Tn5 transposon mutagenesis in Acidovorax citrulli for identification of genes required for pathogenicity on cucumber

Authors


E-mail: wxshan@nwsuaf.edu.cn

Abstract

An Acidovorax citrulli–cucumber pathosystem was established through which A. citrulli mutants with altered pathogenicity, generated by transposon mutagenesis, were identified on cucumber cotyledons. The A. citrulli group I strain FC440 was shown to grow faster in cucumber leaf tissues than a group II strain and was used for Tn5 transposon mutagenesis. A total of 2100 Tn5 insertional mutants were generated, and analysis of the mutant library showed that the transposon insertions were single, independent and stable. A conserved non-flagellar type III secretion system (NF-T3SS) ATPase gene hrcN was identified and confirmed to be essential for pathogenicity and functionality of NF-T3SS in Acitrulli. Comparative sequence analysis of the HrcN protein and its homologues in other representative bacterial plant pathogens revealed that the NF-T3SS of Acitrulli is close to that of Ralstonia solanacearum and Xanthomonas campestris, but distant from that of Pseudomonas syringae and Erwinia amylovora. The generated Tn5 insertional mutant collection is valuable for identification of genes required for A. citrulli pathogenesis, and the established A. citrulli–cucumber pathosystem will facilitate an improved understanding of A. citrulli biology and pathology.

Introduction

Acidovorax citrulli is the causal agent of bacterial fruit blotch (BFB), a sporadic but devastating disease of cucurbits worldwide (Schaad et al., 1978, 2008; Willems et al., 1992). Since the first natural outbreaks in 1987 and 1989 in commercial watermelon (Citrullus lanatus) fields on the Mariana Islands in the South Pacific and in Florida, USA (Wall & Santos, 1988; Somodi et al., 1991), BFB has spread worldwide and has been observed to cause diseases on other cucurbits, such as honeydew melon (Cucumis melo) (Isakeit et al., 1997), citron melon (Citrullus lanatus var. citroides) (Isakeit et al., 1998; Pallen et al., 2005), pumpkin (Cucurbita pepo) (Langston et al., 1999; Walcott et al., 2004), squash (Cucurbita maxima) (Walcott et al., 2004) and cucumber (Cucumis sativus) (Martin et al., 1999; Burdman et al., 2005; Liu et al., 2009).

Acidovorax citrulli is a Gram-negative, rod-shaped bacterium that can be transmitted through seed. Under high relative humidity and high temperature, the bacteria spread rapidly throughout transplant greenhouses and in the field, leading to seedling lesions, blight or fruit rot (Schaad et al., 2003). BFB has great potential to cause significant economic losses to cucurbit production, and has been responsible for losses of up to 50–90% of marketable fruits in some watermelon fields (Latin & Rane, 1990; Somodi et al., 1991). Currently, strategies for managing BFB are limited and there are no available resistant commercial cultivars, making BFB a serious threat to the cucurbit industry worldwide (Bahar & Burdman, 2010). Understanding the complex interactions between the host plant and the pathogen is critical for the development of effective disease control measures. Unfortunately, the genetic and biochemical mechanisms employed by A. citrulli to ensure infection and colonization in the host plant are largely unknown.

Current knowledge on the biology and pathology of Acitrulli is limited. One of the important findings is the identification of pathogen differentiation in population structure and host range, which is useful for screening for BFB resistance and understanding pathogenicity mechanisms of A. citrulli, and for providing the framework for further investigation of the evolutionary, ecological and epidemiological significance of the pathogen (Walcott et al., 2000). Since the end of the 1980s, BFB has been a serious threat, mainly for watermelon. In 1991, it was reported that Acitrulli isolated from the 1989 BFB outbreak in Florida produced a hypersensitive response on tobacco and tomato, whereas the subspecies type strain did not (Somodi et al., 1991). Subsequent discoveries have aroused serious concern worldwide: trends of devastating BFB outbreaks in other non-watermelon cucurbits (Isakeit et al., 1997, 1998; Langston et al., 1999; Martin et al., 1999), and the discovery of two subgroups (group I and group II) within Acitrulli with significant differences in host range and pathogenicity (O’Brien & Martin, 1999; Walcott et al., 2000, 2004; Burdman et al., 2005). It was found that group I strains could moderately infect all cucurbit hosts, and group II strains were more aggressive on watermelon than on other hosts. These discoveries indicated that the population structure of the pathogen has been changing and the host range of Acitrulli has been expanding.

Another important finding is the identification of genes required for Acitrulli pathogenicity. The complete genome sequence of the group II strain AAC00-1 of A. citrulli was determined in 2007 (released by the Joint Genome Institute; GenBank accession number NC_008752) which facilitated the study of the genetic basis of BFB pathogenesis. The type III secretion system (T3SS) has been discovered in A. citrulli through genome sequencing (GenBank accession number AY898625) and was confirmed to be essential for pathogenicity in A. citrulli on the host plants watermelon and melon by gene disruption experiments of the T3SS structural genes hrcR and hrcC (Ren et al., 2009; Johnson et al., 2011). In addition, the type IV secretion system has been identified as required for virulence (Bahar et al., 2009). However, successful establishment of A. citrulli in the host tissues requires the coordinated activity of many genes, for which the identities and modes of action are still largely unknown.

Several research groups have used A. citrulli–watermelon or A. citrulli–melon interaction models (Bahar et al., 2009; Ren et al., 2009; Johnson et al., 2011) to explore the molecular basis of BFB pathogenesis. The reported findings have focused mainly on the pathogenesis on these two hosts. However, it should be noted that both group I and II strains resulted in higher percentages of symptomless seedlings in watermelon and melon than in cucumber in seed transmission assays; the percentages of symptomless seedlings with group I and II strains were 13% and 79%, respectively, for watermelon; 9% and 90%, respectively, for melon; and only 1·7% and 46·2%, respectively, for cucumber (Burdman et al., 2005). These data suggest cucumber is probably a more suitable host for investigating BFB pathogenesis. With a relatively small and sequenced genome (Huang et al., 2009) and a higher transformation efficiency (Selvaraj et al., 2010) than other cucurbits, cucumber emerges as a model species in cucurbits. It is therefore very likely that the A. citrulli–cucumber interaction system will promote an increased understanding of the molecular basis of the pathogenicity of this pathogen.

The aim of this study was to explore the A. citrulli–cucumber interaction as a model pathosystem for the study of the molecular mechanisms of BFB pathogenesis. The Acitrulli–cucumber system was used to determine pathogenicity and identify mutants of A. citrulli with altered pathogenicity. In addition, the HrcN protein of Acitrulli was compared with its homologues in other representative plant pathogenic bacteria.

Materials and methods

Bacterial strains, plasmids and growth conditions

The bacterial strains and plasmids used in this study are listed in Table 1. Acidovorax citrulli group I strain FC440 and group II strain AW0601 were originally isolated from diseased cantaloupe (Cucumis melo var. cantalupensis) and watermelon (Citrullus lanatus) in Xinjiang, China, respectively (Xu, 2007). Acidovorax citrulli strains were grown in nutrient broth (Difco) or on King’s medium B (KMB) (King et al., 1954) at 28°C. Escherichia coli cells were cultivated at 37°C in Luria–Bertani (LB) liquid medium (Sambrook & Russell, 2001). Plasmids were introduced into E. coli by electroporation (Sambrook & Russell, 2001) and into A. citrulli by diparental conjugation (De Bruijn & Rossbach, 1994). Antibiotics were added to the media at the following final concentrations, if not stated otherwise: ampicillin, 30 μg mL−1; kanamycin, 50 μg mL−1; tetracycline, 10 μg mL−1.

Table 1.   Bacterial strains and plasmids used in this study
Strain/plasmidCharacteristicsReference/source
  1. AmpR, KanR and TetR = resistant to ampicillin, kanamycin and tetracycline, respectively.

Acidovorax citrulli strains
 FC440AmpR; wildtype, group I strainN. W. Schaad
 AW0601AmpR; wildtype, group II strainB. S. Hu
 FC440(ΔhrcN)AmpR; KanR; FC440 mutant defective in hrcNThis work
 FC440(ΔhrcN +pHC60)AmpR; KanR; TetR; FC440 (ΔhrcN) strain containing expression vector pHC60This work
 FC440(ΔhrcN + hrcN)AmpR; KanR; TetR; FC440 (ΔhrcN) complemented with hrcN gene expressed by vector pHC60This work
Escherichia coli strains
 DH10BFendA1 recA1 galE15 galK16 nupG rpsL ΔlacX74 Φ80lacZΔM15 araD139Δ(ara, leu)7697 mcrAΔ(mrr-hsdRMS-mcrBC) λ-Invitrogen
 S17-1λ pirE. coli 294, thi RP4-2-Tc::Mu-Kan::Tn7 chromosomally integratedSimon et al. (1983)
Plasmids
 pRLl063aBroad-host-range plasmid Kan::Tn5 with Vibrio fischeri luxAB as reporterWolk et al. (1991)
 pHC60TetR; expression vector; low copy numberCheng & Walker (1998)
 pHC60 hrcNTetR; pHC60 containing a 1341-bp fragment with the hrcN gene; used to complement FC440(ΔhrcN)This work

Growth of plants and bacteria

Cucumber cv. Nongcheng No. 3 and Nicotiana benthamiana were grown in growth chambers in plastic pots under a 16-h photoperiod at 25°C with standard potting soil. Fruits of cucumber cv. Jingyanmini No. 2 were purchased from the market. Inoculated plant materials were kept under the same growth conditions supplemented with 95% humidity. Bacterial inoculum was prepared based on the linear regression equation between the bacterial suspension concentration of A. citrulli and OD600 value, which was = 4 × 10−10+ 0·1128, where x was the concentration of bacterial cells suspension (CFU mL−1) and y was the OD600 value (Peng et al., 2007). To prepare inoculum, a fresh colony of bacteria was inoculated into 2 mL nutrient broth and incubated at 28°C with shaking at 220 rpm. After 18 h of incubation, cells from 1 mL culture (OD600 ≈ 1) were collected by centrifuging at 1000 g for 10 min, and resuspended in 2 mL ddH2O. The optical density of the cell suspension was adjusted to an OD600 value of 0·5, which corresponds to a concentration of approximately 9·68 × 108 CFU mL−1.

Plant inoculations

Fully expanded leaves of 6-week-old N. benthamiana plants were used for hypersensitive response assays by infiltration of bacterial suspension (9·68 × 108 CFU mL−1) from the abaxial side using a blunt syringe. Visible hypersensitive reaction (HR) was observed 48 h post-infiltration. On cucumber fruits, 100 μL inoculum (containing approximately 9·7 × 107 bacterial cells) were injected through the peel for investigation of BFB symptom development. Sterile distilled water was used for controls. Bacterial pathogenicity assays were performed on cotyledons of 6-day-old cucumber seedlings by scratching the adaxial surface with a flamed needle and then inoculating with 20 μL bacterial suspension (approximately 1·9 × 107 bacterial cells). Five days after inoculation, BFB severity was evaluated based on a 0–4 scale: 0, no symptoms; 1, slight water-soaked lesion or chlorotic area on cotyledons; 2, extensive water-soaking on cotyledons; 3, collapse of the infected cotyledons or spread of the infection to true leaves; 4, collapse of the whole seedling. Each A. citrulli strain tested included at least six replicates (three plants) per experiment. Mutants with altered pathogenicity were screened using the wildtype FC440 strain as the control. Except for the screening of altered-pathogenicity mutants from the mutant library, each experiment mentioned above was performed at least three times for each tested strain. Two-sided t-tests were conducted using the statistical software R (version 2·13·0) (R Development Core Team, 2011).

Quantification of in planta bacterial growth

To determine levels of bacterial growth in planta, leaves of 3-week-old cucumber seedlings were vacuum-infiltrated with bacterial suspensions of 105 CFU mL−1 following the previous reported procedure (Katagiri et al., 2002) with some modification. Detached leaves were placed in the bacterial suspension and a vacuum was applied to 95 kPa followed by a slow release to infiltrate the leaves uniformly. Leaves were then transferred to plastic trays and the petioles were covered with cotton balls soaked in sterile distilled water. The trays were incubated in growth chambers for 5 days. To measure bacterial multiplication within leaves, three leaf discs from cucumber leaves (surface-disinfected with 70% ethanol solution before use) were ground in sterile ddH2O, and serial dilutions were spread onto KMB plates supplemented with ampicillin for genetic selection. Bacterial population size was examined up to 5 days (bacterial populations in leaves were sampled immediately after infiltration and on the 3rd and 5th day thereafter) post-inoculation. Relative growth rates were calculated based on the initial pathogen population infiltrated into the plant tissue (0 days after infiltration). Two-sided t-tests were performed using statistical software R (version 2·13·0). Three independent experiments were carried out for each strain.

Molecular manipulation

Routine molecular manipulations were carried out using standard procedures (Sambrook & Russell, 2001). Unless otherwise stated, all molecular biology reagents, including restriction enzymes, DNA polymerase and T4 DNA ligase, were from TaKaRa Biotech Inc. Kits for plasmid and PCR product purification were purchased from Axygen Scientific Inc. Kits for genomic DNA isolation from bacterial cells were purchased from Tiangen Biotech Inc. Oligonucleotide primers and DNA sequencing were synthesized or performed by the Beijing Genomics Institute.

Generation of Acitrulli Tn5 transposon mutants by diparental conjugation

A suicide vector carrying the transposon Tn5, pRLl063a Kan::Tn5 (Wolk et al., 1991), was used for A. citrulli mutagenesis. A broad-host E. coli strain S17-1 (Simon et al., 1983) was used as the host of the suicide vector. Diparental conjugation was carried out following a standard procedure (De Bruijn & Rossbach, 1994) with minor modifications. A 2-mL culture was started with a fresh colony of wildtype FC440 in nutrient broth supplemented with ampicillin and was grown at 28°C with shaking at 220 rpm until log-phase growth. The cultures were diluted 1:10 in fresh medium and incubated for another 5 h. The same procedure was used for the donor strain S17-1 pRL1063a, except the culture was grown at 37°C in LB broth supplemented with kanamycin. The A. citrulli FC440 and S17-1 pRL1063a cells were pelleted at 1000 g for 10 min, then resuspended in 0·01 m PBS to an OD600 value of 0·3 for A. citrulli FC440 (4·68 × 108 CFU mL−1) and 0·6 for S17-1 pRL1063a. To generate Tn5 insertion mutants, equal volumes of A. citrulli FC440 and S17-1 pRL1063a cells were mixed and 100 μL mixture was pipetted onto sterile mating filter membranes (25 mm nitrocellulose filter membrane) placed on KMB plates without antibiotics and incubated for 48 h at 28°C. In each set of mating experiments, 100 μL of A. citrulli FC440 and S17-1 pRL1063a cells were spotted separately onto the mating membrane as controls. Cells from the mating filters were selected by culturing in 1 mL nutrient broth supplemented with ampicillin and kanamycin at 28°C with shaking at 220 rpm for 24 h. Each culture mixture was further streaked individually onto solid KMB containing ampicillin and kanamycin and incubated for an additional 48 h. One colony from each mating was chosen for the third round of selection. Candidate mutants were stored at −80°C with 15% glycerol. To verify the exconjugants after three rounds of selections, colony PCR assays were used for detection of luxAB in the genome of the exconjugants resulting from introduction of suicide vector pRLl063a and for verification of Acitrulli, using primer pairs of LUXP1/LUXP2 (Li et al., 2004) (5′-TCGGCTTGGTATCGC/CTTAGGTCCATTCTCA-3′) and SEQID4m/SEQID5 (5′-GTCATTACTGAATTTCAACA/CCTCCACCAACCAATACGCT-3′) (Schaad et al., 2000; Walcott et al., 2003), respectively. Both plasmid DNA of pRL1063a and cells of wildtype strain FC440 were used as positive controls. For determination of the transposition frequency, cells washed from the mating filter in 1 mL nutrient broth supplemented with ampicillin and kanamycin were serially diluted in 2 mL selection broth followed by incubation at 28°C with shaking at 220 rpm for 6, 12 and 24 h. Finally, 100 μL of each cell culture were plated onto selection plates and the number of colonies resistant to both antibiotics was scored after 48 h of incubation at 28°C. Transposition frequency was defined as the number of Tn5-containing exconjugants divided by the total number of recipient cells of A. citrulli FC440.

Southern blot analyses of A. citrulli Tn5 mutants

There are five PstI restriction sites in the pRL1063a vector, including three in the region introduced into the A. citrulli genome through transposition of Tn5. By digestion with PstI, a 3·8-kb fragment from plasmid pRL1063a and a fragment of at least 3·6 kb from the Tn5 insertion region in the A. citrulli genome carrying the luxAB gene are expected to be released. For Southern blot analyses, 1 μg total A. citrulli genomic DNA and 0·1 μg pRL1063a plasmid DNA were restriction-digested with PstI, resolved in 1·0% agarose gel in 1× TAE and transferred to Hybond N+ nylon membrane (Amersham–Pharmacia) using 20× SSC as the transfer agent following a standard procedure (Sambrook & Russell, 2001). The membrane was baked at 80°C for 2 h to fix the DNA. A 1·6-kb luxAB fragment was PCR-amplified from plasmid pRL1063a using primers LUXP1 and LUXP2, as described above, and used as a probe for Southern hybridization. Approximately 100 ng purified PCR product was 32P-labelled with random primers. Membrane hybridizations were done at 65°C for 16 h using 6× SSPE with 5× Denhardt’s solution (1× Denhardt’s solution contains 0·02% Ficoll 400, 0·02% PVP, 0·02% BSA) and 0·5% SDS as the hybridization buffer as described by Sambrook & Russell (2001). Following hybridization, the membranes were washed at 65°C twice in 2× SSC with 0·1% SDS for 15 min, and twice in 0·2× SSC with 0·1% SDS for 15 min. The signals were exposed and visualized on a FLA-7000 phosphor-imager (Fuji Photo Film Co. Ltd).

Cloning of Tn5-tagged regions of A. citrulli mutants and sequence analysis

Because pRL1063a contains a replication origin (oriT) that functions in E. coli, though not in Acitrulli, DNA sequences surrounding the transposon insertion sites can be recovered from the genome of Acitrulli mutants using a plasmid rescue procedure (Wolk et al., 1991). Briefly, the genomic DNA of the exconjugant was extracted and digested with EcoRI, which cuts the genome frequently but does not cut the transposon Tn5, keeping the KanR gene and the oriT fragment intact. Digestion products were then self-ligated and transformed into Ecoli DH10B by electroporation (Sambrook & Russell, 2001). Colonies resistant to kanamycin were sequenced using primers MUP/MDP (5′-TATCAATGAGCTCGGTACCC/AGGAGGTCACATGGAATATCAGAT-3′) (et al., 2005). The obtained sequences were subjected to blast searches against the genome sequence of Acitrulli AAC00-1 (GenBank accession number NC_008752). To explore if GC content and gene density bias existed in the surveyed insertion sites, the background GC content of each kilobase was calculated without overlap, and gene density was computed every 50 kb with 1 kb moving each step. The calculation was performed by Perl 5 (Wall, 2011) and illustrated by R (version 2·13·0) based on the Acitrulli genome. To detect if position bias existed, chi-square tests were performed manually.

Construction of an hrcN expression vector and genetic complementation in Acitrulli

Expression vector pHC60 (Cheng & Walker, 1998) was used for genetic complementation assays of candidate genes. Escherichia coli strain S17-1 was used as the host of the expression vector. For the generation of an hrcN expression construct, hrcN was amplified by PCR from Acitrulli strain FC440 using primers ATPF/ATPR (5′-GCCGGTACCGGATGACGATGCATGACG/CTAGTCTAGAGGCATCAACCCGAGATGTCGG-3′) (KpnI site and XbaI site underlined, respectively), and cloned into pHC60. The resulting construct, pHC60 hrcN, was transferred into E. coli strain S17-1 and conjugated into strain FC440 (ΔhrcN), giving FC440 (ΔhrcN+hrcN). Plasmid pHC60 was transferred into FC440 (ΔhrcN) following the procedures used for strain FC440 (ΔhrcN+pHC60) (see above).

Protein phylogenetic analysis

The protein sequences of T3SS ATPase (HrcN), flagellum-specific ATP synthase (FliI), and F0F1 ATP synthase subunit beta (ATP_beta) used in this study are listed in Table 2. Amino acid sequences were aligned by ClustalW (Larkin et al., 2007), and phylogenetic analyses were performed using the neighbour-joining method via mega 4 (Tamura et al., 2007) with default parameters, bootstrapped 1000 times.

Table 2.   Bacterial FliI, ATP_beta and HrcN protein sequences
DesignationaDescriptionOrganismNo. of residuesAccession no.b
  1. aThe two-letter abbreviations of the source organism (e.g. Ac, Acitrulli) are followed by protein designations.

  2. bAccession numbers in this table are retrieved from the GenBank database.

Ac FliIFlagella-specific ATP synthaseAcidovorax citrulli474ABM34932
Ps FliIFlagella-specific ATP synthasePseudomonas syringae pv. tomato DC3000452AAO55479
Xc FliIFlagella-specific ATP synthaseXanthomonas campestris pv. campestris458YP_001903624
Ac ATP_betaF0F1 ATP synthase subunit betaAcidovorax citrulli476YP_968753
Ps ATP_betaF0F1 ATP synthase subunit betaPseudomonas syringae pv. tomato DC3000459NP_795317
Xc ATP_betaF0F1 ATP synthase subunit betaXanthomonas campestris pv. campestris468YP_001905199
Ac HrcNType III secretion system ATPaseAcidovorax citrulli442ABM31070
Ps HrcNType III secretion system ATPasePseudomonas syringae pv. tomato DC3000419AAG33879
Xc HrcNType III secretion system ATPaseXanthomonas campestris pv. campestris438YP_001904474
Rs HrcNType III secretion system ATPaseRalstonia solanacearum439NP_522431
Ea HrcNType III secretion system ATPaseErwinia amylovora454CBX79348

Results

Pathogenicity determination of A. citrulli on cotyledons of cucumber

As a model plant for the cucurbits, cucumber was selected as a host of Acitrulli for the study of its biology and pathogenicity. Two A. citrulli strains, FC440 and AW0601, representing groups I and II, respectively, were used for plant infection assays. Scratch-wound inoculation of cotyledons indicated that BFB severity was significantly different between the two group strains at the same bacterial concentration. Strain FC440 showed significantly more aggressiveness than AW0601 on cucumber (P = 3·851E-04) (Fig. 1a,b). Monitoring the multiplication of bacteria within leaf tissues further confirmed that strain FC440 grew significantly faster than strain AW0601 (P = 3·914E-05 and 1·148E-03 on the 3rd and 5th days, respectively; Fig. 1c). Group I strain FC440 was therefore chosen for use in subsequent experiments to establish an Acitrulli–cucumber pathosystem.

Figure 1.

 Pathogenicity determination of Acidovorax citrulli on cucumber (Cucumis sativus cv. Nongcheng No. 3). (a) Water-soaked lesions developed 5 days post-inoculation (dpi) on cotyledons of 6-day-old cucumber seedlings by scratch-wound inoculation with 20 μL bacterial cell suspension at 9·68 × 108 CFU mL−1. Left panel: cotyledons treated with strain FC440; right panel: cotyledons inoculated with strain AW0601. (b) Comparison of strains FC440 and AW0601 in causing bacterial fruit blotch severity. Symptom severity was scored 5 dpi according to the following scale: 0, no symptoms; 1, slight water-soaked lesion or chlorotic area on cotyledons; 2, extensive water-soaking on cotyledons; 3, collapse of the infected cotyledons or spread of the infection to true leaf; 4, collapse of the whole seedling. Error bars indicate standard errors from three independent experiments. (c) Multiplication of wildtype strains FC440 and AW0601 within cucumber leaves. Detached leaves were infiltrated with bacterial suspensions (∼105 CFU mL−1), and relative growth rate (RGR) was determined at the indicated time points. Student’s t-test showed the differences between treatments. Error bars indicate standard deviation. Three independent experiments were performed with similar results. Bacterial fruit blotch (BFB) symptoms formed on cotyledons 5 dpi (d), on true leaf 8 dpi (e) and on flesh of fruit 8 dpi (f). A droplet of 20 μL bacterial suspension (9·68 × 108 CFU mL−1) of strain FC440 were wound-inoculated on cotyledons of 6-day-old seedlings or leaves of 3-week-old seedlings of cucumber cv. Nongcheng No. 3, respectively. To the right of the main vein on the true leaf: dH2O negative control. Cucumber cv. Jingyanmini No. 2 fruits were injected with 100 μL bacterial suspension (9·68 × 108 CFU mL−1).

Strain FC440 was shown to successfully infect and colonize cotyledons, true leaves and fruits of cucumber by wound or needle injection inoculations (Fig. 1d–f). Symptoms on both cotyledons and true leaves were similar to those described previously (Martin et al., 1999; Burdman et al., 2005). Initial symptoms included chlorotic, water-soaked lesions that later dried out to form necrotic spots of light brown, dead tissue. Typically, small water-soaked spots developed around inoculation sites in the leaves, and large chlorotic areas formed on the surface of fruits. Water-soaked lesions were observed on cotyledons of 6-day-old seedlings 5 days post-inoculation, or on true leaves of 3-week-old seedlings 8 days post-inoculation. It took at least 8 days for visible lesions to develop on cucumber fruits. Therefore, the group I strain FC440 and cotyledons of 6-day-old cucumber seedlings were used for pathogenicity assays and mutant screening.

Generation of Acitrulli mutants by Tn5 transposon mutagenesis

It was difficult to achieve transposition in A. citrulli. Electroporation with the suicide vector pRL1063a failed to produce double antibiotic-resistant colonies (AmpR and KanR). With diparental conjugation of wildtype FC440 and S17-1 pRL1063a with 12 or 24 h conjugation followed by plating appropriate dilutions on selective agar media plates it was difficult to obtain positive colonies. When conjugation was extended to 48 h, the background on selection plates could be very low (no colonies) or very high, even diluted at 10−10 (too many colonies to select). To facilitate selection of exconjugants, the first round of selection was performed in liquid medium to enrich the exconjugants. Through a combination of 48 h conjugation and 24 h selection in double antibiotic liquid medium, an average of 1·16 × 104 positive colonies were recovered per mating, giving a transposition frequency of 5·04 × 10−4 per cell of A. citrulli FC440. However, this was most likely an overestimation because the initial recipient cells had multiplied for several generations and the numbers of exconjugants may also have increased during such a long conjugation and selection process. The results appeared to be in agreement with those for several other strains of A. citrulli for which transposon mutation also proved difficult (Tingchang Zhao, Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, personal communication). The duration of conjugation was extended to 48 h and exconjugant growth in selective liquid medium to 24 h. To avoid inclusion of possible redundant colonies in the mutant collection and to minimize the effort of mutant screening in downstream experiments, particularly in screening for altered-pathogenicity mutants, one individual colony was selected from colonies produced by each diparental conjugation mating after the third round of selection. An average of 98·4% matings produced ampicillin- and kanamycin-resistant colonies after three rounds of selection, and 98·3% of these colonies were verified to be A. citrulli by PCR amplification. A total of 2100 transconjugants were generated and stored at −80°C.

The quality of the library was assessed by Southern blotting analyses of five randomly selected candidate mutants. A single chromosome DNA fragment hybridized with the Tn5 probe, suggesting single Tn5 insertions in the chromosome (Fig. 2a). The transposon boundaries of 10 randomly selected mutants were sequenced in order to localize the insertion sites in the A. citrulli FC440 genome (Supplementary Table S1). Sequence analysis indicated that all 10 inserts generated 9-bp duplications at the boundaries, consistent with a Tn5 transposition. The majority of insertion sites were located inside recognized or hypothetical ORFs. The high percentage of intragenic insertion events implied a high probability of obtaining various kinds of functionally disabled mutants, including those with altered pathogenicity.

Figure 2.

 Analysis of transposon insertion mutants of Acidovorax citrulli FC440. (a) Southern blot analysis of five randomly selected individual transposition colonies. PstI-digested genomic DNA (approximately 1 μg) isolated from wildtype A. citrulli (lane 1) and its Tn5 insertional mutants (lanes 2–6). Tn5 (3809 bp) was released from plasmid pRL1063a by PstI digestion and used as positive control (100 ng) (lane 7). (b) GC content and gene density bias of the insertion sites in the A. citrulli genome were analysed with 10 randomly selected individual Tn5 mutants by Perl scripts and illustrated by R based on the A. citrulli genome.

Library screening using the developed pathogenicity assays on cucumber cotyledons led to the identification of six altered-pathogenicity mutants from 1436 screened mutants, and three of them lost their abilities to cause BFB symptoms on susceptible hosts (cucumber, melon, watermelon, squash and pumpkin) and to induce HR in non-host tobacco. The remaining three showed reduced pathogenicity on host plants. Cloning and sequence analysis of the insertion sites revealed several interesting genes that probably contribute to pathogenicity. These included the glucose-inhibited division protein A gene, the flagellar hook-associated protein 3 gene, a predicted secretion-associated protein gene, and a predicted T3SS ATPase gene (genes Aave-0051, Aave-4431, Aave-0473 and Aave-0463, respectively, in the sequenced strain of AAC00-1; GenBank accession number NC_008752). The disruption of the former two individual genes led to reduced pathogenicity of Acitrulli on its hosts, whereas the disruption of the latter two caused pathogenicity deficiency in the pathogen. GC content and gene density bias were not found in the 10 randomly selected candidates (Fig. 2b). The insertion sites appeared to be clustered within the 4·1–5·2 Mb region of the Acitrulli genome, but a chi-square test could not reject the hypothesis of their uniform distribution (χ² = 7 < χ²0·10(4) = 7·779).

To assess the stability of the Acitrulli Tn5 mutants, transposon boundaries of the above 10 mutants, which remained pathogenic, were analysed after three rounds of inoculations of cucumber cotyledons using single colonies. The results showed that all tested mutants were pathologically and genetically stable.

Role of the HrcN protein in pathogenicity and NF-T3SS functionality in Acitrulli

One of the altered-pathogenicity mutants was shown to be disrupted with a predicted T3SS ATPase, as mentioned above, which is homologous to the NF-T3SS HrcN protein. The transposon insertion site was at bp 938 of the 1329-bp hrcN gene. This mutant, designated FC440 (ΔhrcN), lost the ability to cause BFB symptoms in cucumber (Fig. 3a, seedling 3) and failed to induce HR in tobacco (Fig. 3b, circle 3). The predicted HrcN protein in strain FC440 (GenBank accession number JF701985) shared 99·3% sequence identity with Aave-0463 in strain AAC00-1, indicating a very high level of conservation between the two subgroup strains in Acitrulli.

Figure 3.

 Analysis of Acidovorax citrulli mutant disrupted with NF-T3SS ATPase HrcN on host and non-host plants. Response of (a) 6-day-old cucumber (cv. Nongcheng No. 3) cotyledons (photo taken 5 days after scratch-wound inoculation) and (b) 6-week-old leaf of non-host plant Nicotiana benthamiana (photo taken 48 h after infiltration) to inoculation with A. citrulli wildtype FC440 (2); FC440 (ΔhrcN) carrying the empty vector (3); FC440 (ΔhrcN+hrcN) (i.e. FC440 [ΔhrcN] expressing hrcN gene) (4); ddH2O was used as negative control (1).

Typical and conserved characteristics of ATPase enzymes (Walker et al., 1982) were found in the predicted ATPase HrcN sequence, including two Walker box motifs, Walker box A motif173 PAGVGKS179 and Walker box B motif258 LLMMD262. Comparative sequence analyses of NF-T3SS ATPase (HrcN), F0F1 ATP synthase subunit beta and flagellum-specific ATP synthase (FliI) between Acitrulli and representative plant pathogenic bacteria, including Pseudomonas syringae and Xanthomonas campestris, showed that the predicted HrcN of Acitrulli fell in the NF-T3SS ATPase cluster, and was distant from the other two clusters (Fig. 4).

Figure 4.

 Phylogenetic analysis of HrcN-related ATPases. The abbreviations of NF-T3SS ATPase, FliI and ATP_beta represent non-flagellar type III secretion system ATPase, flagella-specific ATP synthase, and F0F1 ATP synthase subunit beta, respectively. Ac, Ea, Ps, Rs and Xc represent Acidovorax citrulli, Erwinia amylovora, Pseudomonas syringae, Ralstonia solanacearum and Xanthomonas campestris, respectively.

To verify that the transposon-disrupted hrcN gene plays a crucial role in Acitrulli pathogenicity and NF-T3SS functionality, and to exclude the possibility of a polar effect of the transposon insertion on genes adjacent to hrcN, a genetic complementation experiment was carried out by transforming the pHC60 plasmid carrying the wildtype hrcN gene into the mutant FC440 (ΔhrcN). The complemented strain FC440 (ΔhrcN+hrcN) showed restored ability to cause water-soaked lesions on cucumber cotyledons and to induce HR in tobacco leaves (Fig. 3a seedling 4 and Fig. 3b circle 4). Both the sequence comparison and genetic complementation assays confirmed that the predicted T3SS ATPase was a NF-T3SS ATPase HrcN.

Comparison of the NF-T3SS of Acitrulli with those of Ralstonia solanacearum and X. campestris

The difference between two NF-T3SS groups of plant pathogenic bacteria (Alfano & Collmer, 1996) may be derived from the distinct T3SSs they possess. To evaluate the phylogenetic relationships between NF-T3SS in Acitrulli and other representative plant pathogenic bacteria, the NF-T3SS ATPase HrcN, a conserved and essential inner-membrane component in NF-T3SS, was chosen for comparative sequence analysis. The results showed that the NF-T3SS of Acitrulli, R. solanacearum and X. campestris are closely related to each other and are distinct from those of P. syringae and E. amylovora (Fig. 4), indicating that the NF-T3SS of A. citrulli belongs to group II with those of R. solanacearum and X. campestris (Alfano & Collmer, 1996).

Discussion

This study set up a model pathosystem for A. citrulli in which cucumber cotyledons were used for pathogenicity assays. The established Acitrulli–cucumber pathosystem provides a valuable model to dissect interactions between Acitrulli and cucurbits.

The established pathosystem was proved to be efficient by successful identification of pathogenicity-deficient Acitrulli insertional mutants generated by Tn5 transposon mutagenesis. A mutation in the predicted T3SS ATPase gene hrcN in Acitrulli led to loss of pathogenicity. Further phylogenetic analysis and genetic complementation experiments confirmed that NF-T3SS is essential for Acitrulli pathogenicity, consistent with the results from characterization of the interaction between Acitrulli and its original natural host melon or watermelon (Ren et al., 2009). Comparative analysis of the conserved and essential NF-T3SS component HrcN protein and its homologues showed that the NF-T3SS of Acitrulli was in group II, which includes that of R. solanacearum and X. campestris (Alfano & Collmer, 1996).

It was shown that the group I strain of Acitrulli was more aggressive on cucumber than the group II strain, and the group I strain of the Acitrulli–cucumber model is probably more suitable for investigating Acitrulli–cucurbit interactions. Previous investigations showed that cucumber can be infected by Acitrulli under natural and experimental conditions, and typical BFB symptoms were characterized mainly on cucumber foliage; no symptoms on plant stems or fruits were reported (Martin et al., 1999; Burdman et al., 2005; Liu et al., 2009). In the present study, typical symptoms were observed on cucumber fruits following bacterial inoculation by injection. Besides the capability to infect cucumber, Acitrulli strains in the two subgroups were reported to differ in pathogenicity on the host plant (O’Brien & Martin, 1999; Walcott et al., 2000, 2004; Burdman et al., 2005), with the group I strains being more aggressive than the group II strains on cucumber, as determined by both seedling pathogenicity and seed transmission assays (Burdman et al., 2005). The present study also investigated differential BFB severity between strains of the two subgroups following wound inoculation on cucumber cotyledons. To exclude the possible influence of differences in motility, capability of surface attachment and adhesion between strains on pathogenicity determination, an infiltration method was employed for inoculation and to quantify the growth rates of the two subgroup strains within tissues. The data confirmed that the group I strain grew significantly faster than the group II strain within cucumber tissues, which could lead to more severe BFB development. Quantification of bacterial populations in inoculated leaf tissues further confirmed the difference in BFB severity between the strains on cotyledons, and verified that the pathogenicity assays on cucumber cotyledons were stable and repeatable. Large-scale screening for pathogenicity mutants requires an efficient procedure for pathogenicity determination. In view of the existence of certain percentages of symptomless mutants on seedlings, as revealed by seed transmission assays discussed above, and the tedious screening procedures reported, the simplified method for pathogenicity determination by cotyledon inoculation of cucumber seedlings will facilitate large-scale identification of pathogenicity genes in the BFB pathogen.

Identification of pathogenicity-related genes in the pathogen has been a major focus in the field of plant–pathogen interactions. To identify pathogenicity-associated genes in A. citrulli, transposon-mediated insertional mutagenesis has been applied to create mutants followed by in planta screening for altered-pathogenicity genes on original hosts of wildtype strains with seed-transmission assays (Bahar et al., 2009; Ren et al., 2009). The present study generated a high-quality, Tn5-mutated library of Acitrulli, which provided plenty of genetically stable single-insertion mutants of Acitrulli. The investigation of some mutants excluded the concern of potential genetic instability problems resulting from expression of transposase in the subsequent generations of in vivo transposition systems (Goryshin et al., 2000). Thus, the newly developed protocol for Acitrulli mutagenesis will facilitate the use of genetic approaches to the study of Acitrulli biology and pathology. The valuable mutant resource generated can be applied in a wide range of research on biology and pathology of Acitrulli. The quality of the mutant collection and the more efficient library screening procedure was confirmed by successful identification of several genes required for pathogenicity, including the predicted NF-T3SS ATPase gene in A. citrulli.

The NF-T3SS ATPases shared substantial sequence similarities with FliI, the ATPase component of the flagellar assembly, and to a lesser degree with the catalytic beta subunit of the rotary F0F1 ATP synthase (Woestyn et al., 1994; Akeda & Galan, 2004). Both the flagellar and non-flagellar T3SSs secrete macromolecules from the site of synthesis (cell cytoplasm) to their sites of action (Pallen et al., 2005), but the process of secretion and the mode of gene action between the two types of mechanisms (flagellar apparatus and needle complex) can be very different. In animal pathogenic bacteria the flagellar T3SS ATPase (FliI) is not essential for flagellar T3SS, but is required for the initial entry of export substrate into the export gate (Minamino & Namba, 2008). However, in the plant pathogen X. campestris pv. vesicatoria, the secretion via NF-T3SS could not occur without the ATPase (Lorenz & Buttner, 2009). In the present study, sequence comparison and genetic complementation assays indicated that the predicted T3SS ATPase was a NF-T3SS ATPase HrcN and was essential for both A. citrulli pathogenicity on the host plant and the HR reaction on a non-host plant, suggesting that the HrcN protein was essential for functionality of NF-T3SS. The identification of HrcN not only verified the requirement of T3SS for pathogenicity in A. citrulli on watermelon and melon, as previously reported (Ren et al., 2009; Johnson et al., 2011), but also confirmed the A. citrulli–cucumber interaction system to be effective for studying the pathogenicity mechanism of BFB.

NF-T3SS is used by proteobacteria for pathogenic or symbiotic interaction with plant and animal hosts (Marie et al., 2003). Without NF-T3SS, phytopathogenic bacteria are unable to defeat basal defences, grow in plants, produce disease symptoms in hosts, or evoke the hypersensitive response (HR) in non-hosts. Based on their possession of similar genes, operon structures and regulatory systems, the NF-T3SSs within phytopathogens are further divided into two groups (Alfano & Collmer, 1996). Being categorized in the same group, the inclusion of NF-T3SSs in the bacterial plant pathogens R. solanacearum and X. campestris will further aid understanding of the function of NF-T3SS in the pathogenesis of A. citrulli.

Acknowledgements

We gratefully acknowledge Dr Baishi Hu of Nanjing Agriculture University for A. citrulli strains, Professor Sanfeng Chen of China Agricultural University for plasmid pRL1063a and E. coli strain S17-1, and Dr Yanzhang Wang of the Institute of Plant Physiology and Ecology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, for plasmid pHC60. We also thank Junli Quan, Biao Gu, Meixiang Zhang, Keke Shangguan and Hua Cao for valuable discussions during the course of this study. This work was supported by the Scientific Research Program of the Higher Education Institution of Xinjiang (#XJEDU2005S07) and the 111 Project from the Education Ministry of China (#B07049).

Ancillary