Characterization of genes required for the pathogenicity of Pectobacterium carotovorum subsp. carotovorum Pcc21 in Chinese cabbage

Pectobacterium carotovorum subsp. carotovorum is a well-known plant pathogen that causes severe soft rot disease in various crops, resulting in considerable economic loss. To identify pathogenicity-related factors, Chinese cabbage was inoculated with 5314 transposon mutants of P. carotovorum subsp. carotovorum Pcc21 derived using Tn5 transposon mutagenesis. A total of 35 reduced-virulence or avirulent mutants were isolated, and 14 loci were identified. The 14 loci could be functionally grouped into nutrient utilization (pyrD, purH, purD, leuA and serB), production of plant cell-wall-degrading enzymes (PCWDEs) (expI, expR and PCC21_023220), motility (flgA, fliA and flhB), biofilm formation (expI, expR and qseC), susceptibility to antibacterial plant chemicals (tolC) and unknown function (ECA2640). Among the 14 genes identified, qseC, tolC and PCC21_023220 are novel pathogenicity factors of P. carotovorum subsp. carotovorum involved in biofilm formation, phytochemical resistance and PCWDE production, respectively.


INTRODUCTION
Pectobacterium carotovorum subsp. carotovorum causes soft rot disease in cabbage, potato, onion, radish and other crops during cultivation, transportation and storage, resulting in considerable economic damage. It exists on plant surfaces and in soil, and may enter the host through wound sites or natural openings. After successfully invading a plant, it resides in the intercellular space or vascular tissue, where it produces plant cell-wall-degrading enzymes (PCWDEs) in a disease-promoting environment (Perombelon, 2002;Toth et al., 2003).
Extracellular maceration enzymes, such as polygalacturonase (Peh), pectate lyase (Pel), cellulase (Cel), xylanase and protease (Prt), are important pathogenic factors of soft rot. During pathogen infection, these enzymes degrade plant structures composed of pectin, cellulose and hemicelluloses fibres, leading to plant cell necrosis and tissue maceration (Abbott & Boraston, 2008). A number of regulatory factors that control these enzymes are also important in soft rot pathogenesis. P. carotovorum subsp. carotovorum uses a cell-density-dependent cell-to-cell communication system called quorum sensing (QS) to regulate many genes and pathogenesis. In the QS system of Gram-negative bacteria, ExpI/ExpR is responsible for the synthesis of acylhomoserine lactone (AHL), which regulates the production of extracellular enzymes (Henke & Bassler, 2004;Nasser et al., 1998;Pirhonen et al., 1993;Whitehead et al., 2002). Additionally, flhD, gacA and hexA participate in the regulatory pathway for PCWDE production (Cui et al., 2008). Several more genes, including cysQ, hel, yheL and trkA, are reported to be involved in the production/secretion of PCWDE (Laasik et al., 2005).
In addition to the above-mentioned factors, there are further requirements for P. carotovorum subsp. carotovorum survival in a host. Motility and nutrient uptake are important for bacterial pathogenicity in host plants. Pirhonen et al. (1991) reported that transposon mutants of Erwinia carotovora subsp. carotovora that were impaired in the motility-related genes had reduced virulence. Components of the flagella export apparatus are evolutionarily related to the type III secretion system (Aizawa, 2001) used for the transport of virulence factors into host plant cells; and the development of flagella systems is closely associated with the regulation or secretion of virulence factors (Chatterjee et al., 2009). Nutrient uptake from host plant tissue by plant pathogens is also a critical process in pathogenesis (Urbany & Neuhaus, 2008). There are some reports that purine and pyrimidine metabolism is related to pathogenicity (Chatterjee & Sonti, 2005;Kim et al., 2003) and gluconate metabolism is required for virulence of the soft rot pathogen P. carotovorum subsp. carotovorum (Mole et al., 2010).
Soft rot bacteria must attach to the plant surface and resist antibacterial chemicals produced by plants. Recently, it was reported that production of cellulose enhances plant surface colonization and is required for resistance to chlorine treatments in Dickeya dadantii (Prigent-Combaret et al., 2012). In this bacterium, bcs operon is involved in the cellulose synthesis (Jahn et al., 2011). However, the bacterial genes associated with attachment or survival are not well characterized in P. carotovorum subsp. carotovorum, which is a necrotroph. The interactions of necrotroph pathogens with plants are not as well understood as those of biotrophs. Previous studies have focused on PCWDE enzyme production in P. carotovorum subsp. carotovorum (Cui et al., 2008;Laasik et al., 2005), but many more factors are considered to be involved in its complex pathogenesis.
In this study, we used Tn5 transposon mutagenesis to select mutants of P. carotovorum subsp. carotovorum with altered ability to produce soft rot symptoms. Among a total of 5314 transposon-mediated mutants screened on Chinese cabbage, 35 mutants with reduced or no pathogenicity were isolated and 14 of those with mutations at different loci were characterized. The altered genes were associated with various functions, such as nutrient utilization, QS, motility, secretion and regulation. We report the identification of new candidate pathogenicity factors in the genus Pectobacteria.

METHODS
Bacterial strains, media and cultivation conditions. The bacterial strains and plasmids used or constructed in this study are shown in Table 1. P. carotovorum subsp. carotovorum Pcc21 (isolated in Korea) and its mutants were cultured and tested at 28 uC in LB medium (1 % tryptone, 0.5 % yeast extract, 1 % NaCl, pH 7.5). Escherichia coli were cultured at 37 uC. To compare mutant bacterial growth, Chinese cabbage extract, Chinese cabbage exudates and M9 salts minimal medium supplemented with various carbon sources were used. To prepare enzyme assay samples, the strains were cultured overnight in LB medium at 28 uC. When required, medium was supplemented with rifampicin, kanamycin or ampicillin at a final concentration of 50 mg ml 21 , or with designated chemicals at the appropriate concentration.
Transposon mutagenesis. A P. carotovorum subsp. carotovorum Pcc21 mutant library was constructed using an EZ-Tn5 ,R6kcori/ KAN-2.Tnp Transposome TM kit (Epicentre Biotechnologies) according to the manufacturer's protocol. Briefly, the transposome complex was introduced into P. carotovorum subsp. carotovorum Pcc21 by electroporation and the cells were spread on an LB agar plate supplemented with kanamycin and rifampicin. A total of 5314 singlecolony-forming mutants were selected and stored at 280 uC.
Complementation of mutants. To construct 14 plasmids for complementation analysis of each mutant, genes corresponding to the mutated regions of each mutant were amplified with PCR using specific primers listed in Table S1 (available in Microbiology Online). Analyses of the mutated genes with the database of BioCyc (Caspi et al., 2010) and of OperonBD (Pertea et al., 2009) predicted that the mutated genes in Pcc21-M2, Pcc21-M5, Pcc21-M8, Pcc21-M9, Pcc21-M10, Pcc21-M15 and Pcc21-M16 are monocistronic, but those in Pcc21-M4, Pcc21-M13, Pcc21-M14, Pcc21-M19, Pcc21-M20, Pcc21-M21 and Pcc21-M22 are polycistronic. In the case of polycistronic genes, the whole operon containing the mutated gene was amplified and cloned. Pcc21-M4 and Pcc21-M20 belonged to the same operon. The mutated gene in Pcc21-M22 was the last gene of the operon, and both the whole operon and the corresponding gene were amplified and cloned for further experiments (Table S2). The PCR products were purified and cloned into pGEM-T easy vector (Promega) or into pGFPuv vector (Clontech), replacing the GFP gene with one of the restriction enzyme sets (SphI-SpeI, PstI-ApaI or XbaI-SpeI). The constructed plasmids, pDH200-213, are listed in Table S2. The constructed plasmids were transformed into each corresponding mutant by an electroporation or conjugation method, as described previously (Lee et al., 2008;Sambrook et al., 1989). The E. coli S17-1 l pir strain was used as a conjugal donor.
Pathogenicity test. Fresh, woundless Chinese cabbage was rinsed twice with 70 % ethanol. All mutants were diluted and fixed at a cell density equivalent to an optical density at 600 nm (OD 600 ) of 0.5. A small scar was made on the Chinese cabbage with forceps, and 10 ml of each mutant was inoculated onto the injured section. The Chinese cabbages were placed into sterilized plastic boxes with sufficient moisture and stored at 28 uC.
DNA sequence determination and analysis. Mutant genomic DNA was digested with EcoRI or KpnI, which did not have recognition sites within the transposon. Fragmented genomic DNA was self-ligated and used to transform E. coli EC100 pir+ by the CaCl 2 method (Sambrook et al., 1989). Plasmids containing the mutated region were isolated using a DNA-spin plasmid DNA isolation kit (Intron Biotechnology), and the inserts were sequenced using the specific primers KAN-2 FP-1 and R6KAN-2 RP-1 (Epicentre). To identify nucleotide sequences similar to the mutated regions, we searched the GenBank database using BLAST (http://www. ncbi.nlm.nih.gov/). Mutant growth in Chinese cabbage extract and exudate. To prepare disinfected Chinese cabbage extract, Chinese cabbage was ground using a mixer. Debris was removed by centrifugation at 5800 g for 30 min, and the supernatant was collected. This step was performed at least five times. Finally, the collected supernatant was sterilized using a Stericup (0.22 mm pore size, GP Express Plus Membrane, Millipore). Chinese cabbage exudate was prepared according to the following method. White leaf regions were rinsed with water and cut uniformly. Chinese cabbage exudate was collected by centrifugation at 3300 g for 5 min and filtered through filter paper to remove debris. The filtrate was sterilized by passing through a 0.22 mm pore CA membrane filter (Corning), which eliminated small debris and bacteria. Sterilized Chinese cabbage extract or exudate containing appropriate antibiotics was inoculated with each mutant, and growth was measured with a Lambda 25 UV/VIS spectrometer (PerkinElmer).
In vivo multiplication analysis. To track the mutant multiplication path on Chinese cabbage, we introduced pGFPuv plasmid (Clontech) into the WT Pcc21 and into each mutant, using the CaCl 2 method (Sambrook et al., 1989). These strains were inoculated onto Chinese cabbage using the same method as applied in the pathogenicity test. GFP expression was measured every 12 h using a luminescent image analyser (JP/LAS-3000, FujiFilm). Extracellular enzyme assays. Bacterial cells were removed from overnight cultures and the supernatants were used for enzyme assays. The compositions of the media for the semiquantitative enzyme activity assays were as follows: Pel assay medium contained 1 % polygalacturonic acid (PGA), 1 % yeast extract, 0.38 mM CaCl 2 and 100 mM Tris/HCl, pH 8.5; Peh assay medium contained 1 % PGA, 1 % yeast extract, 2.2 mM EDTA and 110 mM sodium acetate, pH 5.5; Cel assay medium contained 1 % carboxymethyl cellulose and 25 mM sodium phosphate, pH 7.0; and Prt assay medium contained 1 % skim milk and 0.1 % yeast extract (Chatterjee et al., 1995;Wery et al., 2003). All assay media were supplemented with 0.8 % agarose and 0.2 % sodium azide. In each plate, wells were made using a No. 2 cork borer, and the bottoms of the wells were covered with molten 0.8 % agarose containing 0.2 % sodium azide. Overnight cultures grown in LB medium were centrifuged at 5800 g for 5 min, and 30 ml of the supernatant was added to each well. After incubation at 28 uC for 16-18 h, 4 N HCl was poured onto the Peh and Pel enzyme assay plates, and the clear zones were measured. The Cel assay plates were developed with 0.1 % Congo red solution for 15-30 min and washed several times with 1 M NaCl until clear zones were visible around the holes. The Prt assay plates were measured after 36 h of incubation, revealing clear zones without any further treatment.
Pel activity was measured as described by Starr et al. (1977) with some modifications. Briefly, 10 ml supernatant was mixed with 990 ml Pel reaction mixture (0.1 M Tris/HCl, pH 8.5, 2.2 mM CaCl 2 and 5.75 mg ml 21 PGA) and then incubated at room temperature. Absorbance changes were measured at 235 nm for 10 min using a Lambda 25 UV/VIS spectrometer (PerkinElmer). The activity was calculated as A 235 nm h 21 OD 600 nm 21 and a relative ratio of each mutant versus WT strain was calculated.
Motility test on agar plates and transmission electron microscopy. Motility tests were performed on semi-solid LB medium (0.3 % agar). The plates were inoculated with bacteria by stabbing and incubated at 28 uC. Motility was compared after 12 and 24 h of incubation (Sperandio et al., 2002).
For transmission electron microscopy, cells freshly grown on Tryptic Soy Agar plates were resuspended in sterile water. Cells were negatively stained with 0.1 % (w/v) uranyl acetate and observed with a transmission electron microscope (Carl Zeiss LEO 912 AB).
Quantitative biofilm formation assessment. To assess biofilm formation, cells were cultured overnight at 28 uC in LB and then diluted with fresh LB medium to a cell density equivalent to an OD 600 of 0.2-0.3 (~10 8 c.f.u. ml 21 ). The wells of 96-well flat-bottom polystyrene microplates (SPL) were filled with 200 ml of diluted cells

Isolation of pathogenicity-defective mutants and identification of mutated regions
To isolate novel pathogenicity-related genes, we used a simple mutagenesis system based on a transposontransposase synaptic complex (transposome). Tn5 transposons contain an R6Kc conditional origin of replication, and the Tn903 kanamycin resistance gene was employed as a random mutational element. After screening 5314 mutants, we isolated 35 mutants that showed reduced or no pathogenicity in Chinese cabbage.
The transposon insertion sites from the 35 clones were sequenced, and a BLAST search of the GenBank database showed that the 35 mutants carried Tn5 transposons in 14 different loci (Table 2). Pathogenicity-related loci identified in the P. carotovorum subsp. carotovorum genome were predicted to encode proteins involved in various functions, such as nutrient utilization, QS, motility, secretion and regulation. Of the 35 mutants, 14 typical mutants representing individual genes were selected and we carried out complementation tests to confirm the ORFs of the genes. We introduced constructed clones into corresponding mutants and then a pathogenicity test was performed (Table S2). The pathogenicity against Chinese cabbage of all mutants except Pcc21-M4 and Pcc21-M20 containing the corresponding WT gene or operon was successfully restored (data not shown). Introduction of pDH201 into Pcc21-M4 and Pcc21-M20, representing purH and purD genes, in one operon failed and the reasons remain unclear.

Bacterial multiplication in Chinese cabbage
Most mutants did not induce soft rot symptoms in Chinese cabbage. However, strains with mutations in flhB, qseC or tolC homologous genes showed soft rot symptoms 24-48 h later than the other mutants (Table 3). We tested whether bacterial growth is a requirement for visual soft rot symptoms or not by monitoring bacterial growth with bacteria carrying the GFP gene. Green fluorescence was observed only around areas showing soft rot symptoms, but not observed beyond those symptoms. Thus, it appears that when the bacteria multiply, soft rot symptoms occur at the same time (Fig. 1). All mutants multiplied under adequate nutritional conditions (such as in the presence of plant extracts), but limited nutrient availability (such as in the presence of plant exudates) inhibited the growth of several mutants, particularly those carrying mutations in the pyrD and purH homologous region (Table 3).

Mutants with altered nutrient utilization
The ability of the mutants to use a carbon source was investigated by cultivating the mutants in M9 minimal medium with the separate addition of glycerol, sucrose, fructose, glucose and mannitol. Strains carrying mutations at five different genes, Pcc21-M2, Pcc21-M4, Pcc21-M14, Pcc21-M19 and Pcc21-M20, did not grow on minimal medium with all five different carbon sources (Table 3). Sequence analysis of five mutated genes revealed that they are associated with nutrient utilization. Pcc21-M2 had transposon insertion in a class 2 dihydroorotate dehydrogenase homologous region, which is related to pyrimidine synthesis (Table 2) (Aleksenko et al., 1999;West, 2005). Mutants Pcc21-M4 and Pcc21-M20 had transposon insertions in two purine biosynthesis protein genes, bifunctional phosphoribosylaminoimidazole carboxamide formyltransferase/IMP cyclohydrolase and phosphoribosylglycinamine-glycine lygase, respectively (Chatterjee & Sonti, 2005;Dougherty et al., 2006). Pcc21-M14 and Pcc21-M19 had mutations in 2-isopropylmalate synthase and phosphoserine phosphatase, which are involved in the biosynthesis of leucine and serine, respectively (Table 2) (Patte et al., 1999;Wang et al., 2007;Zou et al., 2007). These results suggest that the virulence deficiency of mutants with insertions at five different loci is due to an inability to produce pyrimidine, purine, leucine or serine.
P. carotovorum subsp. carotovorum replicates in the intercellular space until conditions favour disease development. It is not known whether non-pathogenic mutants continue to multiply in vascular tissue or not. The 14 mutants carrying GFP did not multiply efficiently in the Chinese cabbage intercellular space as expanded GFP expression was not observed in Chinese cabbage (Fig. 1). However, all mutants (excluding the five mutants carrying transposon insertions in nutrient utilization-related genes) multiplied in Chinese cabbage exudates (Table 3), suggesting that other factors, as well as nutrient utilization, may be involved in the pathogenicity-defective mutants.

Mutants with altered production and secretion of exo-enzymes
To determine whether the loss in pathogenicity in specific mutants is caused by reduced production or secretion of exo-enzymes, the 14 mutants were tested for the production and secretion of four different PCWDEs. Only three of the 14 did not produce or secrete exo-enzymes (Table 4 and Fig.  2). These mutants lost over 90 % of Pel activity, which is a major PCWDE of plant pathogen bacteria, compared with the WT strain. Two mutants carried transposon insertions in the QS-associated genes, expI and expR, and the other had an insertion in the PCC21_023220 gene with an unknown function. The disruption of expI or expR affected the production or secretion of exo-enzymes because exoenzyme production and secretion are controlled by a QS signal produced when the bacteria reach a specific cell density. The disrupted gene in Pcc21-M10 (expI) is involved in the synthesis of AHL, which is a major QS signal. We also isolated strains with mutations in expR (Pcc21-M15), which is known as a AHL receptor, the transcription regulator of rsmA (rsm, repressor of secondary metabolites) ( Table 2) (Burr et al., 2006;Chatterjee et al., 1995;Cui et al., 2006;Fray et al., 1999). Pcc21-M16 had a transposon insertion at the locus PCC21_023220. The PCC21_023220 gene encodes a putative DNA-binding protein with a YheO-like PAS (perarnt-sim) domain. PAS domains are found in a variety of prokaryotes and eukaryotes as important signalling modules (Gu et al., 2000;Taylor & Zhulin, 1999). These findings suggest that PCC21_023220 may be involved in the production or secretion of exo-enzymes through unknown pathways as a regulatory protein (Table 4). Many bacteria carry this gene, but its function has not been determined. This is the first report that PCC21_023220 controls the production or secretion of PCWDE.
It is interesting to note that P. carotovorum subsp. carotovorum Pcc21 has more than eight pectate lyases, three polygalacturonases, two cellulases and several putative PCWDE-associated genes (GenBank accession no. NC_018525.1). An insertional mutation in just one of these genes was not sufficient to cause deficient pathogenicity. None of our mutants had insertions in the gene encoding PCWDE.

Mutants with altered motility
Motility is an important determinant for the pathogenesis of Pectobacteria (Cui et al., 2008;Hossain et al., 2005). We investigated the motility of all 14 mutants on semi-solid agar plates. All mutants, except Pcc21-M5, Pcc21-M9 and Pcc21-M21, were motile. Mapping of the transposon insertions of non-motile mutants showed insertions in flgA, fliA and flhB, which code for proteins involved in flagella biosynthesis (Ferris et al., 2005;Kutsukake et al., 1990;Sperandio et al., 2002). Pcc21-M5 had a mutation in the flgA homologous region, which codes for the flagellar basal body P-ring. Pcc21-M9 had a mutation in the fliA homologous region, which encodes a sigma factor for flagella synthesis, and Pcc21-M21 had a mutation in flhB, which encodes a flagella biosynthesis protein (Table 4). In Pcc21-M22, a transposon was located in a qseC homologous gene, which encodes the sensor kinase in a two-component regulatory system in E. coli (EHEC) and Salmonella Typhimurium (Clarke et al., 2006;Hadjifrangiskou et al., 2011). In the absence of QseC, QseB is constitutively phosphorylated and represses virulence-associated genes, including those for type 1 pili, curli and flagella (Hadjifrangiskou et al., 2011). However, Pcc21-M22 has a movement radius nearly equal to that of WT Pcc21. Transmission electron microscopy of these mutants revealed that Pcc21-M5, Pcc21-M9 and Pcc21-M21 had no flagella, while Pcc21-M22 retained flagella (Fig. 3).

Mutants with altered biofilm formation
Biofilm formation is crucial for the pathogenesis of plant and animal bacterial pathogens. We evaluated the biofilm formation of mutants using the crystal violet staining assay, which has been used to assess biofilm formation by a wide variety of bacteria (Koczan et al., 2009;O'Toole & Kolter, 1998;Pérez-Mendoza et al., 2011;Wakimoto et al., 2004). After growing for 48 h in a 96-well assay plate, P. carotovorum subsp. carotovorum Pcc21 demonstrated visual evidence of biofilm formation, whereas there was an absence of biofilm formation by Pcc21-M10 (expI), Pcc21-M15 (expR) and Pcc21-M22 (qseC). Next, we quantified the amount of biofilm formation. The amount of biofilm formed by the Pcc21-M22 (qseC) mutant was only 20 % of the amount formed by the WT bacteria (Table 4). This result supports recent findings that qseC is involved in biofilm formation (Durham-Colleran et al., 2010;González Barrios et al., 2006;Novak et al., 2010). The mutation in the qseC gene also reduced pathogenicity. Interestingly, biofilm  Pathogenicity of P. carotovorum formation by P. carotovorum subsp. carotovorum Pcc21 is perhaps controlled by two different QS signal pathway systems; one is expI and expR, and the other is qseC and PCC21_042560. However, the QS signal system genes expI and expR appeared to control both PCWDE and biofilm production, and mutations in these genes abrogated pathogenicity. Three mutants with defective flagella biosynthesis exhibited reduced biofilm formation. The flgA and fliA mutants produced less than 50 % of the biofilm produced by WT P. carotovorum subsp. carotovorum Pcc21. These results suggest that flagella play a role in biofilm formation in P. carotovorum subsp. Carotovorum, as in E. coli. It has been known that motility is important for normal biofilm formation in E. coli (Pratt & Kolter, 1998).

Mutants with different resistance to antibacterial plant chemicals
Plants produce a variety of chemicals with antimicrobial properties (Federici et al., 2004;Simões et al., 2009), and many bacteria have evolved with mechanisms that tolerate or eliminate plant-derived toxins (Federici et al., 2004). Given that none of the mutants grew efficiently in Chinese cabbage, the effects of plant-derived chemicals on bacterial growth were evaluated. Of the mutants, only the tolC mutant showed increased sensitivity to half of the plantderived chemicals that were evaluated (Table 5). The MICs of berberine, rhein and genistein in the tolC mutant strain were at least 128-, 64-and 64-fold lower, respectively, than the MICs in WT bacteria (Table 5). Rhein is an antimicrobial of rhubarb. Genistein is an isoflavonoid produced as a precursor in the biosynthesis of antimicrobial pytoalexins and phytoanticipins. The alkaloid berberine is a common component of several plant species, particularly in the Berberidaceae family. Although berberine has antibiotic properties, it is exported by bacterial multidrug resistance pumps. The susceptibility of the tolC mutant to berberine increased 128-fold, while susceptibility to phenolic compounds, such as coumaric acid, cinnamic acid and pyrithione, was not affected. Different susceptibilities to diverse phytochemicals suggest that TolC may play a role in antibacterial resistance via efflux systems. Recently, several researchers suggested a relationship between TolC and a multidrug resistance efflux pump in other plant-pathogenic bacteria (Al-Karablieh et al., 2009b;Barabote et al., 2003). Considering that Pcc21-M8 (tolC) showed decreased exo-enzyme production in some tests rather than that of the WT strain, the type I secretion system and the type II secretion system are likely to be involved in P. carotovorum subsp. carotovorum infection of host cells.  The Pcc21-M13 mutant showed PCWDE production comparable to the WT and retained motility, but produced about 60 % of biofilm compared to the WT (Table 4). Pcc21-M13 has a transposon insertion in the ECA2640 homologous gene and the function of this uncharacterized gene in pathogenicity remains to be studied.
In summary, we identified three novel genes associated with the pathogenicity of P. carotovorum subsp. carotovorum: qseC, tolC and PCC21_023220. The function of PCC21_023220 as a regulator remains unknown. We are currently investigating the role of PCC21_023220 as a regulator of PCWDE production. Although many genes that function in P. carotovorum subsp. carotovorum pathogenicity have been identified, many more uncharacterized genes important for pathogenesis are waiting to be found.