Mode of antagonism of Brevibacillus brevis against Botrytis cinerea in vitro

Authors


Seddon Department of Agriculture and Forestry, MacRobert Building, University of Aberdeen, 581, King Street, Aberdeen AB24 5UA, Scotland, UK (e-mail: b.seddon@abdn.ac.uk).

Abstract

Aims: To assess the activity of Brevibacillus brevis (formerly Bacillus brevis) Nagano and the antibiotic it produces, gramicidin S, against the plant pathogen Botrytis cinerea.

Methods and Results: Germination and growth of Bot. cinerea were assessed in the presence of B. brevis or gramicidin S in liquid media, on solid media and on leaf sections of Chinese cabbage. Germination was 10-fold more sensitive to gramicidin S than growth. Inhibition of Bot. cinerea was greater in liquid media compared with on solid media. Activity of gramicidin S against Bot. cinerea on leaf sections was much lower than in vitro. In vitro inhibition of Bot. cinerea by B. brevis Nagano was similar to equivalent levels of gramicidin S.

Conclusions: Antibiosis, via gramicidin S, is the mode of antagonism exhibited by B. brevis Nagano against Bot. cinerea in vitro.

Significance and Impact of the Study: The mode of antagonism of B. brevis against Bot. cinerea was elucidated. The differing activity of gramicidin S against Bot. Cinerea in vitro and on leaf sections indicates one mechanism by which biocontrol activity may differ between laboratory and field conditions.

INTRODUCTION

Botrytis cinerea is an important pathogen of numerous plant genera (Jarvis 1977) and is known to cause extensive damage to protected crops (Jarvis 1992). Botrytis cinerea can cause grey mould on seedlings and mature foliage of Chinese cabbage (Brassica campestris subsp. pekinensis) (Anon 1984). Several Bacillus and related species have been tested as potential biological control agents (BCAs) as they produce a range of antibiotics and form resistant spores (Edwards et al. 1994). One of these, Brevibacillus brevis (formerly Bacillus brevis Nagano (Shido et al. 1996)), produces a single cyclic decapeptide antibiotic, gramicidin S, which has been shown to be fungicidal (Murray et al. 1986).

In vitro studies of antagonism are generally the preliminary studies in most biocontrol work. These tests are used to screen a large number of possible BCA isolates (Merriman et al. 1974) to determine the mode of antagonism (Dennis and Webster 1971a, b, c), or to rate the efficacy of individual isolates (Utkhede and Sholberg 1986). The suspected mode of antagonism, antibiosis, for B. brevis Nagano could be tested by comparing the activity of B. brevis Nagano against Bot. cinerea with that of pure gramicidin S and the antibiotic-negative mutant, B. brevis E-1 (Iwaki et al. 1972). In this way, a clear demonstration of the role of gramicidin S could be made.

In planta experiments allow the potential biocontrol system to be tested within the infection court of the pathogen and as such, the tests are nearer to in situ conditions than in vitro experiments (Janisiewicz and Roitman 1988). Leaf discs or small detached leaves are frequently used for in planta studies (Baker et al. 1983). The level of antagonism can be assessed by a variety of methods depending on the disease symptoms experienced in the host–pathogen interaction under study. Visual, discrete local lesions can be counted (Last and Hamley 1956) or the size of individual spreading lesions can be measured (Andrews et al. 1983). Botrytis cinerea produces local lesions on a number of hosts. However, in some host–pathogen interactions, the macroscopic symptoms do not lend themselves to quantification, and it is necessary to assess antagonism occurring on the leaf surface microscopically by measuring percentage germination of conidia or germ-tube growth. Yoder and Whalen (1975) stained leaf sections with aqueous phenolic Rose Bengal, which resulted in a proportion of ungerminated conidia being washed from the leaf during staining and subsequent washings. Thus, the true percentage germination could not be determined without knowledge of the number of ungerminated conidia washed from the leaf prior to microscopic observation. The optical brightener, Calcofluor, used in the experiments described here, has proved especially useful as it is non-toxic and is readily adsorbed by micro-organisms (Dolan and McNicol 1986). Calcofluor, unlike natural fluorescent material associated with plant surfaces, is excited by short wavelength u.v. light. There are several benefits of labelling Bot. cinerea conidia with Calcofluor. Firstly, labelled conidia are distinguishable from any natural contaminants present on leaves. Secondly, labelled conidia can be observed non-destructively over a period of time and thirdly, labelled conidia can be observed on plant material without loss of ungerminated conidia. Calcofluor labelling also has no detectable effect on the viability or virulence of Bot. cinerea conidia (Edwards 1993).

The aim of this paper was to test the activity of B. brevis Nagano and gramicidin S against Bot. cinerea in vitro and on leaf surfaces.

MATERIALS AND METHODS

Bacterial cultures

Stocks of B. brevis Nagano and B. brevis E-1 were originally supplied by Professor Y. Saito (Hyogo College of Medicine, Japan). Micrococcus luteus NCIMB 8166 was supplied by the National Culture Collections of Industrial and Marine Bacteria Ltd. (Aberdeen, UK). All isolates were maintained as frozen stocks. Bacterial cultures were plated out onto Nutrient Agar (Oxoid) and incubated at 37°C for 5 days (30°C for 2 days for M. luteus). Bacteria were transferred to 5 ml half-strength Nutrient Broth (Oxoid), which contained glycerol (10% v/v), and stored at –70°C. Bacteria were spread onto Nutrient Agar plates from frozen stocks. These plates were incubated overnight at 37°C (30°C for M. luteus) and used to inoculate 20 ml Nutrient Broth (Oxoid) in a 100 ml conical flask. These seed cultures were incubated overnight at 37°C (30°C for M. luteus), 150 rev min–1. A 0·1 ml aliquot of B. brevis seed culture was used to inoculate 500 ml tryptone soya broth (Oxoid) in 2 l conical flasks. Flasks were then incubated at 37°C and 150 rev min–1 for 14 days. Cultures were used either in this form or as washed spores. To wash spores, cultures were spun in a bench centrifuge at 3000 g for 10 min. Pellets were resuspended in 15 ml sterile distilled water (SDW) and spun at 2000 g for 10 min. This washing procedure was repeated five times. Spore concentrations were determined using a Thoma counting chamber (Weber Scientific International, Teddington, UK) and adjusted to 108–1010 spores ml–1. During the counting of spores, the percentage of phase-bright spores was checked and was consistently found to be greater than 90%. Spores were stored at –20°C until used.

Fungal cultures

A strain of Bot. cinerea (YCC) was isolated from a field crop of Chinese cabbage in Yorkshire, UK. To maintain pathogenicity, YCC was used to inoculate Chinese cabbage seedlings. Chinese cabbage cv. Granaat untreated seed was washed in 90% ethanol for 1 min, then in 4% sodium hypochlorite for 5 min, and rinsed three times in SDW. Individual seeds were aseptically placed into 60 ml sterile plastic universals containing 10 ml Dowson’s Basal Medium (1 g diammonium hydrogen phosphate, 0·2 g potassium chloride, 0·2 g magnesium sulphate heptahydrate and 15 g Agar no. 3 (Oxoid) in 1 litre SDW). The seeds were placed in a growth cabinet at 15°C, 100% relative humidity (RH), on an 18 h day/6 h night cycle. Seedlings were inoculated with Bot. cinerea conidia after 5–6 days. A 100 μl drop of SDW was placed on a malt extract agar (Oxoid) plate of sporulating Bot. cinerea, mixed with the conidia and placed on the leaf of a Chinese cabbage seedling. Once good sporulation had occurred, infected seedlings were frozen at –20°C and used as stock. Conidia from frozen seedlings were subcultured onto malt extract agar plates to check for purity. Conidia from these plates were subsequently used as inocula for conidial harvests or further stocks. For conidial harvests, malt extract agar plates were centrally inoculated with conidia and incubated at 25°C for 3–4 days. Mycelial plugs from the edge of these colonies were used to inoculate Medium X + 10% Sucrose plates (Harrison 1978), and incubated at 25°C until mycelia had reached the edge of the plates (4–5 days). Plates were transferred to 25°C under near-u.v. light for 2 days, and then returned to the laboratory and left for 14 days at room temperature before spores were harvested. Conidia were harvested by addition of 10 ml SDW to each plate and conidia dislodged using a sterile, blunt-ended spatula. The conidial suspension was poured through two layers of sterile Whatman 105 lens tissue (Whatman International Ltd, Maidstone, UK), centrifuged at 3000 g for 10 min, resuspended in 15 ml SDW and centrifuged at 3000 g for 10 min. This washing procedure was repeated twice. The concentration of conidia was determined using an Improved Neubauer counting chamber (Weber) and adjusted to 106–108 conidia ml–1. The conidial suspension was frozen to –20°C until required.

Fluorescent staining of Botrytis cinerea conidia

For specific observations on germination, some conidia of Bot. cinerea were stained with the vital stain, Calcofluor (Sigma), based on the method of Dolan and McNicol (1986). These conidia were harvested as detailed above, except that they were resuspended in a 0·5% aqueous solution of Calcofluor after the first wash. The Calcofluor solution was prepared in SDW and filter sterilized. The conidial suspension was held at 4°C for 4 h, after which time conidia were then washed three times using 15 ml SDW with centrifugation at 3000 g for 10 min. The determination of concentration, and storage, was as above.

Ethanol extraction and bioassay of gramicidin S

Plant material (2 g), or the pellet from 20 ml bacterial culture (centrifuged at 3000 g for 10 min), was added to 20 ml ethanol in a 50 ml round-bottomed flask fitted with a reflux condenser. The sample was boiled for 10 min and allowed to cool in a fume cupboard. Upon cooling, the sample was again centrifuged at 3000 g for 10 min. The supernatant fluid was then bioassayed either directly or after decimal dilution in ethanol. A 2 ml aliquot of an overnight culture of M. luteus was used to seed 150 ml Nutrient Agar that had been autoclaved and cooled to approximately 45°C. The medium was poured into a sterile 250 × 250 × 20 mm tissue culture plate (Global Science and Technology Ltd, Auckland, NZ). The agar was left to dry for 30 min before wells were made using a sterile 7 mm diameter cork borer. Five replicates of each test solution and a range of gramicidin S standards (0–100 μmol l–1) were randomly positioned on the plate. The diameters of zones of inhibition were measured after 24 h of incubation at 30°C. Standards used were either aqueous or ethanolic, depending on the nature of the sample, and where necessary, test solutions containing more than 100 μmol l–1 gramicidin S were diluted 10-fold and re-assayed.

Inhibition of Bot. cinerea germination in vitro

Inhibition on glass slides.

A 1 ml volume of glucose peptone broth (GPB) (4 g bacteriological peptone (Oxoid), 20 g glucose in 1 litre SDW, pH adjusted to 6·0 with 1 N sodium hydroxide) containing gramicidin S (0–20 μmol l–1) and 1 × 105 conidia of Bot. cinerea was vortexed on mixing for 10 s, before 40 μl were placed on a sterile glass slide. Slides were incubated at 25°C, 100% RH, for 16 h, after which time they were covered with a coverslip and observed at 100 × magnification using a phase-contrast microscope (M14 Photoplan, Vickers Instruments, York, UK). The number of germinated conidia within a count of 100 conidia was determined on triplicate slides for each sample. A conidium was determined as having germinated if germ tube emergence had commenced. This method was repeated with B. brevis Nagano and E-1 spores instead of the antibiotic, with final concentrations of bacterial spores in the range 2 × 104–4 × 105 ml–1.

Inhibition in liquid media.

A 20 ml volume of GPB containing gramicidin S (0–20 μmol l–1) in 100 ml conical flasks was autoclaved at 121°C for 15 min. Flasks were inoculated with 0·1 ml of Bot. cinerea conidial stock to give a final concentration of 105 conidia ml–1. Tests were carried out in triplicate flasks that were incubated at 25°C, 100 rev min–1, for 5 days. Biomass was determined by filtering the cultures with pre-weighed lens tissue. Flasks were washed twice with 20 ml SDW and these washings were also passed through the lens tissues. After filtration, lens tissues were dried overnight at 80°C and re-weighed. Twenty uninoculated flasks of GPB were also filtered. Confidence limits for the weight variation of these controls were calculated. This experiment was repeated with B. brevis Nagano and E-1 using 2 × 103–2 × 105 spores ml–1.

Inhibition under different incubation conditions.

Conical flasks were prepared and inoculated as above. After various time periods (0–16 h), 40 μl were removed from each flask and placed on sterile glass slides. These samples were incubated at 25°C, 100% RH, for the remainder of the 16 h experiment. After 16 h, the percentage of germination was determined as described above.

Inhibition of Bot. cinerea growth in vitro

Inhibition in liquid culture.

Conical flasks (100 ml volume), each containing 20 ml GPB and a range of gramicidin S concentrations (0–100 μmol l–1), were prepared as outlined above. Two 10 mm mycelial plugs of Bot. cinerea from GPA (GPB plus 15 g Agar no. 3 (Oxoid)) plates were placed into each conical flask and incubated at 25°C, 100 rev min–1, for 5 days. Cultures were then filtered through pre-weighed lens tissue and dried at 80°C, overnight, before being weighed. Confidence limits for the weight variation of 20 uninoculated controls were calculated. This experiment was repeated with B. brevis Nagano and E-1 using 2 × 104–2 × 106 spores ml–1.

Inhibition on solid medium.

Ten bottles, each containing 250 ml GPA and gramicidin S (0–1 mmol l–1), were prepared and autoclaved. Each 250 ml bottle was then used to pour 10 plates. Plates were inoculated with 12 mm diameter plugs of Bot. cinerea. Plates were incubated for 5 days at 25°C, and the final diameter of each colony was then measured and recorded.

Inhibition of Bot. cinerea germination and gramicidin S on leaf surfaces

Inhibition of germination.

Leaf material for these experiments was commercially-grown organic Chinese cabbage. The outer leaves were discarded and 50 × 30 mm sections from the white leaf base were removed from inner leaves. These sections were inoculated with two 10 μl drops of Calcofluor-labelled Bot. cinerea conidia (106 ml–1) on the adaxial surface, either side of the central vein. Triplicate leaf sections were incubated for 5 days at 15°C, 100% RH, in the dark, then dried at 60°C and observed under a fluorescence microscope (Polyvar, Reichert-Jung, Vienna, Austria) with an incident exciter filter 330–380 nm and 400 × magnification. The percentage of germination was determined. Treated leaves were sprayed, using a sterile Devilbiss atomiser no. 15 (Sunrise Medical Ltd, Wollaston, UK), with gramicidin S solutions (0–500 μmol l–1) or B. brevis cultures (0–2 × 107 spores ml–1) until run-off, and air-dried. Sections were then removed and inoculated with Calcofluor-labelled conidia of Bot. cinerea.

Gramicidin S adsorption to the leaf surface.

Six leaves of equivalent age and size were cut from 6-week-old Chinese cabbage cv. Graanat plants grown in a greenhouse. Leaves were trimmed to produce six leaf sections of equal size, all weighing 1·0 g. Each section was added to a sterile glass universal bottle containing 10 ml aqueous gramicidin S ranging from 0 to 50 μg ml–1 (0–41 μmol l–1). The bottles were agitated for 1 h before the solutions were bioassayed for gramicidin S. The leaf sections were resuspended in 10 ml ethanol and boiled under reflux for 10 min in a round-bottomed 50 ml flask with a water-cooled condenser. This extraction procedure was repeated twice before the leaves were discarded; the combined ethanol extracts were then evaporated down to 2 ml and bioassayed for gramicidin S, as described previously.

Gramicidin S activity in ethanolic leaf extracts.

Four leaves from a commercially-grown, organic Chinese cabbage were trimmed to produce four leaf sections of equal size, all weighing 1·0 g. The leaf sections were placed in a round-bottomed 100 ml flask, with a water-cooled condenser, with 40 ml ethanol and boiled under reflux for 10 min. After cooling, seven 4·5 ml aliquots of the ethanolic extract were placed into sterile test tubes. To these, 0·5 ml gramicidin S ethanolic stock solutions were added to give a final concentration of 0–100 μg ml–1 (0–82 μmol l–1). These solutions were mixed for 30 min before being bioassayed.

RESULTS

Gramicidin S bioassay

Zones of inhibition were converted to gramicidin S concentration using a linear regression of log (GS concentration) against (diameter of inhibition zone)2. All regressions used had an r2 greater than 0·95. The sensitivity limit of the bioassay was 5 μmol l–1. For the in vitro experiments, a B. brevis Nagano spore stock was diluted to 107 spores ml–1 and its gramicidin S concentration was 410 μmol l–1. Ethanol extraction of the B. brevis E-1 (gramicidin S-negative mutant) spore stock (107 spores ml–1) did not produce a zone on the bioassay plate and therefore, contained less than 5 μmol l–1 gramicidin S. For the in planta experiments, a B. brevis Nagano spore stock was diluted to 2 × 107 spores ml–1 and its gramicidin S concentration was 530 μmol l–1. Ethanol extraction of the B. brevis E-1 spore stock (107 spores ml–1) did not produce a zone on the bioassay plate and therefore, contained less than 5 μmol l–1 gramicidin S.

Inhibition of Bot. cinerea germination in vitro

Inhibition on glass slides.

Percentage germination of Bot. cinerea was determined after 16 h of germination in GPB on glass slides at 25°C, 100% RH, in the presence of a range of gramicidin S, B. brevis Nagano and E-1 spore concentrations (Table 1). Germination and germ-tube growth were markedly reduced by 10 μmol l–1 gramicidin S, and complete inhibition of germination occurred at 15 μmol l–1 gramicidin S. In the absence of gramicidin S, Bot. cinerea germinated and formed branched septate hyphae after 16 h. Conidia which did germinate in the presence of 10 μmol l–1 gramicidin S produced short, swollen germ tubes of uneven diameter. Brevibacillus brevis Nagano caused complete inhibition of Bot. cinerea germination at 3 × 105 spores ml–1. This spore concentration contained 12 μmol l–1 gramicidin S. The antibiotic-negative B. brevis E-1 failed to inhibit Bot. cinerea germination even at the highest concentration of 4 × 105 spores ml–1.

Table 1.   Germination of conidia of Bostrytis cinerea in the presence of gramicidin S, Brevibacillus brevis Nagano and B. brevis E-1 on glass slides. Mean (S.D.) from triplicate glass slides Thumbnail image of

Inhibition in liquid media.

Percentage inhibition of inoculated conidia was based on the mean weight increase of uninoculated controls. Flasks with a biomass value within the 95% confidence limits of uninoculated controls (–2·91 to 5·43 mg) had a percentage inhibition of 100%. Inhibition of germination as measured by biomass after 5 days of incubation (Table 2) was greater than as measured by percentage germination after 16 h on glass slides (Table 1). Botrytis cinerea germination in liquid culture was completely inhibited by 5 μmol l–1 gramicidin S. Brevibacillus brevis Nagano (105 spores ml–1), which contained 4 μmol l–1 gramicidin S, gave a level of inhibition equivalent to 5 μmol l–1 of pure antibiotic. Brevibacillus brevis E-1 caused no inhibition of Bot. cinerea, even at the highest concentration of 2 × 105 spores ml–1.

Table 2.   Germination of Botrytis cinerea conidia in the presence of gramicidin S, Brevibacillus brevis Nagano and B. brevis E-1 in liquid culture. Mean (S.D.) from triplicate flasks Thumbnail image of

Inhibition under different incubation conditions.

Conidia that were transferred immediately onto a glass slide from a flask were inhibited by gramicidin S to a lesser extent than those that remained in flasks for the full 16 h (Fig. 1). Complete inhibition on glass slides occurred at 15 μmol l–1 gramicidin S, and at 5 μmol l–1 gramicidin S in flasks, indicating that the method of incubation did affect the sensitivity of Bot. cinerea conidia to gramicidin S. Percentage germination of Bot. cinerea conidia in the absence of gramicidin S was unaltered by the transfer time from flask to glass slide. The mean percentage germination and standard deviation for these controls was 90·6 ± 1·12.

Figure 1.

 Germination of Botrytis cinerea in the presence of gramicidin S after transfer from flasks to glass slides. Transfer time (min): (▮), 0; (□), 0·25; (▴), 10; (○), 60; (●), 960

Inhibition of Bot. cinerea growth in vitro

Inhibition in liquid culture.

Biomass values within the 95% confidence limits of uninoculated controls (21·0–33·2 mg) had no significant growth and hence, a percentage inhibition of 100%. The average weight of uninoculated controls (27·1 mg) was deducted from the filtered mass value to give the biomass for each flask.

Complete inhibition of Bot. cinerea growth from inoculated mycelial plugs in liquid culture occurred at 50 μmol l–1 gramicidin S (Table 3). This was 10-fold greater than the gramicidin S concentration required for complete inhibition of germination of Bot. cinerea under the same conditions. Brevibacillus brevis Nagano gave levels of inhibition of Bot. cinerea growth similar to the pure antibiotic, in that 106 spores ml–1 (containing 41 μmol l–1 gramicidin S) totally inhibited mycelial growth of Bot. cinerea. Brevibacillus brevis E-1 at the highest concentration of 2 × 106 spores ml–1 resulted in negligible inhibition of Bot. cinerea growth (7·6%).

Table 3.   Growth of Botrytis cinerea from mycelial plugs in the presence of gramicidin S, Brevibacillus brevis Nagano and E-1 in liquid culture. Results are mean (S.D.) from triplicate flasks Thumbnail image of

Inhibition on solid medium.

Botrytis cinerea growth on solid medium was inhibited 24% by 10 μmol l–1 gramicidin S, but increasing the concentration further had little effect on the level of inhibition, even at 1 mmol l–1 gramicidin S (Table 4).

Table 4.   Growth of Botrytis cinerea from mycelial plugs in the presence of gramicidin S, on GPA plates. Results are mean (S.D.) of colony diameter from 10 replicates Thumbnail image of

Inhibition of Bot. cinerea germination and gramicidin S on leaf surfaces

Inhibition of germination.

Percentage germination of conidia on untreated leaves of chinese cabbage was 72%. Germination was significantly (Unpaired Student t-test, P < 0·05) reduced on leaves treated with 100 and 500 μmol l–1 gramicidin S (Table 5). Short, swollen germ tubes were observed with both these concentrations. Percentage germination on control leaves was lower than on the control glass slides, and the variation between replicate counts was greater on the leaf sections than on the glass slides. Results for the effects of B. brevis Nagano were similar to those for the pure antibiotic. Percentage germination was only significantly (Unpaired Student t-test, P < 0·05) reduced with the two highest spore concentrations, 4·0 and 20 × 106 spore ml–1. The ability of gramicidin S and B. brevis Nagano to inhibit Bot. cinerea germination was much lower on the leaf surface than in vitro. Treatment of leaves with the B. brevis E-1 culture had no effect on percentage germination of conidia of Bot. cinerea.

Table 5.   Germination of Botrytis cinerea conidia on leaf sections sprayed with a range of gramicidin S, Brevibacillus brevis Nagano and E-1 concentrations. Mean (S.D.) from triplicate leaf sections Thumbnail image of

Gramicidin S adsorption to the leaf surface.

From the original and final concentration of gramicidin S in solution, the concentration of gramicidin S bound to the leaf was determined by subtraction (Table 6). The 1 g leaf section adsorbed a significant amount of gramicidin S. The amount ranged from 57 to 31% as the gramicidin S concentration in the original solution increased from 100 to 500 μg. Ethanol extraction of gramicidin S bound to the leaf material was attempted. The maximum concentration of gramicidin S recoverable was 28·5–78·5 μg ml–1. When bioassayed, none of the extracts produced a zone of inhibition. The sensitivity limit of the bioassay was 5 μg ml–1. The percentage of adsorbed gramicidin S that was non-detectable was therefore 93–100%.

Table 6.   Concentration of gramicidin S bound to 1 g of Chinese cabbage leaf after one hour incubation Thumbnail image of

Gramicidin S activity in ethanolic leaf extract.

The ethanolic leaf extract was mixed with a range of gramicidin S solutions (0–1000 μg ml–1) so as to dilute the gramicidin S 10-fold. None of the resulting solutions produced zones of inhibition. The detection limit of the bioassay was 5 μg ml–1 so therefore, 95–100% of gramicidin S was undetected within the leaf extract.

DISCUSSION

Many strains of Bot. cinerea have a wide host range (Jarvis 1977) but some strains vary in their pathogenicity towards individual hosts (Kovacs and Tuske 1983). It was therefore necessary to obtain an isolate of Bot. cinerea that was the causative agent of grey mould of Chinese cabbage. It was also necessary to obtain a fresh isolate from the field, as pathogenic fungi tend to lose virulence during repeated subculturing within the laboratory.

In vitro experiments showed that gramicidin S is sporicidal to conidia of Bot. cinerea and is less inhibitory towards mycelial growth. Grey mould is a polycyclic disease, and the rapid spread of the disease is via conidial dispersal and germination. The sensitivity of Bot. cinerea conidia to gramicidin S would indicate that B. brevis could effectively reduce disease development by reducing the number of viable conidia. The inhibition exhibited by B. brevis Nagano spores, and the total lack of inhibition exhibited by the antibiotic-negative strain, B. brevis E-1, indicated that the mode of antagonism exhibited was antibiosis due to the presence of gramicidin S. The amount of inhibition by B. brevis Nagano spores was directly correlated to the level of gramicidin S present, further supporting the conclusion that antibiosis was the mode of antagonism in vitro.

Usually, antagonistic micro-organisms are field isolates and as such, little is known about them. Antibiosis in these cases may be indicated by the isolation within the culture filtrate of the antibiotic involved (Gueldner et al. 1988; Walker et al. 1998). In the case of B. brevis Nagano, the culture filtrate would have a low activity as almost all of the gramicidin S is bound to spores (Edwards 1993). The availability of pure gramicidin S and an antibiotic-negative mutant simplified the determination of the mode of antagonism involved. Various workers have used antibiotic-negative mutants to determine whether antibiosis was the mode of antagonism occurring between a specific antagonist and pathogen in vitro (Thomashow and Weller 1988).

Inhibition of Bot. cinerea germination as determined by biomass within flasks after 5 days of incubation was greater than inhibition of germination as quantified by percentage germination after 16 h incubation. By measuring percentage germination of conidia in the presence of gramicidin S after 16 h of incubation within shaking flasks, on glass slides, and on transfer from flask to glass slide during the incubation period, it was shown that differences in gramicidin S sensitivity are the result of differences in incubation conditions and not due to differences in methods of quantifying inhibition. Mycelial growth of Bot. cinerea on solid media was only slightly inhibited (25–30%) by gramicidin S even at the highest concentration used (1 mmol l–1). This compares with inhibition of mycelial growth in liquid culture that was totally inhibited by 50 μmol l–1 gramicidin S.

Results in planta showed the same trend as those in vitro; both pure gramicidin S and B. brevis Nagano, but not B. brevis E-1, were active against Bot. cinerea germination. However, the levels of gramicidin S required for inhibition of conidial germination on leaf surfaces were much higher than on glass slides. On glass slides, Bot. cinerea germination was totally inhibited by 15 μmol l–1 gramicidin S whereas on the leaf surface, germination was only partially inhibited by 500 μmol l–1 gramicidin S. The pure culture of B. brevis Nagano containing 540 μmol l–1 gramicidin S gave similar results, while B. brevis E-1 failed to have any effect on germination of Bot. cinerea, again indicating that gramicidin S is the active antifungal metabolite under these conditions.

The percentage germination of conidia on untreated control leaf sections was lower than for glass slide germination controls. This was probably due to the lower nutritional status of the leaf surface compared with GPB, or to the presence of antifungal substances from the leaf itself (Blakeman 1991). The standard deviations of results from leaf section experiments were greater than those from glass slides. This is probably due to the greater number of variables in planta compared with in vitro. On the leaf surface, the nutritional status, the presence of plant antifungal substances and antagonism from natural phyllosphere micro-organisms would vary between leaves and between different sites on a leaf (Blakeman 1991). Immersion of Chinese cabbage leaf sections in solutions of gramicidin S showed that a high proportion of the antibiotic was either adsorbed to, or absorbed by the leaf surface or metabolized. Detectable levels of gramicidin S could not be recovered by ethanolic extraction of the leaf sections. The activity of gramicidin S within ethanolic extracts of Chinese cabbage leaves was examined. Gramicidin S (up to 100 μg ml–1) did not produce a zone of inhibition when bioassayed within this extract. This, and the poor inhibition of Bot. cinerea germination on the leaf surface, would suggest that some ethanol-soluble leaf component binds to gramicidin S and that the antibiotic is inactive in this form. Hence, much higher concentrations of gramicidin S are required for inhibition of Bot. cinerea in planta than in vitro.

This paper highlights a number of important examples of how in vitro screening for biocontrol agents can provide erroneous results, both positive and negative. Firstly, if B. brevis had been screened for antibiotic activity using culture filtrate it would have given poor results, as the majority of the gramicidin S is spore-bound. Secondly, gramicidin S is highly sporicidal in vitro, although how much so is dependent on cultural conditions. However, due to strong binding of the antibiotic to the leaf surface, it is much less active within the actual infection court of the pathogen. From these results it is concluded that B. brevis is likely to have only limited activity towards Bot. cinerea on Chinese cabbage unless other media can be developed which induce the formulation of higher levels of gramicidin S.

Acknowledgements

The authors would like to thank Thelma McKay for her technical assistance.

Ancillary