Journal of Applied Bacteriology 1995, 70, 97-108

Antibiotic production and biocontrol activity by Bacillus subtilis CL27 and Bacillus pumilus CL45

C. Leifertl, H. Li194, Siripun Chidbureel, S. Hampson", Suzanne Workman', D. Sigee", H.A.S. Epton" and Agnes Harbour3 'Department of Plant and Soil Science, University of Aberdeen, Aberdeen, 'Department of Cell and Structural Biology, University of Manchester, Manchester, and 3E.R. Johnsons of Whixley Ltd, Whixley, Yorkshire, UK

4918/05/94: received 16 May 1994, revised 10 August 1994 and accepted 11 August 1994

C. LEIFERT, H. L I , S. CHIDBUREE, S. HAMPSON, S. WORKMAN, D. SIGEE, H.A.S. EPTON AND

A , H A R B o u R . 1995. Bacillus subtilis CL27 and B . pumilus CL45 showed similar activity against Botrytis cinerea in in vttro plate assays. In a seedling bioassay, however, B. subtilis CL27 had activity similar to a commercial fungicide while B. pumilus CL45 failed completely to prevent seedling damping-off caused by Bot. cinerea.

Antibiotic production by the two Bacillus strains was found to depend on the growth substrate and highest antibiotic production was found on media based on homogenized cabbage tissue. Antibiotic activity was found to depend on the pH and nutrient concentration in the assay medium. Antifungal antibiotics produced by B . subtilis CL27 and B . pumilus CL45 in different fermentation media were separated by thin layer chromatography. As suspected from the activity spectrum, three antibiotics (one with activity against Alternaria brassicicola, one with activity against Botrytis cinerea and one with activity against both fungi) could be detected in the fermentation broth of CL27, but only one in the fermentation broth of CL45. The two antibiotics produced by strain CL27 with activity against A. brassicicola were identified as peptides since their bands on the T L C plates developed a green to blue/green colour after treatment with 4,4'-tetramethyldiamino-diphenylmethane (TDM) reagent. The third antibiotics produced by strain CL27 and antibiotic produced by CL45 had a similar Rf-value and appeared not to be peptides based on the reaction with TDM. However, they showed a slightly different activity spectrum when tested against a range of different fungi.

Antibiotic production was clearly indicated as the mode of action of in vivo biocontrol by strain CL27 against damping off caused by Bot. cinerea of Astilbe micro-plants, because a u.v.-induced antibiotic negative mutant strain CL27b showed no activity in seedling bioassays in vivo. Also the mutant strain CL27a which produced the two peptide antibiotics but had lost the ability to produce the non-peptide antibiotic, showed greatly reduced in vivo activity.

INTRODUCTION

Many Bacillus strains are known to suppress fungal growth in vitro by the production of one or more antifungal anti- biotics (Katz and Demain 1977). Some of these antibiotic- producing strains were also shown to suppress fungal plant diseases in vivo (Swinburne et al. 1975; Baker et al. 1983; Singh and Deveral 1984; Utkhede and Sholberg 1986; Pusey et 01. 1986,1988; Fravel 1988; Edwards and Seddon

Correspondence to: Dr C. Leifert, Department of Plant and Soil Science, University of Aberdeen, Cruickshank Building, Aberdeen A B 9 ZUD, U K .

Permanent address: Department of Biology, Yunnan Normal University, Kunmmg, Yunnan 650092, PR China

1992). The antibiotics produced in vitro were generally assumed to be the compounds responsible for biocontrol in vivo. Bacillus spp., however, produce a range of other metabolites including biosurfactants (Edwards and Seddon 1992), chitinase and other fungal cell wall-degrading enzymes (Priest 1977; Pelletier and Sygusch 1990; Frand- berg and Schnurer 1994a,b), volatiles (Fiddaman and Rossall 1993, 1994) and compounds which elicit plant resistance mechanisms (Schonbeck et al. 1980; Kehlenbeck et al. 1994). These metabolites have also been implicated in biocontrol. Antibiotic production on artificial media in vitro alone cannot therefore provide sufficient proof of the involvement of antibiotics in biocontrol activity in vivo.

98 C . LEIFERT ET A L .

Many of the antifungal antibiotics produced in vitro were identified as peptide antibiotics (Katz and Demain 1977). Most antibiotics, however, were identified from strains which were isolated in the search for new pharmaceuticals (Kleinauf and von Dohren 1982) and most of these strains are without or have not been tested for biocontrol activity. Few antibiotics produced by biocontrol agents have been identified, but from the limited information available it appears that the antibiotics are the same or similar to those investigated as novel pharmaceuticals (McKeen et al . 1986; Gueldner et al. 1988). Antibiotic production and activity by biocontrol strains was only demonstrated in vitio, however, and there is therefore no direct evidence that the same anti- biotics are produced and responsible for disease control when biocontrol agents are applied to plant tissues. The value of in vitro studies into the mode of action was also questioned because antibiotic activity produced by different strains in in vitro plate assays correlated very little with in vivo biocontrol activity (Fravel 1988; Leifert et al. 1992, 1993b).

Differences in biocontrol activity between strains were usually attributed to differences in growth and antibiotic production in plunta. There is no information, however, that confirms this view because of the difficulty in studying antibiotic production in planta. As a result it is not known: (1) whether different or additional antibiotics are produced in planta; and (2) whether the ability to control diseases is correlated to the production of antibiotics or other metabo- lites. There is also very little information about the effect of environmental factors such as nutrients and pH on bio- control activity and the production and activity of anti- fungal antibiotics.

This study had three aims:

T o establish the importance of antibiosis as a mode of action for biocontrol in antibiotic-producing Bacillus strains. This involved a comparison of in vivo bio- control activity of B . subtilis CL27 and B . pumilus CL45 (two strains with similar in vitro activity against Botrytis cinerea) and two mutant strains of CL27 with either no or reduced antibiotic production. Strains CL27 and CI,45 have previously been shown to have a different range of activity when tested against various plant pathogenic fungi (Danby et al. 1994). To separate and tentatively identify different antibiotics produced by the two Bacillus strains by thin layer chro- matography (TLC) and to determine the activity of individual antibiotics against a range of different fungi. T o test the effect of nutrient composition and pH on antibiotic production and activity. This involves a com- parison of antibiotic production by the two strains on defined, leaf extract and conventional nutrient media and the testing of activity of fermentation broth

extracts in agar media with different pH and nutrient concentration or composition.

MATERIALS AND METHODS

Plant materlal and biocontrol actlvlty trials

The plants used to determine biocontrol activity in vivo were Astilbe microplants produced by shoot tissue culture at Neo Plants Ltd, Freckleton, Lancashire (see Leifert et al. 1993a, for the tissue culture methods used). Micro- propagated plants provide a realistic assay system as they are extremely susceptible to attack by grey mould (Botrytis cinerea), especially during acclimatization under high humidity conditions. In commercial practice they have therefore to be treated with fungicides at regular intervals (4-7 d) to avoid total loss of crops (Leifert et a / . 1994).

Rooted shoots were removed from the tissue culture media and planted into 3 x 3 cm compost plugs in poly- styrene trays. Plants were sprayed at weekly intervals with fungal spores and antagonists (see below for the different antagonists and preparations used) or fungicide suspensions using a Humbrol atomizer (Humbrol Ltd, Hull, UK). Two fungicide treatments commonly used in commercial plant tissue culture nurseries were used. Treatment 1 involved a weekly rotation of Octave (prochloraz + manganese; Shering, Berlin, Germany) and Repulse (chlorothalonil ; ICI, Farnham, UK) and treatment 2 a rotation of Repulse and Bavistin (carbendazim ; BASF AG, Ludwigshafen, Germany). To avoid cross-contamination with antagonists and fungal spores, propagation chambers were created by attaching a metal frame to the tray and sealing trays indi- vidually in polythene bags. The humidity inside the propa- gation chambers was monitored with a Vaisalla 100 humidity probe and was found to be between 89 and 95% (Vaisalla OY, PL26, Helsinki, Finland). Ten weeks after planting Astilbe plants were assessed for the number of dead, partially necrotic and healthy leaves and for visible fungal growth on leaves.

Bacterial strains used and preparation of bacterial inoculum

The bacterial strains used were B . pumilus strain CL45 (NCIMB 40489) and B . subtzlis strain CL27 (NCIMB 40491) which had previously been isolated from Brassica leaves (Leifert et al . 1992). Two mutant strains of CL27 with altered antibiotic production but similar an vitro growth rates (designated CL27a and CL27b) were produc- ed by u.v.-irradiation of suspensions of vegetative Bacillus

ANTIBIOTICS OF BIOCONTROL BAClLLUS 99

cell suspensions. CL27a showed reduced antibiotic pro- duction (smaller inhibition zones in in vitro assays) and was later found to produce only two of the three antibiotics pro- duced by CL27. CL27b had lost the ability to produce all three antibiotics.

Cabbage broth 5 (CB5; 50 g cabbage leaf extract I - ' water) was prepared as described by Leifert et al. (1993b). Two hundred ml of CB5 was then dispensed into 500 ml Erlenmeyer flasks and inoculated with different Bacillus strains. Inocula for CB5 were prepared by suspending bac- terial growth from a 24 h NA culture in quarter strength Ringer's solution. The turbidity was then adjusted to an absorbance of 1.0 at 625 nm using an ULTRO SPEC 4051 spectrophotometer (LKB Biochrom Ltd, Cambridge, UK) and 0.1 ml used to inoculate each flask. Antagonists were then grown at 20°C on a Gio Gyrotory Shaker (New Brunswick Scientific, Edison, NJ, USA) for 4 d.

The fermentation broth was then either used directly or separated into (1) bacterial cells and (2) cell-free culture filtrates by centrifugation for 50 min at 3000 g (Mistral 3000i centrifuge, Fisons, UK) and subsequent filtration of the supernatant fluid through a 0-2 prn pore size cellulose nitrate filter (Sartorius, Gottingen, Germany). The pellet containing the bacterial cells was subsequently re- suspended in an equal amount of quarter strength Ringer's solution. Polystyrene trays (9 plants/tray) were then sprayed at weekly intervals with 20 ml of either (1) fermen- tation broth with bacterial cells, (2) cell-free culture fil- trates, or (3) aqueous suspensions of the bacterial cells.

Preparation of fungal inocuium

Fungal inocula were used in seedling bioassays (Astilbe plants were sprayed every 7 d with 2 ml of a fungal suspen- sion containing lo3 spores ml-') and for in vitro sensitivity tests. The Bot. cinerea strain used was isolated from infect- ed Astilbe plants at Neo Plants Ltd (Freckleton, Lancs, UK), the nursery supplying the micropropagated plantlets, while the strain of Alternaria brassicicola was isolated from rotting cold stored Dutch white cabbages. Fungal stock cultures were grown on cabbage agar 5 (CA5; 50 g of homogenized cabbage tissue, 10 g of agar, 11 of distilled water, prepared as described by Leifert et al. 1993b) at 4°C in the dark. Malt extract agar (MA) was used for spore production. Botrytis cinerea was grown at 20°C for 5 d in the dark followed by 3 d under u.v.-light (Phillips TL40W/ 08; maximum radiant power at 350 nm) to induce sporu- lation and Alternaria brassicicola was grown at 20°C for 14 d in the dark prior to use in experiments. Spores were har- vested by pouring 10 ml of quarter strength sterile Ringer's solution onto a fungal plate and suspending spores with a bacteriological loop. Spore suspensions were filtered

through a double layer of muslin and adjusted to the required concentration with a haemocytometer.

Growth, sporulation and antibiotic production by Bacllius spp. on different growth media

Cabbage broth (50 g 1-' [CBS] or 100 g 1-' [CBlO] of homogenized cabbage leaves), nutrient broth (NB ; Oxoid), nutrient broth plus 0.005 g 1-' MgS04 (NB + Mg), nutri- ent broth plus 1% glucose (NB + 1% glucose), a defined medium (DM; 2 g 1-' NH4C1, 6 g 1-' Na,HPO,, 3 g 1- ' KH,P04, 3 g 1-' NaCl, 0.05 g 1-' MgSO, . 7H20, 5 mg 1-' L-methionine, 10 mg I - ' glycine, 2.5 mg I - ' L-aspartic acid, 10 g I - ' agar) and D M plus 1% glucose (DM + G) were used as growth media. One hundred ml of medium was transferred to 250 ml Erlenmeyer conical flasks, autoclaved and then inoculated with 0.1 ml of a bac- terial suspension from a 24 h nutrient agar culture of the Bacillus strains which had been adjusted to a turbidity of 1.0 with a spectrophotometer. Amino acids were filter sterilized and added to the medium after autoclaving. Four ml of medium were sampled aseptically at various times after inoculation. The number of cfu was determined by plating of dilution steps (quarter strength Ringer's solution) of the sampled medium on nutrient agar. The number of Bacillus spores was determined by mixing 0.1 rnl of absol- ute alcohol with 0.1 ml of the culture medium in wells of microwell plates (0.3 ml wells; Nunc Ltd, Copenhagen, Denmark) for 20 min (to kill vegetative cells) followed by serial dilution and plating of dilution steps on nutrient agar. Crude antibiotic extracts from the various media were made by centrifuging the liquid for 30 min at 3000 g with a Mistral 3000i centrifuge (Fisons, UK). The supernatant fluid was filtered through a 0-2 pm cellulose nitrate filter (Sartorius, Gottingen, Germany) to obtain cell-free fil- trates. Thirty p1 of the crude extract or the different dilu- tions were placed in 4 mm diameter wells in CA5 seeded with Bot. cinerea or A. brassicicola (see below).

In vltro assay for sntlfungal sctivlty

Twenty ml of liquid CA5 or potato dextrose agar (PDA) was cooled to 40°C (Bot. cinerea spores are very sensitive to heating) and inoculated with 0.2 ml of a fungal suspension containing 4 x lo5 spores mi-' . Fungal seeded medium was poured into Micro-detection dishes (Proteus Molecular Design Ltd, Marplc, Chcshire, UK). Antibiotic activity assays were performed in Microdetection plates because they allow the detection of lower concentrations of anti- biotics than assays carried out with conventional Petri dishes (Danby et al. 1994). After the agar had set 4 mm wells were cut into the agar with a cork borer and filled

100 C . LEIFERT ET A L .

with 30 pl of the cell-free culture filtrates from batch cul- tures of Bacillus spp. or redissoved individual antibiotic preparations obtained from T1.C plates (see below). Inhibi- tion zone diameters around colonies or wells were measured 2 d after inoculation with Botrytis and 4 d after inoculation with Alternaria seeded media.

When the effect of pH on the activity of antibiotics in crude extracts was determined the pH of the assay medium (CA5) was adjusted after autoclaving to values of between 3 and 9 with HCI and NaOH.

Separation of antifungai antibiotics produced by Bacillus spp. by thin layer chromatography (TLC)

Fifty /A of a cell-free culture filtrate from a 3 d batch culture of the different media used (see Leifert et al. 1993b for details of the culture method) was spotted onto silica gel 60 T L C plates (Art. 5721, Merck). T L C plates were devel- oped with various solvents and solvent mixtures, but the best separation was achieved using the solvent system described by Swinburne et al. (1975), i.e. butan-l- 01 : acetic acid : H 2 0 ; 3 : 1 : 1.

After development T L C plates were dried overnight in a laminar flow cabinet and then at 70°C for 5 h in a drying oven (to remove acetic acid residues from the plates which prevented fungal growth on the plates during detection of antibiotics). T o detect antibiotic activity on T L C plates a 4 mm layer of molten fungal spore-seeded CB5 agar was poured on T L C plates after development. The fungal species used were either Bot. cinerea or A. brassicicola and the concentration of spores in the medium was 4 x lo5 ml 1 - ' medium. After the agar had set the plates were incu- bated at 20°C in the dark for 4 d. After 4 d A. brassicicola had produced sufficient mycelium and spores to allow visual detection of zones of inhibition.

Botrytts cinerea, on the other hand, did not produce suf- ficient mycelium to allow visual detection of zones of inhi- bition. Fungal growth was therefore visualized by spraying plates with aesculin solution (Lund and Lyon 1975) fol- lowed by further incubation at 20°C for 5 h. Areas on the TLC plates with viable mycelium dcvcloped a brown/grey/ black colour due to the hydrolysis of aesculin by the /3- glucosidase. Zones of inhibition remained white.

Preparations of individual antibiotics were made by col- lecting the appropriate areas on the silica-gels, extraction of antibiotics with sterile distilled water adjusted to pH 3 (HCI) and filtration of the resulting suspension through a 0.2 pm cellulose nitrate filter.

Detection of peptides on thin layer chromatography (TLC) plates

Peptides on the TLC plates were visualized by treatment with TDM using a modification of the Chlor-o-Tolidin

method of Rydon and Smith (von Arx et al. 1976). Since one antibiotic of strain CL27 remained negative by this method it was repeated after hydrolysation of the prep- aration with 6 mol I - ' HCI at llO"C, a procedure required to detect acidic cyclic peptide antibiotics produced by Bacillus (Wakayama et al. 1984).

RESULTS

In vItfo and In vivo activity of antibiotic-producing Bacillus strains against Botrytis cinerea

When the number of healthy, partially necrotic and dead leaves was assessed and when the amount of fungal growth present on the plants was scored after 10 weeks, only the treatment involving Bacillus subtilis CL27 showed control of Botrytis cinerea similar to that obtained by fungicide treatment 1 (Octave/Repulse) (Fig. 1). Strain CI,27a which had lost the ability to produce one of the antibiotics pro- duced by CL27 (see below) showed intermediate control. The second fungicide treatment (Repulse/Bavistin) also showed reduced protection. This was later found to be due to carbendazim resistance and increased chlorothalonil tol- erance in the Botrytis strains found in commercial micro- propagation nurseries (Workman and Leifert, unpublished). Strain CL27b which showed no antibiotic production in vitro (see below) and strain CL45 which produced inhibi- tion zones similar to those of CL27 could not prevent disease development.

Crude fermentation broth with Bacillus cells, washed Bacillus cell preparation and cell-free culture filtrates of the fermentation broth inhibited fungal spoilage to a similar level for strain CL27. For strain CL27a crude fermentation broth with Bacillus cells and cell-free culture filtrates gave better control than the washed Bacillus cell preparations (Fig. 1).

Growth, sporulation and antibiotic productlon under different conditions in vltro

The production of antibiotics in batch culture differed, depending on the bacterial strain used, incubation time and the fermentation medium (NB, NB + Mn, NR + 1%1 glucose, CB5, CB10, DM and DM + 1% glucose, PDB) used. Culture filtrates of CL27 and CL45 again had very high activity (inhibition zones of around 40 mm) against Bot. cinerea when cabbage extracts (CB5, CB10) were used as fermentation media, but much lower activity (inhibition zones < 15 mm) when NB or DM were used as media (Figs 2 and 3). In CB5 and CB10, both CL27 and CL45 developed anti-Botrytis activity after 2 d, before the first

ANTIBIOTICS OF BIOCONTROL EAClLLUS 101

T "I s 40

c.

0 5 '5 20

4 0

v) a > m

CL27 CL27a CL27b CL45 Control

Fig. 1 Percentage of (a) dead leaves, (b) green leaves without visible signs of necrosis and (c) score of the amounts of Botrytir growth on plants 3 months after weaning of plants into soil. Botrytis growth scores: 6, all plants show visible mycelium growth; 5, 7 80% ; 4, > 60% ; 3, > 40% ; 2, > 20% ; 1, 1-20% of plants showing visible Botrytis growth. Treatments: C and S, fermentation medium with Bacillus cells (a); C, cell-free culture filtrates (a); S, washed Bacillus cells in quarter strength Ringer's solution (a); F1, fungicide rotation 1 (Octave/Repulse) (m); F2, fungicide rotation 2 (Repulse/Bavistin) (0); UT, untreated control plants (0). All treatments had been sprayed at weekly intervals with Botrytis cinerea spores. Means and standard deviations are from 54 plants per treatment

newly formed Bacillus spores could be detected. On NB and DM antibiotic activity also developed after 2 d but an increase in the number of spores could only be detected after 9 d (Figs 2 and 3). Addition of manganese to NB resulted in increased anti-Botrytis activity with CL45 and greatly increased sporulation with both Bacillus strains. No

sporulation could be detected (the number of spores remained below 10') when 1% glucose was added to NB and DM (Figs 2 and 3). No anti-Botrytis activity could be detected in PDB with both strains.

Isolate CL27 but not CL45 showed activity against A. brassicicola. Activity against Alternaria was highest in cell- free culture filtrates from NB + Mg and CBlO fermenta- tions of strain CL27 (20 mm inhibition zones) and similar in all other media (inhibition zones of between 15 and 10 mm) (Figs 2 and 3). Similar to the activity against Botrytis, anti-Alternarza activity preceded spore formation by 24 h in CB5 and CBlO and 7 d in NB and DM, and addition of 1% glucose did not stop the development of antifungal activity (Figs 2 and 3). Addition of manganese to NB resulted in increased anti-Alternaria activity and sporu- lation of strain CL27 (Fig. 2). No anti-Alternaria activity could be detected in PDB with both strains.

Growth rates of the two U.V. mutants (CL27a and CL27b) were similar to CL27 (Fig. 4). However, only CL27a (which did not produce one of the three antibiotics produced by CL27, see below) retained the ability to produce spores while CL27b (which did not produce all three antibiotics, see below) failed to sporulate on all media tested (Fig. 4). Activity against Botrytis but not Alternaria was reduced with CL27 growing in cabbage extract medium, while CL27b showed no antibiotic activity in the in vitro assays used.

Analysis of antibiotic production by TLC analysis

When TLC plates were developed with samples of cell-free culture filtrates of Bacillus strains CL27 or CL45 grown in CB5 medium and over-layered with Bot. cinerea seeded CA5, large inhibition zones (30-40 mm) occurred at an Rf- value of between 0.38 and 0.40 (Table 1). CB5 filtrates from the culture of strain CL27 also resulted in a small inhibition zone (3 mm) at an Rf-value of 0.56. However, this additional band was not present when cell-free filtrates from cultures of CL45 in CB5 were examined. When T L C was used to investigate the products of CL27 grown in NB and NB + Mn, only the small band of inhibition against Botrytis could be detected at an Rf-value of 0.56. Again there were no zones of inhibition when NB and NB + Mn medium inoculated with CL45 were examined (Table 1).

When TLC plates were overlaid with A. brassicicola seeded CA5 medium, two zones of inhibition (Rf-values 0.56 and 0.61) could be detected when NB, NB + Mn and CB5 inoculated with strain CL27 were examined (Table 1). No inhibition of A. brassicicola was found with NB, NB + Mn and CB5 which had been inoculated with CL45.

There was no detectable activity against Alternaria and Botrytis in PDB culture filtrates of both bacterial strains, CL27 and CL45.

102 c. LEIFERT E r A L .

Activitv aaainst:

9 r

y 5L

Total Spores

6!, j I , I , ,:, ;;;; 0 2 4 6 8 1 0 1 2 1 4

9 r

t I 7 Total Spores 4 0 0,NB

A A , N B t G 0 , N B t M n

0 2 4 6 8 1 0 1 2 1 4

9 r Total SDores O,DM -. \ H 0 , D M t G

- 6f

5= L F tr

0 2 4 6 8 1 0 1 2 1 4 l ime (d)

Bot. nit. - 50 r 0 0,CBS . O.CB10 t

0 2 4 6 8 1 0 1 2 1 4

E

2

- 30

Activity against: Bot. Alt.

0,NB . 0 , N B t M n A A , N B t G

_ - 0 2 4 6 8 1 0 1 2 1 4

Activity against: Bot. Alt.

0,DM - 40 m ~ . D M + G E

.P 20 .- 4- = I -

Bands at Rf-values of 0.56 and 0.61 showed a green/ blue/black colour after treatment with T D M reagent, sug- gesting that these two compounds are peptides. Since there was no colour development at an Rf-value of 3.9 even after hydrolysation with HCI, the anti-Botrytis antibiotic is unlikely to be a peptide.

Strain CL27a produced the two peptide antibiotics in similar quantities to the wild-type but failed to show activ- ity a t an Rf-value of 3.9 (Table 1). None of the antibiotics was detected on T L C plates inoculated with cell-free extracts from all media tested with strain CL27b.

Fig. 2 Growth and sporulation of Bacillus subtilis isolate CL27 and level of antagonistic activity against Ahernaria brassicicola and Botrytis cinerea in different liquid media (spore numbers on NB + G and DM + G) remained below lo5 cfu ml-'). Antifungal activity was assayed in Sens-acute plates containing fungal seeded CA5. Means are of three determinations

Effect of pH and nutrients on antibiotic activity

The activity of antifungal antibiotics formed in liquid batch culture of B. subtilis (measured as the inhibition zone diam- eter on fungal seeded CA5 medium) was affected by the pH (Figs 5 and 6) and the concentration and type of nutrients in the assay medium (Table 2).

Cell-free extracts from fermentations of CL27 and of CL45 in CBlO and of CL45 in NB + Mn showed the highest activity (largest inhibition zones) on Botrytis seeded CA5 between pH 5.6 and 6.0. Activity decreased slightly

A N T I B I O T I C S OF B I O C O N T R O L BACILLUS 103

50

- 40 E E -

30- 8 C .o 20 &?

= 10- 4- .- r

Total Spores O,CBB ' 0,CBlO .

- -

-

0 2 4 6 8 1 0 1 2 1 4

en=n=B=a

Total Spores z 7 a O,NB 0 . o,NB+Mn 0) 0

0 2 4 6 8 1 0 1 2 1 4 Time (d)

Fig. 3 Growth and sporulation of Bactllus pumilus isolate CL4S and level of antagonistic activity against Alternaria brasstcrcola and Botrytis cinerea in different liquid media (there was no activity against Botrytts and Alternaria in NB cell-free culture filtrates; symbols overlap). Antifungal activity was assayed in Sens-acute plates containing fungal seeded CAS. Means are of three determinations

Total Spores a o.CL27a m 0,CL27 6

9

8 r

1-

E 'c : 7 0 Fig. 4 Growth and sporulation of Bacillus

subtilis isolate CL27 and CL27a and level of antagonistic activity against Alternaria brassicicola and Botrytis cinerea in different fermentation media. Antifungal activity

rn 0 A

6

was assayed in microdetection plates 6

i

0 2 4 6 8 1 0 1 2 1 4 5 " ~ ' " ' " ' ' 1 ' 1

Total Spores 1 : :;;:a

- containing fungal seeded CA5. Means are 0 2 4 6 8 1 0 1 2 1 4 of three determinations Time (d)

i Activity against: Bot. Alt. a 0,CBB rn 0,CBlo

l-!- 0 2 4 6 8 1 0 1 2 1 4

- 40 E E -

Activity against: Bot. Ah. a O,NE . O,NB+Mn

Activity against: Bot. Alt. a o.CL27a . O.CL27

- 40 E E -

Activity against: Bot. Alt.

o.CL27a O,CL27

104 c. LEIFERT E r A L .

Rf-value ( 5 0.05)t Fungus used to Bacterial Growth visualize activity strain medium 0.39 0.56 0.61

Botrytis cinerea CL27 CL27 CL27 CL27a CL27a CL45 CL45 CL45

CB5 35 NB - N B + M n - CB5 NB + Mn - CBS 42 NB -

N B + M n -

-

7 7 Alternariu brassicicola CL27 CB5 - CL27 NB 4 4

10 9 CL27 NB + Mn - CL27a CB5 6 7

9 9 CL27a NB + Mn - CL45 CB5 CL45 NB CL45 NB + Mn

-

-

- - - - - -

- - - Table 1 Rf-values and size of inhibition zone diameter (mm) formed on T L C plates following development of cell-free culture filtrates of Bacillus subtilis strain CL27 and Bacillus pumilus strain CL45*

* 50 pl of a 3-d bacterial fermentation broth was developed and activity visualized by overlaying TLC plates with Botrytis cinerea and Alternaria brassicicola seeded CA5 medium. Means are from three determinations.

Solvent system used: Butan-1-01 : acetic acid : H,O; 3 : 1 : 1.

CBlO 50 (a)

40

30

20

10

0 3 4 5 6 7 8 9

501(c’ 40

I

20 ’i NB + Mn

@,CF A,Rf = 0.56 0, Rf = 0.61

CBlO 50 (b)

30 401 @,CF 0, Rf = 0.39 A,Rf = 0.56 U,Rf ~ 0 . 6 1

L

l0l nm-L -~ 3 4 5 6 7 8 9

NB + Mn @,CF A,Rf = 0.56 0 , R f ~ 0 . 6 1

30

,@.@--.-@ 20

3 4 5 6 7 8 9 3 4 5 6 7 8 9 PH PH

Fig. 5 Effect of assay medium pH on the activity of cell-free culture filtrates (CF; solid symbols) and individual antibiotics purified by TLC (open symbols) produced by Bacillus subtilis strain CL27 after 2 d of growth in different liquid media. Activity was assayed in fungal seeded CA5. (a, c) Activity against Botrytis cinerea; (b, d) activity against Alternaria brassicicola. (a, b) Antibiotics produced in CB10; (c, d) antibiotics produced in NB + Mn

ANTIBIOTICS OF BIOCONTROL BAClLLUS 105

30 401

CBlO 0, CF 0, Rf = 0.39

Individual antibiotics purified by T L C required very specific pH-ranges for activity. The major anti-Botrytis antibiotics from strains CL27 and CL45 (Rf-value 3.9) showed the highest activity at pH values below 6 while the two peptide antibiotics with activity against Alternaria were only active at pH values of 5.6 and above (Figs 5 and 6).

With increasing concentration of nutrients, in the form of cabbage tissue or potato extract and dextrose in the assay medium, the activity of antibiotics also decreased (Table 2). This effect was more pronounced in potato dextrose agar as the assay medium, where on full strength PDA only very small inhibition zones were formed, while on quarter strength PDA, inhibition zones were 3.5 times larger (Table 2).

3 4 5 6 7 8 9 PH

0

Flg. 6 Effect of assay medium pH on the activity of cell-free culture filtrates (CF, solid symbols) and individual antibiotics purified by TLC (open symbols) produced by Bacillus pumilus strain CL45 after 2 d of growth in different fermentation media. Activity was assayed in Borrytis cinerea seeded CA5. (a) Antibiotics produced in CBlO; (b) antibiotics produced in NB + Mn

when the pH was reduced to 3.1 and to very low values when the pH was increased to values above 6.0 (Fig. 5). In NB + Mn extracts of fermentations with strain CL27 activity could only be detected at pH 5.6 and above and increased slightly with increasing pH (Fig. S), while extracts from NB + Mn inoculated with CL45 were only active at pH 5.6 and below (Fig. 6).

Activity range of crude fermentation broth extracts from Bacillus spp.

When the activity of individual antibiotics purified by T L C against different fungal genera was determined, different activity patterns were found for the three antibiotics pro- duced by strain CL27 and strain CL45 (Table 3). Activity patterns for the two peptide antibiotics produced by CL27 were very similar for all fungi except Bot. cinerea which was only inhibited by the antibiotic with an Rf-value of 0.56. The two antibiotics with an Rf-value of 0.39 produced by strains CL27 and CL45 also had very similar activity pat- terns but differed when tested against a Zygorrhynchus strain.

DISCUSSION

Both cell-free fermentation broth filtrates and washed Bacillus cells of Bacillus subtilis strain CL27 prevented grey mould disease development on Astilbe. This and the finding that an antibiotic-negative mutant completely lost biocontrol activity and a partially deficient mutant showed reduced activity strongly suggests that antibiosis is the mode of action of biocontrol by Bacillus subtilis strain CL27. Additionally, the concentrations of antibiotics in the culture medium were sufficient to control disease even in the absence of Bacillus cells.

The finding that Bacillus pumilus strain CL45 had no in vivo activity, although it produced similar anti-Botrytis antibiotic activity to strain CL27 in liquid media in vitro, is surprising and it is possible only to speculate about reasons. On the basis of its activity range, the antibiotic produced by strain CL45 is different from that of CL27 and its lon- gevity on the leaf surface might be different, possibly because of inactivation as described for gramicidin S, an antibiotic produced by B. brevis (Edwards and Seddon 1992). The set of antibiotics produced by CL27 provides activity over a wide range of pH-values, while the antibiotic

106 C. LEIFERT ET A L

Table 2 Effect of nutrient concentration in the assay medium on anti-Burrpis activity in cell-free extracts of CBlO

Inhibition zone diameter (mm) Nutrient

Assay medium used concentration (g I-') Mean S.D. fermentation medium*

Cabbage extract agar 200t 35 2 loot 38 2 sot 41 3

Potato Dextrose agar 40 8 1 20 13 2 10 26 2

* Bacillus subtilis CL27 was grown for 48 h on the fermentation medium. Means and standard deviations (s.D.) are from five determinations. t Fresh cabbage leaves were used.

produced by CL45 provides activity only below pH 6 and this could also be a reason for the lack of biocontrol activity by CL45. However, proof for such mechanisms would require the development of methods which allow the pro- duction and activity of Bacillus antibiotics on leaf surfaces to be monitored.

Bacteria in the genus Bacillus are known to produce anti- fungal peptide antibiotics (Katz and Demain 1977) and peptide antibiotics were shown to be produced in vitro by B. subtilis and B. brevis strains which had in vivo activity against fungal plant diseases (McKeen et al. 1986; Gueld- ner el al. 1988; Edwards and Seddon 1991). Two of the antibiotics produced by strain CL27 were also shown to be peptides. On the basis of the tests used, the antibiotics with an Rf-value of 0.39 produced by both CL27 and CL45 appeared not to be peptides. The similar Rf-value and size of inhibition zone on the T L C plate suggest that the major anti-Botrytis antibiotics formed by both Bacillus strains CL27 and CL45 are similar compounds. However, their different activity pattern showed that they are not identical.

Mutants of strain CL27 which had lost the ability to produce the non-peptide antibiotic but retained the ability to produce the peptides were frequently found (seven iso-

lates; C. Leifert, unpublished results). This indicated that the pathways of antibiotic synthesis were quite different. CL27 mutants which only croduced one of the two peptide antibiotics could not be obtained, however, suggesting that they may be produced on the same biochemical pathway.

The production of antibiotics was dependent on the composition of the growth medium. The use of cabbage leaf extract agar as a substrate resulted in the highest anti- biotic production. I t was also the only medium on which the major anti-Botrytis antibiotic of CL27 (Rf-value of 0.39) was produced. The ability to colonize and utilize the nutrients in wounds and early leaf infection sites of Bot. cinerea might therefore be fundamental to biocontrol with this organism which was originally isolated from cabbage leaf surfaces (Leifert et al. 1992).

The failure of Bacillus strains to produce the main anti- Botrytis antibiotic (Rf-value 0.39) on nutrient broth and the defined medium and the absence of antibiotic production on potato dextrose broth suggests that the antibiotics found (especially the non-peptide) may be novel, since PDA has been used by most other authors to select Bacillus spp. which produce anti-fungal antibiotics (Katz and Demain 1977; Pusey and Wilson 1984; Singh and Deveral 1984;

Table 3 Size of inhibition zone diameter (mm) formed when individual antibiotics purified by TLC were assayed against

Fungal species strain 0.39 0.56 061 different fungal genera in microdetection - plates containing fungal seeded CA5

medium* Botrytrs cinerea CL27 31

Alternurtu brasstcrcula CL27 12 12

Aspergillus P 1 t CL27 6 6

Antibiotic tested (Rf-value ( f 0.05)*) Bacterial

9 CL45 32 -

CL45

CL45

CL45 13

CL45 24

-

-

- - -

-

- - -

zygorrhynchus P23t CL27 7 8

Phoma P13t CL27 21

-

- - - - - -

* Means are from three determinations. t See Danby ei al. (1994) for a detailed description of the strains.

ANTIBIOTICS OF BIOCONTROL BACILLUS 107 ~

McKeen et a / . 1986; Utkhede and Sholberg 1986; Wilson and Wisniewski 1989). This can only be established, however, by identification of the compounds in future investigations.

Compared with cabbage agar (where antibiotic activity was similar at different concentrations of cabbage extract), activity decreased rapidly when the concentration of potato extract and dextrose in PDA was increased. This clearly indicated that compounds which inhibit antibiotic activity are present in this medium. Another factor affecting the activity of all antibiotics identified in this study was the pH of the medium. The pH and inhibitory compounds present in certain plant tissues are therefore likely to affect anti- biotic production and biocontrol activity in vdvo.

The production of the two peptide antibiotics appeared to be closely connected to sporulation in nitro since the U.V.

mutant strain CL27b and four other antibiotic-deficient strains isolated (Leifert, unpublished) had also lost the ability to sporulate. Further proof for such a close link comes from the finding that the addition of manganese to nutrient broth increased both the production of the peptide antibiotics and sporulation. Antibiotic production has fre- quently been described as a function or early event of spore formation since cultural conditions which suppressed sporulation often also reduced antibiotic production. Also sporulation-negative mutants often failed to produce peptide antibiotics (Bernlohr and Novelli 1959; Schmitt and Freese 1968; Ray and Bose 1971; Haavick and Thomassen 1973; Lee et af. 1975). This does not appear to be true in general since there are a few reports of antibiotic- deficient mutants which continued to sporulate (Ray and Bose 1971 ; Haavick and Thomassen 1973). The main anti- Botrytis antibiotics of strain CL4S and CL27 (Rf-value of 0.39) also appeared to be synthesized independently of sporulation, since the second mutant, CL27a, had lost the ability to produce the antibiotic, but sporulated at a rate similar to the wild type.

I t should be pointed out, however, that in this study sporulation was monitored only by assessing the number of cfu tolerant to ethanol. Resistance to ethanol occurs only after the sporulation process is complete. Early events of sporulation might well have taken place in treatments where no alcohol-resistant cfu developed. This is clearly indicated by the finding that addition of glucose inhibited the development of ethanol-resistant spores but not the production of antibiotic activity on D M and NB.

However, since antibiotic production appears to depend at least in part on sporulation by strain CL27, methods which promote sporulation on leaf surfaces are likely to increase the activity of the organism in vivo. Unless a con- tinuous cycle between vegetative growth, sporulation and spore formation can be encouraged on a leaf surface, treat- ment with strain CL27 is likely to remain similar to a

chemical treatment where the spray has to be reapplied at regular intervals to ensure protection. Unlike some other Bacillus biocontrol agents (Edwards and Seddon 1991) there are no selective enumeration methods for B. subtilis strain CL27. It is currently not possible to monitor growth, sporulation and spore germination of strain CL27 on plant surfaces. Once selective plating methods or molecular alter- natives for the monitoring of Bacillus population dynamics are developed it should be possible to determine in detail the nutritional and environmental requirements for differ- ent steps in the life cycle of Bacillus spp. and to improve the biocontrol activity in vivo.

ACKNOWLEDGEMENTS

This project was financially supported by the MAFF Open Contract Scheme (CSA 1710) and SANDOZ UK Ltd.

REFERENCES

Baker, C.J., Stavely, J.R. and Mock, N. (1983) Biocontrol of bean rust by Bacillus subtilis under field conditions. Plant Disease 69,

Bernlohr, R.W. and Novelli, G.D. (1959) Antibiotic production as a function of spore formation in Bacillus lichenifrmis. Nature

Danby, S., Hampson, S.B., Joshi, S., Sigee, D.C., Epton H.A.S. and Leifert, C. (1994) Activity of antibiotics produced by Bacil- lus subtilis and Bacillus pumilus against common fungal contami- nants of plant tissue cultures. In Physiology, Growth and Development of Plants in Culture ed. Lumsden, P.G., Nicholas, J.R. and Davis, W.J. pp. 404-408. Dortrecht: Kluwer Aca- demic Publishers.

Edwards, S.G. and Seddon, B. (1991) Development of a new selective medium for the retrieval of Bacillus brevir from the environment. Journal of Applied Bacteriology 71, xxx.

Edwards, S.G. and Seddon, B. (1992) Bacillus brevis as a bio- control agent against Botrytis cinerea on protected Chinese cabbage. In Recent Advances in Botrytis Research ed. Verhoeff, K., Malathrakis, N.E. and Williamson, B. pp. 267-271. Wagen- ingen : Pudoc Scientific Publishers.

Fiddaman, P.J. and Rossall, S. (1993) The production of anti- fungal volatiles by Bacillus subtilis. Journal of Applied Bacte-

Fiddaman, P.J. and Rossall, S. (1994) Effects of substrate on the production of antifungal volatiles from Bacillus subtilis. Journal of Applied Bacteriology 76, 395-405.

Frandberg, E. and Schniirer, J. (1994a) Evaluation of a chromo- genic chitooligosaccharide p-nitrophenyl N,W-diacetyl chito- biose, for the measurement of chitinolytic activity of bacteria. Journal of Applied Bacteriology 76, 259-263.

Frandberg, E. and Schniirer, J. (1994b) Chitinolytic properties of Bacillus pabuli K1. Journal of Applied Bacteriology 76, 361-367.

Fravel, D.R. (1988) Role of antibiosis in the biocontrol of plant diseases. Annual Review in Phytopathology 26,75-91.

770-772.

184, 1256-1257.

~i010gy 74, 119-126.

108 C . LEIFERT ET A L .

Gueldner, R.C., Reilly, C.C., Pusey, P.L., Costello, C.E. and Arrendale, R.F. (1988) Isolation and identification of iturins as antifungal peptides in biological control of peach brown rot with Bacillus subtilis. Journal of Agricultural and Food Chemistry

Haavic, H.I. and Thomassen, S. (1973) A Bacitracin-negative mutant of Bacillus lichenformis which is able to sporulate. Journal of General Microbiology 76, 45 1-454.

Katz, E. and Demain, A.L. (1977) The peptide antibiotics of Bacillus: chemistry, biogenesis, and possible functions. Bacte- riological Reviews 41, 449-474.

Kehlenbeck, H., Krone, C., Oerke, E.-C. and Schonbeck, F. (1994) The effectiveness of induced resistance on yield of mil- dewed barley. Journal of Plant Diseases and Protection 101, 11-21.

Kleinauf, H. and von Dohren H. (1982) Peptide Antibiotics Bio- synthesis and Function, pp. 78-81. Berlin: Walter de Gruyter.

Lee, S.G., Littau, V. and Lipmann, F. (1975) The relation between sporulation and the induction of antibiotic synthesis and of amino acid uptake in Bacillus brevis. Journal of Cell Biology 66, 233-242.

Leifert, C., Sigee, D., Epton, H.A.S., Stanley, R. and Knight, C. (1992) Isolation of bacteria antagonistic to postharvest fungal diseases of cold stored Brassica spp. Phytoparasitica 20, 143s- 148s.

Leifert, C., Clark, E. and Rothery, C.A. (1993a) Micro- propagation; the propagation of plants by tissue culture. Bio- logical Sciences Review 5 , 31-35.

I,eifert, C., Sigee, D.C., Stanley, R., Knight, C. and Epton, H.A.S. (1993b) Biocontrol of Botrytis cinerea and Alternaria brassicicula on Dutch white cabbage by bacterial antagonists at cold-store temperatures. Plant Pathology 42, 270-279.

Leifert, C., Morris, C. and Waites, W.M. (1994) Ecology of microbial saprophytes and pathogens in field grown and tissue cultured plants. C R C Critical Reviews in Plant Science 13, 139- 183.

Lund, R.M. and Lyon, G.D. (1975) Detection of inhibition of Erminia carotovora and Erwinia herbicola on thin layer chro- matograms. Journal of Chromatography 110, 193-196.

McKeen, C.D., Reilley, C.C. and Pusey, P.L. (1986) Production and partial characterisation of antifungal substances antago- nistic to Monilinia fructicola from Bacillus subtilis. Phytopathol-

Pelletier, A. and Sygusch, J. (1990) Purification and character- isation of three chitosanase activities from Bacillus megaterium

36,366-370.

ou 76, 136-139.

P1. Applied and Environmental Microbiology 56, 844-848. Priest, F.G. (1977) Extracellular enzyme synthesis in the genus

Bacillus. Bacteriological Reviews 41, 71 1-753. Pusey, P.L. and Wilson, C.L. (1984) Postharvest biological

control of stone fruit brown rot by Bacillus subtilis. Plant Disease 68, 753-756.

Pusey, P.L., Wilson, C.L., Hotchkiss, M.W. and Franklin, J.D. (1 986) Compatibility of Bacillus subtilis for postharvest control of peach brown rot with commercial fruit waxes, dicloran and cold-storage conditions. Plant Disease 70, 587-590.

Pusey, P.L., Hotchkiss, M.W ., Dulmage, H.T., Baumgardner, R.A. and Zehr, E.I. (1988) Pilot tests for commercial pro- duction and application of Bacillus subtilis (B-3) for postharvest control of peach brown rot. Plant Disease 72,622-626.

Ray, B. and Bose, S.K. (1971) Polypeptide antibiotic-negative spore former mutants of Bacillus subtilis. Journal of General Microbiology 17, 491-498.

Schmitt, R. and Freese, E. (1968) Curing of sporulation mutant and antibiotic activity of Bacillus subtilis. Journal of Bacteriology

Schonbeck, F., Dehne, H.-W. and Beicht, W. (1980) Untersuchungen zur Aktivierung unspezifischer Resistenz- mechanismen in Pflanzen. Zeitschrift fur Pjanzenkrankheiten und Pflanzenschutz 87,654-666.

Singh, V. and Deveral, B.J. (1984) Bacillus subtilis as a control agent against fungal pathogens of citrus fruit. Transactions of the British Mycological Society 83, 487-490.

Swinburne, T.R., Barr, J.G. and Brown, A.E. (1975) Production of antibiotics by Bacillus subtilis and their effect on fungal colo- nists of apple leaf scars. Transactions of the British Mycological Society 65, 21 1-217.

Utkhede, R.S. and Sholberg, P.L. (1986) In vitro inhibition of plant pathogens : Bacillus subtilis and Enterobacter aerogenes in vivo control of two postharvest cherry diseases. Canadian Journal of Microbiology 32,963-967.

von Arx, Faupel, M. and Brugger, M. (1976) Eine Modification der Chlor-o-Tolidin Farbereaction fur die Dunnschichtchroma- tographie. Journal of Chromatography 120,224-228.

Wakayama, S., Ishikawa, F. and Oishi, K. (1984) Mycocrein, a novel antifungal peptide antibiotic produced by Bacillus cereus. Antimicrobial Agents and Chemotherapy 26,939-940.

Wilson, C.L. and Wisniewski, M.E. (1989) Biological control of postharvest diseases of fruits and vegetables: an emerging tech- nology. Annual Review of Phytopathology 27, 425-441.

96, 1255-1265.