Marine algae were collected from the intertidal region of Ulleungdo, Gyeongsangbuk-do, Republic of Korea (37°27′36.24″N, 130°51′42.69″E). Fungi were isolated and cultivated as described previously (Chung et al., 2021a). In brief, marine algal samples were cut into segments and grown in potato dextrose agar (PDA; BD, Franklin Lakes, NJ, USA) supplemented with 3% sea salt, 0.01% (w/v) ampicillin, and 0.01% (w/v) streptomycin. After cultivation at 20°C for 14 days, fungal isolates were obtained, stored in a 20% glycerol solution at −80°C, and deposited in the National Marine Biodiversity Institute of Korea (MABIK).

The fungal culture broth was prepared as described previously (Chung et al., 2021b). In brief, five agar plugs were obtained from the fungal colonies grown in PDA at 28°C for 5 days using a cork borer and cultured in 100 ml of potato dextrose broth (PDB; BD) at 28°C for 14 days. The cell-free supernatants (fungal culture broth) were collected by passing them through Miracloth (Millipore, Billerica, MA, USA), followed by 0.22-μm pore size syringe filtration (Corning, New York, USA), and stored at −20°C until use.

Sixteen phytopathogenic bacteria (Acidovorax avenae subsp. cattleyae, Acidovorax konjaci, Agrobacterium tumefaciens, Burkholderia glumae, C. michiganensis subsp. michiganensis, Erwinia amylovora, Erwinia pyrifoliae, Pectobacterium carotovorum subsp. carotovorum, Pectobacterium chrysanthemi, P. syringae pv. actinidiae, P. syringae pv. lachrymans, R. solanacearum, Xanthomonas arboricola pv. pruni, Xanthomonas axonopodis pv. citri, Xanthomonas euvesicatoria, and X. oryzae pv. oryzae) were used for antibacterial activity tests in vitro and in vivo as described in a previous study (Le et al., 2021). The bacterial names, related plant diseases, and cultivation conditions are listed in Table 1. The bacterial strains were cultured in tryptic soy agar (TSA; BD) and tryptic soy broth (TSB; BD) at 30°C unless indicated otherwise.

We isolated 55 marine fungi from various sources (seawater, algae, and sediment) and tested for in vitro antibacterial activity using the broth microdilution method as described in a previous study (Nguyen et al., 2018). Three phytopathogenic bacteria, namely, A. tumefaciens (At), R. solanacearum (Rs), and X. arboricola pv. pruni (Xap), were used to examine the antibacterial activity of the fungal filtrates. These bacterial strains were incubated in TSB at 25 or 30°C (Table 1) at 150 rpm for 1 or 2 days, and the bacterial cultures were diluted to obtain an optical density of 0.1 (ca. 1 × 10⁸ CFU/ml) at 600 nm.

One hundred eighty and 100 μl of TSB containing an individual phytopathogenic bacterium at 1 × 10⁵ CFU/ml (except for 1 × 10⁶ CFU/ml for Xanthomonas spp.) were pipetted into the first and the second well of a 96-well microtiter plate, respectively. Twenty microliters of the fungal culture filtrate was added to the first well to treat the phytopathogenic bacterium with 10% (v/v) fungal broth. Then, 100 μl of the solution in the first well was transferred to the second well by two-fold dilution to make 5% (v/v) fungal broth. One hundred microliters of the solution in the second well was discarded to make a final volume per well of 100 μl. The streptomycin sulfate solution in H₂O (by serial two-fold dilutions at a range from 0.78 to 100 μg/ml) and 10% PDB were added to TSB containing phytopathogenic bacteria as positive controls for antibacterial activity and microbial growth, respectively. In addition, 10% PDB added to TSB without bacterial inoculation was included as a negative control. After incubation at 30°C for 72 h, bacterial growth was evaluated by the naked eye and a microplate reader (HIDEX; Turku, Finland) at 600 nm. The minimum inhibitory concentration (MIC) was defined as the lowest concentration of the culture broth capable of inhibiting bacterial growth. This experiment was performed in triplicate for each sample.

Only the fungal strain showing higher antibacterial activities in vitro was selected and examined for in vivo antibacterial activity against tomato bacterial wilt caused by Rs. One h after the inoculation of Rs in tomato plants, 20 ml of the diluted fungal culture broth (dilutions: 1:5, 1:15, and 1:45) was applied to the plants by soil drenching. Soil drenching is a common application method for pesticides to treat bacterial wilt caused by a soil-borne pathogen, Rs (Lee et al., 2012). The preparation of bacterial inoculum and the rest of the experimental procedures are described in Section 2.8.

Patulin biosynthetic compounds (patulin, gentisyl alcohol, toluquinol, m-cresol, m-hydroxybenzyl alcohol, m-hydroxybenzaldehyde, m-hydroxybenzoic acid, gentisyl aldehyde, and gentisic acid) were purchased from Sigma-Aldrich Co. (St. Louis, MO, USA). To assess the MIC against the phytopathogenic bacteria in vitro, stock solutions of all the compounds were prepared at 20 mg/ml using methanol.

Based on the results of the in vitro antimicrobial activity against phytopathogenic bacteria, strain UL-Ce9 was selected for fungal identification because this strain showed higher activity than other fungal strains tested. The isolation of fungal genomic DNA, polymerase chain reaction (PCR), sequencing of genetic markers, and phylogenetic analysis were performed as described in a previous study (Chung et al., 2021b). UL-Ce9 conidia were cultured in PDB at 28°C, 150 rpm for 3 days. Genomic DNA was extracted from the harvested, frozen, and ground mycelia using phenol:chloroform:isoamyl alcohol (25:24:1) (Sigma-Aldrich). PCR was conducted using the primers Bt2a (5′-GGTAACCAAATCGGTGCTGCTTTC-3′) and Bt2b (5′-ACCCTCAGTGTAGTGACCCTTGGC-3′) to amplify the beta-tubulin gene (benA) (Visagie et al., 2014). The 401 bp-PCR products were purified using a gel extraction kit (Qiagen, Hilden, Germany) and sequenced using the same primers, as described in a previous study (Chung et al., 2021b).

For phylogenetic analysis, closely related sequences were identified using the benA sequence as a query for a BLAST search in GenBank, aligned, and edited manually using the MEGA 6 software (Tamura et al., 2011). The aligned sequences were subjected to neighbor-joining analysis, followed by bootstrapping with 1,000 replicates in MEGA 6.

To isolate the active compounds from the fungal strain UL-Ce9, the fungus was incubated in 1 L of PDB at 28°C for 14 days and filtered using four layers of sterile cheesecloth. The culture supernatants were partitioned sequentially twice with ethyl acetate (EA, 1 L) and butanol (1 L). Water in the EA extraction solution was removed by adding Na₂SO_4. Two organic solvents and one aqueous layer were concentrated to dryness using a rotary evaporator. The antibacterial activities of the individual extracts were examined against Rs, and the extract with the lowest MIC value (highest antibacterial activity) was selected for the next steps to identify antibacterial metabolites.

The EA extract (2.6 g) was subjected to column chromatography in a silica gel column (4.5 cm diameter and 58 cm length, Kiesel gel 60, 230 g, 70–230 mesh; Merck, Darmstadt, Germany) and eluted with CHCl₃/methanol [9/1 (v/v), 800 ml]. All collected fractions were subjected to thin-layer chromatography (TLC; Kiesel gel 60 GF 254, 0.25 mm layer thickness, Merck) using CHCl₃/methanol (9/1, v/v) as a developing solvent system. Fractions (F1–F6) with similar patterns were pooled and assayed in vitro for the growth inhibition of Rs.

The fractions showing antibacterial activity were recombined based on similar TLC patterns and used for TLC bioautography (Wang et al., 2021). The fractions were separated in a thin-layer plate (Kisel gel 60 GF 254, 0.25 mm layer thickness, Merck) with the developing solvent CHCl₃/methanol (9/1, v/v). The TLC plate was sterilized using ultraviolet (UV) light and placed on TSA inoculated with Rs (1.0 × 10⁶ CFU/ml). After 3 days of incubation at 30°C, the antibacterial activity of the fractions was confirmed.

To isolate the antibacterial compounds (CP1, CP2, and CP3), the fractions were subjected to prep-TLC (Kiesel gel 60 GF 254, 0.5 mm thickness, Merck). The developing solvents CHCl₃/MeOH (9/1, v/v) and CHCl₃/MeOH (8/2, v/v) were used to isolate CP1 and CP2, respectively. The purity of the isolated compounds was validated by high-performance liquid chromatography (HPLC, Waters Corp., Milford, Massachusetts, United States). Compounds CP1–CP3 were subjected to HPLC on an Atlantis^® T3 column (5 μm, 4.6 × 250 mm, Waters Corp.) with 0.1% trifluoroacetic acid (Sigma-Aldrich) in water (eluent A) and 0.1% trifluoroacetic acid in acetonitrile (eluent B) as a mobile phase at a flow rate of 1 ml/min. Detection was performed using a Waters 996 detector set at 254 nm.

To determine the chemical structures of CP1–CP3, ¹H-NMR (nuclear magnetic resonance) and ¹³C-NMR spectra were measured in methanol-d₄ (Cambridge Isotope Laboratories, Inc., Tewksbury, MA, USA) using a Bruker Avance III HD 500 MHz instrument (Bruker BioSpin GmbH, Rheinstetten, Germany).

Two types of formulations, namely, a wettable powder (T-WP) and a soluble concentrate (T-SL), were prepared using toluquinol (also known as methylhydroquinone; Sigma-Aldrich). To prepare T-WP, 20 g of toluquinol suspended in a minimal volume of acetone was combined with 15 g of synthetic hydrated silicon dioxide (white carbon; Rhodia Asia Pacific Pte. Ltd., Singapore). After evaporating the acetone in a fume hood, 5 g of CR-100 (a wetting agent; Yoosung Chemical R&T Co. Ltd.), 5 g of CR-SDS (a surfactant; Yoosung Chemical R&T Co. Ltd.), and 55 g of kaolin (a dispersant) were added. Finally, the mixture was ground using an electric powder grinder. To prepare T-SL, 10 g of toluquinol suspended in 40 ml of ethanol was mixed with 20 g of CR-MC33 (an emulsifier; Yoosung Chemical R&T Co. Ltd.) and water to obtain 100 ml of the final product.

Rice seeds (8 g) were treated with a sterilizer (1:2,000 diluted Sportak EC (Prochloraz 25%), Hankook Samgong, Seoul, Korea) at 30°C for 1 day and soaked in freshwater at 30°C for 4 days. The seeds (5 seeds/pot) were sown in plastic pots (8 cm in diameter and 7 cm in height) and grown under a light regime of 14-h light/10-h dark at 28–36°C and 90% relative humidity for 6 weeks.

The causative agent of bacterial leaf blight of rice X. oryzae pv. oryzae (Xoo) was cultured in TSA at 30°C for 2 days, and the cells were resuspended in 10 mM MgCl₂ solution (pH 7.0). Cell density was measured using a UV spectrophotometer at 600 nm, and the bacterial inoculum suspension was adjusted to 2.0 × 10⁷ CFU/ml. One day before the inoculation of Xoo, 20 ml of diluted T-WP (1:1,000, 1:2,000, and 1:4,000 dilutions), T-SL (1:500, 1:1,000, and 1:2,000 dilutions), toluquinol (33.3, 100, and 300 μg/ml), gentisyl alcohol (33.3, 100, and 300 μg/ml), and patulin (33.3, 100, and 300 μg/ml) were applied to the rice by foliar spraying until runoff onto the leaves. Streptomycin sulfate (33.3, 100, and 300 μg/ml) and Tween 20 (500 μg/ml) were used as positive and non-treated controls, respectively.

To infect rice, scissors were dipped in the Xoo suspension and used to cut the tips (3–4 cm from the tips) of rice leaves. The rice plants were grown under a light regime of 12-h light/12-h dark at 28–36°C and 90% relative humidity. Disease severity (DS) was recorded 14 days after bacterial inoculation (DAI). The experiment included three biological replicates (three leaves per treatment), and the entire experiment was performed twice.

Disease control efficacy against tomato bacterial wilt was examined as described in a previous study (Im et al., 2020). Tomato seeds (Seokwang, Farm Hannong Co, Seoul, Republic of Korea) were grown in pots (4.5 cm in diameter) containing horticulture nursery soil (Punong Co., Gyeongju, Republic of Korea) until the tomato plants reached the 4–5 true-leaf stage. The plants were then transplanted into pots (7 cm in diameter and 6 cm in height) and grown until the 7–8 true-leaf stage.

Ralstonia solanacearum, the causative agent of bacterial wilt, was cultured in TSA at 30°C for 2 days, and the cells were harvested using sterile water and a cell scraper. The cell density was measured using a UV spectrophotometer at 600 nm, and the bacterial inoculum suspension was adjusted to 1.5 × 10⁸ CFU/ml. One day before the inoculation of Rs, 20 ml of diluted T-WP (1:1,000, 1:2,000, and 1:4,000 dilutions) and T-SL (1:500, 1:1,000, and 1:2,000 dilutions) were applied to the tomato plants by soil drenching. The commercial pesticide Sungbocycline (oxytetracycline calcium alkyltrimethyl ammonium 17%, 1:1,000 dilution in accordance with manufacturer's recommendations; Sungbo Chemicals Co, LTD., Seoul, Republic of Korea) and Tween 20 (250 μg/ml) were used as positive and non-treated controls, respectively.

Twenty milliliters of the Rs suspension were drenched into the soil, and the plants were maintained at 30°C and 95% humidity for 2 weeks. DS was recorded at 9 and 14 DAI. The experiment included three biological replicates (three plants per treatment), and the entire experiment was performed twice.

To evaluate the disease control efficacy against rice leaf blight and tomato bacterial wilt, DS was visually estimated on a 6-point scale (0 = no symptoms, 1 = one leaf partially wilted, 2 = one to two leaves wilted, 3 = two of three leaves wilted, 4 = four or more leaves wilted, and 5 = death of the entire plant) as described in a previous study (Winstead and Kelman, 1952). The control efficacy values for the individual plant diseases were calculated as follows:

Experimental data were analyzed using a one-way or two-way analysis of variance, followed by Tukey's multiple comparison tests or the Bonferroni post-test (GraphPad Prism 5.0).

The in vitro antibacterial activities of the marine fungi were studied against At, Rs, and Xap. Among the tested fungal strains, UL-Ce9 isolated from the marine alga, Ceramium sp., exhibited the highest antibacterial activity against At, Rs, and Xap. The MIC value of UL-Ce9 culture broth was 5% (v/v) against all the phytopathogenic bacteria examined.

The culture broth of fungi showing high antibacterial activities in vitro was subjected to an in vivo antibacterial activity test against the bacterial wilt of tomato. The uninoculated tomato plants did not show any wilting symptoms. In contrast, the tomato plants inoculated with Rs displayed typical wilting symptoms. This indicates that the tomato plants were infected with Rs. Among the marine fungi examined, the culture broth (1:5 dilution) of UL-Ce9 completely suppressed the disease (100% ± 0.00 disease control efficacy). Moreover, the disease control efficacies of UL-Ce9 filtrates at dilutions of 1:15 and 1:45 were 83.3 ± 2.3% and 76.2 ± 3.3%, respectively (Supplementary Figure 1).

UL-Ce9 was isolated from the blades of the marine alga Ceramium species. The UL-Ce9 β-tubulin gene sequence (deposited in GenBank; accession number OQ559567) showed the highest nucleotide similarity with that of Penicillium griseofulvum CMV005F8 (99.75%) and the P. griseofulvum type strain CBS 185.27 (99.24%; E-values = 0). The UL-Ce9 β-tubulin sequence was also closely related to Penicillium dipodomyicola NRRL 13487 with 94.64% sequence similarity (E-value = 1e-167). A neighbor-joining phylogenetic analysis using the β-tubulin sequence showed that UL-Ce9 formed a clade with these two P. griseofulvum strains, supported by a 100% bootstrap value. Therefore, UL-Ce9 was identified as P. griseofulvum (Figure 1).

The antibacterial activities of the EA, butanol, and water extracts of UL-Ce9 were examined against Rs in vitro, and the MIC values were 3.9, 7.8, and 250 μg/ml, respectively. As the EA extract showed the highest antibacterial activity, it was used for the isolation, purification, and identification of the antibacterial compounds of P. griseofulvum (Figure 2A).

More than 81 fractions obtained via chromatography were grouped into six fractions (F1–F6) based on separation patterns from the TLC analysis (Figure 2A). The antibacterial activities of F1 (210 mg), F2 (186 mg), F3 (250 mg), F4 (120 mg), F5 (162 mg), and F6 (250 mg) were assessed against Rs, and the MIC values were 31.2, 3.9, 3.9, 3.9, and 3.9 μg/ml for F1, F2, F3, F4, and F5, respectively. F6 did not show any antibacterial activity against Rs.

Compounds 1 (CP1) and 2 (CP2) were purified from fractions F13 and F45 by prep-TLC. Active CP1 and CP2 displayed one major spot in the TLC analysis, which was detected at an R_f value of 0.7 and 0.5, respectively. The purities of CP1 and CP2 were validated by HPLC at 254 nm, and the absorption maxima of CP1 and 2 were observed at 276 and 291 nm, respectively (Figure 2B).

The liquid chromatography-electrospray ionization-mass spectrometry (LC-ESI-MS) of CP1 displayed an ion peak [M + H]⁺ at m/z 155.1, suggesting a molecular formula of C₇H₅O₄, similar to that of patulin. Further evidence was obtained from the two proton signals at δ 4.53 (1H, dd, 16.8 and 3.0 Hz, and H-5a) and δ 4.34 (1H, dd, 16.8 and 3.0 Hz, and H-5b) in the ¹H-NMR spectrum and the signals of one oxygenated methine at δ 87.94 (C-4) in the ¹³C-NMR spectrum (Table 2). Two methine protons at δ 6.21 (1H, m, and H-2) and δ 6.19 (1H, m, and H-6) were attached to a double bond in the ¹H-NMR spectrum, and one carbon showed resonance at δ 168.78 (C-1) in the ¹³C-NMR spectrum. In the ¹H-NMR spectrum, the singlet signal was observed at δ 5.92 (1H and s; Figure 3). These data suggest that CP1 is patulin (Yoo and Park, 1995).

The ESI-MS of CP2 displayed an ion peak [M]⁺ at m/z 140, suggesting a molecular formula of C₇H₈O₃, similar to that of gentisyl alcohol. The ¹H-NMR showed aromatic protons at δ 6.61 (1H, d, 8.55, and H-3), 6.54 (1H, dd, 3.0, 8.55, and H-4), and 6.75 (d, 2.75, H-6) and methylene protons at δ 4.59 (2H, s, and H-7; Table 2). These data displayed phenolic hydroxyl groups at δ 4.96 (s and 7-OH). The presence of an OH group at C-2 and C-5 was confirmed by the fragment ion at m/z 122 (100; Figure 3). Based on these data, CP2 was identified as gentisyl alcohol (Ishikawa et al., 1986).

Compound 3 showed one spot in the TLC analysis with an R_f value of 0.6 when irradiated with UV light at 254 and 365 nm. CP3 showed the maximum UV absorption at 290 nm (Figure 2B) by HPLC analysis. The ¹H-NMR showed aromatic protons at δ 6.84 (1H, dd, 8.60, 2.88, and H-3), 6.59 (1H, dd, 8.6, 0.46, and H-4), and 6.53 (dd, 2.88, 0.4, and H-6) and a methyl group at δ 2.15 (3H, s, and H-7) (Table 2 and Figure 3). Based on these data, CP3 was identified as toluquinol (Huang et al., 1989).

Because patulin, gentisyl alcohol, and toluquinol are associated with patulin biosynthesis (Puel et al., 2010), we investigated the antibacterial activities of related metabolites (m-cresol, m-hydroxybenzyl alcohol, m-hydroxybenzaldehyde, m-hydroxybenzoic acid, gentisyl aldehyde, and gentisic acid) in vitro (Table 3).

Overall, compared with patulin, gentisyl alcohol, and toluquinol, the other six metabolites exhibited lower antibacterial activities against all phytopathogenic bacteria examined. In particular, the MIC values of m-cresol, m-hydroxybenzyl alcohol, m-hydroxybenzaldehye, m-hydroxybenzaldehyde, and gentisic acid were over 200 μg/ml, except 100 μg/ml for m-hydroxybenzaldehyde and 12.5 μg/ml for gentisic acid against Xoo. Gentisyl aldehyde showed higher antibacterial activity than those of the other five metabolites, with MIC values similar to those of toluquinol against B. glumae (Bg), Pectobacterium carotovora subsp. carotovora (Pcc), P. chrysanthemi (Pc), P. syringae pv. actinidiae (Psa), and Rs.

Patulin showed the highest antibacterial activities against most phytopathogenic bacteria except A. avenae subsp. cattleyae (Aac), A. konjaci (Ak), Xap, X. euvesicatoria (Xe), and Xoo. The lowest MIC values (3.13 μg/ml) of patulin were observed against C. michiganensis subsp. michiganensis (Cmm), Rs, and Xoo, followed by Xap (MIC 6.25 μg/ml). The highest antibacterial activity of gentisyl alcohol was observed against Xoo (MIC 3.13 μg/ml). In addition, it showed strong inhibitory effects on the growth of Rs and Xap (MIC 6.25 μg/ml). A metabolite with the highest antibacterial activity against Xoo was toluquinol (0.78 μg/ml). Compared with gentisyl alcohol, toluquinol showed higher activity against nine phytopathogenic bacteria (Ak, At, Pcc, Pc, Psa, Psl, Xap, Xe, and Xoo).

Due to the potentially negative effects of patulin on human health (Pal et al., 2017), we prepared two types of pesticide formulations, namely, T-SL and T-WP, using toluquinol to assess the disease control efficacy against bacterial plant diseases.

Patulin, gentisyl alcohol, and toluquinol, as well as T-SL and T-WP, were evaluated for their efficacy against rice leaf blight caused by Xoo (Figure 4). Overall, all metabolites and formulations effectively suppressed the development of leaf blight of rice, similar to the positive control, streptomycin sulfate, except for gentisyl alcohol at 33.3 μg/ml. Gentisyl alcohol was less effective than patulin to suppress leaf blight of rice at 33.3 μg/ml (P < 0.05). There were no differences in the efficacies of different concentrations of each compound/formulation. For example, the efficacies of T-SL were 36.9 ± 17.9% and 45.0 ± 16.2% at 1:1,000 and 1:500 dilutions, respectively, which were not statistically different (P > 0.05). Similarly, T-WP exhibited control efficacies of 47.1 ± 10.0% and 56.4 ± 9.4% at 1:1,000 and 1:500 dilutions. At 100 μg/ml, the control efficacies of patulin, gentisyl alcohol, and toluquinol were 56.4 ± 9.4%, 39.8 ± 15.8%, and 45.6 ± 5.0%, respectively. Notably, patulin showed phytotoxicity in rice plants at 300 μg/ml, causing leaf desiccation and leaf spots on the leaves 2 days after treatment. These data suggest that toluquinol and its formulations exert strong suppressive effects on leaf blight of rice, similar to the currently used chemical (streptomycin) for disease management.

The disease control efficacy of toluquinol against tomato bacterial wilt caused by Rs was investigated using two toluquinol formulations, T-SL and T-WP (Figure 5). Overall, both the toluquinol formulations showed higher control efficacies, even at the lowest concentrations, compared with that shown by the positive control, a commercial pesticide Sungbocycline (oxytetracycline), at 9 DAI (P < 0.05). At 14 DAI, the control efficacies of the most diluted T-SL (1:2,000) and T-WP (1:4,000) formulations slightly decreased compared with those at 9 DAI, similar to that of Sungbocycline (P > 0.05). In contrast, at higher concentrations, T-SL (1:500) and T-WP (1:1,000) completely suppressed the bacterial wilt of tomato (100 ± 0.00% efficacy) at both 9 DAI and 14 DAI, whereas the control efficacy of Sungbocycline was 66.7 ± 7.2% and 51.11 ± 7.7%, respectively. These results suggest that toluquinol suppresses bacterial wilt of tomato more effectively than the currently used pesticides.