The diversity of Campylobacter spp. throughout the poultry processing plant

Campylobacteriosis is the leading food‐borne disease in developed countries, and poultry are a major source for human infection. The diversity of Campylobacter on chicken carcasses during processing may lead to isolates that are able to survive abattoir processing. This has important implications for public health and adds a further layer to the complexity of the epidemiology of campylobacteriosis. The diversity of the Campylobacter spp. populations on broiler carcasses was studied at three different stages of processing (post‐bleed, post‐scald and post‐chill) in three UK processing plants, using the pulsed‐field gel electrophoresis (PFGE) KpnI enzyme. One hundred and sixty Campylobacter strains from 3 processing plants were identified as C. jejuni (92.3%) with 27 PFGE subtype profiles recovered from carcasses at the post‐bleed point. Change in populations was identified when carcasses move towards the end of poultry processing. Seven C. jejuni genotypes were able to survive the scalding tank stage process, and 5 genotypes surviving the entire poultry process. Confirmation by PFGE gives information on the genotypic profiles of C. jejuni on chicken carcasses and how they change according to the temperatures exposed to during processing. Diversity within C. jejuni populations produces genotypes that adapt to tolerate the processing environment, and these may be capable of causing human disease. Understanding more about the genotypes that survive the processing will have important implications for public health.

immune response to Campylobacter (Humphrey et al., 2014;Williams et al., 2013). After colonization, Campylobacter spp. remain in the birds throughout their life cycle until slaughter, despite interventions both on the farm and in the processing plant, such as strict biosecurity and disinfection . Large numbers of Campylobacter spp. are present throughout the processing plant making avoidance of cross-contamination between birds and flocks almost impossible . This explains the high correlation between poultry products and human infections (Coward et al., 2008;Stern, 2008). This viewpoint is supported by the EFSA (2011) that found more than 80% of chicken carcasses at retail sale in the United Kingdom were contaminated by Campylobacter spp.
The processing plant has been recognized as a high-risk area which is involved with cross-contamination between birds. García-Sánchez et al., (2017) reported that the de-feathering and evisceration machines were very effective in spreading Campylobacter spp. between carcasses. Herman et al., (2003) found that the same Campylobacter spp. were identified both during rearing on the farm and on the final product after processing and concluded that a correlation exists between the contamination of broiler chickens during rearing and after processing. This indicates that Campylobacter is able to survive during processing and control measures are not always effective.

Molecular epidemiological techniques: for examples, ribotyping,
PFGE, flagellin typing (fla typing) and multilocus sequence typing (MLST) (Dingle et al., 2001;Wassenaar & Newell, 2000) have shown Campylobacter genotypes are widely diverse, leading to an increased survival potential under environmental stressors such as processing. In one study, Parkhill et al., (2000) concluded that thermophilic Campylobacter spp., can survive in the environment via genetic diversity and a high rate of variation in homopolymeric tracts. Furthermore, diversity in the genotypes provides a bacterial population with genome plasticity that may increase the adaptation for survival in a hostile environment (de Boer et al., 2002). In addition, Newell et al., (2001) stated that in the UK, individual broiler chickens can be colonized by more than one subtypes of C. jejuni or C. coli but in low number of campylobacter of the same subtypes. Therefore, identifying Campylobacter spp. in poultry by comparing the genetic profiles can help to understand the transmission routes of these bacteria, allowing the development of suitable preventive methods (Johnsen et al., 2006).

Analysis of Campylobacter genotypic diversity from processing plants and broiler chickens showed both low diversity among
Campylobacter isolates and only two different profiles have been identified. There is a large diversity among broiler flocks with four different profiles, when campylobacters on carcasses were analysed during process using PFGE (Normand et al., 2008). Although some subtypes were found in post-chill stages and had survived processing, it was concluded that Campylobacter genetic diversity declined through processing (Hunter et al., 2009). The objectives of this study were to identify if there was a change in Campylobacter populations on carcasses through the processing plant by examining the diversity among Campylobacter isolates collected at post-bleed, post-scald and post-chill stages of the processing chain at three UK commercial poultry processors to identify Campylobacter genotypes that are capable of surviving.

Impacts
• A high level of genetic diversity was found among the Campylobacter isolates with 27 PFGE profiles recovered from chicken carcasses at the post-bleed point using PFGE KpnI enzyme.
• Seven C. jejuni subtypes were able to survive the scalding stage process (exposing the birds to a temperature of around 52°C, loosen the feathers).
• Changes in Campylobacter populations were identified when carcasses move towards the end of poultry processing with 5 C. jejuni subtypes surviving the entire poultry process. This emphasizes the importance of reducing Campylobacter population in poultry abattoirs.

| Bacteria and culture conditions
As described by Elvers et al., (2011), neck skin samples were plated onto modified charcoal cefoperazone deoxycholate agar (mCCDA; Oxoid Ltd). Plates were incubated at 41.5°C for 48 hr in a microaerobic atmosphere generated using the CampyGen gas generating sachets (Oxoid Ltd.). Colonies of Campylobacter spp. were identified on the basis of colony morphology on mCCDA, on which they appeared as a greyish metallic colony with a tendency to spread, and 2-4 colonies from each sample were taken. These were streaked onto two separate Columbia blood agar plates (BA; Oxoid Ltd.) containing 5% (w/v) defibrinated horse blood. One plate was incubated at 41.5°C for 48 hr in a microaerobic atmosphere generated using CampyGen (Oxoid Ltd.), and the second plate was incubated at 41.5°C for 48 hr aerobically to rule out any possibility of the F I G U R E 1 Dendrograms representing relatedness among PFGE profiles of Kpnl digests of 160 Campylobacter isolates detected on broiler carcasses in post-bleed point process from processing plants (B), (K) and (H). The dendrogram was constructed by using Dice Coefficients matrix based on the UPGMA. The strains grouped into clusters of 80% genetic similarity indicated as I II III IV*. Clusters were defined as isolates whose restriction digests had a similarity coefficient ≥80%. Each letter and number represent a different genotype. Each colour on the right-hand side of the figure represents a different processing plant, plant B is in green, K is in blue and H is in pink. The coloured blocks on the left-hand side of the figure indicate the different PFGE profiles. *The isolates were closely related, allowing them to be grouped into clusters of at least 80% genetic similarity. The similarity value at each branching in the dendrogram represents the average similarity of profiles in the branch, and 100% means the unique profiles were identical. The software GelCompare II was used [Colour figure can be viewed at wileyonlinelibrary.com] colonies being contaminants. Any colonies that did not grow aerobically were presumed to be Campylobacter spp. and were further confirmed using Gram staining and Oxidase positivity. Bacteria were stored on Cryobeads (Microbank™, Pro-Lab diagnostics) at −80°C until required for further confirmation using PCR. When a culture was needed a single bead was removed from the Cryovial and placed onto a BA plate and incubated microaerobically as described above.

| Deoxyribonucleic acid (DNA) extraction
The DNA from the samples was extracted using the QIAamp® DNA extraction Kit (Qiagen). The manufacturer's protocol for purification of total DNA (Spin-Column protocol) was followed https://www. qiagen.com/gb/resou rces/resou rcede tail. Eluted DNA was stored at −20°C until use.

| Campylobacter speciation using multiplex real time PCR
A multiplex real-time PCR (RT-PCR) was used to identify the species of forty-eight isolates from plant B, which represented (2x) 24 PFGE Campylobacter genotypes chosen from post bleed (12 PFGE genotypes), post-scald (7 PFGE genotypes) and post-chill (5 PFGE genotypes) stage samples. Using a combination of in-house

designed primers for Campylobacter jejuni, Campylobacter coli and
Campylobacter upsaliensis and previously published primers for the 16S rRNA gene of Campylobacter (Lund et al., 2004). The primers and probes used in the multiplex PCR are described in Table 1. F I G U R E 2 Dendrograms representing relatedness among PFGE profiles of Kpnl digests of Campylobacter jejuni genotypes detected on broiler carcasses at the post-bleed point in the process from commercial processing plant K. Clusters were defined as isolates whose restriction digests had a similarity coefficient ≥80%.
Isolates with similar profiles indicated close genetic relationships, allowing them to be grouped into clusters of at least 80% genetic similarity. The similarity value at each branching in the dendrogram represents the average similarity of profiles in the branch, and 100% similarity means the unique profiles were identical. The software GelCompare II was used F I G U R E 3 Dendrograms representing relatedness among PFGE profiles of Kpnl digests of Campylobacter jejuni genotypes detected on broiler carcasses at the post-bleed point in the process from commercial processing plant H. Clusters were defined as isolates whose restriction digests had a similarity coefficient ≥80%. Isolates with similar profiles indicated close genetic relationships, allowing them to be grouped into clusters of at least 70% genetic similarity.
The similarity value at each branching in the dendrogram represents the average similarity of profiles in the branch, and 100% similarity means the unique profiles were identical. The software GelCompare II was used [Colour figure can be viewed at wileyonlinelibrary.com] F I G U R E 4 Dendrograms representing relatedness among PFGE profiles of Kpnl digests of Campylobacter jejuni genotypes detected on broiler carcasses at the post-bleed point in the process from commercial processing plant B. Clusters were defined as isolates whose restriction digests had a similarity coefficient ≥80%.
Isolates with similar profiles indicated close genetic relationships, allowing them to be grouped into clusters of at least 80% genetic similarity. The similarity value at each branching in the dendrogram represents the average similarity of profiles in the branch, and 100% similarity means the unique profiles were identical. The software GelCompare II was used [Colour figure can be viewed at wileyonlinelibrary.com] Mx3005p system (Agilent Technologies). The data analysis was performed using the MxPro software (Agilent Technologies).

| Pulsed-field gel electrophoresis
All the DNA preparations, restriction endonuclease digestions and pulsed-field gel electrophoresis (PFGE) were carried out as described by Michaud et al., (2001) and PFGE Standard Operating protocol for PulseNet (2013).

| Pulsed-field gel electrophoresis analysis
Hundred and sixty isolates were randomly chosen to be representative of Campylobacter diversity across three commercial processing plants (B, K, H). The Kpnl PFGE restriction profile patterns were compared using GelCompar II gel analysis software (Applied Maths, Sint-Martens-Latem, Belgium). Based on band position, the similarity between profiles was identified and derived from the Dice correlation coefficient, with maximum tolerance of 2% which was used to compensate for between-gel variance and 5% optimization. A coefficient matrix was used to generate dendrograms based on the unweighted pair group method using arithmetic averages (UPGMA) (de Boer et al., 2000). The estimation of genetic diversity of the Campylobacter populations was calculated using Simpson's index as described by Hunter (1990).

F I G U R E 5
Dendrograms representing relatedness among PFGE profiles of Kpnl digests of Campylobacter jejuni genotypes detected on broiler carcasses at the post-scald point in the process from commercial processing plant B. Clusters were defined as isolates whose restriction digests had a similarity coefficient ≥80%. Isolates with similar profiles indicated close genetic relationships, allowing them to be grouped into clusters of at least 80% genetic similarity. The similarity value at each branching in the dendrogram represents the average similarity of profiles in the branch, and 100% similarity means the unique profiles were identical. The software GelCompare II was used [Colour figure can be viewed at wileyonlinelibrary.com] F I G U R E 6 Dendrograms representing relatedness among PFGE profiles of Kpnl digests of Campylobacter jejuni genotypes detected on broiler carcasses at the postchill point in the process from commercial processing plant B. Clusters were defined as isolates whose restriction digests had a similarity coefficient ≥80%. Isolates with similar profiles indicated close genetic relationships, allowing them to be grouped into clusters of at least 80% genetic similarity. The similarity value at each branching in the dendrogram represents the average similarity of profiles in the branch, and 100% similarity means the unique profiles were identical. The software GelCompare II was used [Colour figure can be viewed at wileyonlinelibrary. com]

| RE SULTS
Campylobacter was detected in 490/521 (94%) of the neck skins analysed, irrespective of processing plant. From these isolates, a subset was analysed from all processing plants B (n = 60), H (n = 45) and K (n = 55), using multiplex RT-PCR and 160 of these isolates were identified as C. jejuni and further characterized using PFGE.

| Genotype diversity from single birds
One hundred and sixty C. jejuni isolates were examined that be- PFGE profiles between the slaughterhouses (B, H and K) for each bird carcass examined showed a considerable diversity among C. jejuni isolates with one carcass having three genotypes present ( Figure 1). However, there were also some carcasses carrying only one genotype and the profiles C7 and C23 were more common. The PFGE of C. jejuni genotypes were more diverse at the first point of processing. Figures 2, 3 and 4 show a relationship between the isolates at each processing plant.  Table 2). The isolates were shown to be highly genetically diverse from the birds at post-bleed (Figure 4), but there was less genetic F I G U R E 7 The survival of Campylobacter jejuni subtypes throughout plant B at the three different processing stages. Each coloured bar is a genotype as indicated by the key at the bottom of the graph. The left side of vertical axis represents the percentages of the C. jejuni populations assigned to specific genotype at specific stage of processing. These populations of each Campylobacter genotype change during processing stages, and the surviving genotype makes up more of the populations [Colour figure can be viewed at wileyonlinelibrary.com] diversity among C. jejuni strains at post-chill ( Figure 6). Table 3 represents the number of C. jejuni isolates from plant B and the number of C. jejuni genotypes detected from post-bleed, post-scald and post-chill of the chicken processing line, respectively. Figures 4, 5 and 6 show a relationship between these isolates at each sample point.

| Diversity within the poultry processing plant
The diversity index value of C. jejuni strains from post-bleed was 0.908, and this index was lower at the post-scald and post-chill point in the processing line (0.837 and 0.746, respectively; Table 3). This difference was still evident as most genotypes were found commonly on carcasses in early processing rather than later in the post-chill stage.
Thus, C. jejuni subtypes obtained from post-bleed were statistically more highly diverse than those sampled from post-scald and post-chill (0.016 and 0.002, respectively) when using the Fisher's exact test.

| Identification of Campylobacter jejuni PFGE genotypes surviving processing
The

| D ISCUSS I ON
Campylobacter has a long history of being associated with chickens, with the number of human cases continuing to rise despite interventions (WHO, 2018). There is a need to understand more about the relationship between Campylobacter and chickens and how the bacterium survives through poultry processing. The PFGE

micro-restriction profiles revealed a considerable diversity among
Campylobacter isolates obtained earlier in the processing chain.
The total number of different PFGE profiles recovered from carcasses at the post-bleed point in the three processing plants was high (n = 27), with two to four different types being isolated from a single carcass. This indicates that there are significant genetically diverse populations of Campylobacter on carcasses from poultry processing plants, including those from the same flock (Table 2). These results are in agreement with Marotta et al., (2014) who found a high level of genotypic diversity of Campylobacter populations present in chicken carcasses throughout poultry production. The results demonstrated that individual carcasses could harbour multiple strains of Campylobacter spp. at the post-bleed point.
Campylobacter isolates from abattoirs H, B and K were highly diverse at the beginning of the slaughter process, probably because samples originally came from different broiler farms and there was carry over of the Campylobacter populations between the rearing cycles and/or flocks (Marotta et al., 2014). The possibility exists that cross-contamination could also occur from the transport crates or the auto-killer, as they have been shown previously to be contaminated (Mead et al., 1994;Stern et al., 1995). However, another study suggested that cross-contamination during poultry processing was undetectable on the end product . The present study showed that the isolates from abattoirs fell into four major PFGE groups clustering at 80% similarity, thus indicating all isolates were related irrespective of the abattoir or the original farm suggesting that those that survived this part of the process are the Campylobacter genotypes that are more heat tolerant and so are therefore more likely to be recovered from the end product.
Further changes in levels of C. jejuni population diversity on carcasses after the scalding tank occurred when carcasses moved through the system into the chilling stage. Results showed a change in C. jejuni genotype populations; in particular, three PFGE genotypes C. jejuni C5, C. jejuni C7 and C. jejuni C23 were more dominant at this stage (Figure 7). Cold stress may act as a selective force which permits cold-tolerant Campylobacter to survive (Chan et al., 2001).
This difference was still evident as most genotypes were found commonly on carcasses in early processing rather than later in the post-chill stage. These findings are supported by the study of Hunter et al., (2009) who demonstrated that chilling can be a factor in the reduction of Campylobacter diversity.
Interestingly, C. jejuni genotypes C. jejuni C4, C. jejuni C5, C. jejuni C6, C. jejuni C7 and C. jejuni C23 were found in both scalding and chill stages in abattoir B (Figure 7). Thus, these five genotypes are able to survive the higher temperature of 55°C and low temperature 6°C which they are exposed to during processing. This result indicates that survival at low and high temperatures could be related as previously reported by Zhang and Rock (2008) and Hughes et al., (2010) and that the bacterium can survive in a wide range of physical environments through the controlled biophysical properties of their membrane phospholipids. Some isolates could be detected throughout poultry processing and some were only detected up to the scalding stage and were not recovered post bleed. While the levels fluctuated these results show that only the more resilient isolates can survive ( Figure 7). This is in agreement with Newell et al., (2001), who found that the resistance of some Campylobacter subtypes during poultry processing varied between strains and the more robust strains survived through carcass chilling. Taken together, these results suggest that elimination of Campylobacter genotype populations on carcasses at the final stage may be related to the number of subtypes that are better able to adapt to the stress of poultry processing. It suggests that the carcasses containing more than one genotype may have had an impact on the diversity of the organism to produce strains that survive throughout the process. This could lead to identifying particular genetic elements related to the ability to survive heat and cold stress within the abattoir environment. The findings also suggest that C. jejuni genotypes are able to survive under these conditions and are may be capable of causing human disease.

| CON CLUS ION
Different C. jejuni genotypes can survive at various stages of the process, and this may help to identify the factors responsible for survival at each stage. This knowledge may help to identify methods for reducing Campylobacter on chicken carcasses and improve processing plant hygiene and public health.

ACK N OWLED G EM ENTS
The authors acknowledge the staff of the Food Standard Agency funded project at the Veterinary School, University of Bristol for their cooperation and assistance in collection of the samples from processing facilities. We also acknowledge the University of Bristol for financial support.

AUTH O R CO NTR I B UTI O N S
H. M. designed and carried out the experiment and wrote the manuscript. L K. W. and E. VK. helped in analysis and interpretation of the data and provided critical feedback and helped shape the research.