In Vivo Transfer and Microevolution of Avian Native IncA/C2 blaNDM-1-Carrying Plasmid pRH-1238 during a Broiler Chicken Infection Study

ABSTRACT The emergence and spread of carbapenemase-producing Enterobacteriaceae (CPE) in wildlife and livestock animals pose an important safety concern for public health. With our in vivo broiler chicken infection study, we investigated the transfer and experimental microevolution of the blaNDM-1-carrying IncA/C2 plasmid (pRH-1238) introduced by avian native Salmonella enterica subsp. enterica serovar Corvallis without inducing antibiotic selection pressure. We evaluated the dependency of the time point of inoculation on donor (S. Corvallis [12-SA01738]) and plasmid-free Salmonella recipient [d-tartrate-fermenting (d-Ta+) S. Paratyphi B (13-SA01617), referred to here as S. Paratyphi B (d-Ta+)] excretion by quantifying their excretion dynamics. Using plasmid profiling by S1 nuclease-restricted pulsed-field gel electrophoresis, we gained insight into the variability of the native plasmid content among S. Corvallis reisolates as well as plasmid acquisition in S. Paratyphi B (d-Ta+) and the enterobacterial gut microflora. Whole-genome sequencing enabled us to gain an in-depth insight into the microevolution of plasmid pRH-1238 in S. Corvallis and enterobacterial recipient isolates. Our study revealed that the fecal excretion of avian native carbapenemase-producing S. Corvallis is significantly higher than that of S. Paratyphi (d-Ta+) and is not hampered by S. Paratyphi (d-Ta+). Acquisition of pRH-1238 in other Enterobacteriaceae and several events of plasmid pRH-1238 transfer to different Escherichia coli sequence types and Klebsiella pneumoniae demonstrated an interspecies broad host range. Regardless of the microevolutionary structural deletions in pRH-1238, the single carbapenem resistance marker blaNDM-1 was maintained on pRH-1238 throughout the trial. Furthermore, we showed the importance of the gut E. coli population as a vector of pRH-1238. In a potential scenario of the introduction of NDM-1-producing S. Corvallis into a broiler flock, the pRH-1238 plasmid could persist and spread to a broad host range even in the absence of antibiotic pressure.

carbapenemase-producing bacteria in livestock might be underestimated in Europe, due to the voluntary basis for screening at the European Union level (12). One of the most common mechanisms leading to carbapenem resistance is the production of carbapenem-hydrolyzing ␤-lactamases, mainly encoded by the genes bla VIM , bla IMP , and bla NDM (which are responsible for the production of class B metallo-␤-lactamases), bla KPC (class A ␤-lactamases), and bla OXA-48 (class D ␤-lactamases) (13). Worrisome is the worldwide spread of these enzymes by mobile genetic elements, like integrated conjugative elements and plasmids (14). Among carbapenem-resistant/nonsusceptible bacteria, NDM-1-producing bacteria are usually not more virulent. However, due to many nosocomial outbreaks, they are regarded as the most harmful ones. This is linked to the broad geographical reservoirs of NDM-1 in many unrelated bacterial species, due to the location of the bla NDM-1 genes on broad-host-range plasmids (15). A recent study has revealed the localization of the bla NDM-1 gene on type 1 IncA/C 2 plasmid pRH-1238 (referred to as pRH-1738 by Fischer et al. [11]) in an avian native Salmonella enterica subsp. enterica serovar Corvallis strain (12-SA01738) isolated from a wild bird (Milvus migrans) in 2012 in Germany (11). The discovery of this first completely sequenced bla NDM-1 -fosA3-IncA/C plasmid (GenBank accession number KR091911.1) (16) is of great value due to its host and potential for dissemination into livestock production. This is additionally emphasized by the broad host range of IncA/C plasmids, allowing replication not only in Enterobacteriaceae but also in other bacterial species, such as Pseudomonas and Photobacterium damselae (14). Genome analysis of the pRH-1238 plasmid revealed the coexistence of several resistance genes [bla NDM-1 , bla CMY-16 , fosA3, sul1, sul2, strA, strB, aac(6=)-Ib, aadA5, aphA6, tet(A), mphA, dfrA17, and floR], facilitating resistance to carbapenems, fosfomycins, aminoglycosides, co-trimoxazole, tetracyclines, and macrolides (16). The above-mentioned studies and recent reports on VIM-1-producing Escherichia coli and S. Infantis isolates in swine and poultry farms (17,18) showed that the spread and persistence of carbapenemase-producing bacteria in wild birds and livestock are a reality. With recent reports of VIM-1-producing S. Infantis being simultaneously found in swine and minced pork meat in Germany (19), the concerns of human exposure via the food chain are additionally highlighted.
Commercial poultry production is a continuously evolving livestock branch, characterized by fast turnovers and production pressure, which, combined with poor management, could lead to the misusage of antimicrobials (20). This might also favor commercial broiler production acting as a niche for selection of multidrug-resistant bacteria. Furthermore, the intestinal tract of broiler chicken offers a cohabitat for different Enterobacteriaceae and E. coli, which is described as a major opportunistic pathogen in chickens with a potential for zoonotic transfer to humans (21). Therefore, it is of relevance to explore if and to what extent different genera and clonal lines might act as potential recipients of the bla NDM-1 -carrying plasmid pRH-1238. This is an important concern due to previous confirmation of the presence of multidrug-resistant and NDM-1-producing S. Corvallis bacteria in a wild bird. Knowing the broad host range of bla NDM-1 -carrying plasmids, our aim was to obtain an insight into a potential scenario for this introduction under experimental conditions but conditions that still mimic the rearing management practices common to commercial broiler production. Therefore, we aimed to investigate the intraspecies transfer (Salmonella to Salmonella) and interspecies transfer (Salmonella to endogenous gut microflora) capacities of this bla NDM-1 -carrying plasmid without inducing antibiotic selection pressure in a broiler chicken infection study. The objective of our study was (i) to determine the excretion dynamics of an avian native donor strain (S. Corvallis [12-SA01738]) and a poultryassociated recipient strain [D-tartrate-fermenting (D-Ta ϩ ) S. Paratyphi B (13-SA01617), referred to here as S. Paratyphi B (D-Ta ϩ )] at different inoculation time points, (ii) to analyze the in vivo broad-host-range capacity of the pRH-1238 plasmid, and (iii) to analyze the microevolution of the pRH-1238 plasmid (GenBank accession number KR091911.1) using whole-genome sequencing (WGS). p.i., 2 animals were shedding S. Corvallis, and on the 28th day of life 11 animals were shedding S. Corvallis.
Fecal excretion of transconjugants. In groups 1 and 3, the earliest Enterobacteriaceae transconjugants were detected at 3 days p.i., whereas in group 2, the earliest  detection was after 2 days p.i. (Table 1). E. coli transconjugants were detected in all groups, whereas Klebsiella transconjugants were detected only in group 2. During the 21 days p.i., S. Paratyphi B (D-Ta ϩ ) transconjugants were not detected.
On the basis of consensus sequence mapping of the pRH-1238 plasmid progeny to the reference sequence of pRH-1238 from the S. Corvallis reisolates in all groups, a deletion in transfer (Tra) region 1 (Tra1) (ϳ50 to 60 kb in size) was observed (Fig.  4). In one strain (G1-11d-T10), this led to the loss of bla CMY-16 . Another noteworthy result was the high percent sequence identity among the pRH-1238 progeny from selected Enterobacteriaceae transconjugants, in contrast to S. Corvallis reisolates ( Table 2 and Fig. 5).

DISCUSSION
Recent publications have reported the occurrence of carbapenem-nonsusceptible Enterobacteriaceae in wild birds, livestock, and food products and their spread, related to plasmid-mediated carbapenemases (11,(17)(18)(19). As carbapenems are not licensed for use in livestock, it is assumed that the occurrence of carbapenemase-producing bacteria is triggered by coselective pressure, since plasmids carrying bla NDM-1 , like the plasmid chosen for use in this study, commonly harbor multiple but variable resistance determinants (16,22). Still, current research shows that the spread of certain plasmidmediated resistance genes in broiler chickens is also possible without antibiotic selective pressure (20,23). Therefore, for understanding the mechanisms contributing to the spread of carbapenem resistance or carbapenem-nonsusceptible isolates in vivo, the objective of our animal trial was to explore the broad-host-range capacity and stability of a conjugative bla NDM-1 -carrying plasmid, IncA/C 2 plasmid pRH-1238, hosted by an S. Corvallis strain in chickens without antimicrobial selection pressure, representing the nonuse of carbapenems in livestock. With the help of WGS, such a setup enabled us to obtain an insight into the microevolution of the plasmid in vivo.
Challenge strain excretion. During our study, we observed prolonged fecal excretion of NDM-1 carbapenemase-producing S. Corvallis (12-SA01738), contrary to that of S. Paratyphi B (D-Ta ϩ ) (13-SA01617). Statistical analysis of data from group 1 [simultaneous inoculation of S. Corvallis and S. Paratyphi B (D-Ta ϩ )] revealed that the fecal excretion of S. Corvallis was significantly higher toward the end of the trial (the 16th, 21st, 25th, and 28th days of life) (Fig. 1) and was not hampered by the later inoculation of S. Paratyphi B (D-Ta ϩ ), as in group 2 (the difference between S. Corvallis excretion in groups 1 and 2 was statistically significant only by 8th day of life) ( Fig. 1 and 2). Because of previous studies reporting the invasiveness of S. Paratyphi B (D-Ta ϩ ) toward epithelial cells and macrophages and their presence in ceca, the liver, and the spleen (24), the decreased excretion of S. Paratyphi B (D-Ta ϩ ) observed in our in vivo trial is a noteworthy finding. Although this serovar is reported to be broiler associated, we have not observed a competitive advantage, contrary to the findings for S. Corvallis. On the other hand, the prolonged excretion of NDM-1-producing S. Corvallis in the absence of antibiotic pressure is an important concern due to its resistome and the broad host range of the pRH-1238 plasmid.
In vivo transfer of bla NDM-1 -harboring plasmid pRH-1238. In our study, we demonstrated the in vivo transfer of IncA/C 2 bla NDM-1 -carrying conjugative plasmid pRH-1238 from avian native S. Corvallis to E. coli strains belonging to phylogroups A, B1, and D, represented by four E. coli multilocus sequencing types (ST-117, ST-156, ST-2040, and ST-2485) and a K. pneumoniae isolate (ST-1106) ( Table 2). At the individual level, on particular sampling days, we observed pRH-1238 acquisition not only in different E. coli strains but also in different Enterobacteriaceae genera (Table 2). This, together with their rapid onset of excretion (Table 1), demonstrates the broad host range and the high Innermost circle, pRH-1238 coordinates; second-innermost circle, GC content of pRH-1238 reference plasmid; gray circles, group 1; green circles, group 2; yellow circles, group 3; outermost annotations, the resistome (black), bla genes (red), and deletions (dark gray) in Tra1 and the adjacent region, obtained using BRIG (50). Note in strain G1-11d-T10 pRH-1238 progeny the loss of the bla CMY-16 gene.
transferability of this multidrug resistance-conferring plasmid, leading to multidrug resistance acquisition in one horizontal gene transfer event. The affected species and genus (E. coli and Klebsiella) underline the importance of this concern due to their clinical relevance and ubiquitous distribution in the environment, acting as potential reservoirs of bla NDM-1 (25). This deserves attention, especially in commercial broiler production, where contamination pressure due to continuous rearing cycles as well as short interservice breaks could lead to the continuous propagation of pRH-1238 within a mixed bacterial population. With the previous detection of avian native NDM-1 carbapenemase-producing S. Corvallis in wild birds, such an entry scenario in commercial broiler production would presumably lead to rapid and diverse bla NDM-1 dissemination within a broiler flock even without antibiotic pressure. This might also lead to environmen- tal contamination, as has been observed for extended-spectrum ␤-lactamase (ESBL)/ AmpC-producing E. coli strains (26). The broad host range and the high level of transferability without antibiotic pressure should be kept in mind with the implementation of preventative measures. Instead of relying only on selective and coselective pressure as a measure to minimize carbapenemase-producing bacteria, further approaches assessing quantification of resistance gene dissemination with and without selective antibiotic pressure should also be considered.
The detection of enterobacterial transconjugant strains until the end of the trial (in group 1, from the 10th day of life onwards; in group 2, from the 9th day of life onwards; and in group 3, from the 13th day of life onwards) (Table 1) underlines that fact the plasmid acquisition has a presumably low or negligible fitness cost (27). Intestinal bacteria serve as reservoirs or even vectors for antibiotic resistance plasmids (28), findings which are further emphasized by the plasmid and resistance gene acquisition from the gut microflora observed in challenge strains ( Table 2).
As we did not detect NDM-1-producing S. Paratyphi B (D-Ta ϩ ) transconjugants, we assume that the host's E. coli population has an important influence on the reception and further spread of the pRH-1238 plasmid. This might be linked to the dense and diverse population and host gut adaptation of E. coli, serving as native recipients of pRH-1238. Although our in vitro filter mating conjugation experiments indicated a high transfer rate of the pRH-1238 plasmid to S. Paratyphi B (D-Ta ϩ ) (see Table S2 in the supplemental material), its absence in vivo might be linked to (i) serovar colonization dynamics, (ii) the abundance, diversity, and interference of E. coli strains, and (iii) the detection limit of the method used in this study (ϳ100 CFU/g). The majority of E. coli NDM-1 producers belonged to phylogroup A (represented by ST-2040); however, ST-117 and ST-156 strains were also detected. Besides being associated with poultry, strains of these STs are also described to be a potential source of not only ␤-lactam resistance genes but also polymyxin resistance genes (29)(30)(31). In a recent publication, a human-acquired mcr-1-carrying ST-117 strain of avian origin was characterized, highlighting the capability of this ST for resistance gene acquisition (30). The observed dominance of E. coli strains belonging to phylogroup A might resemble their occurrence in the gut or their ability to serve as native recipients for pRH-1238, as described for certain clonal lines dominant in the spread of the plasmid-mediated oqxAB gene encoding quinolone resistance (23).
Furthermore, bla CMY-16 is a variant of the bla CMY-2 lineage, which has been described to be the most common plasmid-mediated AmpC enzyme common to different Enterobacteriaceae worldwide (32). Therefore, the introduction of the pRH-1238 plasmid into a broiler flock should be assessed as well in light of the potential further dissemination of not only bla NDM-1 but also bla CMY-16 , which might be additionally propagated due to the use of cephalosporins in commercial poultry production. For future understanding, it is of interest to predict the dissemination potential of plasmid-mediated resistance genes relevant to public health and questioning the genera or serovars dominant in this exchange. Such data could contribute to a wider picture, broaden our knowledge for carbapenem resistance risk assessment, and serve as an asset for future approaches minimizing the spread of antimicrobial resistance in vivo.
Plasmid content variability in S. Corvallis reisolates. The observed native plasmid variability (plasmids of ϳ310 kb [IncHI2], 180 kb pRH-1238 [IncA/C 2 ], and Ͻ20 kb [ColRNAI]) was predominant in S. Corvallis reisolates from groups 1 and 2 ( Fig. S2 and  S3). This observation leads us to the assumption that the simultaneous and initial inoculation of S. Corvallis led to certain rearrangement mechanisms in native plasmid content, observed as the complete loss of the ϳ310-kb IncHI2 and Ͻ20-kb ColRNAI plasmids or partial region deletions in the ϳ310-kb IncHI2 plasmid (up to ϳ100 kb) and in the ϳ180-kb pRH-1238 plasmid (up to ϳ50 to 60 kb) (Fig. S2 to S4). We speculate that the earlier (7th day of life) inoculation of S. Corvallis in experimental groups 1 and 2 led to this occurrence. In a recent study by Card et al. (33) with a chemostat which mimicked the broiler microbiome, it seemed that the bacterial community stabilized by day 6. In our case, this unstable microbial population might support mobilome restructuring as well the interaction and subsequent acquisition of bla TEM-1B in S. Corvallis reisolates from group 1 in later stages of the trial (Table 2). Plasmid exchange and certain structural deletions might also be an important part of host adaptation regulation. Previous studies have reported that the acquisition and loss of certain genetic elements in bacteria are stimulated by the adaptation to the new environment, which influences their pathogenicity and might have subsequent consequences for human and animal health (34,35). As our study focused on NDM-producing Enterobacteriaceae detectable on xylose-lysine-deoxycholate (XLD) and chromID Carba agar and we did not conduct metagenomics analysis, we presume that bla TEM-1B might have originated from an E. coli ST-2040 strain. Furthermore, it seems that this sequence type played a significant role in plasmid exchanges (acquisition of pRH-1238 and ColRNAI and transfer of the ColpVC plasmid) with S. Corvallis ( Table 2).
Microevolution of pRH-1238 in S. Corvallis and enterobacterial transconjugants. Besides the plasmid content variability observed after S1 restriction for the IncHI2 (ϳ310-kb) and ColRNAI (Ͻ20-kb) plasmids in S. Corvallis reisolates, the largescale structural changes in pRH-1238 progeny were determined as deletions in Tra1 and downstream (ϳ50 to 60 kb in size) between two resistance islands: ARI-A [harboring sul1, bla NDM-1 , aph6, mphA, and aac6=lb] and ARI-B [harboring sul2, strA, strB, tet(A), floR, fosA3, sul1, aadA5, dfrA7] (16). These deletions did not lead to a significant alteration of the pRH-1238 ␤-lactam resistome, as only one strain did not harbor bla CMY-16 (Table 2 and Fig. 4), due to its position adjacent to Tra1 of pRH-1238. This occurrence is in line with observations indicating the large-scale structural changes often observed in neighboring areas of transposons and insertion sequence elements, indicating that these elements contribute to plasmid genome evolution (36). In a recent in vitro study by Porse et al. (37), deletions in the IncN plasmid (also constituting Tra regions) in recipient E. coli strains were observed, contrary to the findings for native K. pneumoniae and recipient Klebsiella strains. The authors stated that this occurrence might possess a potential competitive benefit for recipient E. coli strains. In contrast to the findings of our in vivo study, the deletions in the pRH-1238 progeny were dominant in S. Corvallis reisolates and not E. coli and K. pneumoniae strains (Fig. 4 and 5), suggesting that these deletions might be host or incompatibility group dependent. Generally, the observed losses of the IncHI2 and ColRNAI plasmids as well as deletions in the pRH-1238 progeny might indicate an evolutionary background in S. Corvallis adaptation which enables maintenance of the pRH-1238 resistome even without antibiotic pressure in wild birds.
A noteworthy observation was a Ͼ400-kb plasmid in sample G2-28d-T1 (Table 2 and Fig. S3) which seemed to be a fusion of IncHI2 and IncA/C 2 plasmid pRH-1238. This mobilome restructuring might be triggered by intrinsic S. Corvallis mechanisms and also linked with the persistence of bla NDM-1 in S. Corvallis. Namely, plasmid fusion and cointegration are frequent phenomena in plasmid evolution and adaptation and prevent, e.g., plasmid incompatibility and facilitate the interaction with a broad range of hosts (38). For a better understanding, it is of interest to explore if these occurrences are triggered by certain metabolic processes in the gut, bacterial stress, or a possible interaction with competitive gut microflora. Interestingly, pRH-1238 progeny from two strains sampled from the cecal contents showed a high percentage of sequence identity to the pRH-1238 backbone ( Table 2 and Fig. 4). Such an occurrence indicates that the S. Corvallis reisolates harboring native pRH-1238 exist in the intestinal tract and continuously disseminate pRH-1238 in vivo. Previous findings have reported on the higher level of colonization of Salmonella in the ceca, leading to higher rates of conjugation, which has been observed for a conjugative extended-spectrum cephalosporin resistance gene-harboring plasmid from S. Newport to E. coli strains and vice versa (39). Furthermore, deletions in pRH-1238 among Enterobacteriaceae transconjugants were minor and not attributed to Tra1, and the sequences of pRH-1238 progeny with these deletions revealed a higher degree of identity to the reference backbone of pRH-1238 (Table 2 and Fig. 5). This indicates that the pRH-1238 acquisition or transfer process itself might not lead to a significant alteration of pRH-1238 in transconjugant strains and that these strains might also serve as long-term reservoirs of pRH-1238 in vivo.
In conclusion, we demonstrated the prolonged fecal excretion of an avian native NDM-1 carbapenemase-producing S. Corvallis strain (12-SA01738) with microevolutionary deletions in the pRH-1238 backbone that preserved the bla NDM-1 gene during a broiler chicken in vivo study. The conjugative pRH-1238 IncA/C 2 bla NDM-1 -carrying plasmid was transferable to different Enterobacteriaceae, expanding its resistance gene pool among gut microflora in the absence of antibiotic pressure throughout the trial. This study shows at the molecular level how the rapid and diverse dissemination of bla NDM-1 -harboring IncA/C 2 plasmids in commercial broiler production can occur even in the absence of selective pressure. Furthermore, it highlights the need for understanding the mechanisms of the interaction of the host microflora and Salmonella serovars and calls for additional efforts in future intervention approaches to avoid the further spread of multidrug resistance plasmids in commercial broiler production.

MATERIALS AND METHODS
Challenge strains. Avian native Salmonella enterica subsp. enterica serovar Corvallis (strain 12-SA01738) of ST-1541 harboring the bla NDM-1 -carrying IncA/C 2 plasmid pRH-1238 (GenBank accession number KR091911.1) was selected as the donor strain. Native D-tartrate-fermenting (D-Ta ϩ ) Salmonella enterica subsp. enterica serovar Paratyphi B (13-SA01617), referred to here as S. Paratyphi B (D-Ta ϩ ), of ST-28, isolated in 2013, with intrinsic resistance to nalidixic acid was selected as the recipient. The pRH-1238 plasmid is the first completely sequenced bla NDM-1 -fosA3-IncA/C plasmid. It is 187,683 bp in size, has a GC content of 51.7%, and contains 173 predicted coding sequences (CDSs). It contains two resistance islands (ARI-A and ARI-B) and two transfer (Tra) regions (Tra1 and Tra2), with bla NDM-1 being located in ARI-A and bla CMY-16 being located in Tra1 (16). Besides pRH-1238, the donor strain harbors two additional plasmids of incompatibility group IncHI2 (ϳ310 kb) and ColRNAI (Ͻ20 kb), whereas S. Paratyphi B (D-Ta ϩ ) was selected as plasmid-free recipient strain. The phenotypic and genotypic properties of the donor and recipient strains are listed in Table S1 in the supplemental material. The selection of S. Paratyphi B (D-Ta ϩ ) was based on its high prevalence in commercial poultry production in Germany (40) as well as optimal in vitro conjugation transfer frequency (CTF) at 42°C (which corresponds to the average body temperature of birds) with S. Corvallis as the donor strain (Table S2). All strains were obtained from the strain collection of the National Reference Laboratory (NRL) for Salmonella in Germany.
In vitro filter mating conjugation experiments. Prior to our in vivo study, in vitro filter mating conjugation experiments with selected Salmonella strains (Table S2) were conducted to determine the average conjugation transfer frequency (CTF) for four potential recipient strains with S. Corvallis (12-SA01738) as the donor. After aerobic growth with gentle shaking at 37°C to obtain an optical density at 560 nm (OD 560 ) value of 0.25, a mixture of the Salmonella donor and recipient at a ratio of 1 to 2 was centrifuged (20,000 ϫ g for 2 min), inoculated on 0.45-m -pore-size filter membranes (Merck Millipore, Germany) that had previously been placed on lysogeny agar (LBA; Thermo Fisher Scientific, Germany), and incubated for 4 h at room temperature (RT), 37°C, or 42°C. Following incubation, the filter membranes were suspended in 4 ml of lysogeny broth (LBL; Thermo Fisher Scientific, Germany), decimally diluted, and plated on transconjugant selective plates (as described in Table S3). All filter mating conjugation experiments were conducted in triplicate in order to determine the average CTF rate (Table S2).
Broiler chicken infection study. For the in vivo trial, 33 broiler chicks (Ross 308) were randomly selected as 1-day-old chicks, without prior determination of the chick sex. Housing, clinical examination, individual labeling, and sampling followed. Animals were randomly divided into three experimental groups (group 1 [G1], G2, and G3), each containing 11 animals (animals T1 to T11) and housed in the facilities for animal experiments at the German Federal Institute for Risk Assessment, Berlin, Germany. In order to evaluate the dependency of the time point of inoculation on excretion of the challenge strains for the 28 days of the experiment, three experimental setups (groups 1, 2, and 3) were assembled. In group 1, the donor and recipient were simultaneously inoculated on the 7th day of life, whereas in group 2 (inoculation on the 7th day of life for the donor and the 10th day of life for the recipient) and group 3 (inoculation on the 7th day of life for the recipient and the 10th day of life for the donor), time-delayed inoculations were used. At the end of the experiment (at the 28th day of life), all animals were handled carefully following electrical stunning before being sacrificed for postmortem cecum extirpation. The experimental design containing the time frame and the related activities is shown in Fig. 6. During the experiment, microambient conditions complied with the hybrid management guide, and the animals were checked daily for evaluation of criteria for health and well-being. The animal trials were approved by the German State Authority for Health and Social Affairs (Lageso; no. 0308/15).
To prevent unintentional cross-reaction with intestinal microbiota, 1-day-old chicks were tested for possible colonization with (i) ESBL/pAmpC-or carbapenemase-producing E. coli using the laboratory protocol provided and recommended by the EURL for antimicrobial resistance (41)  Inoculation challenge and sampling plan. On the day of inoculation, both challenge strains were grown aerobically in LBL at 37°C with gentle shaking to obtain an OD 560 value of 0.35, which corresponded to a bacterial count of 4 ϫ 10 6 CFU per 100 l for both strains used as the inoculum. On day 7, animals were orally inoculated (for group 1, with the donor and recipient strains; for group 2 with the donor strain; for group 3, with the recipient strain), followed by a second inoculation (for group 2, with the recipient strain; for group 3, with the donor strain) on the 10th day of life. After inoculation, a 4-day consecutive sampling was performed, and further sampling was performed two times per week toward the end of trial (Fig. 6). Animals were always sampled individually in a particular time frame with preweighed cotton cloacal swabs (Deltalab, Spain) in order to determine the counts of the excreted challenge strains, expressed as the number of CFU per gram of feces.
Bacterial strain isolation. After suspending the fecal material (ϳ0.2 g) in 5 ml of 0.85% (wt/vol) NaCl, the suspension was subjected to decimal dilution and a 100-l deposition volume per plate was plated with an automatic spiral plater in duplicate on selective agar plates using the spiral colony counting technique with a Whitley automatic spiral plater (Don Whitley Scientific, UK). On the 1st day postinoculation (p.i.), dilutions of 1:10 and 1:10 3 were plated, and these were later adjusted to 1:10 and 1:10 2 on the basis of excretion dynamics. Challenge strain and transconjugant detection was based on growth on xylose-lysine-deoxycholate (XLD) agar (Thermo Fisher Scientific, Germany) with antibiotic supplementation (meropenem [0.125 mg/liter], cefotaxime [1 mg/liter], and/or nalidixic acid [50 mg/liter]), depending on the target strain (donor, recipient, or Salmonella transconjugants), and chromID Carba (bioMérieux, France) for detection of carbapenemase-producing Enterobacteriaceae (CPE) ( Table S3). Colonies suspected of being Salmonella were detected on XLD agar as red-yellow colonies with a black center, and CPE (e.g., E. coli, Klebsiellae) were detected on chromID Carba as purple and blue colonies. In order to further characterize the challenge strain [e.g., to characterize the variability in the plasmid content in S. Corvallis and plasmid acquisition in S. Paratyphi B (D-Ta ϩ )], reisolates from particular chicks within each group [S. Corvallis reisolates in group 1 (chick T10; see below), group 2 (chick T1), and group 3 (chick T3) and S. Paratyphi B (D-Ta ϩ ) reisolates in group 1 (chick T5), group 2 (chick T1), and group 3 (chick T3)] were preserved, whereas when possible four subcolonies (marked I to IV) of presumptive NDM-1 carbapenemase-producing Enterobacteriaceae transconjugants were selected for molecular characterization. Selected strains were denoted on the basis of the group (G1 to G3), day of life (1d to 28d), chick (T1 to T11), and subcolony (I to IV) origin.
Confirmation of presence of pRH-1238 in transconjugants. Transconjugants were screened by PCR amplification of bla NDM-1 and bla CMY-16 using a 1:10-diluted overnight culture as the template as described in previous publications (42,43). PCR mixtures (25-