Exploring Viral Interactions in Clavibacter Species: In Silico Analysis of Prophage Prevalence and Antiviral Defenses

1. Introduction

Phytopathogenic bacteria are one of the main contributors of low agricultural productivity worldwide because they cause devastating diseases in many important crops [1]. Clavibacter is the most economically important gram-positive bacterial plant pathogen, with species that cause severe diseases of maize (C. nebraskensis), wheat (C. tessellarius), tomato (C. michiganensis), potato (C. sepedonicus), bean (C. phaseoli), and pepper (C. capsici) crops [2]. Because they cause significant crop losses, some species are considered quarantine pathogens in different countries [3,4]. Despite considerable efforts to mitigate the impact of Clavibacter-associated diseases through traditional breeding strategies and chemical control measures, their management remains challenging, partly due to the need for a comprehensive understanding of the genetic determinants underlying their pathogenicity and environmental interactions [5,6,7].

Prophages and antiviral defenses represent two interconnected components of bacterial genomes that have received increasing attention in recent years due to their roles in shaping bacterial evolution, pathogenicity and ecological interactions [8,9]. Prophages are viral genomes integrated within bacterial chromosomes or plasmids, which can either enhance or decrease the fitness of the bacteria, impacting their behavior, genetic variability and ecology [10]. Some prophages provide genes encoding toxins that enhance the virulence of pathogenic bacteria, as occurs in Acinetobacter baumannii, Staphylococcus aureus, and Escherichia coli [11,12,13]. Moreover, Bifidobacterium, Lactobacillus, and Streptococcus have been shown to harbor numerous prophages and prophage-like components [14,15,16]. Even though significant research has been conducted on prophages in relation to human bacteria, their significance in the fitness and development of phytopathogenic bacteria is comparable. For example, prophages of some of the most destructive plant pathogens, such as Dickeya, Ralstonia, Xylella, Xanthomonas, Burkholderia, and Pectobacterium, have been implicated in auxiliary genes that encode proteins that inhibit the plant immune response, degradation enzymes and proteins involved in the secretion systems [17,18,19,20]. To make bacterial plant pathogens more competitive, prophages can encode bacteriocins that stop competitors [21], increase the resistance of bacteria to environmental stressors such as metal ions and antibacterial compounds [19,22], or boost metabolic potential to help plants survive when nutrients are limited [23].

On the other hand, bacteria have evolved a plethora of antiviral strategies to resist the invasion of viral and mobile genetic elements (MGEs), thereby influencing their susceptibility to phage predation and horizontal gene transfer [24]. These phage resistance strategies involve diverse mechanisms such as the modification of the surface receptors to prevent phage attachment [25], the action of superinfection exclusion proteins to block phage DNA injection [26], the production of enzymes that degrade phage DNA or modify the bacterial genome [27], the use of cell suicide systems to limit phage spread [28], the DNA interference by the Argonaute system to silence phage DNA [29,30], and the expression of elaborated adaptive immunity mechanisms (e.g. CRISPR-Cas), also called defense systems [31]. In the latter, numerous defense systems with great antiphage power and extensive dispersion in bacteria and archaea have been discovered, but most of their molecular mechanisms are unknown [32,33]. Many of the antiviral systems can be carried by MGEs, such as prophages. In P. aeruginosa, the ϕ297 prophage modifies the structure of O-antigen subunits in lipopolysaccharides, conferring phage resistance[34]. Mycobacteriophages such as Butters, Sbash, and CarolAnn have putative toxin-antitoxin (TA) modules with membrane-associated effectors [35,36]. Some phages encode the stress alarmone guanosine pentaphosphate ((p)ppGpp) for host defense, while others encode Sie membrane proteins or restriction endonucleases [37,38]. The prophage-mediated antiviral systems may play an important role in limiting phage infection and increasing host competence and survival [35].

Despite the well-documented roles of antiviral defenses and prophages in various bacterial species, the distribution, diversity and potential functions in Clavibacter species remain poorly understood. Given the importance of Clavibacter-associated diseases in agriculture and the increasing prevalence of phage-mediated control strategies for bacterial pathogens, the knowledge gap surrounding Clavibacter phages is striking. To date, there have been only three completely sequenced genomes of Clavibacter phages: CMP1 and phage 33, isolated against C. michiganensis, and CN1A phage isolated from C. sepedonicus; and the partially sequenced specific CN77 phage for C. michiganensis [39,40,41]. This emphasizes the necessity to address the impact of the viral elements in Clavibacter biology. In this study, we used bioinformatics and comparative genomics to identify and characterize prophage sequences within publicly available Clavibacter genomes, with the aim to shed light on their genomic features and functional potential. Additionally, we will assess the presence and diversity of antiviral defense mechanisms. Exploring the interplay between Clavibacter and their viral elements will provide valuable insights to understand the impact on pathogenicity, ecology and evolution. Understanding how bacteria defend against viral infections is essential for creating specific strategies to reduce Clavibacter-related diseases and loss to agriculture.

2. Materials and Methods

2.1. Bacterial Genomes and Prophage Identification

A total of 114 genome sequences of the Clavibacter genus were retrieved from GenBank database (https://www.ncbi.nlm.nih.gov/genbank/) (last accessed July 2022), and listed in Table S1. Prophage identification was performed using three tools: the PHASTER (PHAge Search Tool Enhanced Release) [42], the PROPHAGE HUNTER [43], and the command line software VirSorter2 [44]. All potential prophages were manually analyzed to depurate the raw list of prophages. For this, overlapping prophages identified on the same contig were thoroughly analyzed by sequence alignment ClustalW, and they were counted as one, adjusting the prophage length considering the non-overlapping ends. Further analysis depicting prophages type, G+C content, polylysogeny, location, and prophage length was plotted in RStudio (v.4.4.1) using ggplot2 (v.3.5.1). Comparative analysis of the host and the prophage genome sizes were analyzed by correlation with the Kendall method considering significant with a p-value less than or equal to 0.05.

2.2. Annotation of Prophage Genes

Final prophage genomes were annotated using the open-access program RAST: Rapid Annotation using Subsystem Technology [45]. Prophage genomes shorter than 4,000 bp that were not accepted for RAST were annotated with GeneMarkS [46]. Then, prophage genes were evaluated to determine their similarity with bacteriophage genes based on searches with phage-limited BLASTx in the NCBI database. The analysis was restricted to tax IDs relevant to bacteriophages, namely 28883, 10699, 10474, 12333, 38018, 2100421, 186765, 10662, 10744, 79205, 10841, and 102294, using the default settings. Proteins identified as hypothetical in RAST were analyzed with BLASTx and PFAM. For searching integrases on prophage genomes, a database was constructed with phage integrase protein sequences retrieved from UniProt. Subsequently, a BLASTp analysis was performed to identify integrases in the Clavibacter prophages using the integrase database previously constructed. Raw results were filtered based on the following criteria: an E-value threshold of less than 0.0001, a cut-off of 50% identity, and 50% coverage and the presence of integrase-related domains with PFAM, in order to increase the probability of finding divergent integrases.

2.3. Functional Categorization of CDSs

Genes were classified by functional categories using the eggNOG-mapper version 2 [47], to identify the clusters of orthologous genes (COGs) of the CDS of the prophage genes according to eggNOG 5 database (v5). Then, a KEGG pathway enrichment analysis was performed with the advanced clusterProfiler (v4.12.6) package in R. The p-values were adjusted with the Benjamini-Hochberg method, and a p-adjusted ≤ 0.05 was set to identify the significantly enriched pathways. Subsequently, the enhanced terms were displayed through dot plots to show the metabolic pathways significantly enriched in the data.

2.4. Identification of Virulence Factors and Antibiotic Resistance Genes

To identify virulence genes, searches were performed using VirulenceFinder 2.0 [48] on all prophage genomic sequences. An additional survey of antibiotic resistance genes was carried out using the Comprehensive Antibiotic Resistance Database (CARD) with the Resistance Gene Identifier (RGI) option with the default parameters [49]. Detection of Insertion Sequences (IS) in Clavibacter genomes was carried out with MobileElementFinder v1.1.2 using the default parameters [50].

2.5. Comparative Genomic Analyses and Defense System Detection

ANIb (Average Nucleotide Identity using BLAST) was calculated using Pyani (v0.2.12) [51] with default parameters to assess the sequence-level similarity between all prophages. The ANI similarity matrix was used to construct a heatmap using the Pheatmap function (v1.0.12) in R. Hierarchical clustering of rows and columns was performed using the ward.D2/average method, while the distance between rows and columns was calculated using the Euclidean distance. All prophage cluster species levels were aligned, and amino acid synteny was visualized using Clinker [52]. The search for bacterial and phage defense systems was done using Defense Finder (https://defensefinder.mdmlab.fr/) [53].

3. Results

3.1. Identification and Prevalence of Prophages in Clavibacter Strains

To study the prevalence and diversity of prophages, genome sequences of Clavibacter associated with tomato (C. michiganensis n = 71), maize (C. nebraskensis n = 13), lucerne (C. insidious n = 7), and other crops (n = 24) were retrieved from NCBI database and analyzed with three different programs. In the first instance, we identified the presence of prophages in more than 97% (n=111 of 114) of Clavibacter genomes, identifying 123 incomplete prophages with PHASTER, 34 active and 124 ambiguous prophages with Prophage Hunter, and finally, 94 full and 15 partial prophages with Virsorter2, accounting a total of 390 prophages (Figure 1A, Table S2). We performed a manual curation because we observed that several prophages overlapped within their genomic regions and we decided to join all of them in one larger prophage. Manual curation merged several prophages into 21 prophages (highlighted in blue in Table S2), reducing a total of 353 prophages identified in Clavibacter genomes (Figure 1B, Table S3), an observation consistent with previous studies showing that prophages are prevalent in bacteria [54]. We noticed that only 13 out of 353 prophages were detected with two or three programs, strongly supporting their identification.

Further characterization revealed that most Clavibacter strains harbored one prophage (32 out of 114, 28%), followed by two (24 out 114, 21%) and three (21 out 114, 18.4%) prophages, while a higher number of prophages by strain were found less frequently (Figure 2A). In fact, we observed that 17 prophages were detected in CFBP 6488 strain. Because of the diversity of prophage categories, we decided to merge prophages in one category as Complete/Active/Full prophages, resulting in 126 active prophages, and as Incomplete/Ambiguous/Partial prophages in the other grouping 227 cryptic prophages. Then, analysis showed that 281 prophages were located into the chromosome and 72 in plasmids (Figure 2B), where most of the prophages were identified in contigs, indicating that draft genomes are useful for prophage detection. Most of the identified prophages were found in C. michiganensis, as expected due to a higher number of genomes analyzed, and they showed a mean proportion of two phages by genome. In contrast, other Clavibacter species, such as C. californiensis, C. chilensis, and C. insidious, exhibited a higher proportion of 7, 6, and 6 prophages per genome, respectively, although the analyzed genomes were less (Figure 2C-D). Despite many prophages were predicted as active elements (category Complete/Active/Full), further annotation revealed the presence of a few integrases, strongly suggesting that these prophages are cryptic or remanent of domesticated prophages (see below).

Next, the average genome size of prophages was 13.98 kbp, ranging from 1.38 to 123.96 kbp and exhibited an unimodal genome size distribution highly enriched with prophages up to 10 kbp in genome length (Figure 3A-B). This finding is contrary to Bobay et al. [55] in which a bimodal distribution of prophage genome length was observed, suggesting that prophages with larger genomes belong to recently integrated and inactivated prophages, indicating a rapid step of inactivation, while shorter prophages suggest genome degradation due to the phage domestication process. In this context, the prophage genomes frequently accumulate mutations, and those prophage genomic regions that confer beneficial traits to the host are selected at long-term, a mechanism known as phage domestication [56]. Moreover, most of the prophages had an average G+C content of 72.16%, similar to analyzed Clavibacter genomes with 72.56% average GC content, although 145 (41%) prophages showed G+C contents lower than 70% (Figure 3C). Similar G+C content may suggest that most prophage-like elements are the product of long-term interactions with their Clavibacter hosts [57]. Linear regression analysis of host versus prophage genome sizes was very weak (R²=0.065) although was significant (p < 0.001), showing that most prophages were accumulated in genomes between 3.2 and 3.4 Mb in size (Figure 3D). Taken together, these results indicated that Clavibacter genomes exhibit a high prevalence of cryptic prophages, many of them representing domesticated prophages.

All 124 active prophages were analyzed in public databases using BLAST limited to tax ID viruses, showing than only 26 prophages matched to known phage genomes with at least 74.6% identity, although very low coverages were observed. Matched prophages were related to Erwinia phage vB_EamM_Stratton, Gordonia phage BetterKatz, Mycobacterium phage ShrimpFriedEgg, Mycobacterium phage Kykar, and Gordonia phage MagicMan. Because of low coverage, we considered that all active prophages are novel.

3.2. Comparative Genomic Analyses of Clavibacter Prophages

To analyze the genetic diversity of prophages, we created an ANI similarity matrix and constructed a heatmap. This analysis revealed that most of the prophages were grouped into 7 clusters, which shared at least of 90% identity, except for groups B and C, where some pairwise comparisons gave ANI values of at least 80% but were clustered together. The remaining prophage genomes showed identity with a few genomes or remained as single prophages, suggesting they were unique. These results indicated high genetic diversity and revealed novel Clavibacter prophages with unique genomes.

Figure 4. Clustering and heatmap of prophages regions in Clavibacter species. Similar prophages were classified in A to G groups. .

Figure 4. Clustering and heatmap of prophages regions in Clavibacter species. Similar prophages were classified in A to G groups. .

Detailed analysis revealed that most groups of prophages contained at least from two different Clavibacter species, with the exception of cluster D, which was composed of only C. michiganensis prophages. Analysis of C. michiganensis prophages from group A showed low differences in the genomic organization, including some inversions and rearrangements (Figure 5A). A similar pattern was observed for C. michiganensis prophages from cluster D (Figure 5D), showing that the gene organization was conserved with some putative proteins less conserved, reflecting a lower genetic variability among prophages within the group. Strikingly, most of prophages were classified independently of the Clavibacter species, which may suggest these prophages could be acquired by the last ancestor before speciation. In the case of cluster D, only C. michiganensis prophages were grouped together, suggesting a species-specific group of prophages; however, we cannot rule out a potential bias caused by a reduced number of analyzed genomes representing other species than C. michiganensis in our study.

3.3. Functional Annotation of Clavibacter Prophages

In order to gain insights into the nature and function of genes, a total of 6,937 CDS (protein-coding genes) were identified in all the prophage genomes, and 3,858 CDS were filtered after eggNOG analysis representing 55.6% of total CDS. From these, 3,422 CDS were grouped in at least one of the 21 COG functional categories, while 436 out of 3,858 CDS were unclassified, representing 11.3% (Figure 6A, Table S4). Interestingly, after removing redundant descriptions of CDS with COG, we found 594 unique CDS descriptions, indicating that some CDS corresponded to orthologous genes shared among the different prophages. In total, 2,701 CDS were associated with COGs with a defined function, in which the most representative categories were the M group (Cell wall/membrane/envelope biogenesis) alone or in combination with other groups accounting for 26.8% (1,036 out of 3,858 CDS), followed by L group (replication, recombination, and repair) alone or in combination accounting 8.2% (316 out 3,858 CDS). Interestingly, 1,157 CDS (29.9%) were hypothetical proteins (including the S group and the unclassified CDS), which summed up to 3,079 CDS not considered by eggNOG, represented 61% of total proteins with unknown function (4,236 out 6,937 CDS). A complete annotation of prophages was done by combining the results of RAST, BLASTx, CARD, PFAM, and VFDB (Table S5).

Protein-coding genes belonging to the M group and combined categories include several proteins such as glycosyl transferases, cell wall hydrolases, and other enzymes involved in sugar metabolism. Other proteins harbor domains related to dTDP-4-dehydrorhamnose 3,5-epimerase and dTDP-glucose 4,6-dehydratase from the rhamnose pathway, likely playing an important role in the bacterial cell wall structure[58]. In addition, a total of 348 CDS (9% of CDS grouped with COG) are involved in carbohydrate metabolism in G, GK, GM, CG, EG, and EGP categories, which included several ABC-2 type transporters, glycosyl hydrolases, pectate lyases, and cellulases. Interestingly, we found the celA gene, a known virulence factor encoded by the plasmid pCM1 of C. michiganensis [7], was enriched in the G group. Further analysis showed celA was annotated as endoglucanase E1 precursors by RAST and it was present in different prophages, including the phage-plasmids LMG_3663_P4, CFIA_CsR14_P2, 1217_P3, ATCC_10253_P1_3_4, NCPPB382_P1, R1_1_P1_P4, R1_3_P1 and LMG_3663_P5_P1, and the prophages CFBP_7577_P1 and 1106_P1. Other genes related to sugar metabolism were identified, such as UDP-glucose 4-epimerase and glucose-1-phosphate thymidylyltransferase that participate in the production of precursors for the biosynthesis of complex sugars [59,60]. Several CAZymes, mainly of the glycosyltransferase type, were identified in different prophages, such as CASJ001_P1 and CIBA_P1. Furthermore, some Blastx results showed homology with hydrolase proteins from bacteriophages such as Erwinia phage Fifi, suggesting a possible function in the degradation of plant cell wall components [61].

Since the diversity of metabolic functions encoded by prophage genes, we performed a KEGG pathway enrichment analysis. This analysis revealed that metabolic pathways related to carbon, starch, sucrose, and galactose metabolism were enriched in prophage-encoded genes, consistent with findings observed with COG classification. Moreover, biosynthetic pathways involving cofactors, amino acids, nucleotide sugars, and streptomycin were significantly enriched. Interestingly, the streptomycin biosynthesis pathway has only been reported in Streptomyces griseus [62], which belongs to Actinomycetes like Clavibacter. We extracted the genes associated with streptomycin biosynthesis. We found four genes involved in the first metabolic steps of streptomycin biosynthesis (Figure S1), mainly as a unit in the prophage genomes. BLASTn analysis revealed these genes showed high coverages (99%) and at least 70% identity, which suggest these genes might be orthologous acquired by a phage-mediated gene transfer.

On the other hand, prophage genomes were enriched in genes that may confer advantages to adapt to diverse, challenging environments. Several genes related to copper homeostasis and resistance were identified in prophages, such as CopZ, CopG, CopD, and CopC (S group), which are involved in copper translocation. In addition, genes involved in transport and resistance to other metals were identified, such as lead, cadmium, zinc, mercury and arsenic, including transporting P-type ATPases, chaperones, and arsenate-mycothiol transferases, which could confer an advantage to the bacteria in contaminated environments with heavy metals. Other genes encoding ABC (ATP-binding cassette) family transporters related to polysaccharide, cobalt, and Fe³⁺ transport were identified. For antibiotic resistance, we found genes encoding chloramphenicol acetyltransferases in two prophages (LMG26808_P1_3 and CASJ001_P7) using CARD analysis with perfect cut-off, which was also detected in the H group by COG analysis. Chloramphenicol acetyltransferases are known to inactivate chloramphenicol, thiamphenicol, and azidamfenicol via acetylation, being the most prevalent route for acquiring chloramphenicol resistance in bacteria [63]. Moreover, a putative helix_turn_helix multiple antibiotic resistance protein was enriched in the K group, which harbors a MarR domain (PF01047) that is characteristic of Mar proteins, suggesting a possible role in multiple antibiotic resistance [64]. Other related genes were the metallo-beta-lactamase superfamily proteins, which are often involved in the hydrolysis of β-lactam antibiotics.

Regarding virulence factors, analysis with VFDB revealed several hits associated with functions such as immune modulation, nutritional/ metabolic factors, adherence, biofilm, exotoxin, exoenzyme, accessory secretion factor, effector delivery system, stress survival, motility, invasion and regulation. The most abundant categories were immune modulation, including glycosyltransferases, ABC transporters, capsular proteins, carbohydrate metabolism enzymes, acetyltransferases, antiporters and some kinases; followed by nutritional/metabolic factors, the enriched genes were associated mainly with siderophores biosynthesis and transport, transporter ATPases, sugar transporters and metal iron/manganese transporters. Then, for the effector delivery system were identified several proteins associated to type VI secretion system, ATPases, transcriptional regulators, chaperones and hydrolases. For adherence, Clavibacter prophages were found to encode several proteins identified as C5a peptidases, fimbria and pili biosynthesis and assembly, chaperones, metallopeptidases and adhesins (Table S6).

Based on COG, we found only 16 integrases (L group), 6 structural proteins, and 5 terminases, both belonging to the S group. Combined annotation with RAST, BLASTx (limited to viruses-related tax ids) and PFAM allowed the identification of a total of 109 genes encoding phage-related proteins, such as integrases, major and minor capsid proteins, neck, head closure proteins, fibers, tail proteins and head-to-tail adaptor proteins (Figure 7), which are associated to prophage integration, DNA translocation and virion structure [65,66,67]. Despite their identification, phage-related proteins are few in comparison with all CDS for predicted prophages, thus it is possible that they could be lost under the domestication events.

In addition, we found 37 transposasas grouped in the L group. Previous studies have shown that transposases are highly enriched in prophages with shorter genome lengths, and the mathematical modeling including transposition events, revealed that transposable elements could be relevant primarily during the initial event of domestication by inducing mutagenesis and prophage genome degradation; however, in shorter prophages representing advanced domestication events, transposable elements are lost and only beneficial genes are selected [68]. Thereby, we performed the identification of insertion elements (IS), a kind of transposable element, finding a total of 706 IS in 58 out 114 Clavibacter strains (50.87%), being 107 different IS. From these, the most abundant were IS1121, IS1122, ISCmi2 and ISPfr17 (Table S7). However, after a detailed examination, we found that only 29 prophages contained IS. The presence of IS in the prophage could interfere with the activation of viral genes, influence the ability of these viruses to replicate and spreading, and facilitate horizontal gene transfer [69]. IS may be important in initial events of inactivation and domestication, but the IS are generally lost in advanced genomic decay, similar to the observed pattern in Clavibacter prophages.

Taking together, we hypothesized that the underrepresentation of integrases, transposases, and terminases may indicate an advanced stage of domestication. Some studies have shown that integrases are relatively maintained through evolution, maybe mediating horizontal gene transfer of selfish elements, such as satellite phages, or in the process of recombination [68], meanwhile transposases are present in the first step of prophage inactivation but absent in advanced steps of domestication. Despite Prophage Hunter and Virsorter2 algorithms detected active and complete prophages, our analysis may suggest that most Clavibacter prophages probably are cryptic and domesticated prophages with higher levels of genomic decay.

3.5. Diversity of Defense Systems

Considering that identified prophages seem to be domesticated keeping at long-term interaction with the host, we supposed that Clavibacter probably is infected by prophages at a low rate. This reasoning made us speculate that the diversity of the immunity defense system is poor in this bacteria, considering that phage does not represent selective pressure. To explore this hypothesis, we performed a search of defense systems harbored by Clavibacter genomes, and strikingly, we found that 111 out of 114 genomes contained at least one defense system. Moreover, these strains exhibited 35 types of immunity systems (Figure 8, Table S8). Interestingly, none of the analyzed genomes harbored the CRISPR-Cas system, indicating this system is not found in Clavibacter. Meanwhile, the PD-Lambda-1 was present in 95 out 114 of Clavibacter genomes, followed by Abi (AbiE) in 48 genomes and restriction-modification systems (RM), MMB_gp29_gp30, PD-T7-2 in 131, 34, and 30 genomes, respectively. In bacterial restriction-modification, the restriction endonuclease is crucial in giving tolerance against foreign DNA [70]. The MMB gp29-gp30 system is a defense mechanism identified in the MichelleMyBell (MMB) temperate mycobacteriophage, which consists of a toxic protein (gp29) and a putative membrane protein (gp30). The mechanism is unknown, but it has been determined that this protein complex helps bacterial defense against lytic phages [71].

Since phage predation is a potential threat, prophages may encode anti-phage defense genes to protect its host. Thereby, we searched defense-associated genes in prophages, and we found that only 12 prophages were predicted for the AbiU, RM_Type_IIG, MMB_gp29_gp30, AbiE, CapRel, PD-Lambda-1, Hachiman and Gabija systems within their genomes. Other defense systems were less than 9,000 bp upstream or downstream of four prophage genomes (Table S8). Because Hachiman and Gabija have not been reported in prophages, we performed a survey in GenBank in viral genomes, and we detected the presence of only gajAB genes in phage PfaC02b from Pseudomonas. Genomic surveys from other defense systems revealed the presence of darT in Pseudomonas phage Zuri, kwaAB in phage PfAC02a, and the Thoeris anti-defense 1 protein (ths-1) from Bacillus phage Jack Rabbit (data not shown). Detailed analysis with PFAM, RAST and BLAST revealed that Clavibacter prophages harbored genes encoding proteins related to toxin-antitoxin (TA) systems, such as the HicA toxin and the HicB-like antitoxin from the hicAB system; the yoeB-like toxin of the YoeB/YefM systems; and a complete VapBC TA. Toxin-antitoxin was initially found as selfish genetic elements on plasmids but has been identified in bacterial chromosomes and mobile genetic elements as phages [72,73]. These results lead us to speculate that components of several defense systems, including those described only in bacteria might be encoded by prophages.

4. Discussion

Despite the Clavibacter species are devastating and quarantine pathogens in several important economic crops, very few studies have been conducted to explore the contribution of viral elements in their pathogenesis and evolution. To date, several studies in other relevant phytopathogens, such as Ralstonia and Erwinia, have demonstrated the presence of numerous active and inactive prophages, which contribute to genetic variability and confer diverse properties affecting bacterial adaptation to new environments and competition within microbial communities [74,75]. This study provides novel genomic insights about the diversity of prophage-like elements and defense systems across different Clavibacter species genomes. To improve the detection of prophages, we employed three softwares, each using unique criteria and methods to identify prophages. VirSorter2 uses a viral sequence identification system by examining signature genes, gene density and other genomic features, accurately distinguishing viruses [44]. On the other hand, PHASTER employs searches based on sequence homology by comparison with known phage databases, having limitations in identifying unknown phages or atypical sequences [42]. Lastly, prophage Hunter combines sequence similarity and more than 24 genetic features, such as transcriptional orientation, amino acid composition and Watson-Crick ratio, among others, using a scoring system to detect prophages, although it relies on reference data and it is less effective on highly divergent sequences [43]. Despite this, Prophage Hunter allowed the identification of most prophages, at least in part because it uses machine learning models and detailed comparisons with viral genes, improving the detection of latent or atypical prophages than VirSorter and PHASTER, which may overlook due to bias in the information relying on databases. Nonetheless, the results obtained from this analysis clearly showed that prophage-like elements are pervasive in Clavibacter genomes and encode a repertoire of proteins that may enable survival and fitness.

Detailed analysis of most Clavibacter prophages showed them as inactive or cryptic elements, although some prophages were classified as active or full depending on the detection software. Nevertheless, we speculated that all prophages possibly are inactive based on several criteria: only a few encoded integrases, terminases, transposases; few prophage-associated IS were identified, and genome length exhibited an unimodal distribution enriched in short length genomes. Firstly, the low presence of integrases and terminases may indicate an advanced state of domestication, where prophages are no longer active in the production of complete viral particles and depend on alternative mechanisms for their persistence within the bacterial genome. Moreover, previous studies suggested that in degraded prophages, transposable elements, such as transposases and IS, may play an initial role in mutagenesis and genomic degradation during early domestication events but become less frequent as domestication progresses and only genes of selective benefit to the host are maintained [55]. Despite we found more than 100 IS in half of the analyzed Clavibacter, the fact that only 29 prophages contained IS activity on these elements, may imply they were lost in advanced domestication steps, which would be consistent with an evolutionary trend towards the degradation or domestication of these mobile elements in bacterial genomes [68].

Secondly, the unimodal genome length distribution indicates most prophage-like elements have undergone a process of domestication, where the integrated prophages experienced inactivating mutations, genome reorganization and gene loss due to selective pressures at long-term [55]. Finally, the diverse GC content distribution on prophage-like sequences suggested prophage groups are genetically diverse and that those with similar G+C content might be the product of long-term interactions with their Clavibacter hosts [57], supporting our hypothesis that most prophages are under advanced domestication. Unexpectedly, prophages searches for identification failed using BLAST, which may indicate that that Clavibacter prophages are novel, although bias in the identification process might be due to the scarce genomic information regarding Clavibacter phages in databases.

There was a notable variation in the presence and proportion of prophages among different Clavibacter species. Strikingly, the presence of several orthologus prophages across the different Clavibacter strains may suggest that can be derived from a single prophage present in the last ancestor, indicating vertical transmission [55]. A higher number of prophages in specific genomes could be related to species-specific characteristics, such as their ability to acquire exogenous genetic material through prophages, which could influence their evolution, pathogenicity or adaptation [12,75,76]. Moreover, the presence of prophages in plasmids was identified possibly as representing phage-plasmids, which may offer adaptive advantages since their functions as plasmids and phages represent more pathways for the spreading of genes, improving host fitness [77]. Genomic analysis on phage-plasmids showed that their genes overlapped with plasmids and phages, indicating that phage-plasmids mediate gene exchange among prophages and plasmids, including genes associated with core functions, defense systems and antibiotic resistance [78]. Other studies highlight the interaction and impact of phage-plasmids in phytopathogens, such as the Ralstonia solanacearum species complex [57]. Thereby, prophages on plasmids contribute to host bacteria’s genetic diversity and functionality.

The presence of prophage-like elements in Clavibacter strains indicate that they play important roles in bacterial survival, adaptation and fitness, which is consistent with the fact that prophages harbored a repertoire of genes associated with relevant functions, such as the case of virulence factors, antibiotic resistance, tolerance to metals, siderophores, adhesion, among others functions. In Streptococcus pyogenes, the partial loss of genes in prophages maintains key genes for virulence, allowing an advantage in colonization and persistence in their host [16]. Moreover, the annotation of prophage-encoded genes revealed a remarkable frequency of genes associated with sugar and DNA metabolism, indicating an adaptation towards metabolic functions that could benefit the host, optimizing nutrient availability and facilitating adaptation to adverse environments, such as resistance to heavy metals. The presence of genes related to copper and other metal metabolism suggests that prophages may contribute to homeostasis and resistance to toxic elements, which could confer selective advantages to Clavibacter in agricultural environments. Altogether, these genes may contribute to the ability of Clavibacter strains for adapting to new environments [79]. On the other hand, prophages affect the host in diverse ways upon integration, such as causing mutagenesis and gene disruption, modifying levels of gene expression from the host, facilitating recombination acting as a hot spot or protecting from attacks of similar phages [57]. Nonetheless, some studies propose that acquisition and loss of adaptive prophage genes are constantly occurring in bacterial genomes, favoring the flux of genetic information and their evolution and ecological adaptation [55,80].

On the other hand, the presence of diverse immunity systems suggests that Clavibacter is under selective pressure to keep them, indicating that the bacteria is the target of phages. Unlike other bacteria, Clavibacter do not possess an endogenous CRISPR-cas consistent with previous reports [81]. Instead, other defense systems such as PD-lambda systems, Abi, Hachiman and Gabija, among others, form the repertoire of immunity systems in Clavibacter. Georjon et al. [82] showed that defense systems in Actinobacteria are rare systems, such as gp29-gp30 and Wadjet, and often encoded in mobile genetic elements, which was consistent with our results. Indeed, it was shown that 2% of the defense systems identified in Actinobacteria are encoded within prophages [82], indicating that prophages play a significant role as drivers of host immunity.

In summary, in this study we described the occurrence and diversity of prophage-like elements in Clavibacter spp. We demonstrated that Clavibacter prophages are pervasive and most of them are cryptic and probably with high levels of domestication. Interestingly, many prophages were unique unveiling the diversity of prophages infecting Clavibacter species. Moreover, we identified numerous genes conferring advantageous traits that improve host fitness and adaptation, including some genes associated with defense systems. Taken together, our findings highlight that prophages are relevant mediating phage-host interactions, broadening our understanding of Clavibacter biology and evolution.