Plant pathogenic fungi and bacteria used in this study and their origins are listed in Table 1. Fungal pathogens were cultured on potato dextrose agar (PDA, pH 5.6; Difco, Sparks, MD, USA), and incubated in an incubator with 12 h daily illumination at 25°C. Bacterial pathogens were cultured on nutrient agar (NA, pH 6.8; Difco) plate at 30°C. De Man, Rogosa and Sharpe (MRS) medium, each liter containing 10 g Proteose peptone No. 3, 10 g beef extract, 20 g dextrose, 1 g Tween 80, 5 g yeast extract, 2 g ammonium citrate, 5 g sodium acetate, 0.1 g magnesium sulfate, 2 g dipotassium phosphate, and 0.05 g manganese sulfate, pH 6.2 (De Man et al., 1960) was used to culture LAB from fermented vegetables. A MRSSG medium (pH 4.1) was modified from MRS by replacing peptone, beef extract, and dextrose with 1.5% (v/v) surimi (fish soluble liquid, Regrowth Trading, Kaohsiung, Taiwan) and 1.5% (v/v) glycerol. A MRSSS medium (pH 5.3) contained 1.5% surimi and 0.5% sucrose as the main nitrogen and carbon source, repectively. Both MRSSG and MRSSS were also used to culture LAB. Unless otherwise stated, LAB were cultured at 37°C.

Fermented Chinese cabbage (Brassica rapa pekinensis) (catalog number PA050) was purchased from Chengfu Shengji Food (Hsinchu, Taiwan). Fermented spicy mustard (Brassica juncea (L.) Czern) was purchased from Liancheng Agricultural Products Processing Factory (Taichung, Taiwan). Juice of fermented vegetables was serially diluted and plated out onto MRS agar plates, which were incubated in a 37°C incubator for 36 h. Bacterial colonies were picked and subsequently streaked 3 times for single colonies. Antagonistic activity was assayed on a filter paper disc (8 mm in dia.) overlaid on the lawn of Pectobacterium carotovorum subsp. carotovorum (Pcc) ZL3 (Table 1). The PccZL3 strain was pre-cultured on NA for 2 days and a filter paper disc was placed in the center of the plate. LAB strains were cultured in MRS for 5 days and broth cultures (50 µl) were placed on a filter paper disc. The plate was incubated at 30°C for 2 days for the formation of an inhibition zone.

Antagonistic activity was further assayed on radish. White radish (Raphanus sativus L. var. Longipinnatus) purchased from local markets were surface-sterilized with 1% sodium hypochlorite for 5 min, rinsed 3 times with sterile water, air-dried, and cut into slices of 0.5-cm thick. A sterile filter paper disc was placed in the center of the radish slices. Broth cultures of LAB after being grown in MRS for 3 days and mixed with or without chitosan (Charming & Beauty, Taoyuan, Taiwan) were mixed with an equal volume of 3 to 5-day PccZL3 culture suspensions (108 colony-forming unit [cfu]/ml), from which 70 µl was placed on a filter paper disc after being diluted with sterile distilled water to 200× and 400×. Radishes treated with water, MRS, 2% lactic acid (LA, pH 4.1), or MRS containing 2% LA were used as mock controls. All radish slices were placed in a plastic box and incubated at 30°C for 2 days. Each treatment contained four replicates.

Bacterial strains were streaked on the biolog universal growth plates (Biolog, Hayward, CA, USA), incubated at 37°C for 24 h for single colonies, tested for physiological metabolisms using a Biolog GEN III microplate, and identified using Biolog MicroLog/Microstation according to the manufacture’s recommendations. Further identification of bacteria was performed by sequence analysis of a 16 rRNA gene, which was amplified by PCR with the primers fD1 (5′-ccgaattcgtcgtcgacaacagagtttgatcctggctcag-3′) and rp1 (5′-cccgggatccaagcttacggttaccttgttacgactt-3′) (Weisburg et al., 1991). DNA fragments were directly sequenced using an Applied Biosystem 3730 DNA analyzer at Tri-Biotech (Taipei, Taiwan). Similarity searching was performed at the National Center for Biotechnology Information (NCBI) using the BLASTX program. Sequence data from this article has been deposited in the EMBL/GenBank data libraries under accession numbers: MN396542 (J02) and MN396543 (J13).

LAB strains were cultured in a modified MRS by altering the main carbon or nitrogen source to determine the most effective combination for antagonistic activity. To identify an effective nitrogen source, both proteose peptone and beef extract in MRS were omitted and replaced with 2% (w/v) each of soya peptone, casein, skim milk, tryptone, glutamic acid, and tyrosine. Two agricultural wastes, surimi and chicken feather meal, were also tested. To identify an effective carbon source, dextrose was replaced with 2% (w/v) each of fructose, sucrose, dextrin, maltose, glycerol, and corn meal. Broth cultures of LAB strains were grown in MRS or modified MRS for 3 to 5 days and 1% (w/v) chitosan was added. Broth cultures were diluted with sterile water to 400×, mixed with an equal volume of bacterial suspensions collected from PccZL3 (10⁸ cfu/ml), and applied onto radish as described above.

Assays for spore germination were performed by placing spore suspensions (3 × 10⁴ spores/ml) prepared from fungal pathogens: Alternaria brassicicola ABA31, Colletotrichum higginsianum PA01, or Fusarium oxysporum f. sp. rapae Focn38 (Table 1) and mixed with an equal volume of LAB broth cultures on a glass slide, which was placed on the top of a L-shaped rod and incubated in a glass Petri dish for 8-12 h in the complete darkness. Spores mixed with water were used as a mock control. Unless otherwise indicated, 100 randomly selected spores were examined microscopically. Sclerotia (n = 50) produced by Sclerotium rolfsii SR grown on PDA at 28°C for 7 days were soaked in water or LAB broth cultures for 12 h and transferred to 1.5% water agar medium for germination. The antagonistic activity against Xanthomonas campestris pv. campestris (Xcc) 70 was assessed by placing 20 µl of LAB broth cultures on an 8-mm filter paper placed in the center of bacterial lawn and incubating at 30°C for 2-3 days. Bacterial lawn was prepared by mixing Xcc70 suspensions with NA (45°C), which was then poured on Petri dish. Each treatment contained at least three replicates and all experiments were repeated twice. Percentage (%) of inhibition was calculated by dividing the relative difference of inhibition rates between mock control and treatment over that of the control and multiplied by 100.

The effect of LAB against PccZL3 was assessed on detached cabbage leaf petioles. Ten microliters of PccZL3 suspensions (10⁸ cfu/ml) after being mixed with broth cultures of LAB to 2×, 100×, 200×, and 400× were applied to longitudinal section of leaf petioles. The treated leaves were kept in a plastic box for 2 days for lesion development. Petioles treated with water and PccZL3 only were used as controls.

Biocontrol against A. brassicicola ABA31 was performed on cabbage seedlings and detached leaves. Cabbage (Brassica oleracea L. Capitata group) cv. Gaofeng seeds were purchased from Known-You Seed (Kaohsiung, Taiwan). Seeds were sterilized in 1% sodium hypochlorite for 5 min, rinsed 3 times with sterile water, and incubated on Petri dishes for 3 days. Seedlings were planted in peat moss (BVB no. 4, Bas Van Buuren, Coldenhovelaan, The Netherland) in an 8 × 16 plastic tray for 10 days, transplanted to 3-inch pots filled with peat moss, and maintained in a greenhouse. Plants were applied with 1000× Hyponex No. 2 fertilizer (Hyponex, Marysville, OH, USA) once a week for 3 to 4 weeks. The effect of LAB grown in a modified MRS and mixing with chitosan on the control of cabbage black spot disease was evaluated on detached cabbage leaves. The third and fourth leaves of 28-day-old cabage plants were surface-sterilized with 1% sodium hypochlorite, rinsed 3 times with sterile water, air-dried, and used for inoculation. Leaves were inoculated by placing 5 µl each of LAB broth cultures and ABA31 suspensions (3 × 10⁴ spores/ml) on detached cabbage leaves. The leaves were kept in a damp Petri dish for 5 days for lesion development. Biocontrol against ABA31 was also assessed in greenhouse conditions. Broth cultures of LAB were applied to the leaves of 28-day-old cabbage plants by homogeneously spraying to run-off with an air compressor (SIL-AIR 30 HA, Werther International, via Brunelleschi, Reggio Emilia RE, Italy) with air pressure set at 1 atm. The compressor was at 10 to 15 cm away from leaves. Broth cultures of LAB were serially diluted 200× and applied uniformly to cabbage seedlings 1 day before or after the inoculation or co-applied with conidial suspensions (3 × 10⁴ spores/ml) prepared from ABA31. The effect of LAB application frequency was evaluated by spraying broth cultures of LAB over cabbage seedlings at days 14, 21, and 28 (three applications), days 21 and 28 (two applications), or day 28 (one application) after transplant. Cabbage seedlings were inoculated by spraying spore suspensions of ABA31 at day 29.

Biocontrol against Xcc70 was performed by uniformly spraying broth cultures of LAB over cabbage seedlings with an air compressor and challenged with Xcc70 suspensions (10⁸ cfu/ml) by a spraying or scissor-cutting method. All test plants were covered with plastic bags for 24 h and maintained in a greenhouse after plastic bags were removed. Disease severity on the 3rd and 4th leaves was recorded 7 days post inoculation. Disease severity was determined using the following scale: 0, no lesions; 1, lesion areas between 1 and 5%; 2, lesion areas between 5.1 and 10%; 3, lesion areas between 10.1 and 20%; 4, lesion areas between 20.1 and 30%; and 5, between 30.1 and 50% (Conn et al., 1990). Disease severity was calculated by the following formula: Disease severity (%) = [Ʃ (disease index × number of infected leaves)/total number of leaves × 5] × 100.

Each treatment contained at least four replicates and experiments were conducted at least two times. Data were analyzed by ANOVA using SAS/STAT software version 9.0 (SAS Institute Inc., Cary, NC, USA). Significance of treatments was determined using Fisher’s least significant difference test (P ≤ 0.05).

In total, 22 LAB strains were recovered from fermented vegetable juices based on the ability to inhibit PccZL3 using filter paper disc overlaid plate assays. The antagonistic activity was examined further on radish root slices (Fig. 1). LAB strains were cultured in MRS, mixed with PccZL3 suspensions, and together placed on a filter paper overlaid on radish. PccZL3 mixed with water and 1% chitosan induced necrotic lesions on radish. In the presence of chitosan, mixing 2% lactic acid with PccZL3 significantly inhibited the formation of lesions, while mixing MRS medium with PccZL3 had no effects. Two LAB strains designated J02 and J13 (400× dilution, v/v) reduced soft rot severity on radish when mixing with chitosan. In the absence of chitosan, J02 and J13 failed to effectively suppress the development of soft rot lesions (Table 2). Chitosan alone dissolved in 2% lactic acid also had suppressive effects on soft rot severity, which were significantly less effective than chitosan/LAB.

J02 exhibited similarity with Lactobacillus pentosus based on the analyses using a Biolog GEN III MicroStation system and comparison with Biolog’s database. J13 showed similarity with Leuconostoc fallax. Further analysis revealed that J02 16S rRNA sequences shared 99% identity to those of L. pentosus deposited in the database. J13 16s RNA sequences shared 99% identity to those of L. fallax. Phylogenetic analysis suggested that J02 was closely related to L. pentosus and J13 was closely related to L. fallax (Supplementary Fig. 1).

Since chitosan increased the antagonistic ability, 1% chitosan was added into LAB broth cultures for further tests thereafter. J02 or J13 cultured in MRS and mixed with chitosan suppressed the expansion of soft rot lesions on radish roots (#2 treatments in Fig. 2A). To improve the antagonistic effects and to identify a cheaper ingredient for future application, eight nitrogen sources were tested. When J02 and J13 were cultured in a modified MRS, in which both peptone and beef extract in MRS were omitted and replaced with other nitrogen source, the lesion suppression activities varied between J02 and J13 and among different nitrogen sources. J02 was more effective in the suppression of soft rot than J13. Among eight nitrogen sources tested, J02 cultured in MRS or the modified MRS using casein or surimi as the main nitrogen source (pH 4.1) resulted in a strong suppression in lesion sizes. The suppressive ability on the lesion expansion was comparable to J02 cultured in MRS containing peptone and beef extract as nitrogen sources. Similar results were observed when using J13 (pH 5.3) as the test strain. It was noticed that glutamic acid or chicken feather meal when used as the main nitrogen source to culture J13 but not J02 had no effects on disease suppression. Using tyrosine as the nitrogen source to grow J13 resulted in a strong disease suppression; however, tyrosine had no effects when it was used to grow J02. MRS or modified MRS with different nitrogen or carbon source alone had no suppressive effects (data not shown). For both J02 and J13, the disease suppressive ability increased with the increasing concentration of surimi as the nitrogen source and dextrose as the carbon source to culture LAB (Fig. 2B). Modified MRS with surimi and 2% glycerol or sucrose alone had no suppressive effects (data not shown).

Eight different carbon sources were used to culture LAB in order to identify the effective yet cheaper carbon source for future application. Using surimi as the nitrogen source and fructose, sucrose, dextrin, or glycerol as the carbon source to culture J02 (pH 4.1) resulted in a great suppression on soft rot lesions (Fig. 3A). Glycerol combined with surimi to culture J02 resulted in the most effective on disease suppression. Using maltose or corn meal as the carbon source to culture J02 had less or no effect, however. When culturing J13, fructose or sucrose combined with surimi (pH 5.3) resulted in a greater suppression on soft rot compared to other test carbon sources. When using dextrin, glycerol, or corn meal as the carbon source, J13 failed to effectively suppress soft rot severity on radish. In contrast, sucrose combined with surimi to culture J13 resulted in the most effective on disease suppression. The disease suppressive ability of J02 increased with the increasing concentration of glycerol surimi as the carbon and nitrogen sources, respectively (Fig. 3B). Similarly, the disease suppressive ability of J13 increased with the increasing concentration of sucrose and surimi as the carbon and nitrogen sources, respectively. Because chitosan could be completely dissolved in J02 but not in J13 broth cultures, J02 was used for most of the experiments thereafter.

To test the ability of LAB to inhibit different plant pathogens of cruciferous vegetables, a MRSSG medium containing 1.5% surimi and 1.5% glycerol (pH 4.1) as the main nitrogen and carbon sources, respectively, was modified from MRS and used to culture J02. Broth cultures of J02 after being mixed with 1% chitosan suppressed the germination of spores collected from ABA31 even at a 400-fold dilution (Table 3). J02 broth/chitosan had less effect on the suppression of spores collected from C. higginsianum. J02 broth/chitosan suppressed the germination of spores from F. oxysporum f. sp. rapae Focn38 only at a 200-fold or lower dilution; Focn38 spores even germinated slightly better (negative inhibition values) than mock controls in some tests at a 400-fold dilution. The suppressive activity against C. higginsianum was observed when J02 broth/chitosan was applied at a 10 or 20-fold dilution. J02 broth/chitosan could suppress the germination of sclerotia produced by S. rolfsii at a 2-fold dilution and further dilutions of J02 broth had little or no suppressive activity. Similarly, broth cultures of J02 only when applied at a 2-fold dilution had antagonistic activity against Xcc70 in the presence of chitosan.

J02 was cultured in MRSSG (pH 4.1) for 3 days and broth cultures were used for biocontrol experiments after adding 1% chitosan. Inoculation assays revealed that A. brassicicola induced small lesions on detached cabbage leaves after being treated with J02 broth cultures and mixing with chitosan (Fig. 4A). In contrast, A. brassicicola induced large necrotic lesions on cabbage treated with water or MRS only. Quantitation of disease severity revealed that J02 broth even at a 400-fold dilution could reduce black spot lesions by more than 70% compared to those of water or MRS mock controls (Fig. 4B). At a-800-fold dilution, J02 broth reduced cabbage black spot severity by ~50%. Greenhouse trails also revealed that J02 broth at a 200-fold dilution reduced black spot severity on cabbage seedlings considerably (Fig. 5A). Application of J02 broth at a 200-fold dilution 24 h before inoculation or co-application with fungal inoculum reduced black spot severity considerably compared to water controls (Fig. 5B). Application of J02 broth and chitosan 24 h post inoculation was less effective in the reduction of black spot severity on 28-day-old cabbage seedlings. Application of J02 broth at 200× once (at day 28), twice (at days 21 and 28), or three times (at days 14, 21, and 28) significantly reduced black spot severity (Fig. 5C). Application of 2% lactic acid also reduced black spot severity on detached leaf assays.

As stated above, broth cultures of LAB reduced soft rot lesions caused by PccZL3 on radish. Inoculation assays on detached cabbage leaf petioles also revealed that application of J02 broth cultures (pH 4.1) even at a 400-fold dilution plus 1% chitosan completely (100%) suppressed soft rot severity (Fig. 6A). Application of J13 broth plus chitosan was less effective in the reduction of soft rot severity compared to J02. J02 broth cultures and chitosan after being applied onto cabbage leaves three times before inoculation significantly reduced the severity of cabbage black rot caused by Xcc70 using either a scissor-cutting inoculation (Fig. 6B) or a spraying inoculation (Fig. 6C) method.