Curtobacterium allii sp. nov., the actinobacterial pathogen causing onion bulb rot

A Gram-stain-positive, aerobic, and non-spore-forming bacterial strain, 20TX0166T, was isolated from a diseased onion bulb in Texas, USA. Upon testing its pathogenicity on onion bulb, it produced pathogenic response which makes it first species of pathogen belonging to the phylum actinobacteria detected in onion. Phylogenetic analysis of the 16S rRNA gene sequence revealed that the strain belonged to the genus Curtobacterium and was most similar to Curtobacterium flaccumfaciens LMG 3645T (100%), C. pusillum DSM 20527T (99.5%), and C. oceanosedimentum ATCC 31317T (99.5%). The estimated genome size of the novel species was 4.0 Mbp with a G + C content of 70.8%. The orthologous ANI (orthoANIu), ANI based on blast (ANIb), and dDDH values between the novel strain and the closest relative, C. flaccumfaciens LMG 3645T, were 95.7%, 95.4%, and 63.3%, respectively. These values were below the recommended species cut-off threshold of 96% (ANI) and 70% (dDDH), suggesting the strain may be a novel species. Physiologic and phenotypic characters of this novel strain were also unique when compared with the closely related species. The major cellular fatty acids of this strain were anteiso-C15:0 and anteiso-C17:0. Using a polyphasic approach based on phenotypic and genotypic analyses, strain 20TX0166T represents a novel species of the genus Curtobacterium, and the name Curtobacterium allii sp. nov. is proposed. The type strain is 20TX0166T (= LMG 32517T = CIP112023T = NCIMB 15427T).

by Yamada and Komagata (1972). Bacteria in this genus are aerobic, Gram-stain-positive, non-sporeforming, filamentous, and rod-shaped. Curtobacterium spp. are ubiquitous, cosmopolitan, and associated with diverse ecosystems around the world, including soil, water, and plant microbiomes (Aizawa et al. 2007;Chase et al. 2016;Li et al. 2016;Nascimento et al. 2020). Studies conducted to examine the microbial communities of leaf litter found Curtobacterium was the dominant genus, which was highly diverse within itself (Behrendt et al. 2002;Chase et al. 2017;Matulich et al. 2015). Numerous studies on beneficial strains have highlighted their importance as plant growth promoters and biological plant disease control agents (Bulgari et al. 2014;Khan et al. 2019;Lacava et al. 2007;Mayer et al. 2019;Silambarasan et al. 2019;Vimal et al. 2019). Although rare, Curtobacterium strains are also associated with human infections (Francis et al. 2011;Funke et al. 2005;Rivera et al. 2012). Curtobacterium flaccumfaciens (Cf) is a common plant pathogen with the potential to cause significant disease in many legumes, including beans, soybean, and cowpea (Gonçalves et al. 2017;Osdaghi et al. 2018a;Sammer and Reiher 2012). However, no bacterium of this genus or entire phylum actinobacteria were ever reported before as a pathogen of onion (Allium cepa). Due to the extent of plant diseases caused by Cf, many taxonomic studies using rep-PCR, MLSA, and whole genome analyses conducted have revealed high phenotypic and genotypic diversity, and authors from these studies have suggested that many strains previously assigned to this species could potentially be new species (Agarkova et al. 2012;Chen et al. 2021;Gonçalves et al. 2019;Osdaghi et al. 2018b).
In this study, a bacterial strain 20TX0166 T , pathogenic to onion was isolated from rotting onion bulb and characterized using polyphasic approach of taxonomy. Based on phenotypic and genotypic analyses, it is presumably a novel species of the genus Curtobacterium.

Isolation and pathogenicity assay
The bacterial strain 20TX0166 T was isolated from symptomatic tissue of a rotting onion bulb collected in Texas, USA (Fig. S1). For isolation, a 5 mm 2 piece of tissue along the margin of the rotting symptom was cut from the bulb, washed with sterile water, and crushed in 100 µl sterile water. The resulting suspension was streaked on nutrient agar medium and incubated at 25 °C. After successive culturing of single colonies 3 times, the pure culture was stored at -80 °C in 15% aqueous glycerol (v/v). The isolate was tested for pathogenicity by inoculating detached fleshy scales from a red onion bulb with a suspension of the bacterium, and by injecting the bacterial suspension into a yellow onion bulb. For the scale assay, a 10 µl suspension (10 8 CFU/ml) of the bacterium was placed over a wound made at the centre of the piece of scale (~ 3 cm × 4 cm) with a sterile needle, and the inoculated scale pieces were incubated at 25 °C for 10 days. For the bulb assay, a 0.5 ml aliquot of the bacterial suspension (10 8 CFU/ml) was injected into the upper shoulder of the bulb, and the bulb incubated at 25 °C for 12 days. Each assay was conducted with three replicate samples (detached bulb scale pieces or whole bulbs) and the experiment was repeated. An onion pathogenic Pantoea ananatis strain (PNA 97-1R) was used as a positive control treatment, whereas sterile phosphate-buffer saline (PBS) solution was used as a negative control treatment (Khanal et al. 2022). The legume pathogenic strain C. flaccumfaciens LMG 3645 T was also used to test for its pathogenicity in onion and compare along with the novel strain 20TX0166 T .

16S rRNA analysis and phylogeny
Genomic DNA of strain 20TX0166 T was extracted using the DNeasy Power Soil kit (Qiagen, MD, USA) and stored at − 20 °C. The universal primers 27F (5′-AGA GTT TGATCMTGG CTC AG-3′), 534R (5′-CGG TTA CCT TGT TAC GAC TT-3′), and 1492R (5′-CGG TTA CCT TGT TAC GAC TT-3′) were used to amplify and sequence the 16S rRNA gene (Walker et al. 2015;Webster et al. 2006;Weisburg et al. 1991;Wilson et al. 1990). Sequences generated by the three primers were trimmed and assembled to make a consensus sequence of 1,424 nucleotides (nt) using Geneious Prime 2020 (https:// www. genei ous. com/). The partial 16S rRNA gene sequence of strain 20TX0166 T was deposited and made publicly available (GenBank OK275102). The gene sequence of the novel strain was compared with those of the type strains of Curtobacterium species using reference RNA sequences (refseq_rna) database in BLAST (https:// blast. ncbi. nlm. nih. gov/ Blast. cgi) (Altschul et al. 1990). To examine the phylogenetic relationship of strain 20TX0166 T with related known species, the 16S rRNA gene sequences of eight Curtobacterium species were downloaded from GenBank and aligned using MUSCLE v3.8.425 (Edgar 2004) in Geneious Prime 2020. Using the aligned sequences (1388 nt), a maximum likelihood phylogenetic tree was constructed using PhyML 3.0 (Guindon et al. 2010) and visualized using MEGA v10.2.6 (Kumar et al. 2018). In PhyML 3.0, the HKY85 substitution model with 1,000 bootstrap replicates was used for the phylogenetic analysis (Lefort et al. 2017).

Whole genome analysis and phylogeny
Using the genomic DNA, the genome of the novel species was sequenced and assembled at CD Genomics (Shirley, NY, USA). Pair-end sequencing with an average sequencing depth of 100X was performed using the Illumina NovaSeq 6000 platform. The genome was compiled after trimming and assembling sequences from the raw fastq files using the default settings of Unicycler v0.4.8 (Wick et al. 2017) and SPAdes v3.13.0 (Bankevich et al. 2012). The genome was annotated functionally using PROKKA v1.14.5 (Seemann 2014). The annotated genome was used to generate a graphical illustration of the circular genome using DNAplotter v18.1.0 ( Fig. S2) (Carver et al. 2009). The genome sequence of the novel strain was compared with that of eight validly published Curtobacterium species to determine the average nucleotide identity (ANI) using the Orthologous ANI Tool (OAT software v0.93.1) based on OrthoANIu algorithm (Yoon et al. 2017) and pairwise ANI based on BLAST (ANIb) (Goris et al. 2007) using Python-based software-pyANI (Pritchard et al. 2016). Similarly, the digital DNA-DNA hybridization (dDDH) values were calculated using Type (Strain) Genome Server-TYGS (https:// tygs. dsmz. de/) (Meier-Kolthoff and Göker 2019). A phylogenetic tree based on whole genome sequence was created in the TYGS webserver, comparing the novel strain with the same eight species mentioned earlier. The Genome BLAST Distance Phylogeny method (GBDP) was used to calculate intergenomic distances. The GBDP method utilized the d5 distance formula and 100 distance replicates for each bacterial strain (Meier-Kolthoff et al. 2013). A phylogenetic tree was estimated using FASTME 2.1.6.1 followed by Subtree Pruning and Regrafting post-processing. To resolve the species delineation of 35 strains deposited as C. flaccumfaciens in the NCBI Genome database (Chen et al. 2021;Flanagan et al. 2013;O'Leary and Gilbertson 2020;Osdaghi et al. 2022;Tarlachkov et al. 2021;Vaghefi et al. 2021;Zhang et al. 2021), we used an approach adapted from Zhao et al. (2021), in which the ANI and dDDH values were computed after comparing with the C. flaccumfaciens type strain LMG 3645 T and the proposed C. allii type strain 20TX0166 T (Table S1).

Genome annotation and comparative genomics
The Rapid annotation and subsystems technology (RAST) online service was used to automatically annotation the bacterial genomes (Aziz et al. 2008). The annotated genomes were visualized, and comparative analyses of gene-associated features were performed using the SEED viewer (Overbeek et al. 2014). The strain 20TX0166 T was compared against 16 closely related Curtobacterium strains (Table S2), including strains from seven different species of Curtobacterium that are not known to be plant pathogens. C. flaccumfaciens is represented by two legume pathogenic strains, type strain LMG 3245 T and strain P990 (Chen et al. 2021). The remaining seven strains (CFBP 1384, Cff1037, CFBP 2403 were reported as Curtobacterium flaccumfaciens; but on our phylogenetic analysis, they are diverse enough to represent multiple novel species potentially but deposited as C. flaccumfaciens in the NCBI Genome database. Of these seven strains, six cause disease in plants except the strain Jub65, which was isolated from a nematode gut (O'Leary and Gilbertson 2020; Osdaghi et al. 2022;Vaghefi et al. 2021;Zhang et al. 2021).

Physiological and chemotaxonomic analyses
Phenotypic analyses were performed following the protocols described by Schaad et al. (2001) unless otherwise stated. A 24-h-old culture of strain 20TX0166 T was tested for the Gram reaction using the standard Gram staining technique and KOH test (Suslow 1982). Cells of the strain were stained with malachite green to evaluate whether the strain formed spores. Cultures were grown for 7 days in nutrient agar tubes in the absence of oxygen using a layer of mineral oil, to determine the facultative anaerobic status of the strain. Motility was observed by culturing the bacteria in a semi-solid growth motility test medium (per l, tryptone 10 g, sodium chloride 5 g, and agar 5 g). Catalase activity was tested with 3% hydrogen peroxide and oxidase activity was tested using Oxistrips and Oxidrops (Hardy Diagnostics, CA, USA). Imaging was performed using cultures that were grown for 24 h at 28 °C in yeast extract peptone glucose (YPG) broth (per l, glucose 10 g, yeast extract 5 g, peptone 5 g) and visualized using calibrated magnification with a 15C CCD camera on a JEOL 1200Ex transmission electron microscope (TEM) operated at 100 kV at the Texas A&M Microscopy and Imaging Centre following previously described protocols on negative staining (Scarff et al. 2018). A pectolytic assay to assess potato soft rotting capacity was conducted using flame-sterilized, peeled potato slices, and the tobacco hypersensitivity response was tested in planta on a leaf of each of three replicate tobacco plants along with relevant positive and negative control strains of bacteria and buffer (Schaad et al. 2001).
Biolog Gen III microplates, API 20NE strips (bio-Mérieux), and API Coryne strips (bioMérieux) were utilized to assess various biochemical and physiological characteristics using a fresh 24-h-old culture grown in NBYA medium at 30 °C per the manufacturer's instructions. For Biolog testing, an IF-A suspension [optical density (OD 600 ) of 0.022] was prepared and dispensed into the Gen III microplates, which were incubated at 30 °C. Results were read after 24, 48, and 72 h. For the API Coryne test, a bacterial suspension (turbidity greater than that of the McFarland #6 standard) was prepared using a 24-h-old culture, dispensed into the test strips, incubated at 33 °C, and results read after 24 and 48 h. For the API 20NE assay, a bacterial suspension (OD 600 of 0.09) was prepared in 0.9% sterile saline, dispensed into the test strips, incubated at 30 °C, and results read after 24 and 48 h. The OD 600 was measured using a Fluostar Omega microplate reader (BMG LABTECH Inc., NC, USA). These three tests were carried out a minimum of 3 times each, and results were compared with the results of type strains of the six closely related species of Curtobacterium. Among these species, C. flaccumfaciens is a known plant pathogen, though not known to be pathogenic to onion. The remaining five species are not known to be plant pathogens. Fatty acid analysis of the novel strain and Curtobacterium flaccumfaciens LMG 3645 T was conducted using 24-h-old culture grown on neutral (pH 7.0) 1/10 tryptic soy (TS) agar media (per l, tryptic soy broth 3 g, agar 15 g) incubated at 35 °C. The fatty acid methyl ester (FAME) profile was determined using the Sherlock Microbial Identification System (Sherlock V6.0B) at ESML Analytical Inc. (Cinnaminson, NJ, USA), following the previously described methods (Sasser 1990).

Pathogenicity assay
In both, scale and bulb pathogenicity assays, the bacterial strain 20TX0166 T caused a pathogenic response, inducing necrosis of the fleshy scales ( Fig. S3) and whole bulbs (Fig. 1). Two isolates (20TX0166b from fleshy scale and 20TX0166c from bulb) with similar morphology to the original isolate were re-isolated from each of the two pathogenicity tests. When these re-isolated daughter strains were tested further using the scale and bulb assays, the symptoms they produced were identical to that of the original strain. After sequencing the 16S rRNA gene, the daughter strains (GenBank OM863553 and OM863552) were confirmed as identical to the original strain. Our results showed that C. flaccumfaciens LMG 3645 T did not cause any pathogenic response on onion.
Phylogenetic analysis based on 16S rRNA gene sequence Upon searching on BLAST, the 16S rRNA gene sequence (GenBank accession number OK275102) of the strain 20TX0166 T shared 100% similarity with C. flaccumfaciens LMG 3645 T ( Table 1). The novel strain sequence also had > 99% similarity with the sequences of type strains of six other species of Curtobacterium. Such a high level of similarity makes it difficult to assign strain 20TX0166 T as one of the previously documented species based on 16S rRNA gene sequence similarity. In the phylogenetic tree created using 16S rRNA gene sequences, the novel strain did not produce separate branching from the C. flaccumfaciens LMG 3645 T , indicating the strain was most closely related to this species (Fig. 2). However, previous studies on Brevundimonas and Streptomyces have reported that multiple bacterial species with distinct ecological niches can share identical 16S rRNA gene sequences (Antony-Babu et al. 2017;Jaspers and Overmann 2004). Also, in a taxonomical classification study conducted using 1142 actinobacterial type strain genomes (Nouioui et al. 2018), it was suggested that similarity based on a single gene sequence cannot be fully relied for species identification. Thus, the genome of the novel strain isolated from rotting onion was analysed further to confirm the species identity.

Comparative genomics
The comparison of subsystem features within the genome was done using the RAST server and SEED viewer (Table S2) to indicate the presence or absence of gene-associated biological functions. The degree of similarity between the compared strains was somewhat congruent with the phylogenetic relatedness based on dDDH and ANI. Of the total 277 subsystem features listed across the 17 strains, 175 were common to all the strains representing the core genome of the genus Curtobacterium. Of the remaining 102 subsystem features, 64 were absent in the novel strain 20TX0166 T and sporadically distributed as strain-specific features. As anticipated, the novel strain showed more similarity with eight plant pathogenic strains than the non-pathogenic strains. Further, the novel strain shared maximum similarity with four strains: LMG 3645 T , P990, CFBP 1384, and Cff1037. The first two are C. flaccumfaciens strains, while the latter two are deposited as C. flaccumfaciens in the public database but need to be reclassified as they belong to the species cluster of the novel strain (Fig.  S4). Interestingly, the two subsystem features; alkanesulfonates utilization subsystem representing alkanesulfonates ABC transporter ATP-binding protein and formaldehyde assimilation: ribulose monophosphate (RuMP) pathway subsystem representing 6-phospho-3-hexuloisomerase (PHI) enzyme were found only in the novel strain, making it unique among others. It is known that ATP binding cassette (ABC) transporter in the alkanesulfonates uptake system plays a pivotal role in the pathogenesis of Xanthomonas citri in citrus canker disease (Sampaio et al. 2017;Tófoli de Araújo et al. 2013). Since onion is rich in antimicrobial sulfur compounds, genes related to sulfur metabolism conferring thiosulfinate tolerance are vital for causing center rot in onion by Pantoea ananatis (Stice et al. 2020). The presence of alkanesulfonates transporter protein capable of assimilating organic sulfur could be a key to the virulence of this strain in onion. Since formaldehyde is highly cytotoxic and found naturally in animals and plants, including onions, another unique genomic trait of the novel strain capable of assimilating formaldehyde could be of interest of pathogenesis (Nowshad et al. 2018). This mechanism found primarily in methylotrophic bacteria is valued in the biotechnology industry, as it not only detoxifies but also generates energy using single carbon organic compounds such as formaldehyde (Yurimoto et al. 2009). Genes regulating the RuMP pathway and PHI enzymes found in many bacterial pathogens other than methylotrophs that enable the detoxification of the formaldehyde generated in host-pathogen interfaces are essential for pathogenesis (Chen et al. 2016). The novel strain pathogenic to onion is very likely to benefit from the prevalence of this gene. These findings from genome analysis pave a new avenue to assess the factors contributing to the virulence of this novel strain that would give a deeper understanding of Gram-stain-positive plant pathogenic bacteria.

Fig. 3
Phylogenetic tree based on the whole genome sequence of strain 20TX0166 T of Curtobacterium isolated from a rotting onion bulb in Texas, USA, and whole genome sequences of closely related species of Curtobacterium. The tree was constructed using the TYGS web server, and the text was reformatted using MEGA v10.2.6 and Inkscape v1.1. Pseudo-boot-strap values at the branching nodes indicate the percentage of 100 replicates. The scale bar refers to the number of nucleotide substitutions per site. The superscript T denotes the type strain of the species used and the GenBank RefSeq assembly accession number of each strain is in parentheses Physiological and chemotaxonomic analyses The cells of strain 20TX0166 T were Gram-stainpositive (Fig. S5), non-spore-forming rods (Fig. S6), obligately aerobic, non-motile, catalase positive, and oxidase negative. The TEM images showed that bacterial cells were rod-shaped with an absence of flagella, 1.5-2.0 µm long × 0.6-1.0 µm wide (Fig. 4). No soft rotting of potato slices was observed after 24 h at 22 °C, indicating no pectolytic activity by strain 20TX0166 T . The tobacco hypersensitivity test was negative, in which the strain did not elicit a hypersensitive response, indicating the bacterium is non-pathogenic to tobacco.
Growth was observed at 4 to 39 °C but not at 41 °C on all media tested. The optimum temperature for growth was 25 °C and, although growth was observed at 4 °C, the rate of growth was slow as it took 5-7 days to form visible colonies at that temperature. After 24 h of incubation at 25 °C on YPGA and NBYA media, the bacterium produced 0.5-1.0 mm wide, smooth, creamy, round, and entire colonies. Colony size increased to 2-3 mm in 3 days and became somewhat mucoid with light yellow pigmentation (Fig. S7). Bacterial growth was observed at a pH of 5-10 with an optimal pH of 7. Growth was observed on YPGA medium with up to 9% NaCl, while optimum growth was observed at 0 to 1% NaCl.
Significant differences were noted in the Biolog, API Coryne, and API 20NE test reactions between the novel strain and C. flaccumfaciens LMG 3645 T , as well as the type strains of other related species (Table 2). The comprehensive results of the Biolog, API Coryne, and API 20NE of the novel strain and the closely related species are provided in Table S3. The notable features in these assays that make the novel strain atypical among the related species are briefly listed here. In the Biolog, both the strain 20TX0166 T and C. flaccumfaciens LMG 3645 T gave positive reactions in N-acetyl-d-galactosamine, d-arabitol, myo-inositol, l-alanine, and d-glucuronic acid, which makes them distinct from rest of the five species compared. In contrast, the strain 20TX0166 T had negative reactions in d-malic acid, l-malic acid, bromo-succinic acid, gamma-amino-butyric acid, and alpha-hydroxy-butyric acid; weakly positive reactions in N-acetyl neuraminic acid and methyl pyruvate; and positive reactions in d-glucose-6-PO4, d-fructose-6-PO4, gelatin, l-pyroglutamic acid, and l-galactonic acid lactone. In the API Coryne test, the typical features of the novel strain were positive reactions with pyrazinamidase, alkaline phosphatase, and the ability to ferment l-ribose, d-mannitol, d-lactose, and glycogen; while negative reactions with gelatin hydrolysis and fermentation of d-maltose. Likewise, in API 20NE test, the novel strain assimilated potassium gluconate, but not d-mannitol, N-acetyl-glucosamine, malic acid, and trisodium citrate. These results further support that the onion strain is a novel species of Curtobacterium.
The major fatty acids profile from FAME analysis were anteiso-C 15:0 and anteiso-C 17:0 . Iso-C 15:0 , iso-C 16:0 , summed feature 8 (C 18:1 ω7c and/or C 18:1 ω6c), and iso-C 17:0 were detected in lower amounts (Table 3). At least 11 more fatty acid profiles were detected in trace amounts (< 1% of total). Though the cellular fatty acid composition of the novel strain was fairly similar to C. flaccumfaciens LMG 3645 T , four profiles detected in trace amounts in the former were not detected in the latter. In an earlier study (Aizawa et al. 2007), 11-cyclohexyl undecanoic acid (ch-C 17: 0 ) was detected as a major fatty acid component in three species of Curtobacterium (C. flaccumfaciens, C. pusillum, and C. ammoniigenes), whereas it was not detected in both the novel strain and C. flaccumfaciens in our study. Comparing the fatty acid profiles from previous studies (Kim et al. 2009(Kim et al. , 2012, ch-C 17: 0 was not detected in any of the five other Curtobacterium species (C. citreum, C. luteum, C. albidum, C. herbarum, and C. oceanosedimentum). These results suggest the novel strain belongs to the genus Curtobacterium, but there was enough variation to classify the strain 20TX0166 T under any predefined species, supporting the evidence that it is a novel species.

Conclusion
In brief, based on physiological, chemotaxonomic, genotypic, and phylogenetic characterizations of strain 20TX0166 T , it is found to be different from all the most closely related species of the genus Curtobacterium. Thus, we suggest that strain 20TX0166 T should be assigned as a novel species of Curtobacterium, for which the name Curtobacterium allii is proposed.