electroporation, and transformants were selected for kanamycin resistance. Finally, construction of ΔmotA in NCTC12658 was done by introducing the plasmid pCV1206-1 into C. jejuni NCTC12658 by electroporation. The mutants were verified by PCR and sequencing. The nonmotile phenotype of NCTC12658 ΔmotA was further verified by lack of swarming ability of the strain in soft motility agar together with a microscopic analysis.

Motility assay.

Motility assays were performed essentially as described previously (49). Briefly, C. jejuni strains were grown under MA conditions at 37°C overnight (ON) on BA plates, harvested in BHI, and adjusted to an optical density at 600 nm (OD₆₀₀) of 0.1. One microliter of cell suspension was spotted on a predried heart infusion broth plate containing 0.25% agar (soft motility agar), and plates were incubated under MA conditions at 37°C for 24 h. Average motility was determined from five plates by measuring the growth zone diameter at three different locations on each plate. The assays were performed at least twice.

Propagation of bacteriophages.

Phages were propagated using the double-layer method as previously described (36). Briefly, the propagating strain was harvested in CBHI after overnight growth on BA plates. The cell suspension was adjusted to an OD₆₀₀ of 0.350, incubated under MA conditions at 37°C for 4 h to allow cells to recover, and subsequently mixed with phages suspended in sodium chloride-magnesium sulfate buffer (SM buffer) (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 10 mM MgSO₄, 0.01% [wt/vol] gelatin). Phages and bacteria were incubated for 15 min to allow phage adsorption, and 0.6 ml of the suspension was mixed with 5 ml melted NZCYM overlay agar (0.6% agar) tempered to 45°C and gently poured onto NZCYM agar plates (1.2% agar) supplemented with vancomycin (10 μg/ml). Following incubation overnight under MA conditions at 37°C, plates with confluent (at least 50%) or complete lysis were flooded with 5 ml cold SM buffer, and phages were eluted by gentle shaking (80 to 100 rpm) overnight at 4°C. Finally, phages in SM buffer were collected, filtered through an 0.22-μm sterile syringe filter (Millipore), and stored at 4°C.

Titration of bacteriophages and host range analysis.

Bacteriophage titration was performed as previously described (36). Briefly, C. jejuni wild-type cells or mutants were prepared as described under “Propagation of bacteriophages.” Bacterial lawns were made by mixing 500 μl of bacterial culture with 5 ml NZCYM overlay agar (0.6% agar) at 45°C and poured onto a predried NZCYM plate containing vancomycin (10 μg/ml). The plates were allowed to set and then dry in the fume hood for 45 to 60 min before 10-fold serial dilutions (10⁰ to 10⁻⁷) of the F341 phage stock were spotted (3 × 10 μl) on the lawns. Once spots had dried into the plates, these were incubated for 20 to 24 h under MA conditions at 37°C. Phage titers were calculated by counting PFU from the dilutions where single plaques were present. Spots with no single plaques and full lysis zones at high phage concentrations (10⁰ and 10⁻¹ dilutions, corresponding to approximately 10⁷ to 10⁶ PFU in each spot) were interpreted as “lysis from without.” Experiments were performed at least twice.

Bacteriophage adsorption assay.

Bacteriophage binding assays were performed as previously described by Scott et al., with minor changes (50). Briefly, C. jejuni grown on BA plates overnight was harvested in CBHI and centrifuged at 6,000 × g for 5 min, followed by resuspension of the pellet in CBHI. This washing procedure was repeated twice, and the final cell suspension was adjusted to an OD₆₀₀ of 0.4. Bacteriophages were added to the bacterial suspension at a final concentration of log₁₀ 4 to 5 PFU/ml (multiplicity of infection [MOI] of approximately 0.0001) and incubated at 37°C with gentle shaking for 90 min. Samples were taken at 0, 30, 60, and 90 min, filtered through an 0.22-μm sterile syringe filter (Millipore), and stored at 4°C until enumeration of free phages in the supernatant (see “Titration of bacteriophages and host range analysis”). Experiments were repeated twice. For each experiment, the phage titer at time zero was accounted as 100% free phages, and from this background level, the percentage of free phages was calculated for the remaining time points and the mean percentages of free phages and standard deviations thereof were determined. For each time point (30, 60, and 90 min), a one-way t test was performed to determine if phage adsorption was statistically significantly different (P < 0.05) from that at time zero.

SYBR gold staining of bacteriophages.

The phages were stained with SYBR gold nucleic acid gel stain (Invitrogen) as described previously (51) with minor modifications. All handling with the SYBR gold stain was done with careful protection from light. Briefly, the original SYBR gold stock solution (10,000×) was diluted 1,000-fold (in MilliQ water) into a working solution and kept in the freezer for up to 1 month. SM-buffered phage lysate was double filtered through an 0.22-μm sterile syringe filter (Millipore), to remove cell debris, and treated with a final concentration of 1 μg/ml DNase and RNase A for 30 min at 37°C to remove extracellular DNA and RNA. This was followed by staining of the phages at a final concentration of 2.5% (vol/vol) diluted SYBR gold solution and incubation in the dark overnight (at least 8 h) at 4°C. Stained phages were purified by ammonium acetate purification prior to confocal fluorescence microscopy.

Ammonium acetate purification and up-concentration of bacteriophages by ultracentrifugation.

Ammonium acetate purification of phages was performed according to the method of Ackermann (52). Briefly, the SYBR gold-stained phages with a titer of at least 10⁸ PFU/ml were transferred to Beckman Coulter 3P tubes (Beckman) and ultracentrifuged at 274,000 × g for 1 h in an L-70 ultracentrifuge (Beckman). The supernatant was discharged, and the sediment was carefully dissolved in sterile 0.1 M ammonium acetate. The centrifugation procedure was repeated twice more, and the final sediment was dissolved in 0.5 to 1.0 ml ammonium acetate and kept at 4°C until use.

FM 4-64 staining of Campylobacter.

Serial dilutions of C. jejuni strains were prepared in liquid culture (CBHI) to obtain a culture of an OD₆₀₀ around 0.1 (∼10⁸ CFU/ml) after overnight (ON) growth under MA conditions at 37°C. Staining of C. jejuni with the lipophilic styryl/lipid membrane dye FM 4-64 (Molecular Probes) (1.0 mg/ml in dimethyl sulfoxide [DMSO]) was performed as described previously, but with minor changes (53). Briefly, C. jejuni cells were incubated with FM 4-64 at a final concentration of 0.05 mg/ml and allowed to stain for 10 to 20 min at 37°C. All handling with the FM 4-64 stain was carried out with careful protection from light.

Confocal fluorescence microscopy.

Confocal fluorescence microscopy was used to study the phage infection mechanism by visualizing the phage DNA. Prior to microscopy, SYBR gold-stained phages (∼10⁹ PFU/ml) and FM 4-64-stained C. jejuni (∼10⁸ CFU/ml) were mixed at an MOI of approximately 10. To observe the progress of the infection, 10-μl samples were taken out on an object glass at specific time points and immediately mixed with 10 μl 2% agarose tempered to 55°C. A cover glass was put on top, and the phage-C. jejuni mixture was squeezed onto the object glass. Agarose as a mounting material was used to prevent C. jejuni cells from moving but without killing the cells or disrupting cell surface structures such as the flagella. To study the interaction between phages and C. jejuni, confocal fluorescence microscopy was performed using an Inverted Point Scanning Confocal SP5 II microscope from Leica Microsystems with five photomultiplier detectors and equipped with an argon laser. Fluorescence excitation/emission maxima of SYBR gold are ∼495/537 nm, whereas the excitation/emission spectra for FM 4-64 are 515/640 nm. This allowed a joined excitation light but different emission lights whereby the two stains were differentiated. The wavelength used for excitation was 496 nm. The “green” fluorescence of the SYBR gold stain was detected using an emission filter of 505 to 545 nm, and the “red” FM 4-64 stain was detected using an emission filter of 620 to 660 nm.

Imaging of bacteriophage adsorption and infection by electronic microscopy.

The process of adsorption to, and infection of, C. jejuni NCTC11168 by phage F341 was visualized by use of transmission electron microscopy (TEM). C. jejuni NCTC11168 was harvested as described under “Propagation of bacteriophages.” After 4 h of growth under MA conditions at 37°C, the cells were mixed with phage F341 at an MOI of approximately 10. At multiple time points, 5-μl samples were taken from the phage-bacterium mixture and dispensed onto freshly glow-discharged 0.25% Pioloform-coated copper grids (Athene type, 3 mm; Agar Scientific, United Kingdom) for 5 min. Subsequently, samples were negatively stained with 1% aqueous uranyl acetate that was immediately removed. The staining procedure was repeated once more. Samples were observed at 80 kV on a JEOL 1220 transmission electron microscope (JEOL, United Kingdom), and images were acquired with an SIS MegaView III digital camera with iTEM software.

RESULTS

Phage F341 is independent of CPS for infection of C. jejuni.

Recently, we showed that a group of phages are dependent on capsular polysaccharides (CPSs) for infection of C. jejuni NCTC11168 and identified O-methyl phosphoramidate attached to GalfNAc in CPS as a receptor of phage F336 (36, 37). With the aim of identifying phages independent of CPS for infection, we screened our phage collection (43) for phages forming plaques on lawns of an acapsular C. jejuni NCTC11168 kpsM mutant. We found that one of our phages, called F341, which belongs to group III (type CP81) of Campylobacter phages indeed formed plaques on the capsule-deficient kpsM mutant and even with a 10- to 100-fold-increased efficiency of plaquing (EOP) compared to wild-type C. jejuni NCTC11168 (Table 3), hence showing that phage F341 infection is independent of CPS. This conclusion was confirmed using the propagating strain of phage F341, C. jejuni NCTC12658, and a corresponding kpsM mutant, as the wild type and the acapsular mutant were infected by phage F341 at the same efficiency (Table 4). Titration of the F341 phage stock showed a 100-fold-higher phage titer using lawns of C. jejuni NCTC12658 compared to NCTC11168, suggesting that phage F341 infection is more efficient using the propagating strain. Finally, we noted that lack of the capsule did not affect the motility of C. jejuni (Fig. 1). Thus, we identified phage F341 to be independent of the capsule for infection of C. jejuni NCTC11168 and NCTC12658.

TABLE 3.

Phage F341 infection of C. jejuni NCTC1168 and derived mutants

Mutant genotype	Relevant phenotype	Plaque formationa	Lysis zoneb	Motilityc	EOPd
Wild type	Wild type	106–107	Yes	Motile	1
kpsM	No capsule	107–108	Yes	Motile	10–100
ΔflaAB	No flagellar filaments	No plaques	No	Nonmotile	—e
ΔmotA	Paralyzed flagella	No plaques	No	Nonmotile	—
ΔflgP	Paralyzed flagella	No plaques	No	Nonmotile	—
kpsMΔflaAB	No capsule, no flagellar filaments	No plaques	Yes	Nonmotile	—
kpsMΔmotA	No capsule, paralyzed flagella	No plaques	Yes	Nonmotile	—
ΔflgP+flgP	Complemented, flagellar rotation	106–107	Yes	Motile	1
cj1295	Changes in flagellar glycosylation	106–107	Yes	Motile	1
cj1316c	Changes in flagellar glycosylation	106–107	Yes	Motile	1
cj1324	Changes in flagellar glycosylation	106–107	Yes	Motile	1

^aPlaque formation was determined as PFU per ml, and numbers are noted as intervals based on data from at least 10 experiments, except for ΔflgP+flgP, cj1295, and cj1324, 3 replicates of which have been included.

^bLysis zones were determined as complete lysis from a drop (10 μl) of concentrated phage suspension (10⁰ to 10⁻¹).

^cNonmotile strains were defined as growth zones less than 6 mm in soft motility agar.

^dEfficiency of plaquing relative to wild type.

^e—, no plaques.

TABLE 4.

Phage F341 infection of C. jejuni NCTC12658 wild type and mutants

Mutant genotype	Relevant phenotype	Plaque formationa	Motilityb	EOPc
Wild type	Wild type	109	Motile	1
kpsM	No capsule	109	Motile	1
motA	Paralyzed flagella	5 × 105–1 × 106	Nonmotile	0.0005–0.001

^aPlaque formation was determined as PFU per ml.

^bNonmotile strains were defined as growth zones less than 6 mm in soft motility agar.

^cEfficiency of plaquing relative to wild type.

Phage F341 requires a motile flagellum for infection of C. jejuni.

In addition to CPS, motility and flagella have been suggested to be important for phage infection of C. jejuni (35). Thus, knowing that CPS was not necessary for phage F341 infection, we constructed a C. jejuni NCTC11168 ΔflaAB mutant lacking the flagellar filaments and tested whether phage F341 was able to infect this strain. Interestingly, we found that phage F341 did not form any plaques on lawns of the flaAB mutant (Table 3), hence suggesting that phage F341 is dependent on flagella and/or motility for infection of C. jejuni.

To further investigate if infection by phage F341 is dependent on the flagellum or motility per se, we constructed two additional flagellar mutants and examined their susceptibility to phage F341. C. jejuni NCTC11168 ΔmotA has a paralyzed, i.e., nonrotating, flagellum by lack of the motor stator protein MotA (54, 55), and C. jejuni NCTC11168 ΔflgP likewise carries a paralyzed flagellum as previously described (56). Similar to the flaAB mutant, no plaques were observed on lawns of either the motA mutant or the flgP mutant, thus showing that phage F341 cannot infect C. jejuni NCTC11168 carrying intact but paralyzed flagellar filaments. Complementing flgP confirmed that phage F341 requires rotating flagella to complete successful infection, as both motility and plaque formation were reverted to wild-type levels (Table 3).

Similar to the flagellar single mutants, phage F341 did not form any single plaques on lawns of double mutants of C. jejuni NCTC11168 (kpsMΔflaAB and kpsMΔmotA) lacking both capsule and motility (Table 3; Fig. 1). But, we observed that 10⁶ to 10⁷ F341 phage were able to completely lyse bacterial lawns of the double mutants, but only when phage were present at such high concentrations in one spot. This suggests that phage F341 causes lysis from without, defined as adsorption-induced lysis by phage binding without DNA injection (57, 58).

Plaque assays with a motA mutant of the propagating strain C. jejuni NCTC12658 resulted in a 3- to 4-log reduction of EOP compared to the wild type; thus, the ability to infect is also severely reduced in C. jejuni NCTC12658 lacking motA. Furthermore, plaques formed on the C. jejuni NCTC12658 motA mutant were smaller than plaques on the corresponding wild type, supporting the altered infectivity (Table 4). Altogether, we showed that phage F341 is dependent on flagellum-driven motility for efficient infection of C. jejuni NCTC12658 and requires motile flagella in order to infect C. jejuni NCTC11168.

The requirement of functional flagella for infection is linked to adsorption of phage F341.

To further investigate the dependency on motile flagella for phage F341 infection, we analyzed the adsorption of phages to C. jejuni NCTC11168 wild type and mutants. This was determined by examining and comparing the percentages of free phages not bound to bacterial cells over a 90-min period. For comparison, the propagating strain C. jejuni NCTC12658 was included. In accordance with the differences in the plaque sizes and quantity, we found that phage F341 adsorbed with increased efficiency to the propagating strain NCTC12658 compared to C. jejuni NCTC11168 (Fig. 2). Moreover, phage F341 showed distinctly increased adsorption to the C. jejuni NCTC11168 kpsM mutant compared to the corresponding wild type, as the percentage of free phages was lower for the kpsM mutant. We could not detect any significant phage adsorption to the flaAB or motA mutants (Fig. 2), which again is consistent with the plaque assay data. Finally, phage F341 showed significant adsorption to C. jejuni double mutants (kpsMΔflaAB and kpsMΔmotA), although with lower efficiency than that for the kpsM mutant (Fig. 2), suggesting that phage F341 is able to adsorb to C. jejuni in the absence of functional flagella, but only if the capsule is also lacking. These data support our hypothesis that many phage particles adsorbing to C. jejuni double mutants (kpsMΔflaAB or kpsMΔmotA) at the same time, but without being able to infect the cells, cause lysis from without. Collectively, our data show that phage F341 needs motile flagella for efficient infection and that the requirement for motility is associated with phage adsorption, suggesting that phage F341 uses flagella as a receptor.

FIG 2.

Phage adsorption illustrated as percentage of free phages not adsorbed to C. jejuni. Results are calculated from 3 individual experiments, where phage titers are normalized to values at time zero and the percentages of free phages at the remaining time points are shown as the mean percentages of free phages and standard deviations thereof. Asterisks indicate statistically significant differences (P < 0.05) in phage adsorption compared to that at 0 min. wt, wild type.

Role of flagellar glycans for phage F341 infection.

C. jejuni flagella are highly decorated by carbohydrates, and each flagellin subunit is posttranslationally modified by O-linked glycosylations at up to 19 different serine or threonine residues (24, 59). In addition, different O-linked flagellar carbohydrates such as pseudaminic acid (Pse5Ac7Ac) and legionaminic acid (Leg5Ac7Ac) and derivatives thereof have been identified so far in C. jejuni NCTC11168 (25, 60). Previous studies of mutants in central genes of the O-linked glycosylation locus such as the pseB gene, required for synthesis of Pse5Ac7Ac and derivatives thereof, suggest that O-linked glycosylation may be required for formation of functional flagella and thus motility (23, 61, 62). Here, we speculated that phage F341 may interact with these glycans during binding to the flagellum. However, due to the requirement of functional flagella for infection, our mutant analysis is limited to genes with known function in the O-linked glycosylation pathway that still produce functional flagella in C. jejuni NCTC11168. Based on this criterion, we tested if phage F341 was able to form plaques on lawns of C. jejuni NCTC11168 cj1295, cj1316c, and cj1324 mutants that still are motile but lack specific flagellar glycans. The cj1295 mutant lacks the di-O-methyl glyceric acid derivative of Pse5Ac7Ac, and the cj1316c mutant flagellum lacks Pse5Ac7Am, while the cj1324 mutant lacks the derivatives Leg5Am7Ac, Leg5AmNMe7Ac, and Pse5Ac7Am (25, 63–65). Interestingly, we found that phage F341 could infect these mutants as efficiently as the wild-type strain (Table 3), hence showing that binding of phage F341 to C. jejuni is not dependent on Leg5Am7Ac, Leg5AmNMe7Ac, Pse5Ac7Am, or the di-O-methyl glyceric acid derivative of Pse5Ac7Ac.

Microscopic analysis of phage F341 interacting with C. jejuni NCTC11168.

To study the interaction between phage F341 and C. jejuni cells in more detail, we used confocal fluorescence microscopy. C. jejuni cells were stained by membrane lipid dye FM 4-64 (emission maxima, 640 nm), and phage DNA was stained using SYBR gold (emission maxima, 537 nm), allowing separation of the two fluorescent colors in the microscope. Since C. jejuni is highly motile, we developed a method to mount live C. jejuni cells to the object glass by using 2% agarose. Others have observed free phage particles appear as small dots in the microscope, while phage DNA injected into the cell leads to release of the DNA-bound fluorophore (38, 66). In the absence of phage particles, only red cells and no green fluorescent cells were observed (Fig. 3A), while coincubation of phages and bacteria for 30 min resulted in phage DNA injection seen as green fluorescence inside red C. jejuni cells (Fig. 3B). In contrast, phage F341 appeared to stick to the surface of the kpsMΔmotA double mutant, as green fluorescence covered the entire cell surface, not leaving any red fluorescent bacterial membrane visible (Fig. 3C), confirming our hypothesis that phage F341 binding to the double mutants causes lysis from without. Also, we observed green fluorescence concentrated at or near the cellular poles of C. jejuni NCTC11168 2 min after mixing the bacteria and phages, but due to rapid diffusion of the fluorescence inside the cell, it could not be captured in the pictures. However, the colocation of phages with the polar flagella is in agreement with the phage plaque assays and adsorption assays that point to the flagellar filaments as the phage receptor.

FIG 3.

Confocal fluorescence microscopy of phage F341 infection of C. jejuni NCTC11168 as overlays of emission filters of 505 to 545 nm and 620 to 660 nm. (A) C. jejuni NCTC11168 without phages. No green fluorescence is observed. (B) C. jejuni NCTC11168 20 to 30 min post-infection with phage F341. Phage infection is observed as intracellular green fluorescence. (C) C. jejuni NCTC11168 ΔmotA plus ΔkpsM double mutant 2 min post-infection with phage F341. Green fluorescence covers the bacterium, indicating massive phage adsorption.

Visualization of phage adsorption to the flagellum.

To further investigate the phage-Campylobacter interaction, we performed transmission electron microscopy (TEM). Immediately after mixing phage F341 and C. jejuni NCTC11168, no interaction was observed (data not shown). However, after 10-min coincubation of phages and C. jejuni cells, we observed binding of phage F341 to the flagella of C. jejuni (Fig. 4A and B), and after 30 min, this interaction could still be observed, suggesting an ongoing adsorption of phages to C. jejuni NCTC11168 (Fig. 4C). Interestingly, the TEM analysis demonstrated that noncontracted phage F341 adsorbs to the flagellum by perpendicular attachment of the tail (Fig. 4C) and that the tail fibers seem to stabilize and participate in this interaction (Fig. 4D). We also noticed that nonadsorbed phages bound to each other in cluster-like formations in which the tail fibers seem to interact (Fig. 4E), as also observed by others (for an example, see reference 33). Moreover, nonadsorbed phages with contracted tail tips and deformed empty heads adsorbed to spherical vesicles were also seen as previously reported for C. jejuni phages (32, 33, 67) (Fig. 4E). Thus, we demonstrated that phage F341 specifically interacts with C. jejuni NCTC11168 flagella during infection and that F341 uses the phage tail fibers to establish this interaction.

FIG 4.

Transmission electron micrographs of phage F341 interaction with C. jejuni NCTC11168. (A) Ten-minute coincubation of C. jejuni NCTC11168 and phage F341. Initial phage attachment with noncontractile tail to the flagellum is visible. (B) Closeup of the square in panel A. (C) Thirty minutes post-incubation of C. jejuni NCTC11168 and phage F341. Perpendicular phage attachment to the flagellum seen near the polar region of C. jejuni. (D) Closeup of the square in panel C. Tail fibers of phage F341 attached to the flagellum are visible. Arrows point to the tail fibers binding to the flagellum. (E) Phage with contracted tail, empty heads, and a cluster of phage particles. (F) Purified F341. Bars, 100 nm.

DISCUSSION

In this study, we used phage plaque and adsorption assays and mutant constructs as well as different microscopic tools to perform a detailed characterization of bacteriophage F341 infection of C. jejuni NCTC11168. We demonstrate that F341 is a novel flagellotropic phage dependent on motility and a rotating flagellum for successful infection. Thus, C. jejuni motility and the multifunctional flagella not only are important for interaction with the human and avian hosts but also play a novel role in bacteriophage infection of this bacterium.

By use of nonmotile flaAB, motA, and flgP mutants, we demonstrated that functional flagella and motility are required for phage F341 infection of C. jejuni NCTC11168, and we propose that the flagellar filament is the receptor for phage F341. Even though loss of motility has previously been found to result in phage resistance for some Campylobacter phages, thereby suggesting these phages to be flagellotropic (35), the specific infection mechanism and receptors of these Campylobacter phages have not been elucidated. The only specific receptor known for Campylobacter phages is the O-methyl phosphoramidate moiety on the capsule of C. jejuni NCTC11168, shown to be required for infection by at least five phages (36). Campylobacter phages have been grouped into three groups (I and II [type CP220] and III [type CP81]) according to their genome size and ultrastructure (32, 33). Phages belonging to groups I and II have been proposed to be dependent on flagella for infection, whereas capsule-dependent phages are proposed to belong to group III (35, 68). As the Myoviridae phage F341 studied in this work belongs to group III Campylobacter phages (32, 43), it was expected to be dependent on the capsule for infection. However, we found that phage F341 is dependent on motile flagella for infection, hence demonstrating that such grouping does not take the entire C. jejuni phage diversity into account.

C. jejuni motility is required for colonization of the avian host and for establishment of human disease (2, 15), and this study shows that motility is furthermore required for bacteriophage F341 infection. However, not all motile C. jejuni strains are infected by phage F341 (43), and additional host factors such as receptor availability may affect phage infection. Since phage F341 adsorbs to the flagellum, the presence and nature of O-linked glycans covering the flagella may influence phage binding. O-linked glycosylation is complex and varies significantly between strains due to genetic differences, but phase-variable genes of the O-linked glycosylation locus also cause variation within a population of the same strain (25, 26, 60, 62, 64, 69, 70). Approximately 50 genes are suggested to be involved in O-linked glycosylation of C. jejuni NCTC11168 flagella (71), and the flagellar glycosylation of NCTC11168 is both more complex and variable than that of other strains (64). Here, we conclude that binding of phage F341 to C. jejuni is not dependent on particular types of O-linked glycans, while the importance of others, such as the widespread Pse5Ac7Ac, cannot be ruled out. Interestingly, O-linked glycosylation and motility have been linked in Helicobacter pylori, as flagellin of a hypermotile mutant showed increased levels of Pse5Ac7Ac modification (72). But, phage F341 does not infect the hypermotile C. jejuni 11168H (data not shown); however, the O-linked glycosylation of 11168H flagella may be altered, not supporting phage F341 binding. Alternatively, previous nuclear magnetic resonance (NMR) analysis of 11168H has shown that the capsule carries all phase-variable CPS modifications (36), most likely leading to a more dense CPS than that of NCTC11168, used in this study. Here, we observed that removing CPS enhanced plaque size and quantity; hence, a thicker capsule may have the opposite effect, thus preventing phage F341 from forming plaques on the hypermotile strain 11168H.

In this study, we observed that phage F341 infected C. jejuni NCTC11168 less efficiently than the propagating strain NCTC12658, regarding both plaque numbers and sizes (data not shown). This may be explained by phage F341 adsorbing less efficiently to C. jejuni NCTC11168 than to NCTC12658, as a reduced adsorption rate may decrease the number of infection cycles finishing within the same time period, leading to smaller plaques (57). Also, phage F341 did not form any plaques on lawns of a motA mutant of C. jejuni NCTC11168, while the same mutation of C. jejuni NCTC12658 resulted in pinpoint plaques at severely reduced efficiency (EOP, 10⁻⁴), which similarly may be caused by phage F341 adsorbing more efficiently to NCTC12658 than to NCTC11168. In addition, C. jejuni NCTC12658 is slightly more motile than NCTC11168, which may also affect infection efficiency for this flagellotropic phage. A recent study on Agrobacterium sp. strain H13-3 flagellotropic phage 7-7-1 showed that different levels of motility greatly affected phage infection (73). A point mutation causing a change from the charged Glu98 to the neutral Gln of the motA gene led to hypermotility, resulting in more efficient phage infection (73). Thus, alterations in motA sequences of NCTC11168 and NCTC12658 may affect phage adsorption, plaque sizes, and quantities. In addition, O-linked glycosylation of NCTC12658 flagella may also be different from that for NCTC11168, thus influencing the phage binding to the flagella. Future studies of the NCTC12658 genome sequence and the O-linked flagellar glycosylation may reveal those differences.

By use of confocal fluorescence microscopy and TEM, we have demonstrated that phage F341 interacts with the flagella of C. jejuni NCTC11168. Most studies of flagellotropic phages propose that the flagellum helps the attached phages to reach the basal body of the bacterium where DNA injection occurs (40–42, 73), as phages with heads without DNA were observed only at the bacterial body (41). Phage χ was proposed to move to the base of its host during flagellar rotation by forcing the phage to follow the grooves formed by flagellin monomers “like a nut on a bolt” (42). Moreover, the direction of flagellar rotation, counterclockwise versus clockwise, has been shown to be essential for adsorption and subsequent infection for some phages (38, 39, 42). Recently, it was speculated that bacteriophage iEPS5 infecting Salmonella injects its DNA directly into the flagellum and that the DNA is transported through the hollow flagellum by flagellar rotation (38). In theory, this may also apply for flagellotropic Campylobacter phages, as the hollow lumen in the flagellum is wide enough to contain a double-stranded DNA helix, judged from cryo-electron microscopy (cryo-EM) flagellar pictures (74). But, as we did not observe any phages without DNA bound to the flagellum, we find it unlikely that phage F341 injects its DNA into the flagellum. Furthermore, looking at the TEM pictures, it is evident that phage tail fibers attach to the flagellum, and this binding is consistently seen perpendicular to the flagellum. We therefore propose that the initial binding of phage F341 to the flagellum allows the phage to slide along the rotating flagellum, which enables the phage to bind to the bacterial surface where the DNA is then injected. As the phage does not bind to CPS, such a secondary receptor may be another surface component, or alternatively, DNA injection may occur directly at the polar region where the flagellum protrudes from the membrane (75). Also, our data suggest that binding to the flagellum is a prerequisite for DNA injection, since binding to acapsular nonmotile mutants does not result in infection. We furthermore speculate that the initial binding to the flagellum may cause a conformational change of the phage tail that enables DNA injection.

In conclusion, we have shown that phage F341 is a novel flagellotropic phage dependent on interaction with a rotating flagellum for infection of C. jejuni NCTC11168 and demonstrated the interaction of the tail fibers of a flagellotropic phage with the flagellum of Campylobacter. The detailed characterization has broadened our understanding of phage infection mechanisms in C. jejuni, and future studies of O-linked glycans as well as the genome sequence of phage F341 will allow us to further characterize the interaction between phage F341 and the C. jejuni flagellum. Interestingly, bacteriophages have evolved to target the flagellum, being the most essential structure of C. jejuni required for colonization of the chicken and establishment of human disease, and the study underlines the diverse role of flagella in C. jejuni physiology. Even though C. jejuni colonizes the chicken gut to high numbers, the viscous environment of the mucus may hinder bacteriophages from getting across to their host, and the flagellum may act as a tentacle utilized by flagellotropic phages to locate a bacterial host in this ecological niche.