The genomic DNA (gDNA) of C. michiganensis LMG7333^T was extracted using the solution-based HiGene Genomic DNA Prep kit (BIOFACT, Daejeon, Korea), according to the manufacturer’s instructions. Using extracted gDNA, whole genome sequencing of LMG7333^T was conducted using combined methods of two sequencing platforms, PacBio RSII and Illumina HiSeqXten (Macrogen, Seoul, Korea), for genome sequence completion. The reads obtained from PacBio RSII were assembled by Hierarchical Genome Assembly Process (Chin et al., 2013). HiseqXten reads were also used to correct assembly by Pilon v1.21 (Walker et al., 2014). Based on the assembled sequences, gene prediction and annotation were conducted by Prokka v1.13 (Seemann, 2014). Then, InterProScan v5.30–69.0 (Jones et al., 2014) and psiblast v2.4.0 (Camacho et al., 2009) with EggNOG DB v4.5 (Huerta-Cepas et al., 2016) and the Clustering Orthologous Groups (COG) database v2014 (Tatusov et al., 2000) were used to confirm the predicted proteins. Circos program v0.69.3 (Krzywinski et al., 2009) was used for drawing circular maps of the genome.

The completed whole genome sequence of C. michiganensis LMG7333^T type strain was first compared by aligning the EasyFig program (Sullivan et al., 2011) and NCBI blast program with those of C. michiganensis NCPPB382 (GenBank accession #: AM711865, AM711866, and AM711867) (Gartemann et al., 2008) and UF1 (GenBank accession #: CP033724, CP033725, and CP033726). Then, it was also compared with completed genome sequences of nine other Clavibacter species, including several type strains such as C. sepedonicus ATCC33113^T (GenBank accession #: AM849034) (Bentley et al., 2008), C. capsici PF008^T (GenBank accession #: CP012573, CP012574, and CP012575) (Bae et al., 2015), C. insidiosus R1-1 (GenBank accession #: CP011043, CP011044, CP011045, and CP011046) (Lu et al., 2015), C. nebraskensis NCPPB2581^T (GenBank accession #: HE614873), C. insidiosus LMG3663^T (GenBank accession #: MZMO01000001, MZMO01000002, and MZMO01000003), C. tessellarius ATCC33566^T (GenBank accession #: MZMQ01000001 and MZMQ01000002) (Li and Yuan, 2017; Li et al., 2018), C. phaseoli CFBP8627^T (GenBank accession #: QWGV00000000), C. michiganensis subsp. chilensis CFBP8217^T (GenBank accession #: QWGS01000000), and C. californiensis CFBP8216^T (GenBank accession #: QWEE01000000) (Méndez et al., 2020). The genome sequence of Leifsonia xyli subsp. cydontis DSM46306^T (GenBank accession #: CP006734) (Monteiro-Vitorello et al., 2013), which belongs to the same family as genus Clavibacter, was used as an out group. To understand the phylogenetic relationships among the analyzed strains, OrthoANI values were calculated, and UPGMA dendrograms were generated using the Orthologous ANI Tool of ChunLab (Lee et al., 2016). Moreover, the pan-genome analysis was conducted, and results were shown through a Venn diagram (Bardou et al., 2014).

C. michiganensis strain LMG7333^T was grown at 26°C in KB medium (20 g proteose peptone No. 3, 1.5 g K₂HPO₄, 6 mL 1 M MgSO₄, 16 mL 50% glycerol, and 15 g agar in the total volume of 1 L) for 3 days and inoculated into tomato plants using three different inoculation methods, which are leaf-clipping inoculation, stem injection, and root-dipping inoculation. Ten-day-old and 3- and 4-week-old tomato plants (Solanum lycopersicum L. cv. “Betatini”) were used for the root-dipping inoculation, leaf-clipping inoculation, and stem injection, respectively. For the leaf-clipping method, the bacterial inoculum was prepared and adjusted to a concentration of 2 × 10⁸ CFU/mL (OD₆₀₀ = 0.4). Sterilized scissors were dipped into this inoculum, and then half of the leaves were cut using scissors (Hwang et al., 2020b). Three leaves located in different directions of each plant and at least five plants were used for each treatment. For stem injection, a bacterial concentration was adjusted to 10⁹ CFU/mL (OD₆₀₀ = 2.0), and the 20 μL bacterial suspension was injected into tomato stems above 2 cm from the cotyledon after making a hole with a needle. At least four plants were used for each treatment. For the root-dipping inoculation method, 10-day-old seedlings were pulled out from soil, and their roots were dipped in 1 mL of bacterial suspension (10⁹ CFU/mL) for 30 min. The seedlings were then replanted into pots. Five plants were used for each treatment. C. michiganensis LMG7333^T was resuspended in sterilized 10 mM MgCl₂ buffer, and this buffer was used as a negative control in all inoculation methods.

After inoculation with C. michiganensis LMG7333^T by stem injection, tomato stems were collected at 0, 7, 14, 21, and 28 days after inoculation (dai). For sample collection, 3 cm lengths of inoculated stems were harvested from four different sites, which are the inoculated site and 10, 15, and 20 cm above from the inoculated site. Then, 2 mL of 10 mM MgCl₂ was added to ground samples and the supernatant was dotted onto KB media after 10-fold serial dilution. Three plants were used for each repetition, and experiments were repeated at least three times.

For statistical analysis of bacterial growth of Clavibacter michiganensis LMG7333^T in planta, Duncan’s multiple range test was conducted (p < 0.05).

The complete genome sequence of C. michiganensis LMG7333^T was obtained through the combination of PacBio RSII and Illumina HiSeqXten sequencing methods, and the genome coverage was 320X. From HiSeqXten, we got 9,154,590 reads, and after assembly, the N50 value was 3,271,838. Finally, we obtained three contigs, one circular chromosome (GenBank accession #: CP080437), and two plasmids, pCM1 and pCM2 (GenBank accession #: CP080438 and CP080439). Their genome sizes and G + C contents are 3.27 Mb with 72.74% G + C content, 32 kb with 67.75% G + C content, and 73 kb with 66.99% G + C content, respectively (Figures 1A,B). This genomic composition of C. michiganensis LMG7333^T is similar to those of NCPPB382 and UF1 strains, of which the complete genome sequences are available. However, their sizes are different from one another, particularly, plasmid sizes (Figure 1B). Genome annotation revealed 3,033 protein-coding sequences (CDSs) as well as 45 tRNAs and 6 rRNAs in the chromosome and 32 and 67 CDSs in pCM1 and pCM2, respectively (Figure 1B). The annotated proteins were categorized into 21 COG. Approximately 40% of CDSs were categorized into two groups for the function unknown (S; 891 CDSs) and the general function only (R; 349 CDSs), which are poorly characterized, in the COG database (Figure 2). COG groups for carbohydrate transport and metabolism (G; 255 CDSs); transcription (K; 205 CDSs); amino acid transport and metabolism (E; 198 CDSs); inorganic ion transport and metabolism (P; 179 CDSs); translation, ribosomal structure, and biogenesis (J; 146 CDSs); and cell wall/membrane/envelope biogenesis (M; 137 CDSs) were found to be dominant groups. In contrast, only two CDSs were categorized into cell motility groups (N) (Figure 2). Some CDSs were not grouped into specific COG or grouped to two or more COG groups. There are no CDSs for flagella formation, which is consistent with a previous study reporting that C. michiganensis LMG7333^T did not have flagella (Oh et al., 2016).

As shown in Figure 1B, three C. michiganensis strains differed in the sizes of a chromosome and two plasmids. Here, three genome sequences were compared to examine how different they were. First, the sequence identity of chromosomal DNA among the three strains was almost 100% identical although the query cover between LMG7333^T and NCPPB382 strains was 97% (Figure 3). As with other C. michiganensis strains, C. michiganensis LMG7333^T has about 120 kb of pathogenicity island (PAI) region from 2,334,274 to 2,464,147 bp on its chromosome (Figure 3 and Supplementary Figure 1). In the chp region of PAI, the putative virulence genes, such as the chp, ppa, and pel gene families, were intact like those in the NCPPB382 strain. Moreover, the tomA gene was found in the tomA region of PAI (Supplementary Figure 1). Two other important virulence genes, celA, and pat-1 genes are located from 7,321 to 9,561 bp in pCM1 and from 7,833 to 8,675 bp in pCM2 of C. michiganensis LMG7333^T (Figure 3). The presence of an intact PAI region, celA and pat-1 genes, suggests that C. michiganensis LMG7333^T can also infect tomato plants to cause typical disease symptoms like the NCPPB382 stain.

The size and sequence query cover of the pCM1 plasmid of LMG7333^T was the most different from those of two other strains although the sequence identity in the matched regions was almost 100% identical. Its size was the biggest among the three strains, and the matched regions were only 63% with UF1 pCM1 and only 39% with NCPPB382 pCM1 (Figure 3). Among unmatched regions in pCM1 of LMG7333^T, some regions were unique to this strain, whereas others were only matched with that of NCPPB382. Most of the genes in these unmatched regions are annotated as hypothetical proteins, indicating that the origin and functions of these genes are unknown. In the public database, C. michiganensis NCPPB382 pCM1 has genes for cellulase (celA), portioning protein (parA and parB), and replication origin (repA) and several hypothetical proteins. When the sequence of pCM1 was compared with these annotated genes of pCM1 in C. michiganensis NCPPB382, we found that C. michiganensis LMG7333^T pCM1 also contained these annotated genes except the traG gene encoding putative conjugal transfer protein (Załuga et al., 2014). Gene annotation also revealed that the espI gene, which is a component of the type VII ESX-1 secretion system (Zhang et al., 2014), was found in the pCM1 of LMG7333^T but not NCPPB382, whereas both pCM2 plasmids have the espI homologous gene.

The size and sequence of LMG7333^T pCM2 plasmid were almost 100% identical to that of the UF1 strain, whereas the matched regions with NCPPB382 pCM2 were only approximately 37 kb of 73 kb, which is only 55% of the total plasmid size with 91–99% sequence identity (Figure 3). In the matched region of pCM2 plasmids, all three C. michiganensis strains possess pat-1 genes, encoding a putative serine protease, which was known as a plasmid-borne virulence factor of C. michiganensis (Burger et al., 2005). In the case of C. michiganensis NCPPB382, there are more pat-1 paralogs, phpA and phpB, in pCM2. However, C. michiganensis LMG7333^T lacks these two genes. The unmatched region covered by 45% of the total plasmid size was unique to each strain.

Overall, these results indicate that, although three C. michiganensis strains have similar genome size and major virulence factors, like PAI, celA, and pat-1, in their genomes, the plasmid sequences are varied, and both pCM1 and pCM2 of LMG7333^T contain unique DNA sequences mostly annotated as hypothetical proteins. Moreover, genome comparison among three C. michiganensis strains using complete genome sequences indicates that C. michiganensis LMG7333^T genome sequence can be used as another reference genome of C. michiganensis.

Based on genomic information, average nucleotide identity (ANI) values were calculated among C. michiganensis LMG7333^T and nine other Clavibacter bacteria, including type strains of other Clavibacter species and Leifsonia xyli subsp. cydontis as an outgroup to measure the genetic distance among genomes. Based on ANI data, three C. michiganensis strains were clustered into the same clade with more than 99% ANI value, and non-pathogenic Clavibacter californiensis was the closest to C. michiganensis with almost 95% ANI value (Figure 4 and Supplementary Table 1). There was the second clade, including C. insidiosus, C. nebraskensis, C. phaseoli, and C. michiganensis subsp. chilensis, which is close to the clade of C. michiganensis. Type strains of C. capsici PF008^T, C. sepedonicus ATCC33113^T, and C. tessellarius ATCC33566^T are phylogenetically distinct from the C. michiganensis clade.

Using genome sequences of pathogenic Clavibacter strains, pan-genome analysis was also conducted. Three C. michiganensis strains were first compared, and they have 2,959 common genes. C. michiganensis LMG7333^T only has seven unique genes in its genome. Moreover, it has 8 and 84 genes overlapped with C. michiganensis NCPPB382 and UF1, respectively (Figure 5A). Among the seven unique genes in C. michiganensis LMG7333^T, only one gene was annotated as RNA helicase, and the other six genes were annotated as hypothetical proteins. The specific genes in other two C. michiganensis strains were also annotated as hypothetical proteins. C. michiganensis LMG7333^T has 91 unique genes compared with C. michiganensis NCPPB382. Among these genes, 50 genes are located in the chromosome of the type strain, and the other 41 genes exist in two plasmids, 17 and 24 genes in pCM1 and pCM2, respectively. Among these 91 genes, only 17 genes were annotated (Supplementary Table 2).

BLAST analysis with 91 genes unique to C. michiganensis LMG7333^T compared with C. michiganensis NCPPB382 was conducted to identify the possible origin of these genes. Based on the analysis results, these 91 genes exhibited significant homology with diverse organisms, including other Clavibacter species (Supplementary Table 2). One gene exists in only C. michiganensis LMG7333^T, and nine genes are found in only two C. michiganensis strains, LMG7333^T and UF1. Twenty-seven genes were shared with other C. michiganensis strains, MSF322 and VL527 (Méndez et al., 2020). Twenty-two genes are shared with other Clavibacter bacteria, such as C. insidiosus (10), C. capsici (7), C. nebraskensis (4), and C. sepedonicus (1). The remaining 32 genes are shared with diverse organisms, 11 families, and 18 genera. Most of these genes are shared with bacteria, including four families, Actinomycetaceae (1), Cellulomonadaceae (1), Microbacteriaceae (18), and Micrococcaceae (2), which belong to the phylum Actinobacteria. In particular, the 18 genes shared with the Microbacteriaceae family, including Clavibacter genus, were shared with six genera, Microbacterium (5), Cryobacterium (1), Curtobacterium (8), Leifsonia (2), Leucobacter (1), and Rathayibacter (1). Finally, among 91 C. michiganensis LMG7333^T genes, 77 genes are found only in the Microbacteriaceae family. These results indicate that the C. michiganensis LMG7333^T might obtain most foreign genes from the same bacterial family.

The C. michiganensis LMG7333^T genome was also compared with those of four other Clavibacter type strains. Type strains of five Clavibacter species have 2,345 core genes. Interestingly, each Clavibacter has 138–316 specific genes in their genomes (Figure 5B). Most of these genes were annotated as hypothetical proteins and categorized mostly into unknown function genes, carbohydrate transport, and metabolism as well as transcription proteins in the COG database (Supplementary Table 3). When pan-genome analysis was conducted, duplicated genes were regarded as one gene. Therefore, the total CDS number (Figure 1B) and CDS number of pan-genome analysis (Figure 5B) were slightly different.

In our previous study, C. michiganensis LMG7333^T and its mutant strain were used to investigate the gene function involved in virulence (Hwang et al., 2019). For the virulence assay or mutant screening, the root-dipping method was the best way to observe typical wilting symptoms within a short period. Here, we examined the virulence of C. michiganensis LMG7333^T on tomato plants in various ways. Tomato leaves infected with C. michiganensis LMG7333^T using the leaf-clipping method exhibited wilting symptoms from the cut part of leaves (Figure 6A). As bacteria are directly inoculated into the plants, it was successfully spread along the leaf veins and induced wilting symptoms on the inoculated leaves. Eventually, this bacterium spread throughout the plant and caused typical wilting symptoms even on non-inoculated leaves and branches (Figure 6A). In contrast, no disease symptoms and no physiological disorders were found in buffer-treated tomato plants.

Stem inoculation after artificial wounding, which is a conventional method for studying C. michiganensis, induced large canker lesions at the site of inoculation (Figure 6A). The inoculated bacteria were gradually spread along the xylem toward the tip, and approximately from 21 dai, leaves on the tomato tip exhibited wilting symptoms (Figure 6B). Movement of C. michiganensis LMG7333^T within the stem was confirmed by bacterial growth in planta. From 7 dai, bacteria were detected at 10 cm above the inoculation site, and from 14 dai, it was isolated at the topper site of the stem, indicating that C. michiganensis LMG7333^T could move and colonize in areas higher than the inoculated site (Figure 6B). Altogether, typical bacterial cankers at the site of injection as well as wilting symptoms at the branch and leaves were observed in stem-inoculated tomato plants.

Disease symptoms through this bacterial infection were also caused by the root-dipping methods (Figure 6A). Although this method has not been widely used, disease symptoms were shown within 10 days by the root-dipping method in 8–10-day-old tomato plants. Because this pathogen can enter the tomato plant through its roots during natural infection, the virulence or pathogenicity assay in tomato seedlings by root infection may be more significant than the stem inoculation (Figure 6A). The observation that C. michiganensis LMG7333^T infection in tomato plants resulted in typical disease symptoms by three different inoculation methods indicates that this pathogen with the completed whole genome sequence might be valuable as a standard strain for C. michiganensis study and its interaction with the plants.