Campylobacter fetus Plasmid Diversity: Comparative Analysis of Fully Sequenced Plasmids and Proposed Classification Scheme

Abstract

Campylobacter fetus is an animal pathogen that contains 2 mammal-associated subspecies: Campylobacter fetus subsp. fetus (Cff) and Campylobacter fetus subsp. venerealis (Cfv) including its biovar intermedius that exhibit different biochemical traits and differences in pathogenicity. Although plasmids are important in the horizontal transfer of antimicrobial resistance genes and virulence factors, C. fetus plasmids are understudied. Here, the closed sequences of 12 plasmids from Spanish C. fetus isolates were compared with the publicly available DNA sequences of C. fetus plasmids and other members of the Campylobacterales order. Sizes of C. fetus plasmids from Spanish isolates ranged between 4 and 50 kb and most of them (10/12) were potentially conjugative. Comparative analysis of the plasmids’ gene content revealed a close genetic relationship between the plasmids of C. fetus isolated in Spain and those from other geographical regions, while being clearly distinct from plasmids of other Campylobacter species. Furthermore, C. fetus plasmids were grouped into two main clusters regardless of their geographic location or lineage. The distribution pattern of relaxase, replicase, and single-stranded DNA binding SSB protein encoding genes showed a clustering comparable to that resulting from plasmid whole gene content analysis, suggesting its potential use for the classification of C. fetus plasmids. Most of the larger plasmids harbored mobile genetic elements. These results can help to better understand the evolutionary dynamics and pathogenic implications of C. fetus plasmids.

Introduction

Campylobacter fetus is an important animal pathogen that includes three subspecies: Campylobacter fetus subsp. fetus (Cff), Campylobacter fetus subsp. venerealis (Cfv), and Campylobacter fetus subsp. testudium (Cft) (WOAH 2021). Despite the close similarity of their genomes, Cff and Cfv differ in host specificity, clinical signs, and pathogenicity. Cff can colonize several mammalian hosts; in sheep and cattle it causes abortions and in immunocompromised humans, it can cause systemic infections (Sprenger et al. 2012). However, Cfv is a pathogen restricted to the bovine genital tract and along with its biovar intermedius (Cfvi) is the causative agent of bovine genital campylobacteriosis (BGC), a bovine sexually transmitted disease that causes infertility, early embryonic death, and abortions. BGC produces significant economic losses for the beef cattle sector and is endemic in Spain (EFSA et al. 2017; Nadin-Davis et al. 2021; Pena-Fernández et al. 2021). Cft is genomically different and is mainly isolated from reptiles, although it has also been isolated from immunocompromised humans (Gilbert et al. 2016).

Whole-genome studies of C. fetus subspecies have described gene determinants involved in host adaptation and pathogenicity, such as the Type IV secretion system (T4SS) and the filamentation induced by cyclic AMP (FIC) proteins. These gene determinants are typical of Cfv, but can be transferred via plasmids or integrative and conjugative elements (Kienesberger et al. 2014; van der Graaf-van Bloois et al. 2016; Sprenger et al. 2017). The role of plasmids in host adaptation and pathogenicity in other zoonotic Campylobacter spp. (C. jejuni and C. coli) is very important, as these mobile genetic elements transfer antimicrobial resistance (AMR) genes, virulence factors (VF) and fitness determinants (van Vliet et al. 2021). Plasmid characterization and classification are vital to understand plasmid dissemination and their contribution to bacterial genetic diversity. Although several authors have described the presence of plasmids in C. fetus and their composition (Moolhuijzen et al. 2009; Nadin-Davis et al. 2021), they are mostly sequenced at the scaffold or contig level, thus hampering the development of a scheme to classify them. The main objective of this study was the identification and complete sequencing of the plasmids present in Spanish C. fetus isolates, as well as the detection of the replication genes, AMR genes, and VF they may carry. Additionally, a comparative study was conducted with publicly available fully sequenced plasmids of C. fetus isolates from different geographical locations, and a classification scheme was proposed.

Results and Discussion

Gene Content of Spanish C. fetus Plasmids and Comparative Analysis with Plasmids of the Campylobacterales Order

Twelve plasmids were identified and fully sequenced from six out of seven Spanish C. fetus genomes (Table 1). The only isolate not harboring plasmids was Cff004, taxonomically identified as Cff and isolated from feces. The six plasmid-harboring isolates carried one (n = 1), two (n = 4), or three plasmids (n = 1). Among these plasmids, 10 were characterized as potentially conjugative, while the remaining 2 were predicted as mobilizable but nonconjugative small plasmids (4,406 bp). Previous research suggested that Cfv and Cfvi tend to carry a higher number of plasmids than Cff, which supports the hypothesis of the poor ability of Cfv biovars to avoid the entry of exogenous DNA (Nadin-Davis et al. 2021). Nonetheless, the analysis of a larger number of C. fetus isolates may provide additional and more detailed information on this phenomenon. In our study, Spanish C. fetus plasmids also showed a highly variable size, ranging from a few thousand bp to more than 50 kb. Smaller plasmids had a limited number of genes (7 to 8), while larger plasmids had an average of 45 genes (Table 1).

Despite the considerable diversity, all the Spanish C. fetus plasmids clustered together with other C. fetus plasmids in the database. The 10 larger plasmids (>26 kb) of the Spanish isolates grouped with other similarly sized C. fetus plasmids; the 2 smaller nonconjugative C. fetus plasmids (4.4 kb) were most closely related to 2 C. fetus small plasmids and together with them formed a large cluster composed by other small plasmids (<6.5 kb) from 13 different species (Fig. 1). When the 19 publicly available C. fetus plasmids and the 12 plasmids from this study were compared, 2 main clusters were observed in the dendrogram based on the presence/absence of genes (Fig. 2a), i.e. 1 large cluster (divided into 3 subclusters) that included the larger plasmids, and a second cluster containing the smaller mobilizable but nonconjugative plasmids. This analysis also revealed that C. fetus plasmids clustered regardless their geographical location or lineage (Cff, Cfv, or Cfvi) (Fig. 2a), in contrast to the observed trend in C. fetus chromosomes, which generally clustered based on lineage and geographic location, as documented previously (Farace et al. 2021; Pena-Fernández et al. 2024).

Comparative Sequence Analysis of the Plasmids of C. fetus Isolates from Spain and Other Countries

BLAST analysis of the 12 plasmids from the Spanish C. fetus isolates revealed a high degree of homology between some of them (supplementary table S1A, Supplementary material online) and with several C. fetus plasmids from other countries retrieved from the database (supplementary table S1B, Supplementary Material online). Matches with plasmids from species other than C. fetus were only found between pCfv002_P3 and pCfvi011_P1 and 2 plasmids from Campylobacter vicugnae (Miller et al. 2017, 2024). Otherwise, the remaining C. fetus plasmids from Spain produced no matches with >70% coverage and ≥85% sequence identity with plasmids from species other than C. fetus. Plasmids that shared the highest homology according to the BLAST search also clustered together based on the presence/absence of genes (supplementary table S1, Supplementary Material online, Fig. 2a), and, in many cases, showed highly conserved gene arrangements as shown in the Mauve alignments (supplementary fig. S1, Supplementary Material online). Thus, within the identified subclusters in Cl 1 and Cl 2, the Spanish C. fetus plasmids were highly similar among themselves. However, 2 plasmids within Cl 1.2 had a unique composition and differed largely in size (pCfv002_P3, 26 kb and pCfv020_P1, 53 kb). Notably, they shared only part of their structure with the rest of plasmids in Cl 1.2 (supplementary fig. S1B, Supplementary Material online), and one of them (pCfv020_P1, size 53,911 bp) also shared half of its sequence (∼27.4 kb) with plasmids from Cl 1.1 (≥98% identity) (supplementary fig. S1A, Supplementary Material online). This suggests that plasmid pCfv020_P1 could be a hybrid (or recombinant plasmid) between the Spanish C. fetus plasmids from Cl 1.1 and Cl 1.2.

Most of the plasmids of Cl 1 (15/26) had multicopy insertion sequences (supplementary table S2, Supplementary Material online, supplementary fig. S1, Supplementary Material online). Specifically, seven plasmids from Spanish isolates carried 1 to 3 copies of the composite insertion sequence (IS) element IS200/IS605 while 9 of 19 public plasmids carried composite ISs of the family IS200/IS605 and occasionally also IS200/IS607. These ISs participate in gene rearrangements thus contributing to the genetic diversity of bacterial populations, as well as to the inactivation of genes due to the disruption of their sequences (He et al. 2015; Nadin-Davis et al. 2021). The effect of these rearrangements can be seen when comparing plasmids in Cl 1.3 (supplementary fig. S1C, Supplementary Material online); here, plasmid pCfv002_P1 carried 2 extra copies of IS200-IS605 flanking a block of 25.3 kb in opposite orientation in comparison with other similar plasmids (pCfv020_P2 and pCfv022_P1). The inversion of this gene set mediated by these transposases may contribute to the dynamic reorganization of the genetic material within the genomes.

Finally, the smaller plasmids pCfvi018_P2 and pCfvi027_P2 were identical (100% coverage and identity, supplementary fig. S1D, Supplementary Material online) and grouped within Cl 2 with other plasmids of small size and unknown function. Overall, these plasmids did not show any BLAST hits with coverage higher than 20% with plasmids from non-C. fetus species, suggesting that transfer of genetic material between the Spanish C. fetus and other Campylobacter species may not occur regularly like previously suggested by other authors (van der Graaf-van Bloois et al. 2023).

Proposed Scheme for C. fetus Plasmids Classification

Identification and classification of plasmids can help to trace the exchange of genetic material such as AMR genes or VF between populations. The genes coding for relaxase, replicase, and single-stranded DNA binding (SSB) protein play a significant role in the replication and plasmid transfer processes, facilitating the horizontal transmission of genetic material in bacteria (Al-Trad et al. 2023; Garcillán-Barcia et al. 2023). A high diversity in genes coding for relaxase and SSB was observed in the C. fetus plasmids (Fig. 2b), i.e. six relaxase gene variants, three of them (Relax-15, Relax-16, and Relax-17) only found in the Spanish plasmids; and five different ssb gene types. Conversely, the Spanish C. fetus plasmids harbored only three types of replicase genes. Overall, the hierarchical clustering based on the occurrence profile of the relaxase, replicase, and ssb genes (Fig. 2b) was similar to the plasmid clustering pattern observed when all plasmid genes were considered for comparison (Fig. 2a). These findings suggest that the relaxase, replicase, and the ssb genes could be suitable for C. fetus plasmids classification.

Analysis of Plasmid Encoded Virulence-associated Genes

All plasmids included in Cl 1 except plasmid mp1 (n = 25) carried genes encoding T4SS. This secretion system allows the transfer of genetic material between bacteria during conjugation and comprises at least 10 genes (VirB/VirD4) (Gorkiewicz et al. 2010). Notably, none of the C. fetus plasmids analyzed in this study had a complete T4SS, as none of them contained VirB7 and only 10 (5 from Spanish isolates) carried the remaining VirB/VirD4 genes (supplementary table S2, Supplementary Material online). Incomplete T4SS has been frequently reported in C. fetus chromosomes (van der Graaf-van Bloois et al. 2016; Nadin-Davis et al. 2021). However, the presence of plasmids harboring most of the VirB/VirD4 genes may help to compensate the lack of genes on the chromosomes. Notably, plasmids within Cl 1.1 did not carry VirD4 and VirB9. A previous study (Gorkiewicz et al. 2010) concluded that inactivation of VirB9 and VirD4 genes decreased bacterial killing and invasiveness. Therefore, the transfer of plasmids carrying these genes could contribute to the virulence properties of C. fetus strains.

Also involved in C. fetus virulence are the FIC proteins, whose function is related to the modification of host proteins, although it has also been suggested that some Cfv FIC proteins may act as antitoxins (Sprenger et al. 2017). van der Graaf-van Bloois et al. (2016) documented the presence of FIC-encoding genes on the chromosome of C. fetus subspecies but not on plasmids. Conversely, Nadin-Davis et al. (2021) reported their presence in both Cfv and Cfvi plasmids. In this study, the plasmids previously analyzed by Nadin-Davis et al. were reannotated with our in-house database. They were then analyzed together with the Spanish C. fetus plasmids, and genes encoding FIC proteins not reported by Nadin-Davis et al. were identified in plasmids of clusters Cl 1.2 and Cl 1.3 regardless their subspecies (supplementary table S2, Supplementary Material online). This additional finding was attributed to the utilization of different databases. Other genes encoding toxin–antitoxin (TA) products (yafQ, brnA/brnT), which play an important role in bacterial colonization and survival in the host as well as stabilization of plasmids (Sprenger et al. 2017), were found in most Cl 1 plasmids (except Cl 1.4) and in the 2 Cl 2 plasmids derived from Argentinian strains (supplementary table S2, Supplementary Material online).

Finally, despite the increasing trend toward the emergence of resistance in C. fetus strains (van der Graaf-van Bloois et al. 2023), no genes coding for AMR were detected in the plasmids of the Spanish C. fetus isolates or in the publicly available C. fetus plasmids. Likewise, no CRISPR-Cas genes were detected in the C. fetus plasmids.

Conclusions

Campylobacter fetus plasmids, unlike chromosomes, clustered independently of their geographic location of isolation or lineage. This underlines the need for a classification scheme to understand the genetic information flow mediated by plasmids in the population. Markers such as genes encoding relaxases, replicases, and the SSB in C. fetus plasmids have revealed patterns consistent with the clustering of plasmids according to their whole genetic composition and therefore represent promising targets for plasmid classification. Furthermore, most of the larger C. fetus plasmids harbor genes directly related to virulence and host adaptation. These findings underline the importance of understanding plasmid dynamics in C. fetus evolution and pathogenesis, which may have significant implications for public and animal health.

Materials and Methods

Selection of Spanish C. fetus Isolates and Culture Conditions

Seven Spanish C. fetus isolates from bull preputial wash (n = 6) and feces (n = 1) that had been previously sequenced using Illumina technology and taxonomically characterized in our laboratory as follows: 1 Cff, 3 Cfv, and 3 Cfvi (supplementary table S3, Supplementary Material online) (Pena-Fernández et al. 2024) were selected for long-read whole-genome sequencing (WGS) using Oxford Nanopore technologies (ONT, Oxford, UK). Isolates were subcultured on 5% sheep blood-enriched Columbia agar (COS, Biomerieux, Marcy-l'Étoile, France) and incubated at 37 °C under microaerobic conditions (5% O₂, 10% CO₂, and 85% N₂, GENbox Microaer, Biomerieux, Marcy-l'Étoile, France) for 48 h.

Whole-genome Sequencing, Hybrid Assembly, Plasmids Identification and in Silico Characterization

Genomic DNA extraction was performed from single colony pure cultures using the NZY Microbial gDNA Isolation kit (NZYtech, Lisbon, Portugal). Libraries were prepared using the Rapid barcoding Kit (SQK-RBK004) and sequenced in R9.4.1 flow cells (FLO-MIN106) using the MinION Mk1C (ONT) device. ONT raw reads were subjected to base-calling using Guppy v6.2.11 in HAC mode. Adapter removal was then performed using Porechop v0.2.4 (Wick et al. 2017), reads shorter than 1,000 bp were discarded using Filtlong v0.2.0, and only the best 1,000 Mbp reads were kept for subsequent analysis.

The long-read sequences produced by ONT in this study and the short-read sequences previously generated by Illumina (Pena-Fernández et al. 2024) were used for a hybrid assembly using Unicycler v0.4.8 (Wick et al. 2017) with default parameters to overcome possible gaps during the assembly of the short-read sequences and obtain completely closed chromosomes and plasmids. Assembled genomes were analyzed with RFPlasmid v0.0.18 (van der Graaf-van Bloois et al. 2021) where plasmid sequences were selected with a P-vote ≥ 0.6, and circularization was assessed with Bandage v0.9.0 (Wick et al. 2015).

Plasmids were screened for AMR genes and VF using BLASTn and ABRicate v.0.8.10 (T. Seemann, https://github.com/tseemann/abricate) against ResFinder (Florensa et al. 2022) and VFDB (accessed on May 11, 2023) (Liu et al. 2022) and hits were filtered at 90% coverage and 90% identity. With Plascad (Che et al. 2021) predictions were made for plasmids categorization into 3 groups (potentially conjugative, mobilizable nonconjugative, and nonmobilizable) based on the presence of specific proteins involved in DNA transfer, such as relaxase, T4CP, and T4SSs. A plasmid harboring relaxase, T4CP, and T4SSs is predicted as potentially conjugative, but its conjugative capacity needs to be experimentally demonstrated. Those containing only relaxase are termed mobilizable nonconjugative. Plasmids lacking all these genes are classified as nonmobilizable.

Plasmids Genome Analysis and Comparative Genomics

To ensure a comprehensive comparative analysis of the plasmids at the host species level, only complete sequenced plasmids available in the databases were included in this study. Thus, all plasmids categorized as order Campylobacterales in Plasmid Atlas (pAtlas, v1.6.1) (Jesus et al. 2019) (accessed on May 10, 2023) were retrieved. Excluding 2 plasmids (NC_008790.1, NZ_CP020479.1) that were fully sequenced but had linear topology, the remaining were fully sequenced and closed. The dataset was completed with 10 other completely sequenced C. fetus plasmids available in GenBank (accessed on May 10, 2023), resulting in a total of 181 plasmids of 4 genera: 9 Arcobacter, 116 Campylobacter, 52 Helicobacter, and 4 Sulfuricurvum (see supplementary table S3, Supplementary Material online). These included 19 C. fetus plasmids corresponding to 13 genomes from bovine origin and different geographical locations (Table 1).

Gene prediction and annotation of the C. fetus plasmids was performed with Prokka v1.11 (Seemann 2014) with default parameters using an in-house curated database of C. fetus plasmids. Gene composition of each plasmid was assessed using Roary v3.13.0 (Page et al. 2015) with a minimum percentage of identity of 80. For the comparative analysis of plasmids, a dendrogram based on the gene presence/absence output of Roary was constructed with RaxML v8.2.4 (Stamatakis 2014). The resulting tree was visualized and annotated using the website iTOL v6 (Letunic and Bork 2007).

To assess the homology among the Spanish C. fetus plasmids, a custom database containing their sequences was built and the homology percentage among them was calculated by performing BLASTn analysis using Geneious Prime 2003.0.3. Only hits with a query coverage ≥50% and identity ≥90% were considered. Furthermore, to evaluate the percentage of homology between C. fetus Spanish plasmids and other publicly available plasmids, a BLASTn analysis was performed using the online NCBI BLAST tool (Altschul et al. 1990), considering only plasmids with query coverage ≥70% and identity ≥85%. Highly similar plasmids (>90% query coverage and >95% identity) were aligned and plotted using Mauve via Geneious Prime 2003.0.3.

A classification scheme based on relaxase, replicase, and ssb genes was developed to characterize the C. fetus plasmids. First, the C. fetus plasmids were screened for these genes against an in-house developed Campylobacter plasmid database (Utrecht University, manuscript in preparation) using BLASTn and ABRicate v.0.8. A hierarchical clustering based on the distribution of relaxase, replicase, and ssb genes was built using the unweighted pair-group method with arithmetic mean based on the Jaccard distance matrix. The presence of genes belonging to the T4SS (VirB/VirD4) and FIC, as well as mobile genetic elements (transposases) and genes related to the TA system, were retrieved from Prokka annotation data.

Supplementary Material

Supplementary material is available at Genome Biology and Evolution online.

Acknowledgments

The authors express their gratitude to Montserrat Agüero and Iratxe Pérez, from the Laboratorio Central de Veterinaria de Algete, for generously providing the C. fetus bacterial strains, and to Beatriz Oporto, from NEIKER, for her excellent technical assistance in the long-read Oxford Nanopore sequencing.

Author Contributions

A.H., G.A., L.v.G., B.D., A.Z., and J.A.W. conceived the study and participated in its design. N.P.F. performed laboratory analysis. N.P.F., L.v.G., M.O., and J.L.L. and participated in the bioinformatics analysis. N.P.F. wrote the manuscript, with interpretation of results and discussion inputs from all authors. All authors read and approved the final manuscript.

Funding

This study was supported by the research projects RTA2017-00076-00-00 funded by MCIN/AEI/10.13039/501100011033 and FEDER “A way to do Europe,” as well as the Department of Economic Development, Sustainability, and Environment of the Basque Government (Project URAGAN 21-00012). N.P-F. was the recipient of the grant for doctoral formation Pre2018-086113 funded by MCIN/AEI/10.13039/501100011033 and by “ESF Investing in your future.”

Data Availability

The 12 Spanish C. fetus plasmids sequenced in this study are available at GenBank under the accession numbers described in Table 1, associated with the BioProject PRJNA1019261.

Ethics Declarations

Sample collection was carried out by veterinary practitioners strictly following Spanish ethical guidelines and animal welfare regulations (Real Decreto 53/2013). The collection of this material, being considered a routine veterinary practice, did not require the approval of the Ethics Committee for Animal Experimentation. Informed oral consent was obtained from the farm owners at the time of sample collection. All methods were performed in accordance with the relevant guidelines and regulations and complied with ARRIVE guidelines (Percie du Sert et al. 2020).

Literature Cited

Author notes