Prevalence, Development, and Molecular Mechanisms of Bacteriocin Resistance in Campylobacter

Abstract

Bacteriocins (BCNs) are antimicrobial peptides produced by bacteria with narrow or broad spectra of antimicrobial activity. Recently, several unique anti-Campylobacter BCNs have been identified from commensal bacteria isolated from chicken intestines. These BCNs dramatically reduced C. jejuni colonization in poultry and are being directed toward on-farm control of Campylobacter. However, no information concerning prevalence, development, and mechanisms of BCN resistance in Campylobacter exists. In this study, susceptibilities of 137 C. jejuni isolates and 20 C. coli isolates to the anti-Campylobacter BCNs OR-7 and E-760 were examined. Only one C. coli strain displayed resistance to the BCNs (MIC, 64 μg/ml), while others were susceptible, with MICs ranging from 0.25 to 4 μg/ml. The C. coli mutants resistant to BCN OR-7 also were obtained by in vitro selection, but all displayed only low-level resistance to OR-7 (MIC, 8 to 16 μg/ml). The acquired BCN resistance in C. coli could be transferred at intra- and interspecies levels among Campylobacter strains by biphasic natural transformation. Genomic examination of the OR-7-resistant mutants by using DNA microarray and random transposon mutagenesis revealed that the multidrug efflux pump CmeABC contributes to both intrinsic resistance and acquired resistance to the BCNs. Altogether, this study represents the first report of and a major step forward in understanding BCN resistance in Campylobacter, which will facilitate the development of effective BCN-based strategies to reduce the Campylobacter loads in poultry.

Campylobacter species, including C. jejuni and C. coli, represent one of most common bacterial causes of human gastroenteritis in the United States (36). Human Campylobacter illnesses are caused primarily by C. jejuni (∼90%) and secondarily by C. coli (∼10%). This group of pathogenic organisms causes watery diarrhea and/or hemorrhagic colitis in humans and is associated with Guillain-Barré syndrome, an acute flaccid paralysis that may lead to respiratory muscle compromise and death (29). There are estimated to be more than two million cases of campylobacteriosis in the United States each year, and the annual medical and productivity costs resulting from Campylobacter infection are estimated at 1.5 billion to 8.0 billion dollars (4). Poultry, particularly broiler chickens, are considered a major source of human campylobacteriosis (17). Thus, on-farm control of Campylobacter in poultry would reduce the risk of human exposure to this pathogen and have a significant impact on food safety and public health (19). In particular, resistance of Campylobacter to clinical antibiotics, including fluoroquinolones and macrolides, the major drugs of choice for treating human campylobacteriosis, raises an urgent need for novel strategies to prevent and control Campylobacter colonization in poultry (19, 27).

To date, three general strategies to control Campylobacter in poultry on the poultry farm have been proposed, and these include (i) reduction of environmental exposure (biosecurity measures), (ii) an increase in poultry's host resistance to reduce Campylobacter carriage in the gut (e.g., competitive exclusion, vaccination, and host genetics selection), and (iii) the use of antimicrobial alternatives to reduce and even eliminate Campylobacter from colonized chickens (e.g., bacteriophage therapy, reviewed by Lin [19]). However, effective implementation of biosecurity measures relies on a better understanding of risk factors and sources of Campylobacter for poultry (43). In contrast to biosecurity measures, the other two general intervention approaches are currently not commercially available and are still under development. Notably, recent breakthroughs in the discovery and characterization of potent anti-Campylobacter bacteriocins (BCNs) may lead to an effective measure for on-farm control of Campylobacter in poultry (19). BCNs are short antimicrobial peptides (AMPs) produced and exported by most bacterial species examined to date for the apparent purpose of destroying their competitors (35). Many BCN-producing bacteria (e.g., lactic acid bacteria) are commensals in the intestine. Therefore, the intestinal BCN-producing bacteria may achieve competitive advantage and function as an innate barrier against pathogens in the host. The natural and low-toxicity BCNs have been proposed as promising candidates for novel antimicrobials (5). Several anti-Campylobacter BCNs produced by chicken commensal bacteria, such as OR-7 from Lactobacillus salivarius (38), E-760 and E50-52 from Enterococcus faecium (25, 40), and bacillocins from Paenibacillus polymyxa (41), have displayed potent killing effects in vitro. Oral administration of these BCNs dramatically reduced C. jejuni colonization in poultry intestines (5 to 8 log₁₀ CFU reductions). Thus, these natural anti-Campylobacter BCNs have been proposed as effective alternatives to therapeutic antibiotics and are being developed for on-farm control of Campylobacter (25, 38, 40, 41).

Although the anti-Campylobacter BCNs described above have been demonstrated to be very effective in reducing C. jejuni colonization in poultry, several critical issues (e.g., production and resistance development) need to be addressed for future regulatory approval and public acceptability of this intervention measure. In this study, we examined prevalence, development, and molecular mechanisms of BCN resistance in Campylobacter by using molecular and genomic approaches. Our findings strongly suggest that there is limited likelihood in the way of resistance development for the anti-Campylobacter BCNs, such as OR-7 and E-760. Microarray and random transposon mutagenesis studies indicated that the multidrug efflux pump CmeABC contributes to intrinsic resistance of Campylobacter to the BCNs. In addition, the CmeABC efflux pump also works synergistically with other BCN resistance mechanisms to confer acquired BCN resistance in Campylobacter.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

The major bacterial strains and plasmids used in this study are listed in Table 1. Among 157 strains used for the prevalence survey in this study, 137 isolates were C. jejuni, while 20 isolates were C. coli. These Campylobacter strains were isolated from different hosts, including humans (15), bovines (5), chickens (121), turkeys (1), and pigs (4), as well as from the environment, including trapped mice (5), bird droppings (5), and lagoons (1). All of these strains were collected from 16 geographically diverse areas in the United States. These C. jejuni and C. coli strains were routinely cultured in Mueller-Hinton (MH) broth (Difco) or on MH agar at 42°C under microaerobic conditions, which were generated by a CampyGen Plus (Oxoid) gas pack in an enclosed jar. When needed, MH media were supplemented with kanamycin (Kan), chloramphenicol (Cm), or BCNs at the specified concentrations. Escherichia coli cells were grown at 37°C in Luria-Bertani (LB) medium with agitation and supplemented with 30 μg/ml of Kan or 20 μg/ml of Cm.

TABLE 1.

Major bacterial strains and plasmids used in this study

Strain or plasmid	Description	Source or reference
Strains
Campylobacter spp.
JL106	C. coli strain isolated from a human; OR-7 MIC = 64 μg/ml	Humana
JL20	C. coli strain isolated from a pig; OR-7 MIC = 0.5 μg/ml	This study
JL25	C. coli strain isolated from a pig; OR-7 MIC = 0.5 μg/ml	This study
S3B	C. jejuni strain isolated from a chicken; OR-7 MIC = 0.5 μg/ml	27
JL349	JL20 derivative, BCNr mutant obtained from single-step selection in vitro using OR-7 as a selective agent; OR-7 MIC = 32 μg/ml	This study
JL241	C. jejuni NCTC 11168 isolated from a human; OR-7 MIC = 0.5 μg/ml	32
JL199	NCTC 11168 derivative, cmeB::Kan	22
JL4	NCTC 11168 derivative, cmeR::Cm	20
JL341	NCTC 11168 derivative, BCNr mutant generated by natural transformation using genomic DNA of C. coli JL106; OR-7 MIC = 8 μg/ml	This study
JL360	JL341 derivative, cmeB::Kan	This study
JL424	JL360 containing shuttle vector pCME	This study
JL416	JL341 derivative, Cj1125c::Cm	This study
JL372	JL341 derivative, Cj1687::Cm	This study
JL377	JL341derivative, Cj0035c::Cm	This study
JL28	C. jejuni 81-176 isolated from a human; OR-7 MIC = 0.5 μg/ml	3
JL3	81-176 derivative, cmeB::Kan	22
JL219	81-176 derivative, Cj0035c::Cm	10
JL358	81-176 derivative, BCNr mutant generated by natural transformation using JL349 genomic DNA; OR-7 MIC = 16 μg/ml	This study
K15A2	JL358 perR::Kan; OR-7 MIC = 4 μg/ml	This study
JL412	K15A2/pPerR	This study
E. coli
DH5α	F− φ80lacZΔM15 Δ(lacZYA-argF)U169 recA1 endA1 hsdR17 (rK−, mK+) phoA supE44 thi-1 gyrA96 relA1 λ−	Invitrogen
JL402	E. coli DH5α containing pPerR	This study
JL48	Conjugation helper strain, DH5α containing plasmid RK2013	1
Plasmids
pGEM-T Easy	PCR cloning vector, Ampr	Promega
pCj1125c	pGEM-T Easy containing 1.7-kb Cj1125c gene of JL241	This study
pcmCj1125c	pCj1125c with Cm resistance gene inserted in Cj1125c gene	This study
pCj1687	pGEM-T Easy containing 2.0-kb Cj1687 gene of JL241	This study
pcmCj1687	pCj1687 with Cm resistance gene inserted in Cj1687 gene	This study
pRY111	E. coli-C. jejuni shuttle vector, Cmr	22
pPerR	pRY111 derivative containing a 1.45-kb perR gene plus its promoter region	This study
pCME	pUOA18 shuttle vector derivative containing a wild-type cmeABC operon	23
pRK2013	IncP Tra RK2+ ΔrepRK2 repE1+ Kanr	7

^aIsolated from human feces and kindly provided by Qijing Zhang (Iowa State University).

Bacteriocins.

The BCNs OR-7 and E-760 were purified from L. salivarius NRRL B-30514 and E. faecium NRRL B-30745, respectively, as described in recent publications (25, 38). These BCNs were dissolved in sterile H₂O and stored at −20°C prior to use. The amino acid sequences of OR-7 (55 amino acid [aa] residues) and E-760 (62 aa residues) BCNs were consistent with class IIa bacteriocins based on the N-terminal regions of the peptides (25, 38). Both OR-7 and E-760 were also resistant to high temperature (e.g., 90°C for 15 min for OR-7) and a wide pH range (e.g., 3.0 to 9.1 for OR-7) (25, 38).

BCN susceptibility testing.

The susceptibilities of C. jejuni and C. coli isolates to BCNs OR-7 and E-760 were determined by a standard microtiter broth dilution method with an inoculum of 10⁶ bacterial cells/ml as described previously (22). MICs were determined by the lowest concentration of specific BCN showing complete inhibition of bacterial growth after 2 days of incubation at 42°C.

In vitro selection of BCN^r C. jejuni and C. coli.

BCN OR-7 was used as the selective agent to obtain spontaneous BCN-resistant (BCN^r) mutants in vitro. Briefly, three C. jejuni strains (NCTC 11168, 81-176, and S3B) and two C. coli isolates (JL20 and JL25) were grown in BCN-free MH broth to the late log phase. The cultures were centrifuged, and the pellets were suspended in MH broth to a final concentration of approximately 10¹⁰ cells/ml. The cell suspensions were plated in triplicate on MH agar plates containing 4, 8, or 16 μg/ml of OR-7. After 2 days of incubation, the OR-7-resistant (OR-7^r) colonies were enumerated, and we calculated the frequency of the emergence of BCN resistance, i.e., the ratio of CFU on BCN-containing plates to the CFU on BCN-free MH agar plates. The single-step in vitro-selected OR-7^r mutants were randomly selected and used for MIC testing together with the corresponding parent strain. The experiment was repeated twice with triplicate measurements within each single independent experiment.

DNA isolation and natural transformation.

Chromosomal DNA was isolated from Campylobacter by using the Wizard genomic purification kit (Promega) according to the manufacturer's instructions. Natural transformation (biphasic method) was performed following a standard procedure (6). The natural transformation efficiency was expressed as transformants per μg DNA per recipient CFU. The OR-7 resistance trait in JL106 (a naturally OR-7-resistant C. coli strain; Table 1) and JL349 (an in vitro-selected OR-7-resistant C. coli mutant obtained in this study; Table 1) were transferred to C. jejuni NCTC 11168 and 81-176 by natural transformation, respectively, creating OR-7^r C. jejuni mutants JL341 and JL358 (Table 1). The insertional mutation of cmeB was transferred to the NCTC 11168-derived OR-7-resistant mutant (JL341) by natural transformation, generating mutant JL360 (Table 1).

Whole-genome microarray analysis.

We used microarray techniques to compare the transcriptome of wild-type C. jejuni NCTC 11168 with that of its OR-7^r derivative, JL341 (OR-7 MIC = 8 μg/ml) (Table 1). The microarray glass slides (C. jejuni OciChip) were purchased from Ocimum Biosolutions (Indiana). The bacterial RNA isolation, cDNA synthesis and labeling, microarray hybridization, and data collection and analysis were detailed in our previous publication (13). Notably, false-discovery rate (FDR) multiple-testing adjustment was applied to the microarray probabilities. The hybridization experiments were repeated using RNA isolated from five independently grown cultures. In this study, we chose a P value of <0.05 and a ≥2.0-fold change as the cutoff for significant differential expression between C. jejuni NCTC 11168 and its derivative BCN^r mutant JL341.

In vivo random transposon mutagenesis.

The BCN^r C. jejuni JL358, an 81-176 derivative (Table 1), was subjected to in vivo transposon mutagenesis by using EZ::Tn5 <KAN-2> transposome (Epicentre) as detailed in our previous publication (19). Briefly, one microliter of EZ::Tn5 <KAN-2> Transposome was used to electroporate C. jejuni JL358 competent cells. The Kan^r transformants were individually picked and inoculated in 96-well microplates. Following 2 days of incubation, cultures of mutants were replicated into microtiter plates containing 4 μg/ml of OR-7 (4-fold reduction of MIC of the parent strain). Those mutants that could not grow in OR-7-containing media were selected from the initial plates and subjected to a second screening to confirm increased sensitivities of the mutants to OR-7. To confirm specific genetic linkage between the transposon insertion and the increased BCN susceptibility of each mutant, backcrossing of the transposon mutations into the parent strain was performed using natural transformation (6). The MICs of BCN for the backcrossed mutants together with parent strain C. jejuni JL358 were determined. The specific transposon insertion site of each mutant was determined by directly sequencing the genomic DNA (24). Sequence analysis was performed using the DNAStar software package.

PCR and real-time qRT-PCR.

Key PCR primers used in this study are listed in Table 2. PCR was performed in a 50-μl mixture containing each deoxynucleoside triphosphate at a concentration of 200 nM, each primer at a concentration of 200 nM, 2.0 nM MgCl₂, 50 ng of C. jejuni genomic DNA, and 2.5 U of Platinum Taq DNA polymerase (Invitrogen). Real-time quantitative reverse transcription-PCR (qRT-PCR) was performed as described previously (13) by using gene-specific primers. Cycling conditions varied according to the estimated annealing temperature of the primers and the expected sizes of the products (available upon request).

TABLE 2.

Key oligonucleotide primers used in this study

Primer	DNA sequence (5′-3′)a	Product size (bp)b	Gene amplifiedb
Cj1125cF	GCCCGCTAGAATGTCTTTGA	1,733	Cj1125c
Cj1125cR	ATCTAACCCGGGACGATTTT
Cj1116cF	GGAACTCATTGATGAAATGCAA	1,982	Cj1116c
Cj1116cR	CCCTACCATCTATAGGTGCAAAA
CmAfeI-F	GCGAGCGCTTGCTCGGCGGTGTTCCTTT	811	cat
CmAfeI-R	GCGAGCGCTGCGCCCTTTAGTTCCTAAAG
Cj1687F1	TCTTTGGCATCTTTGGCTTT	2,000	Cj1687
Cj1687R1	TGCGATTTTGATGTTTCC
Cj0630cF	CAACGAAAAACAAAGCAA	1,350	Cj0630c
Cj0630cR	TGTTTTTAAGTTCTTCGATTTTTGC
PerRF2	AAACAAGTAAGGTGGAA	1,662	Cj0032 (perR)
PerRR2	AGTGCAATCAGATAGTAAA

^aThe underlined and italic sections within the primer sequences represent the AfeI restriction sites.

^bProduct sizes and amplified genes refer to those of the relevant primer pairs.

Isogenic mutant construction and complementation in trans.

To construct an isogenic Cj0035c (a putative efflux pump gene) mutant of NCTC 11168, genomic DNA of an isogenic Cj0035c mutant of 81-176 (JL219) (Table 1) was extracted and used for natural transformation with NCTC 11168 (JL341) as a host strain, creating mutant JL377 with insertional inactivation in Cj0035c.

Isogenic Cj1125c and Cj1687 mutants of NCTC 11168 were constructed by insertional mutagenesis as described in a previous publication (20). Briefly, to construct an isogenic Cj1125c mutant of NCTC 11168, an approximately 1.7-kb fragment was PCR amplified from genomic DNA of JL341 by using primer pair Cj1125cF and Cj1125cR (Table 2). The PCR product was cloned into the pGEM-T Easy vector (Promega), resulting in construct pCj1125c. The Cm resistance gene cassette was PCR amplified from plasmid pUOA18 (45) by using Pfu polymerase (Stratagene) and primers CmAfeI-F and CmAfeI-R (Table 2). The resulting blunt-ended PCR product was purified and ligated into the pCj1125c vector, which was digested with SwaI prior to ligation, to generate mutant construct pcmCj1125c. The construct pcmCj1125c, which serves as a suicide vector, was introduced into NCTC 11168 by natural transformation (44). One transformant, designated JL416, was selected on an MH agar plate containing 5 μg/ml of chloramphenicol. Similarly, Cj1687 mutant JL372 was constructed in an NCTC 11168 background by using primers Cj1687-F and Cj1687-R (Table 2). Insertional mutagenesis of Cj1125 and Cj1687 was confirmed by PCR using the specific primer pairs Cj1125cF/Cj1125cR and Cj1687-F/Cj1687-R (Table 2), respectively. A similar site-directed mutagenesis approach was used to create isogenic Cj0630c (DNA polymerase III, delta subunit) and Cj1116c (a putative membrane-bound zinc metallopeptidase) mutants of NCTC 11168.

To complement the perR (a transcriptional regulator gene) mutation in K15A2, a 81-176 derivative obtained from the random transposon mutagenesis study described above, the complete perR gene, together with its 200-bp upstream and 100-bp downstream regions, was amplified using PerRF and PerRR primers (Table 2) in conjunction with Pfu polymerase (Stratagene). The blunt-ended PCR product was purified and ligated to the SmaI-digested shuttle vector pRY111 (47). A ligation mixture was then introduced into E. coli DH5α, creating construct JL402. The pPerR plasmid from JL402 was transferred into K15A2 by triparental conjugation using DH5α/pRK2013 as a helper strain (1). The complemented strain JL412, together with other related strains, was tested for BCN susceptibility. To complement the cmeB mutation in JL360 (JL341 background), the plasmid pCME bearing the cmeABC operon (23) was extracted and transferred into JL360 by natural transformation, creating the complemented strain JL424.

RESULTS

Prevalence of BCN resistance in C. jejuni and C. coli isolates from various sources.

All C. jejuni isolates displayed low OR-7 MICs, ranging from 0.25 to 1.0 μg/ml. Among the 137 tested C. jejuni isolates, 84 showed OR-7 MICs of 1.0 μg/ml, while 4 and 49 isolates showed MICs of 0.25 and 0.5 μg/ml, respectively. C. coli appeared to display greater intrinsic BCN resistance than C. jejuni; only 40% of C. coli isolates (8 out of 20) showed MICs of ≤1 μg/ml (2 isolates with MICs of 0.5 μg/ml and 6 isolates with MICs of 1 μg/ml). All other C. coli isolates displayed higher MICs, i.e., 2 μg/ml (8 isolates), 4 μg/ml (3 isolates), and 64 μg/ml (1 isolate from a human, and this isolate was designated JL106; Table 1). Interestingly, each tested Campylobacter strain displayed an identical MIC for another BCN, E-760.

Frequency of in vitro emergence of OR-7-resistant Campylobacter.

Three C. jejuni strains (NCTC 11168, 81-176, and S3B) and two C. coli isolates (JL20 and JL25) that displayed the same MIC of OR-7 (0.5 μg/ml) were chosen for examination of the frequency of in vitro emergence of BCN resistance. As shown in Table 3, the OR-7^r mutants were not obtained by using C. jejuni strains under different tested selection pressures. In contrast, C. coli strains displayed higher frequencies of emergence of BCN resistance (1.0 × 10⁻⁸ to 9.0 × 10⁻⁶) in the presence of different concentrations of OR-7 by a single-step selection; the MICs of selected OR-7^r mutants ranged from 8 to 32 μg/ml. Independent in vitro selection experiments showed findings similar to those presented in Table 3 (data not shown). We also intended to select mutants with higher levels of OR-7 resistance by using the cells grown in the presence of sublethal concentration of OR-7. However, we failed to select OR-7^r mutants with MICs of >64 μg/ml after extensive efforts.

TABLE 3.

Frequency of the emergence of BCN OR-7-resistant Campylobacter in vitro^a

BCN concn (μg/ml)	Frequency of emergence of OR-7 resistance for strainb
	C. jejuni			C. coli
	81-176	11168	S3B	JL20	JL25
4	<9.2 × 10−9	<2 × 10−10	<1.5 × 10−10	3.2 × 10−7	9.0 × 10−6
8	<9.2 × 10−9	<2 × 10−10	<1.5 × 10−10	8 × 10−8	4.0 × 10−7
16	<9.2 × 10−9	<2 × 10−10	<1.5 × 10−10	1.0 × 10−8	1.2 × 10−8

^aThe spontaneous OR-7^r mutants were selected on MH agar plates containing different concentrations of OR-7 by using late-log-phase Campylobacter culture grown in OR-7-free MH broth.

^bFrequency values are the ratios of CFU on BCN-containing plates to the CFU on BCN-free MH agar plates. The values are means of triplicate measurements in a single independent experiment.

The acquired BCN resistance could be transferred at intra- and interspecies levels in Campylobacter.

C. jejuni and C. coli are well known for their exceptional ability to acquire exogenous DNA by natural transformation, which is considered a major mechanism mediating horizontal transfer of antibiotic resistance in Campylobacter. Thus, we examined if natural transformation contributes to horizontal transfer of BCN resistance in Campylobacter. As shown in Table 4, the acquired BCN resistance in C. coli JL106 and JL349 could be transformed to C. coli isolates such as JL20 and JL25 with a transformation frequency of about 4 ×10⁻⁷ CFU/μg DNA/recipient cell. Interestingly, the OR-7 resistance of JL106 and JL349 also could be easily transferred into strains of different species, such as C. jejuni 81-176 and NCTC 11168, with a transformation frequency of ∼10⁻⁸ CFU/μg DNA/recipient cell. The MICs of BCN for the selected BCN^r transformants were up to 16 μg/ml (Table 4).

TABLE 4.

Horizontal gene transfer of OR-7 resistance among Campylobacter spp.^a

Genomic DNA donor (MIC in μg/ml)	Recipient strain	Transformation frequency (transformants per μg DNA per recipient CFU)b	Highest MIC of selected transformants (μg/ml)c
C. coli JL106 (64)	C. jejuni NCTC 11168	1.2 × 10−8	8 (JL341)
	C. coli JL20	7.1 × 10−7	ND
	C. coli JL25	3.6 × 10−7	ND
C. coli JL349 (32)	C. jejuni 81-176	1.0 × 10−8	16 (JL358)
	C. coli JL20	4 × 10−7	16
	C. coli JL25	4 × 10−7	16

^aGenomic DNA extracted from a specific OR-7^r C. coli strain was used to transform Campylobacter by standard biphasic natural transformation (6).

^bThe values are from a representative experiment with duplicate measurements within the independent experiment.

^cND, not determined.

Transcriptional profiling of BCN^r mutant JL341.

DNA microarray analysis was performed to compare the transcriptome of the BCN^r mutant JL341 to that of its parent strain, NCTC 11168. The microarray analysis revealed that 9 genes were upregulated and 10 genes were downregulated in OR-7^r mutant JL341 (Table 5). Since upregulation of some genes, such as those involved in lipopolysaccharide (LPS) modification and peptide degradation, is a common mechanism used by enteric pathogens to confer resistance to AMPs (48), the upregulated genes from this microarray analysis are of particular concern, and these include genes encoding multidrug efflux pump CmeABC, putative drug efflux pumps Cj1687 and Cj0035c (belonging to a major facilitator superfamily [MFS]), putative membrane-bound zinc metallopeptidase Cj1125c, and galactosyltransferase PglA (Table 5). Upregulation of these genes was also confirmed by qRT-PCR using specific primers (Table 5). The overexpression of CmeABC in JL341 was further confirmed by β-galactosidase promoter fusion assay and immunoblotting using specific antibodies against CmeB and CmeC as described in our previous publication (22; data not shown). PCR amplification and sequencing of the cmeABC operon from JL341 did not reveal any mutations compared to its parent strain, NCTC 11168.

TABLE 5.

Differentially expressed genes (≥2-fold changes) in NCTC 11168 OR-7^r mutant JL341 identified by DNA microarray and real-time qRT-PCR

Gene	Function categories and description of product	P value	Fold change by microarray	Fold change by qRT-PCRa
Upregulated genes
Cj0366c (cmeB)	Inner membrane transporter of CmeABC efflux system	5.93E−6	4.9	6.7
Cj0365c (cmeC)	Outer membrane component of CmeABC efflux system	2.72E−5	4.0	7.4
Cj0630c	DNA polymerase III, delta subunit	4.63E−4	13.5	ND
Cj1726c	Putative homoserine O-succinyltransferase	8.31E−4	2.0	ND
Cj1116c	Putative membrane-bound zinc metallopeptidase	4.00E−3	4.8	4.0
Cj1125c	Putative galactosyltransferase (WlaG/PglA)	5.50E−3	12.8	8.0
Cj0176c	Putative lipoprotein	7.05E−3	2.0	ND
Cj1687	Putative efflux transporter (MFS family)	2.14E−2	2.0	2.0
Cj0035c	Putative efflux transporter (MFS family)	2.19E−2	3.4	2.0
Downregulated genes
Cj0508	Penicillin-binding protein	2.37E−5	−4.5	ND
Cj0093	Putative periplasmic protein	1.49E−4	−2.5	ND
Cj0628	Putative lipoprotein	1.57E−4	−2.9	ND
Cj0045c	Putative iron-binding protein	6.60E−4	−2.0	ND
Cj0091	Putative lipoprotein	8.90E−4	−2.5	ND
Cj0629	Putative lipoprotein	1.16E−3	−2.7	ND
Cj1423c	Putative sugar-phosphate nucleotidyltransferase	1.49E−3	−2.7	ND
Cj1650	Hypothetical protein	1.80E−3	−17.9	ND
Cj1714	Small hydrophobic protein	4.56E−3	−2.4	ND
Cj1539c	Putative anion-uptake ABC transport system protein	9.39E−3	−2.1	ND

^aMeans of three independent experiments. ND, not determined.

The isogenic cmeB, Cj0035c, Cj1125c, and Cj1687 mutants of JL341 were successfully obtained using natural transformation or site-directed mutagenesis. However, our extensive efforts to construct isogenic Cj0630c and Cj1116c mutants of JL341 were unsuccessful, probably due to the essential role of these gene products in Campylobacter growth. Except for the cmeB mutant, none of the generated isogenic mutants displayed increased susceptibilities to OR-7. Inactivation of cmeB in JL341 led to a 4-fold reduction in the MIC of BCN OR-7 (Table 6). Complementation of the cmeB mutant with plasmid pCME fully restored the MIC back to the level of the parent strain (Table 6). To test if CmeABC contributes to intrinsic resistance of C. jejuni NCTC 11168 to BCN, isogenic cmeB mutant was compared with the wild-type strain for susceptibility to OR-7. As shown in Table 6, inactivation of CmeB in the wild-type strain also significantly increased susceptibility of the mutant (JL199) to OR-7. Compared to the case for its parent strain, overexpression of CmeABC in JL4 due to cmeR mutation led to a slight but consistent increase in the MIC of OR-7 (1 μg/ml) (Table 6).

TABLE 6.

CmeABC efflux pump contributes to both intrinsic and acquired OR-7 resistance in C. jejuni NCTC 11168

Strain	Description	MIC of OR-7 (μg/ml)
JL241	Wild-type NCTC 11168	0.5
JL199	Isogenic cmeB mutant of NCTC 11168	0.125
JL4	Isogenic cmeR mutant of NCTC 11168	1
JL341	NCTC 11168 derivative, BCNr mutant	8.0
JL360	Isogenic cmeB mutant of JL341	2
JL424	JL360 containing pCME for complementation of cmeB	8

Identification of genes contributing to OR-7 resistance by random transposon mutagenesis.

A complementary genomic approach for microarray analysis, random transposon mutagenesis, was used to identify genes involved in BCN resistance for this study. A library containing 2,496 Kan^r mutants was generated to screen the mutants with an increased susceptibility to BCN OR-7. Six mutants displaying greater susceptibility to OR-7 than parent strain JL358 were identified (Table 7). Backcrossing of the transposon mutations into the parent strain by natural transformation further confirmed that the BCN-sensitive phenotype in each mutant was linked to the gene with a specific transposon insertion. Direct sequencing of the mutant genomic DNA by using transposon-specific primers (24) revealed a specific transposon insertion site in each mutant (Table 7). All the transposon insertions occurred in the coding regions of corresponding genes. Five mutants had transposons inserted in different sites of the genes encoding the multidrug efflux pump CmeABC, which has been characterized in our previous study (22). This finding clearly indicated that CmeABC was involved in acquired BCN resistance in JL358. However, the expression level of CmeABC in JL358 is comparable to its sensitive parent strain 81-176 as demonstrated by immunoblotting and LacZ-promoter fusion assays (data not shown). In addition, no mutations occurred in the cmeABC operon in JL358, based on sequencing analysis (data not shown). Similar to the finding in NCTC 11168 (Table 6), inactivation of CmeABC in 81-176 also led to increased susceptibility to OR-7 (4-fold MIC reductions).

TABLE 7.

Identification of transposon mutants with increased sensitivity to BCN OR-7

Strain	MIC of OR-7 (μg/ml)	Locus designation	Tn location (ORF size in bp)a	Function of inserted gene product
JL358	16	NAb	NA	NA
K15A2	4	Cj0322 (perR)	54 (411)	Fur family regulator
K17E10	4	Cj0367c (cmeA)	557 (1,104)	Component of CmeABC pump
K16H6	4	Cj0366c (cmeB)	790 (3,123)	Component of CmeABC pump
K11A10	4	Cj0365c (cmeC)	107 (1,479)	Component of CmeABC pump
K1H1	4	Cj0366c (cmeB)	1,679 (3,123)	Component of CmeABC pump
K15G10	4	Cj0365c (cmeC)	1,213 (1,479)	Component of CmeABC pump

^aThe number indicates the nucleotide before which the transposon (Tn) is inserted. ORF, open reading frame.

^bNA, not applicable.

The mutant K15A2 has a transposon insertion in perR, a gene encoding the Fur family regulator. Complementation of the perR mutation in K15A2 (construct JL412) partially restored BCN resistance, to an OR-7 MIC of 8 μg/ml. In addition, inactivation of PerR in wild-type C. jejuni 81-176 also resulted in increased susceptibility to OR-7 (2-fold MIC reductions), indicating that PerR is also involved in intrinsic resistance to BCN.

DISCUSSION

This comprehensive survey used Campylobacter strains from varied origins and geographically diverse regions. This study provides more compelling evidence demonstrating the potent killing activity of BCNs OR-7 and E-760 against Campylobacter, which has been reported in previous publications (40, 41). The survey also indicated that BCN^r Campylobacter strains are rarely detected among clinical, poultry, and environmental isolates, suggesting that development of BCN resistance in Campylobacter is uncommon, even though Campylobacter strains frequently encounter various BCNs produced by commensal bacteria in the intestine (40, 41). The limited in vitro emergence of OR-7^r Campylobacter described in this study (Table 3) demonstrated that no OR-7^r C. jejuni mutants were selected under selection pressure. Only C. coli could develop BCN resistance in vitro, with all mutants displaying only low-level resistance to BCN. It is still unknown why BCN resistance in C. coli develops more quickly than in C. jejuni. Nevertheless, our findings strongly indicate that among Campylobacter strains, there is very limited development of resistance to anti-Campylobacter BCNs, such as OR-7 and E-760. These findings support a recent theory that bacteria have not developed a highly effective mechanism to resist BCNs and other endogenous AMPs during evolution, which is likely due to multiple targets of natural AMPs (33). The best-studied BCN is nisin, which is ribosomally produced by Lactococcus lactis. Nisin exerts bactericidal effects via at least two modes of action: targeting the membrane-bound cell wall precursor lipid II, consequently resulting in inhibition of peptidoglycan synthesis, and membrane pore formation, resulting in membrane damage and depolarization (16). Resistance to nisin in some bacteria has been reported (28, 39); however, high levels of nisin resistance in bacteria were not observed although nisin has been used as a food preservative for a half century (9). It has been also reported that the acquired bacterial resistance to BCNs, such as occurs with pediocin PA-1 and nisin A, is unstable and exacts an associated fitness cost (11). The stability of the acquired resistance to the anti-Campylobacter BCNs in Campylobacter is still unknown and needs to be examined in future studies.

AMPs, including various BCNs, have been considered potential natural “peptide antibiotics” to combat bacterial infections (2, 5, 8, 14). Elucidating the underlying mechanisms of AMP resistance in bacteria could help us to develop “smarter” antibiotics (33). Therefore, in this study, we also determined genetic loci involved in BCN resistance in Campylobacter by using complementary genomic approaches. Both microarray and random transposon mutagenesis demonstrated the role of multidrug efflux pump CmeABC in resistance of Campylobacter to BCN OR-7. Active extrusion of antimicrobials by multidrug-resistant (MDR) efflux pumps plays the vital role in antimicrobial resistance expressed by many Gram-negative bacteria (34). It has been observed that MDR efflux pumps are also involved in AMP resistance in several bacteria (30, 42, 48). The best-studied bacterium is Neisseria gonorrhoeae, whose mtrCDE multidrug efflux pump contributes to both intrinsic AMP resistance and acquired AMP resistance; overexpression of MtrCDE mildly increased resistance to human AMPs (2- to 4-fold increase in MIC) (37). In Campylobacter, the CmeABC MDR efflux pump contributes to resistance to various antimicrobials, including both structurally diverse antibiotics and natural antimicrobials, such as bile salts, which are present in the intestine (22, 23, 46). It is important to mention that the CmeABC efflux pump is widely distributed and produced in C. jejuni and C. coli; overexpression of CmeABC (mediated by mutations in either repressor CmeR or the CmeABC promoter region) led to only slightly increased resistance to clinical antibiotics, including fluoroquinolones (2-fold MIC increase) and macrolides (4-fold MIC) (13, 22, 23), which is different from what has been observed for many other Gram-negative bacteria in which overexpression of a MDR efflux pump alone could result in dramatically increased drug resistance (34). However, a significant contribution of CmeABC to acquired antibiotic resistance in Campylobacter is reflected by its synergy with other nonefflux resistance mechanisms which do not require overexpression of CmeABC and/or any mutations in the regulatory elements of CmeABC (1, 20, 21, 22, 24, 27).

In this study, we observed that CmeABC also contributed to both intrinsic BCN resistance and acquired BCN resistance in Campylobacter. Notably, the findings from this study also indicated that involvement of CmeABC in BCN resistance does not require overexpression of CmeABC, although CmeABC is found to be overexpressed in NCTC 11168-derived BCN-resistant mutant JL341. This notion is also supported by the finding that the isogenic CmeR mutant of NCTC 11168 (JL4) that overproduced CmeABC led to an only 2-fold increase in the MIC of OR-7 (Table 6). Therefore, our findings indicated that the CmeABC efflux pump works synergistically with other nonefflux BCN resistance mechanisms to confer BCN resistance in Campylobacter. Altogether, this study provides further evidence explaining why CmeABC is essential for Campylobacter colonization in the intestine (24) and further highlighting the multifunction nature of CmeABC and the critical role of this efflux pump in Campylobacter pathobiology.

Although CmeABC plays an important role in BCN resistance, inactivation of CmeABC alone in the OR-7^r JL341 did not lead to a susceptibility level comparable to that of its wild-type parent strain, NCTC 11168 (JL241 MIC, 0.5 μg/ml; Table 6), strongly suggesting that other factors work together with CmeABC to contribute to the acquired BCN resistance observed in JL360. Based on the microarray work in this study, several upregulated genes are of particular interest. In addition to the putative efflux transporters Cj1678 and Cj0035c, Cj1116c, a putative peptidase, may involve proteolytic cleavage of BCNs, because degradation of AMPs by protease is one of mechanisms used by bacteria to resist endogenous AMPs (48). In Salmonella enterica serovar Typhimurium, the membrane protease PgtE, which is regulated by the two-component regulatory system PhoP/PhoQ, is responsible for degrading cationic AMPs (12). The zinc-dependent membrane metalloproteases ZmpA and ZmpB in Burkholderia cenocepacia degrade the antimicrobial peptide protamin and human cathelicidin LL-37 as well as other alpha-helical cationic AMPs (18). PglA (Cj1125c) is a galactosyltransferase that is involved in N-linked protein glycosylation, a unique surface carbohydrate modification mechanism that was observed in C. jejuni (26); such surface modification may be required for BCN resistance in Campylobacter. In this study, we successfully achieved mutations in most of these interesting genes. However, none of these mutants displayed increased susceptibilities to BCN compared to that of the parent strain, JL341. We were not able to generate a mutant with a mutation in Cj1116c, strongly suggesting that Cj1116c is an essential gene for Campylobacter physiology. Additional work, such as production and purification of recombinant Cj1116c, is still needed to determine whether Cj1116c could function as a peptidase to degrade the anti-Campylobacter BCNs.

Random transposon mutagenesis also identified another gene, perR, that is involved in BCN resistance. PerR is a transcriptional regulator controlling transcription of the genes encoding the oxidative stress resistance proteins (such as catalase KatA, superoxide dismutase SodB, and alkyl-hydroxyperoxidase AhpC) (31). Transcriptional profiling analysis of an isogenic perR mutant identified 104 genes that belong to the PerR regulon (31). PerR activates several genes encoding proteins responsible for capsule biosynthesis, which include acetyl coenzyme A (acetyl-CoA) carboxylase AccA, 3-oxoacyl-synthase FabH, and fatty acid/phospholipid synthesis protein PlsX (31). Thus, inactivation of perR in JL358 may affect capsule synthesis and such surface remodeling may increase the susceptibility of JL358 to the BCN OR-7. This speculation needs to be examined in the future.

The results obtained from this study provide helpful information for the risk assessment of on-farm control of Campylobacter by using the anti-Campylobacter BCNs and represent the first and major step forward in understanding the genetic mechanisms of Campylobacter resistance to BCNs. However, it is still unknown if higher selection pressure (e.g., therapeutic usage of bacteriocins) will promote the emergence of BCN^r Campylobacter mutants in vivo. If so, can Campylobacter develop high-level bacteriocin resistance in response to therapeutic treatment with BCN? In addition, it is unclear if BCN^r Campylobacter can persist in the absence of selection pressure. To obtain solid answers to these questions, multiple laboratory experiments using chickens should be performed to examine the dynamic change of Campylobacter populations in response to BCN treatment and to determine the in vivo stability of BCN resistance in Campylobacter. These studies are expected to provide important information that may help to avoid a rapid loss of efficiency of BCN and to design more sustainable and smarter peptide antibiotics. In addition, examination of the molecular basis of BCN resistance in Campylobacter may help us to develop more sustainable and effective BCN-based intervention strategies against Campylobacter colonization in chickens. In this study, we revealed that active transport of multidrug efflux pump CmeABC confers resistance to BCN in Campylobacter. Thus, inhibition of this pump by efflux pump inhibitors will significantly increase susceptibility of Campylobacter to BCN. Previous studies (15, 21) have showed that inhibition of efflux pump CmeABC in Campylobacter spp. resulted in increased susceptibility to various antimicrobials. Notably, such efflux pump inhibitors also could dramatically increased Campylobacter susceptibility to intestinal bile salts by inhibiting CmeABC, leading to reduced colonization of Campylobacter in chickens (15, 21). Therefore, oral administration of the anti-Campylobacter BCNs together with such efflux pump inhibitors by using an appropriate delivery system (e.g., encapsulation) would enhance the therapeutic effect of anti-Campylobacter BCNs. Further answers need to be determined in our future studies.