Genotypic and Phenotypic Properties of Cattle-Associated Campylobacter and Their Implications to Public Health in the USA

Since cattle are a major source of food and the cattle industry engages people from farms to processing plants and meat markets, it is conceivable that beef-products contaminated with Campylobacter spp. would pose a significant public health concern. To better understand the epidemiology of cattle-associated Campylobacter spp. in the USA, we characterized the prevalence, genotypic and phenotypic properties of these pathogens. Campylobacter were detected in 181 (19.2%) out of 944 fecal samples. Specifically, 71 C. jejuni, 132 C. coli, and 10 other Campylobacter spp. were identified. The prevalence of Campylobacter varied regionally and was significantly (P<0.05) higher in fecal samples collected from the South (32.8%) as compared to those from the North (14.8%), Midwest (15.83%), and East (12%). Pulsed Field Gel Electrophoresis (PFGE) analysis showed that C. jejuni and C. coli isolates were genotypically diverse and certain genotypes were shared across two or more of the geographic locations. In addition, 13 new C. jejuni and two C. coli sequence types (STs) were detected by Multi Locus Sequence Typing (MLST). C. jejuni associated with clinically human health important sequence type, ST-61 which was not previously reported in the USA, was identified in the present study. Most frequently observed clonal complexes (CC) were CC ST-21, CC ST-42, and CC ST-61, which are also common in humans. Further, the cattle associated C. jejuni strains showed varying invasion and intracellular survival capacity; however, C. coli strains showed a lower invasion and intracellular survival potential compared to C. jejuni strains. Furthermore, many cattle associated Campylobacter isolates showed resistance to several antimicrobials including ciprofloxacin, erythromycin, and gentamicin. Taken together, our results highlight the importance of cattle as a potential reservoir for clinically important Campylobacter.


Introduction
Campylobacteriosis is estimated to affect over one million individuals in the United States annually, with symptoms ranging from mild diarrhea to more serious neuropathies [1]. Of major concern are C. jejuni, a species responsible for the majority of human campylobacteriosis, and C. coli, which has been exhibiting increased resistance to antimicrobials [2,3]. Both C. jejuni and C. coli can readily and asymptomatically colonize major food-animals such as poultry and cattle, subsequently contaminating foods of animal origin including meats and dairy products [4]. Therefore, both species highlight the potential public health impact of Campylobacter contamination of food animals.
Although it is known that chickens constitute a major reservoir for Campylobacter spp., the occurrence of these pathogens in other food animals such as cattle and its potential impact on human health remain largely uncharacterized. The latter can be partially attributed to the sporadic nature of Campylobacter infections and difficulties in isolating these fastidious pathogens. However, since cattle are a major source of food and the multi-faceted cattle industry engages people from farms to processing plants and meat markets, it is conceivable that both live cattle and contaminated cattle products could contribute significantly to Campylobacter infections in humans. Furthermore, cattle-associated Campylobacter also pose an additional indirect public health risk [5,6]. For example, contamination of surface and ground water may occur with waste run-off from cattle farming and processing operations. Thus, it is important to further investigate the epidemiology of Campylobacter in the cattle population in order to assess associated risks to public health.
Recent studies have shown that the contributions of non-poultry associated Campylobacter to human infections were considerable and warrant investigation [7,8]. For example, evidence collected using Multilocus Sequence Typing (MLST) showed that cases of human Campylobacter infections in Finland could be attributed equally to cattle and poultry [9]. Moreover, another study reported that cattle were the source of human infections in 35% of the cases examined in Lancashire, England [8]. This is not surprising since up to 80% of cattle herds and 40-60% of the individual animals shed Campylobacter [10,11,12]. Consequently, the role of cattle as reservoirs for these pathogens might be important for understand-ing the epidemiology of Campylobacter infections. However, particularly in the US, the prevalence of Campylobacter in cattle, their characteristics and relationship to isolates from humans have not been extensively described in the peer reviewed literature.
Host cell invasion and intracellular survival and resistance to antibiotics are important characteristics that affect Campylobacter infections in humans. However, the potential role of these properties in facilitating human infections with Campylobacter isolated from cattle is not clear. Although epidemiological studies deploy typing analysis to glean information about the relationships between cattleassociated Campylobacter and clinically-important human isolates, the capability of cattle-associated Campylobacter to invade and persist in the human host might require further assessment, especially when previously unknown sequence types are identified. This also applies to antimicrobial resistance properties, which have been posing a serious concern in Campylobacter collected from animal hosts along with the possibility of the transmission of these isolates to humans through the food chain. Therefore, epidemiological studies concerning cattle-associated Campylobacter would gain from attempts to amend molecular typing analysis with in vitro invasion studies using human intestinal cell lines and phenotypic assays for determining antibiotic resistance.
Since molecular typing data of cattle-associated Campylobacter in the United States are limited and little is known about their impact on human health, it is important to adopt a multiphasic approach to characterize cattle-associated Campylobacter by using a combination of molecular typing and in vitro assays. Therefore, in this study, we determined the genotypes of C. jejuni and C. coli isolated from cattle slaughtered for meat purpose in different geographical locations in the U.S. and investigated their antimicrobial susceptibility profiles as well as virulence associated phenotypes such as their potential for invasion and persistence in human intestinal epithelial cells.

Materials and Methods
Distribution of sampling sites and collection and processing of fecal samples from beef cattle A total of 944 fresh fecal samples (10 g each) were collected during the summer and early fall of 2008 from cattle presented to slaughter which included both feed-lot (n = 482) as well as mature cows and bulls (n = 462) culled from milking and breeding herds ( Table 1). The samples were collected from colon on the conveyer belt during the normal harvest process from slaughtered cattle. To account for potential spatial heterogeneity between different beef processing plants, sampling efforts included seven plants that were distributed across four major geographical locations [North (N), East (E), Midwest (M), and South (S)] in the U.S. The time interval of sample collection per plant spanned two consecutive days and was divided into four periods during which at least four different lots of cattle were sampled in order to capture the genetic diversity of cattle-associated Campylobacter. To further limit sampling bias, the number of samples collected per plant was calculated in terms of the total number of animals expected for slaughter on each day, and a statistically calculated sample size to detect at least a single sequence type of Campylobacter should it be present in each group of animals at a prevalence of greater than 2.5% with a confidence of 95%. Samples were stored on ice immediately after collection and shipped overnight to the laboratory for Campylobacter isolation.
Isolation and identification of Campylobacter species from fecal samples To isolate Campylobacter spp., 1 g of each fecal sample was enriched in Preston broth for 48 h at 42uC under microaerobic conditions (5% O 2 , 10% CO 2 , and 85% N 2 ) [13]. From the enrichments, an inoculum (100 ml) was spread onto modified Cefoperazone Charcoal Deoxycholate Agar (mCCDA) plates, which were then incubated for an additional 48 h at 42uC under microaerobic conditions [14]. Three to five colonies suspected as Campylobacter were selected from each plate and sub-cultured onto Muller-Hinton agar plates, after which DNA was extracted from each isolate using the Genomic DNA Purification Kit (Epicenter, Madison, WI) as described by the manufacturer. The DNA was then quantified using the NanoDrop 1000 Spectrophotometer V3.7.1 (Fisher Scientific, Pittsburgh, PA) and subjected to speciesspecific PCR analysis to confirm the identity of the isolates as described elsewhere [15]. PCR analysis targeted a 16S rRNA gene fragment (F) 59-ATCTAATGGCTTAACCATTAAAC-39 and (R) 59-GGACGGTAACTAGTTTAGTATT-39, mapA, (F) 59-CTA-TTTTATTTTTGAGTGCTTGTG-39 and (R) 59-GCTTTAT-TTGCCATTTGTTTTATTA-39 and ceuE, (F) 59ATTTGAAA-ATTGCTCCAACTATG-39 and (R) 59-TGATTTTATTATTT-GTAGCAGCG-39 which indicated the specific detection of Campylobacter spp., C. jejuni, and C. coli, respectively [16,17]. All PCR products were resolved on a 1.5% agarose gel containing 0.5 mg/ml of ethidium bromide. The size of the PCR products was determined using a 1 Kb DNA ladder and detection was confirmed by comparison to PCR products generated from C. jejuni 81-176 (wild-type strain) and C. coli (ATCC 33559), which were used as positive controls in all PCR analysis. Negative controls (reactions with no DNA templates) were included in all PCR analyses to ensure specific product amplification.

Pulsed Field Gel Electrophoresis (PFGE) of Campylobacter isolates
To determine their genotypic relatedness, the Campylobacter isolated from fecal samples were analyzed using PFGE analysis as described in Ribot et al. [18]. Briefly, C. jejuni and C. coli isolates were harvested from MH agar plates and suspended to an OD 610 of 1.4 in 1X PBS. To prepare agarose plugs, OD adjusted suspensions were gently mixed with 1% SeaKem Gold agarose (SKG, Fisher scientific, Pittsburgh, PA) that was pre-melted in TE (10 mM Tris, 1 mM EDTA, pH 8.0). The plugs were then incubated with shaking (200 rpm) in lysis buffer [50 mM Tris, 50 mM EDTA (pH 8.0), 1% sarcosine, 0.1 mg ml 21 of proteinase K] for 1 h at 55uC. After lysis the plugs were washed four times and suspended in 5 ml of fresh TE. The plugs were then sliced and digested overnight with SmaI at room temperature. The digested slices were loaded onto a 1% SKG agarose gel and DNA fragments were separated by electrophoresis for 20 h using the CHEF Mapper system (Bio-Rad, Hercules, CA) followed by post staining with ethidium bromide. The resulting PFGE patterns were documented and analyzed using the BioNumerics 5.1 software (Applied Maths Inc, Austin, TX). Similarity and clustering analysis of the PFGE patterns were performed using the Dice Coefficient and the unweighted pairgroup method with arithmetic averages (UPGMA) with optimization of 1% and position tolerance of 1.5% [7], respectively. The PFGE analysis was also performed on C. jejuni 81-176 and C. coli (ATCC 33559), which were used as controls for facilitating gel to gel comparison. Furthermore, Lambda Ladder PFG Marker (50-1,000 kb, New England BioLabs, Ipswich, MA) was used as a molecular marker. A cut-off similarity value of 75%, was used to determine the sub-clusters of macro restriction profiles (MRPs).
Determining the antimicrobial-resistance properties of the Campylobacter isolates . Campylobacter isolates were grown to mid-log phase and then suspended in Mueller-Hinton broth to achieve an OD 600 of 0.05. For each isolate, one hundred microliter of the suspension were transferred to each well in the Sensititre Campy plates, including a control well that did not contain any antibiotics, while C. jejuni 81-176 and C. coli (ATCC 33559) were used for quality control. The plates were then incubated under microaerobic conditions at 42uC for 24 h after which the minimum inhibitory concentration (MIC) was measured. The MIC for each antibiotic was defined as the absence of bacterial growth in the well with the lowest concentration of the antibiotic and susceptibility was interpreted according to the Clinical and Laboratory Standards Institute (CLSI) [19].

Multilocus sequence typing (MLST) of Campylobacter isolates
To investigate the clonality of the Campylobacter isolates and to assess similarity to strains associated with human infections, MLST analysis was performed on a total of 112 isolates (62 C. jejuni and 50 C. coli). Nine C. jejuni isolates were not typed because either they yielded incomplete profiles or last culturability upon storage. However, since the C. coli isolates showed limited genetic diversity using MLST, the tested number of isolates (n = 50) was considered adequate to meet the objectives of this study. These isolates were selected to be representative of different PFGE clusters. MLST analysis was conducted as described by Dingle et al. [20]. Briefly, loci from seven housekeeping genes (aspA, glnA, gltA, glyA, pgm, tkt, and uncA) were amplified using gene specific primers by PCR and anticipated sizes of the amplicons were confirmed by agarose gel electrophoresis. The Campylobacter MLST oligonucleotides [20] were obtained from Integrated DNA Technologies (Coralville, IA). PCR products were purified (QIA quick 96 PCR purification kit, QIAGEN, Valencia, CA), sequenced in both directions as described previously [21], and, forward and reverse sequences were then aligned using ClustalW (www.ebi.ac.uk/clustalw). Allelic profiles were determined by performing BLAST analysis using the single-locus query function, while sequence types (STs) were assigned using the allelic profile query function available in the MLST Campylobacter database (http://pubmlst.org/campylo-bacter/). STs were then traced to their respective clonal complexes using BURST at http://pubmlst.org/.

Invasion and Intracellular survival potential of the Campylobacter isolates in INT407 cells
Representative C. jejuni (n = 19) and C. coli (n = 12) that belonged to clonal complexes associated with human infections as well unassigned clonal complexes including the newly identified STs were further tested for their virulence associated phenotypes by assessing their ability to invade and survive within human intestinal epithelial cells. The invasion studies were conducted as described in Konkel et al. [22] and Prasad et al. [23]. Briefly, 10 5 cells ml 21 of INT407 (human embryonic intestine, ATCC CCL 6) were seeded into each well of a 24-well tissue culture plates after suspension in Eagle's Minimum Essential Medium (MEM, Fisher scientific, Pittsburgh, PA) supplemented with 10% fetal bovine serum (FBS, Fisher scientific, Pittsburgh, PA). The plates were then incubated at 37uC in a humidified incubator with 5% CO 2 until semi-confluent mono-layers were obtained. In preparation for infection with Campylobacter, the INT407 mono-layers were washed three times and covered in MEM supplemented with 1% FBS. Similarly, Campylobacter cultures were washed three times and suspended in MEM supplemented with 1% FBS to obtain 10 7 bacteria ml 21 . One ml of bacterial suspension was added to each well containing the INT407 cell monolayer, achieving a 1:100 multiplicity of infection (MOI) then incubated for 3 h. After 3 h of incubation with bacteria, cells were treated with gentamicin (150 mg/ml) and incubated for additional 2 h. The infected monolayers were then washed three times with MEM, lysed using 0.1% Triton X-100 (Fisher scientific, Pittsburgh, PA), serially diluted (10-fold) in MEM and 100 ml of each dilution were spread on MH agar plates. The agar plates were then incubated for 48 h at 42uC under microaerobic conditions, after which colony forming units (CFU) were counted to determine the number of Campylobacter that invaded the monolayers. Each isolate was tested in duplicate per assay, and the experiment was repeated three times on separate occasions. C. jejuni 81-176 (highly invasive) [24], C. jejuni NCTC11168 (poorly invasive) [24] and C. coli (ATCC 33559) were used as controls in all invasion assays.
To assess intracellular survival [22], Campylobacter cultures and the INT407 cells were processed as described above. However, after treatment with gentamicin, MEM containing 3% FBS and bacteriostatic concentration of gentamicin (10 mg ml 21 ) were added to each well to allow strict quantitation of intracellular bacteria. After further incubation at 37uC for 24 h, the monolayers were washed three times, lysed and serially diluted in MEM as described earlier and plated on MH agar to determine CFU of Campylobacter surviving within the intestinal cells. In addition, we also cultured the supernatant of gentamicin treated monolayers to ensure the quality Table 1. Campylobacter species isolated from cattle fecal samples that were collected from four major geographical locations in the USA. of the gentamicin protection assay. The experiment was repeated three times on separate occasions. C. jejuni 81-176, C. jejuni NCTC11168, and C. coli (ATCC 33559) were used as controls in all intracellular survival assays.

Statistical analysis
A Chi square test was used to evaluate data collected from the Campylobacter prevalence analysis. The data derived from cell culture assays were analyzed using one-way ANOVA followed by Tukey's Multiple Comparison Test. A P value of ,0.05 was considered statistically significant for all experiments. Measurements expressed as mean 6 SE (standard error) were averages of at least three replicas.

Results
Occurrence and distribution of Campylobacter spp. in feces sampled from cattle The occurrence and distribution of Campylobacter in cattle feces was investigated in order to determine the role of cattle as a potential reservoir for these bacteria. Campylobacter were detected in 181 (19.2%) out of 944 fecal samples, while a total of 213 isolates were retrieved from the Campylobacter-positive fecal samples. The C. jejuni and C. coli constituted the majority of the isolates and were detected in 71 (7.5%) and 132 (14%) of the samples, respectively (Table 1). There was no significant difference (P,0.12) in the number of C. jejuni isolated from fecal samples obtained from; either feed lot (6.2%) or the cull cattle (8.9%), while C. coli was more frequently isolated from cull cattle (17%) as compared to feed lot cattle (11%) (P,0.01). Furthermore, the occurrence of Campylobacter in cattle feces varied according to geographic location of the sampling sites. Specifically, Campylobacter species were retrieved from 32.8% of the fecal samples collected from the South, which was significantly higher (P,0.05) than those from the North (14.8%), Midwest (15.83%), and East (12%) ( Table 1).
C. jejuni and C. coli strains were genotypically diverse across different geographic regions PFGE analysis was performed on C. jejuni and C. coli isolated from feces in order to determine the genetic diversity of these bacteria and the relationship between the isolates retrieved from different geographic locations. PFGE analysis was successfully performed on 67 out of 71 C. jejuni, while 4 isolates could not be typed using the aforementioned method. Regardless, analysis of the PFGE profiles of the C. jejuni suggested that the isolates possessed diverse genotypes, especially when comparing profiles of isolates belonging to different geographic locations (Fig. 1A). Furthermore, there were 7 main clusters (Fig. 1A) and by using a cut-off similarity value of 75%, profiles of the C. jejuni were classified to15 sub-clusters. With the exception of 4 clusters that included isolates that were collected from 2 to 3 different geographic locations, C. jejuni profiles were mostly found to form geographically homogenous groupings. For example, one cluster was composed of three isolates collected from the East, the Midwest, and the South, respectively, while others included 4 isolates, 3 of which were isolated from the Midwest and one from the South (see arrows in Fig. 1A). Overall, it was interesting to observe that profiles of the isolates from the North did not group with those collected from the East and the South.
Similarly, PFGE analysis was successfully performed on 115 out of 132 C. coli isolates, while the remaining isolates could not be typed using the aforementioned method (Fig. 1B). Unlike the C. jejuni, PFGE profiles of certain C. coli isolated from all four geographic locations were found to share a 100% similarity. Specifically, analysis grouped the C. coli isolates into 4 main clusters and by using a cut-off similarity value of 75%, 21 sub-clusters were identified, 2 of which included identical profiles (100% similar) for isolates from different geographic locations (See arrows Fig. 1B). For example, a cluster was composed of 5 isolates (4 from the North and 1 from the East) that displayed identical profiles, while another cluster similarly contained identical profiles of 4 isolates (3 from the North and 1 from the South). However, other clusters contained profiles of C. coli isolates from different geographic locations that were not identical, yet exhibited high similarity (between 79 to 95%) than that of the cut-off (75%). Similar to PFGE analysis of C. jejuni, many C. coli isolated from the same geographic location showed profiles that tend to cluster together.
Antimicrobial susceptibility of C. jejuni and C. coli isolates To better assess the potential public health impact of the Campylobacter spp. associated with cattle, the isolates were assayed for their potential to resist antibiotics that are of both clinical and veterinary importance. Antimicrobial susceptibility was assessed on 66 (the 5 remaining isolates lost cultivability after storage) C. jejuni isolates using commercially available Sensititre Campy plates. The C. jejuni isolates were resistant to different antimicrobials including, ciprofloxacin (MIC: 4-64 mg ml 21 ), erythromycin (MIC: 32 mg ml 21 ), tetracycline (MIC: 16-64 mg ml 21 ), and clindamycin (MIC: 8-16 mg ml 21 ) (Fig. 2A). However, resistance to tetracycline was observed for the majority of the tested C. jejuni (72.7%), while resistance to nalidixic acid was observed for only 27.3% of the isolates ( Fig. 2A). Furthermore, 10 (15.1%) tested C. jejuni were resistant only to tetracycline, while 34 (51.5%) isolates were resistant to 3 or more antimicrobials, including ciprofloxacin, tetracycline, nalidixic acid, erythromycin, and clindamycin. Overall, a higher percentage of C. jejuni isolates from the South region showed decreased resistance to most of the antimicrobials (See asterisk Table 2). The MRPs and antimicrobial resistance profiles of the C. jejuni isolates analyzed by MLST are summarized in Table S1.
Since, C. coli isolates belonging to different PFGE clusters were highly similar, we tested representative isolates from each cluster for a total of 63 isolates. Although the range of MIC to disparate antibiotics was at instances different from those observed for C. jejuni, the C. coli isolates also exhibited resistance to the several of the antimicrobials tested (Fig. 2B).  (Table 3). Furthermore, resistance to telethromycin was observed for the majority of the tested C. coli (66.7%), while resistance to flourofenicol was observed for only 25.4% of the isolates (Fig. 2A). Of the 63 tested C. coli, 3 (4.8%), 9 (14.3%) and 48 (76.2%) isolates were resistant to only one (e.g. nalidixic acid), two (e.g. ciprofloxacin and nalidixic acid), and three or more antimicrobials, respectively. Furthermore, Figure 1. Dendrogram showing the Pulsed-field gel electrophoresis patterns for SmaI restricted cattle-associated C. jejuni (A) and C. coli (B). Similarity analysis was performed using the Dice coefficient, and clustering was performed by the unweighted pair-group method with arithmetic averages UPGMA (optimization, 1% and position tolerance, 1.5%). Clustering cut-off was 75% similarity, which was represented by dashed line. Numbers on bootstraps represent Cophenetic correlations. N = North, S = South, M = Midwest, and E = East. Cj = C. jejuni, CC = C. coli. doi:10.1371/journal.pone.0025778.g001 a higher percentage of C. coli isolates from the South showed increased resistance to all the antimicrobials ( Table 2).

MLST analysis identifies STs belonging to human clonal complexes
To determine the genotypic properties of Campylobacter spp. isolated from cattle as well as their relationship to those of human origin, MLST analysis was performed on selected C. jejuni and C. coli isolates. A total of 112 isolates, 62 C. jejuni and 50 C. coli that represented the main PFGE clusters and different geographic locations, were selected for MLST analysis. Thirteen new C. jejuni STs were identified and designated as STs 4922 to 4932, 5447, and 5448 (Table 3). These new STs were recognized following query with the PubMLST database. Furthermore, three new  Table 2. Antimicrobials resistance of C. jejuni and C. coli isolated from four different locations.  (Table 3). Interestingly, we also detected in our study three isolates belonging to clonal complex ST-61, and two of them were identified as ST-61, which is clinically important and has not been reported previously in cattle in the USA according to the PubMLST database. The most frequently observed clonal complexes (CC) were CC ST-21, CC ST-42, and CC ST-61, which are also common in humans. Six isolates in our collection had incomplete allelic profiles and could not be assigned to any ST.

Number of resistant isolates (%) and range of MICs in mg/ml
In contrast to C. jejuni isolates, C. coli isolates showed very limited diversity (Table 3). Of the 50 C. coli isolates analyzed (Table 3), a Table 3. Distribution of multilocus sequence types among cattle C. jejuni and C. coli.  Invasion and intracellular survival potential of cattle associated C. jejuni and C. coli To investigate the virulence-associated potential of the cattleassociated C. jejuni and C. coli, we tested selected isolates for their ability to invade and survive in human INT407 intestinal epithelial cells. C. jejuni and C. coli isolates that belonged to all the detected CCs were selected for these assays; in addition isolates that exhibited disparate antibiotic resistance profiles and/or originated from different geographic locations were also included in this analysis. All tested C. jejuni isolates (n = 19) invaded INT407 (Fig. 3A); however, the invasion potential varied between isolates, ranging from an average of 2.75610 1 to 8.0610 4 CFU ml 21 (Fig. 3A). The invasion capability of 17 of the 19 isolates tested was significantly lower than that of the highly invasive strain C. jejuni 81-176 (P,0.01); however, 11 of these isolates invaded the INT407 cells with higher numbers as compared to the poorly invasive C. jejuni NCTC11168 (Fig. 3A). Furthermore, two C. jejuni isolates Cj-M-13 and Cj-N-33 exhibited high invasion potential that matched that of C. jejuni 81-176 ( Figure 3A). Interestingly, all tested C. jejuni isolates were capable of intracellular survival, albeit the potential varied between isolates, ranging from an average of 8.0610 1 to 1.3610 4 CFU ml 21 (Fig. 3B). Additionally, the same aforementioned two isolates showed a significantly higher (P,0.01) potential for intracellular survival as compared to that of C. jejuni 81-176.
On the other hand, only nine C. coli isolates could invade the INT407 cells with average numbers ranging between 2.75610 1 and 1.34610 4 CFU ml 21 (Fig. 4A). There was a significant difference in the invasion capabilities between the C.coli isolates (P,0.05). Furthermore, only seven isolates were capable of intracellular survival (Fig. 4B). Three isolates that did not invade were not tested for intracellular survival. In general C. coli isolates were less invasive and displayed reduced intracellular survival compared to C. jejuni.
Our data also show that invasive Campylobacter possessed a variable antimicrobial phenotype that ranged from complete susceptibility to resistance to 7 different antimicrobials (Table 4). However, it was notable that C. jejuni isolates with high invasion potential (Cj-M-13 and Cj-N-33) were also resistant to multiple antimicrobials (see asterisks Table 4; Fig. 3A).

Discussion
We investigated the occurrence of two important Campylobacter spp. (C. jejuni and C. coli) in cattle that are slaughtered for meat production across 4 major geographic locations in the USA. This is in contrast to previous studies of cattle-associated Campylobacter in the USA, which have been mostly geographically confined and/or limited to investigating the prevalence of certain species as well as other phenotypic characteristics such as antibiotic resistance [12,25,26]. Therefore, our approach was unique as it included analysis of genotypic diversity and antimicrobial resistance properties as well as attempts to understand the potential role of cattle-associated Campylobacter in virulence related phenotypes that might impact human health. Significantly, our findings highlight the importance of cattle as a reservoir for genotypically diverse, antimicrobial resistant, and potentially virulent Campylobacter in the USA.
Although the association of Campylobacter spp. with cattle in the USA has been investigated previously, available literature reported differing prevalence numbers for these pathogens. For example, Hoar et al [26] isolated Campylobacter from 5% of the fecal samples collected from cattle, while Sato et al [27] and Gharst et al [28] showed that these bacteria occurred in 27.9% and 23.4% of their samples, respectively. These differences in the prevalence of cattleassociated Campylobacter can be attributed to several factors, including methods for isolation, sample size and type (e.g. dairy versus feedlot), seasonal variations, and geographical location [2]. However, the latter factor, geographical location, was a prominent difference in the aforementioned prevalence studies, which were conducted in California, Wisconsin and Southeastern region of the USA, respectively [26,27,28]. Since the prevalence of cattleassociated Campylobacter in the USA seemed to vary according to the geographic location, it was necessary in this study to investigate this assumption as well as the overall occurrence of these bacteria in cattle. Our results showed that the occurrence of Campylobacter in Southern USA (prevalence of 32.8%) was significantly (P,0.05) higher than those observed in Northern (14.8%), Midwestern (15.83%), and Eastern (12%) locations. The precise reasons that contribute to the observed differences in geographic distribution of the Campylobacter are not clear. Nevertheless, previous studies have suggested that environmental conditions such as temperature, humidity, and sunlight might contribute to increases in human infections with Campylobacter [29]. Subsequently, the climate might be a driver of the relatively higher prevalence of cattle-associated Campylobacter in the Southern USA, which is generally warmer as compared to the other locations. Interestingly, Stanley et al. [30] suggested that indirect-temperature dependent factors such as migratory animals might impact the occurrence of Campylobacter in dairy cattle. Additionally, C. coli have been previously reported to predominantly occur in samples collected from broiler flocks during warmer months [15]. Strikingly higher prevalence of C. coli (83 isolates) in our samples from the South (Table 1) further support a role for climate in affecting the geographic prevalence of Campylobacter. Regardless, the overall prevalence of cattle-associated Campylobacter in this study was 19.2%, falling between the percentages (0.8 to 46.7%) reported for cattle from other countries [31]. However, it is important to note that samples in our study were collected only over a two-day period from a limited number of locations in each region. Furthermore, to emphasize a public health relevance of cattle as potential reservoir for Campylobacter (discussed below), our sampling efforts focused on the seasons that normally witness an increase in Campylobacter incidences [8]. As such our prevalence data don't account for possible seasonal and temporal variations and, from this study alone, the extent to which differences between geographic locations contributed to Campylobacter prevalence in cattle cannot be conclusively ascertained. Nevertheless, our prevalence data provide evidence that like in other countries cattle in the USA might constitute a considerable reservoir for these pathogens, promoting their potential persistence in cattle-derived products and environment, which might pose a risk for human consumers and cattle-handlers.
The genus Campylobacter contains species that are known for their adaptation capabilities, readily acquiring genetic material necessary for survival and persistence [32]. High genotypic diversity has been reported in Campylobacter isolated from humans, chickens, sheep, pigs, and cattle [33,34,35,36]. This is important since the genotypic diversity of these bacteria might serve as an  indicator for the value of a reservoir in providing a suitable niche for the persistence of these microorganisms and their evolution [36]. As elegantly demonstrated by Wilson et al. [36], the broiler gastrointestinal tract constitutes such a niche, allowing multiplication of C. jejuni and enhancing its genetic diversity. Subsequently, it was important to investigate the genetic diversity of Campylobacter in cattle in order to assess the potential attributes of this reservoir. Our PFGE analysis of the cattle-associated Campylobacter revealed that all C. jejuni and C. coli isolates were generally grouped into 7 and 4 major clusters, respectively (Fig. 1). Remarkably, some C. jejuni isolates possessed genotypes that were 100% similar on 3 different occasions (see stars in Fig. 1A). Similarly, several C. coli exhibited 100% similar genotypes and were encountered on seven occasions (see asterisks in Fig. 1B). Taken together, these observations confirm a high genetic diversity within the cattle-associated C. jejuni and C. coli isolated in the USA, which also corroborates the findings of studies conducted on cattle in other countries, including Spain [35], the United Kingdom [37], Denmark [38], Finland [9], and Turkey [39]. Additionally, our PFGE analysis revealed clustering of some C. jejuni isolates (more than 75% similarity) that originated from different geographic locations. Likewise, some clusters included C. coli isolates with identical genotypes originating from different geographic locations (see arrows Fig. 1). This indicated that isolates with identical genotypes can occur in different herds, which highlights the possibility for certain strains to be transmitted among disparate herds/individuals. However, the latter requires further investigation to elucidate potential routes and mechanisms of transmission. Our PFGE analysis shows that cattle can be considered a potential reservoir for genetically diverse Campylobacter that is worth investigating in order to determine its relevance to public health in the USA.
An outstanding feature in livestock associated Campylobacter is its ability to resist antibiotics (reviewed in Luangtongkum et al. [40], an increasingly important emerging threat to public health [41]. Furthermore, previous studies have demonstrated that cattleassociated Campylobacter in the USA can potentially resist front-line therapeutic drugs for treating human infections. For example, 31.8% and 44% of C. coli collected from calves and feedlot cattle were resistant to erythromycin and ciprofloxacin, respectively [31]. Additionally, 47.7% and 49.1% of C. jejuni isolated from dairy and feedlot cattle, respectively, were resistant to tetracycline [25,42]. In agreement with these studies, our antibiotic resistance analysis showed that a high percentage of the cattle-associated strains were resistant to erythromycin, ciprofloxacin, tetracycline and others (Fig 2A and Table 2). Remarkably, 23.7% and 12.1% of the cattleassociated C. jejuni that were analyzed in our study exhibited resistance to ciprofloxacin and erythromycin, respectively ( Fig. 2A). This was an important and marked difference when compared to previous studies in the US that reported a relatively lower frequency of resistance to ciprofloxacin (1.8% to 5%) and erythromycin (0.4% to 2.9%) in C. jejuni isolated from different cattle operations [25,31,42]. Our data further support previous findings and predictions that livestock-associated Campylobacter are becoming increasingly resistant to important antibiotics [41,43]. These observations highlight the need for rigorous surveillance of antibiotics used in cattle operations in order to facilitate interventions and curb further emergence of antibiotic resistant Campylobacter.
Understanding the contributions of cattle to human infections with Campylobacter relies heavily on: i) assessing the relationship between the cattle-associated Campylobacter and those isolated fromor previously implicated in human infections and ii) directly examining the potential of these isolates to invade and persist in the human intestine. However, these two criteria were rarely investigated together in previous research that focused on cattleassociated Campylobacter and are, to our knowledge, absent in similar studies conducted in the USA. To meet the aforementioned criteria, we typed a pool of the cattle-associated Campylobacter using MLST and further analyzed a subset of isolates for their properties in an in vitro surrogate for the human host (i.e. INT407 intestinal cell line). Since interpretation of the MLST typing depends on pre-existing online data bases (PubMLST) to assign sequence types (STs) to the isolates, it is possible to compare the cattle-associated Campylobacter from this study to others from different sources (e.g. humans) and locations, including other countries. Subsequently, our MLST analysis confirmed that cattleassociated C. jejuni (n = 62) were diverse and belonged to disparate STs (n = 25; Table 3), while only 8 STs were assigned to C. coli (n = 50) (Table 3). Furthermore, we identified 12 C. jejuni that belonged to unassigned clonal complexes (n = 10), which along with newly discovered alleles (n = 3) and STs (n = 13) (Table 3) emphasize the genetic diversity of this bacterium in cattle, possibly indicating that certain C. jejuni might be highly associated with this potential source of infection. The latter is supported by reports showing that CC ST-21 and ST-61 were predominant in cattle samples from other countries, including the UK, Canada, New Zealand, and Finland [9,33,44,45]. The detection of CC ST-61 and CC ST-21 (Table 3) in this study was of particular interest, since these complexes constitute two of six CCs that harbor 60% of C. jejuni associated with human disease [20,37]. Although all tested cattle-associated C. jejuni were capable of invading the human INT407 cells with variable efficiency, two isolate (Cj-M-13) and (Cj-N-33) that belong to CC ST-21 not only exhibited an invasive potential that was comparable to the hyper-invasive strain C. jejuni 81-176 [46] but also showed a higher capability to survive in the intracellular milieu (Fig. 3). Furthermore, these two isolates were resistant to 7 out of 9 tested antimicrobials, including ciprofloxacin, erythromycin, and gentamicin (Table 4). Along the same lines, 11 out of 19 C. jejuni with varying invasive capacities were able to either match or outperform the intracellular survival potential of C. jejuni 81-176 (Fig. 3). No specific association was observed with either invasion or intracellular survival of the tested C. jejuni isolates and their MRPs, antimicrobial resistance properties or their geographical distributions. Since many of the tested C. jejuni (13 out of 19) were moderate to highly invasive ( Table 4) and given this bacterium's aforementioned capability for natural transformation, the occurrence of invasive strains in a reservoir that is witnessing a rise in antibiotic-resistant Campylobacter constitutes a cause for caution and increased surveillance. Since invasion and intracellular survival of pathogens facilitates evading the immune system of the host and allows for persistent and recurrent infection [47], it can be argued that the cattleassociated C. jejuni would pose a plausible risk to human health.
Furthermore, MLST data and invasion studies also showed relatively lower clinical relevance of selected cattle-associated C. coli isolates, which all belonged to CC ST-828, a complex that include strains which are mainly isolated from agricultural and environmental sources and some from clinical cases [48]. Although the invasion and intracellular survival capacities of C. coli were overall lower than those of C. jejuni, one cattle associated C. coli (CC-S-10) invaded the cells with approximately 10 4 CFU per ml, while also surviving inside the cells in relatively high numbers (Fig. 4). Taken together, our data confirm the observations of Sheppard et al. [48] that ruminants also are among the most likely sources of human infections with C. coli.
We conclude that cattle in the USA constitute a suitable niche for the persistence of Campylobacter and a potentially important source for human infections with these pathogens. The potential of cattle-associated Campylobacter to resist antibiotics, invade and persist in human cells, and their genetic classification into clinically important clonal complexes warrant highlighting the contribution of cattle to the epidemiology of these pathogens in the USA.

Supporting Information
Table S1 Antimicrobial resistance profiles, Sequence Types, and MRP clusters of C.jejuni isolates.