Genetic Basis and Functional Consequences of Differential Expression of the CmeABC Efflux Pump in Campylobacter jejuni Isolates

The CmeABC multidrug efflux transporter of Campylobacter jejuni plays a key role in antimicrobial resistance and is suppressed by CmeR, a transcriptional regulator of the TetR family. Overexpression of CmeABC has been observed in laboratory-generated mutants, but it is unknown if this phenotype occurs naturally in C. jejuni isolates and if it has any functional consequences. To answer these questions, expression of cmeABC in natural isolates obtained from broiler chickens, turkeys and humans was examined, and the genetic mechanisms and role of cmeABC differential expression in antimicrobial resistance was determined. Among the 64 C. jejuni isolates examined in this study, 43 and 21 were phenotypically identified as overexpression (OEL) and wild-type expression (WEL) levels. Representative mutations of the cmeABC promoter and/or CmeR-coding sequence were analyzed using electrophoretic mobility shift assays and transcriptional fusion assays. Reduced CmeR binding to the mutated cmeABC promoter sequences or decreased CmeR levels increased cmeABC expression. Several examined amino acid substitutions in CmeR did not affect its binding to the cmeABC promoter, but a mutation that led to C-terminal truncation of CmeR abolished its DNA-binding activity. Interestingly, some OEL isolates harbored no mutations in known regulatory elements, suggesting that cmeABC is also regulated by unidentified mechanisms. Overexpression of cmeABC did not affect the susceptibility of C. jejuni to most tested antimicrobials except for chloramphenicol, but promoted the emergence of ciprofloxacin-resistant mutants under antibiotic selection. These results link CmeABC overexpression in natural C. jejuni isolates to various mutations and indicate that this phenotypic change promotes the emergence of antibiotic-resistant mutants under selection pressure. Thus, differential expression of CmeABC may facilitate Campylobacter adaptation to antibiotic treatments.


Introduction
Multidrug efflux pumps play key roles in bacterial physiology, conferring intrinsic and acquired resistance to diverse toxic compounds.There are multiple types of drug efflux systems in bacteria, but in gram-negative bacteria, the resistance-nodulation-cell division (RND) family of efflux pumps is of primary importance for antimicrobial resistance [1].ArcAB-TolC, MexAB-OprM, and MexXY-Z in Escherichia coli and Pseudomonas aeruginosa are examples of well-characterized RND efflux pumps that extrude bile salts, organic cations, detergents, and various classes of antimicrobials [2][3][4][5].These efflux pumps have been associated with intrinsic and acquired resistance to antimicrobial compounds.CmeABC, also a member of the RND family, is the predominant efflux pump in Campylobacter jejuni [6] and plays a key role in the resistance to structurally diverse compounds, such as bile salts, ciprofloxacin, erythromycin, ethidium bromide, and various detergents [6,7].Synergistic effects of CmeABC with other resistance mechanisms (such as target gene mutations) contribute to high-level resistance to macrolides, tetracyclines, and fluoroquinolones in Campylobacter [8][9][10][11].Due to its significant role in bile resistance, CmeABC is essential for Campylobacter colonization and adaptation in the intestinal tract of animals [7].
CmeABC is comprised of an inner membrane transporter (CmeB), a periplasmic fusion protein (CmeA), and an outer membrane protein (CmeC).The three components are encoded by a three-gene operon [6].Transcription of this operon is repressed by CmeR, a TetR family transcriptional regulator [12].The cmeR gene is located immediately upstream of the cmeABC operon.The CmeR protein contains a C-terminal ligand-binding domain and a N-terminal DNA-binding domain [7].The DNA-binding domain interacts specifically with a 16-base inverted repeat within the promoter region of the cmeABC operon [13].This binding inhibits the transcription of the cmeABC operon.However, mutation of CmeR or alteration of the promoter sequences affects the binding of CmeR, resulting in increased expression of CmeABC [12,14].Additionally, cmeABC expression is inducible by bile and this induction is through the interaction of bile with the ligand-binding pocket of CmeR, which triggers a conformational change in the DNA-binding domain, releasing CmeR from the cmeABC promoter [15,16].
Gastroenteritis caused by Campylobacter is estimated to affect 845,024 people and cause 8,463 hospitalizations per year in the United States [17].C. jejuni and C. coli are the most common Campylobacter species associated with foodborne disease and are commensals in avian species (particularly poultry), swine, and ruminants [18].Campylobacter contamination frequently occurs with poultry meat and unprocessed milk.Thus, undercooked poultry and unpasteurized milk are common vehicles for foodborne transmission of Campylobacter to humans [18,19].Clinically, campylobacteriosis is manifested as diarrhea, abdominal cramps, and fever, which typically resolves in 1 week without medical intervention.However, when antimicrobial treatment is indicated with severe or prolonged cases, or in immunocompromised patients, fluoroquinolones and macrolides are the drugs of choice [18,20].Increasing resistance to these antibiotics in Campylobacter is problematic, especially to fluoroquinolones, as Campylobacter is highly adaptable to fluoroquinolone treatment and acquisition of mutations associated with fluoroquinolone resistance does not impose a fitness cost on this organism [21][22][23].In all clinically relevant antibiotic resistance, CmeABC plays an important role as inactivation of cmeABC rendered Campylobacter much more susceptible to various antimicrobials [6].
Considering the significance of CmeABC in Campylobacter pathobiology, its varied expression levels are expected to affect antimicrobial resistance.Under toxic conditions or in the adaptation to harsh environments, enhanced expression of cmeABC may confer a survival advantage on Campylobacter.The advantage may occur directly through increased extrusion of toxic substrates, which reduces their harmful effect and increased frequency of emergence of antimicrobial resistant mutants.Although inactivation of cmeABC or overexpression of this efflux pump has been examined under experimental conditions by using insertional mutagenesis or stepwise selection of mutants on antibiotic containing plates [12,14], it is unknown if differential expression of cmeABC occurs in naturally occurring isolates and if the differential expression has any functional consequences.In this study, we investigated the expression of cmeABC in C. jejuni isolates from turkeys, chickens, and humans, examined the mechanisms associated with the differential expression, and measured the functional consequences associated with the differential expression.

Immunoblotting
Isolates were initially screened for CmeABC expression by immunoblotting with polyclonal antibodies against CmeABC.All clinical C. jejuni isolates, NCTC 11168 and 11168ΔcmeR were cultured in MH broth.Samples were pelleted, and re-suspended in SDS loading buffer for a final concentration of 5 x 10 9 CFU/mL.The protein samples were analyzed by SDS-PAGE and immunoblotting as described previously using antibodies against CmeA, CmeB, CmeC [6] and MOMP [9].Bands for CmeA, CmeB, and CmeC from the clinical isolates were compared to those of NCTC 11168 and 11168ΔcmeR by visual inspection and densitometric analysis using the AlphaEaseFC Software (Version 3.2.3Rev C; Innotech).Primary classification as wild typelevel (WEL) or overexpression level (OEL) of CmeABC was based on analysis of CmeB by densitometry and CmeA was utilized as a secondary factor for the classification.The 64 clinical C. jejuni isolates were analyzed on 8 immunoblots with NCTC 11168 and 11168ΔcmeR used as controls for WEL and OEL, respectively.The threshold for CmeABC overexpression based on densitometric analysis was a 2-fold increase for the CmeB band in relation to the expression level in NCTC 11168.Each immunoblot was examined individually to ensure the threshold for overexpression was met.This phenotypic classification was further confirmed by measuring the cmeB transcript using real time RT-PCR.OEL isolates showed overexpression of CmeB on immunoblotting and/or real time RT-PCR expression of cmeB.CmeA expression based on a threshold of 2-fold for overexpression was used for isolates that remained between the two phenotypic groups after real time cmeB expression levels were analyzed.
The expression level of CmeR was also evaluated by immunoblotting in selected clinical isolates that harbored mutations in the cmeR gene.The whole cell samples were prepared from isolates M63885, CT9:7, CB2:6, CB2:8, CB2:11, S13530, T37957A, X7199, CT2:2, NCTC 11168, and 11168ΔcmeR.Samples were loaded onto a 12% SDS PAGE gel for electrophoresis 16srRNA-R GTAACTAGTTTAGTATTCCGG [26] mapA-F GAGTGCTTGTGCAACTAAAC [27] mapA-R ATAGCATCTTGAGTTGCTCC [27] GSF CTAAATGGAATCAATAGCTCC [12] GSR1 GCACAACACCTAAAGCTAAAA [12] PF AAAAGGATCCTAAATGGAATCAATAGCTCC (BamHI) [12] 16S-F TACCTGGGCTTGATATCCTA [34] 16S-R GGACTTAACCCAACATCTCA [34] cmeB-F ACGATTCAACCTTTTCCCAGC [34] cmeB-R TTTGCTACTTGAGCAATCGCTTC [34] F3 ATTTTCAATCAACCAGAAGCTG [16] R3 TCCAATTGGCAAGATGTCTATC [16] 1 Restriction sites are indicated by underlined sequence doi:10.1371/journal.pone.0131534.t001 in Lameilli buffer at 80V for 30 minutes followed by 200V for 60 minutes.The gel and PVDF membrane were equilibrated in Towbin transfer buffer for 30 minutes, assembled onto the transfer apparatus, and transferred at 60 V for 40 minutes in Towbin transfer buffer.The membrane was blocked in blocking buffer (5% skim milk with 0.01% Tween-20 in PBS) at 4°C on a rocker.Then it was incubated with rabbit anti-CmeR diluted 1:100 in blocking buffer for 90 minutes at room temperature on a rocker.The membrane was washed 3 times for 10 minutes in washing buffer (0.01% Tween-20 in PBS) and further incubated with goat anti-rabbit IgG conjugated with horseradish peroxidase (KPL) (1:1000 in blocking buffer) for 1 hour at room temperature.After three washings, the membrane was developed with the 4CN Horseradish Peroxidase Substrate system (KPL).Densitometric analysis for CmeR was performed in the same manner for CmeABC.SDS-PAGE and immunoblotting were also performed for the recombinant mutant CmeR named rCmeR-tr to determine if the protein was recognizable by CmeR antibodies.This recombinant protein was derived from mutations in strain CT2:2 (Table 3).The recombinant CmeR named rCmeRSS was used as a control.Both rCmeRSS and rCmeR-tr were loaded at 400 ng onto a 12% SDS-PAGE gel.Electrophoresis and immunoblotting were performed using the same methods used for immunoblotting of CmeR in clinical isolates.

Real time RT-PCR
Transcription of cmeB and cmeR was detected by real time RT-PCR to confirm the results of immunoblotting.RNA was isolated from 24-hour cultures of clinical isolates, NCTC 11168, and 11168ΔcmeR as described previously (13).Real time RT-PCR was performed for all isolates for cmeB and for selected isolates for cmeR as described previously (12,13,16,34).The primers used for real time RT-PCR are shown in Table 1.Relative expression based on NCTC 11168 was calculated with the Pfaffl Method [35].Overexpression of cmeB was defined as greater or equal to 3 fold of NCTC 11168 expression.cmeB expression measured by RT-PCR was compared with the result of immunoblotting for final determination of a CmeABC phenotype.For those isolates where RT-PCR and immunoblotting data did not align, less weight was given to real time RT-PCR data due to reported variability in cmeB coding sequence [36].

DNA sequencing
DNA sequencing was performed to determine if isolates were carrying mutations in cmeR, the Cj0369c-cmeR promoter or the cmeABC promoter.All clinical isolates classified as having a phenotype of CmeABC overexpression, some selected clinical isolates with the wild-type expression levels, and 81-176 were amplified with RIGA-F and RIGA-R primers (Table 1) for sequence analysis of the region between cmeR to the 5'end of cmeA, which covers the whole ORF of cmeR and the entire promoter of cmeABC.Additionally, the predicted promoter region of Cj0369c-cmeR (the two genes share a single promoter located upstream of Cj0369c) was examined to determine if mutations in the promoter region were involved in differential CmeR expression in clinical isolates.The Cj0369c-cmeR promoter was amplified with Cj0370-F and Cj0368c-R (Table 1).GenBank accession numbers are found in S1 Table.
Electrophoretic mobility shift assays (EMSA) EMSA was used to assess the binding of CmeR to the cmeABC promoter sequences from various clinical isolates.These clinical isolates were observed to contain sequence polymorphisms within the cmeABC promoter region.A recombinant CmeR named rCmeRSS [37] with cysteines 69 and 166 replaced with serine was used for EMSA as described previously [12,38].The Cys-Ser substitution does not affect the binding activity, but significantly improves the stability of recombinant CmeR as the Cys residues are sensitive to oxidation during in vitro binding assay.The 170-bp cmeABC promoter sequences were amplified from genomic DNA of NCTC 11168, 81-176, CT3:7, CT1:1, CT1:9, CT9:20, and X7199 with primers GSF and GSR1 [12] (Table 1).The amplified products were purified (QIAquick PCR Purification kit, Qiagen) and then labeled with DIG-11-dd-dUTP using the DIG Oligonucleotide 3' End Labeling kit (Roche).The labeled promoter DNA was used as probes in EMSA.
The promoter probes (0.05 pmol each) were incubated with 0, 60, 120, and 180 ng of rCmeRSS in 22 μL of binding buffer according to the method of Alekshun et.al. and Lin et. al. 2005 [12, 38].The reaction mixtures were incubated for 30 minutes at room temperature and Promega DNA loading buffer was added to each reaction.Samples were separated by electrophoresis at 200V for 45 minutes on a 6% polyacrylamide gel in 0.25X TBE Buffer and transferred to a positively charged nylon membrane by vacuum [12].Chemiluminescent detection using alkaline phosphatase-conjugated anti-DIG antibody and CDP-Star (Roche) was performed as previously described [12].

Construction of promoter fusions and β-galactosidase assays
The observed sequence polymorphisms in the promoter sequences of cmeABC were assessed for their impact on cmeABC transcription by constructing transcriptional fusions with a promoter-less lacZ gene.Genomic DNA templates (NCTC 11168, 81-176, CT1:1, CT 1:9, CT9:20, CT3:7, M32506 and X7199) were used for amplification of a 578-bp sequence containing the cmeABC promoter with primers PF [12] and PX (Table 1).These PCR products were purified (QIAquick PCR Purification kit), digested with XbaI and BamHI (Promega), and re-purified using the QIAquick kit.Vector pMW10 [32] was purified (QIAprep Spin Miniprep kit, Qiagen) from DH5αpMW10 (Table 2), digested with the same enzymes, and re-purified.Vector and PCR product inserts were ligated with T4 ligase (Roche) and transformed into DH5α.Transformants were selected on LB agar plates supplemented with kanamycin (30 μg/mL).Plasmid constructs (Table 2) were purified from DH5α and sequenced with pMW10-F and pMW10-R (Table 1) to confirm the appropriate sequences and fusion.
Sequence polymorphisms were also observed in the Cj0369c-cmeR promoter.To determine if these mutations affected CmeR expression, a second set of promoter fusions was created using the same method.Briefly, primers 370BamH-F and 369XbaI-R1 (Table 1) were used to amplify a 238-base pair segment containing the Cj0369c-cmeR promoter from genomic DNA templates of NCTC 11168, 81-176 and X7199.Construction of plasmids, transformation into DH5α and sequencing was performed as described above.Plasmid constructs pMW11168-R, pMW81-176-R and pMWX7199-R were electroporated into C. jejuni 81-176 (Table 2).The empty vector, 81-176pMW10, from the prior assay was used as a background control.β-galactosidase assays for 3 independent experiments and statistical analysis were performed as described in the assays with the cmeABC promoters.

Construction, purification, and functional analysis of various CmeR variants
Sequence polymorphisms were detected in CmeR among the analyzed isolates.To determine if the sequence variations affected the DNA binding activity of CmeR, we generated various forms of recombinant CmeR using site-directed mutagenesis.Plasmid pQECmeRSS [37] was used as a template for site directed mutagenesis, which was done using the Stratagene Quik-Change II kit.Site-specific primers (Table 1) were used to produce the amino acid changes in CmeR from isolates CT2:2, M63885, CT9:7, and T37597A.All amino acid substitutions, corresponding nucleotide sequences and protein names are listed in Table 3. Isolate M63885 contains 2 amino acid substitutions in CmeR, which were introduced simultaneously into the pQECmeRSS template.The M63R17-F and M63R17-R primers were used to mutate the threonine to isoleucine at residue 3 and the M63475-F and M63475-R primers were used to mutate the glutamate to lysine at residue 159.This mutated plasmid was named pQECmeR-IK.Isolates CT9:7 and T37957A both contain a single amino acid substitution in CmeR.Primers T97547G-F and T97547G-R were used to change the proline to arginine at residue 159 as observed in strain CT9:7, creating the plasmid named pQECmeR-R.Primers T3X250A-F and T3X250A-R introduced the glutamate to lysine substitution at residue 84 as observed in strain T37957A.This plasmid was named pQECmeR-K.Isolate CT2:2 contains 2 amino acid substitutions in CmeR and a nucleotide insertion.This mutated CmeR was created using 2 rounds of mutagenesis.The T insertion after nucleotide 583 was introduced into template pQECmeRSS with primers T22583-F1 and T22583-R1 along with the first amino acid substitution at residue 144, a glycine to alanine substitution, with primers r431GA-F and r431GA-R to create an intermediate plasmid.The mutations were confirmed in the intermediate plasmid prior to introduction of the final mutations.In the second round of mutagenesis, this intermediate was used as a template to introduce the final amino acid change, a serine to glycine substitution at residue 207 with primers r619621-F and r619621-R to create plasmid pQECmeR-tr.
All mutations were introduced into the respective templates by one cycle of 95°C for 30 seconds followed by 16 cycles of 95°C for 30 seconds, 55°C for 30 seconds, and 68°C for 4 minutes.The amplified product was cooled on ice for 2 minutes before Dpn-I digestion of parental DNA at 37°C for 1 hour.Each product was transformed into JM109 or XL1-Blue (Agilent) and the transformants were selected on LB agar supplemented with ampicillin (100 μg/mL).The specific mutations were confirmed by sequencing with primers pQETypeIII/IV-F and pQEReverse-R (Table 1).
For functional analysis, EMSA was used to assess the ability of the mutated versions of CmeR to bind to the cmeABC promoter.Binding by the purified mutant proteins rCmeR-tr, rCmeR-R, rCmeR-K or rCmeR-IK was compared to binding by the rCmeRSS protein.The 11168 cmeABC promoter probe (0.05 pmol) was incubated with 0, 60, 120, and 180 ng of rCmeRSS or one of the mutant rCmeR proteins in 22μL of reaction buffer according to the method of Alekshun et.al. and Lin et. al. 2005 [12, 38].The reaction mixtures were incubated for 30 minutes at room temperature and Promega DNA loading buffer was added to each reaction.Electrophoresis, transfer, and detection were performed using the same methods as described for the EMSA assay.

Antimicrobial susceptibility testing
Agar dilution test was performed in MH agar and MH agar supplemented with 12,500 μg/mL of sodium choleate according to the CLSI-recommended method [31].Ciprofloxacin, kanamycin, erythromycin, tetracycline, ampicillin, clindamycin, chloramphenicol, ethidium bromide, sodium choleate, cholic acid and taurocholic acid were tested.In addition to the 21 WEL and 43 OEL clinical isolates, 3 laboratory strains classified as WEL isolates, NCTC 11168, ATCC33560, and 81-176, were also tested.At least 2 experiments were performed for each isolate.Distribution of the minimum inhibitory concentration (MIC) around the median was tested with the Brown Forsythe test (SAS version 9.2) comparing the 43 OEL and 24 WEL isolates.The significance level set at 0.05.
Total cell counts for each set of cultures (36 parallel cultures) were estimated from 5 random wells.The 5 wells were selected using the Random Integer Set Generator (Random.org).From each of these wells, a 10 μL sample of culture was removed and serially diluted to 10 −7 .A 90 μL sample from the 10 −5 , 10 −6 , and 10 −7 dilutions was spread to MH agar.Plates were incubated for 2 days for total CFU counts.Counts from the 5 random wells were averaged to determine the total count for each set of 36 parallel cultures.
To determine the number of spontaneous mutants, all 36 wells were plated to MH agar containing 4 μg/mL of ciprofloxacin.Plates were checked after 2 days for colony size and incubated for an additional day to ensure colonies were large enough for counting.Mutation rate was calculated by the Ma-Sandri-Sarkar Maximum Likelihood Estimator method using the Fluctuation Analysis Calculator [47].Mutations per culture (m) for WEL and OEL were transformed by the natural logarithm (ln(m)) and compared with the Student's t-test using GraphPad InStat (Version 3.06) with the significance level set at 0.05.

In vitro ciprofloxacin treatment
Inactivation of CmeABC reduced the emergence of ciprofloxacin-resistant mutants in C. jejuni under antibiotic selection, while inactivation of cmeR (resulting in overexpression of cmeABC) increased mutant emergence, suggesting that expression levels of CmeABC influences the emergence of antibiotic resistant mutants.To assess if differential expression of cmeABC in naturally occurring isolates affects their adaptation to antibiotic treatment, we examined the emergence frequencies of ciprofloxacin-resistant mutants in isolates of different CmeABC phenotypes.Three WEL (CB6:8, F15871, CT10:18) and three OEL isolates (T37957A, CT7:20, CB8:14) were used for this experiment, which were cultured on non-selective media (MH agar) and then treated with 4 μg/mL of ciprofloxacin using the method of Han et.al. 2008 [48] with some modifications.Briefly, MH broth supplemented with 4 μg/mL of ciprofloxacin was inoculated to an initial concentration of 10 7 CFU/mL in a 20 mL of culture with 3 replicates per isolate.Cultures were incubated for 3 days.Samples (0.5 mL) were taken on days 0, 1, 2, and 3post inoculation for enumeration of ciprofloxacin-resistant mutant and total cell counts.Total counts were cultured on MH agar according to the plate drop method [49] for days 0, 1, 2, and 3.For days 0 to 2 mutants were counted by direct plating 100 μL of culture onto MH-ciprofloxacin (4 μg/mL) agar in duplicate and according to the plate drop method [49] onto MH-ciprofloxacin (4 μg/mL) agar.For day 3, total plate counts and mutants were all determined by the plate drop method.The serial dilutions used in the plate drop method were dilutions 10 −2 to 10 −7 for day 0 and 10 −3 to 10 −8 for days 1 to 3. Two independent experiments were performed for each isolate.The cell counts were calculated for each culture and log transformed.Data analysis was performed with Graph Pad Prism (Version 6.0c) with multiple unpaired t-tests and Holm-Šídák method for multiple comparisons.The significance level was set at 0.05.
In the second experiment, cultures were inoculated to an initial density of 10 6 CFU/mL.Samples were collected in the same manner as the first experiment.For each day total counts were cultured on MH agar according to the plate drop method [49].Mutants were counted by direct of plating 100 μL of culture to MH-ciprofloxacin (4 μg/mL) agar in duplicate and serial dilutions for plate drop method [49] onto MH-ciprofloxacin (4 μg/mL) agar.The serial dilutions used in the plate drop method were dilutions 10 −2 to 10 −7 for day 0 and 10 −3 to 10 −8 for days 1 to 3. Two independent experiments were performed for each isolate.Data analysis was performed as described in the prior experiment.

Phenotypic classification of isolates
Initial screening for phenotypic classification of CmeABC expression was done through immunoblotting (Fig 1) and real time RT-PCR for expression of cmeB.Analysis of the 64 C. jejuni isolates for cmeB expression identified 43 isolates with overexpression levels (OEL) of CmeABC and 21 isolates with wild-type expression levels (WEL) of CmeABC.The region spanning from cmeR to cmeA was sequenced for all OEL isolates, 4 of the WEL isolates, and C. jejuni strain 81-176 to identify genetic mutations that were potentially associated with the differential cmeABC expression (S1 Table ).All isolates except 1 OEL isolate harbored mutations in the sequenced region compared to the same region in strain NCTC 11168.
To refine the identification of mutations mediating differential cmeABC expression, the cmeABC promoter was analyzed for mutations unique to the OEL isolates and cmeR was analyzed for DNA polymorphisms resulting in amino acid changes.The CmeR binding site of the cmeABC promoter contains an A to T substitution at base 10 of the 16 base inverted repeat in strain 81-176 (Fig 2A) that is considered a natural variation (81-176 variation) [6].Among the clinical isolates, 34 carried this mutation.Isolates that contained only the 81-176 variation in the CmeR binding site were excluded from further analysis.There were 14 OEL isolates carrying mutations within the CmeR binding site of the cmeABC promoter other than the 81-176 A to T substitution (Fig 2A).These isolates also carried amino acid mutations in CmeR.All of the observed amino acid mutations were also seen in isolates with no mutations in the cmeABC promoter.These 14 isolates were categorized as cmeABC promoter mutants.
Analysis of cmeR for DNA polymorphisms found numerous mutations leading to amino acid changes in CmeR.DNA polymorphisms that did not result in amino acid changes in CmeR were excluded.Three isolates, M63885, T37957A, and CT9:7 harbored unique mutations in CmeR (Table 3).Other OEL isolates carried amino acid changes in CmeR that were also seen in WEL isolates.Five OEL isolates and one WEL isolate with a unique combination of substitutions at residues 144, 183 and 207 were also selected for analysis (Table 3).None of these isolates, except X7199, had mutations in the cmeABC promoter.Together these 9 clinical isolates were categorized as cmeR mutants.For each probe, the amount of rCmeRSS used for the each reaction was 0 (lanes 1 and 5), 60 (lanes 2 and 6), 120 (lanes 3 and 7), and 180 ng (lanes 4 and 8), respectively.The rCmeRSS-DNA complexes are indicated with a "C" and the unbound promoter probe is indicated with a "P".doi:10.1371/journal.pone.0131534.g002substitution at base 2 of the inverted repeat and 7 isolates with an A to G substitution at base 14.The 81-176 variation (A to T at base 10) is found in 7 isolates carrying the substitution at base 14 and one isolate carrying the substitution at base 2. Two isolates carrying the 81-176 variation and a 5 base deletion in the cmeABC promoter were detected.The last 3 bases of the CmeR binding site and the following 2 bases are absent in isolate CT9:20 (Fig 2A ).Isolate X7199 has a 5 base deletion 5' to the CmeR binding site in addition to an A to G substitution at base 14 of the CmeR binding site (Fig 2A).These 14 isolates are considered cmeABC promoter mutants.

Mutation of the CmeR binding site affects cmeABC expression in clinical isolates
From the cmeABC promoter mutants, 6 representative sequences were selected for analysis by EMSA and transcriptional fusion (Fig 2A).The promoters from the isolates X7199 and CT9:20 were selected for their unique deletions.The M32506 and CT3:7 promoters both contain the A to G mutation at position 14, while the CT3:7 promoter also contains the 81-176 variation.The CT1:1 and CT1:9 promoters carry the G to A mutation at position 2.
The ability of CmeR to bind to the mutant cmeABC promoter sequences was assessed by EMSA.The selected mutant cmeABC promoter sequences from clinical isolates were paired with either the 11168 promoter (from strain NCTC 11168) ( To further quantify the effect of the promoter mutations on cmeABC expression, transcriptional fusion of the mutant cmeABC promoters to the promoterless lacZ gene was performed in the presence and absence of CmeR.In the 81-176 wild-type background, CmeR is expressed and binds to the cmeABC promoter, repressing its transcription.The 81-176ΔcmeR strain is an isogenic mutant that does not express CmeR, resulting in a loss of repression for the cmeABC promoter.Without this repressor, cmeABC is overexpressed.Transcription from the 11168 promoter in the 81-176 wild-type background (CmeR is present) was defined as the basal level of cmeABC transcription and used as a control.In the presence of CmeR (81-176 wild-type background) (Fig 3A), all examined cmeABC promoters were transcribed.Compared with the base (11168 promoter), expression from the 81-176 promoter, CT1:1 promoter, CT1:9 promoter, CT3:7 promoter, M32506 promoter, X7199 promoter was 2.4-fold (p = 0.0096), 4.6-fold (p = 0.0232), 5.5-fold (p = 0.0073), 5.6-fold (p = 0.0425), 6.3-fold (p = 0.0036), and 5.4-fold (p = 0.0150) higher, respectively.These increases are indicative of decreased repression by CmeR due to the mutations in the cmeABC promoters.Notably, the CT9:20 promoter with the deletion of the last 3 bases of the CmeR binding site, had the highest increase in transcription at 8-fold (p = 0.0208) over the basal levels, consistent with its most obvious reduction in binding by CmeR on EMSA.
Transcription from the CT1:1 and CT1:9 promoters, which had the same sequence (Fig 2A) but was carried in different isolates, was not significantly different when compared to each other.This indicates that choosing a single representative isolate is sufficient to evaluate the other isolates in the group.The transcription from the X7199 and CT3:7 promoters, which had the same two point mutations in the CmeR-binding site (Fig 2A ), was not significantly different between the two promoters, although there was an additional mutation (5-base pair deletion) upstream of the CmeR binding site in the X7199 promoter.These results suggest that the substitutions in the CmeR binding site alone (represented by CT3:7) is sufficient to alter expression of cmeABC.
In the absence of CmeR, transcription for the 11168 and 81-176 promoters were elevated, and the levels of transcription among the promoters except X7199 were not significantly different (p > 0.

Varied expression levels of cmeR in clinical isolates
Several DNA polymorphisms were detected in the cmeR gene of the clinical isolates, resulting in amino acid changes in this regulatory protein (Table 3).Immunoblotting of whole cell proteins was performed to determine the CmeR expression level from 8 OEL isolates and 1 WEL isolate harboring representative mutations.The anti-CmeR antibody detected the CmeR protein from 8 of the 9 cmeR mutants (Fig 4A).The remaining cmeR mutant isolate (CT2:2) did not produce a band reactive with the antibody (Fig 4B , lane 2), suggesting that the CmeR protein was not translated in this isolate.CT2:2 contained a T insertion after base 583 in cmeR, resulting in a frame shift and premature truncation (Table 3).Additional immunoblotting failed to detect any portion of the truncated CmeR from CT 2:2.To confirm the results of immunoblotting, cmeR expression was evaluated by real time RT-PCR.Real time expression levels of cmeR varied dramatically among the 9 isolates.Expression ranged from 0.006 to 33 fold of that in NCTC 11168 (Table 3), however this was not correlated with CmeR expression levels on immunoblotting (S2 Table ).The transcript level of cmeR from CT2:2 were negligible at 0.006 fold of that in NCTC 11168, consistent with the lack of protein expression as detected by immunoblotting (Fig 4).
cmeR and Cj0369c form a two-gene operon and share a single promoter located in front of Cj0369c [13].The predicted promoter for the Cj0369c-cmeR operon contains an inverted repeat with two half sites separated by a 12-base spacer that may represent an unknown regulatory mechanism [13].Sequence analysis of this region was performed on several isolates to determine if mutations occurred and if they could be correlated with the varying cmeR expression levels identified by real time RT-PCR.Mutations of the Cj0369c-cmeR promoter were found in some isolates after comparison to the sequence of NCTC 11168 and were divided into 2 groups (S2 Table ).The first group carried a single base deletion one base 5' to the second half site of the inverted repeat in strains 81-176, T37957A, and E46972.The second group, consisting of isolates X7199, W52546, and S13530, contained a T insertion in the second half site of the inverted repeat after the eighth base and a G to A substitution at base 5 of the spacer.The Cj0369c-cmeR promoter from CT2:2 had no mutations, suggesting that transcription from this promoter is unlikely the source of the decreased production as observed by immunoblotting and real-time PCR.
Transcriptional fusion of representative Cj0369c-cmeR promoters from strains 81-176, X7199, and NCTC 11168 was performed to determine if the observed polymorphisms affected transcription.The 81-176, NCTC 11168, and X7199 promoters produced low levels of transcription (mean Miller units in the range of 8-15), which were not significantly different (p > 0.05) (S1 Fig) , suggesting that these mutations were not associated with the expression levels of cmeR.However, this was not consistent with the real time cmeR expression data (Table 3).The mutations observed in the Cj0369c-cmeR promoter for X7199 and the S13530 were identical but, cmeR expression was 10.9 and 0.2 fold of NCTC 11168 respectively.The reason for this discrepancy is unclear.
Truncation, but not amino acid substitution affected CmeR binding to the cmeABC promoter The amino acid changes in CmeR observed in clinical isolates were categorized into 5 groups represented by the 9 cmeR mutant isolates (Table 3).Site-directed mutagenesis and recombinant CmeR production were performed for 3 representative isolates with amino acid substitutions and a single isolate with substitution and truncation of CmeR (Table 3).The 4 proteins produced were named rCmeR-tr, rCmeR-IK, rCmeR-K and rCmeR-R.The rCmeR-IK from M63885 contains 2 amino acid substitutions at residues 6 and 159 replacing threonine with isoleucine and glutamate with lysine, respectively.The rCmeR-K from T37957A and rCmeR-R from CT9:7 contain single amino acid substitutions of the glutamate at residue 84 for lysine in rCmeR-K and the proline 183 residue for arginine in rCmeR-R.The rCmeR-tr from isolate CT2:2 contains a glycine to alanine substitution at residue 144 and a nucleotide insertion after base 583 resulting in pre-mature truncation of CmeR to 193 amino acids (the full-length CmeR is 210 amino acids).The S207G substitution observed in CT2:2 occurs downstream of the truncation and was not expected to affect CmeR function.rCmeR-tr was detected by immunoblotting with the anti-CmeR antibody and presented as a band of 23

Differences in antimicrobial susceptibility
In vitro antimicrobial susceptibilities did not differ between the OEL and WEL isolates in the presence or absence of bile for most of the tested antimicrobials except for chloramphenicol.Without bile, the median MIC for chloramphenicol was 4 μg/mL for both the OEL and WEL isolates (Table 4).However, the distribution of MICs around the median was significantly different between the two phenotypic groups (p < 0.05).The WEL isolates had chloramphenicol MICs of 2 to 4 μg/mL with 67% of the isolates at 4 μg/ml, while the OEL isolates had MICs ranging from 2 to 16 μg/mL with 21% of the isolates at 8 and 16 μg/mL (Table 4).With bile in the testing media, the MIC distribution between the two phenotypes was also significantly different (p <0.05) (Table 4).Although the median remains at 4 μg/mL for both groups, 81% of the OEL isolates were at this MIC, while 58% of the WEL isolate were at this MIC.Both OEL and WEL isolates have MIC ranges of 2 to 8 μg/mL after addition of bile.Compared to the non-bile media, addition of bile shifted the MIC to the upper end for WEL isolates and shifted the MIC to the median the for OEL isolates.

Emergence of ciprofloxacin-resistant mutants
The spontaneous mutation rate to ciprofloxacin was examined for selected isolates using the fluctuation assay.Although there was a trend that the mutation rate was higher in OEL than in WEL, the difference was not statistically significant (p> 0.05) (S3 Fig) .However, OEL isolates showed increased emergence of ciprofloxacin-resistant (Cip R ) mutants during in vitro treatment ( Fig 6).Two experiments were performed with the initial inoculum levels of 10 7 and 10 6 CFU/mL, respectively.For the inoculums at 10 7 CFU/mL, there were no significant difference in the mean numbers of pre-existing Cip R mutants between the WEL and OEL cultures on day 0 (Fig 6A).Cip R populations from both WEL and OEL cultures expanded over days 1 to 3. The mean Cip R mutant populations were 0.9 logs higher for OEL on day 1, 1.5 logs higher on day 2, and 2 logs higher than WEL on day 3.However, the means were not significantly different between the OEL and WEL groups (p > 0.05).
For the inoculum of 10 6 CFU/mL, the means for pre-existing Cip R mutants on day 0 were not significantly different for the WEL and OEL cultures.The mean Cip R mutant numbers on day 1 were 2 logs (p = 0.0175) higher than the mutants in the WEL cultures.This trend continued on days 2 and 3 with OEL means being 3.7 logs (p = 0.0053) and 4.2 logs (p = 0.0016) higher than the WEL means, respectively.These results indicate OEL cultures produced significantly higher numbers of Cip R mutants than the WEL cultures during ciprofloxacin treatment.

Discussion
This study demonstrates that differential expression of CmeABC naturally occurs in Campylobacter isolates derived from different host species.The variable expression was linked to multiple mechanisms including mutations in the cmeABC promoter region and coding sequences of CmeR as well as decreased expression of CmeR, the repressor for the cmeABC operon.Additionally, there are unidentified mechanisms that also modulate cmeABC expression as some OEL isolates did not have any mutations in the known regulatory elements for cmeABC.Differential expression of cmeABC was associated with altered susceptibility to chloramphenicol and enhanced the emergence of Cip R mutants under selection with ciprofloxacin.These findings suggest that differential expression of CmeABC may be selected under natural conditions and may facilitate Campylobacter adaptation to various environments.
Of the 64 isolates examined in this study, 43 (67%) were phenotypically classified as OEL isolates.CmeABC is normally repressed by CmeR, which binds to the promoter region of cmeABC [12].Thus, mutations in CmeR and/or the cmeABC promoter sequence were investigated to determine the genetic basis associated with overexpression of cmeABC in these clinical isolates.Those mutations that occurred in the CmeR binding site or resulted in amino acid changes in CmeR and were absent from the majority of the WEL isolates were selected for detailed analysis.Additionally, those isolates harboring the same mutation as the one in the CmeR binding site of strain 81-176 were excluded from analysis as this mutation has been characterized previously [12].Using these selection criteria, we selected 14 isolates harboring promoter mutations and 9 isolates harboring CmeR mutations for detailed analysis.Most of the detected amino acid substitutions in CmeR did not affect its function as determined by EMSA.However, a single nucleotide insertion at the 3'end of the cmeR gene resulted in a frame-shift and presumably led to truncation of the CmeR protein in CT2:2.CmeR was not detected by immunoblotting in CT2:2 (Fig 4B ) and the cmeR transcript level was also significantly reduced.Additionally, CT2:2 did not contain any mutations in the predicted Cj0369c-cmeR promoter region, excluding the possibility that lack of cmeR expression was due to altered transcription initiation.These findings suggest that the single nucleotide insertion could have destabilized the cmeR transcript and/or the frame shift rendered the CmeR protein unstable in C. jejuni, leading to the lack of CmeR in this isolate.However, a recombinant version of the truncated CmeR (rCmeR-tr) was successfully generated (Fig 5A ), suggesting that this truncated CmeR is stable in the E. coli host.Interestingly, rCmeR-tr failed to bind to the promoter DNA of cmeABC as determined by EMSA (Fig 5B ), suggesting that even if this truncated version is made in C. jejuni, it would not be able to control the expression of cmeABC.The lack of CmeR production and the inability of the truncated CmeR to bind to promoter DNA fully explain the overexpression of cmeABC in isolate CT2:2.
It is interesting that rCmeR-tr lost the ability to bind DNA despite the fact that the truncation occurred in the C-terminal end of CmeR and the DNA-binding domain remained intact.This could be explained using the known structural information of CmeR.CmeR functions as dimer in vivo and the crystal structure of CmeR identified that α helices 6, 8, 9, and 10 are involved in dimer formation [12,15,50].The truncation in rCmeR-tr occurred in the α10 helix of CmeR.Thus, the truncation in rCmeR-tr likely affects dimer formation and ultimately the function of CmeR.This result suggests that the C-terminal sequence of CmeR is also important for its interaction with target DNA.Recently, C. jejuni ATCC 33560, a quality control strain used for antimicrobial susceptibility testing in C. jejuni, was found to contain a frame shift mutation in cmeR, which led to truncation of the CmeR protein [51].The truncation occurs in α helix 8 of CmeR and presumably results in non-functional CmeR [51].Together, these findings indicate that frame-shift mutations in CmeR occur under natural conditions.As CmeR is a pleiotropic regulator (regulating other genes in addition to cmeABC) [13], truncation of CmeR likely affects multiple functions in C. jejuni.
The majority of the mutations in the cmeABC promoter were found to affect cmeABC expression.The cmeABC promoter contains a 16 base inverted repeat that serves as the specific binding site for CmeR [12].Mutations in the CmeR binding site within the cmeABC promoter have been described previously after in vitro stepwise selection with erythromycin [14] and ciprofloxacin [12].This study is the first to describe the occurrence of this type of mutation in C. jejuni isolates from natural sources and various hosts including humans, turkeys, and chickens.These mutations inhibited binding by CmeR, resulting in increased transcription from the cmeABC promoter.This was demonstrated by reduced binding of the mutant promoters by CmeR on EMSA (Fig 2B, panels I, II, III,V, and VI) and increased transcription of cmeABC as determined by transcriptional fusion assays (Fig 3).In addition to the substitution, deletions within the CmeR binding site were found in one isolate, CT9:20.This promoter showed the greatest inhibition of CmeR binding on EMSA and the largest increase in transcription in the presence of CmeR (Fig 3A).These findings indicate that mutations in the cmeABC promoter commonly occur and these mutations influence the expression of this multidrug efflux pump.
Multiple mutations were also identified in a single isolate.For example, isolate X7199 harbored a substitution in the CmeR binding site and a 5 base deletion upstream of the CmeR binding site (Fig 2A).This isolate also contained mutations in cmeR (Table 3).The cmeABC .This discrepancy between the results of EMSA and transcriptional fusion suggests that the EMSA assay has a lower sensitivity than the transcriptional fusion assay, or alternatively, there is another regulatory mechanism that also modulates cmeABC expression.Indeed, when the X7199 promoter was examined in the absence of CmeR using transcriptional fusion (Fig 3B ), its expression was significantly increased compared to its own expression level in the presence of CmeR (Fig 3C).This result is consistent with the EMSA result and suggests that the X7199 promoter is still under the repression by CmeR.
In some isolates, the OEL phenotype was not linked to the known mechanisms modulating cmeABC expression as there were no mutations detected in CmeR or the promoter region.Additionally, even though some isolates harbored mutations in CmeR, these mutations did not affect the function of CmeR.Furthermore, the regulation of the X7199 promoter cannot be fully explained by a CmeR-only-dependent mechanism.These findings suggest that there may be additional regulatory mechanisms modulating cmeABC expression.Previously, Lin et.al.
2005 also described a CmeR-independent mechanism modulating cmeABC expression [16].Bile is a known inducer of CmeABC and mediates increased expression by altering the confirmation of CmeR, resulting in disassociation of CmeR from the cmeABC promoter [15,16].It was noticed that in the absence of CmeR, the expression of cmeABC was further induced by taurocholate, suggesting this bile compound induced expression of cmeABC through a CmeRindependent mechanism [16].Taken together, observations from this study and previous reports suggest that multiple mechanisms modulate the expression of cmeABC.
`The functional consequence of cmeABC overexpression was evaluated in relation to antimicrobial treatments.Antimicrobial susceptibility was unaffected by cmeABC overexpression for most of the tested antibiotics, with the exception of chloramphenicol.This was not surprising as a previous study using genetic manipulation revealed that overexpressing cmeABC by inactivating cmeR had a modest effect on MICs, but inactivation of cmeABC significantly increased the susceptibility of C. jejuni to antibiotics [6,12].For chloramphenicol, the OEL isolates displayed a larger range of MIC values than WEL isolates, with more MICs distributed above the median value (Table 4).This suggests that overexpression of CmeABC had an effect on the MIC distribution of chloramphenicol.Interestingly, when chloramphenicol MIC was measured in the presence of bile, the MICs of the WEL isolates shifted above the median value, while the MICs of the OEL isolates shifted toward the median value.This difference is probably due to the fact that bile is an inducer for cmeABC and the possibility that induction was more prominent in the WEL isolates than in the OEL isolates.For the OEL isolates, cmeABC was already overexpressed due to less inhibition by CmeR or other unidentified mechanisms.Thus, the bile-mediated induction through CmeR is expected to be less effective in the OEL isolates than in the WEL isolates.
Fluoroquinolones resistance in Campylobacter is mediated by DNA gyrase mutations and the function of CmeABC [6,8,9].These two mechanisms function synergistically in mediating resistance to fluoroquinolones [9].Additionally, CmeABC promotes the emergence of fluoroquinolone-resistant mutants under antibiotic selection [8,23].In this study, we examined the correlation between the OEL phenotype and ciprofloxacin resistance.It was found that the basal spontaneous mutation rate was not affected by overexpression of cmeABC as measured by the fluctuation assay.However, the OEL isolates showed higher level of emergence of CipR mutants under antibiotic selection (Fig 6).This was consistently shown by using two inoculation doses (106 and 107 CFU/mL).For the 107 CFU/mL inoculum, the difference between the OEL and WEL groups were obvious, but were not statistically significant.For the 106 CFU/mL inoculum, the numbers of CipR mutants from the OEL isolates were significantly higher than those from the WEL mutants.The lack of statistical significance with the 107 CFU/mL inoculum was probably due to the presence of pre-existing CipR mutants in the inoculum (Fig 6) that somewhat reduced the difference between the OEL and WEL groups.Thus, reducing the inoculum to 106 CFU/mL allowed clear detection of differences between the two groups.These results suggest that the OEL phenotype may facilitate Campylobacter to adapt to fluoroquinolone treatment by promoting the emergence of resistant mutants.This finding has practical implication as fluoroquinolone antibiotics are used for both human medicine and animal production.Thus, the detection of a large number of C. jejuni with an OEL phenotype from different host species might be the result of antibiotic usage that has served as a selection force for the OEL phenotype.
In summary, this study reveals that overexpression of CmeABC commonly occurs in C. jejuni isolates derived from various host species.The overexpression is mediated by multiple mechanisms including mutations in the cmeABC promoter sequence and in the CmeR coding sequence.Additionally, results from this study suggest that there are other unidentified mechanisms that modulate the expression of CmeABC.Overexpression of cmeABC promotes the development of resistant mutants upon treatment with fluoroquinolone antibiotics and may contribute to the survival and persistence of C. jejuni in animal reservoirs where antibiotics are commonly used.These findings provide new insights into the adaptive mechanisms of C. jejuni and further highlight the potential to control antibiotic resistant Campylobacter by targeting CmeABC.

Fig
Fig 2A illustrates the region of the cmeABC promoter containing the CmeR binding site.Two categories of substitutions were observed in the CmeR binding site: 5 isolates with a G to A

Fig 3 .
Fig 3. Effect of various mutations in CmeR and the promoter region on transcription of cmeABC.The names of the promoters used in the transcriptional fusions and β-galactosidase assays are indicated under each panel.Each promoter was assayed in the wild-type 81-176 background (A) and the 81-176ΔcmeR background (B).The data represent means with standard deviation from three independent experiments.The relative difference in transcription (fold change) due to repression by CmeR for each promoter is shown in (C) and was determined by comparison of transcription in the absence of CmeR (B) to the presence of CmeR (A).The unpaired Student's t-test with Welch's correction was used for comparison of the means with significance set at 0.05.doi:10.1371/journal.pone.0131534.g003 05) (Fig 3B).Comparisons for each promoter in the presence (WT) and absence (ΔcmeR) of CmeR showed that deletion of CmeR significantly increased expression (p <0.05) of the cmeABC promoters from 11186 and 81-176 (Fig 3C).Transcription in the absence of CmeR (81-176 ΔcmeR background) showed a 6.3-fold increase for the 11168 promoter (p = 0.014), a 2.5-fold increase for the 81-176 promoter (p = 0.0039), and a 2.6-fold increase for the X7199 promoter (p = 0.0367) (Fig 3C).The smaller change for the 81-176 and X7199 promoters compared to the 11168 promoter is likely due to already elevated expression in the wild-type background mediated by the CmeR binding site mutations (Fig 3A).Compared to their transcription in the cmeR wild-type background, the expression of CT1:1, CT1:9, CT3:7, M32506, and CT9:20 promoters were not significantly different (p > 0.05), suggesting that inactivation of cmeR did not further increase transcription from these promoters (Fig 3C).

Fig 6 .
Fig 6.Emergence of ciprofloxacin-resistant mutants from WEL (circle) and OEL (triangle) isolates during treatment with ciprofloxacin.In panel A, the experiment was performed with an initial inoculum of 10 7 CFU/ml of each isolate, while in panel B, the inoculum was 10 6 CFU/mL for each isolate.The culture medium was MH broth containing 4 μg/mL of ciprofloxacin.Three WEL and OEL isolates were used in each experiment with cultures prepared in triplicate.Each point represents the number of ciprofloxacin-resistant mutants from a single culture.Bars represent mean log 10 CFU/mL for each group.Means for each phenotypic group were compared for each day with multiple unpaired Student' t-tests and Holm-Šídák method for multiple comparisons.The significance level was set at 0.05.doi:10.1371/journal.pone.0131534.g006 promoter in this isolate was identical to the promoter in CT3:7 except for a 5-base pair deletion upstream of the CmeR binding site (Fig 2A).While the CT3:7 promoter demonstrated reduced binding to CmeR on EMSA (Fig 2B, panel V), the X7199 promoter had similar binding as the 81-176 cmeABC promoter (Fig 2B, panel IV).However, both the X7199 and CT3:7 cmeABC promoters demonstrated similar, elevated expression by transcriptional fusion assay compared with the 11168 promoter in the presence of CmeR (Fig 3A)

Table 1 .
Oligonucleotide primers used for PCR and real time RT-PCR.1

Table 2 .
Bacterial strains or plasmids used in this study.

Table 3 .
Mutation, expression, and phenotypes of CmeR in selected C. jejuni isolates.

Table 4 .
Chloramphenicol MIC distribution (% for each MIC) among the tested isolates.