Comparative Genomics and Phylogenetic Analyses Suggest Several Novel Species within the Genus Clavibacter, Including Nonpathogenic Tomato-Associated Strains

Clavibacter spp. are economically important bacterial plant pathogens infecting a set of diverse agricultural crops, such as alfalfa, corn, pepper, potato, tomato, and wheat. A number of plant-pathogenic members of the genus (e.g., C. michiganensis sensu stricto and C. sepedonicus, infecting tomato and potato plants, respectively) are included in the A2 (high-risk) list of quarantine pathogens by the European and Mediterranean Plant Protection Organization (EPPO). Although tomato-associated members of Clavibacter spp. account for a significant portion of the genetic diversity in the genus, only the strains belonging to C. michiganensis sensu stricto (formerly C. michiganensis subsp. michiganensis) cause bacterial canker disease of tomato and are subjected to the quarantine inspections. Hence, discrimination between the pathogenic and nonpathogenic Clavibacter sp. strains associated with tomato seeds and transplants plays a pivotal role in the accurate detection and cost-efficient management of the disease. On the other hand, detailed information on the genetic contents of different lineages of the genus would lead to the development of genome-informed specific detection techniques. In this study, we have provided an overview of the phylogenetic and genomic differences between the pathogenic and nonpathogenic tomato-associated Clavibacter sp. strains. We also noted that the taxonomic status of newly introduced subspecies of C. michiganensis (i.e., C. michiganensis subsp. californiensis, C. michiganensis subsp. chilensis, and C. michiganensis subsp. phaseoli) should be reconsidered.

ABSTRACT

Members of the genus Clavibacter are economically important bacterial plant pathogens infecting a set of diverse agricultural crops (e.g., alfalfa, corn, potato, tomato, and wheat). Tomato-associated Clavibacter sp. strains account for a great portion of the genetic diversity of the genus, and C. michiganensis sensu stricto (formerly C. michiganensis subsp. michiganensis), causing bacterial canker disease, is considered one of the most destructive seed-borne agents for the crop worldwide. However, current taxonomic descriptions of the genus do not reflect the existing diversity of the strains, resulting in unsatisfactory results in quarantine surveys for the pathogens. In this study, we used all the available genome sequences of Clavibacter sp. strains, including the type strains of newly described subspecies, to provide precise insight into the diversity of tomato-associated members of the genus and further clarify the taxonomic status of the strains using genotypic and phenotypic features. The results of phylogenetic analyses revealed the existence of nine hypothetical new species among the investigated strains. None of the three new subspecies (i.e., C. michiganensis subsp. californiensis, C. michiganensis subsp. chilensis, and C. michiganensis subsp. phaseoli) is included within the tomato-pathogenic C. michiganensis sensu stricto lineage. Although comparative genomics revealed the lack of chp and tomA pathogenicity determinant gene clusters in the nonpathogenic strains, a number of pathogenicity-related genes were noted to be present in all the strains regardless of their pathogenicity characteristics. Altogether, our results indicate a need for a formal taxonomic reconsideration of tomato-associated Clavibacter sp. strains to facilitate differentiation of the lineages in quarantine inspections.

IMPORTANCE Clavibacter spp. are economically important bacterial plant pathogens infecting a set of diverse agricultural crops, such as alfalfa, corn, pepper, potato, tomato, and wheat. A number of plant-pathogenic members of the genus (e.g., C. michiganensis sensu stricto and C. sepedonicus, infecting tomato and potato plants, respectively) are included in the A2 (high-risk) list of quarantine pathogens by the European and Mediterranean Plant Protection Organization (EPPO). Although tomato-associated members of Clavibacter spp. account for a significant portion of the genetic diversity in the genus, only the strains belonging to C. michiganensis sensu stricto (formerly C. michiganensis subsp. michiganensis) cause bacterial canker disease of tomato and are subjected to the quarantine inspections. Hence, discrimination between the pathogenic and nonpathogenic Clavibacter sp. strains associated with tomato seeds and transplants plays a pivotal role in the accurate detection and cost-efficient management of the disease. On the other hand, detailed information on the genetic contents of different lineages of the genus would lead to the development of genome-informed specific detection techniques. In this study, we have provided an overview of the phylogenetic and genomic differences between the pathogenic and nonpathogenic tomato-associated Clavibacter sp. strains. We also noted that the taxonomic status of newly introduced subspecies of C. michiganensis (i.e., C. michiganensis subsp. californiensis, C. michiganensis subsp. chilensis, and C. michiganensis subsp. phaseoli) should be reconsidered.

INTRODUCTION

A number of plant-pathogenic bacterial species are reported to have nonpathogenic lineages which usually exist in the same ecological niche as their pathogenic counterparts (1–3). Nonpathogenic lineages or strains typically have genetic contents similar to those of their pathogenic relatives but lack some of the key pathogenicity determinants (e.g., pathogenicity islands, virulence genes, and plasmids) (4–6). As far as economically important quarantine plant-pathogenic bacteria are concerned, the presence of nonpathogenic strains in commercial seeds or propagative parts of plants will interfere in the accurate detection of the pathogens, leading to false-positive results in the quarantine inspections and unsatisfied seed producers and traders (7). This is due in part to the fact that most of the nonpathogenic bacterial strains are phenotypically similar to their pathogenic relatives; therefore, they are not differentiable from each other on the culture media (7). Furthermore, most of the molecular detection protocols (e.g., PCR primers, probes, and antibodies) are designed based on the general features of a given species/subspecies/pathovar rather than focused on the pathogenicity determinants of the pathogen (7). As a consequence, contradictions in the results of quarantine inspections will lead to economic losses for seed producers and will have negative impact on transportation of plant materials on a global scale (7, 8).

Clavibacter spp. are economically important Gram-positive bacterial plant pathogens infecting a set of diverse agricultural crops, e.g., alfalfa, corn, pepper, potato, tomato, and wheat (9). Until recently, the genus Clavibacter was considered to include only one species, C. michiganensis, comprising five plant-pathogenic subspecies, i.e., C. michiganensis subsp. insidiosus, C. michiganensis subsp. michiganensis, C. michiganensis subsp. nebraskensis, C. michiganensis subsp. sepedonicus, and C. michiganensis subsp. tessellarius (9, 10). Furthermore, all the tomato- and pepper-associated Clavibacter sp. strains were classified as members of C. michiganensis subsp. michiganensis regardless of whether they were pathogenic or nonpathogenic on the host of isolation. However, using multilocus sequence analysis (MLSA) and multilocus sequence typing (MLST), Jacques et al. (11) showed that tomato-associated nonpathogenic Clavibacter sp. strains are phylogenetically distinct from the pathogenic counterparts in the species. Differentiation of the pathogenic and nonpathogenic strains of C. michiganensis has always been an ongoing challenge for official sanitary agencies, quarantine inspectors, and seed providers (12), since false-positive results would lead to the rejection of seed/seedling lots on an economically significant scale (7, 8). This led to the belief that a comprehensive complete genome sequence-based reconsideration of C. michiganensis sensu lato (all the former members of C. michiganensis as described by Davis et al. [9]) is warranted to shed light on the genetic diversity, genomic repertories, and taxonomic status of the pathogenic and nonpathogenic tomato-associated strains of the species (11).

Following the emergence of high-throughput molecular-phylogenetic techniques, many Clavibacter sp. strains which were often previously misidentified based on phenotypic features were assigned to novel taxa. For instance, tomato-associated nonpathogenic members of C. michiganensis sensu lato were assigned to two new subspecies, C. michiganensis subsp. californiensis and C. michiganensis subsp. chilensis (13). Additionally, C. michiganensis subsp. phaseoli and C. michiganensis subsp. capsici were identified as the causal agents of bacterial bean leaf yellowing on common bean (Phaseolus vulgaris) and bacterial canker of pepper (Capsicum annuum), respectively (14, 15). Furthermore, nonpathogenic peach-colored strains isolated from the tomato phyllosphere were reported to be distinct from the tomato-pathogenic members of Clavibacter spp. (16).

Recently, a reclassification of Clavibacter spp. into five new species and a new combination was proposed based on genomic information, e.g., average nucleotide identity (ANI) and digital DNA-DNA hybridization (dDDH) indices (17–19). The original subspecies of C. michiganensis sensu lato were elevated to the species level and designated C. michiganensis (here referred to as C. michiganensis sensu stricto, formerly C. michiganensis subsp. michiganensis), C. tessellarius, C. insidiosus, C. nebraskensis, and C. capsici, as well as C. sepedonicus as new combination (17, 19). However, due to the lack of genomic information from the newly proposed subspecies (C. michiganensis subsp. californiensis, C. michiganensis subsp. chilensis, and C. michiganensis subsp. phaseoli) as well as several taxonomically undetermined strains, additional investigations are warranted to further clarify the taxonomy of the genus. Moreover, strains associated with solanaceous vegetables contain a large fraction of diversity within the Clavibacter sp. members, and much of the molecular, phylogenetic, and genomic information for these strains remains unexplored. For the tomato-associated strains of Clavibacter spp., comparative genomics on a wide collection of nonpathogenic and pathogenic strains would further elucidate the genetic diversity of these bacteria, resulting in the development of genome-informed specific molecular markers (e.g., specific conventional PCR and real-time PCR primers, as well as loop-mediated isothermal amplification) for the detection and differentiation of the pathogenic and nonpathogenic strains in the quarantine posts.

The objectives of the present study were to (i) investigate the genetic diversity of tomato-associated Clavibacter sp. strains using the genome sequences of all available nonpathogenic and pathogenic strains and (ii) provide a novel taxonomic overview into the status of tomato-pathogenic and nonpathogenic strains within the genus. For this aim, we used the genome sequences of 40 Clavibacter sp. strains, including the type strains of three newly described subspecies (C. michiganensis subsp. californiensis, C. michiganensis subsp. chilensis, and C. michiganensis subsp. phaseoli), as well as additional atypical nonpathogenic strains isolated from tomato plants around the globe (20). Draft genome sequence-based phylogenetic analyses revealed a higher diversity among the nonpathogenic strains of Clavibacter spp. than has previously been reported, delineating them into several new species. On the other hand, our data revealed that the two individual subspecies C. michiganensis subsp. chilensis and C. michiganensis subsp. phaseoli need to be considered members of one species, according to the 99% genome similarity among the type strains. Furthermore, comparative genomics among the pathogenic and nonpathogenic tomato-associated strains, as well as the type strains of the remaining species/subspecies within the genus, revealed several pathogenicity determinant genes presenting only in C. michiganensis sensu stricto, which could be considered suitable genomic targets for the development of specific detection methods for the tomato pathogen.

RESULTS

Pathogenicity and host range.

Tomato and pepper plants inoculated with the standard strain of C. michiganensis sensu stricto (ICMP 22049) showed the expected disease symptoms at 10 to 12 days postinoculation. Although tomato plants inoculated with strain ICMP 22049 showed wilting and plant death (see Fig. S1a in the supplemental material), pepper plants inoculated using the same strain showed only stem canker symptoms on the site of inoculation, with no wilting or plant death, in the same time frame (Fig. S1b). However, neither C. michiganensis subsp. phaseoli nor C. michiganensis subsp. chilensis induced symptoms on the inoculated plant species, i.e., common bean, cowpea, pepper, mung bean, and tomato (Table 1). Furthermore, we could not reisolate C. michiganensis subsp. phaseoli and C. michiganensis subsp. chilensis from the stem, petiole, and leaf tissues 5 to 10 cm above the inoculation site on the stem. This could be an indication of the fact that C. michiganensis subsp. phaseoli and C. michiganensis subsp. chilensis were unable to endophytically colonize the evaluated plant species. For the orange-pigmented tomato-associated strains CFBP 8615 and CFBP 8616, although no symptoms were observed on common bean, cowpea, pepper, mung bean, and tomato plants, bacterial colonies similar to those originally inoculated were consistently reisolated from the leaf tissues of common bean cultivar Navy plants inoculated with CFBP 8616 (Table 1). Furthermore, the standard strain ICMP 22049 was consistently reisolated from the symptomatic pepper and tomato plants on yeast extract-peptone-glucose agar (YPGA) medium, and their identity was confirmed using the genus-specific primer pair CMR16F1/CMR16R1 (data not shown). Similar results were obtained in both replications of the experiments, while the negative-control plants remained healthy.

TABLE 1.

Results of pathogenicity tests and host range assays of the type strains of Clavibacter michiganensis subsp. phaseoli and C. michiganensis subsp. chilensis, as well as two atypical peach-colored Clavibacter sp. strains, on different annual crops under greenhouse conditions^a

Taxon	Strain	Pathogenicityb on:
		Pepper		Common bean			Cowpea	Mung bean	Tomato
		cv. Sereno	cv. Aziz	cv. Red kidney	cv. Pinto	cv. Navy
C. michiganensis subsp. phaseoli	CFBP 8627T	−	−	−	−	−	−	−	−
C. michiganensis subsp. chilensis	CFBP 8217T	−	−	−	−	−	−	−	−
Clavibacter sp.	CFBP 8615	−	−	−	−	−	−	−	−
Clavibacter sp.	CFBP 8616	−	−	−	−	−c	−	−	−
C. michiganensis sensu stricto	ICMP 22049	+d	+d	−	−	−	−	−	+

^aNone of the evaluated bacterial strains was pathogenic on the tested plants, while the nonpathogenic strain CFBP 8616 was reisolated from common bean cv. Navy tissues.

^b−, negative; +, positive.

^cThe inoculated bacterial strain was reisolated from the asymptomatic leaf tissues of the test plants.

^dThese plants showed only stem canker symptoms on the site of inoculation, with no systemic wilting and plant death.

Phylogenetic analyses.

A neighbor-joining phylogenetic tree constructed using the genome sequences of 40 Clavibacter sp. strains (Table 2) via the ANI Calculator online service with the all-versus-all strategy revealed high genetic diversity among tomato-associated nonpathogenic strains of the genus (Fig. 1). ANI values between different pairs of strains varied from 87% to 100% among the Clavibacter sp. strains (Table 3). While all tomato-pathogenic strains of C. michiganensis sensu stricto clustered in a monophyletic clade showing 99 to 100% ANI with one another, nonpathogenic strains isolated from tomato were scattered in several clades, most of which had <96% ANI values with the other clades (Table 3). The closest nonpathogenic clade to the C. michiganensis sensu stricto group consisted of three strains, i.e., the type strain of C. michiganensis subsp. californiensis (CFBP 8216^T), and the nonpathogenic strains LMG 26808 and CFBP 7493. the ANI value between the type strain of C. michiganensis sensu stricto (LMG 7333^T) and the type strain of C. michiganensis subsp. californiensis was 95% in all the calculating strategies, retaining them at the threshold of species definition (21). Nevertheless, the dDDH value (57.70%) between the type strains of C. michiganensis sensu stricto and C. michiganensis subsp. californiensis was far below the threshold for species delineation (70%) with this method (Table 3). Altogether, given the differences in their pathogenicities and biochemical characteristics (13) as well as the below-threshold genomic similarity, the two taxa C. michiganensis sensu stricto and C. michiganensis subsp. californiensis could be considered separate species. Furthermore, the ANI between the type strains of C. michiganensis sensu stricto and C. michiganensis subsp. californiensis and the cluster which included the nonpathogenic strains LMG 26808 and CFBP 7493 was 94 to 95% (Table 3). The dDDH values between these type strains and LMG 26808 and CFBP 7493 were also 57 to 58%, indicating them as separate species (Fig. 1).

TABLE 2.

Clavibacter sp. genome sequences used for the comparative genomics and phylogenetic analyses^a

Previous nameb	New namec	Strain	Host of isolation	Pathogenicity	Date	Country of Isolation	GenBank accession no.	Reference
Clavibacter sp.	New species IV	CFBP 8615d	Solanum lycopersicum	NPe	2015	Iran	QWGT01000000	20
Clavibacter sp.	new species IV	CFBP 8616d	Solanum lycopersicum	NP	2015	Iran	QWGU01000000	20
C. michiganensis subsp. californiensis	New species II	CFBP 8216Td	Solanum lycopersicum	NP	2000	USA	QWEE01000000	20
C. michiganensis subsp. chilensis	New species III	CFBP 8217Td	Solanum lycopersicum	NP	2007	Netherlands	QWGS01000000	20
Clavibacter sp.	New species III	CFBP 7491d	Solanum lycopersicum	NP	NDf	ND	QWEB01000000	20
Clavibacter sp.	New species I	CFBP 7493d	Solanum lycopersicum	NP	ND	ND	QWEC01000000	20
C. michiganensis subsp. insidiosus	C. insidiosus	CFBP 1195	Medicago sativa	Alfalfa	1964	UK	QWDZ01000000	20
C. michiganensis subsp. insidiosus	C. insidiosus	CFBP 6488	Medicago sativa	Alfalfa	1998	Czech Republic	QWEA01000000	20
C. michiganensis subsp. nebraskensis	C. nebraskensis	CFBP 7577	Zea mays	Corn	ND	ND	QWED01000000	20
C. michiganensis subsp. phaseoli	New species III	CFBP 8627T	Phaseolus vulgaris	NP	2009	Spain	QWGV01000000	20
Clavibacter sp.	New species VIII	CASJ009d	Solanum lycopersicum	NP	2011	USA	MDHJ00000000.1	22
Clavibacter sp.	New species V	CFBP 8019d	Solanum lycopersicum	NP	2011	USA	NZ_MDJZ00000000.1	22
Clavibacter sp.	New species IX	CFBP 8017d	Solanum lycopersicum	Wheat	2006	Netherlands	MDJY00000000	22
Clavibacter sp.	New species VI	CF11	soil	NP	2011	China	JROD00000000	23
Clavibacter sp.	New species I	LMG 26808d	Solanum lycopersicum	NP	ND	Netherlands	AZQZ00000000	28
Clavibacter sp.	C. insidiosus	CFBP 7494d	Solanum lycopersicum	Wheat	1999	Chile	MDJW00000000	22
Clavibacter sp.	New species VII	DOAB 609	Triticum aestivum	ND	1976	USA	LQXA00000000	50
C. michiganensis subsp. michiganensis	C. michiganensis	LMG 7333T	Solanum lycopersicum	Tomato	1957	Hungary	NZ_MZMP00000000	51
C. michiganensis subsp. michiganensis	C. michiganensis	NCPPB 382	Solanum lycopersicum	Tomato	1956	UK	AM711867.1	26
C. michiganensis subsp. michiganensis	C. michiganensis	CASJ004	Solanum lycopersicum	Tomato	1999	USA	MDHE00000000	22
C. michiganensis subsp. michiganensis	C. michiganensis	CAYO001	Solanum lycopersicum	Tomato	2001	USA	MDHL00000000	22
C. michiganensis subsp. michiganensis	C. michiganensis	CASJ005	Solanum lycopersicum	Tomato	2001	USA	MDHF00000000	22
C. michiganensis subsp. michiganensis	C. michiganensis	CA00001	Solanum lycopersicum	Tomato	2000	USA	MDHK00000000	22
C. michiganensis subsp. michiganensis	C. michiganensis	CASJ008	Solanum lycopersicum	Tomato	2002	USA	MDHI00000000	22
C. michiganensis subsp. michiganensis	C. michiganensis	CASJ006	Solanum lycopersicum	Tomato	2002	USA	MDHG00000000	22
C. michiganensis subsp. michiganensis	C. michiganensis	CASJ003	Solanum lycopersicum	Tomato	1999	USA	MDHD00000000	22
C. michiganensis subsp. michiganensis	C. michiganensis	VKM Ac-1403	Solanum lycopersicum	Tomato	2017	USA	FVZG00000000	52
C. michiganensis subsp. michiganensis	C. michiganensis	CASJ002	Solanum lycopersicum	Tomato	1999	USA	MDHC00000000	22
C. michiganensis subsp. michiganensis	C. michiganensis	CA00002	Solanum lycopersicum	Tomato	2000	USA	MDHM00000000	22
C. michiganensis subsp. michiganensis	C. michiganensis	CASJ007	Solanum lycopersicum	Tomato	2011	USA	MDHH00000000	22
C. michiganensis subsp. capsici	C. capsici	PF008T	Capsicum sp.	Pepper	ND	South Korea	NZ_CP012573	15
C. michiganensis subsp. capsici	C. capsici	CFBP 7576d	Solanum lycopersicum	Pepper	1997	ND	MDJX00000000	22
C. michiganensis subsp. insidiosus	C. insidiosus	LMG 3663T	Medicago sativa	Alfalfa	1955	USA	MZMO00000000	51
C. michiganensis subsp. insidiosus	C. insidiosus	R1-1	Medicago truncatula	Alfalfa	2009	USA	NZ_CP011043	29
C. michiganensis subsp. nebraskensis	C. nebraskensis	NCPPB 2581T	Zea mays	Corn	1971	USA	NC_020891.1	50
C. michiganensis subsp. nebraskensis	C. nebraskensis	DOAB 395	Zea mays	Corn	2014	Canada	LSOE00000000	50
C. michiganensis subsp. nebraskensis	C. nebraskensis	DOAB 397	Zea mays	Corn	2014	Canada	LAKL00000000	53
C. michiganensis subsp. sepedonicus	C. sepedonicus	ATCC 33113T	Solanum tuberosum	Potato	ND	Canada	NC_010407.1	54
C. michiganensis subsp. sepedonicus	C. sepedonicus	CFIA-CsR14	Solanum tuberosum	Potato	ND	Canada	MZMN00000000	51
C. michiganensis subsp. sepedonicus	C. sepedonicus	CFIA-Cs3N	Solanum tuberosum	Potato	ND	Canada	MZMM00000000	51
C. michiganensis subsp. tessellarius	C. tessellarius	ATCC 33566T	Triticum aestivum	Wheat	1978	USA	MZMQ01000000	51

^aThe first 10 sequences were obtained in this study and announced previously (20), while the remaining ones were retrieved from the NCBI GenBank database.

^bThe original nomenclature of the taxon.

^cEither revised taxonomy of the strains (17) or their new taxonomic status as proposed in this study. Although new names of six species (i.e., C. capsici, C. insidiosus, C. michiganensis, C. nebraskensis, C. sepedonicus, and C. tessellarius) were formally described previously (17), a formal taxonomic description is needed for the hypothetical new species I to IX, which are based on the ANI/dDDH values (Table 3) and comparative genomics (Table 4).

^dThe strain was isolated from tomato plant but was nonpathogenic on the host of isolation.

^eNP, nonpathogenic.

^fND, not determined.

FIG 1.

Average nucleotide identity (ANI)-based neighbor-joining phylogenetic tree of 40 Clavibacter sp. strains constructed using the ANI calculator online service. Different colors represent hypothetical new species (I to IX). Seven hypothetical novel species were determined among tomato-associated nonpathogenic Clavibacter sp. strains. Furthermore, based on the ANI/dDDH indices, the type strains of C. michiganensis subsp. chilensis and C. michiganensis subsp. phaseoli belong to the same novel species. *, C. michiganensis subsp. californiensis.

TABLE 3.

Average nucleotide identity and digital DNA-DNA hybridization values among the type and/or representative strains of different lineages defined within the genus Clavibacter^a

No.	Taxon	Strain	ANIb (lower diagonal) or dDDH (upper diagonal) (%)
			1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17
1	C. michiganensis sensu stricto	LMG 7333T		57.70	58.70	48.00	48.60	49.70	48.00	49.80	49.50	44.80	43.10	40.20	40.00	37.50	37.20	36.80	34.60
2	C. michiganensis subsp. californiensis	CFBP 8216T	95/95/95		57.80	46.40	47.90	48.60	47.40	48.10	47.90	44.80	43.70	40.50	40.40	38.50	38.40	38.00	35.70
3	Clavibacter sp.	LMG 26808	95/94/95	94/95/95		45.80	47.80	48.40	47.10	47.90	47.70	44.30	42.60	39.60	39.50	37.10	37.00	36.40	34.50
4	C. sepedonicus	ATCC 33113T	92/92/92	92/92/92	92/92/92		45.10	46.20	45.20	47.00	46.60	43.90	42.30	39.10	39.10	36.40	36.20	36.00	34.00
5	C. insidiosus	LMG 3663T	93/93/93	93/93/93	92/92/93	92/92/92		64.60	59.90	51.20	51.00	47.10	43.30	40.50	40.20	37.90	37.40	37.20	34.80
6	Clavibacter sp.	CFBP 7494	93/93/93	92/93/93	93/92/93	92/92/92	96/96/96		60.20	52.20	52.10	46.00	44.10	40.70	40.60	38.10	37.60	37.50	34.90
7	C. nebraskensis	NCPPB 2581T	92/93/92	92/93/92	92/92/92	92/92/92	95/95/95	95/95/95		51.40	51.10	45.30	43.30	40.70	40.70	37.70	37.20	36.90	34.90
8	C. michiganensis subsp. chilensis	CFBP 8217T	93/93/93	93/93/93	92/93/93	92/93/92	93/93/93	93/93/94	93/93/93		87.50	47.00	45.80	43.40	42.90	40.20	39.50	39.20	37.10
9	C. michiganensis subsp. phaseoli	CFBP 8627T	93/93/93	93/93/93	92/92/93	92/93/92	93/93/93	93/93/93	93/93/93	99/99/99		47.10	45.70	43.30	42.90	40.00	39.50	39.10	37.00
10	Clavibacter sp.	CFBP 8615	92/92/92	92/92/92	91/92/92	91/92/92	92/92/92	92/92/92	92/92/92	92/92/92	92/93/92		48.00	40.90	40.50	38.20	37.60	37.70	35.80
11	Clavibacter sp.	CFBP 8019	91/92/91	91/92/91	91/91/91	91/92/91	91/92/91	91/91/91	91/91/91	92/92/92	92/92/92	93/93/93		40.10	40.00	36.90	36.80	36.80	34.40
12	C. capsici	PF008T	90/91/90	90/91/91	90/90/90	90/91/90	90/91/90	90/90/90	90/91/90	91/92/91	91/92/91	90/91/91	90/91/90		58.50	38.40	38.50	37.70	34.50
13	Clavibacter sp.	CF11	90/90/90	90/90/90	90/90/90	90/90/90	90/90/90	91/90/90	91/90/90	91/91/91	91/91/91	90/90/90	90/90/90	95/95/95		38.10	38.50	37.60	34.40
14	C. tessellarius	ATCC 33566T	90/90/89	90/91/90	89/89/89	89/90/89	90/90/89	90/90/90	90/90/89	91/91/90	91/91/90	90/90/90	89/90/89	90/91/90	90/90/90		49.00	57.70	33.50
15	Clavibacter sp.	DOAB 609	89/89/89	89/90/90	89/89/89	89/89/89	89/89/89	90/90/89	89/89/89	90/90/90	90/90/90	89/90/89	89/89/89	90/90/90	90/90/90	93/93/93		47.20	33.60
16	Clavibacter sp.	CFBP 8017	89/89/89	89/90/90	89/89/89	89/89/89	89/89/89	90/90/89	89/89/89	90/90/90	90/90/90	89/90/89	89/89/89	90/90/89	90/90/89	95/95/95	92/92/92		33.30
17	Clavibacter sp.	CASJ009	88/89/89	89/90/88	88/88/88	88/89/88	88/89/88	88/88/88	88/89/88	89/90/89	89/90/90	89/90/89	88/89/88	88/89/89	88/88/88	88/89/88	88/88/87	87/88/87

^aA combination of ANI and dDDH indices was used to designate a taxonomic status for a given phylogenetic clade, where the “new species” status was assigned to a clade only when both the ANI and dDDH values were below the accepted thresholds (≤95% and ≤70% for ANI and dDDH, respectively [21]).

^bANI values were calculated using three different algorithms, i.e., JSpeciesWS, ANI calculator, and OrthoANIu, which are presented from left to right, respectively.

Clavibacter sepedonicus strains formed a monophyletic cluster separate from all the other lineages by ANI values of <93%, which is in coherence with its elevation at the species level (17, 18). Clavibacter insidiosus and C. nebraskensis strains clustered in a monophyletic group showing 95% ANI between the type strains of the species (Table 3). Here also, the dDDH value (59.90%) between the type strains of these two taxa was far below the threshold for species definition (70%) with this method (Table 3), supporting their elevation into separate species (17–19). The two taxa are also different in their hosts of isolation and pathogenicity patterns. While strain CFBP 7494 showed only a 64.60% dDDH value with the type strain of C. insidiosus, the 96% ANI (on the upper edge of species definition) prevents differentiation of this strain from the C. insidiosus species. Strain CFBP 7494 was isolated from tomato but was nonpathogenic on this plant species, while it has been shown that it induces disease symptoms on wheat plants under greenhouse conditions (22). Further evidence, including a comprehensive field survey and host range assay, is needed to elucidate the prevalence and exact taxonomic status of strain CFBP 7494. Surprisingly, the type strains of C. michiganensis subsp. phaseoli and C. michiganensis subsp. chilensis shared 99% ANI with one another and 98% ANI with CFBP 7491, isolated from tomato seeds. These three strains had ANIs below 93% with all the remaining clades, suggesting a novel species within the genus. A high dDDH value (87.50%) also confirmed the close relationship between the type strains of C. michiganensis subsp. phaseoli and C. michiganensis subsp. chilensis (Table 3).

Two peach-colored strains, CFBP 8615 and CFBP 8616, shared 100% ANI with one another, while they differed from all the remaining clades, with ANI values of <93%. Furthermore, nonpathogenic strain CFBP 8019 was determined to be the phylogenetically closest strain to the peach-colored strains, with 93% ANI values between the two clades. These ANI values are far below the accepted threshold (95 to 96%) for the definition of prokaryotic species (21), suggesting that strains CFBP 8615 and CFBP 8616 could be defined as forming a new species separated from CFBP 8019, while strain CFBP 8019 itself belongs to a new stand-alone species (Fig. 1). Low ANI values were also confirmed by dDDH values, which were <48% between the peach-colored strains and all the remaining clades (Table 3). Strain CF11, isolated from soil in a tomato-growing greenhouse (23), as well as the type strain of the pepper pathogen C. capsici (PF008^T) clustered in a monophyletic clade, while they differed from one another with 95% ANI and 58.50% dDDH values. Hence, CF11 could be proposed as forming a new species within the genus, while the elevation of former C. michiganensis subsp. capsici to the species level (C. capsici) was confirmed, as proposed by Li et al. (17). The type strain of C. tessellarius showed <93% ANI with the type strains of all the other subspecies/species, confirming the wheat pathogen as a stand-alone species. However, neither strain CFBP 8017 nor strain DOAB 609, which were clustered in a shared clade with the type strain of C. tessellarius, could be included within this species. The ANIs of CFBP 8017 and DOAB 609 with the type strain of C. tessellarius were 95% and 93%, respectively, while the dDDH values between the same strains were 57.70% and 49.00%, respectively (Table 3). Thus, each of the CFBP 8017 and DOAB 609 strains could be defined as representing novel species. The nonpathogenic strain CASJ009 also had ANI values of <90% with all the Clavibacter sp. strains evaluated in this study, indicating that this strain also represents a novel species within the genus (Fig. 1; Table 3).

Comparative genomics.

Comparative genomics data obtained using the RAST online service revealed that the genome size among the Clavibacter sp. strains varied between 3,024 kbp in CFBP 8019 and 3,420 kbp in LMG 26808, with G+C contents ranging from 72.0% in LMG 26808 to 73.7% in C. tessellarius ATCC 33566^T. Furthermore, the number of coding sequences (CDS) varied from 2,629 in C. michiganensis subsp. chilensis (CFBP 8217^T) to 3,181 in DOAB 609. Genomic characteristics of Clavibacter spp. in a panel of 20 representative strains, which were selected on the basis of the phylogenetic analyses (as detailed above) to cover all lineages/clades of the genus, are shown in Table 4. The number of subsystems varied from 260 in CF11 to 345 in the reference strain of C. michiganensis sensu stricto, NCPB 382, and the type strain of C. sepedonicus, ATCC 33113^T. Although the feature counts were similar in most of the subsystems among the pathogenic and nonpathogenic Clavibacter sp. strains, differences in the siderophore-producing subsystems were observed, and the siderophore assembly kit was detected only in the nonpathogenic strain CFBP 8616 (Table 4).

TABLE 4.

Genomic characteristics of Clavibacter sp. strains used in this study^a

Parameter or subsystem feature	C. michiganensis sensu stricto (NCPPB 382)	C. capsici (PF008T)	Clavibacter sp. (CFBP 8615)	Clavibacter sp. (CFBP 8616)	C. michiganensis subsp. californiensis (CFBP 8216T)	Clavibacter sp. (CFBP 7491)	Clavibacter sp. (CFBP 7493)	Clavibacter sp. (CFBP 7494)	Clavibacter sp. (CFBP 8019)	Clavibacter sp. (CFBP 8017)	Clavibacter sp. (DOAB 609)	Clavibacter sp. (CF11)	Clavibacter sp. (LMG 26808)	C. michiganensis subsp. phaseoli (CFBP 8627T)	C. michiganensis subsp. chilensis (CFBP 8217T)	Clavibacter sp. (CASJ009)	C. insidiosus (LMG 3663T)	C. nebraskensis (NCPPB 2581T)	C. sepedonicus (ATCC 33113T)	C. tessellarius (ATCC 33566T)
Genome size (bp)	3,297	3,056	3,129	3,094	3,193	3,288	3,275	3,313	3,024	3,172	3,296	3,118	3,420	3,052	3,044	3,268	3,387	3,063	3,258	3,318
GC content (%)	72.7	73.6	73.2	73.2	72.7	73.0	72.9	73.3	73.5	73.5	73.2	73.6	72.0	73.5	73.5	73.6	72.7	73.0	72.6	73.7
No. of CDS	2,979	2,725	2,807	2,730	2,784	2,917	2,897	2,956	2,676	3,014	3,181	3,002	3,097	2,642	2,629	3,054	3,091	2,739	3,047	2,956
No. of subsystems	345	326	323	319	338	316	342	330	326	263	266	260	340	315	311	341	332	325	345	317
No. of RNAs	51	51	47	47	48	50	48	50	48	48	53	49	57	50	49	51	52	51	51	52
No. of CDS for:
Cofactors, vitamins, pigments	177	158	169	157	165	139	173	161	158	107	116	99	175	167	130	181	165	139	169	176
Virulence, disease, defense	34	38	37	36	38	36	36	32	38	26	25	25	40	35	34	42	36	30	33	22
Resistance to antibiotics/toxic compounds	19	23	22	21	23	21	21	17	24	17	16	15	25	21	19	27	21	17	20	18
Invasion and intracellular resistance	15	15	15	15	15	15	15	15	14	9	9	10	15	14	15	15	15	13	13	4
Potassium metabolism	13	12	12	12	13	12	13	13	16	5	5	5	9	9	13	8	13	9	10	11
Phages, prophages, transposable elements	0	0	0	0	2	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0
Iron acquisition and metabolism	16	17	13	25	16	8	19	18	13	5	10	5	17	10	10	12	15	14	17	16
Siderophoresb
Yersiniabactin	2	0	0	0	2	0	2	0	0	0	5	0	2	0	0	0	2	2	0	0
Aerobactin	3	3	3	3	3	3	4	3	3	0	0	0	3	3	3	3	3	3	3	3
Nucleosides and nucleotides	87	72	93	92	89	76	100	76	81	80	79	75	74	98	68	90	79	80	99	65
Protein metabolism	211	201	209	192	189	185	209	204	208	166	162	167	205	178	198	192	205	178	213	185
Cell division and cell cycle	26	21	8	8	23	24	24	23	21	0	0	0	8	23	8	22	24	20	22	23
Motility and chemotaxis	1	2	2	2	2	3	1	2	2	0	0	0	3	3	3	3	2	1	2	2
Regulation and cell signaling	33	30	19	20	22	22	25	30	33	14	17	18	36	28	17	29	31	33	35	24
Secondary metabolism	6	18	5	5	4	5	11	12	11	9	9	9	12	0	1	11	5	5	13	11
Fatty acids, lipids, and isoprenoids	88	78	84	103	82	108	78	88	59	44	44	42	79	85	76	82	65	65	80	101
Nitrogen metabolism	7	10	6	6	10	6	7	6	6	6	6	6	7	6	6	11	7	7	6	11
Stress response	67	73	76	69	74	67	64	70	78	27	27	28	68	71	61	82	74	71	78	64
Metabolism of aromatic compounds	10	8	9	8	9	14	10	10	8	4	3	3	10	10	10	18	10	9	10	10
Amino acids and derivatives	224	190	223	212	233	221	210	192	199	217	221	203	236	193	195	224	219	229	232	220
Sulfur metabolism	20	16	16	11	9	12	10	16	17	6	5	5	13	10	11	26	17	9	22	18
Phosphorus metabolism	38	26	25	27	28	26	27	23	25	28	27	27	36	25	26	25	23	23	25	24
Carbohydrates	285	266	255	268	267	283	247	268	237	181	194	175	278	263	261	382	266	266	269	281

^aIndividual genomes were analyzed using the online annotating service RAST, and protein-encoding sequences (CDS), functions of the genes, and subsystems represented in the genomes were determined for each genome using the SEED-Viewer comparative environment. Features corresponding to cell wall and capsule, DNA metabolism, dormancy and sporulation, membrane transport, respiration, RNA metabolism, and miscellaneous groups are not included since they were common among all the evaluated taxa.

^bThe siderophore assembly kit was detected only in Clavibacter sp. strain CFBP 8616.

One-versus-one BLASTn- and BLASTp-based explorations using the complete genome sequence of C. michiganensis sensu stricto NCPPB 382 as the reference genome versus the individual Clavibacter sp. strains revealed the lack of pathogenicity determinant genes/clusters in all the tomato-associated nonpathogenic strains evaluated in this study (Table 5). For the chp gene cluster (i.e., loci CMM_0034 to CMM_0077 in the NCPPB 382 genome sequence [accession no. AM711867.1]), only a fraction of the genes were detected in the nonpathogenic strains (Fig. 2). For instance, a sugar phosphate isomerase (CMM_0034) was present in all the nonpathogenic strains. A putative phosphotransferase (CMM_0065) and ATPase (parX = CMM_0066) were found in CFBP 7491, CFBP 7493, and LMG 26808. A hypothetical protein produced by the CMM_0054 locus and a transcriptional regulator protein secreted by CMM_0055 were found only in the type strain of C. michiganensis subsp. californiensis. A serine protease (ppaD = CMM_0075) and a putative ATPase (CMM_0067) were found only in CASJ009 and CFBP 7491, respectively, while a putative DNA invertase (CMM_PS_07) were found in both of the last strains. Among the pathogenicity determinant genes inside the chp gene cluster, ppaA (CMM_0041), pelA1 (CMM_0043), pelA2 (CMM_0051), chpC (CMM_0052), and chpG (CMM_0059) were found in none of the evaluated nonpathogenic strains. Non-chp pathogenicity determinant genes clvG (CMM_1963), clvF (CMM_1964), clvA (micA = CMM_1967), and perF (CMM_2382) were also not detected in the evaluated nonpathogenic bacterial strains. Interestingly, a subtilisin-like serine protease (sbtA = CMM_0070) was found in a number of nonpathogenic strains (Table 5), while nucleotides 1 to 600 were missing in all the nonpathogenic members. The expansin-encoding gene expA (CMM_1480) was found in CFBP 8017, DOAB 609, CFBP 7493, and LMG 26808.

TABLE 5.

Results of one-versus-one BLASTn/BLASTp searches using the genome sequence of Clavibacter michiganensis sensu stricto NCPPB 382 (accession no. AM711867.1) against all other genome sequences shown in Table 2^a

Locus tag on NCPPB 382 genome	Gene or cluster	Length (bp)	Query coverage and sequence similarity (%)b
			C. michiganensis sensu stricto (NCPPB 382)	Clavibacter sp. (CFBP 8615)	Clavibacter sp. (CF11)	Clavibacter sp. (CFBP 8017)	Clavibacter sp. (DOAB 609)	C. michiganensis subsp. californiensis (CFBP 8216T)	C. michiganensis subsp. chilensis (CFBP 8217T)	C. capsici (PF008T)	Clavibacter sp. (CFBP 7491)	Clavibacter sp. (CFBP 7493)	Clavibacter sp. (LMG 26808)	Clavibacter sp. (CFBP 7494)	Clavibacter sp. (CFBP 8019)	Clavibacter sp. (CASJ009)	C. michiganensis subsp. phaseoli (CFBP 8627T)	C. insidiosus (LMG 3663T)	C. nebraskensis (NCPPB 2581T)	C. sepedonicus (ATCC 33113T)	C. tessellarius (ATCC 33566T)	Reference
CMM_0070	sbtA	3,102	[100]100	[75]75	[98]82	ND	[97]81	[82]82	[76]94	[97]82	[97]89	ND	ND	ND	[98]82	[77]82	[96]89	ND	[99]80	ND	ND	43
CMM_0013	srtA	801	[100]100	[100]92	[100]87	[100]87	[100]87	[100]97	[100]91	[100]87	[100]91	[100]97	[100]97	[100]90	[100]89	[100]89	[100]91	[100]92	[100]92	[100]93	[100]88	43
CMM_1480	expA	762	[100]100	ND	ND	[77]81	[77]82	ND	ND	ND	ND	[100]99	[100]99	ND	ND	ND	ND	ND	ND	ND	NDc	26
CMM_1673	xysA	1,296	[100]100	[98]90	[99]87	[99]92	[97]91	[99]95	[81]76	[100]88	[83]76	[99]94	[99]94	[99]93	[99]89	[48]80	[81]75	[99]95	[99]92	ND	[99]92	43
CMM_1674	xysB	2,010	[100]100	[92]93	[100]89	[100]89	[54]75	[100]93	[100]93	[100]90	[100]93	[94]95	[99]94	[100]92	[100]92	[95]90	[93]93	[100]90	[100]91	ND	[99]89	43
CMM_2443	celB	1,608	[100]100	ND	ND	[100]89	[100]87	ND	ND	[100]88	ND	[100]91	[100]91	[100]91	ND	ND	ND	[100]92	[64]85	[100]91	[100]85	26
CMM_2691	NAd	1,002	[100]100	[99]89	[100]89	[100]89	[99]90	[100]91	[100]89	[99]89	[100]89	[100]94	[100]94	[89]92	[99]90	[97]83	[100]88	[100]94	[98]92	[99]89	[100]90	26
CMM_2692	NA	1,410	[100]100	[90]93	[100]86	[100]86	[100]86	[91]96	[77]90	[100]89	[30]86	[100]91	[100]91	[100]90	[100]87	ND	[90]90	[100]92	[100]91	[100]86	[100]88	26
CMM_2871	NA	1,491	[100]100	ND	ND	ND	ND	ND	ND	ND	ND	[100]90	[100]90	ND	ND	ND	ND	[92]88	[98]92	ND	[100]92	26
CMM_2645	vatr1	624	[100]100	[100]94	[100]92	[100]92	[100]91	[85]98	[100]93	[100]92	[100]94	[100]98	[100]98	[100]96	[100]93	[100]89	[100]94	[100]96	[100]96	[100]94	[100]92	25
CMM_2969	vatr2	1,323	[100]100	[100]92	[100]92	[99]92	[100]92	[100]96	[100]94	[100]92	[100]94	[100]95	[100]95	[100]94	[100]93	[100]90	[100]94	[100]93	[100]93	[100]93	[100]91	25
pCM1_0020	celA	2,241	ND	ND	[46]84	[46]75	[47]75	ND	ND	[45]77	ND	ND	ND	ND	ND	ND	ND	[97]90	[76]84	ND	[45]75	26

^aPutative pathogenicity determinant genes/regions described in the literature were subjected to the analyses. While nine chromosomal genes (i.e., chpC, chpG, clvA [micA], clvF, clvG, pelA1, pelA2, perF, and ppaA) as well as the pCM2 plasmid-borne pathogenicity-associated gene pat-1 were determined as present exclusively in C. michiganensis sensu stricto, the remaining genes were detected in different phylogenetic lineages regardless of their pathogenicity and host range.

^bThe values in brackets refer to the query coverage of the locus, while the other values refer to the sequence similarity between the reference strain (NCPPB 382^T) and the strain in question. Query coverage of <50% was not considered reliable data. ND, not detected (absence of the target gene/cluster).

^cA putative expansin protein was found in this species (GenBank accession no. OQJ63896.1), in which the nucleotide sequence was different from that of CMM_1480 expA.

^dNA, not assigned.

FIG 2.

Results of comparative genomics on the two main pathogenicity determinant regions of Clavibacter michiganensis sensu stricto NCPPB 382 (i.e., chp and tomA) against the nonpathogenic tomato-associated strains, as well as the type strains of five species that are pathogenic on other plants. While only a fraction (<10%) of the clusters were detected in the nonpathogenic strains (A), variations were observed in the patterns of the genes (B).

For the tomA gene cluster (CMM_0078 to CMM_0112 in the genome sequence of NCPPB 382), a β-glucosidase-related gene (bglC = CMM_0083) was found in the type strains of C. michiganensis subsp. chilensis, C. capsici, and C. michiganensis subsp. phaseoli, as well as strains CFBP 8615, CFBP 8616, CFBP 7493, LMG 26808, and CFBP 7494. A putative alpha-glucosidase gene (aglA = CMM_0106) was found in CFBP 7491, CFBP 8019, and CASJ009. Furthermore, a putative ABC-type sugar transport permease gene (CMM_0108) was found in CFBP 7491 and CASJ009, while the srtA gene (CMM_0013), which encodes a putative sortase enzyme, was found in all the evaluated nonpathogenic strains in this study (Table 5).

We also assessed the presence of a set of eight virulence genes, i.e., celB (CMM_2443), pelA1 (CMM_0043), pelA2 (CMM_0051), xysA (CMM_1673), xysB (CMM_1674), CMM_2691, CMM_2692, and CMM_2871, which are responsible for cell wall degradation at the later stages of tomato infection by C. michiganensis sensu stricto. The polygalacturonase-encoding locus CMM_2871 was found in tomato-associated strains CFBP 7493 and LMG 26808, as well as the type strains of C. insidiosus (LMG 3663^T), C. nebraskensis (NCPPB 2581^T), and C. tessellarius (ATCC 33566^T). Furthermore, celB (CMM_2443), which is a homolog of the plasmid-born celA gene (24), was present in tomato-associated nonpathogenic strains CFBP 8017, DOAB 609, CFBP 7493, LMG 26808, and CFBP 7494. The four loci CMM_1673, CMM_1674, CMM_2691, and CMM_2692 were present in all tomato-associated strains, except for CMM_2692, which was absent in CASJ009, while the query coverage varied between 30 and 100% and the sequence similarity varied from 75 to 100% among the strains. Surprisingly, the virulence-associated transcriptional regulator genes vatr1 (CMM_2645) and vatr2 (CMM_2969), which regulate C. michiganensis sensu stricto virulence during infection (25), were present in all the strains evaluated in this study regardless of their pathogenicity status (Table 5).

Orthologous gene clusters were determined using the OrthoVenn online service through four-versus-four and five-versus-five designations of the representative strains from different phylogenetic lineages (Fig. 3A to D). Type strains of the five former C. michiganensis sensu lato subspecies shared 2,157 proteins in their genome sequences (Fig. 3A). Although none of the type strains of C. michiganensis sensu stricto and C. nebraskensis showed unique proteins in their sequences, the type strains of C. sepedonicus, C. tessellarius, and C. insidiosus showed 4, 12, and 16 unique proteins among their genome sequences, respectively. Furthermore, when the two phylogenetic neighboring clades of C. michiganensis sensu stricto (i.e., C. michiganensis subsp. californiensis and CFBP 7493) were compared with the type/reference strains of C. michiganensis sensu stricto (LMG 7333^T and NCPPB 382), 7 and 11 unique proteins were detected in the genome sequences of CFBP 7493 and C. michiganensis subsp. californiensis strain CFBP 8216^T, respectively (Fig. 3B). Unique and shared proteins in the type strains of C. michiganensis subsp. californiensis, C. michiganensis subsp. phaseoli, and C. michiganensis subsp. chilensis as well as the atypical peach-colored strain CFBP 8615 are shown in Fig. 3C and D.

FIG 3.

Venn diagrams constructed using the OrthoVenn online service, showing the distribution of shared gene families (orthologous clusters) among different sets of Clavibacter sp. strains.

Plasmids, phages, and bacteriocins.

No integrative plasmid (episome) was detected using the PlasmidFinder online service in the draft genome sequences of bacterial strains investigated in this study, except for LMG 26808, in which two Enterobacteriaceae plasmids, IncL/M(pOXA-48) and IncR, were identified. Surprisingly, sequences homologous to the plasmid-born celA gene were detected in the sequences of the type strains of C. insidiosus (LMG 3663^T) and C. nebraskensis (NCPPB 2581^T), with query coverage of 97% and 76%, respectively, and sequence similarity of 90% and 84% to the sequence of the reference strain NCPPB 382 (Table 5). On the other hand, plasmid profiling detected the two expected plasmids pCM1 (≈27 kb) and pCM2 (≈70 kb) in the tomato-pathogenic strain C. michiganensis sensu stricto ICMP 22049 (data not shown). However, type strains of C. michiganensis subsp. phaseoli and C. michiganensis subsp. chilensis, as well as the two peach-colored strains CFBP 8615 and CFBP 8616, did not carry any detectable plasmid (data not shown).

The PHASTER online service was used to detect prophage sequences within the bacterial genomes. Altogether, five hypothetical prophage groups, i.e., Gordon_Schwabeltier (accession no. NC_031255), Gordon_Smoothie (NC_030696), N15 (NC_001901), P1 (NC_005856), and Phi92 (NC_023693), were detected in the Clavibacter sp. strains investigated in this study (Table 6). Prophages were detected in tomato-associated strains CFBP 8615, CFBP 7491, LMG 26808, and CASJ009, as shown in Table 6. While strain LMG 26808 contained three prophages, only one prophage per strain was detected in strains CFBP 8615, CFBP 7491, and CASJ009. The phage Phi92, which was originally isolated from a pathogenic Escherichia coli strain, was detected in CFBP 8615, LMG 26808, CASJ009, C. sepedonicus (ATCC 33113^T), and C. tessellarius (ATCC 33566^T), while each of the remaining four prophages was detected in only one strain (Table 6).

TABLE 6.

Prophages within the genome sequences of Clavibacter sp. strains detected using the online service PHASTER^a

Taxon	Strain	Region length (kb)	Completeness	No. of proteins	Position (contig: bp)	Phage (GenBank accession no.)	G+C content (%)
Clavibacter sp.	CFBP 8615	6.9	Incomplete	8	38: 3643–10586	Phi92 (NC_023693)	71.28
Clavibacter sp.	CFBP 7491	6.8	Incomplete	14	4: 13543–20391	Gordon_Schwabeltier (NC_031255)	66.29
Clavibacter sp.	LMG 26808	9.9	Incomplete	10	13: 128032–137979	Phi92 (NC_023693)	71.79
		7	Incomplete	11	15: 712–7762	N15 (NC_001901)	46.49
		23.7	Incomplete	10	15: 5546–29310	P1 (NC_005856)	48.19
Clavibacter sp.	CASJ009	7.7	Incomplete	9	1009722–1017492	Phi92 (NC_023693)	71.29
C. insidiosus	LMG 3663T	6.7	Incomplete	16	scaffold2: 5055–11844	Gordon_Smoothie (NC_030696)	66.95
C. sepedonicus	ATCC 33113T	7.1	Incomplete	10	42745–49937	Phi92 (NC_023693)	70.71
		7.7	Incomplete	8	660037–667813	Phi92 (NC_023693)	71.20

^aFive prophage groups, i.e., Gordon_Schwabeltier, Gordon_Smoothie, N15, P1, and Phi92, were detected in strains CFBP 8615, CFBP 7491, LMG 26808, CASJ009, LMG 3663, and ATCC 33113. No prophage was detected within the genome sequences of the strains ATCC 33566T, CF11, CFBP 7493, CFBP 7494, CFBP 8017, CFBP 8019, CFBP 8216T, CFBP 8217T, CFBP 8616, CFBP 8627T, DOAB 609, NCPPB 2581T, NCPPB 382, and PF008T.

In silico screening for bacteriocins and antibiotic peptides revealed distinct differences between the pathogenic and nonpathogenic tomato-associated Clavibacter sp. strains. The lantibiotic Michiganin A was detected in all 12 C. michiganensis sensu stricto strains (data not shown) but was not detected in the nonpathogenic tomato-associated strains or in the pathogenic strains on other plant species. Furthermore, sactipeptides (peptides with cysteine sulfur-to-α-carbon cross-links) were the most common group of bacteriocins among all the Clavibacter sp. strains. Indeed, except for tomato-pathogenic C. michiganensis sensu stricto strains, all the strains which contained bacteriocins had at least one type of sactipeptides (Table 7). Linear azol(in)e-containing peptides (LAPs) were detected in both the pathogenic and nonpathogenic strains, while thiopeptides which are commonly produced by Actinobacteria were found in the two phylogenetically closely related nonpathogenic strains CFBP 7493 and LMG 26808. Enterocin_AS_48, a circular bacteriocin produced by Enterococcus spp., was detected exclusively in strain CF11, which was originally isolated from soil in a tomato-growing greenhouse (23).

TABLE 7.

In silico screening for bacteriocins and antibiotic peptides among the Clavibacter sp. genome sequences analyzed in this study^a

Taxon	Strain	Contig no.	Start codon	End codon	Class
C. michiganensis sensu stricto	NCPPB 382	1	2211104	2233501	63.1; Michiganin A (Lantibiotic)
C. capsici	PF008T	1	2406476	2426476	Sactipeptides
Clavibacter sp.	CFBP 8615	14	7079	27079	LAPs
		91	1570	18430	Sactipeptides
Clavibacter sp.	CFBP 8616	166	4888	15112	Sactipeptides
C. michiganensis subsp. californiensis	CFBP 8216T	201	9178	10822	Sactipeptides
Clavibacter sp.	CFBP 7491	224	8131	11869	Sactipeptides
Clavibacter sp.	CFBP 7493	1	25925	45925	Thiopeptide, LAPs
		226	-8455	11545	Sactipeptides
Clavibacter sp.	CFBP 7494	9	267758	287758	LAPs
		15	89075	109075	Sactipeptides
Clavibacter sp.	CFBP 8019	11	186443	206443	Sactipeptides
		17	190604	210604	LAPs
Clavibacter sp.	LMG 26808	6	43154	63154	Thiopeptide, LAPs
		10	130865	150865	Sactipeptides
Clavibacter sp.	CF11	17	47765	67765	Sactipeptides
		2	247508	267508	LAPs
		13	138056	158368	150.1, Enterocin_AS_48
Clavibacter sp.	DOAB 609	20	4781	24781	Sactipeptides
		30	31400	51400	LAPs
Clavibacter sp.	CFBP 8017	33	39236	59236	LAPs
		46	11546	31546	Sactipeptides
C. michiganensis subsp. phaseoli	CFBP 8627T	212	−9916	10084	Sactipeptides
C. michiganensis subsp. chilensis	CFBP 8217T	NDb	ND	ND	ND
Clavibacter sp.	CASJ009	ND	ND	ND	ND
C. insidiosus	LMG 3663T	1	2292719	2312719	Sactipeptides
C. nebraskensis	NCPPB 2581T	1	2422475	2442475	Sactipeptides
C. sepedonicus	ATCC 33113T	1	2663669	2683669	Sactipeptides
		1	3119744	3139861	294.1, Plantathiazolicin (Plantazolicin)
C. tessellarius	ATCC 33566T	1	1820699	1840699	Sactipeptides

^aThe lantibiotic Michiganin A was detected in all 12 C. michiganensis sensu stricto strains (Fig. 1) but was not found in nonpathogenic tomato-associated strains or in the pathogenic strains on other plant species.

^bND, not detected.

DISCUSSION

In this study, using phylogenetic analyses, comparative genomics, and pathogenicity assays, we provide novel insight into the diversity of Clavibacter sp. strains, with a special focus on tomato-associated members of the genus. Phylogenetic analyses accomplished with ANI and dDDH calculations revealed a higher genetic diversity of Clavibacter sp. strains than has so far been assumed (17). We also aimed to decipher the phylogenetic positions of the three newly described subspecies of C. michiganensis sensu lato (i.e., C. michiganensis subsp. californiensis, C. michiganensis subsp. chilensis, and C. michiganensis subsp. phaseoli). Although our results confirm that these three subspecies are no longer included in C. michiganensis sensu stricto, we still used their original names in this study to avoid confusion. A formal taxonomic study would provide appropriate epithets for these taxa. On the other hand, BLAST-based comparative genomics revealed that several genes (i.e., vatr1, vatr2, xysA, xysB, and srtA) which had previously been identified as pathogenicity determinants were present in all the pathogenic and nonpathogenic tomato-associated strains, indicating further complexities in the functions of these genes (25, 26). Nonpathogenic counterparts of actinobacterial plant pathogens have frequently been isolated from a set of taxonomically diverse plant species which were distant from the main host of the pathogen. For instance, both the pathogenic and nonpathogenic Curtobacterium flaccumfaciens strains phylogenetically closely related to the common bean pathogen C. flaccumfaciens pv. flaccumfaciens were isolated from solanaceous annual crops, i.e., eggplant, pepper, and tomato (2). While the pathogenic and nonpathogenic strains of C. flaccumfaciens are not differentiable using routine molecular techniques, e.g., MLSA (27), all the nonpathogenic members of C. michiganensis sensu lato could be differentiated from the tomato-pathogenic C. michiganensis sensu stricto strains (11).

Nonpathogenic strains of Clavibacter spp. were consistently reported to associate with seeds, transplants, and aerial portions of tomato plants (11, 13, 22, 28). However, until recently, only a few genome sequences from nonpathogenic Clavibacter sp. strains were available (22, 28), limiting our understanding of the putative role of these bacteria on the host plants and their environment. In a preliminary complete genome sequence-based comparative study, Zaluga et al. (28) investigated the genome of tomato-associated nonpathogenic strain LMG 26808 and provided initial insights into the genetic bases of differences between the pathogenic and nonpathogenic members of C. michiganensis sensu lato. However, it has been noted that LMG 26808 is phylogenetically very close to the C. michiganensis sensu stricto clade, leaving a greater portion of nonpathogenic Clavibacter sp. diversity uninvestigated (28). Our results revealed that strain LMG 26808 as well as two other strains, i.e., CFBP 7493 and CFBP 8216^T, are phylogenetically closely related and fall into a monophyletic clade along with the pathogenic members of C. michiganensis sensu stricto (Fig. 1). The genomic contents of the two clades represented by LMG 26808/CFBP 7493 as “hypothetical new species I” and CFBP 8216^T as “hypothetical new species II” varied in the pathogenicity-related genes sbtA, expA, and celB and the locus CMM_2871, while there was no difference between the strains LMG 26808 and CFBP 7493 in the evaluated genomic areas (Table 5). More specifically, the expA gene (CMM_1480), which is responsible for expansin production (29), and a polygalacturonase encoded by the CMM_2871 locus at the final stages of infection were found in the CFBP 7493 and LMG 26808 strains but not in the type strain of C. michiganensis subsp. californiensis (Table 5).

Recently, Li and colleagues (17) reevaluated the taxonomy of C. michiganensis sensu lato and proposed the reclassification of each of the former C. michiganensis subspecies into species status. Since only two genome sequences of C. insidiosus were available at that time (both isolated in the United States) (29), we sequenced two further “Old World” strains (i.e., CFBP 1195 and CFBP 6488, isolated in the United Kingdom and the Czech Republic, respectively) to gain a precise vision of the intraspecies diversity of the alfalfa pathogen (20). Our analyses confirm the existence of C. capsici, C. insidiosus, C. michiganensis sensu stricto, C. nebraskensis, C. sepedonicus, and C. tessellarius as stand-alone species. While the 95% ANI did not solely support the separation of the alfalfa and maize pathogens, the 59.90% dDDH between the type strains of the two taxa, as well as their distinct host plants, could be considered evidence for the separation of C. insidiosus and C. nebraskensis (Tables 2 and 3). Strain CFBP 7494, which was isolated from tomato seeds and causes disease symptoms on wheat plants under greenhouse conditions (22), clustered in a monophyletic clade with the alfalfa-pathogenic strains and still fell into the C. insidiosus species, with 96% ANI and 64.60% dDDH. While the phylogenetic position of strain CFBP 7494 was clarified in these analyses, only further investigations using a larger collection of strains will shed a light on the genetic content, biological characteristics, and taxonomic status of tomato-associated wheat-pathogenic members of Clavibacter spp.

Draft genome sequences of the three new subspecies of C. michiganensis sensu lato revealed their phylogenetic position, highlighting inaccuracy in the nomenclature of C. michiganensis subsp. chilensis and C. michiganensis subsp. phaseoli. Type strains of these two subspecies shared 99% ANI and 87.50% dDDH with one another, indicating a synonymy and orientating toward the proposal of a novel unique new species (Fig. 1 and Table 3). The type strain of C. michiganensis subsp. phaseoli (CFBP 8627^T) was isolated from common bean seeds in Spain and reported to cause bacterial bean leaf-yellowing disease in greenhouse assays (14). However, we could not observe any symptoms on inoculated common bean plants even when three different cultivars were evaluated in the pathogenicity tests (Table 1). This could be attributed to differences in the environmental conditions between the two assays and probably also differences in the susceptibility of common bean cultivars used in the two studies. Further field surveys are needed to decipher potential natural occurrence under the field conditions and the putative frequency and prevalence of the common bean-associated Clavibacter sp. strains.

Comparative genomics revealed that tomato-associated C. michiganensis sensu lato strains are adapted to a nonpathogenic lifestyle, which is reflected by the lack of pathogenicity gene clusters present in the pathogenic members (Table 5; Fig. 2). Although the absence of almost all of the 129-kb chp/tomA region was common among all the nonpathogenic strains, some of the putative virulence factors were present in the nonpathogenic strains, suggesting contributions of these genes in the endophytic lifestyle of the bacteria. An in-depth comparative analysis with newly sequenced Clavibacter sp. genomes allowed us to illustrate a more precise insight underlying genetic contents of these bacteria. For instance, the expA gene was detected in the nonpathogenic strains CFBP 7493 and LMG 26808 as well as the wheat-pathogenic strains CFBP 8017 and DOAB 609 but not in the type strain of C. tessellarius (Table 5). Microbial expansins are found in the genomes of several plant-pathogenic bacteria, and it is assumed that they provide particular advantages to xylem-dwelling phytopathogens (29, 30). Expansin enhances cellulose breakdown by cellulase enzymes in the later stages of pathogen invasion (31). These observations correlate with the initial assumptions that nonpathogenic Clavibacter sp. strains must have lost or never contained prominent virulence determinants (e.g., the 129-kb chp/tomA region) responsible for disease induction in tomato plants. With the availability of genome sequences covering a broader diversity of nonpathogenic Clavibacter sp. strains, one would assume that the gene flow and evolutionary pathways of pathogenicity determinants in the genus are similar to those that have previously been estimated for plant-pathogenic xanthomonads (4).

In conclusion, our results obtained from the analyses of 40 genome sequences provide a comprehensive insight into the genetic diversity of Clavibacter spp. and confirm the recent taxonomic revision of the genus. However, phylogenetic analyses suggest that the recently described subspecies C. michiganensis subsp. chilensis and C. michiganensis subsp. phaseoli should be classified as members of the same novel species (13, 14). Taking together all the phylogenetic, genomic, and pathogenicity data, nine hypothetical novel species could be identified within Clavibacter spp., seven of which (i.e., hypothetical new species I, II, III, IV, V, VIII, and IX as shown in Fig. 1) were isolated from asymptomatic tomato tissues or seed lots. These findings raise the question whether the current taxonomy of tomato-associated Clavibacter sp. strains is technically applicable for quarantine purposes and emphasize at the same time the need for more detailed taxonomic investigations among the phylogenetically diverse tomato-associated Clavibacter sp. strains. Indeed, the only pathogenic lineage of tomato-associated strains is C. michiganensis sensu stricto, while the seven nonpathogenic lineages need to be designated novel formal taxa. This would help the plant pathology agencies and tomato seed industry inspectors to specifically target the enemy and neglect the nonpathogenic lineages. Only a formal taxonomic study would address this issue, with delineation of appropriate epithets and species descriptions for these new taxa. On the other hand, the nine pathogenicity determinant genes (Table 5) would be appropriate targets for the development of novel genome-informed detection methods for differentiation of tomato-pathogenic and nonpathogenic strains.

MATERIALS AND METHODS

Bacterial strains and genome sequences.

Draft genome sequences of 10 Clavibacter sp. strains (Table 2) were prepared using the shotgun genome sequencing facility of the Illumina HiSeq X platform. The culture media, bacterial growth conditions, genomic DNA preparation, sequencing procedure, and genome annotation were described previously (20). In this framework, we investigated the type strains of C. michiganensis subsp. californiensis (CFBP 8216^T), C. michiganensis subsp. chilensis (CFBP 8217^T), and C. michiganensis subsp. phaseoli (CFBP 8627^T), two C. insidiosus strains (CFBP 1195 and CFBP 6488), one C. nebraskensis strain (CFBP 7577), and the nonpathogenic peach-colored (i.e., CFBP 8615 and CFBP 8616) and yellow-pigmented (i.e., CFBP 7491 and CFBP 7493) strains. Furthermore, all the publicly available genome sequences of Clavibacter spp., until April 2019, were retrieved from the NCBI GenBank database and included in the phylogenetic analysis and comparative genomics. Table 2 describes the 40 Clavibacter sp. strains used in this study, their origins of isolation, and their pathogenicity features.

Pathogenicity tests and host range.

Due to the close phylogenetic relationships between the type strains of C. michiganensis subsp. phaseoli (CFBP 8627^T) and C. michiganensis subsp. chilensis (CFBP 8217^T), these two strains, as well as the atypical peach-colored strains recently isolated from tomato (16), were subjected to the pathogenicity tests and host range assays under greenhouse conditions. Pathogenicity tests were performed on bell pepper (cv. Sereno), chili pepper (cv. Aziz), common bean (cv. Red kidney, Pinto, and Navy), cowpea (Vigna unguiculata cv. Partow), mung bean (Vigna radiata cv. Mashhad), and tomato (cv. Sunseed 6189) plants. The plant growth conditions, inoculation procedure, and incubation environment were the same as detailed previously (32, 33). Inoculated plants were periodically monitored for the appearance of disease symptoms up to 30 days postinoculation. Positive- and negative-control plants were treated in the same manner using the standard strain of C. michiganensis sensu stricto (ICMP 22049, isolated from a symptomatic tomato plant in Iran in 2015 [16]) and sterile distilled water, respectively. Koch’s postulates were satisfied by reisolating the inoculated strains on yeast extract-peptone-glucose agar (YPGA) medium from all inoculated plants. Confirmation of the identity of the reisolated bacteria was made by determining Gram reaction and colony characteristics on yeast extract-dextrose-calcium carbonate (YDC) agar medium as well as by using the genus-specific primer pair CMR16F1/CMR16R1 (34) as described previously (16). The pathogenicity tests were conducted twice.

Phylogenetic analyses.

Average nucleotide identity (ANI) was calculated among all the Clavibacter sp. genome sequences included in this study. The ANI was estimated using both one-versus-one and all-versus-all strategies via different algorithms, i.e., JSpeciesWS (http://jspecies.ribohost.com/jspeciesws/) (35), ANI calculator (http://enve-omics.ce.gatech.edu/g-matrix/) (36), and OrthoANIu (https://www.ezbiocloud.net/tools/orthoaniu) (37). An ANI-based neighbor-joining phylogenetic tree was constructed using the ANI calculator online service, and the genome sequence of Leifsonia xyli subsp. cynodontis (DSM 46306; accession no. NC_022438.1) was used as an outgroup in the tree. Additionally, the Genome-to-Genome Distance Calculator online service (http://ggdc.dsmz.de/distcalc2.php) was used to calculate digital DNA-DNA hybridization (dDDH) values, which infer the genome-to-genome distances between pairs of genomes based on the Genome BLAST Distance Phylogeny (38). A combination of ANI and dDDH indices was used to designate a taxonomic status to a given phylogenetic clade, where the “new species” status was assigned to a clade only when both ANI and dDDH values were below the accepted threshold (≤95% and ≤70% for ANI and dDDH, respectively) (21).

Comparative genomics.

Twenty strains representing the entire genetic diversity of Clavibacter spp. based on the ANI/dDDH data, host of isolation, and pathogenicity characteristics were subjected to the comparative genomics analyses. Type strains of all the C. michiganensis sensu lato species/subspecies, as well as all the individual strains sharing ≤95% and ≤70% ANI and dDDH values, respectively, with the other taxa were selected for comparative genomics analyses. Genome length (bp), G+C content (%), and total numbers of protein-coding sequences (CDS), RNA genes, and pseudogenes were determined for all the genomes.

The online annotating service RAST (Rapid Annotations using Subsystems Technology) (http://rast.nmpdr.org/) (39) was used for fully automated annotation of the bacterial genomes, and the obtained information was used to reconstruct metabolic networks and subsystems. A subsystem is a set of functional roles that the annotator considers related categories. Subsystems represent a collection of functionally related protein families that make up a metabolic pathway (e.g., iron acquisition and metabolism), a complex (e.g., the ribosome), or a class of proteins (e.g., bacteriocins) (40). Subsequently, the genomes were transferred to the comparative environment of the SEED-Viewer (http://www.theseed.org/wiki/Main_Page) (41) for comparative genomics analyses. The SEED-Viewer was used for the identification of protein-encoding sequences (CDS), assigning functions to the genes, and prediction of represented gene clusters in the genomes. The distribution of the genes among various clusters and specific protein-encoding genes within each cluster were estimated using the same service. Furthermore, BLASTn/BLASTp-based investigation was performed to decipher whether the pathogenicity determinant genes/clusters are present in the genomes (26). Using the complete genome of C. michiganensis sensu stricto NCPPB 382, one-versus-one BLASTn/BLASTp searches were done against the sequences of the pathogenicity island (a 129-kb low-G+C region which includes chp and tomA clusters) as well as several individual genes proposed to have effective contributions to the virulence of C. michiganensis sensu stricto (22, 25, 26, 28, 42, 43). Proteins with amino acid sequence similarities higher than 50% and with a query coverage higher than 70% were considered homologs (28).

We also screened the genome sequences for the presence of hypothetical bacteriocin-encoding genes/clusters using the web-based tool BAGEL4 (http://bagel4.molgenrug.nl/) (44). BAGEL4 combines direct mining for the structural genes with indirect mining for bacteriocin-associated genes. Furthermore, the online service PlasmidFinder 2.0 (https://cge.cbs.dtu.dk/services/PlasmidFinder/) (45) was used for the screening of all the genomic sequences for the presence of integrative plasmids/episomes. Identification and annotation of prophage sequences within bacterial genomes were performed using the online service PHASTER (PHAge Search Tool Enhanced Release) (http://phaster.ca/) (46). Given the fact that identification of overlaps among the orthologous clusters can enable us to elucidate the function and evolution of proteins across multiple species, genome-wide comparisons and visualization of orthologous clusters were performed using the online service OrthoVenn (http://www.bioinfogenome.net/OrthoVenn/) (47). The analyses were conducted on the “bacteria” section of the platform using default settings (E value, 1e−5; inflation value, 1.5). Regarding the numeric limitation of OrthoVenn in the handling of bacterial genomes (up to six genomes per run), different series of the strains were evaluated using the same parameters.

Plasmid profiling.

In order to further investigate the genetic contents of nonpathogenic tomato-associated Clavibacter sp. strains, we evaluated the plasmid profiles of the type strains of C. michiganensis subsp. chilensis (CFBP 8217^T) and C. michiganensis subsp. phaseoli (CFBP 8627^T) as well as the two peach-colored strains CFBP 8615 and CFBP 8616 isolated from tomato. Plasmids were isolated according to the procedure described by Kotchoni et al. (48) with minor modifications. The tomato-pathogenic C. michiganensis sensu stricto strain ICMP 22049 was used as positive control (49). Bacterial strains were grown in 50 ml Luria-Bertani (LB) medium on a shaker at 150 rpm and 27°C for 48 h. Bacterial cells were harvested by centrifugation at 13,000 × g for 1 min at room temperature, and the pellet was resuspended in 200 μl of “solution I” of the protocol of Kotchoni et al. (48), mixed well, and incubated at 37°C for 20 min. Subsequently, 400 μl of freshly prepared “solution II” was added to the microtubes and mixed well by inverting gently four to six times to avoid breaking the plasmid(s), and then 200 μl of “solution III” was immediately added, mixed very gently, and incubated at 4°C for 15 min without any intervention. The mixture was centrifuged at 10,000 × g for 5 min, the supernatant was transferred to a new microtube, and a 0.6 volume of isopropanol was added to the supernatant, mixed gently by inverting four to six times, kept at room temperature for 10 min, and then centrifuged at 10,000 × g for 5 min, and the supernatant was discarded. The pellet containing precipitated plasmid DNA was washed with 400 μl of 70% (vol/vol) ethanol and centrifuged at 10,000 × g for 3 min at room temperature. The supernatant was removed, and the pellet was air dried. Finally, plasmid DNA was resuspended in 50 μl sterile distilled water containing 10 mg/ml RNase A. The presence of plasmids was analyzed on a 0.6% agarose gel as described previously (27).

Data availability.

The data sets generated for this study were previously published (20) and can be found in the NCBI GenBank Database (see Table 2).