Genome and Methylome analysis of a phylogenetic novel Campylobacter coli cluster with C. jejuni introgression

The intriguing recent discovery of Campylobacter coli strains, especially of clade 1, that (i) possess mosaic C. coli / C. jejuni alleles, (ii) demonstrate mixed multilocus sequence types (MLSTs) and (iii) have undergone genome-wide introgression has led to the speculation that these two species may be involved in an accelerated rate of horizontal gene transfer that is progressively leading to the merging of both species in a process coined ‘despeciation’. In an MLST-based neighbour-joining tree of a number of C. coli and C. jejuni isolates of different clades, three prominent Campylobacter isolates formed a seemingly separate cluster besides the previously described C. coli and C. jejuni clades. In the light of the suspected, ongoing genetic introgression between the C. coli and C. jejuni species, this cluster of Campylobacter isolates is proposed to present one of the hybrid clonal complexes in the despeciation process of the genus. Specific DNA methylation as well as restriction modification systems are known to be involved in selective uptake of external DNA and their role in such genetic introgression remains to be further investigated. In this study, the phylogeny and DNA methylation of these putative C. coli / C. jejuni hybrid strains were explored, their genomic mosaic structure caused by C. jejuni introgression was demonstrated and basic phenotypic assays were used to characterize these isolates. The genomes of the three hybrid Campylobacter strains were sequenced using PacBio SMRT sequencing, followed by methylome analysis by Restriction-Modification Finder and genome analysis by Parsnp, Smash++ and blast. Additionally, the strains were phenotypically characterized with respect to growth behaviour, motility, eukaryotic cell invasion and adhesion, autoagglutination, biofilm formation, and water survival ability. Our analyses show that the three hybrid Campylobacter strains are clade 1 C . coli strains, which have acquired between 8.1 and 9.1 % of their genome from C. jejuni . The C. jejuni genomic segments acquired are distributed over the entire genome and do not form a coherent cluster. Most of the genes originating from C. jejuni are involved in chemotaxis and motility, membrane transport, cell signalling, or the resistance to toxic compounds such as bile acids. Interspecies gene transfer from C. jejuni has contributed 8.1–9.1% to the genome of three C. coli isolates and initiated the despeciation between C. jejuni and C. coli . Based on their functional annotation, the genes originating from C. jejuni enable the adaptation of the three strains to an intra-intestinal habitat. The transfer of a fused type II restriction-modification system that recognizes the CAYNNNNNCTC/GAGNNNNNRTG motif seems to be the key for the recombination of the C. jejuni genetic material with C. coli genomes.


INTRODUCTION
The two most common disease-causing Campylobacter species are C. jejuni and C. coli, accounting for about 85 and 15% of all human Campylobacter infections, respectively [1]. Leading to abdominal cramps, watery or bloody diarrhoea, and post-infectious complications such as Guillain-Barré syndrome and reactive arthritis [2], these species are thought to cause about 200000 reported cases of campylobacteriosis every year [3].
Despite the fact that C. coli and C. jejuni share 86.5% identity in their seven multi-locus sequence typing (MLST) housekeeping genes (aspA, glnA, gltA, glyA, pgm, tkt and uncA), their identities as phylogenetically discrete species is confirmed by whole genome sequencing [4,5]. MLST further divides C. coli into three clades which, interestingly, also colonize distinct niches and have differing gene exchange behaviours. C. coli clade 1 is most frequently isolated from clinical samples as well as farm animals and undergoes high levels of intraclade and interspecies genetic exchange. Isolates from the closely related clades 2 and 3, which are highly distinct from clade 1, are commonly found in water environments and water birds and undergo little genetic exchange (5-10 times less than C. jejuni) [6][7][8].
Ecological separation is thought to be the primary factor for the maintenance of the phylogenetically distinct groups of C. coli and C. jejuni [5]. For example, C. coli is found to dominate in swine, while C. jejuni dominates in chicken and cattle [9]. Furthermore, different C. jejuni clonal complexes have been found to dominate in distinct wild bird species [7,10,11].
Recent observations of C. coli and C. jejuni isolates that possess mosaic C. coli/C. jejuni alleles [7], that represent mixed C. coli/C. jejuni multilocus sequence types (MLSTs) [5], and that have undergone genome-wide introgression [12], however, suggest that there may be a breakdown in the ecological barriers that have historically kept the two species separate.

Impact Statement
Campylobacter jejuni and Campylobacter coli are the most important food-associated bacterial pathogens for acute enteritis. Therefore, there is a need for monitoring along the food production chain from the poultry farm to the diseased individual. A phenomenon has been described for a subgroup of C. coli strains where genes are taken up from another bacterial species, C. jejuni. This progressive process has been observed to lead to the fusion of the two bacterial species in the course of evolution into a common bacterial species. Therefore, this process is called 'despeciation'. The mechanisms for this have been poorly described so far. By analysing the DNA sequences and also the DNA methylation of such hybrid strains, we reveal possible mechanisms for the uptake of genes from other bacterial species during despeciation. In particular, the role of restriction-modification systems, which degrade or spare DNA based on its methylation patterns, is discussed. By increasing the awareness of the existence of these hybrid strains, especially in the agricultural niche, it will also be possible to deal with these phenomena in a more structured way in diagnostics. Further studies investigating targeted DNA uptake from the other bacterial species should be performed based on the present data. In particular, C. coli clade 1 isolates have been observed to progressively accumulate C. jejuni DNA, with 10 and 23% of its core genome replaced by C. jejuni DNA in sequence type (ST)-828 and ST-1150 clonal complexes, respectively [12]. The impact of the increased horizontal gene transfer between the species is also seen in the emergence of multihost C. jejuni lineages that are capable of colonizing both birds and mammals [13]. Thus, the rapid expansion of a single C. coli lineage found in both agricultural animals and human disease [6] could be demonstrated. Furthermore, horizontal gene transfer plays a role in the widespread acquisition of antimicrobial resistance [14]. This progressive introgression of the C. jejuni and C. coli genomes has been hypothesized to eventually lead to a merging of the C. jejuni species, particularly with C. coli clade 1 isolates in the agricultural niche, in a process termed 'despeciation' .
In bacteria, DNA methylation plays a role in discriminating between self and extraneous DNA, which is a prerequisite for protecting the host genome from extraneous -sometimes invasive -DNA by restriction-modification (RM) systems [15]. RM systems typically consist of two components: a restriction endonuclease recognizing a specific (methylated) DNA motif and an associated DNA methyltransferase that methylates the same DNA and prevents or prepares its cleavage by the restriction endonuclease [16]. In the context of introgression, it is therefore likely that specific methylated DNA motifs and associated RM systems also play a role. Thus, it will be necessary to determine whether the selective activity of an RM system can facilitate the incorporation of C. jejuni DNA fragments into specific C. coli strains.
The three isolates meC0280, meC0281 and meC0467 reported here are Campylobacter strains that were isolated from turkey meat slaughtered in Berlin-Brandenburg, Germany. Despite being classified as C. coli by MALDI-TOF MS analysis [17], MLST-based analysis showed that this cluster is more closely related to C. jejuni and may have a possible hybrid species origin. In light of the suspected, ongoing genetic introgression between the C. coli and C. jejuni species, this study aimed to genotypically and phenotypically characterize these three isolates in order to investigate the potential hybrid origin of these isolates.

DNA extraction and MLST
Genomic DNA of all Campylobacter isolates was extracted using the QIAamp DNA Mini Kit (Qiagen) according to the manufacturer's instructions.
The MLS type was established using amplification and sequencing primers reported previously (https:// pubmlst. org/ campylobacter/ info/ primers. shtml). The cycling conditions were 94 °C for 1 min, followed by 35 cycles of 94 °C for 2 min, 50 °C for 1 min and 72 °C for 1 min, followed by a final elongation step of 72 °C for 5 min [18]. Amplicons of the seven genes included in the C. coli/C. jejuni MLST scheme were sent for sequencing to Microsynth Seqlab using 10 pmol of the respective sequencing primer. The mega x software for Linux was used for calculation of an MLST-based evolutionary history using the neighbourjoining method [19,20]. Evolutionary distances were computed using the Maximum Composite Likelihood method [21]. Apart from the three potential hybrid isolates, the remaining sequences of the MLST-STs were taken from the pubMLST database.

Alignment to C. coli-and C. jejuni-specific gene markers
From a study by Méric and co-workers, the gene sequences of 21 C. coli-and 27 C. jejuni-specific markers were obtained [22]. These gene sequences were aligned to the genomes of meC0280, meC0281 and meC0467 to determine their presence in their genomes and thereby determine the likely origin of these isolates. The alignments were done using the NCBI blast tool optimized for 'More dissimilar sequences (discontiguous megablast)' [23].

Library preparation and genome sequencing
High-molecular-weight DNA from all three isolates (meC0280, meC0281 and meC0467) was extracted using Qiagen Genomic Tip/100 G Kit (Qiagen) according to the manufacturer´s instructions. An SMRTbell template library was prepared according to the instructions from Pacific Biosciences following the Procedure and Checklist -Greater than 10 kb Template Preparation.
Briefly, for preparation of 10 kb libraries, 8 µg genomic DNA was sheared using g-tubes (Covaris) according to the manufacturer's instructions. DNA was end-repaired and ligated overnight to hairpin adapters applying components from the DNA/Polymerase Binding Kit P6 (Pacific Biosciences). Reactions were carried out according to the manufacturer's instructions. BluePippin Size-Selection to 4 kb was performed according to the manufacturer's instructions (Sage Science). Conditions for annealing of sequencing primers and binding of polymerase to purified SMRTbell template were assessed with the calculator in RS Remote (Pacific Biosciences). SMRT sequencing was carried out on the PacBio RSII (Pacific Biosciences). In total one SMRT cell per strain was sequenced. In parallel, short-read sequencing was performed from the same DNA on a MiSeq (llumina).

Genome assembly and bioinformatics analysis
The generated SMRT cell data were assembled using the 'RS_HGAP_Assembly.3' protocol included in SMRT Portal version 2.3.0 using default parameters. For PacBio long-read assemblies, 60841 post-filtered reads with an average read length of 13024 bp were used for strain meC0280, 90268 post-filtered reads with an average read length of 12937 bp were used for strain meC0281, and 59866 post-filtered reads with an average read length of 14066 bp were used for strain meC0467. Chromosomal contigs were trimmed, circularized and adjusted to dnaA (chromosomal replication initiation protein DnaA) as the first gene. Extrachromosomal elements were trimmed and circularized. The validity of the assembly was checked using the 'RS_Bridgemapper.1' protocol included in SMRT Portal version 2.3.0. Finally, each genome was errorcorrected against InDel errors by a mapping of generated Illumina reads onto the respective PacBio genome using the Burrows-Wheeler Alignment tool (BWA) [24] with subsequent variant and consensus calling using VarScan 2 [25]. A consensus concordance of QV60 could be confirmed for each of the genomes. Finally, automated genome annotation was performed using Prokka 1.8 [26].

Phylogenetic analysis
The three genomes generated in this study (meC0280, meC0281 and meC0467), 20 C. coli clade 1 genomes, one C. coli clade 3 genome and four C. jejuni genomes were analysed using the Parsnp application. The phylogenetic tree was midpoint-rooted and visualized using Figtree (http:// tree. bio. ed. ac. uk/ software/ figtree/).

Sequence analysis
For identification of methylated bases and modification motifs, the RS_Modification_and_Motif_Analysis.1 protocol within SMRT Portal version 2.3.0 was used with standard parameters on the basis of the assembled genomes of strains meC0280, meC0281 and meC0467. Putative restriction modification systems have been identified using the Restriction-ModificationFinder-1.0 server (available at https:// cge. cbs. dtu. dk/ services/) based on the Restriction Enzyme database (REBASE, www. rebase. neb. com) [56]. The consensus sequences were illustrated as sequence logos obtained by the WebLogo 2.82 server (http:// weblogo. berkeley. edu/).

Smash++ analysis
To analyse the composition of the putative hybrid strains meC0280, meC0281 and meC0467, we used the novel software tool Smash++ [27]. To this end, we divided these genomes into non-overlapping segments of 1000 bp each and used Smash ++ to match them against the genomes of C. jejuni NCTC 11168 and C. coli BfR-CA-9557. The aim was to identify 1,000 bp genome segments that are either more likely to be C. coli, more likely to be C. jejuni, or almost equally likely to be C. coli or C. jejuni.
For the matching procedure, Smash ++ was run with filter size of 100, sampling steps of 180 and compression level of 3. As output, position, size and relative redundancy of similar regions in each 1000 bp genome segment of hybrid strains and C. coli or C. jejuni homology were obtained. To measure the similarity, we converted relative redundancy (RR), which is a dissimilarity measure with values between 0 and 2 [27], to a similarity measure 1−RR/2, with values between 0 and 1, and used it for the calculations explained in the following by an example. Consider a segment x in strain meC0280 -with 1000 bp length -is similar to a segment y in the C. coli genome -again with 1000 bp length -and the relative redundancy of x rather than y is 1.4, i.e. RR x:y =1.4, and also RR x:y =1.6. We have Sim x:y =1−RR x:y /2=0. 3 and Sim x:y =1−RR x:y /2=0.2, meaning that 0.3 of x is similar to 0.2 of y. The total similarity is, then, calculated as in which x and y denote the lengths of x and y , respectively.
The genome segments with a larger C. jejuni identity were then blasted against the genome of C. jejuni NCTC 11168 using Geneious Prime 2019.3.2 (https://www. geneious. com), in order to identify the genes that potentially integrated into the genome of the hybrid strains.

Phenotypic assays
In addition to the genomic analyses, the three hybrid strains (meC0280, mec0281 and meC0467) were phenotyped in comparison to C. coli (BfR-CA-9557 and RM 2228) and C. jejuni (NCTC 11168 and 81116=NCTC 11828). Growth curves, motility, eukaryotic cell adhesion, eukaryotic cell invasion, autoagglutination, biofilm formation and water survival were studied comparatively (see Data S3, available in the online version of this article).

Species confirmation and MLST
Species identification of the three hybrid strains as C. coli was performed using the MALDI Biotyper system (Bruker Daltonics). Results with MALDI Biotyper identification score values ≥2.000 were considered correct. Additionally multiplex PCR was used to discriminate between C. jejuni and C. coli [28]. MLST of the three presumptive C. coli isolates meC0280, meC0281 and meC0467 resulted in  previously; instead they branch off at the base of the C. jejuni clade (Fig. 1). Due to this unique MLST-based phylogeny, the complete genomes of these three isolates were sequenced.  Cj81116_1523 Gene identifier in the re-annotated genome.

C. coli-and C. jejuni-specific genes
Méric et al. identified 21 genes specific for C. coli and 27 genes specific for C. jejuni based on analysis of 192 genomes that can be used to differentiate the individual microbial species [22]. blast searches of these 48 genes within the genome sequences of the (at this point of the study 'potential') hybrid strains showed that the hybrid strain genomes aligned to all the 21 C. coli-specific gene markers (100%) with percentage identities averaging 93% (ranging between 84 and 98%) and query coverages averaging 100% (ranging between 98 and 100%). In contrast, the hybrid strain genomes aligned to only eight or nine of the 27 C. jejuni-specific gene markers (37%) with percentage identities of the aligned gene markers averaging 82% (ranging between 72 and 98%) and query coverages averaging 62% (ranging between 30 and 100%).
These results indicate that the hybrid strains are C. coli isolates that have genes from C. jejuni integrated into their genome.

Core genome-based dendrogram
The core genome-based phylogeny also clearly shows that the isolates meC0280, meC0281 and mec0467 are essentially C. coli isolates of clade 1. Nevertheless, the evolutionary distance clearly shows that these three isolates, in particular meC0280, are much more closely related to the C. jejuni isolates than, for example, C. coli BfR-CA-9557 (Fig. 2).

Smash++ analysis
The Smas h++ algorithm fragmented the genomes of meC0280, meC0281 and meC0467 into 1724, 1912 and 1662 1000 bp fragments. Of these, only 157 (9.11%), 155 (8.11%) and 137 (8.24%), respectively, showed a higher similarity to C. jejuni, while 1225 (71.06%), 1179 (61.66%) and 1127 (67.81%) were assigned to C. coli. No clear assignment was possible for 342 The PacBio SMRT Analysis software was used to identify methylation motifs. The sequence consensus of the motif is shown in column 2 whereas the reverse complement of the motif, the partner motif, is shown in column 10. IUPAC ambiguity codes represent non-uniform positions. The position of the modified base within the motif and the type of methylation are indicated in columns 3 and 4. Column 5 denotes the percentage of a motif's occurrences in the genome (column 7) for which a methylation has been detected (column 6). Column 8 lists the average modification quality (in Phred Q-scores) and column 9 the average coverage of motifs detected as modified.
*QV, quality value (19.84%), 578 (30.23%) and 398 (23.95%) fragments (Fig. 3b). Segments with a higher similarity to the C. jejuni genome are scattered over the entire genome of the hybrid strains and do not form a coherent cluster (Fig. 3a). The increased number of genome parts of mecC0281, which can equally be assigned to both C. jejuni and C. coli, is increased due to the presence of two ECEs in this isolate. Again, the hybrid strains are rather C. coli isolates of clade 1, containing at least 8.11-9.11% C. jejuni DNA. blast analysis of these 'Smash++'-fragments revealed 141, 136 and 124 genes, respectively, which probably originate from C. jejuni (see Data S1, Tables 1-3). Note: the Smash ++ algorithm analysed 1000 bp segments for their similarity to comparable segments in C. coli and C. jejuni. Due to the fact that some genes are also significantly larger than 1000 bp, sometimes up to 2000 bp or occasionally 3000 bp, the number of genes is reduced compared to the segments. These genes encode proteins involved in various functions (illustrated in Fig. 4 using the example of meC0280). What stands out is that many of the genes taken up from C. jejuni play a role in chemotaxis and motility (e.g. fliD), membrane transport, regulation and cell signalling, and resistance to antibiotics and toxic compounds. As compared to the whole genome of the clade 1 C. coli isolate BfR-CA-9557, the chemotaxis and motility genes taken up from C. jejuni in meC0281 account for 13% instead of the expected 5.7%, membrane transport genes account for 10% instead of 3.3%, genes for regulation and cell signalling account for 3% instead of 1.1%, and resistance to antibiotics or toxic compounds genes account for 5 % instead of 1.9%.
This suggests that the uptake of genetic material from C. jejuni is selective and not merely random and lineages with these Fig. 4. Functional subsystems identified in 141 genes of the hybrid strain meC0280 originating from C. jejuni. This pie chart illustrates the functional subsystems of the 141 genes of the hybrid strain meC0280, which presumably originated from C. jejuni. In particular, the two categories motility and chemotaxis as well as membrane transport dominate. Isolate meC0280 was chosen as an example because with 141 genes of putative C. jejuni origin compared to the other two strains, it contains the most genes of putative C. jejuni origin.
genes seem to have acquired selective advantages. Most of these genes do not belong to the core genome and are absent in at least one of the three genomes. Only 23 genes with a probable C. jejuni origin are present in all three genomes (Fig. 5, Data S1, Tables S4-S10). These shared genes will be discussed in more detail.
One of the shared genes encodes a fused type II RM system consisting of a DNA-methyltransferase M and a restriction endonuclease R. It is annotated as Cco280III in meC0280 (annotation according to REBASE; ORF 1382; position in genome 1002805-1006899; locus tag: DSM101856_01047; Table 2) and recognizes the motif CAYNNNNNCTC and the corresponding partner motif GAGNNNNNRTG. (Table 3, Fig. 6). In C. coli meC0281 and meC0467 homologous genes encoding a fused type II RM system, Cco104626I and Cco104627ORFBP, are present. The difference between these two is that REBASE predicts the same recognition motif for Cco104626I (meC0281, ORF 1065, position in genome 771536-775630; locus tag: DSM104626_00799) as in meC0280, whereas CACNNNNNGT is the predicted recognition motif of Cco104627ORFBP (meC0467, ORF 977, position in genome 727644-731651; locus tag: DSM104627_00741, Data S2). Canonical type II RM systems are typically composed of two homodimeric R (restriction) subunits and a separate M (methylation) protein. Both act independently and recognize the same methylated palindromic DNA motif [29]. However, in the case of Cco280III and the homologues in the other two hybrid strains, the RM system consists of a single protein, in which the restriction subunit R is fused with the DNA methyltransferase subunit M. This RM system is conserved in the context of this analysis. For the type II RM system encoded by cj1051c in C. jejuni it could be shown that it reduces the transformation efficiency for plasmids [30]. Additionally, Beauchamp and co-workers identified the so-called Campylobacter transformation system methyltransferase (CtsM) [31], which is also present in C. coli BfR-CA-9557 (clade 1) [32]. CtsM recognizes the RA m6 ATTY motif, and DNA originating from a ctsM knockout mutant transforms C. jejuni significantly less effectively than DNA derived from ctsM-expressing bacteria [31]. Thus, Cco280III and homologues in other hybrid strains may bear the function to prevent disadvantageous changes in the chromosome but promote uptake of advantageous DNA, e.g. from C. jejuni strains. From this point of view, DNA methylation could be a key factor in the emergence of C. coli-C. jejuni hybrid strains.
Another gene present in all three hybrid genomes is fucP, which encodes a fucose permease that enables a subgroup of C. jejuni isolates to metabolize l-fucose [33]. The presence of fucP has been associated with livestock (bovine) habitats of C. jejuni, and it has also been deduced that fucose permease is an important prerequisite for residing in the mucosal layer [34].
Among the genes common to all three hybrid strains with putative C. jejuni origin are the three multidrug efflux RND transporter permease subunit genes cmeB, cmeE and cmeF. Both multidrug efflux pumps, CmeABC and CmeDEF, comprise a periplasmic fusion protein, CmeA or CmeD, an inner membrane efflux transporter, CmeB or CmeE, and an outer membrane protein, CmeC or CmeF. In particular, the genes of the two periplasmic fusion proteins, CmeA or CmeD, seem to have been taken up from C. jejuni and integrated into the genome of the hybrid strains. CmeABC and CmeDEF are involved in bile resistance [35]. Bile acids, especially deoxycholic acid, chenodeoxy-cholic acid and glycocholic acid, have been demonstrated to induce the expression of cmeA, cmeB, cmeC and cmeE [36]. The uptake of these genes from C. jejuni might be an adaptation to an intestinal bile acidcontaining habitat. Bile acids in turn cause oxidative stress in the bacterial cells, so that adaptations to oxidative stress must also be understood in the context of adaptation to an intestinal habitat containing bile acids.
Another common gene is mutS2. It encodes the recombination inhibitory protein MutS2. MutS2 has been shown to play an important role in repairing oxidative DNA damage and it has anti-recombination activity in Helicobacter pylori. MutS2 maintains the integrity of the genome by suppressing homologous and homoeologous DNA recombination [37,38].
The gene encoding elongation factor 4 lepA is a further common gene in the three hybrid strains. EF4/LepA binds to the post-translocation as well as to the pretranslocation ribosomal complexes and regulates the elongation cycle and thus protein synthesis especially under specific stress conditions [39]. It has been demonstrated that EF4/LepA is retained at the inner cell membrane of Escherichia coli and released into the cytoplasm at high intracellular ionic strength or low temperature [40]. EF4/LepA protects cells from moderate stress by allowing stress-paused translation to resume, but at high stress levels it acts in a mechanism that accelerates cell death by accumulation of reactive oxygen species [41].
Other common proteins of C. jejuni origin involved in amino acid metabolism and transcription are pantoate-beta-alanine ligase, chorismate-binding protein, cation:dicarboxylase symporter family transporter, and DNA-directed RNA polymerase subunit beta RpoB.
As already observed in the genome of clade 1 isolate BfR-CA-9557, three homologues to the iron transport protein TonB are present in all three hybrid isolates [32]. The TonB-dependent ferri-enterochelin receptor gene cfrA (cj0755) is also one of the common genes apparently taken up from C. jejuni. The outermembrane receptor protein CfrA binds ferri-enterochelin and initiates the uptake via the inner-membrane ABC transporter system CeuBCDE [42]. The presence of cfrA (cj0755) and other putative C. jejuni iron-uptake systems was associated with better livestock adaptation [34].
The remaining common genes that were apparently taken up from C. jejuni can be divided into two categories. The 'respiration and phosphate metabolism' category includes the genes for formate dehydrogenase subunit alpha, FAD-binding oxidoreductase, phosphate ABC transporter ATP-binding protein, and esterase-like activity of phytase family protein.
The category 'DNA synthesis and replication' includes the genes encoding the type I DNA topoisomerase and the ribonucleoside-diphosphate reductase subunit alpha.

Phenotypic assays
In addition to genome analysis, the hybrid strains were examined for growth behaviour, motility, invasion and adhesion ability, biofilm formation and autoagglutination in comparison to two C. jejuni (NCTC 11168 and 81116) and two C. coli (BfR-CA-9557 and RM 2228) reference strains. In phenotypic testing, differences of varying significance were observed for some of the parameters both within the hybrid strains and in comparison to the C. jejuni and C. coli reference strains (see Data S3).
Statistical analysis revealed that strain meC0467 was significantly more motile than strain meC0280 (P<0.05). Strain meC0467 demonstrated a significantly (P<0.05) lower adhesion to Caco-2 cells compared to the two other hybrid strains. Also, strain meC0281 showed a significantly higher tendency to autoagglutination than the strains meC0280 and meC0467 (P<0.05). Finally, it was observed that strain meC0467 formed significantly more biofilm than meC0280 and meC0281 (P<0.001). No significant differences were detected between the hybrid strains with respect to water survival and Caco-2 cell invasion.
In the comparison of growth dynamics, meC0280 in particular stands out, which grows about twice as fast as NCTC 11168, 81116, BfR-CA-9557, meC0281 and meC0467. Only C. coli RM 2228 also demonstrates an average generation time in the range of 120-140 min. In relation to the reference strains, significant differences between two of the hybrid isolates and the C. coli and C. jejuni reference strains were also observed in their motility. Strain meC0280 was significantly (P<0.05) less motile compared to C. jejuni NCTC 11168 as well as 81116, but meC0467 was significantly more motile than C. coli BfR-CA-9557. Differences in adhesion to Caco-2 cells in the sense of stronger or weaker host cell adhesion were likewise observed within the reference strains, so that general tendencies were not discernible. No significant differences were observed in Caco-2 cell invasiveness.
Concerning the above-mentioned strong differences in the tendency to autoagglutination within the hybrid strains, which is greater than that between the reference strains, no general statement can be made regarding the autoagglutination comparison with the reference strains. Regarding biofilm formation, strain meC0475 stood out, showing significantly (P<0.05) higher biofilm formation in our experiment compared to all other isolates tested except NCTC 11168. The C. jejuni strains survived statistically better in water than the C. coli strains (P<0.05). Here, two of the hybrid strains, meC0281 and meC0467, showed a statistically significant difference to the C. jejuni strains (P<0.05) and hence a greater similarity to the C. coli strains (P>0.05), which is in accordance with the higher C. coli genome content of the hybrid strains.

Conclusions
This study on three genome-sequenced C. coli-C. jejuni hybrid strains shows that the process of despeciation of C. coli and C. jejuni continues to progress. C. coli isolates from clade 1 were already known to have incorporated genetic material from C. jejuni. In some of the isolates described here as hybrid strains, the C. jejuni genome proportion is significantly higher (approx. 8-9 %) than in non-hybrid clade 1 C. coli isolates. Clade 2 and clade 3 C. coli isolates are found almost exclusively in environmental waters. If one understands this as a reference point, then the uptake of C. jejuni genetic information occurs mainly in the framework of adaptation to intestinal habitats in livestock and thus to conditions with bile acids and thus oxidative stress.
A type II restriction-modification system (Cco280III in meC0280) which, according to our analyses, itself originally is derived from C. jejuni and that recognizes the motif CAYNNNNNCTC/GAGNNNNNRTG, may play a key role in the uptake of C. jejuni genetic material. While appropriately methylated DNA fragments are tolerated and consecutively integrated into the chromosome, DNA sequences are degraded without appropriate methylation.

Funding information
The authors received no specific grant from any funding agency.