Acidovorax citrulli strain AAC00-1 was used in this study. The bacterium was routinely cultured on Luria-Bertani (LB) agar for 48 h at 28°C. For preparation of cultures for VBNC induction experiments, single colonies were picked to inoculate 10 mL of LB broth that were incubated at 28°C for 12 h with shaking (120 rpm) until they reached the exponential growth phase, as determined in preliminary growth curve experiments. At this point 2 mL aliquots of the bacterial suspension were transferred to 200 mL of LB broth and incubated for 12 h under the same conditions. The cells were then harvested by centrifugation at 10,610 g for 5 min, washed twice with sterilized 0.85% NaCl solution (w/v) to remove traces of the growth media, and resuspended and adjusted to an OD₆₀₀ of 0.5 in sterilized 0.85% NaCl solution. The resulting suspensions were centrifuged and resuspended as described above to prepare a 20-fold concentrated cell suspension containing approximately 10⁹ CFU/mL. These cells were used in the VBNC induction experiments described below.

The VBNC cells of A. citrulli strain AAC00-1 were induced as previously described (Jiang et al., 2016). The maximum non-inhibitory concentration of copper sulfate (CuSO₄) for the growth of A. citrulli in LB broth was determined prior to the induction experiments. Briefly, 100 μL aliquots of 12 h-incubated liquid cultures (∼10⁸ CFU/mL) were inoculated into 10 mL LB broth containing a range of CuSO₄ concentrations: 0, 0.01, 0.1, 0.5, 1.0, 1.5, 2.0, and 2.5 mM. The cultures were then incubated for 12 h at 28°C with shaking (120 rpm) and the number of cells in each treatment was estimated by plating on LB agar and counting the number of colonies that grew after incubation at 28°C for 48 h. Each treatment consisted of three replicates, and two independent experiments were carried out.

Viable but non-culturable induction was performed using AAC00-1 suspensions generated in 150 mL 1× AB salts solution (1 g L^-1 NH₄Cl, 0.3 g L^-1 MgSO₄, 0.15 g L^-1 KCl, 0.01 g L^-1 CaCl₂, 2.5 mg L^-1 FeSO₄) amended with 0 (negative control), 0.5, 5, 10, or 50 μM CuSO₄. The final cell concentration in the induction microcosms was approximately 10⁷ CFU/mL. The flasks were incubated at 28°C without shaking for 210 days, during which 3 mL aliquots were collected and evaluated at regular intervals. The number of cells in each sample was counted by flow cytometry (FCM) and by plating on LB agar. When the number of culturable cells in the suspension was less than 0.1 CFU/mL, it was considered to contain no culturable cells and that all viable cells were in the VBNC state (Whitesides and Oliver, 1997; Baffone et al., 2003). For each treatment, there were two independent experiments with three replicates per experiment.

The ratio of viable cells to the total number of cells was assessed using the LIVE/DEAD^® BacLight^TM Bacterial Viability Kit (Invitrogen, Carlsbad, CA, United States), which identifies dead cells according to the integrity of cell membranes. Preliminary experiments indicated that when the two dyes were used simultaneously, a high level of dead cells could dramatically affect the total number of cells detected (those stained by SYTO9). Hence, samples were stained separately to avoid interference. The preliminary experiments also indicated that the recommended concentration of PI, 30 μM (cells:PI, v:v ratio of 1:1), produced sufficient fluorescent intensity in the detection area, but the recommended concentration of SYTO9, 6 μM (cells:SYTO9, v:v ratio of 1:1) produced strong fluorescence beyond the detection area. Therefore different ratios (cells:SYTO9, v:v ratio of 1:1, 2:1, 3:1, and 4:1) were evaluated to determine the optimal concentration of SYTO9, which was found to be 4 μM (cells:SYTO9, v:v ratio of 2:1). Having established the optimal dye concentration, the total number of cells was calculated by comparison with a known number of fluorescent beads in the Trucount tubes (BD Biosciences, San Diego, CA, United States). Twelve-hour-incubated A. citrulli cells washed twice with and resuspended in sterilized 0.85% NaCl solution were incubated for 15 min in the dark in the presence of SYTO9 or PI. Excess dye was then removed by washing in 0.85% NaCl, and 800 μL of each sample was transferred to Trucount tubes for analysis. Cells treated with 75% (v/v) ethanol for 10 min were used as the dead cell control. A total of 10,000 events were assessed for each sample using the FACSCalibur flow cytometry system (BD Biosciences, San Diego, CA, United States) with the following settings: amplifier mode set to Log; voltage mode of forward scatter (FSC) set to E00; photomultiplier tube for side scatter (SSC), FL1 for SYTO9 and FL2 for PI, set at 435, 686, and 600 V, respectively; and flow speed set to medium. The number of A. citrulli cells in the microcosm was calculated according to the ratio of cells to beads as follows:

The number of VBNC cells produced during the induction experiments was assessed using 3 mL samples from each induction flask. After washing the cells twice in sterilized 0.85% NaCl solution to eliminate traces of Cu²⁺, 800 μL samples were placed in Trucount tubes for FCM analysis. The cells were then stained using a 2:1 cell:dye ratio (v/v) (667 μL cells and 333 μL SYTO9) for SYTO9 (4 μM) and 1:1 (500 μL cells and 500 μL PI) for PI (30 μM). After incubation in the dark for 15 min, the cells were washed in 0.85% NaCl to eliminate excess dye before 800 μL samples were transferred to 5 mL round-bottom polystyrene tubes (12 mm × 75 mm style, BD Biosciences, San Diego, CA, United States) for analysis by FCM. The remaining bacterial cell suspension (∼1 mL) was then used to prepare a 10-fold serial dilution that was plated on LB agar to assay for culturability. The resulting colonies were counted after 48 h incubation at 28°C. The number of VBNC cells was then calculated by subtracting the number of culturable cells from the number of viable cells, which in turn was calculated by subtracting the number of dead cells stained with PI from the total number of cells stained with SYTO9. Each treatment consisted of three replicates with the entire experiment being conducted twice.

Different strategies were used to resuscitate VBNC A. citrulli cells that were induced by a range of CuSO₄ concentrations (5, 10, and 50 μM). The first treatment involved the removal of the induction factor (Cu²⁺) by addition of the chelator ethylene diamine tetraacetic acid (EDTA disodium salt). The effect of different concentrations of EDTA on the growth of A. citrulli in LB was determined prior to the resuscitation experiments with the same method used to assess the effects of CuSO₄ (see above). The EDTA concentrations tested were 0 (negative control), 5, 10, 50, 100, and 200 μM. The experiment was carried out twice, with three replicates per treatment per experiment. In the EDTA resuscitation treatment, 10 mL of VBNC suspensions were transferred from the induction flasks to sterilized tubes containing 100 μL of EDTA at different concentrations to produce final concentrations of EDTA at onefold (5, 10, and 50 μM), 1.5-fold (7.5, 15, and 75 μM), or twofold (10, 20, and 100 μM) of the concentrations of Cu²⁺ used during induction. The resuscitation microcosms were incubated at 28°C with shaking (120 rpm), and 100 μL samples were checked every 2 days by plating on LB agar. Other treatments tested for resuscitation were growth in an oligotrophic (M9 minimal medium) or in a rich (LB) medium. In these experiments, the VBNC cells from 10 mL induction microcosms were harvested by centrifugation at 10,610 g for 5 min and washed twice with 0.85% NaCl before being resuspended in an equal volume of the growth medium (10 mL). The resuscitation microcosms were then incubated at 28°C with shaking (120 rpm) and samples were checked daily by plating on LB agar. If no culturable cells were recovered after 7 days of incubation, it was assumed that no resuscitation occurred. Another resuscitation approach involved resuspending VBNC cells in cell-free supernatant (CFS) collected from the exponential phase of an AAC00-1 culture or in AB salts amended with casein hydrolysate. The CFS was prepared by centrifuging a liquid culture of AAC00-1 grown in LB broth for 12 h and passing the supernatant through a sterilized 0.2 μm Minisart filter (Sartorius AG, Göttingen, Germany). The VBNC cells from 10 mL induction microcosms were then harvested by centrifugation and washed twice with 0.85% NaCl and resuspended in either 10 mL of the CFS, or 10 mL of an AB salts solution amended with 100 μg/mL casein hydrolysate. The resulting cultures were incubated at 28°C with shaking (120 rpm) and checked every 2 days for culturability by plating on LB agar. The last resuscitation treatment involved transferring VBNC cells to M9 liquid media amended with 10% (v/v) fresh watermelon seedling juice in an attempt to simulate conditions found in vivo. The watermelon seedling juice was prepared as follows: 3-week-old watermelon (cv. Ruihong; kindly supplied by the Institute of Vegetables and Flowers Chinese Academy of Agricultural Sciences) seedlings were grinded and centrifuged at 10,610 g for 5 min, and the supernatant was transferred and passed through a sterilized 0.2 μm Minisart filter (Sartorius AG, Göttingen, Germany). The samples were incubated at 28°C with shaking (120 rpm) and checked every 2 days for culturability by plating on LB agar. Each EDTA treatment contained three replicates and the other treatments contained six replicates. There were two independent experiments for all the resuscitation experiments.

Fifteen colonies were randomly picked from each treatment of the resuscitated cells and verified by PCR using the A. citrulli specific primers SEQID4^m (5′-GTCATTACTGAATTTCAACA-3′) and SEQID5 (5′-CCTCCACCAACCAATACGCT-3′) (Schaad et al., 2000). The PCR was conducted with a MyCycler^TM Thermal Cycler 580BR (Bio-Rad, Hercules, CA, United States) in 25 μL reaction volumes containing 2.5 μL 10×Taq reaction buffer, 2.4 μL dNTPs (2.5 mM each), 1 μL each primer (10 μM stock), 0.25 μL TaqDNA polymerase (5 U/μL), and 1 μL of a cell suspension that was previously incubated at 100°C for 10 min. ddH₂O was added to complete the reaction volume to 25 μL. The thermocycler reaction conditions were as follows: denaturation at 95°C for 5 min, followed by 35 cycles of 95°C for 30 s, 53°C for 30 s and 72°C for 30 s, and a final extension at 72°C for 5 min. The resulting PCR fragments were analyzed by electrophoresis on a 1.5% (w/v) agarose gel, with single 246 bp fragments indicating a positive result.

Virulence assays were carried out by seed-transmission and cotyledon infiltration assays with watermelon (cv. Ruihong). Seed-transmission experiments were carried out for both VBNC cells and resuscitated cells as previously described (Bahar et al., 2009). Briefly, VBNC cells (∼10⁷ cells/mL) in 10 mL induction suspensions were washed twice with 0.85% NaCl to remove the copper sulfate and resuspended with the same volume of 0.85% NaCl solution. Resuscitated cells were prepared by picking single 48-h-old colonies from LB plates and inoculating them into 10 mL LB broth. After incubation of 12 h at 28°C with shaking (120 rpm), the suspensions were adjusted to an OD₆₀₀ of 0.5, corresponding to ∼10⁸ CFU/mL. The resulting suspensions were diluted to 10⁶ CFU/mL with sterilized 0.85% NaCl solution and used as inoculum in seed transmission assays. Twenty-four watermelon seeds were incubated for 2 h at 28°C with shaking (120 rpm) in the bacterial suspensions (VBNC cells at ∼10⁷ cells/mL and resuscitated cells at 10⁶ CFU/mL). The inoculated seeds were then dried on sterilized filter paper for 12 h at 28°C before being sown in pots (six seeds per pot) filled with a soil mixture containing perlite, vermiculite and chicken manure at a ratio of 4:4:1 (w/w/w). The pots were kept in a greenhouse (20–28°C) with a 14 h photoperiod for 14 days, after which the fresh shoot weight of each plant was measured and the mean weight value of the seedlings was used to evaluate the virulence among different treatments of resuscitated cells. Non-VBNC induced A. citrulli cells that were collected at the exponential phase (growth in LB broth for 12 h) were used as positive control for disease induction, while treatment of seeds with a 0.85% NaCl solution was used as negative control. Two independent experiments were carried out with a total of 34 to 45 replicates (plants) per treatment.

Cotyledon infiltration experiments were carried out with a 100-fold concentrated suspension of VBNC A. citrulli cells. For these assays, cells from 100 mL VBNC induced suspension were harvested by centrifugation at 10,610 g for 5 min, washed twice with 0.85% NaCl solution to remove the Cu²⁺ ions and then resuspended with 1 mL 0.85% NaCl solution. Fourteen-day-old watermelon seedlings were inoculated by injecting 25 μL of a 100-fold concentrated suspension of induced VBNC A. citrulli cells (∼10⁹ cells/mL) into the back of the cotyledons. The plants were incubated in the greenhouse and leaf samples were collected at 7 and 14 days post-inoculation (dpi) for bacterial isolation. Leaf samples were surface sterilized by immersion in a 1% (v/v) NaOCl solution for 1 min and washed in sterilized distilled water three times before being dried on sterilized filter paper. The dried leaves were then ground for 1 min at 30 Hz/s with an automated grinder (Retsch GmbH, Haan, Germany), and homogenized in 1 mL sterilized 0.85% NaCl solution by vortexing. The resulting suspension was streaked onto LB agar and candidate colonies that developed after 48 h incubation at 28°C were checked by PCR using the SEQID4^m/SEQID5 primer set as described above. Two independent experiments were carried out with 6 replicates (cotyledons) per treatment in each experiment.

Data were analyzed using the SPSS 19.0 software (IBM, Armonk, NY, United States). Statistical significance (p < 0.05) was determined using the Student’s t-test, one-way or two-way analysis of variance (ANOVA) and Tukey’s honest significant difference (HSD). The chi-squared test (p < 0.05) was used to compare the resuscitation frequencies of different induction conditions and different resuscitation methods.

The LIVE/DEAD^® BacLight^TM Bacterial Viability Kit effectively stained cells with a cells:SYTO9 ratio (v:v) of 1:1 and 2:1 (corresponding to final concentration of SYTO9 of 6 and 4 μM, respectively, according to the manufacturer’s instructions) (Figure 1A). However, when the cells:SYTO9 ratio (v/v) was 1:1 and the concentration of SYTO9 was 6 μM, it produced strong fluorescence beyond the detection area (Figure 1D), so a ratio of 2:1 (cells:SYTO9, v:v) was chosen in this study (Figure 1E). PI was used as the manufacturer’s instructions. Additionally, A. citrulli cell concentrations in the range of 10⁶ to 10⁷ CFU/mL yielded the best detection rates (Figure 1B). However, since the VBNC induction process was assessed over a long period, 10⁷ CFU/mL was selected for subsequent experiments. The preliminary tests also confirmed that FCM was highly specific, easily distinguishing between fluorescent beads and other debris in the Trucount tube (Figure 1C), and that cells stained by SYTO9 or PI could be clearly distinguished from unstained cells (Figure 1E,F).

The minimum inhibitory concentration (MIC) of copper sulfate for A. citrulli strain AAC00-1 was 1.5 mM, and among tested concentrations, Cu²⁺ concentrations equal and lower than 100 μM had no significant effect on growth in LB broth after 12 h (Figure 2). These results indicated that Cu²⁺ concentrations as high as 50 μM would be suitable for the induction experiments, which were conducted in AB salts solution. The number of total cells, viable cells, culturable cells, and VBNC cells present in the different treatments was assessed over a period of 210 days by combination of FCM and dilution plating methods. The results of these experiments are summarized in Figure 3, while the raw data is detailed in Supplementary Table S1. The highest concentration of copper (50 μM CuSO₄) induced the VBNC state in a short period of time, with all cells losing culturability after 3 h. The number of viable A. citrulli cells decreased from 3.28 × 10⁷ cells/mL to 9.54 × 10⁵ cells/mL. Lower concentrations of copper also induced the VBNC state, although they required longer exposure times: 5 days in the case of 10 μM CuSO₄ and 15 days at 5 μM, with viable cell counts of 1.24 × 10⁶ cells/mL and 1.89 × 10⁶ cells/mL, respectively. In contrast, the lowest concentration of copper (0.5 μM CuSO₄) had no effect on culturability compared to the negative control even after 210 days of exposure, with the number of culturable cells being 6.22 × 10⁴ CFU/mL for CuSO₄-treated cells and 5.79 × 10⁴ CFU/mL for control cells. In spite that the 50 μM treatment had the most rapid effect, with VBNC cell counts of 1.87 × 10⁶ cells/mL, the 10 and 5 μM treatments produced significantly more VBNC cells by the end of the 210 days induction period, with VBNC cell counts of 1.09 × 10⁷ cells/mL and 1.01 × 10⁷ cells/mL, respectively. There were no significant differences between the 10 and 5 μM treatments. All the log-transformed data are shown in Supplementary Table S1 and statistical significance (p < 0.05) was according to the Student’s t-test (for culturable cells of control and 0.5 μM copper sulfate treatment samples that were exposure 15 days or longer in the copper sulfate) and one-way or two-way ANOVA.

All resuscitation methods tested were able to induce resuscitation with the exception of the M9 treatment, which failed to resuscitate any VBNC A. citrulli cells, regardless of the Cu²⁺ concentration used for VBNC induction (Table 1). EDTA and LB broth were the most effective resuscitation methods. The concentrations of EDTA used for resuscitation had no drastic effect on the growth of A. citrulli, although slight but significant differences (p < 0.05) were found between 200 μM EDTA and 10 and 50 μM EDTA (Supplementary Figure S1). Resuscitation experiments revealed no significant differences (p < 0.05) among 5, 10, and 50 μM induction treatments, and at the three concentrations of EDTA.

As mentioned above, M9 medium treatment was ineffective to induce resuscitation (Table 1). The other tested growth media were more efficient at resuscitating VBNC cells that were exposed to stress conditions for a shorter period of time (e.g., higher CuSO₄ concentrations). For example, LB broth had resuscitation frequencies of 3/12, 7/12, and 7/12 for the 5, 10, and 50 μM CuSO₄ induction treatments, respectively, while casein hydrolysate and M9 supplemented with watermelon seedling juice failed to resuscitate VBNC cells from the 5 μM CuSO₄ treatments. Despite this, when successful resuscitation occurred, the average concentrations of recovered culturable cells were often higher than the concentrations of culturable cells obtained with the EDTA method. For example, resuscitation with LB yielded bacterial concentrations of 2.68 × 10⁸, 3.15 × 10⁷, and 3.81 × 10⁷ CFU/mL for the 5, 10, and 50 μM CuSO₄ induction treatments respectively. Similarly, resuscitation with CFS of AAC00-1 cells yielded bacterial concentrations of 6.00 × 10⁸, 1.56 × 10⁷, and 1.18 × 10⁸ CFU/mL for VBNC cells induced at 5, 10, and 50 μM CuSO₄, respectively. Using casein hydrolysate, the average concentrations of resuscitated cells were 3.80 × 10⁶ and 1.07 × 10⁸ CFU/mL, for VBNC cells induced with 10 and 50 μM CuSO₄, respectively, while for the same VBNC induced cells, M9 supplemented with watermelon seedling juice lead to resuscitation of 2.67 × 10⁷ and 7.77 × 10⁷, respectively (Log-transformed data are shown in Table 1).

The virulence of the VBNC A. citrulli cells produced by exposure to Cu²⁺ was tested by watermelon seed-transmission assays or following injection of 25 μL aliquots of VBNC cells (∼10⁹ cells/mL) into the cotyledons of 14-days-old watermelon seedlings. Watermelon seedlings emerging from seeds that were treated with VBNC A. citrulli cells, showed no symptoms in seed-transmission assays (data not shown). Similarly, cotyledons injected with VBNC cells did not develop typical BFB symptoms 14 days post-inoculation (dpi). Furthermore, no culturable A. citrulli cells were isolated from any of the cotyledons injected with VBNC cells after this period. In contrast, cotyledons injected with A. citrulli AAC00-1 cells that did not undergo through the VBNC state, induced typical water-soaked symptoms at 2 dpi, extensive necrotic lesions at 7 dpi and totally necrotic and dry at 14 dpi (Figure 4).

In contrast to VBNC cells, resuscitated cells were able to induce symptoms in watermelon seed-transmission assays. In these assays, disease severity negatively correlates with the seedling shoot fresh weight (Bahar et al., 2009). Fourteen days post-inoculation, the mean fresh shoot weight of watermelon seedlings which emerged from seeds inoculated with resuscitated cells from all treatments varied from 0.32 to 0.43 g, and did not significantly differ from seedling emerging from seeds inoculated with non-VBNC treated AAC00-1 cells (0.29 g). All inoculation treatments significantly (p < 0.05) differ from non-inoculated controls that were treated with a 0.85% NaCl solution (Figure 5). Taken together these results indicate that VBNC A. citrulli cells completely lacked virulence, but once resuscitated, they fully regained pathogenicity ability.