Fusaric Acid-Producing Strains of Fusarium oxysporum Alter 2,4-Diacetylphloroglucinol Biosynthetic Gene Expression in Pseudomonas fluorescens CHA0 In Vitro and in the Rhizosphere of Wheat

Abstract

The phytotoxic pathogenicity factor fusaric acid (FA) represses the production of 2,4-diacetylphloroglucinol (DAPG), a key factor in the antimicrobial activity of the biocontrol strain Pseudomonas fluorescens CHA0. FA production by 12 Fusarium oxysporum strains varied substantially. We measured the effect of FA production on expression of the phlACBDE biosynthetic operon of strain CHA0 in culture media and in the wheat rhizosphere by using a translational phlA′-′lacZ fusion. Only FA-producing F. oxysporum strains could suppress DAPG production in strain CHA0, and the FA concentration was strongly correlated with the degree of phlA repression. The repressing effect of FA on phlA′-′lacZ expression was abolished in a mutant that lacked the DAPG pathway-specific repressor PhlF. One FA-producing strain (798) and one nonproducing strain (242) of F. oxysporum were tested for their influence on phlA expression in CHA0 in the rhizosphere of wheat in a gnotobiotic system containing a sand and clay mineral-based artificial soil. F. oxysporum strain 798 (FA⁺) repressed phlA expression in CHA0 significantly, whereas strain 242 (FA⁻) did not. In the phlF mutant CHA638, phlA expression was not altered by the presence of either F. oxysporum strain 242 or 798. phlA expression levels were seven to eight times higher in strain CHA638 than in the wild-type CHA0, indicating that PhlF limits phlA expression in the wheat rhizosphere.

Antibiosis is an important mechanism used by plant-beneficial microorganisms to overcome the effects of soil-borne fungal pathogens (44). The polyketide metabolite 2,4-diacetylphloroglucinol (DAPG) is one of the most effective antimicrobial metabolites produced by strains of fluorescent pseudomonads (25, 40) and is effective against bacteria, fungi, and helminths (13, 18, 23, 30, 34, 39). In Pseudomonas fluorescens Q2-87 and CHA0, the four DAPG biosynthesis genes, phlACBD, are organized as an operon and are indispensable for the production of DAPG as well as monoacetylphloroglucinol (MAPG), which is either a precursor to or a degradation product of DAPG. This operon is followed by a gene coding for a putative efflux protein (phlE) and flanked by the divergently transcribed phlF gene, which encodes a repressor protein of DAPG synthesis (5, 11, 38).

Environmental factors influence the production of antimicrobial compounds such as DAPG in fluorescent pseudomonads. Variation in the biocontrol performance of these bacteria has been attributed to changes of biotic and abiotic factors associated with field location and cropping time (14, 44). Complex biotic factors, such as plant species, plant age, host cultivar, and infection with the plant pathogen Pythium ultimum, can significantly alter phlA expression (33). DAPG production can be influenced by the carbon sources and minerals present in the bacterial environment. Fe³⁺ and sucrose increase production of DAPG and MAPG in P. fluorescens F113 (16), but in P. fluorescens Pf-5 and CHA0, production is stimulated by glucose (15, 34). In strain S272, the highest DAPG yield was obtained with ethanol as the sole carbon source (48). Furthermore, in P. fluorescens CHA0, the production of DAPG is stimulated by Zn²⁺, Cu²⁺, and NH₄Mo²⁺ (15). Microbial metabolites play an important role in the regulation of DAPG in that its synthesis is autoinduced and repressed by other bacterial extracellular metabolites of strain CHA0 (38).

The fungal toxin fusaric acid (FA) is also a potent inhibitor of DAPG synthesis in P. fluorescens CHA0 (14, 38). FA was first isolated by Yabuta et al. (47) from Fusarium heterosporum Nees as a compound that inhibited the growth of rice seedlings. FA is produced by many Fusarium spp. and is toxic to various plants, fungi, and bacteria (1, 8, 19, 20, 22, 31). Synergistic interactions of FA with other naturally co-occurring mycotoxins can cause animal toxicity (3, 12, 41, 46).

In the present study, we screened 12 Fusarium oxysporum strains for FA production. Our objective was to determine whether the ability of these fungi to produce FA in culture media and on wheat roots correlates with the suppression of DAPG production in CHA0. The phlF mutant CHA638, lacking the DAPG pathway-specific repressor PhlF, was used to investigate the mode of action of the microbial interaction.

MATERIALS AND METHODS

Microorganisms, plants, and culture conditions.

P. fluorescens CHA0 is a DAPG-producing strain isolated from the rhizosphere of tobacco grown in a Swiss soil naturally suppressive to Thielaviopsis basicola (43). The phlF mutant CHA638 was constructed by gene replacement (38). Plasmid pME6259 is a derivative of pME6010 and contains a phlA′-′lacZ translational fusion and a tetracycline resistance gene (38). pME6010 is present at approximately six copies per chromosome and is maintained stably in P. fluorescens in the absence of antibiotic selection (21). Earlier work has shown that retention of the reporter plasmid pME6259 by strain CHA0 is 98.8 ± 0.9% after 20 generations of growth in Luria-Bertani (LB) (36) broth (33). All strains were routinely cultivated at 27°C on King's B agar (KB) (26) and in LB with the addition of tetracycline (125 μg/ml) as required, unless otherwise mentioned.

F. oxysporum strains 233, 234, 235, 236, 237, 238, 240, 241, 242, and 410 were isolated from wheat roots (Triticum aestivum cv. Arina) by the Swiss Federal Research Station for Plant Production, Changins, Switzerland. Strains 798 and 801 (obtained from Centraal Bureau voor Schimmelculture, Baarn, The Netherlands) were isolated from soil cultivated with wheat in Grignon, France. All F. oxysporum strains were routinely cultured on 1.5% malt agar plates (Difco, Detroit, Mich.) at 24°C.

Seeds of winter wheat (T. aestivum cv. Arina) were surface disinfected for 15 min in 7% sodium hypochlorite (vol/vol), followed by 10 min in 10% H₂O₂ (vol/vol), and then thoroughly rinsed with sterile double-distilled water. Seeds were pregerminated for 3 days on 0.85% water agar at 24°C in darkness.

Production of FA by F. oxysporum strains.

Malt broth (2%), Czapek Dox medium (Oxoid, Hampshire, United Kingdom), and R2 broth (7) amended or not with 89 mg of ZnSO₄·7H₂O (Merck, Darmstadt, Germany) (250 ml in 500-ml baffled flasks) per liter were inoculated with one 7-mm agar plug taken from 5- to 7-day-old malt agar cultures. Cultures were incubated for 7 days at 24°C on a rotary shaker (180 rpm). Fungal biomass (mycelia and microconidia) was collected by centrifugation (2,200 × g) for 15 min. One part of the supernatant was filtered through a 0.8-μm membrane (Nalgene, Rochester, N.Y.) and stored at 2°C. In addition, 50 ml of the supernatant was acidified to pH 2 with approximately 400 μl of 2 M HCl, mixed with 50 ml of ethylacetate, and shaken vigorously for 1 min. The organic phase was separated from the aqueous phase by filtering through a silicone-coated filter paper (catalog no. 484015; Macherey and Nagel, Düren, Germany) and brought to dryness in vacuo. The residue was dissolved in 1 ml of methanol and analyzed by high-performance liquid chromatography (HPLC) using a Hewlett Packard 1090 liquid chromatograph (Hewlett Packard Co., Palo Alto, Calif.) equipped with a reverse-phase column (4 by 125 mm) packed with Nucleosil 120-5-C18 (Macherey-Nagel, Oensingen, Switzerland) and set at 50°C. The samples (10 μl) were eluted with a linear methanol gradient from 25 to 100% in 0.43% o-phosphoric acid over 12 min. FA was detected by monitoring A₂₇₀. The retention time was approximately 2 min, with a mobile-phase flow rate of 0.7 ml/min. The samples were quantified against a standard curve of synthetic FA (Acros Organics, Geel, Belgium). The fraction containing the FA standard from the HPLC was collected, and the presence of FA was qualitatively and quantitatively confirmed by mass spectrometry (MS) on a TSQ3000 triple quadrupole mass spectrometer with electrospray ionization (Thermo Finnigan, Bremen, Germany). The experiment was carried out twice with similar results. Data from all trials were analyzed for a trial-by-treatment interaction with analysis of variance using Systat version 9.0 (Systat Inc., Evanston, Ill.). Data could not be pooled, and the two individual trials are presented separately (see Table 1). Means of three replicate cultures were separated using Fisher's protected (P = 0.05) least significant difference test.

TABLE 1.

Production of the phytotoxin FA by different strains of F. oxysporum^a

F. oxysporum strain	FA production (nmol/ml of medium)
	Malt		Czapek Dox		R2 without zinc
	Expt 1	Expt 2	Expt 1	Expt 2	Expt 1	Expt 2
233	<0.2	<0.2	6.8 a	<0.2	5.7 a	104.6 a
234	<0.2	<0.2	0.8 a	<0.2	<0.2	<0.2
236	<0.2	<0.2	5.9 a	<0.2	<0.2	<0.2
237	<0.2	<0.2	<0.2	0.9 a	<0.2	<0.2
238	1.4 a	<0.2	<0.2	0.5 a	<0.2	<0.2
241	110.0 b	47.1 a	100.0 a	114.7 a	550.0 b	308.3 b
410	<0.2	<0.2	1.6 a	<0.2	<0.2	<0.2
798	<0.2	10.3 ab	2,633.7 b	3,492.4 b	0.9 a	1.4 a
801	<0.2	2.0 b	1,962.3 c	786.1 c	2.3 a	7.7 a

^aThe F. oxysporum strains were cultivated in 250 ml of liquid malt, Czapek Dox, or R2 without zinc for 7 days at 24°C and 180 rpm. The detection limit for production of FA was 0.2 nmol/ml. Values have been adjusted to reflect the 35% efficiency of extraction. Each value is the mean of three replicate cultures. Means for the same experiment within the same column followed by the same letter were not significantly different at P ≤ 0.05 according to Fisher's protected least significant difference test.

Influence of FA and fungal filtrates on phlA′-′lacZ expression in vitro.

P. fluorescens CHA0 and its mutant derivative CHA638, carrying a translational phlA′-′lacZ fusion on plasmid pME6259, were grown in 24 ml of a minimal glucose-ammonium medium (OSG) (38) amended with 6 ml of fungal filtrate (described above) or Czapek Dox medium without tetracycline. Exponential-growth-phase LB cultures of the bacterial strains were diluted to an optical density at 600 nm (OD₆₀₀) at 27°C of 0.01, and 30-μl aliquots were used for inoculation. Cultures were incubated with rotary shaking at 160 rpm. When appropriate, cultures were amended with FA. A fresh stock solution of synthetic FA was prepared immediately prior to use by dissolving it in 20% (vol/vol) ethanol (pH 6.5). Controls received the same amount of the solvent. In all experiments, β-galactosidase activities were determined throughout the exponential and stationary growth phases by the method of Miller (32). Bacterial growth was estimated from the OD at 600 nm after the correlation with CFU was established by dilution plating on KB agar. The experiment was repeated three times, and the results were similar. Values for an OD₆₀₀ of 1.5 and 2.5 of one representative experiment are presented (see Table 2). The means of three replicate cultures were separated using Fisher's protected (P = 0.05) least significant difference test.

TABLE 2.

Influence of F. oxysporum culture filtrates on phlA expression in P. fluorescens CHA0 and CHA638^a

P. fluorescens strain	F. oxysporum strain	FA concn in bacterial growth medium (μM)	β-Galactosidase activity (U per 108 CFU)
			OD600 = 1.5	OD600 = 2.5
CHA0	None	0	1,510 a	2,100 a
	240	0	1,520 a	2,000 a
	241	25	1,480 a	1,910 a
	242	0	1,460 a	1,950 a
	410	0	1,640 a	1,920 a
	798	700	220 b	590 b
	801	160	200 b	1,520 c
CHA638	None	0	1,850 a	1,720 a
	240	0	1,660 a	1,390 a
	241	25	1,590 a	1,590 a
	242	0	1,540 a	1,560 a
	410	0	1,740 a	1,680 a
	798	700	1,660 a	1,490 a
	801	160	1,680 a	1,470 a

^aP. fluorescens CHA0 (wild type) and CHA638 (phlF mutant) carrying a translational phlA′-′lacZ fusion on plasmid pME6259 were tested. The F. oxysporum strains were cultivated in Czapek Dox medium for 7 days at 24°C and 180 rpm. Filtrates from F. oxysporum cultures were added to bacterial growth medium OSG at a ratio of 20% (vol/vol). The FA content of the culture filtrates was determined, and the final FA concentration in the bacterial growth medium is indicated. β-Galactosidase activities were determined in bacterial cultures grown to an OD₆₀₀ of 1.5 ± 0.1 or 2.5 ± 0.1, as indicated. Data represent means of three replicate cultures. Means for the same P. fluorescens strain within a column followed by a different letter are significantly different at P ≤ 0.05 according to Fisher's protected least significant difference test.

Gene expression assay in the rhizosphere of wheat.

The influence of the two F. oxysporum strains 242 and 798 on the expression of phlA in P. fluorescens CHA0 and its phlF mutant CHA638 was monitored on wheat roots. Plants were grown in artificial soil containing vermiculite and quartz sand of different fractions (24). The soil was moistened with double-distilled water (10%, wt/wt). One-liter Erlenmeyer flasks were filled with 300 g of artificial soil, plugged with cotton wool stoppers, and autoclaved for 30 min at 121°C. After autoclaving, 30 ml of sterile modified Knop's nutrient solution (24) was added. Cultures of F. oxysporum strains 242 and 798 grown in Czapek Dox as described above were centrifuged (2,200 × g) for 15 min, and the resulting pellet was resuspended in sterile Knop's solution. Microconidia were separated from mycelia by filtering through sterile glass wool and quantified by using a hemocytometer. The artificial soil was inoculated with 10⁴ conidia per g of soil.

Overnight LB broth cultures of CHA0, CHA0/pME6259, and CHA638/pME6259 were washed with 0.9% NaCl solution and diluted with 0.9% NaCl to give a final concentration of 10⁷ CFU/ml. The seeds were pregerminated as described above. Wheat seedlings were soaked in the bacterial suspension for 60 min. Five seedlings were transferred to each flask, covered with soil, and incubated in a growth chamber with 16 h of light at 18°C (160 μE/m²/s), followed by 8 h of darkness at 13°C. The flasks were arranged in a completely randomized design. Seven days later, plants were removed from the flasks and shaken gently to discard loosely adherent soil. Roots with tightly adherent soil (in this article defined as the rhizosphere) were shaken in sterile saline solution (0.9%) at 300 rpm for 5 min.

Rhizosphere colonization with bacteria and Fusarium spp. was determined from the resulting suspension by plating serial dilutions on King’s B agar or on malt agar amended with rifampin (100 μg/ml) (24). β-Galactosidase activity was measured using 200 μl of the rhizosphere suspension. Control treatments inoculated with the wild-type strain, CHA0, were used to subtract background activity from calculations. Plants were washed with tap water, screened for quantifiable visible lesions, blotted dry, and weighed. Representative samples of root and shoot tissue from all treatments were surface sterilized and plated onto Komada's medium (27) to check for endophytic infection with F. oxysporum. The experiment was repeated twice with four replicate flasks (five plants/flask) per treatment. An analysis of variance was performed using Systat version 9.0 (Systat Inc., Evanston, Ill.) to analyze trial-by-treatment interaction. Data could not be pooled, and two individual trials are presented separately in Table 3. Means of four replicate flasks were separated using Fisher's protected (P = 0.05) least significant difference test.

TABLE 3.

Expression of a phlA′-′lacZ fusion in P. fluorescens strains CHA0 and CHA638 in the wheat rhizosphere with and without F. oxysporum strain 242 or 798^a

P. fluorescens strain	F. oxysporum strain	β-Galactosidase activity (U per 108 CFU)		Plant fresh wt (g)		Bacterial rhizosphere colonization (107 CFU per g)
		Expt 1	Expt 2	Expt 1	Expt 2	Expt 1	Expt 2
None	None	nd	nd	2.7 a	2.3 a	nd	nd
	242	nd	nd	3.1 a	2.4 a	nd	nd
	798	nd	nd	3.1 a	2.7 a	nd	nd
CHA0/pME6259	None	65 a	34 a	2.6 a	2.6 a	2.7 a	1.4 a
	242	304 b	56 b	2.8 a	2.6 a	0.9 a	1.1 a
	798	38 c	22 c	2.6 a	2.6 a	1.9 a	1.1 a
CHA638/pME6259	None	460 b	287 d	2.9 a	2.6 a	2.5 a	1.5 a
	242	514 b	365 d	3.0 a	2.7 a	1.7 a	1.0 a
	798	420 b	252 d	3.0 a	2.6 a	2.4 a	0.8 a

^aP. fluorescens strains CHA0 (wild type) and CHA638 (phlF mutant) carrying a phlA′-′lacZ translational fusion on plasmid pME6259 were tested. F. oxysporum strain 242 or 798 was added to the gnotobiotic system at 10⁴ microconidia per g of soil. Plants were grown for 14 days in artificial soil under gnotobiotic conditions. Fresh weights of all plants (always five) in one flask are presented. Values represent the means of four replicate flasks with five plants each. Means within a column followed by a different letter were significantly different at P ≤ 0.05 according to Fisher's protected least significant difference test. nd, not detected.

β-Galactosidase assays and plasmid stability.

β-Galactosidase assays were conducted by the method of Miller (32). Activities were expressed as units per 10⁸ CFU. Unit values were calculated by the method of Miller (32), following the formula 1,000 × [OD₄₂₀/(time [minutes] × sample volume [milliliters])] = units of β-galactosidase. The stability of plasmid pME6259 was evaluated by replica plating onto KB agar supplemented with tetracycline (125 μg/ml). Plates were incubated at 27°C for 48 h, and the percentage of tetracycline-resistant colonies, indicative of the presence of pME6259, was determined.

RESULTS

Production of FA by 12 strains of F. oxysporum in four different media.

We extracted FA from F. oxysporum cultures grown in four different media (malt broth, Czapek Dox medium, and R2 broth with and without zinc). The FA detection limit was 0.2 nmol per ml of medium. F. oxysporum strains 235, 240, and 242 did not produce detectable amounts of FA in any of the four media (data not shown). The other nine strains did produce FA in at least one of the media tested (Table 1). In malt liquid medium, only strain 241 produced significant amounts of FA in both experiments (Table 1). Strains 238, 798, and 801 produced small amounts in only one of the two experiments. In Czapek Dox broth, strains 233, 234, 236, 237, 238, 241, and 410 produced small but detectable amounts of FA. Strains 801 and 798 produced large amounts of FA in this medium, ranging between 0.8 and 3.5 mM. In the saccharose-based medium R2, strain 241 produced large amounts of FA, reaching a concentration up to 550 μM. In this medium, FA production by strains 233, 798, and 801 was also detected (Table 1). The addition of zinc at a concentration of 310 μM to R2 medium abolished production of FA in all strains tested (data not shown).

Repression of a phlA′-′lacZ fusion in P. fluorescens CHA0 in vitro depends on FA concentration.

The bacteria were grown in the presence of 100, 300, 500, and 750 μM FA under the same conditions (e.g., temperature, medium, inoculation, and rpm) as in the experiments with the culture filtrates of F. oxysporum (Table 2). The plasmid-borne translational phlA′-′lacZ gene fusion was expressed from the mid-exponential to the early stationary growth phase in the absence of added FA (Fig. 1). β-Galactosidase activities at mid-exponential growth (OD₆₀₀ of 0.638) were approximately 120 U per 10⁸ CFU and between 2,300 and 2,420 U per 10⁸ CFU when the OD₆₀₀ was ≥2 (early stationary growth phase). Reporter gene expression was not affected significantly by the addition of 100 μM FA to the growth medium. A concentration of 300 μM FA reduced the β-galactosidase activity of strain CHA0/pME6259 to about half the amount in the control. FA concentrations of 500 and 750 μM blocked expression of the phlA′-′lacZ fusion almost completely (Fig. 1). The addition of FA did not affect bacterial growth (data not shown).

FIG. 1.

Repression of expression of a phlA′-′lacZ translational fusion by the addition of synthetic FA at different concentrations to a growing culture of P. fluorescens CHA0 harboring pME6259. Strain CHA0/pME6259 was grown in OSG medium amended with Czapek Dox at a ratio of 4:1 at 27°C. Different concentrations of FA were added to the medium: none (◊), 100 μM (□), 300 μM (×), 500 μM (⧫), and 750 μM (▪). Throughout growth, β-galactosidase activities and OD₆₀₀ were determined. Values are the means of three replicate cultures. Addition of FA to the medium had no effect on cell growth. Error bars representing the standard errors of the means may be obscured by symbols.

Influence of culture filtrates of F. oxysporum on expression of a phlA′-′lacZ fusion in P. fluorescens strains CHA0 and CHA638.

Addition of F. oxysporum culture filtrates stimulated bacterial growth to different extents (data not shown). In order to avoid effects of bacterial growth caused by the addition of different F. oxysporum culture filtrates, β-galactosidase activities are always reported for a certain cell density (OD₆₀₀) rather than for incubation time (Fig. 1 and 2 and Table 2).

FIG. 2.

Influence of addition of filtrates of F. oxysporum strain 801 grown in Czapek Dox medium for 7 days at 24°C at 180 rpm and differing in FA content on phlA′-′lacZ expression in a growing culture of P. fluorescens CHA0. Filtrates of strain 801 were added to OSG medium at a ratio of 1:4, giving final FA concentrations of 400 μM (□) and 160 μM (▪). The control treatment received the same amount of Czapek Dox (○). β-Galactosidase activities and OD₆₀₀ of strain CHA0/pME6259 were determined throughout growth. Values are the means of three replicate cultures. Error bars represent the standard errors of the means. Some of the error bars are too small to be shown.

Six strains of F. oxysporum were selected which produced either large amounts of FA (strains 798 and 801) or minimal (strain 241) or undetectable (strains 240, 242, and 410) amounts of FA in Czapek Dox medium after 7 days of growth (Table 1). Filtrates from F. oxysporum cultures grown in Czapek Dox medium for 7 days (Table 1, experiment 2) were added to the bacterial growth medium (OSG) at a ratio of 20% (vol/vol), which gave final FA concentrations of 0, 25, 0, 0, 700, and 160 μM for strains 240, 241, 242, 410, 798, and 801, respectively. The effect of the fungal culture filtrates on phlA expression in strains CHA0 and CHA638 during the exponential and early stationary growth phases was monitored. In the control treatment, Czapek Dox medium was added to OSG medium at the same ratio as the fungal filtrates (20%, vol/vol). The addition of Czapek Dox to OSG medium did slightly enhance bacterial growth compared to growth in plain OSG (data not shown).

Culture filtrates of F. oxysporum strains 240, 241, 242, and 410 did not affect the β-galactosidase activities of strain CHA0/pME6259 at both cell densities reported (OD₆₀₀ of 1.5 and 2.5) compared to those of the control (Table 2). Addition of culture filtrates of strains 798 and 801 repressed phlA expression to different extents. At an OD₆₀₀ of 1.5, the β-galactosidase activities of strain CHA0/pME6259 grown in the presence of filtrate from strain 798 or 801 were dramatically repressed and did not exceed 220 U per 10⁸ CFU, which represents a sevenfold phlA repression. At an OD₆₀₀ of 2.5, β-galactosidase activity measured in CHA0/pME6259 grown with 801 filtrate was still significantly lower than that of the control, but repression was only 30%, whereas repression in the presence of 798 filtrate was 70% (Table 2). The repressing effects of culture filtrates 798 and 801 on the phlA′-′lacZ fusion were abolished when the fusion was expressed in the phlF mutant CHA638 (Table 2).

In a follow-up experiment, we tested the effect of culture filtrates of F. oxysporum strain 801 with different FA contents on phlA expression in P. fluorescens CHA0. Filtrates from cultures grown in Czapek Dox (Table 1) were used to obtain final FA concentrations of 400 and 160 μM in the bacterial growth medium. Average β-galactosidase activity in P. fluorescens CHA0/pME6259 remained below 680 U per 10⁸ CFU in medium amended with culture filtrate of experiment 1 (400 μM) (Fig. 2). phlA expression in medium with filtrate from experiment 2 (160 μM) was repressed to a smaller extent; the maximal β-galactosidase activity in this treatment was 1,550 U per 10⁸ CFU at an OD₆₀₀ of 2.7, whereas that of the control was >2,000 U per 10⁸ CFU.

Expression of a phlA′-′lacZ fusion in strains CHA0 and CHA638 in the rhizosphere of wheat in the presence of two different F. oxysporum strains.

After 2 weeks on wheat roots, retention of plasmid pME6259 was 97.3 ± 2.1% in wild-type strain CHA0 and 65.8 ± 4.3% in phlF mutant CHA638, based on the viable bacteria reisolated from the rhizosphere that grew on tetracycline-amended agar. β-Galactosidase activities were calculated per CFU still carrying reporter plasmid pME6259. Plant fresh weight did not differ significantly between the treatments in both experiments shown in Table 3 and ranged between 2.6 and 3.1 g in experiment 1 and between 2.3 and 2.7 g in experiment 2. No visible lesions were detected on the wheat plants, and no F. oxysporum could be reisolated on Komada's medium from the inside of wheat roots, stems, or leaves, indicating the absence of an endophytic or pathogenic interaction (data not shown). Colonization of the rhizosphere with F. oxysporum was around 1.9 × 10⁵ CFU per g of rhizosphere and did not differ significantly between the experiments (data not shown). No statistical differences could be found for bacterial rhizosphere colonization between the treatments in the two experiments (Table 3). Rhizosphere colonization was estimated at from 0.9 × 10⁷ to 2.7 × 10⁷ CFU per g of rhizosphere in experiment 1 and ranged between 0.8 × 10⁷ and 1.5 × 10⁷ CFU per g of rhizosphere in experiment 2. No P. fluorescens cells were detected in the treatments without bacteria added (Table 3).

In both experiments, β-galactosidase activity measured in strain CHA0 was lowered significantly in the treatment with F. oxysporum strain 798, i.e., to 60% of that of the control treatment (Table 3). In the treatment with F. oxysporum 242, β-galactosidase activity measured in CHA0/pME6259 was enhanced two- to fourfold compared to the treatment without Fusarium added. β-Galactosidase activities were seven- to eightfold higher in the phlF mutant CHA638 than in the wild type when Fusarium was not added. The amendment with F. oxysporum strain 242 or 798 had no significant influence on β-galactosidase activity in CHA638 (Table 3), suggesting that the effects of F. oxysporum on the wild-type CHA0 are mediated by PhlF.

DISCUSSION

F. oxysporum is found worldwide. This fungus can persist almost indefinitely in soil due to its ability to survive winter in the mycelial or chlamydospore state, with the result that normal crop rotation is not a practical control measure (6). The species occurs as a soil saprophyte as well as a wilt pathogen of many crop plants, and many of the strains isolated produce FA (2, 4, 10, 35, 42). In this study we addressed in vitro and in the wheat rhizosphere the impact of different F. oxysporum strains on a gene critical for the biocontrol activity of the model organism P. fluorescens CHA0.

FA production by F. oxysporum is strain dependent.

Production of FA varied markedly between F. oxysporum strains and media (Table 1). Sucrose has been reported to favor FA production (10, 28), and indeed, F. oxysporum strains 798 and 801 did produce large amounts of FA in Czapek Dox broth, which is a sucrose-based medium. Richard's medium is also supposed to favor FA production (9, 37). In the slightly modified form of this medium (R2) used in our experiments (7), only four strains produced detectable amounts of FA. The addition of zinc to this medium abolished FA production, as has been shown for other Fusarium strains (9, 14, 17).

DAPG biosynthesis in CHA0 is repressed in vitro by the fungal metabolite FA.

Synthetic FA represses biosynthesis of DAPG in CHA0, with an almost complete block at 500 μM (Fig. 1), as observed by Schnider-Keel et al. (38). Here we show that an FA concentration of 300 μM gave an intermediate inhibition of phlA expression, and at a concentration of 100 μM, no significant repressive effect could be observed in vitro (Fig. 1).

Strains of the species F. oxysporum produce a wide range of secondary metabolites such as phytotoxins, pigments, and mycotoxins and cause diverse responses in plants, animals, humans, and microorganisms (29). In our fungal filtrate experiments, three lines of evidence show that FA is the secondary metabolite most likely to be responsible for phlA repression. (i) Repression of gene expression increased with increasing contents of FA in the fungal filtrates. The levels of repression correspond to those reached by similar concentrations of synthetic FA added to the medium (Table 2, Fig. 1). (ii) Moreover, addition of the same amount of two different fungal filtrates from the same Fusarium strain, 801, with different FA contents to a growing culture of CHA0 repressed phlA expression differentially. Amendment of one culture filtrate (Table 1, experiment 1) brought the FA concentration in the bacterial growth medium to 400 μM and blocked phlA expression almost completely. Addition of a different filtrate, in which FA concentration was only 160 μM, resulted in partial repression only (Fig. 2). These effects correspond well with those obtained with the same concentration of synthetic FA (Fig. 1). (iii) Evidence that an intact phlF gene is needed for repression of phlA by FA has been provided by Schnider-Keel et al. (38). Similarly, in our study, repression of phlA expression through the addition of fungal filtrates was abolished in the phlF mutant CHA638 (Table 2).

Interactions between F. oxysporum strains and wheat.

Although the F. oxysporum strains were isolated from wheat roots or from fields cultivated with wheat, the absence of visible lesions, no reduction in plant weight, and lack of detection of the fungus inside the plant material strongly indicate that there are no pathogenic interactions with the wheat cultivar used in our plant experiments. F. oxysporum is a cosmopolitan soil saprophyte, and specialized pathogenic strains are causal agents of vascular wilts and damping-off diseases. They exhibit a high degree of host specificity, and many formae speciales have been described (6). In our case, either the strains are nonpathogenic, or host genotype recognition failed and infection did not take place. However, the strains colonized the wheat rhizosphere, as shown by reisolation of the fungi from the rhizosphere.

Interactions between F. oxysporum strains and CHA0 in the wheat rhizosphere.

F. oxysporum strains 242 (FA⁻) and 798 (FA⁺) altered phlA expression in P. fluorescens CHA0 differentially compared to the control treatment without Fusarium added. To our knowledge, this is the first report to address the impact of a Fusarium strain on a gene critical for biocontrol in the rhizosphere of wheat without any pathogenic interactions. As expected from in vitro experiments, the FA-producing strain 798 also repressed phlA expression of CHA0 in the wheat rhizosphere. Previously, the tomato-pathogenic forma specialis radicis-lycopersici has been shown to repress DAPG synthesis in CHA0 through production of FA in a hydroponic rockwool system (14). Our results show that a nonpathogenic FA-producing strain of F. oxysporum represses phlA in CHA0 in the wheat rhizosphere. We hypothesize that this suppressing effect is caused by the production of FA in the rhizosphere. The latter is supported by the observation that the repressive effect was abolished in the phlF mutant CHA638. PhlF is proposed to be a pathway-specific repressor of DAPG synthesis in that it binds to the promoter(s) of phlA (5, 11, 38). A PhlF-mediated DAPG repression by FA has been demonstrated in vitro (38). It is conceivable that FA might interact with the repressor, leading to increased binding characteristics of the PhlF-promoter complex. However, it cannot be excluded that other, as yet unidentified, regulatory mechanisms act upstream of phlF and that DAPG repression by FA may not be the result of a higher affinity of an FA-PhlF complex to the phlA promoter region. To our knowledge, this is the first report to show a PhlF-mediated DAPG suppression in the rhizosphere.

The repressing effect of F. oxysporum 798 contrasted even with an enhancing effect of strain 242 on phlA gene expression in CHA0. Previously, we have shown that Pythium ultimum can stimulate phlA gene expression in CHA0 on both cucumber and maize (33), an effect that might be an indirect consequence of disease due to an increased release of root exudates from damaged roots. By contrast, in the current study, in which no pathogenic interactions were observed, the change in gene expression is interpreted as a direct effect of the fungus. It is possible that strain 242 produces an inducer that binds to PhlF and thereby prevents the interaction of the repressor protein with the phl promoter. The fact that the presence of strain 242 did not alter phlA expression in the phlF mutant CHA638 supports this hypothesis.

In vitro, phlA expression of the phlF mutant CHA638 was not higher than in the wild type (Table 2). However, on wheat roots, phlA was seven- to eightfold overexpressed in strain CHA638 compared with CHA0. This observation suggests that PhlF generally restricts expression of the DAPG operon in the rhizosphere of wheat more profoundly than in liquid cultures.

Implications for biocontrol.

Even though the F. oxysporum strains used in this study had no pathogenic impact on the plants, the fungi did influence the production of DAPG, a key factor in biocontrol of beneficial bacteria co-occurring in the rhizosphere. F. oxysporum is nearly universally present in soil, which suggests that at least some FA-producing strains will be found in almost any soil sample taken. The suppressive effect of FA on the production of a secondary metabolite critical in biocontrol may affect the performance of FA-sensitive biocontrol bacteria against any soil-borne pathogen. This fact should be taken into consideration when using biocontrol strains in agricultural systems where FA producers are present. Consistent biocontrol of soil-borne diseases might be achieved by applying beneficial microorganisms, or mutants thereof (e.g., phlF mutants), that are insensitive to FA. Efficient formulations offer another approach to improving biocontrol performance: coapplication with strains capable of detoxifying FA and amendments with zinc or a substance that neutralizes repressors (e.g., PhlF) could improve the level and reliability of biocontrol (14, 45).