Multidrug-resistant enterobacteria in newborn dairy calves in Germany

We studied the prevalence of extended-spectrum beta-lactamase (ESBL)-producing Enterobacteriaceae in dairy calves as part of a routine health check protocol. In addition, data regarding antimicrobial use (AMU), farm hygiene, and farm management were collected in order to identify possible risks for ESBL occurrence. Ten farms participated in the study with a median of 781 milking cows (319–1701). All calves investigated were younger than two weeks with an average age of 6.8 (±3.9) days. The farms were visited and samples were collected twice at an interval of 7–11 months. Faecal samples diluted 1:10, were plated onto BrillianceTM ESBL agar in duplicates. After 24 hours at 37°C, colonies were counted and total colony forming units (cfu)/ml calculated. Bacteria species were identified biochemically. ESBL-production was phenotypically confirmed using the MICRONAUT-S β-Lactamases system. Additionally, antimicrobial susceptibility was tested using VITEK® 2 technology. Phylotyping of E. coli isolates and screening for bla genes was performed by PCR. ESBL-producing enterobacteria were detected on all farms and 96.5% of calves investigated shed ESBL-positive bacteria. Of all ESBL-producing isolates, the majority were E. coli (92.9%), followed by Enterobacter cloacae (5.1%) and Klebsiella pneumoniae subsp. pneumoniae (2.0%). The majority of E. coli isolates was clearly assigned to phylogroup C (25.0%), followed by phylogroups A (15.2%) and E (14.1%). CTX-M group 1 was most frequently detected (80.4%). E. cloacae contained blaCTX-M and blaTEM or blaSHV. K. pneumoniae harboured blaSHV only. Besides resistance to penicillins and cephalosporins, the majority of isolates was also resistant to one or more antibiotic classes, with a high proportion being resistant against fluoroqinolones. 52.5% of isolates were further characterised as threefold multidrug resistant gram-negative bacteria (3MDR-GNB) according to the German Commission for Hospital Hygiene and Infection Prevention. None of the isolates were 4MDR-GNB, i.e. none revealed carbapenem-resistance. Penicillins were the most frequently administered antibiotics to calves on most farms and were the predominant substance class at herd level on all farms. Overall, the number of calves treated prior to sampling was rather low (11.7%). Analyses of data regarding the farm management identified weaknesses in biosecurity and cleaning and disinfection. Besides beta-lactam antibiotics being the most commonly used antibiotics no other risk factors could be identified. In summary, the prevalence of ESBL-carriers in dairy calves was exceptionally high and should be motivation to develop strategies for the reduction of multidrug-resistant bacteria in farm animals.

Introduction regularly by the Clinic for Ruminants and Swine in the preceding years as part of an integrated veterinary herd health care service or other scientific projects. The farms kept a median of 781 (319-1701) milking cows and a median of 36  calves that were younger than two weeks. Calves were generally housed in individual pens or hutches for the first two weeks, either indoors (five farms) or outdoors (five farms). Calves on three farms were moved to group pens after 4-7 days. Waste milk was fed to the calves on seven of the ten farms. On one farm the milk was fed to all calves, on the remaining six farms only to bull calves. There was no data available on the proportion of waste milk containing antibiotic residues.

Data collection and animal sampling
The farms were visited twice with an interval of 7-11 months from April 2018 to November 2019. During these visits an interview with the herd manager and an on-site inspection of the farm was carried out. The hygienic status of the farms was recorded and rated using a standardized questionnaire developed for dairy cattle [20]. The questions could either be answered with 'no' or 'yes' (translating to 0 or 3 scoring points) or in the fashion of 'not fulfilled', 'partially fulfilled' or 'fulfilled' (translating to 0, 1 or 3 scoring points). Weighted average ratios (in the following simply referred to as ratios) were calculated to describe the overall hygienic status on the farms as well as different aspects of dairy husbandry, such as biosecurity, feed and water hygiene, birth management and housing of cows and calves. Using scores between 0.00 to 3.00, the hygienic status of the farms was classified as poor (ratio 0.00-1.49, indicated by the colour red), medium (1.50-2.39, yellow) or good (2.40-3.00, green). Data on animal health and antimicrobial treatment were collected through animal scoring and the herd management program. The obtained information on AMU was collected in a database especially designed for this study and was further evaluated regarding gross amounts of used substances on herd level as well as for the different age groups.
On each farm, at each sampling day, faecal samples of 10 randomly selected calves, aged 0-14 days, were collected at each sampling day manually from the rectum or, if observed and clearly assignable to the individual, picked up from the ground. This adds up to a total of 200 samples. The samples were collected during routine health check-ups and transported at 4˚C to the laboratory for further investigation. The presence of bovine coronavirus, bovine rotavirus, Cryptosporidium parvum, and E. coli K99 (F5) were examined using Speed V-Diar 4 TM (Virbac BVT, Bad Oldesloe, Germany) according to the manufacturer's instructions. These results will be reported elsewhere. The ID number, age and gender of each calf were documented. The study was approved by the Animal Welfare Officer of the Veterinary Faculty at Leipzig University.

Microbiological methods
All feacal samples were processed on the same day as collection. One gram of the sample was suspended in 9 ml NaCl-Trytpone broth (NaCl-T; pH 7.0; 1.0 g tryptone, 8.5 g NaCl ad 1000 ml Aqua dest.) and homogenized using a Laboratory Blender Stomacher 400 (Seward Limited, West Sussex, UK) for 60 sec. Homogenates were serially diluted 1:10 in NaCl-T. 100 μl of each suspension was plated in duplicates onto Brilliance TM ESBL agar (Oxoid Deutschland GmbH/ Thermo Scientific, Wesel, Germany) and incubated under aerobic conditions for 24 hours at 37˚C. A pre-enrichment step was not performed. The colonies were counted and total colony forming units (cfu)/ml calculated. All morphologically different colonies from the samples of each farm were isolated and sub-cultured onto Columbia sheep blood agar (7% sheep blood; Oxoid Deutschland GmbH/Thermo Scientific) and Gassner agar (Sifin Diagnostics GmbH, Berlin, Germany). The bacterial species was identified biochemically using the RapID™ ONE System (Thermo Scientific, Wesel, Germany). If test results were inconclusive, isolates were sent to an external laboratory (Diagnosticum, Neukirchen, Germany) for species identification using MALDI TOF. All faecal samples and isolates were stored at -80˚C after processing.
Further characterisation as described in the following, was performed on the first 100 isolates that were obtained.

Determination of bla-genes
Isolates that proved to be phenotypically positive for ESBL-production were screened for the presence of bla CTX-M groups 1 and 9, bla TEM , and bla SHV using PCRs as described by Gröbner et al. [23] and Strauß et al. [24]. The primer sequences are listed in S1 Table. Prior to PCR, one to three colonies of each isolate were suspended in 200 μl phosphate-buffered saline (pH 7.2) and heated at 99˚C for 10 minutes. The lysates were gently centrifuged and diluted 1:10 before being used for PCR.

Phylotyping of E. coli isolates
Phylogroups were assigned using the quadruplex-PCR as described by Clermont et al. [25] which is able to detect strains belonging to group A, B1, B2, C, D, E, F, and clade I. Phylogroup assignment was carried out by scoring the presence or absence of the genes arpA, chuA, yjaA, TspE4.C2, and trpA. The primers used are described in S1 Table. The E. coli strains described above were also used as a control in the phylo-typing PCR. In addition, the following E. coli strains were used: 305/11 (phylogroup A), RL 45 (phylogroup C) and RL 72 (phylogroup E). The latter two were kindly provided by Dr. Jens Hammerl, Federal Institute for Risk Assessment (BfR), Berlin.

Results and discussion
The objectives of this project were to study the prevalence of ESBL-producing enterobacteria in newborn calves on dairy farms, and to identify possible risk factors contributing to a high occurrence of ESBL-carriers. ESBL-producing enterobacteria were detected on all ten farms examined in the study. Of 200 faecal samples, 193 contained ESBL-producing enterobacteria which corresponded to a 96.5% prevalence of animals. This is much higher than reported in other studies performed in Germany, Switzerland, and Israel, which found an overall prevalence of 8.4%-56.2% and up to 26.2% in diseased calves [12,13,18,[26][27][28]. Major differences may relate to farm structures and farm practices, age of animals, and sampling on farms or at the slaughterhouse [12,13,28]. However, a similar amount (97.8%) of calves shedding ESBLproducers was found in the UK [13]. The highest prevalence of ESBL-E is usually found in young animals, notably suckling calves, which corresponds to our findings, but with increasing age ESBL-E prevalence gradually decreases [12,13,28,29]. The average age of sampled calves in this study was 6.8 (±3.9) days. In our study animals older than 14 days were not sampled.
The average amount of ESBL-producing bacteria per g faeces was 1.7x10 9 (±6.7x10 9 ) cfu/g faeces which is more than reported elsewhere [30]. It is noteworthy that a valid comparison of different studies is difficult due to different study designs, sampling techniques, and screening protocols. Although we decided against a pre-enrichment step, the prevalence determined in our study was much higher compared to others that used a pre-enrichment step [13,26,27].
A total of 100 ESBL-producing isolates was further characterized. These included all 87 isolates from the first round of sampling as well as 13 isolates that were found on the first 3 farms during the second round of sampling. The majority were E. coli (92/100; 92.0%), followed by Enterobacter (E.) cloacae (5/100; 5.0%) and Klebsiella (K.) pneumoniae subsp. pneumoniae (2/ 100; 2.0%). One isolate was an AmpC-producing Morganella morganae and thus excluded from the study. This distribution is in accordance to two studies from Switzerland which reported E. coli being the predominant ESBL-producing species in faeces from cattle, whereas ESBL-producing E. cloacae were found to a much lesser extent [12,26]. The occurrence of ESBL-producing K. pneumoniae in cattle faeces has only recently been described [31]. Of all faecal samples investigated, three contained more than one of the above mentioned ESBL-producing species. The majority of E. coli isolates was clearly assigned to phylogroup C (23/92; 25.0%), followed by phylogroups A (14/92; 15.2%) and E (13/92; 14.1%). Phylogroups B1, D, and F were found less frequently with ten (10.9%), eight (8.7%) and six (6.5%) isolates, respectively. None of the isolates belonged to phylogroup B2. For 18 out of 92 (19.6%) E. coli the unambiguous assignment to either phylogroup A or C was not possible using the PCRs of Clermont et al. [25]. These isolates are therefore referred to as members of phylogroup A/C. Overall, isolates belonging to phylogroups C, A/C, and A were predominant. This distribution is in contrast to other studies where phylogroups A and B1 were dominantly found in ESBLE isolates from healthy animals [18,[32][33][34][35][36]. The main reason for the differences compared to other studies might be that the PCR-protocols used elsewhere may not discriminate between phylogroups A and C [18,33,35,36]. However, differences in phylogroup distribution might also be attributed to geographic location, climate, feeding habits or other factors [37].
We screened 99 ESBL-producing isolates for the presence of bla genes. Detailed results can be obtained from Table 1. The majority of E. coli isolates (75/92, 81.5%) harboured bla CTX-M genes, with CTX-M group 1 predominating (80.4%). This is consistent with other German and European studies in which most frequently CTX-M group 1 followed by CTX-M group 9 were detected [13,18,27,35,38]. CTX-M group 9 was detected in five E. coli either as the only bla gene (n = 1) or in combination with other bla genes (n = 4). Only one isolate did not belong to either group 1 or 9. In 45 E. coli (48.9%), bla CTX-M was found in combination with bla TEM . Overall, the prevalence of bla TEM (61.9%) in E. coli either in combination with bla CTX-M or alone was higher compared to other studies [33,39]. Four E. coli (4.3%) were negative for CTX-M, TEM, and SHV. Of the ESBL-producing E. cloacae and K. pneumoniae, one E. cloacae contained bla CTX-M and bla TEM whereas the remaining four and the two K. pneumoniae harboured bla SHV . Studies on EBSL-producing enterobacteria other than E. coli in animals are scarce. In contrast to our findings, bla SHV was the dominant gene in E. coli compared to K. pneumoniae in a recent South African study [31]. To the authors best knowledge there is only one report on E. cloacae carrying bla SHV and bla TEM isolated from a sheep lamb in Switzerland [27].
We further characterised ESBL-producing bacteria for multidrug-resistance (i.e. as 3MDR-GNB or 4MDR-GNB). The definition given by the German Commission for Hospital Hygiene and Infection Prevention [6] is part of a national guideline which is used to guide infection prevention and control measures in human clinical settings [40]. Acknowledging the One Health concept and the risk of bacterial transmission between animals and humans in either direction, data on the occurrence of these MDR bacteria are of high importance for public health and animal health risk assessment. Besides resistance to penicillins and cephalosporins, 60 ESBL-producing isolates (60.6%) were additionally resistant to one or more antibiotic classes as determined by VITEK 1 2 ( Table 2, S2 Table) which is in accordance to other studies in cattle [13,27,35,41] and other food animals [27,35,42]. Concordant to Schmid et al. [13] and Eisenberger et al. [43], a high proportion of isolates (46.5%) was also resistant to fluoroquinolones. Most probably quinolone resistance is plasmid-mediated by genes co-located to bla CTX-M genes. However, even without plasmid-encoded insensitivity a strong association

PLOS ONE
between quinolone resistance and ESBL-production has been described [44,45]. Fiftytwo isolates (52.5%) were further classified as 3MDR-GNB. This finding was similar to results from ESBL-producing E. coli and K. pneumoniae isolated from humans [46]. None of the isolates proved to be 4MDR-GNB, i.e. all were susceptible to imipenem and meropenem (Table 2) which is contrary to a study from the UK that reported an unstable imipenem-resistant phenotype in 1.2% of their E. coli isolates [41]. In Europe, screening for carbapenem-resistance in livestock is still underrepresented and prevalence in European countries was reported as <1% [47]. In Germany, carbapenem-resistant Enterobacteriaceae were mainly associated with swine and chicken whereas only a single report on ertapenem-resistance in E. coli from a beef cattle farm is published [13,47]. The gross amounts (in g) of antimicrobial substances administered on farm level are shown in S3 Table. In contrast to other European countries such as Denmark or The Netherlands, reporting of AMU in dairy calves is not mandatory in Germany according to current jurisdiction [48]. Therefore, the results presented here highly depended on the quality of documentation at farm level. Unfortunately, data regarding the use of antibiotics on one farm could not be evaluated due to inadequate documentation. Of 180 calves on the other farms, for which information on treatment was available, 21 (11.7%) received antibiotic treatment prior to sampling, four of which were treated twice. On two farms none of the sampled calves were treated before. Substances administered to calves and their corresponding indications are given in Table 3. There was no correlation between the treatment of a calf with antibiotics and their status as ESBL-carrier (P = 0.7765).
Of all thirteen antibiotic substance classes documented, 8-11 were used on herd level on all farms, whereas for calves the use was less diverse (2-9 substance classes). Penicillins were most frequently used in calves on all but two farms (Farm 2 and 6), which mainly administered amphenicols (Fig 1). On herd level, penicillins were the predominant substance class on all farms. Similar results were reported for dairy herds in the UK [49] and cattle in Germany [50]. However, van Rennings et al. [50] did not differentiate between production type or age group. Fluoroquinolones were used on six farms which might account for the high amount of resistant isolates in our study. Cephalosporins were not administered to calves on any of the farms. A comparison of AMU among the participating farms regarding gross amounts of antibiotic substance was not possible because a reliable treatment index could not be calculated from the animal numbers obtainable through the herd management program. The feeding of waste milk to the sampled calves had no statistically significant effect on their status as ESBL-carrier (P = 0.1787).
The scores obtained from hygiene analyses during both visits at each farm are shown in Table 4. Scores for overall hygiene ranged between 1.90 and 2.64. On average, the highest scores were achieved for the subsections 'barn climate' and 'milking' whereas the subsections 'biosecurity' and 'cleaning and disinfection' were assessed poorest. Because of the nearly 100% prevalence of ESBLE no correlation to any possible risk factor could be evaluated.
In other studies, several factors contributing to a higher occurrence of ESBLE have been identified: e.g. cleanliness of calf housings and calving pens, crowdedness, not applying teat sealants in cows at dry off, parental treatment of clinical mastitis, feeding milk replacer to calves, or the use of third-and fourth-generation cephalosporins [15,28,35,51,52]. The latter, however, is discussed controversially [52]. Moreover, pig farms located within a small radius of the cattle barn have been associated with higher odds of ESBL occurrence which might be attributed to airborne dust-bound ESBLE [52,53].
One of the aims of this study was to elucidate possible links between farm hygiene, AMU, and MDR enterobacteria. In our study, we determined a 100% ESBL prevalence at farm level. On all farms beta-lactams were the most commonly used antibiotics, but because of the 100% prevalence of ESBLE at farm level no correlation to any possible risk factor could be evaluated.
The small sample number and the restriction to calves younger than two weeks can be seen as limitations of the study. Due to the fact that only the first 100 isolates obtained across the

PLOS ONE
project were further analysed, which represented all ten farms only in the first sampling, whereas in the second sampling only farms 1-3 were included, an overrepresentation of these farms cannot be excluded. The occurrence of ESBL-producing enterobacteria is most likely a multifactorial process. Hence, a longitudinal study including AMU and other relevant management factors, health and ESBLE status of cows and calves is required to identify risk factors contributing to a high prevalence of ESBLE in dairy herds. The high percentage of ESBLE-carriage in calves reported here, highlights the overall need to develop strategies for the reduction of multidrug-resistant enterobacteria in newborn calves.

Conclusions
Our study revealed a 100% prevalence of ESBL-producing enterobacteria in calves at farm level and a prevalence of 96.5% at the individual level. Beta-lactam antibiotics were the most frequently used antimicrobials on the participating farms. This may have contributed to the persistence of ESBL at farm level and possibly to a co-selection of resistance against other antimicrobial substances. Although ESBLE-prevalence has been described to decrease with increasing age of cattle, the high prevalence as well as the high number of ESBL-producing bacteria that are shed, require strategies to prevent the entry of ESBLE into the calf rearing system at an early stage such as prudent use of antimicrobials during drying off and diligent hygiene in calving pens and calf housing. Further investigation is needed, to define the entry point(s) of ESBLE into calf rearing.
Supporting information S1 Table. Primer sequences and sizes of PCR products used for the determination of blagenes and the extended quadruplex phylo-typing method.
(DOCX) 38. Ewers C, Bethe A, Semmler T, Guenther S, Wieler LH. Extended-spectrum β-lactamase-producing and AmpC-producing Escherichia coli from livestock and companion animals and their putative impact on