Bacteriophage CP_F1 was originally isolated from a pig manure sample and was propagated on C. jejuni PT14 (Brathwaite et al., 2013). All strains, associated plasmids and oligonucleotide PCR primers used in this study are listed in Table 1. The C. jejuni strains were routinely grown on blood agar base No. 2 plates (BA plates; Oxoid) supplemented with 5% horse blood (TCS Biosciences) under microaerobic conditions in either a modular atmosphere controlled cabinet (5% CO₂, 5% O₂, 2% H₂, 88% N₂) or in anaerobic jars using gas replacement (7.3% CO₂, 5.6% O₂, 3.6% H₂, 83.5% N₂) at 42°C. Escherichia coli strain TOP10 (Invitrogen) was used for cloning and cultured aerobically at 37°C on Luria agar plates or in Luria broth containing appropriate antibiotic as required.

A determination of bacteriophage infection of the C. jejuni PT14 wild type and mutant strains was performed by employing the traditional spot test assay. A fresh overnight culture of the test Campylobacter strain was collected from a BA plate and emulsified in MH broth supplemented with 10 mM MgSO₄ (Thermo Fisher) to a cell density of approximately 10⁷ CFU/ml. A volume of 500 μl of emulsified cells was added to 5 ml of molten NZCYM (0.6%, Carl Roth) top agar and poured on top of a NZCYM agar plate. After the top agar had solidified, 5 μl aliquots of the phage suspensions, at the routine test dilution of approximately titer of 10⁷ PFU/ml (Frost et al., 1999), were dispensed on top of the soft agar layer. The plate was then incubated at 42°C under microaerobic conditions for 24–48 h after the spots dried into the agar. In a similar way the phage titer was determined by serial 10-fold dilutions of phage suspensions, which were applied as 10 μl droplets in triplicate to the surface of prepared host bacterial lawns and allowed to dry. After incubation the plaques were counted and titers calculated.

In order to assess swarming motility, fresh overnight cultures were taken from a BA plate and resuspended in MH broth. After adjustment of cell suspensions to OD₆₀₀ 0.1, 2 μl of suspension was removed with a pipette tip and stabbed into a swarming agar plate (0.4% Mueller Hinton agar). Inoculated plates were then incubated under microaerobic conditions at 42°C for 48 h. Growth zones were measured for each of triplicate samples after 24 and 48 h to assess cell motility.

The growth from C. jejuni BA plates incubated overnight at 42°C were collected and emulsified in PBS buffer (Merck Millipore) for the adjustment of inoculum concentrations. Flasks containing 50 ml nutrient broth No. 2 (Oxoid) were inoculated with C. jejuni cells at starting concentrations of approximately 10⁵ or 10⁷ CFU/ml. After inoculation, flasks were sealed with cotton stoppers and placed in anaerobic jars (Oxoid). Microaerobic conditions were introduced using the gas replacement method (John et al., 2011). Cultures were then incubated in a shaking incubator at 42°C and 100 rpm. After 2 h incubation, bacteriophage CP_F1 was added at an MOI of 2. Samples were collected at 2 h intervals for determination of viable cell counts and phage titers using the Miles and Misra method. Phage samples were treated with 2 μl chloroform (Thermofisher) and centrifuged at 13,000 × g for 5 min. The supernatant was removed and used for titration. The C. jejuni PT14 wild type and knock-out strains were tested in triplicate to ensure statistical certainty. Growth controls without phage addition served as references.

Phage adsorption rates were determined as described previously (Siringan et al., 2014) with minor modifications outlined as followed. Suspensions of C. jejuni PT14 wild type, flaB mutant and trans-complemented flaB cells that contained approximately 5 × 10⁷ CFU/ml, were inoculated in nutrient broth No. 2 and incubated at 42°C under microaerobic conditions at 100 rpm for 5 h to obtain cells in the exponential growth state. The actual viable count was determined following serial dilution and incubation as described above. For the experiment, bacteriophage CP_F1 was diluted and added to the suspensions to give a final titer of 10⁵ PFU/ml, then briefly mixed and kept static at 42°C under microaerobic conditions. Sampling was performed every 5 min over a period of 30 min. Samples were immediately centrifuged at 13 000 × g for 5 min and the supernatants removed. The titer of free bacteriophages in the supernatants was determined and used to calculate numbers of bound bacteriophages. Bacteriophage adsorption constants were determined using the formula k = -ln (P_t/P₀)/Nt, where P_t = phage titer at time point t (PFU/ml), P₀ = initial phage titer (PFU/ml), N = bacterial density (CFU/ml) and t = time (min).

The efficiency of plating (EOP) of bacteriophage CP_F1 infecting the PT14 mutant strains was determined by enumerating bacteriophages as described above. Subsequently the calculations were made by dividing the bacteriophage titer obtained when applied to the lawns prepared from different mutant strains by the titer obtained when applied to C. jejuni PT14 wild type lawns.

In order to assess the burst size and latent period for bacteriophage CP_F1 infection of C. jejuni PT14 wild type, the flaB mutant and trans-complemented derivatives, single step growth curves were monitored over 4 h. The protocol after Carvalho et al. (2010) was employed for this purpose. Triplicate samples of 10 ml nutrient broth No. 2 were inoculated with 10⁷ CFU/ml cells and grown under microaerobic conditions at 42°C under constant shaking at 100 rpm into early exponential phase (approx. 10⁸ CFU/ml). Bacteriophage CP_F1 was added at an MOI of 0.001. Samples for titer determination were taken every 15 min for 4 h. These were centrifuged at 13,000 × g for 5 min and the supernatants used for titration. Three replicates of each individual experiment were performed and mean values used for presentation of titer development. Non-linear regression was used to determine latent period and burst size of the first burst event.

Genomic DNAs from C. jejuni PT14 non-motile variants were prepared using the GenElute Bacterial Genomic DNA Kit (Sigma Aldrich) following the manufacturer's instructions. DNA sequencing was performed using the Illumina MiSeq platform. The data consisted of 3.1–4.4 million paired-end sequence reads of 250 bp in length. Initial processing of the raw data, mapping of the sequence reads to C. jejuni PT14 (GenBank accession number CP003871) and variant detection were performed using CLC Genomics Workbench version 8.0 (Qiagen, Aarhus, Denmark).

Total RNAs were extracted from CP_F1 infected and uninfected control cultures of C. jejuni PT14. For each treatment, three independent early-log phase cultures (eclipse phase of the infected cultures) growing in nutrient broth No. 2 were harvested 50 min after phage addition and the RNA content extracted using TRIzol Max with Max Bacterial Enhancement Reagent (Invitrogen) according to the manufacturer's protocol. Subsequently ethanol-precipitation and purification using the RNeasy Mini kit (Qiagen, Crawley, UK) according to the manufacturer's instructions were performed including DNase treatment using RNase-free DNase and related reagents (Qiagen). Pure RNA samples were collected in 40 μl of RNase free water and analyzed for quantity/purity (Nanodrop ND1000) and quality (Bioanalyser 2100, Agilent Technologies Inc.). Prokaryote Total RNA Nano series II software, Version 2.3 was used for analysis of the RNA quality. All purified RNA samples showed a RNA Integrity Number (RIN) of 10.0.

Total RNAs were converted to cDNA with random hexamer primers using the Superscript II (Fisher Scientific) reverse transcriptase system according to the manufacturer's protocol. Specific primers for qRT PCR were designed with lengths of 18–24 nt and can be found in Supplementary Table S1. An optical 48 well microtiter plate (Applied Biosystems) was used with 20 μl reaction volumes consisting of Power SYBR Green PCR master mix (Life Technologies), 50 nM gene specific primers and 100 ng of the cDNA template. A StepOne real time PCR system (Applied Biosystems) was programmed for an initial set up of 30 s at 95°C, followed by 40 cycles of 15 s at 95°C and 1 min at 58°C. A melting curve was obtained from 50 to 95°C to control specificities of quantitative PCR reaction for each primer pair. Cycle threshold (CT) values were determined employing the StepOne software version 2.0 (Applied Biosystems). The comparative threshold cycle method was used to calculate change (n-fold) where samples were normalized to the internal control product of the 50 S ribosomal subunit protein L1 gene (rplA), which showed no change in expression levels between phage infected and uninfected cultures of C. jejuni PT14 during acute CP8 infection in previous experiments. Reactions were performed in triplicate. The fold changes were calculated using the 2^−ΔΔCt method. Verification of acute phage infection was confirmed by PCR using synthesized cDNA, employing specific primers targeting the gene for the major capsid protein (mcp) of bacteriophage CP_F1 (homolog of gp23 from phage NCTC12673; Kropinski et al., 2011).

Statistical differences between paired control and treatment groups were assessed by using the Student's t-test with significance p < 0.05. Differences between experimental groups were analyzed by analysis of variance. Viable count and phage titer data were log₁₀-transformed for analysis.

By screening bacteriophage infection of a number of mutant variants of C. jejuni PT14 we have analyzed the effects of flagella related factors on this process. Spot test assays showed a clear dependence on rotating flagella for bacteriophage CP220 (prototype of the Cp220likevirus genus), as no replication and host lysis was observed for flaA or flaAB mutants (Supplementary Figure S1). Similarly, the CP8unalikevirus phage CP_F1 showed decreased infection efficiency on non-motile cells, as infection of flaA and flaAB mutant strains yielded reduced efficiency of plating (EOP) values (flaA 0.08 ± 0.04, n = 3; flaAB 0.12 ± 0.03, n = 3) and highly opaque plaques. Additionally we observed CP_F1 to be strictly dependent on capsular polysaccharide (CPS) as no plaque formation occurred on a kpsM mutant (Figure 1). Interestingly, we found an aberration in infection by bacteriophage CP_F1 of a flaB disruption mutant. CP_F1 was isolated from a pig sample and propagated on strain PT14. It is able to infect and propagate on its host, but yields opaque plaques on soft agar bacterial lawns, which reveal regrowth of the host after infection. We observed similar reactions for phage CP220 (Supplementary Figure S1), suggesting that this reaction is not limited to one class of Campylobacter phage (Javed et al., 2014). However, propagation on a flaB mutant of PT14 produced clear lysis zones that persisted even after prolonged incubation for 24 h (Figure 1). Five individual clones, which all carried the desired disruption of flaB were tested in spot test assays to exclude effects of secondary mutations (de Vries et al., 2015). All showed identical phenotypes toward phage-induced lysis on soft agar lawns and motility (Supplementary Figure S2).

DNA sequencing of PCR amplicons of all five clones carrying the desired gene disruption, revealed an integration of the resistance marker in the flaB gene in the orientation of the flaB reading frame, mitigating any polar effects on the adjacent flaA gene. We also detected no mutations in the flaA gene, as further supported by findings from motility tests. Analysis of the motility of the flaB mutant strains revealed no major reduction in swarming motility (Figure 2), which is in accordance with previous observations (Wassenaar et al., 1991; de Vries et al., 2015).

To analyze the regrowth of Campylobacter cells in lysis zones we collected and sub-cultured surviving cells from the surface of turbid spots. Single colony isolates of these were examined for motility, reaction to phage infection by CP_F1 and DNA sequence analysis of the flaA/B region. We found that the majority of clones (78%, n = 100) showed either complete loss of motility or attenuation of their swarming ability (Supplementary Table S2), which was accompanied by resistance to reinfection by phage CP_F1. Additionally, eight clones were observed to have shifted into the carrier state (Siringan et al., 2014), as demonstrated by plaque formation inside of the growth zone on the motility agar plates and detectable phage propagation after serial sub-culturing of these clones (Supplementary Figure S3). PCR amplification and DNA sequence analysis of the flaA gene of the non-motile isolates revealed no alteration in the gene or non-coding sequences. The non-motile isolates were serially sub-cultured and after each passage the motility and phage resistance of the culture were assessed by inoculation of swarming motility agar plates and plaque assay. After two rounds it was observed that cultures arising from a single clone had recovered motility. In one case the process of escape from bacteriophage infection was reversible, which could be attributable to phase variation. Whole genome sequencing of three non-motile clones exhibiting phage resistance revealed all to have suffered a single nucleotide deletion in the N-terminal region of the rpoN gene coding for the RNA polymerase factor σ⁵⁴ (A911_03265) at nucleotide position 244–250 in a stretch of 7 adenine residues. Further several phase variable genes were shifted to the off state compared to the wild type genome sequences (Table 2). For the flaB mutant, the shift to a phage resistant population could not be observed, which we assume is due to an increased susceptibility of the mutant to phage infection that may also be accompanied by an inability to support the mutation(s) leading to phage resistance.

In order to verify that the observed increase in susceptibility toward phage infection was a result of the disruption of the flaB gene, we constructed a plasmid vector for the introduction of an intact copy of flaB into the pseudogene A911_00230 of the mutant strain. This construct was used to transform the flaB mutant strain and the correct insertion verified by DNA sequencing the target region. Trans-complementation of the disrupted flaB gene led to partial restoration of the phenotype with respect to bacteriophage infection. Overgrowth was restored within phage lysis zones but was less turbid than the wild type. The swarming motility of the complemented strain remained comparable to wild type but the survivors of phage infection were compromised in motility (Figure 2).

In order to analyze the changes in phage susceptibility in greater detail, phage replication experiments were performed in 50 ml liquid medium cultures, using either wild type or the flaB mutant or the flaB trans-complemented strain of C. jejuni PT14 over a period of 24 h (Figure 3). We tested the effects on phage infection at two different host cell densities, above and below the phage proliferation threshold of log₁₀ 7 CFU/ml, which represents the density of bacteria required for the productive replication of bacteriophage (Cairns et al., 2009). At higher cell densities CP_F1 infection of wild type, flaB mutant and flaB complement cultures resulted in an initial fall of approximately 1 log₁₀ CFU/ml in the viable count. This event was followed by a recovery in the viable count of wild type cultures (Figure 3A). In contrast the flaB mutant exhibited a drastic reduction in the growth rate (μ) post the phage-induced crash in viable count (Figure 3B). The growth rates post population crash for infected wild type μ = 0.67 ± 0.02/h and infected flaB mutant μ = 0.31 ± 0.05/h cultures (p < 0.01). The flaB complement culture also recovered faster than the mutant (Figure 3C) exhibiting a significantly greater growth rate of μ = 0.44 ± 0.06/h (p < 0.05). Phage proliferation at high initial cell densities showed significant differences between wild type and flaB phage-infected cultures with significantly greater phage yields at 24 h (p < 0.01).

Infection of wild type C. jejuni PT14 at a bacterial density below the phage proliferation threshold with CP_F1 at an MOI of 2 resulted in minimal differences in the viable bacterial counts between infected and control cultures (Figure 3D). In contrast, the viable count for flaB mutant cultures remained static for a period of 4 h, post phage infection but subsequently these cultures entered exponential growth (Figure 3E). The growth rates of the infected cultures post delay were not significantly different to the uninfected control cultures (control: μ = 0.62 ± 0.10/h; infected: μ = 0.74 ± 0.06/h) suggesting under these conditions there was no fitness disadvantage for the emergent phage resistant sub-population. The delay evident in the phage infected flaB mutant was not present in trans-complemented cultures, which behaved similar to wild type with similar growth rates to the uninfected control at early stages of growth (Figure 3F). Analysis of phage propagation showed no increase in phage numbers during the first 12 h of incubation. However, after 24 h an increase in free virions was detected, implying phage replication. There was a significant difference in the phage titer of approximately 1 log₁₀ PFU/ml (p < 0.001) between infected wild type cultures and the infected flagellin B mutant strain after 24 h (Supplementary Figure S4). Further analyses of the phage propagation parameters were examined using a one-step growth experiment. The latent period for CP_F1 was determined as 90 min with two burst events, marked B1 (90–105 min) and B2 (180–210 min) in Figure 4, evident over the 240 min period. The latent period for the CP_F1 infection process was in a comparable range to other studies (Cairns et al., 2009; Carvalho et al., 2010). The number of phage particles liberated per cell showed an increase for CP_F1 infecting the flaB mutant strain relative to the wild type and flaB complement. A burst size of 1.4 ± 0.4 PFU/cell in the wild type and 4.2 ± 1.2 PFU/cell in the mutant strain were calculated for the first burst event. Differences in the phage yields between the phage infected cultures were statistically significant (p < 0.05). Further, the phage yields were magnified after a second burst event. This resulted in a difference of 1 log₁₀ in PFU between mutant culture and the wild type or the flaB complement at 225 min after the initial addition of the phage. However, the significant change in the replication process was not accompanied by a major difference in the relative efficiency of plating of phage CP_F1 between infection of the PT14 wild type and the flagellin B mutant strain (1.59 ± 0.38, n = 3).

To test if any changes in flagellin gene expression were induced as a response to acute phage infection we performed quantitative RT-PCR analyses. No alterations in the transcript levels of flaA or flaB in wild type PT14 were observed during the eclipse period of phage infection in which phage DNA replication and capsid assembly occurs prior to host lysis. A time point of 50 min post phage addition was selected for transcriptional analysis, since one step growth experiments revealed a latent period of approximately 90 min for CP_F1 infection. Similarly, no change in the expression of flaA was observed in the flaB knock-out strain (Supplementary Figures. S5, S6). These data would suggest that the increase in bacteriophage sensitivity or phage yields of flaB mutants are not linked to a compensatory increase in flaA transcription.

As changes in susceptibility toward phage infection and burst size were observed in the mutant strain, we were interested to examine if a change in adsorption might also be an effect of flaB inactivation. The calculated adsorption constant of CP_F1 binding to the PT14 wild type was in a similar range to that of the Cp8unalikevirus phages CP30 and CP8 (Siringan et al., 2014). However, there were 2-fold and greater increases in the adsorption constants (k) determined for the flaB mutant clones after 30 min incubation (Table 3). Differences in the adsorption constant between the wild type and the mutant were statistically significant for five independent clones based on triplicate experiments (p < 0.01), verifying that the absence of FlaB has an effect on the adsorption process.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.