One thousand and ninety-two C. jejuni strains with high sequencing quality and complete information isolated from Asia, Europe and North America were selected from the BV-BRC database until March 25, 2024 (Olson et al., 2023). Subsequently, their genome files were extracted for our research (Supplementary Table S1).

Campylobacter jejuni genome files extracted from the BV-BRC database were used to construct the core gene phylogenetic tree. Through Snippy, C. jejuni genomes were compared with the genome file of C. jejuni strain NCTC11168 to generate a core Single Nucleotide Polymorphism (SNP) alignment. Subsequently, Gubbins was used to remove SNP sites in the recombinant regions to obtain a high-quality core SNP alignment (Croucher et al., 2014). Then, RAxML was used to generate the final phylogenomic tree (Stamatakis, 2014), which could be visualized via the interactive website tvBOT (Xie et al., 2023).

To obtain the impacts of the R-M system on the dynamics of the transfer of AMR genes, the REase and MTase genes related to R-M system were extracted from the REBASE database (Roberts et al., 2023). Using these genes collected from the REBASE database as reference, we compared the C. jejuni genome files downloaded with them to obtain the distribution of R-M system. We created a sample pool for local screening of C. jejuni strains with the R-M system. BLAST was performed based on the C. jejuni genome files downloaded from the BV-BRC database (Supplementary Table S2). The interactive website tvBOT was used to visualize the distribution of the R-M system in C. jejuni.

Choosing the National Center for Biotechnology Information (NCBI) database as the reference database (Feldgarden et al., 2019), acquired AMR genes in C. jejuni genome files were predicted using the ABRicate tool (Supplementary Table S3). The distribution of AMR genes in C. jejuni was visualized via tvBOT, and to explore the impacts of the R-M system on AMR genes, RStudio packages including ggplot2, RColorBrewer and ggpubr were used to analyze and visualize the differences in the number of AMR genes between C. jejuni with or without the R-M system. The significance was analyzed using the Wilcoxon-Mann–Whitney test.

The recombination_predictions.gff file and per_branch_statistics.csv file were both generated from the Gubbins process. The former one was used to visualize the homologous recombination predictions in C. jejuni via phandango (Hadfield et al., 2017). The latter one contained summary statistics for the number of recombination events reconstructed onto both the internal nodes and the leaves of the phylogenetic tree and along with the.recombination_predictions.embl file which covered detailed information about recombination predictions, we could calculate the total number of recombination events in one C. jejuni strain (Supplementary Tables S4, S5). It was used to analyze the differences of the recombination blocks between C. jejuni with or without the R-M system via RStudio packages including ggplot2, RColorBrewer and ggpubr. The significance was analyzed using the Wilcoxon-Mann–Whitney test.

Campylobacter jejuni genome files downloaded from the BV-BRC database were initially annotated using Prokka (Seemann, 2014). Subsequently, Pangenome analysis was performed using Roary on the generated GFF files (Page et al., 2015). To access the impacts of the R-M system on the distribution of unique and cloud genes in C. jejuni, using RStudio packages including ggplot2, RColorBrewer and ggpubr, we analyzed and visualized the differences of the number of these genes between C. jejuni with or without R-M system based on the gene_presence_absence.csv file obtained from the Roary process (Supplementary Table S6). The significance was analyzed using the Wilcoxon-Mann–Whitney test.

MGEs in C. jejuni genomes were predicted using MobileElementFinder with default parameters (Johansson et al., 2020) (Supplementary Table S7). Then, RStudio packages including ggplot2, RColorBrewer and ggpubr were employed to analyze and visualize the differences in the number of MGEs between C. jejuni with or without the R-M system. The significance was analyzed using the Wilcoxon-Mann–Whitney test.

Covariance is calculated using the following formula: Cov (x, y) in Rstudio. The x is the value of AMR genes or MGEs, and for y, it takes the value 1 when the Type III R-M system is present or 0 otherwise. The covariance value represents the changing trend of two-dimensional random variables. If the value is positive, the changing trend is consistent, or, if not, the changing trend is opposite.

Among C. jejuni strains selected for our study, according to the geographical distribution, 387 (387/1,092, 35.44%), 356 (356/1,092, 32.6%), and 349 (349/1,092, 31.96%) strains were isolated from Europe, North America and Asia, respectively. According to the host origin distribution, 397 (397/1,092, 36.45%), 340 (340/1,092, 31.14%), 167 (167/1,092, 15.29%), 91 (91/1,092, 8.33%), 54 (54/1,092, 4.95%), 35 (35/1,092, 3.21%) and 8 (8/1,092, 0.73%) strains were isolated from patients, poultry, livestock, birds, monkeys, the environment and black bears, respectively.

First, we analyzed the distribution of the R-M system in C. jejuni. Utilizing the REase and MTase genes related to R-M system extracted from the REBASE database, we constructed a sample pool for local screening of C. jejuni strains with the R-M system. BLAST was performed based on the C. jejuni genome files downloaded from the BV-BRC database. We found that 1,043 (1,043/1,092, 95.51%), 1,092 (1,092/1,092, 100%), 334 (334/1,092, 30.59%) and 1,080 (1,080/1,092, 98.90%) strains carry the Type I, Type II, Type III, and Type IV R-M systems, respectively. The Type I and Type III R-M system were distributed unevenly in C. jejuni.

We explored the distribution of AMR genes and the impacts of the R-M systems on the distribution of AMR genes in C. jejuni. Using the ABRicate tool, we identified the acquired AMR genes in C. jejuni and quantified them in the genome of the strains (Figure 1). We divided these strains into groups based on the presence of Type I R-M or Type III R-M systems, as well as whether they are R-M-free. We then analyzed the differences in the number distribution of the AMR genes between these groups (Figure 2). The results revealed that C. jejuni strains carrying the Type III R-M system had a significantly lower number of AMR genes compared to strains without this system (Wilcoxon-Mann–Whitney test, p < 0.0001). To evaluate the directional relationship between the number distribution of AMR genes and the presence of the Type III R-M system, we estimated the covariance value, which shows the relationship between two random variables. The covariance analysis revealed that the covariance value between the Type III R-M group and the Type III R-M-free group was −0.0526, suggesting a negative correlation between the number of AMR genes and the presence of the Type III R-M system. However, there were no significant differences in the distribution of the AMR genes between C. jejuni groups with a Type I R-M system or without it (Wilcoxon-Mann–Whitney test, p = 0.59).

Recombination, as a critical evolutionary force in bacteria, shapes the genomic landscape (Shikov et al., 2023). Using Gubbins, we reconstructed the homologous recombination blocks on the branches of the phylogenetic tree (Figure 3). The total number of the recombination blocks in each C. jejuni strain was counted. Through our research, the median number of homologous recombination blocks in the strains possessing the Type III R-M was 500.5, but in the Type III R-M-free strains, the median number increased by 19.38% to 597.5 (Figure 4, Wilcoxon-Mann–Whitney test, p < 0.0001), and the covariance value between these two groups was −14.1427, implying a negative correlation between them, which is consistent with the conclusion we have gained in our previous study. This suggests that the presence of the Type III R-M system could reduce the homologous recombination frequency in C. jejuni.

Cloud genes are defined as accessory genes found in 0 ~ 15% ofbacterial species. The HGT frequency of a C. jejuni strain can be evaluated by the number of the cloud and unique genes it carries (Xu et al., 2023). We analyzed the distribution of these genes between C. jejuni groups with or without the Type III R-M system to explore how the HGT frequency was affected by this system in C. jejuni population (Figure 5). Our analysis revealed no significant relationship between the number of cloud and unique genes and the presence of the Type III R-M system (Wilcoxon-Mann–Whitney test, p = 0.061 and p = 0.41, respectively).

AMR genes carried by MGEs can spread within C. jejuni population via HGT (Partridge et al., 2018). We compared the number of the MGEs in C. jejuni strains with or without the Type III R-M system to investigate whether the suppression of the acquired AMR gene distribution by Type III R-M system is related to a decline in HGT frequency (Figure 6). The results showed no significant differences in the number of the MGEs between C. jejuni groups with or without the Type III R-M system (Wilcoxon-Mann–Whitney test, p = 0.13).

In conlusion, our results suggest that the Type III R-M system might have limited effects on HGT frequency in C. jejuni.

Recently, C. jejuni AMR has become a severe global issue. Although the causes of C. jejuni AMR are complex, homologous recombination plays a central in enabling bacteria to acquire AMR elements (Partridge et al., 2018; Shikov et al., 2023). By integrating exogenous genetic materials, bacteria can pre-adapt to changing environment, thereby enhancing their viability. However, since gene acquisition can also lead to gene dysfunction and other side effects, bacteria have evolved ‘defense systems’ to protect their genetic integrity. The restriction-modification (R-M) system is the most prevalent defense mechanism, found in 83% of prokaryotic genomes, which is more than double the incidence of CRISPR-Cas systems (Tesson et al., 2022).

In this study, we first characterized the distribution of R-M systems in C. jejuni, revealing an uneven prevalence for both Type I (1,043/1,092, 95.51%) and Type III (334/1,092, 30.59%) systems. Given this disparity, to analyze the impact of R-M systems on AMR gene transmission in C. jejuni, we grouped strains based on the presence or absence of Type I and Type III R-M systems and compared the number of AMR genes between these groups. Our analysis revealed no significant differences in AMR gene distribution between strains possessing the Type I R-M system or not (Figure 2A; Wilcoxon-Mann–Whitney test, p = 0.59). Conversely, a significant negative correlation was identified between the presence of the Type III R-M system and the number of AMR genes in C. jejuni strains (Figure 2B; Wilcoxon-Mann–Whitney test, p < 0.0001; covariance = −0.0526). This finding suggests that the Type III R-M system may act as a barrier to AMR gene acquisition. The presence of that system may suppress the transmission of AMR genes by cleaving target exogenous DNA sequences, corresponding to the function of the REases.

AMR genes can spread within bacterial populations through homologous recombination, exacerbating the AMR problem (Partridge et al., 2018; Shikov et al., 2023). The macrolide resistance gene erm (B) could undergo natural transformation via homologous recombination among Campylobacter strains (Wei et al., 2024). Additionally, the horizontally acquired AMR genes could further be transferred via homologous recombination (Gratia, 2017). One study also shows that the horizontal transfer of chromosomal genes via lateral transduction through homologous recombination might provide significant opportunities for the rapid acquisition of virulence factors (Humphrey et al., 2021). To further explore mechanisms underlying the restriction effect on AMR genes transmission by the Type III R-M system, we conducted a series of analyses. Our results show that the presence of the Type III R-M system is associated with a decrease in the number of homologous recombination blocks in C. jejuni, strains lacking this system exhibiting a 19.38% higher median number of homologous recombination events (Figure 4; Wilcoxon-Mann–Whitney test, p < 0.0001; covariance = −14.1427). Thus, we propose that the Type III R-M system restricts the AMR genes transmission within C. jejuni mainly by limiting the frequency of homologous recombination. The homologous recombination frequency of the exogenous AMR genes in C. jejuni might be suppressed by the Type III R-M REases, which can cleave DNA sequences and restrict the transmission of AMR genes among C. jejuni population, resulting in fewer AMR genes in C. jejuni with the Type III R-M system.

Sharing of genes through HGT contributes importantly to the global dissemination of AMR genes, which could be susceptible to bacterial defense systems (Price et al., 2016; Huo et al., 2015). Proposing that the Type III R-M system might influence the acquisition of AMR genes by interfering with the HGT process, we conducted a series of studies. However, results show that there are no significant differences in the number of the cloud or unique genes between C. jejuni groups with or without the Type III R-M system (Figure 5; Wilcoxon-Mann–Whitney test, p = 0.061 and p = 0.41, respectively). MGEs, such as plasmids and the integrative and conjugative elements, disseminating via HGT, have been identified as key vehicles for the dissemination of AMR determinants (Botelho and Schulenburg, 2021). Through the MobileElementFinder software, we found and calculated the number of insertion sequences and unit transposons carried by C. jejuni strains with or without the Type III R-M system. Nevertheless, our research shows no significant difference between the number of MGEs in C. jejuni with or without that system (Figure 6; Wilcoxon-Mann–Whitney test, p = 0.13). Therefore, we suggest that MGEs might have limited correlation with the Type III R-M system in the transfer of AMR genes.