Identification of Transmission Routes of Campylobacter and On-Farm Measures to Reduce Campylobacter in Chicken

An in-depth analysis was performed on Swedish broiler producers that had delivered chickens with Campylobacter to slaughter over several years, in order to identify possible transmission routes and formulate effective measures to prevent chickens being colonized with Campylobacter. Between 2017 and 2019, 626 samples were collected at farm level and Campylobacter was isolated from 133 (21.2%). All C. jejuni and C. coli isolated from these samples were whole-genome sequenced, together with isolates from the corresponding cecum samples at slaughter (n = 256). Core genome multi-locus sequence typing (cgMLST) analysis, using schemes consisting of 1140 and 529 genes for C. jejuni and C. coli, respectively, revealed that nearby cattle, contaminated drinking water, water ponds, transport crates, and parent flocks were potential reservoirs of Campylobacter. A novel feature compared with previous studies is that measures were implemented and tested during the work. These contributed to a nationwide decrease in Campylobacter-positive flocks from 15.4% in 2016 to 4.6% in 2019, which is the lowest ever rate in Sweden. To conclude, there are different sources and routes of Campylobacter transmission to chickens from different broiler producers, and individual measures must be taken by each producer to prevent Campylobacter colonization of chickens.


Introduction
Campylobacter spp. is the most reported bacterial cause of gastrointestinal disease in humans in Sweden, and in other countries [1,2]. In Sweden, a total of 8132 campylobacteriosis cases were reported in 2018, corresponding to an incidence of 80 cases per 100,000 inhabitants, of which 45% were domestically acquired [3]. In the European Union (EU), 246,571 human cases were reported in 2018, corresponding to an incidence of 64 cases per 100,000 inhabitants [1]. However, the number of cases can be assumed to be significantly underreported [4]. For Sweden, a multiplier of 9.05 is suggested, which would mean that the actual number of people affected by campylobacteriosis during 2018 was 73,595 people, at a cost to society of approximately SEK 649 million (SEK 8825 per case) [5].
Chicken and chicken products are known to be the major sources of Campylobacter infection in humans, through insufficient heat treatment or cross-contamination in the kitchen [1,6]. Campylobacter spp. do not multiply in food, but multiplication is not required to cause disease, as the infective dose

Whole-Genome Sequencing
The 256 whole-genome sequenced C. jejuni (n = 235) and C. coli (n = 21) isolates belonged to 47 different multi-locus sequence type (MLST) profiles, of which 12 were novel profiles ( Table 2). Table 2. Distribution of multi-locus sequence type (MLST) profiles and clonal complexes among isolates of Campylobacter jejuni (n = 235) and C. coli (n = 21).  The most common sequence type (ST) found in this study was ST-257, with 58 of the C. jejuni isolates from three of the four broiler producers (A, C, and D) belonging to this sequence type (ST). ST-583 and ST-918 were the second most common ST, with 16 C. jejuni isolates each. However, ST-918 was only found at one broiler producer (A), whereas ST-583 was found at two broiler producers (A and B). The third most common ST was ST-19, which was found at three broiler producers (A, C, and D). Most of the 12 novel MLST-profiles in this study were found only once, but ST-9714 was found in 10 C. jejuni isolates, all from broiler producer B ( Table 2).
The core genome MLST (cgMLST) schemes developed for C. jejuni and C. coli consisted of 1140 and 529 genes, respectively. Sequenced C. jejuni isolates of the same ST often clustered together in the cgMLST analysis ( Figure 1). However, isolates of ST-45 were diverse, with three different variants being found at broiler producer A and a fourth variant being found at broiler producer B. Isolates of ST-257 that were found at three broiler producers formed two clusters, one for broiler producer A and one for broiler producers C and D ( Figure 1). Isolates of ST-583 formed two clusters, one for broiler producer A and B, respectively. Isolates of ST-19 from three broiler producers (A, C, and D) clustered together, but with fewer allelic differences between isolates from the same broiler producer than between different broiler producers.   Figure 2). Isolates of ST-257 were obtained from cattle feces samples taken near the broiler houses, from sock samples taken inside the broiler houses, from water from drinking water pipes, and from the cecum samples from 26 slaughter batches during four different flock rotations between January to July 2017. After thorough cleaning of the water pipes in summer 2017, ST-257 was not detected again in any sample from producer A during the study period. In addition to ST-257, ST-21 and ST-1326 were also isolated from feces from the nearby cattle ( Figure 2). Furthermore, ST-21 was found in sock samples from inside the broiler house, and in cecum and neck skin samples at slaughter. ST-918 was isolated from cecum samples from more than one flock rotation at slaughter and was also found in transport crates on the arrival at broiler producer A before the chickens were loaded in. Other sequence types isolated from the transport crates were ST-572 and ST-267, of which ST-572 was found in both transport crates and chickens sampled at slaughter (cecum samples). During the 2015-2018 period, Campylobacter was isolated from 40% of the slaughter batches delivered by producer A. In the period January-July 2017, Campylobacter was isolated from 56% of the slaughter batches. After cleaning the water pipes by increasing and decreasing the pressure of water and air in the water pipes in the end of July 2017, Campylobacter was only isolated from 6% of the cecum samples taken at slaughter between August and May 2018. During June to October 2018, the incidence of Campylobacter-positive slaughter batches increased to 41%. Since seven different sequence types (ST-19, ST-21, ST-42, ST-45, ST-538, ST-572 and ST-9715) were identified from the chickens during that period it was considered that there had to be more than one reservoir. In addition to regular cleaning between the rotations, changes of socks under the slippers between the departments were introduced in October 2018. This contributed to fact that Campylobacter was not isolated, either from the sock samples or the cecum samples, from the 15 slaughter batches in November-December 2018. Furthermore, the incidence was significantly decreased (p < 0.00001) to 14% of slaughter batches from producer A in 2019.

Figure 2.
Minimum spanning tree of core genome multi-locus sequence typing (cgMLST) data from Campylobacter jejuni isolated from the internal and external environment at broiler producer A (n = 131). STs are given for all isolates. The numbers on the lines between isolates represent allelic differences. Line length is not proportional to the numbers.

Broiler Producer B
The result of whole-genome sequencing for Campylobacter isolates from producer B reveals a remarkable number (n = 30) of different sequence types of Campylobacter at farm level and at slaughter ( Figure 3). Two C. jejuni sequence types, ST-9198 and ST-9714, were detected in chickens on-farm and at slaughter (cecum samples). Three sequence types, ST-148, ST-583, and ST-9198, indicated possible transmission routes for producer B that were not identified at the other producers. ST-148 was isolated from four different parent flocks and from chickens sampled on-farm in all four compartments at one sampling occasion at producer B. In the cgMLST analysis, there were only three allelic differences between three of the isolates from the parent flocks and isolates from the chickens sampled on-farm ( Figure 3). Campylobacter spp. were isolated from four out of six samples from a water pond next to the farmyard frequently visited by wild birds and wild boar. Whole-genome sequencing resulted in eight different sequence types from the pond, two of which, ST-583 and ST-9198, were also isolated from the chickens ( Figure 3).
Broiler producer B was the only production unit in which C. coli was detected. The same C. coli isolates, ST-829 and ST-4709, were detected in the chickens on-farm and at slaughter (cecum samples) ( Figure 4). In addition, there were only 14 allelic differences between C. coli isolates from chicken on-farm, chickens at slaughter (cecum samples), and parent flocks (ST-2178). From 2015 to 2018, Campylobacter was isolated from 42% of the slaughter batches delivered by producer B. In 2017 Campylobacter was isolated from 57% of the slaughter batches. After installation of protection against wild birds and their droppings above the vent from the broiler house and re-paving the farmyard in December 2017, the incidence was decreased to 46% in 2018. During 2018, the biosecurity was increased in different stages at farm level, such as extending the number of hygiene barriers, including changing footwear three times instead of two, using cotton gloves which were washed after each use. The water pipes were cleaned by increasing and decreasing the pressure of water and air in the water pipes and UV-light treatment of incoming water was installed. Measures were also taken inside the broiler houses such as the sealing of cracks in the floors with silicone and reconstruction of the ventilation on the roof to prevent transmission of condensed water and rainwater to the chickens. In addition, small hatches were installed in the front doors for the removal of dead chickens. The incidence of Campylobacter was decreased (p = 0.1) to 25% of slaughter batches delivered by producer B in 2019.  . Minimum spanning tree of core genome multi-locus sequence typing (cgMLST) data from Campylobacter coli isolated from the internal and external environment at broiler producer B (n = 21). STs are given for all isolates. The numbers on the lines between isolates represent allelic differences. Line length is not proportional to the numbers.

Broiler Producer C
In the past, broiler producer C has delivered fewer chickens colonized with Campylobacter to slaughter than producers A and B. However, during spring-summer 2018, Campylobacter was isolated from a greater proportion of slaughter batches from broiler producer C. ST-257 dominated among the sequenced C. jejuni isolates and cgMLST analysis revealed that isolates from chickens on-farm, chickens at slaughter (cecum samples), and cattle nearby clustered closely together ( Figure 5). This ST-257 was found in 20 of 24 (83%) slaughter batches between May and August from broiler producer C. Different transmission routes were assessed during the visit in August 2018. One of the suspected reservoirs was the nearby cattle. After a fly screen was installed on the roof of the broiler house, the incidence of Campylobacter decreased to 14% (eight of 56) slaughter batches delivered by producer C in 2019. Figure 5. Minimum spanning tree of core genome multi-locus sequence typing (cgMLST) data from Campylobacter jejuni isolated from the internal and external environment at broiler producer C (n = 30). STs are given for all isolates. The numbers on the lines between isolates represent allelic differences. Line length is not proportional to the numbers.

Broiler Producer D
Broiler producer D delivers chickens to the same slaughterhouse as producers A and C, but in the past producer D has delivered fewer chickens colonized with Campylobacter to slaughter, compared with producers A and B. No environmental samples were collected from broiler producer D, but several different STs (ST-19, ST-257, ST-696, ST-1033 and ST-2066) were isolated from the cecum samples at slaughter, of which ST-1033 was found in different compartments in two consecutive flock rotations on-farm ( Figure 6). ST-257 was isolated from cecum samples from one of the flocks, and that isolate differed by only one allele from isolates from producer C (Figure 1). During 2015-2018, Campylobacter was isolated from 13% of the slaughter batches delivered by producer D. In the autumn 2018, possible transmission routes such as biofilms in the water pipes and transport crates were assessed and the importance of hygiene barriers such as changing of footwear at least twice and washing hands before working with chicken was discussed. In 2019, the proportion of Campylobacter-positive slaughter batches decreased (p < 0.00001) to 2%. Figure 6. Minimum spanning tree of core genome multilocus sequence typing (cgMLST) data from Campylobacter jejuni isolated from broilers belonging to producer D (n = 16). STs are given for all isolates. The numbers on the lines between isolates represent allelic differences. Line length is not proportional to the numbers.

Discussion
The results obtained in this study indicate that the greatest challenge in preventing colonization of broiler chickens by Campylobacter is that there are different reservoirs and transmission routes for chicken colonization within and between different producers. Several different reservoirs contributing to Campylobacter colonization of chickens were identified in this study. Cattle in a pasture near the broiler houses were a potential reservoir of Campylobacter spp. for broiler producers A and C. This is consistent with previous findings that proximity of livestock is a risk factor for chickens being colonized with Campylobacter [15,18]. Cattle have also been identified as important reservoirs for human campylobacteriosis, e.g., in the Netherlands, cattle have been identified as the reservoir for 21% of human cases [19], while in Denmark cattle are considered the second most important reservoir [20]. Furthermore, based on MLST data, it has been estimated that 22-55% of all cases of campylobacteriosis in France are caused by ruminants [21]. Since Campylobacter are shed in feces and ubiquitous in the environment, including surface waters, they could be transmitted into broiler houses via vectors such as flies, insects, and rodents, or via vehicles as aerosols or dust [22]. It is believed that flies can act as mechanical vectors and transmit Campylobacter to broiler houses, and therefore fly screens have been effective in reducing transmission [23]. Fly screens were installed on the roof of the broiler houses at producer C after the same sequence type, ST-257, was identified both in cattle feces and in the chickens, after which the incidence of Campylobacter-positive chickens delivered by producer C decreased.
Insufficiently cleaned transport crates is another risk factor for chickens becoming colonized with Campylobacter especially during thinning, when birds in a flock are sent to slaughter in separate batches on two or more slaughter occasions [18,24,25]. In this study, 30% of transport crates dispatched by the slaughterhouse were contaminated with Campylobacter already on arrival at farm level before loading up the chickens to send to slaughter. The Campylobacter isolates from the crates were of several different STs (ST-42, ST-267, ST-572, and ST-918), of which two (ST-572 and ST-918) were also isolated from chickens at slaughter. When transport crates are taken into the broiler houses during thinning, there is a risk of introducing Campylobacter to the chickens and continued colonization of the flock for another 5-7 days. Producers A, C, and D deliver chickens to a slaughterhouse that applies thinning. Insufficiently cleaned transport crates could be a possible explanation for the same Campylobacter strains circulating among these broiler producers.
Previous studies have found little evidence of vertical transmission of Campylobacter from the parent flocks to the chicken via the eggs [26]. Nevertheless, isolates of the same ST (ST-148) from the parent flocks and the chickens on-farm at broiler producer B clustered closely together in the cgMLST analysis. This particular ST has also been detected in chickens from other broiler producers that received chickens from the same hatchery and in human cases of campylobacteriosis during the same period [27]. Several of the human cases were in people working at the slaughterhouse where chickens originating from those parent flocks were slaughtered [27]. While no evidence has been found to support egg-borne transmission of Campylobacter from the parent flock [26,28], it has been suggested that feces containing Campylobacter might contaminate the eggshell and shell membranes of freshly laid fertile eggs. With a short period between laying and hatching, together with optimal temperature and presence of moisture when the chicken emerges from the egg, the chicken might ingest feces containing Campylobacter and become colonized [29]. Otherwise, the majority of flock colonization results from horizontal transmission from the environment and Campylobacter can be easily spread from birds, or other animals, to chickens through a number of routes.
Another potential transmission route detected at broiler producer B was a water pond next to the farmyard that was frequently visited by wild animals. Although several of the C. coli STs from the pond and wild boar did not match any of the chicken isolates at the time, C. jejuni isolates ST-583 and ST-9198 from the pond matched. Water contaminated with Campylobacter is an important source of Campylobacter colonization in both broilers and humans [30,31]. Campylobacter can survive in water for up to several months, depending on the environmental conditions and on the strain [32,33].
Several of the STs detected in this study have also been isolated from humans and chicken products at retail level in Sweden. For example, ST-19, ST-45, and ST-918 were all detected in humans and in chicken products during the early part of 2017 [34]. ST-918 was also the sequence type responsible for the largest outbreak of Campylobacter to date in Sweden (2016-2017) [35]. The most common ST found in the present study was ST-257, which was also the most common ST in humans during August 2018 [36]. ST-257 was isolated from chicken products at retail level during the same period [36], and this ST was found in many reported cases of campylobacteriosis in Sweden during winter 2015 [34]. Furthermore, ST-257 was the second largest cluster in a nationwide Danish study examining around 10% of human cases from 2015 to 2017, where 47% of the clinical isolates formed 104 clusters [37]. In the present study,  [36,38].
A feature in common to all four broiler producers was that the same Campylobacter isolates were detected in several consecutive flocks, i.e., that the Campylobacter isolates either survived in the chicken environment or originated from a common reservoir outside the broiler houses. For broiler producer A, the same Campylobacter isolate (ST-257) was isolated from the water pipes and from the chickens, indicating that the bacterium had managed to survive in the biofilm inside the water pipes between consecutive flocks. The difficulty in efficiently cleaning water pipes has been mentioned in other studies, which isolated Campylobacter from water pipes after disinfection [39,40]. Thorough cleaning of the water pipes by alternating air and water flushing was a measure taken at several of the broiler producers in this study to reduce the risk of next chicken flock being colonized with Campylobacter spp. After proper cleaning of the water pipes at broiler producer A, the ST found in water, ST-257, was not found in any of the later chicken flocks. The study also showed that single STs could occur in one flock rotation, but were removed during cleaning and disinfection between flocks. The novel feature of this study compared with previous studies is that preventive measures at farm level were introduced at each producer continuously during the project. In addition, similar measures were introduced at other chicken producers. The results and the measures implemented to improve hygiene barriers were discussed at meetings of broiler producers belonging to the Swedish Poultry Meat Association. As a result, cleaning water pipes by flushing with air and water was performed by producers throughout Sweden that delivered chickens with Campylobacter to slaughter in more than two consecutive flock rotations. Since a limited number of broiler producers contributed to the majority of the positive slaughter batches in Sweden [13,16], actions within a few broiler producers have a great impact on the prevalence of Campylobacter-positive broilers in the Swedish Campylobacter program. Since the actions applied on the broiler producers in this study were also implemented in other broiler producers, the measures contributed to decreasing the proportion of Campylobacter-positive flocks at national level from 15.4% in 2016 to 4.6% in 2019 [41], which is the lowest ever rate since the Swedish Campylobacter program was started in 1991. At the same time, the rate of reported human campylobacteriosis cases per 100,000 inhabitants in Sweden decreased from 110 in 2016 to 65 in 2019 [42].
The conclusion from this study is that there are different sources and transmission routes for the colonization of chickens with Campylobacter at different broiler producers, and thus individual measures have to be taken at each producer to prevent Campylobacter colonization.

Broiler Producers
Four broiler producers (A-D) were included in the study, each delivering between 70,000 and 480,000 broilers to slaughter around eight times per year. All broilers in the study were conventional broilers without any access to outdoors, and slaughtered at the age of 27 to 37 days. In Sweden, there are about 110 conventional broiler producers in total. Producer A had eight compartments of broilers and Each producer was visited at least once by veterinarians with experience of risk factor analysis of Campylobacter in broilers. During the visits, the broiler houses were carefully inspected, including the design of hygiene barriers and the surrounding environment. Possible risk factors and preventive measures to reduce the proportion of chickens with Campylobacter were discussed with the producer on these visits.

Sampling
A total of 626 samples from the internal and external environment of broiler houses, including different water sources, surrounding animals, transport crates, and parent flocks, were collected between 2017 and 2019. Internal and external surfaces were sampled by sock and swab samples, while water was sampled by filtering at least 50 L through a dialysis filter, as previously described [40]. Before taking samples from drinking water pipes, flushing with water and air at increasing and decreasing pressure was performed in all water pipes in the broiler houses. This removed the biofilm inside the pipes, which were included in the samples. The samples from surrounding animals, including wild animals, consisted of either feces or intestinal contents. Samples from fallow deer were collected in connection with hunting and wild boar either in connection with hunting or from fresh feces on the ground. The mice were trapped by mouse traps as part of the broiler producers' regular pest control. All samples from cattle, dogs, and wild birds consisted of feces samples. Transport crates were sampled at farm level before the broilers were loaded in, and after cleaning and disinfection at the slaughterhouse.
The sock samples consisted of one pair of sterile cotton tubular retention bandages (>7.5 cm × 25 cm Danafast; Mediplast AB, Malmö, Sweden) moistened with 30 mL Cary-Blair transport medium (SVA321645; National Veterinary Institute, Uppsala, Sweden) and placed over covered boots. The farmer was asked to walk around in the area, making sure all parts of the socks were in contact with feces by turning the socks around the shoe covers. Swab samples consisted of sterile non-woven swabs measuring 10 cm × 10 cm (Shaoxing Yibon Medical Co., Ltd, Shaoxing, Zhejiang, China), which were moistened with 30 mL Cary-Blair transport medium before swabbing of surfaces. After sampling, sock and swab samples were placed in sterile blender bags (Standard 400; Grade Products Ltd, Leicestershire, UK) and a further 30 mL of Cary-Blair transport medium was added to the bag, to keep the samples moistened during transport. Sampling at the farm level was performed both by In Sweden, all flocks sent to slaughter are analyzed regarding the presence of Campylobacter within the Swedish Campylobacter program [16]. Ten intact cecae from 10 broilers in a slaughter batch were collected after scalding and defeathering, but before washing and cooling of carcasses. The cecae were placed in plastic jars without transport medium and sent by regular mail, at ambient temperature, to the National Veterinary Institute, Uppsala, where they were analyzed as one pooled sample.
In addition, intestinal contents, feces samples from surrounding animals, and cecae sampled at slaughter were directly cultured on mCCDA plates by taking a loopful of sample. The mCCDA plates were incubated at 41.5 ± 0.5 • C for 48 ± 4 h. The microaerobic atmosphere was generated by CampyGen TM (CN0025; Oxoid) except for the cecum samples, where the microaerobic atmosphere was generated by the Anoxomat system (Mart BV, Lichtenvoorde, Netherlands). Suspected Campylobacter colonies were confirmed and identified to species level by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) (Bruker Daltonics, Billerica, MA, USA). In general, one Campylobacter isolate was cultured per sample except for the water samples, where several isolates were cultured per sample. All Campylobacter isolates identified were stored in Brain Heart Infusion (BHI) broth (CM1135; Oxoid) with 15% glycerol at −70 • C.

Whole-Genome Sequencing
Whole-genome sequencing (WGS) was performed on 256 Campylobacter isolates in total, of which 135 were isolated in this study and 121 were isolated previously at the National Veterinary Institute within the Swedish Campylobacter program. The isolates from the Swedish Campylobacter program were from the same four broiler producers as in this study and from the same period (isolation year in brackets); 62 were from broiler producer A (2017-2018), 23 from B (2017-2018), 20 from C (2018), and 16 from D (2018). A majority of the isolates were C. jejuni (n = 235), while 21 were C. coli.
Genomic DNA was extracted from isolates subcultured twice from single colonies on horse blood agar plates (SVAB341180; National Veterinary Institute) for 48 h at 41.5 • C in a microaerobic atmosphere, using the EZ1 DNA Tissue Kit and the bacterial protocol on an EZ1 Advanced XL (Qiagen, Hilden Germany). The elution volume used was 100 µL and the DNA concentration was measured using the Qubit ds DNA Broad Range assay kit on a Qubit ® 2.0 Fluorometer (Invitrogen, Carlsbad, CA, USA). Sample libraries were prepared using the Nextera XT DNA Library Preparation Kit (Illumina, San Diego, CA, USA). Whole-genome sequencing was performed on an Illumina NextSeq 500 system with 2 × 150 bp paired-end reads using the NextSeq 500 Mid Output kit V2 (Illumina).

Statistical Analysis
The results from the bacterial cultures were analyzed by Fisher's Exact test, performed using a statistical program on the Internet website "Social Science Statistics" (https://www.socscistatistics. com). The tests verified the association between on-farm measures and decrease in colonization of Campylobacter in chickens. A probability level of p < 0.05 was considered statistically significant.

On-Farm Measures
All broiler producers included in this study were visited at the start of the study by one or two veterinarians, and follow-up visits were made to producers A and B. During the visits, the broiler houses were studied, including the design of the buildings, anteroom, hygiene barriers, and the external environment. Possible risk factors were discussed and measures to reduce the proportion of chickens with Campylobacter were proposed and implemented.
Specific measures taken at broiler producer A, where biosecurity was already at a high level, were sock change in addition to the existing shoe change between different compartments, use of separate wheelbarrows for different corridors to transport dead broilers, and cleaning of water pipes in all departments with alternating air and water flushing. At broiler producer B, several additional hygiene barriers were introduced, the farmer made efforts to minimize the number of wild animals around the broiler houses, and the farmyard was scraped and new gravel was laid. In addition, the water pipes were cleaned with alternating air and water flushing and the pH of the drinking water to the chickens was lowered. At broiler producer C, fly screens were installed on the broiler houses after the results of cross-contamination from nearby cattle became known. No physical actions were implemented at producer D, but possible transmission routes and hygiene barriers were discussed with the farmer.