Complete Genome Sequencing Provides Novel Insight Into the Virulence Repertories and Phylogenetic Position of Dry Beans Pathogen Curtobacterium flaccumfaciens pv. flaccumfaciens

Bacterial wilt of dry beans (family Fabaceae) caused by a Gram-positive actinobacterial agent Curtobacterium flaccumfaciens pv. flaccumfaciens is one of the most important biotic constraints affecting edible legumes’ production around the globe (Harveson et al. 2015). C. flaccumfaciens pv. flaccumfaciens is included in the A2 (high risk) list of quarantine pathogens by the European and Mediterranean Plant Protection Organization; hence, it is under strict quarantine control and zero tolerance in several countries (EPPO 2011). C. flaccumfaciens pv. flaccumfaciens is the only member of six pathovars within the species C. flaccumfaciens that is reported to have wide geographic distribution and causes economic yield losses on the respective host plants (Evtushenko and Takeuchi 2006). Host range of the pathogen includes common bean (Phaseolus vulgaris), cowpea (Vigna unguiculata), mung bean (Vigna radiata), soybean (Glycine max), and a number of wild leguminous plants (Huang et al. 2009; Osdaghi et al. 2015a). C. flaccumfaciens pv. flaccumfaciens is well known among plant pathogenic bacteria for its multicolored colony variants possessing yellow, orange, pink, purple, and red colonies on culture media (Osdaghi et al. 2016), while yellow-pigmented strains are more prevalent and aggressive compared with the other variants (Osdaghi et al. 2020a). Disease symptoms include interveinal flaccidity and chlorosis on leaflets leading to tissue necrosis on leaves surrounded by chlorotic margins. Subsequent symptoms are leaf wilting during periods of moisture stress, overall wilting and entire plant death in the case of severe infections and favorite environmental conditions (Harveson et al. 2015). Seed coat discoloration into yellow, orange, or purple is a common symptom in white-seeded common bean cultivars (Harveson 2013).

Bacterial wilt was first observed and described in South Dakota (United States) in 1922 (Hedges 1922). Subsequently, the disease was reported in several Midwestern and Northern states of the United States, as well as Canada and Mexico until the mid-20th century. During the past decades, the pathogen was reported in Australia, Europe, Asia, and Africa leading to substantial economic losses on different dry bean species (EPPO 2011; Osdaghi et al. 2020a). Regarding its seedborne nature, the pathogen is capable of distributing in long distances via infected seed lots, being established in the areas with no history of the disease (EPPO 2011). Since the use of conventional pesticides and chemicals have limited effect on the control of bacterial wilt, use of pathogen-free high-quality seed lots, crop sanitary, and resistant cultivars play a pivotal role in the management of the disease (Harveson et al. 2015).

Despite the economic importance of the bacterial wilt disease in dry beans industry, the causal agent has not so far been investigated for its molecular characteristics and virulence repertories. Indeed, among the Gram-positive bacterial plant pathogens, C. flaccumfaciens pv. flaccumfaciens is the least studied member in terms of genomic features and pathogenicity determinants (Osdaghi et al. 2020b; Thapa et al. 2019). This could be due in part to the fact that molecular manipulation technologies and transformation protocols (e.g., plasmid vectors required for cloning and mutagenesis) have not yet been developed for the bacterium, while all these technologies are available for the other actinobacterial plant pathogens, e.g., the tomato pathogen Clavibacter michiganensis (Meletzus and Eichenlaub 1991). It has been shown that pathogenicity of Clavibacter michiganensis is attributed to a low G+C% content genomic island (GI) harboring virulence-inducing loci. Most of the Gram-positive plant pathogenic bacteria produce lytic enzymes, plasmids, as well as toxic compounds to trigger disease on their host plants (Gartemann et al. 2008). However, none of these features has so far been determined to be involved in the pathogenicity of C. flaccumfaciens pv. flaccumfaciens (Osdaghi et al. 2020a).

By the beginning of the genomic era, high throughput complete genome sequencing initiated a substantial advancement in the understanding of molecular mechanisms underlying plant colonization, pathogenicity, and survival mechanisms of phytopathogenic actinobacteria (Thapa et al. 2019). In 2004, complete genome sequence of sugarcane pathogen Leifsonia xyli subsp. xyli has provided the early perceptions from the genomic repertories of Gram-positive phytopathogens (Monteiro-Vitorello et al. 2004). However, the main expansion in the breadth of knowledge from these bacteria was made by the whole genome sequencing of tomato and potato pathogens Clavibacter michiganensis and Clavibacter sepedonicus, respectively (Gartemann et al. 2008, Bentley et al. 2008). During the subsequent years, whole genome sequences of nearly all plant pathogenic members of the family Microbacteriaceae became available except for the pathovars of C. flaccumfaciens. So far, all the genome sequences assigned as Curtobacterium spp. or C. flaccumfaciens in the public databases belong to nonpathogenic members of the genus, lacking key information about the pathogenicity mechanisms of such an omnivorous cosmopolite group of bacteria (Osdaghi et al. 2018a). Hence, complete genome sequence of the bacterial wilt pathogen will integrate our understanding from the ecology and epidemiology of the pathogen into genomic characteristics and virulence repertories of the bacterium, as well as putative molecular mechanisms underlying disease development in the host plants.

The main purposes of the present study were to (i) obtain the complete genome sequence of a highly virulent bacteriocin-producing strain of C. flaccumfaciens pv. flaccumfaciens, (ii) determine genomic features of the pathogen to provide an insight into its virulence and survival mechanisms, and (iii) evaluate whether or not the key phenotypic features could be traced within the genome of the pathogen. With this aim, we have sequenced the whole genomic DNA of the strain P990 and analyzed the resulting data using comparative genomics against the nonpathogenic counterparts of C. flaccumfaciens as well as previously reported genome sequences of plant pathogenic actinobacteria.

MATERIALS AND METHODS

Bacterial strain, culture media, and DNA extraction.

Yellow-pigmented C. flaccumfaciens pv. flaccumfaciens strain P990 (ICMP 22053), which was isolated from the phyllosphere of asymptomatic bell pepper (Capsicum annuum) in Iran in 2014, was used for complete genome sequencing (Osdaghi et al. 2018a). The strain P990 was shown to have fluidal colonies (means that the colony flowed when culture plates were inclined at 45°) on yeast extract-dextrose-calcium carbonate (YDC) agar medium and capable of producing bacteriocin against other C. flaccumfaciens pv. flaccumfaciens strains on agar-based culture media (Osdaghi et al. 2018b). The bacterial strain was restreaked on yeast-extract peptone glucose agar (YPGA) medium, resuspended in sterile distilled water (SDW) and stored at 4°C for further use. For the long-term storage, the strain was maintained in 15% glycerol at −70°C. Whole genomic DNA extraction was carried out using Expin Combo GP (GeneAll, Tic Tech Centre, Singapore) DNA extraction kit from the 24 h culture in lysogeny broth (LB) medium as recommended by the manufacturer. The quality and quantity of the DNA were spectrophotometrically evaluated and adjusted to 1,500 ng/µl using NanoDrop ND-100 (NanoDrop Technologies, Waltham, MA, U.S.A.) as described previously (Khojasteh et al. 2020). In order to ensure the identity of the bacterial strain, the purified DNA was tested using the C. flaccumfaciens pv. flaccumfaciens-specific PCR primer pair CffFOR2/CffREV4 (Tegli et al. 2002) using the procedure described previously (Osdaghi and Lak 2015).

Pathogenicity and host range assays.

The strain P990 was subjected to pathogenicity and host range assays in greenhouse conditions using the procedure described previously (Osdaghi et al. 2015b). Pathogenicity tests were performed on bell pepper (cultivar Sereno), chili pepper (cultivar Aziz), common bean (cultivars Red Kidney, Pinto, and Navy), cowpea (cultivar Partow), mung bean (cultivar Mashhad), soybean (cultivar Katool), and tomato (cultivar Sunseed 6189) plants. Plant growth conditions, inoculation procedure, and incubation environment were the same as detailed previously (Sedighian et al. 2020). In brief, the plants were inoculated at 5 to 15 days postemergence depending on the stage of growth of the species. Inoculation was made by inserting a sterile dissecting needle dipped into a freshly prepared bacterial suspension (1 × 10⁸ CFU/ml) through the first internode of each plant (Ansari et al. 2019). Inoculated plants were periodically monitored for the appearance of disease symptoms up to 30 days postinoculation (dpi). Positive and negative control plants were treated in the same manner using the type strain of C. flaccumfaciens pv. flaccumfaciens (ICMP 2584^T; isolated from common bean in Hungary in 1957) and SDW, respectively. Koch’s postulates were accomplished by reisolating the inoculated strain on YPGA medium from all inoculated plants showing wilt symptoms. Confirmation of the identity of the reisolated bacteria was made by determining Gram reaction and colony characteristics on YDC agar medium, as well as using the pathovar-specific primer pair CffFOR2/CffREV4 (Tegli et al. 2002).

Furthermore, a cell-free culture filtrate of the strain P990 was evaluated for its effect on common bean seedlings to determine whether any toxic compound is attributed to the pathogenicity of the bacterium. With this aim, the strain P990 was grown in liquid LB medium for 24 h, centrifuged in 11,000 × g for 10 min to remove the bacterial cells, and the supernatant was sterilized by tyndallization through three successive days. In order to approve the sterility of the final substrate, 20 µl of the culture filtrate was spread onto freshly prepared YDC agar plates and incubated in 27°C for 48 h. To prepare different concentrations of the culture filtrate (i.e., 1/2, 1/4, and 1/8 of the original concentration), 62.5, 125, and 250 ml aliquots were separately added to 250 ml of SDW. Subsequently, five common bean seedlings grown on sterilized perlite (6 to 7 days old) pulled out, their roots were rinsed three times in sterile water and dipped into the culture filtrate. The seedlings were incubated under ambient conditions for 5 dpi. Control seedlings were dipped into the same volume and same concentrations of uninoculated LB medium as well as SDW. The substrate was considered as toxic if the seedlings showed wilting symptoms within 5 dpi. To ensure that the symptoms were not attributed to bacterial colonization, Koch’s postulates were accomplished on YPGA medium for the seedlings showing wilt symptoms using the procedure as described above. All the pathogenicity and bioassay tests were conducted at least twice.

Genome sequencing, assembly, and annotation.

Genome sequencing and database construction were performed by Shanghai Ouyi Biomedical Technology Co. (Shanghai, China) using Oxford Nanopore sequencing technologies according to the recommendations from the manufacture. Nanopore data were used for de novo genome assembly using Flye service (Kolmogorov et al. 2019). The assembled genome was subjected to coding gene prediction using online service Prokaryotic Dynamic Programming Genefinding Algorithm, prodigal v2.6.3. Furthermore, RNAscan-SE v1.3.1, RNAmmer v1.2, and Rfam v10.0 services were used to predict tRNA, rRNA and RNA families, respectively. Publicly available databases, i.e., NR annotations, CAZy (carbohydrate-active enzymes), COG (clusters of orthologous groups), eggNOG (evolutionary genealogy of genes: nonsupervised orthologous groups), GO (gene ontology), KEGG (Kyoto encyclopedia of genes and Genomes), VFDB (virulence factor database), Pfam, as well as PHI-base (pathogen and host interaction database) were used for genomics analyses and inferring functional properties of C. flaccumfaciens pv. flaccumfaciens genes. For NR, COG/KOG, GO, Swissprot, eggNOG, and KEGG database annotations, we used diamond software for comparison (Buchfink et al. 2015). We also used HMMER3 software and searched against Pfam database to allocate protein families within the C. flaccumfaciens pv. flaccumfaciens genome (El-Gebali et al. 2019). Subsequently, genome annotation was performed using the GeneMarkS+ v4.6 suite implemented in the NCBI Prokaryotic Genome Annotation Pipeline with default settings (Borodovsky and Lomsadze 2014). Total number of protein-coding genes, RNA genes, and pseudogenes were determined for the genome. Furthermore, the Circos software v0.69 was employed to generate a circular map of the C. flaccumfaciens pv. flaccumfaciens genome (Krzywinski et al. 2009), while IslandViewer 4 was used for the identification and visualization of GIs (Bertelli et al. 2017).

Phylogenetic analyses.

In order to determine the precise phylogenetic position of the strain P990, all the publicly available genome sequences assigned as Curtobacterium spp. were retrieved from the NCBI GenBank database (up to February 2020) and included in the phylogenetic analyses. Average nucleotide identity (ANI) was calculated among all the Curtobacterium spp. 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 (Richter et al. 2016), ANI calculator (Rodriguez-R and Konstantinidis 2016), and OrthoANIu (Yoon et al. 2017). ANI-based neighbor joining phylogenetic trees were constructed using the ANI calculator online service, first for all the Curtobacterium spp. strains, and then for the representative species of all the plant pathogenic actinobacteria using the procedure described previously (Osdaghi et al. 2020b). Additionally, genome-to-genome distance calculator (GGDC, v2.1) online service was used to calculate digital DNA-DNA hybridization (dDDH) value, which infers to the genome-to-genome distances between pairs of genomes based on the genome blast distance phylogeny (Meier-Kolthoff et al. 2013). A combination of ANI and dDDH indices was used to designate a taxonomic status to a given taxon where the “stand-alone species” status was assigned to a taxon when both ANI and dDDH values were below the accepted threshold, i.e., ≤95% and ≤70% for ANI and dDDH, respectively (Kim et al. 2014).

Comparative genomics.

Based on the results obtained from the phylogenetic analyses, P990 was subjected to comparative genomics against two C. flaccumfaciens strains, four phylogenetically closely related Curtobacterium spp. strains to C. flaccumfaciens pv. flaccumfaciens, as well as seven economically important actinobacterial species, which represent the entire genetic diversity of plant-associated actinobacteria based on ANI/dDDH, host of isolation, and pathogenicity characteristics (Thapa et al. 2019). The online annotation service RAST (Aziz et al. 2008) 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 as 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) (Overbeek et al. 2005). Subsequently, the genomes were transferred to the comparative environment of the SEED-Viewer (Overbeek et al. 2014) 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. Distribution of the genes among various clusters and specific protein-encoding genes within each cluster were estimated using the same service.

Furthermore, BLASTx-based investigation was performed to decipher whether the major pathogenicity determinant genes/clusters in plant pathogenic actinobacteria are present in the C. flaccumfaciens pv. flaccumfaciens genome. Hence, one-versus-one BLASTx search was accomplished against the sequences of the pathogenicity island (a 129-kb low G+C region that includes chp and tomA clusters) as well as several individual genes proposed to have an effective contribution to the virulence of plant pathogenic bacteria within the family Microbacteriaceae (Thapa et al. 2019). Proteins with amino acid sequence similarities higher than 50% and with a query coverage higher than 70% were considered homologous. We also screened the genome sequence of P990 for the presence of hypothetical bacteriocin-encoding genes/clusters using the web-based tool BAGEL4 (De Jong et al. 2006). BAGEL4 combines direct mining for the structural genes with indirect mining for bacteriocin-associated genes. Furthermore, the online service PlasmidFinder 2.0 (Carattoli et al. 2014) was used to screen the genome of P990 for the presence of integrative plasmids/episomes. Identification and annotation of prophage sequences within the P990 genome were performed using the online service PHASTER (Arndt et al. 2016). 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 (Wang et al. 2015). The analyses were conducted on the “bacteria” section of the platform using default settings (E-value: 1e-5, and inflation value: 1.5). Regarding the numeric limitation of OrthoVenn to handle the bacterial genomes (up to six genomes per run), different series of the strains were evaluated using the same parameters.

Pairwise genome collinearity alignment of the strain P990 with the above-mentioned representative actinobacterial strains was performed and visualized by BRIG 0.95. BRIG is a Java-based tool for visualizing the comparison of a reference sequence to a set of query sequences. Results are plotted as a series of rings, each representing a query sequence, which are colored to indicate the presence of hits to the reference sequence. BRIG indicates which regions of the reference sequence are present/absent in query sequences (Alikhan et al. 2011). Moreover, Mauve software was used to illustrate locally collinear blocks among the genome of P990 and those of the representative actinobacterial strains mentioned above. Mauve is a Java-based tool for multiple alignments of whole genomes that identifies blocks of sequence homology and assigns each block a unique color. Each genome can then be visualized as a sequence of these colored blocks, facilitating visualization of the genome comparisons. This makes it easy to identify regions that are conserved among the whole set of input genomes (Darling et al. 2010).

Data availability.

The dataset produced in this whole-genome sequencing project is available at the DDBJ/EMBL/GenBank database under accession numbers CP045287 to CP045290. A pure culture of the strain P990 is deposited in the International Collection of Microorganisms from Plants (ICMP, Auckland, New Zealand) and is available under the accession number ICMP 22053. The data that support the findings of this study are openly available in the NCBI GenBank database.

RESULTS

Pathogenicity and host range of P990.

Among the seven crop species inoculated with P990, common bean, cowpea, mung bean, and soybean plants showed typical symptoms of bacterial wilt disease at 6 to 15 dpi. Interveinal chlorosis leading to necrotic areas and overall wilt symptoms were observed on common bean, cowpea, and mung bean plants (Supplementary Fig. S1a). Symptoms on soybean plants were observed later than those on the other crops (i.e., at 15 dpi) and included necrotic lesions (tan spot) on the leaves surrounded by a faint yellow halo at the edges of the necrotic areas. The inoculated P990 strain was consistently reisolated from the symptomatic plants and its identity was confirmed using the primer pair CffFOR2/CffREV4 (data not shown). On the other hand, dipping the common bean seedlings into different concentrations of cell-free culture filtrate of the strain P990 resulted in wilting and plant death at 2 to 4 dpi depending on the concentration of the culture filtrate, where the more concentrated the culture filtrate was the earlier wilting symptoms were observed (Supplementary Fig. S1b). Attempts to isolate the bacterial wilt pathogen from the symptomatic seedlings inoculated with culture filtrate were unsuccessful, indicating the absence of viable C. flaccumfaciens pv. flaccumfaciens cells in the symptomatic tissues. Hence, it could be hypothesized that a toxic compound is responsible for the development of wilt symptoms. Control seedlings dipped into sterile water as well as those seedlings dipped into C. flaccumfaciens pv. flaccumfaciens-free LB medium remained healthy up to 5 dpi.

General feature of P990 genome.

The complete genome sequence of P990 was constructed by employing Oxford Nanopore technology and subsequent de novo genome assembly using Flye online service. P990 has a single circular chromosome consisting of 3,736,959 bp. The GC content of the chromosome is 71.0%. Figure 1A represents the circular diagram and genome features of the P990 chromosome. From the outermost to inner circles, nucleotide numbering (circle 1), COG annotation of forward (circle 2) and reverse (circle 3) strands, noncoding RNAs (rRNA red, tRNA blue, sRNA green; circle 4), GC content (circle 5), and GC skew (circle 6). No significant GC skew was detected in the P990 chromosome, which was in congruence with the results reported for other plant pathogenic actinobacteria (Gartemann et al. 2008). The chromosome of P990 contains a total of 3,672 predicted CDSs. Based on the results obtained from the NCBI prokaryotic genome annotation pipeline, 3,586 CDSs were capable of coding a protein. Among the predicted CDSs, 710 were assigned as hypothetical proteins. A total of 59 RNA genes were detected with 48 tRNAs, eight rRNAs, and three ncRNAs. Furthermore, 70 pseudogenes were detected within the genome, among which 40 were incomplete CDSs, 34 were frameshifted, 12 had multiple problems, and nine pseudogenes had internal stop. Supplementary Figure S2 summarizes the functional classification of gene clusters and orthologous groups resulted from the online services CAZy (A), COG (B), eggNOG (C), GO (D), and KEGG (E). As shown in Supplementary Figure S3a, the genomic content of P990 had the highest similarity to Curtobacterium sp. UNCCL17 (NZ_JMLI00000000.1).

Download as PowerPoint

Plasmids.

The genome of P990 contains a 147,310-bp circular plasmid (pCff1) with 66.1% G+C content. In addition, two smaller plasmid-like circular DNAs, pCff2 (25,142 bp) and pCff3 (22,293 bp), were detected within the genomic contents of the bacterium. Although pCff1 had a set of well-organized protein-coding genes, there was no predicted open reading frames on plasmid-like circular DNAs designated as pCff2 and pCff3; hence, they were excluded from our analyses. Figure 1B shows the circular diagram and features of pCff1 in P990. Data mining among various public databases revealed that homologs of endo-1, 4-beta-glucanase (celA; two copies) in Aspergillus oryzae (yellow koji mold), which is responsible for cellulose hydrolysis, was found in pCff1. Further, homologs of chpC gene (in two copies) and a putative extracellular serine protease (ppaJ) in Clavibacter michiganensis were found in pCff1 (locus tag: pCM1_0023). Investigation of CAZy database showed that there were one carbohydrate-binding module, three glycoside hydrolases, and one polysaccharide lyase in pCff1 (Supplementary Fig. S4a). Further, investigation in COG and eggNOG databases confirmed the presence of at least 10 CDSs in pCff1 contributing to carbohydrate transport and metabolism (Supplementary Fig. S4b and c). A summary of function specification of pCff1 CDSs using different databases is shown in Supplementary Figure S4. Interestingly, 18.3% of the genomic contents of pCff1 had similarity to Clavibacter michiganensis, while 15.6% of the contents of the plasmid was similar to Xanthomonas translucens (Supplementary Fig. S3b).

Genomic islands.

The online service IslandViewer has identified 24 GIs on the chromosome of P990 within the range of 4,051 bp (GI-V) and 17,787 bp (GI-X). Interestingly, 12 out of 24 GIs had <70% G+C% content as listed in Table 1. BLASTx-based screening of the GIs revealed that 18 GIs were C. flaccumfaciens pv. flaccumfaciens-specific, in which >70% of the nucleotide content did not occur in the nonpathogenic Curtobacterium spp. strains, while the remaining six GIs were detected either in Curtobacterium spp. or C. flaccumfaciens (Table 1). The largest of these GIs (17,787 bp) is located between the positions 1,604,094 and 1,621,881 bp of the P990 genome (Fig. 2A). Results of gene predictions for each of these GIs are shown in Table 1. A number of biological functions could be assigned to each of the individual GIs. For instance, proteins predicted in GI-I were responsible for different steps of riboflavin biosynthesis. Further, several DUF (domain of unknown function) families were identified among different GIs. In GI-II no known protein product was identified thus all the CDSs were designated as hypothetical proteins. In order to determine the status of these GIs in the genomes of other nonpathogenic Curtobacterium spp. strains as well as the other actinobacterial plant pathogens, BRIG 0.95 was used to whole genome-based comparisons with P990 as the reference genome. Most of the detected GIs were C. flaccumfaciens pv. flaccumfaciens-specific as shown on the comparative circular map of nine actinobacterial strains (Fig. 2B). For instance, the GIs II, IV, VI, VIII-X, XII-XV, and XX-XXIII could be visually determined as C. flaccumfaciens pv. flaccumfaciens-specific. On the other hand, four GIs were detected on pCff1 as listed in Table 1. Interestingly, a number of putative virulence-associated genes, i.e., pectate lyase (protein id: QFS80865.3), S1 family peptidase (serine protease; protein ids: QIH95653.1, QIH95654.1, and QFS80891.2), and glycosyl hydrolase family (protein id: QFS80892.1), were detected on GIs of pCff1.

TABLE 1. List of genomic islands in the chromosomal DNA and plasmid pCff1 DNA of Curtobacterium flaccumfaciens pv. flaccumfaciens P990 detected using IslandViewer 4

Number	Island start	Island end	Length	Specificity	G+C%	Gene content
Islands on chromosome
I	305,117	315,177	10,060	+	73.38	Hypothetical protein (3), 3,4-dihydroxy-2-butanone-4-phosphate synthase, 6,7-dimethyl-8-ribityllumazine synthase, GNAT family N-acetyltransferase, MFS transporter, phage holin family protein, riboflavin synthase, TetR family transcriptional regulator, winged helix-turn-helix transcriptional regulator
II	445,487	451,733	6,246	+	64.21	Hypothetical protein (14)
III	719,074	724,308	5,234	+	66.76	Hypothetical protein (4), ABC transporter permease, alpha/beta fold hydrolase, ATP-binding cassette domain-containing protein, methyltransferase domain-containing protein, response regulator, transcriptional regulator
IV	748,694	758,824	10,130	+	71.07	Hypothetical protein (3), phosphate ABC transporter permease PstA, phosphate ABC transporter permease subunit PstC, phosphate ABC transporter substrate-binding protein PstS, sortase
V	787,302	791,353	4,051	−	64.98	Hypothetical protein (2), CPBP family intramembrane metalloprotease, DUF1648 domain-containing protein, HTTM domain-containing protein, SdpA family antimicrobial peptide system protein
VI	1,319,540	1,324,587	5,047	+	67.97	30S ribosomal protein S18, 50S ribosomal protein L9, AAA family ATPase, DDE-type integrase/transposase/recombinase, DEAD/DEAH box helicase, DoxX family membrane protein, DUF4209 domain-containing protein, EamA family transporter, GntR family transcriptional regulator, ImmA/IrrE family metallo-endopeptidase, low molecular weight phosphatase family protein, Lsr2 family protein, metalloregulator ArsR/SmtB family transcription factor, NAD(P)-binding domain-containing protein, replicative DNA helicase, single-stranded DNA-binding protein
VII	1,488,523	1,492,942	4,419	−	55.97	TIR domain-containing protein
VIII	1,577,417	1,582,572	5,155	+	68.64	AAA family ATPase, EamA family transporter, HAD-IB family hydrolase, MFS transporter, zinc-binding dehydrogenase
IX	1,584,727	1,589,583	4,856	+	70.81	Beta-ketoacyl-ACP synthase III, helix-turn-helix domain-containing protein, MBL fold metallo-hydrolase, NAD(P)-binding domain-containing protein
X	1,604,094	1,621,881	17,787	+	66.69	DUF11 domain-containing protein, DUF4190 domain-containing protein, helix-turn-helix domain-containing protein, LysR family transcriptional regulator, oxidoreductase, phospholipase, SDR family oxidoreductase, TetR family transcriptional regulator, TIR domain-containing protein, tryptophan–tRNA ligase
XI	1,653,610	1,662,856	9,246	−	70.91	DUF3494 domain-containing protein, GAF domain-containing protein, HupE/UreJ family protein, metallophosphoesterase, ROK family protein
XII	1,728,686	1,736,488	7,802	+	64.42	Cupin domain-containing protein, DUF4231 domain-containing protein, FAD-dependent oxidoreductase, helix-turn-helix domain-containing protein, recombinase family protein
XIII	1,992,203	1,998,551	6,348	+	65.99	Acetyltransferase, glycosyltransferase
XIV	2,258,207	2,264,283	6,076	+	61.64	DUF1911 domain-containing protein, RNA 2’-phosphotransferase, SMI1/KNR4 family protein
XV	2,636,399	2,646,930	10,531	+	62.77	AAA family ATPase, ASCH domain-containing protein, ATP-binding cassette domain-containing protein, carbohydrate-binding protein, GNAT family N-acetyltransferase
XVI	2,683,214	2,688,206	4,992	+	71.94	Alpha/beta fold hydrolase, alternate-type signal peptide domain-containing protein, MerR family transcriptional regulator, signal peptidase I
XVII	2,702,029	2,708,986	6,957	−	72.20	Alpha/beta hydrolase fold domain-containing protein, low temperature requirement protein A, NAD(P)-binding domain-containing protein
XVIII	2,711,815	2,718,382	6,567	−	70.84	ABC transporter permease subunit, BCCT family transporter, extracellular solute-binding protein, glycoside hydrolase, substrate-binding domain-containing protein
XIX	3,255,741	3,269,226	13,485	+	70.53	Alcohol dehydrogenase catalytic domain-containing protein, alpha/beta fold hydrolase, catalase, DUF1345 domain-containing protein, GNAT family N-acetyltransferase, MFS transporter
XX	3,425,557	3,431,948	6,391	+	70.90	Acyltransferase, GAF domain-containing protein, helix-turn-helix domain-containing protein, isoprenylcysteine carboxylmethyltransferase family protein, NAD(P)-dependent oxidoreductase
XXI	3,433,625	3,437,682	4,057	+	73.95	Chromate efflux transporter, glycosyltransferase, heat-shock protein HtpX, helix-turn-helix domain-containing protein
XXII	3,455,324	3,463,089	7,765	+	70.98	ABC transporter permease subunit, extracellular solute-binding protein, glycosyltransferase, heme utilization protein, inositol monophosphatase, substrate-binding domain-containing protein
XXIII	3,463,206	3,469,631	6,425	+	71.13	Cation:dicarboxylase symporter family transporter, glycosyl transferase, LysM peptidoglycan-binding domain-containing protein
XXIV	3,709,160	3,714,691	5,531	−	65.04	DNA mismatch endonuclease Vsr, DUF2075 domain-containing protein, DUF871 family protein, NUDIX domain-containing protein
Islands on plasmid pCff1
1	7,966	23,061	15,095	−	59.92	Hypothetical protein (6), IS481 family transposase, pectate lyase, S1 family peptidase (2), trypsin-like peptidase domain-containing protein, tyrosine-type recombinase/integrase
2	55,043	59,370	4,327	−	67.11	Carbohydrate ABC transporter permease, extracellular solute-binding protein, glycosyl hydrolase family, sugar ABC transporter permease
3	104,029	108,776	4,747	−	68.29	Cell division protein FtsK, conjugal transfer protein, hypothetical protein (5)
4	120,066	146,838	26,772	−	66.39	Dihydroxy-acid dehydratase, DUF1190 domain-containing protein, DUF2167 domain-containing protein, DUF4129 domain-containing protein, DUF4350 domain-containing protein, DUF58 domain-containing protein, gamma carbonic anhydrase family protein (2), glutathionylspermidine synthase family protein, hypothetical protein (3), IS3 family transposase (2), Lrp/AsnC family transcriptional regulator, MFS transporter, MoxR family ATPase, PQQ-dependent sugar dehydrogenase, RDD family protein, stage II sporulation protein M, transposase (2), venom serine protease KN13

View as image HTML

Download as PowerPoint

Phylogenetic analyses.

ANI calculator online service was used to conduct a phylogenetic analysis among the genome sequences of all 45 available actinobacterial strains in the NCBI GenBank database designated as Curtobacterium spp. Neighbor joining phylogenetic tree constructed with all-versus-all strategy revealed high genetic diversity among these strains (Supplementary Fig. S5). ANI values between different pairs of strains varied from 75 to 99% among Curtobacterium spp. (data not shown). The genome of P990 shared ANI and dDDH values of 97 and 74%, respectively, with the pathotype strain of the pathogen CFBP 3418^T, confirming the taxonomic position of the strain as reported previously (Table 2). Using the standard criteria based on ANI and dDDH (Kim et al. 2014), it has been shown that among the 45 evaluated Curtobacterium spp. genome sequences only the strains P990, CFBP 3418^T, MMLR14-002, and MMLR14-014 belong to C. flaccumfaciens sensu stricto, while the two strains UNCCL17 with an unknown source and MCBA15-005 isolated from leaf litter were phylogenetically closely related to C. flaccumfaciens but more likely to belong to a novel species (Fig. 3; Table 2). The C. flaccumfaciens sensu stricto strains were clustered in a monophyletic clade showing 97% ANI with one another. Apart from the strains UNCCL17 and MCBA15-005, the closest clade to C. flaccumfaciens consisted of three strains, i.e., MCBA15-007, MEB126, and UCD-AKU, isolated from leaf litter, Arabidopsis thaliana leaf, and residential carpet, respectively.

TABLE 2. Average nucleotide identity (ANI; lower diagonal) and digital DNA–DNA hybridization (dDDH; upper diagonal) values among the representative plant-associated actinobacterial strains^a

Strain	Number	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
C. flaccumfaciens pv. flaccumfaciens P990	1	…	74	73	63	63	52	52	28	28	26	26	20	20	19	19
C. flaccumfaciens pv. flaccumfaciens CFBP 3418T	2	97	…	74	63	64	53	53	28	28	26	26	19	20	19	19
C. flaccumfaciens MMLR14-014	3	97	97	…	64	65	53	53	28	29	26	26	20	20	19	19
Curtobacterium sp. UNCCL17	4	95	95	96	…	83	50	51	28	28	26	26	20	20	19	19
Curtobacterium sp. MCBA15-005	5	95	96	96	98	…	51	51	53	28	26	26	20	20	19	19
Curtobacterium sp. MCBA15-007	6	94	94	94	93	93	…	90	28	28	26	26	19	20	19	20
Curtobacterium sp. UCD-AKU	7	94	94	94	93	92	99	…	28	28	26	26	19	20	19	20
Curtobacterium sp. BH-2-1-1	8	85	85	84	84	85	85	85	…	35	27	27	20	19	19	19
C. pusillum AA3	9	85	85	85	85	85	85	85	88	…	27	27	20	20	20	19
Curtobacterium sp. MR-MD2014	10	84	84	83	85	84	84	84	85	85	…	40	20	20	20	20
Curtobacterium sp. SGAir0471	11	84	84	83	83	84	84	84	85	85	90	…	20	20	20	20
Rathayibacter tritici NCPPB 1953	12	78	78	78	78	78	78	78	78	78	78	78	…	20	19	20
Leifsonia xyli CTCB07	13	78	78	78	78	78	78	78	78	78	78	78	79	…	20	21
Clavibacter sepedonicus ATCC 33113T	14	78	78	78	78	78	78	78	78	78	78	78	78	78	…	46
Clavibacter michiganensis NCPPB 382T	15	78	78	78	78	78	78	78	78	78	78	78	78	79	92	…

^aANI values were calculated using three different algorithms, i.e., JSpeciesWS, ANI calculator, and OrthoANIu. A combination of ANI and dDDH indices was used to designate a taxonomic status to a given phylogenetic clade. Among the Curtobacterium spp. genome sequences only the strains P990, CFBP 3418^T, MMLR14-002, and MMLR14-014 belong to C. flaccumfaciens sensu stricto.

View as image HTML

Download as PowerPoint

Comparative genomics.

Comparative genomics using the data provided by RAST and SEED viewer online services revealed that the genome size among plant-associated actinobacteria included in this study varied between 2,584 kbp in L. xyli subsp. xyli and 3,906 kbp in Curtobacterium sp. UNCCL17, with the G+C% content varying between 67.7% in L. xyli subsp. xyli and 73.6% in Clavibacter capsici PF008^T (Table 3). Furthermore, the number of CDSs varied between 2,642 in Clavibacter michiganensis subsp. phaseoli CFBP 8627^T and 3,844 in Curtobacterium pusillum AA3. Table 3 presents comparative genomics of P990 against a panel of 13 representative plant-associated actinobacterial strains.

TABLE 3. Genomic characteristics of plant-associated actinobacterial strains used in this study^a

Characteristics	C. flaccumfaciens			Curtobacterium spp.				Clavibacter spp.					Rathayibacter tritici NCPPB 1953	Leifsonia xyli subsp. xyli CTCB07
	P990	CFBP 3418	MMLR14-014	UNCCL17	UCD-AKU	BH-2-1-1	C. pusillum AA3	C. michiganensis NCPPB 382	C. capsici PF008	C. nebraskensis NCPPB 2581	C. sepedonicus ATCC 33113	C. michiganensis subsp. phaseoli CFBP 8627
GenBank accession number	CP045287	NZ_PUEZ00000000	NZ_MJGX00000000	NZ_JMLI00000000	NZ_APJN00000000	NZ_VDTL00000000	NZ_CP018783	AM711867	NZ_CP012573	NC_020891	NC_010407	QWGV00000000	NZ_CP015515	AE016822
Genomic features
Genome size (kbp)	3,736	3,827	3,822	3,906	3,692	3,795	3,973	3,297	3,056	3,063	3,258	3,052	3,354	2,584
GC content (%)	71.0	71.0	71.0	70.8	70.9	71.4	70.9	72.7	73.6	73.0	72.6	73.5	69.5	67.7
Number of CDS	3,598	3,729	3,767	3,837	3,655	3,721	3,844	2,979	2,725	2,739	3,047	2,642	3,289	2,858
Number of subsystems	250	253	249	248	250	257	265	345	326	325	345	315	241	236
Subsystem feature
Resistance to chromium compounds	1	1	1	1	2	0	0	0	0	0	0	0	0	0
Multidrug resistance efflux pumps	2	2	3	2	0	2	2	0	0	0	0	0	0	0
ABC transporters	6	7	6	4	6	8	8	37	26	32	33	40	0	2
Type II secretion system	0	0	0	0	0	0	0	3	3	3	3	3	0	0
Type IV secretion system (Type IV pilus)	11	11	12	10	0	11	11	0	0	0	0	0	0	0
Siderophores	0	0	0	0	0	0	1	5	3	5	3	3	0	0
Siderophore Yersiniabactin	0	0	0	0	0	0	0	2	0	2	0	0	0	0
Siderophore Aerobactin	0	0	0	0	0	0	0	3	3	3	3	3	0	0
Siderophore Enterobactin	0	0	0	0	0	0	1	0	0	0	0	0	0	0
Protein degradation	4	4	4	4	4	7	8	21	21	19	19	22	5	9
Serine endopeptidase	0	0	0	0	0	0	0	1	1	1	1	1	0	0
Metallocarboxypeptidases	0	0	0	0	0	1	1	3	3	3	3	3	2	3
ATP-dependent proteolysis	0	0	0	0	0	0	0	9	9	8	9	9	0	0
Omega peptidases	0	0	0	0	0	1	2	2	2	2	1	3	0	1
Flagellar motility in prokaryota	32	33	33	5	33	32	32	0	0	0	0	0	0	5
Auxin biosynthesis	5	4	5	5	4	4	4	0	0	0	0	0	0	0
Glycolysis and gluconeogenesis	0	0	0	0	0	0	0	13	13	12	13	12	0	0
Di- and oligosaccharides	20	21	20	23	23	20	22	49	38	40	46	44	22	13
Trehalose uptake/utilization	4	4	4	4	4	3	3	0	0	0	0	0	4	0
Maltose and maltodextrin utilization	0	0	0	0	0	0	0	11	12	10	10	16	0	0
Lactose and galactose uptake/utilization	6	7	6	9	6	7	7	16	12	12	15	14	4	6
Sucrose utilization	0	0	0	0	3	0	0	3	0	2	2	0	4	0
Mannitol utilization	0	0	0	0	0	0	0	7	7	7	6	5	0	0
Inositol catabolism	0	0	0	0	0	0	17	16	15	14	0	14	0	8
Propionate-CoA to succinate module	0	0	0	0	0	0	0	5	5	6	5	5	5	6
Xylose utilization	6	5	5	5	8	5	5	18	10	11	15	7	5	2
Fructose utilization	6	6	7	6	7	7	6	13	11	11	11	12	10	6
L-Arabinose utilization	3	3	3	3	3	4	3	12	12	11	12	11	4	0
D-galactarate, D-glucarate and D-glycerate catabolism	0	0	0	0	0	0	0	5	0	0	4	0	0	0

^aIndividual genomes were analyzed using the online annotating service RAST, and protein-encoding sequences (CDS), functions of the genes, and represented subsystems in the genomes were determined for each genome using the SEED-Viewer comparative environment. Values in normal font indicate absence of a given subsystem, italicized values indicate presence of a given subsystem, underlined values indicate low number of subsystems, and bold values indicate high number of subsystems.

View as image HTML

Several lines of evidence highlight differences in metabolic networks and subsystem profiles between C. flaccumfaciens pv. flaccumfaciens and those of nonpathogenic Curtobacterium spp. strains, as well as the other Gram-positive bacterial plant pathogens. For instance, subsystems that are responsible for resistance to antibiotics and toxic compounds, motility and chemotaxis, type IV secretion system, auxin biosynthesis, and trehalose uptake/utilization were found in Curtobacterium spp. strains but not in the other actinobacterial plant pathogens (Table 3). The exceptions were L. xyli subsp. xyli CTCB07 and Rathayibacter tritici NCPPB 1953 in which the motility and chemotaxis, and trehalose uptake/utilization subsystems were found, respectively. In contrast, with some exceptions (as detailed in Table 3) type II secretion system, different types of siderophores, serine endopeptidase, metallocarboxypeptidases, ATP-dependent proteolysis, omega peptidases, glycolysis and gluconeogenesis, maltose and maltodextrin utilization, sucrose utilization, mannitol utilization, inositol catabolism, propionate-CoA to succinate module, as well as D-galactarate, D-glucarate, and D-glycerate catabolism were found in Clavibacter spp., Rathayibacter sp., and Leifsonia sp. plant pathogens but not in Curtobacterium spp. strains.

Surprisingly, the FHIPEP (flagellar/Hr/invasion proteins export pore) family type III secretion system (T3SS, protein id: QHN63420.1) was detected in the genome of P990, which was absent in Clavibacter spp. and Rathayibacter sp. while present in Leifsonia sp. strains. Furthermore, a nonflagellar EscU/YscU/HrcU family T3SS export apparatus switch protein (protein id: QFS80385.2) and the flagellar T3SS pore gene (fliP, protein id: QFS80387.1) were detected in the genome of P990, which were absent in the other plant pathogenic actinobacteria i.e., Clavibacter spp. and Rathayibacter sp. As for the type IV secretion system components, type IV secretion proteins (Rhs, protein ids: QHN62613.1 and QFS80419.2), type IV pilus assembly protein (PilM, protein id: QHN62675.1), as well as type IV pilus twitching motility protein (PilT, protein id: QHN63630.1) were detected in the genome of P990 which were absent in the other actinobacteria evaluated in Table 3.

BLASTx was performed to evaluate the presence of virulence-associated genes in different plant pathogenic Clavibacter strains in the genome of P990. Homologous of virulence-associated genes 1,4-beta-xylanase (xysA, protein id: QHN62540.1), pectate lyase (pelA1 and pelA2), serine protease (chpC, chpG, and pat-1), as well as sortase (srtA) in the tomato pathogen Clavibacter michiganensis NCPPB 382T were detected in the chromosome and plasmid pCff1 of P990. Three genes, i.e., 1,4-beta-xylanase (protein id: QHN62540.1), trypsin-like serine protease (protein id: QHN63486.1), and sortase (protein id: QFS79737.2), were found on chromosome, while pectate lyase (protein ids: QFS80865 and QFS80866) and trypsin-like serine protease (protein id: QFS80891) were detected on plasmid pCff1.

Orthologous gene clusters were determined using OrthoVenn online service through five-versus-five designations of the representative strains from different sets of actinobacteria (Fig. 4A and B). Curtobacterium spp. strains included in the OrthoVenn analyses shared 1,711 proteins in their genome sequences (Fig. 4A). The C. flaccumfaciens pv. flaccumfaciens strains P990 and CFBP 3418T showed one and seven unique proteins, respectively, in their sequences, while 26 proteins were shared between these two strains. Furthermore, 30 proteins were shared among the three C. flaccumfaciens strains, i.e., P990, CFBP 3418T, and MMLR14-014. The higher number of unique proteins in CFBP 3418T in comparison with P990 was likely due to the sequencing and assembling technology of CFBP 3418T where the chromosomal DNA and hypothetical plasmid(s)s were assembled in the same file. On the other hand, the strain P990 had 43 unique protein in the OrthoVenn analyses in comparison with four non-Curtobacterium actinobacterial species, while 1,026 proteins were shared among these six plant pathogens (Fig. 4B). Among the non-Curtobacterium actinobacterial strains, P990 had the highest number of shared proteins (122) with R. tritici NCPPB 1953.

Download as PowerPoint

One-versus-all collinearity test of P990 genome against a set of four Curtobacterium spp. strains was conducted using Mauve. The organization of locally collinear blocks (LCBs) was visualized to determine genome rearrangements and segmentation. To provide a precise scheme from the collinearity of the genomes, LCB weight of all the alignments was adjusted to 150. The order of the LCBs in P990 was generally in congruence with the LCBs in CFBP 3418T; however, the latter strain has a reversion in almost 70% of the genome content (nucleotides 1 to 2,700 kbp). The majority of the LCBs were present in all the five Curtobacterium sp. strains indicating that the gene content within the evaluated strains is preserved (Fig. 5A). Among the three C. flaccumfaciens strains, alignment of the chromosome was broken down into several LCBs in the strain MMLR14-014, indicating multiple translocations and inversions in comparison with the genome of CFBP 3418T. As for the two Curtobacterium sp. strains UNCCL17 and MCBA15-007, the number of LCBs in the strain UNCCL17 was much less than those in the strain MCBA15-007, which was in congruence with their respective phylogenetic distance from the reference strain P990 (Fig. 3). Considering the collinearity of the genome of P990 against four actinobacteria (Fig. 5B), only small portions of the LCBs in the P990 genome were preserved within the actinobacterial plant pathogens indicating significant rearrangements and insertion/deletions between P990 and those of four actinobacterial species.

Download as PowerPoint

No integrative plasmid (episome) and prophage were detected using the PlasmidFinder and PHASTER online services, respectively, in the genome sequence of P990. Furthermore, no acquired antimicrobial resistance gene was found using the ResFinder server. In silico screening for bacteriocins and antibiotic peptides showed that P990 is capable of producing linocin_M18 and linaridin. Although linocin_M18 was detected in the genome sequences of CFBP 3418^T, UNCCL17, and MCBA15-005, linaridin was shown to be unique for P990 among the strains investigated in Figure 3. The C. flaccumfaciens strains MMLR14_002 and MMLR14_014 as well as the Curtobacterium sp. strain MCBA15_005 contained only sactipeptides, while the strain UNCCL17 produced cypemycin in addition to linocin_M18 and sactipeptides. No evidence was found for the presence of bacteriocins in the strains BH-2-1-1 and SGAir0471, while in the strains MCBA15-007, MEB126, UCD-AKU, and C. pusillum AA3 putative bacteriocins with unknown identity were detected (data not shown). The bacteriocins lactococcin_972 and zoocin_A were detected only in the Curtobacterium sp. strains MR-MD2014, DSM 20528, and 9128, which were phylogenetically far from the plant pathogenic members of the genus, indicating distinctive differences in the bacteriocins scheme of the strains designated as C. flaccumfaciens.

DISCUSSION

Among the economically important Gram-positive bacterial plant pathogens, C. flaccumfaciens pv. flaccumfaciens is the least studied member with no information from the virulence repertoires and pathogenicity determinates. In this study, we provide the first complete genome sequence of the C. flaccumfaciens pv. flaccumfaciens strain P990 causing bacterial wilt on dry beans. The bacterium contains a circular chromosome (3,736 kbp) as well as a circular plasmid pCff1 (147 kbp) both carrying a number of virulence-associated genes, e.g., 1,4-beta-xylanase (xysA), pectate lyase (pelA1 and pelA2), serine protease (chpC, chpG, and pat-1), and sortase (srtA). Furthermore, evidences for the production of toxic compounds and bacteriocins were detected using both phenotypic tests (culture filtrate assay) and genomics investigations.

Concerning the plasmid repertories of P990, results obtained in this study were incongruent with those of previously reported data. Indeed, agarose gel-based plasmid profiling using different strains of C. flaccumfaciens pv. flaccumfaciens did not find any evidence for the presence of plasmids (Gross et al. 1979; Osdaghi et al. 2018a), while genome sequencing in this study revealed the presence of at least one circular plasmid (pCff1) as well as two plasmid-like DNAs, with sizes of 25 and 22 kbp, in the genome of P990. Differences between the in vitro and in silico results might be due to the low number of plasmid copies in C. flaccumfaciens pv. flaccumfaciens cells leading to their undetectability using agarose gel-based techniques. Further investigations, e.g., plasmid curing and targeted mutagenesis are warranted to decipher the role of plasmid(s) in the biology of bacterial wilt pathogen. Meanwhile, detection of virulence-associated genes, i.e., pectate lyase, endo-1,4-beta-glucanase, and serine protease, in the pCff1 DNA sequence suggests the putative role of plasmids in the pathogenicity of C. flaccumfaciens pv. flaccumfaciens. In addition, it has been shown that a plasmid determines resistance to arsenate, arsenite, and antimony (III) in C. flaccumfaciens pv. oortii (Hendrick et al. 1984). Although no morphological difference was found between bean-pathogenic and nonpathogenic C. flaccumfaciens strains, the former were resistant to different concentrations of arsenic, while the latter were sensitive to the same concentrations (Osdaghi et al. 2018a). Interestingly, two arsenic transporter genes (protein ids: QHN62312.1 and QFS79252.1) and an arsenate reductase (ArsC, protein ids: QFS79253.1 and QFS80476.1) were detected in the chromosomal genome of P990. It remains undetermined whether the resistance against arsenic is mediated by the detected plasmids in C. flaccumfaciens pv. flaccumfaciens strains similar to those described for other bacterial species (Silver and Misra 1988). Furthermore, homologs of copper resistance protein (CopC, protein ids: QFS78994.1, QFS79968.2, and QFS80310.2), copper homeostasis protein (CutC, protein id: QHN62860.1), and copper resistance protein D (pcoD, protein ids: QFS79544.1 and QHN62928.1) were identified in the genome of P990. All these genes are involved in copper resistance, which may explain why copper-based chemicals are ineffective in field control of the bacterial wilt pathogen (Lamichhane et al. 2018; Osdaghi et al. 2020a).

So far, the lack of publicly available C. flaccumfaciens pv. flaccumfaciens complete genome sequence was limiting our understanding of its virulence mechanisms and pathogenicity determinants. On the other hand, genetic manipulation technologies and transformation vectors have not yet been developed for the bacterium, although it may be possible to use the platforms of the closely related pathogens Clavibacter michiganensis and L. xyli subsp. xyli (Brumbley et al. 2006; Thapa et al. 2019). Similar to its actinobacterial relatives, C. flaccumfaciens pv. flaccumfaciens colonizes the xylem system of the host plant and moves endophytically (Thapa et al. 2019; Zaumeyer 1932). Due to the lack of T3SS in actinobacterial plant pathogens, they use a set of lytic enzymes, toxins, and hormones to disrupt host plant cell walls and acquire nutrients (Thapa et al. 2019). Interestingly, we found a number of T3SS’s components, i.e., FHIPEP family T3SS, EscU/YscU/HrcU family T3SS export apparatus switch protein and flagellar T3SS pore gene fliP in the genome of P990. Since phenotypic feature reflecting the activity of T3SS, e.g., hypersensitivity reaction, has not been reported in C. flaccumfaciens pv. flaccumfaciens, the functionality and contribution of the above-mentioned T3SS components to the virulence of the pathogen need to be evaluated.

Vidaver (1982) has noted damage to xylem and decomposition of middle lamella of C. flaccumfaciens pv. flaccumfaciens-infected plants prior to wilting. Furthermore, high molecular weight glycopeptides were isolated from the pathogen, some of which were considered to cause plant cell wall disruption (Schuster and Coyne 1981). De Bruyne et al. (1992) showed that C. flaccumfaciens pv. flaccumfaciens harbors lytic enzymes, e.g., β-glucosidase, esterase, peptidases, and lipases, all of which have similarity to Clavibacter michiganense genes. Detection of virulence-associated genes 1,4-beta-xylanase (xysA), pectate lyase (pelA1 and pelA2), serine protease (chpC, chpG, and pat-1), and sortase (srtA) in the chromosome and plasmid pCff1 of P990, which were homologous to the corresponding genes in tomato pathogen Clavibacter michiganensis NCPPB 382^T, confirms the activity of cell wall degrading enzymes in C. flaccumfaciens pv. flaccumfaciens and suggests the putative role of these genes in the pathogenicity of the bacterium. Histological studies have also confirmed enzymatic activities in the C. flaccumfaciens pv. flaccumfaciens-infected tissues of dry beans (Nelson and Dickey 1970; Vidaver 1982).

Furthermore, it has been hypothesized that the mechanism of wilting in C. flaccumfaciens pv. flaccumfaciens is the restriction of water movement in xylem by phytotoxic glycopeptides (Gross and Vidaver 1979; Schuster and Coyne 1981). In this study, we found that a cell-free culture filtrate of P990 was capable of inducing wilting on common bean which is in congruence with the hypothesis that a toxic compound is responsible for virulence of the pathogen. Wood and Easdown (1990) have also reported a temperature-resistant substrate derived from the culture filtrate of C. flaccumfaciens pv. flaccumfaciens strains isolated in Australia that induced wilting symptoms on mung bean. However, the biological characteristics of the phytotoxic compounds produced by C. flaccumfaciens pv. flaccumfaciens remain undetermined. In vitro investigations suggest the production of bacteriocins in C. flaccumfaciens pv. flaccumfaciens similar to those reported in the other actinobacterial pathogens (Durbin 1972). Gross et al. (1979) investigated the biological features of bacteriocins produced by 10 C. flaccumfaciens pv. flaccumfaciens strains and named the resulting substrates as “flaccumfacin”. Similarly, genomics investigation confirmed the presence of linocin_M18 and linaridin bacteriocins in the genome of P990. Linocin_M18 is a bacteriocin of Brevibacterium linens M18 that inhibits the growth of several coryneform and other Gram-positive bacteria (Valdés-Stauber and Scherer 1994). It has been shown that the ability of C. flaccumfaciens pv. flaccumfaciens strains to produce bacteriocins against the pathogenic and nonpathogenic actinobacterial strains positively correlated with the virulence vigor and aggressiveness of the same strain on common bean. For instance, among 45 Iranian C. flaccumfaciens pv. flaccumfaciens strains, P990 had the most vigorous suppressive effect on the bacterial species evaluated through bacteriocin production test, while the same strain had the most aggressiveness on common bean, and the most fluidal colonies on culture media at 27°C (Osdaghi et al. 2018c).

Following the development of high throughput molecular-phylogenetic techniques, taxonomy of plant pathogenic actinobacteria which have previously been relied on phenotypic features was reconsidered (Li et al. 2018; Osdaghi et al. 2020b). In this study, ANI- and dDDH-based phylogenetic analyses revealed that the genetic diversity within the bacterial strains designated as C. flaccumfaciens in the literature is higher than that has so far been described. Only two strains, MMLR14-002 and MMLR14-014, were identified as members of C. flaccumfaciens while the remaining strains need to be reclassified as novel taxa. Recently, multilocus sequence analysis using the sequences of five housekeeping genes (i.e., atpD, gyrB, ppk, recA, and rpoB) revealed that the three strains MCBA15-007, MEB126, and UCD-AKU are phylogenetically closely related to the type strain of poinsettia (Euphorbia pulcherrima) pathogen C. flaccumfaciens pv. poinsettiae ICMP 2566T, while the strains MCBA15-005 and UNCCL17 are phylogenetically closely related to the type strain of sugar beet pathogen C. flaccumfaciens pv. betae ICMP 2594T (Osdaghi et al. 2018a). This indicates that the plant pathogenic members of C. flaccumfaciens are less likely to belong to the same species, reinforcing the need to a reconsideration in the taxonomy of phytopathogenic members of the species. A formal taxonomic study would provide appropriate nomenclature for these taxa.

In conclusion, taking together all the phylogenetic, genomics, and pathogenicity data obtained in this study from the analysis of P990 genome sequence along with several plant-associated actinobacteria, we provide a novel insight into the genetic diversity of Curtobacterium spp., which suggest a taxonomic revision on the plant pathogenic C. flaccumfaciens strains. Whole genome sequence of P990 would allow us to predict and test which genes are involved in the pathogenicity of the bacterium, providing an opportunity at the same time to develop state-of-the-art genome-informed detection methods to trace seed infections with lower efforts and cost. Additionally, genomics information will allow us to target C. flaccumfaciens pv. flaccumfaciens pathogenicity determinates in the search for developing wilt-resistant lines and cultivars to aid the dry bean industries around the world.

ACKNOWLEDGMENTS

E. Osdaghi and M. Khojasteh benefited from scientific exchanges and interactions promoted by the Iranian Ministry of Science and Technology.