1615

Journal of Food Protection, Vol. 70, No.7, 2007, Pages 1615-1621

Isolation of Maize Soil and Rhizosphere Bacteria withAntagonistic Activity against Aspergillus flavus and

Fusarium verticillioides

JEFFREY D. PALUMBO,u TERESA L. O'KEEFFE,1 AND HAMED K. ABBAS2

IPlant Mycotoxin Research Unit, U.S. Department ofAgriculture, Agricultural Research Service, Albany, California 94710; and 2Crop Genetics andProduction Research Unit, U.S. Department ofAgriculture, Agricultural Research Service, Stoneville, Mississippi 38776, USA

MS 06-613: Received 30 November 2006/Accepted 10 March 2007

ABSTRACT

Bacterial isolates from Mississippi maize field soil and maize rhizosphere samples were evaluated for their potential asbiological control agents against Aspergillus fiavus and Fusarium verticillioides. Isolated strains were screened for antagonisticactivities in liquid coculture against A. fiavus and on agar media against A. fiavus and F. verticillioides. We identified 221strains that inhibited growth of both fungi. These bacteria were further differentiated by their production of extracellularenzymes that hydrolyzed chitin and yeast cell walls and by production of antifungal metabolites. Based on molecular andnutritional identification of the bacterial strains, the most prevalent genera isolated from rhizosphere samples were Burkholderiaand Pseudomonas, and the most prevalent genera isolated from nonrhizosphere soil were Pseudomonas and Bacillus. Lessprevalent genera included Stenotrophomonas, Agrobacterium, Variovorax, Wautersia, and several genera of coryneform andenteric bacteria. In quantitative coculture assays, strains of P. chlororaphis and P. fiuOl:escens consistently inhibited growthof A. fiavus and F. verticillioides in different media. These results demonstrate the potential for developing individual biocontrolagents for simultaneous control of the mycotoxigenic A. fiavus and F. verticillioides.

Aspergillus fiavus and Fusarium verticillioides arewidespread inhabitants of agricultural soils associated withmany crops, including maize (com, Zea mays L.),'and aretwo of several fungal species that cause ear rot of maize(50). In addition to the economic losses caused by disease,further losses may occur due to contamination of maizewith aflatoxins produced by A. fiavus and fumonisins pro­duced by F. verticillioides (4, 43). These losses affect grow­ers, processors, and livestock producers and are a threat tohuman and animal health (16).

Current methods used to reduce aflatoxin and fumon­isin contamination of maize include modification of culturalpractices to reduce preharvest and postharvest fungal in­fection (34) and introduction of resistance traits into agro­nomically valuable maize cultivars by conventional breed­ing (1, 32, 34) and molecular (19, 34) techniques. Intro­duction of a Bacillus thuringiensis Bt endotoxin transgeneinto maize for insect resistance indirectly reduces mycotox­in contamination of maize (6, 48). In contrast, the use ofchemical pesticides in some cases has resulted in the in­crease in Fusarium and Aspergillus mycotoxins (17).

Several biological control methods for using fungi andbacteria to limit mycotoxin contamination in maize havebeen developed, such as use of nonaflatoxigenic A. fiavusstrains to displace aflatoxigenic strains (2, 11, 18), controlof A. fiavus by antagonistic strains of Kluyveromyces (30),and control of F. verticillioides by endophytic and rhizo­sphere strains of Bacillus (5, 12). Wicklow et al. (51) found

* Author for correspondence. Tel: 510-559-5876; Fax: 510-559-5777;E-mail: palumbo@pw.usda.gov.~

that the fungal endophyte Acremonium zeae produced pyr­rocidine antibiotics that were active against both A. fiavusand F. verticillioides. In most of these cases, the goal ofmycotoxin reduction is met by limiting or delaying growthof toxigenic fungi.

In this study, we hypothesized that single bacterialstrains with activity against both A. fiavus and F. verticil­lioides could be isolated from the maize environment. Us­ing individual strains for biocontrol of multiple targetswould facilitate management of preharvest mycotoxin con­tamination of maize. Bacterial populations from maize soiland rhizosphere samples were screened for phenotypes an­tagonistic against A. fiavus and F. verticillioides. Quanti­tative coculture assays in liquid media were used to deter­mine the extent of the antifungal activity of these bacteriaand to predict the potential for their successful applicationas biocontrol agents in the field.

MATERIALS AND METHODS

Sampling and isolation of bacteria. Soil and maize sampleswere collected from maize fields in Stoneville and Elizabeth,Miss. In November 2004 and April 2005, a total of 40 soil sam­ples were taken from nine field plots. In May 2005, four maizeplants were collected from each of five separate field plots, for atotal of 20 samples. Bacteria were isolated from soil samples byadding 1 g of soil to 10 m1 of sterile 0.05% Tween 80 solution,vortexing for 30 s, sonicating on ice for 15 min in a Bransonmodel 3210 sonication bath, and vortexing for an additional 30 s.To isolate maize rhizosphere bacteria, rOots were aseptically re­moved from each plant sample by cutting at the crown with asterilized razor blade. Each root sample was added to 50 ml ofsterile 0.05% Tween 80 in a l-qt (0.96-liter) polyethylene Ziploc

III

'II1,1

,1616 PALUMBO ET AL. J. Food Prot., Vol. 70, No.7

bag. Bacteria were removed from root and soil surfaces by agi­tating the contents of each bag for 1 min to dislodge soil particlesfrom roots and sonicating on ice for 15 min. Sonicated soil androot samples were serially diluted in 0.05% Tween 80 and spreadplated onto 0.1 X tryptic soy agar (TSA; Difco, Becton Dickinson,Sparks, Md.) containing 100 mg liter-1 cycloheximide. Plateswere incubated at 28°C for 2 days, and individual colonies fromeach sample were randomly selected to screen for antifungal phe­notypes.

Screening for antifungal activity. Bacterial strains were ini­tially screened for antifungal activity against the nor mutant A.flavus strain Papa 827 (40) in potato dextrose broth (PDB; Difco,Becton Dickinson). Individual bacteria were inoculated into singlewells of a 24-well microtiter plate with 1 ml of PDB containing~103 A. flavus Papa 827 conidia and incubated at 28°C for 7 dayswithout shaking. Antifungal activity was assessed visually bycomparing fungal growth in wells treated with bacteria withgrowth in control wells containing only A. flavus Papa 827. Bac­terial strains that inhibited A. flavus growth were screened furtherfor activity against A. flavus Papa 827 and the fumonisin-produc­ing F. verticillioides strain 935 on potato dextrose agar (PDA;Difco, Becton Dickinson). Plates were inoculated in the centerwith 10 fLl of an A. flavus or F. verticillioides conidial suspension(~105 conidia ml- I), and bacterial strains were inoculated aroundthe periphery of the plate, 3 cm from the conidia. Screen plateswere incubated for 7 days at 28°C and assessed visually for an­tifungal activity.

Production of hydrolytic enzymes. Bacterial strains weretested for extracellular hydrolysis of chitin and yeast cell walls indefined minimal agar 813C and 813Y, respectively, as previouslydescribed (39). Colloidal chitin was prepared from crab shell chi­tin (Sigma-Aldrich, St. Louis, Mo.) as previously described (31).Yeast cell walls were prepared from autoclaved baker's yeast (Sac­charomyces cerevisiae) as previously described (39). Chitin andyeast cell walls were added to medium 813C and 813Y, respec­tively, at a concentration of 1% (wtlvol). Bacteria were spot in­oculated and incubated at 28°C for up to 5 days. Enzyme activitythat hydrolyzed chitin or yeast cell walls was measured qualita­tively by the production of visible clearing zones surrounding thebacterial colonies, typically > 1 mm beyond the edge of the col­ony.

Bacterial identification. Total DNA from bacteria was pre­pared from single colonies grown on O.5X TSA using theInstaGene matrix (Bio-Rad, Hercules, Calif.) according to themanufacturer's instructions. The 16S rRNA gene fragments wereamplified in PCRs using 1 to 5 fLl of each cell suspension astemplate and primers 27F (5'-AGAGTTTGATCMTGGCTCAG-3') and 1525R (5'-AAGGAGGTGWTCCARCC-3') (29).PCR conditions were 95°C for 5 min; 35 cycles of 94°C for 30s, 55°C for 30 s, and noc for 90 s; and noc for 5 min. Ampli­fication was confirmed by electrophoretic separation of the frag­ments on a 0.8% agarose gel, and amplified fragments were pu­rified using QIAquick PCR purification columns (Qiagen, Inc.,Valencia, Calif.). Purified fragments were sent to Davis Sequenc­ing, Inc. (Davis, Calif.) for sequencing using the same primers.Bacteria were identified based on similarities of sequences to ho­mologous 16S rRNA gene sequences from bacterial strains des­ignated as type strains in the Ribosomal Database Project database(15).

For identification using the Biolog Microbial IdentificationSystem (Biolog, Inc., Hayward, Calif.), bacteria were first sepa­rated into gram-positive or gram-negative strains using KOH (20).

Gram-positive strains were identified using GP2 MicroPlates(Biolog) and gram-negative strains were identified using GN2MicroPlates (Biolog), according to the manufacturer's instruc­tions. Identification was based on the similarity index of carbonsource utilization by each strain relative to that of identified ref­erence strains in the Biolog GP and GN databases.

Quantification of antifungal activity. A subset of 20 bac­terial strains showing antifungal activity in preliminary screenswas tested for antagonism against A. flavus strain F34 and F.verticillioides strain 935. A. flavus F34 is an afiatoxigenic S strainthat was isolated from almond orchard soil (7). Assays were per­formed in PDB and in maize kernel broth (CKB), which wasprepared by homogenizing frozen sweet maize kernels in distilledwater (100 g liter-I), removing solids by centrifugation at 16,000X g for 15 min, and autoclaving. Bacteria were grown overnightin 0.5X tryptic soy broth (Difco, Becton Dickinson) at 28°C. Tenmicroliters of each culture (~108 CFU ml- I) and ~102 A. flavusF34 or F. verticillioides 935 conidia were added to 10 ml of PDBor CKB in 100-mm-diameter petri dishes. Following incubationat 28°C for 7 days without shaking, fungal mycelium was recov­ered by filtration onto Whatman no. 4 paper filters and dried over­night. Fungal dry weights from six replicate samples (three rep­licates per treatment performed in two independent experiments)were recorded for each combination of A. flavus or F. verticil­lioides + bacterial strain in each medium. Bacterial inhibition offungal growth was quantified relative to control cultures of fungigrown for 7 days without bacteria in each medium. Student's ttest was used to determine the significance of differences betweencontrol and treated samples.

RESULTS

Isolation and screening of bacterial antagonists.Bacteria were isolated from maize field soil samples thatwere taken after the 2004 growing season and at the be­ginning of the 2005 growing season and from maize rhi­zosphere samples taken approximately 4 weeks after plant­ing in 2005. Isolated colonies from each sample were se­lected randomly and screened for fungal growth inhibitionagainst A. flavus Papa 827. This strain contains a mutationin the nor gene, resulting in accumulation of norsolorinicacid, a bright red-orange precursor in the aflatoxin biosyn­thetic pathway. Use of this strain allowed inhibition of fun­gal growth to be easily visualized in liquid coculture withbacteria in multiwell culture plates. Bacterial strains thatshowed inhibition of A. flavus growth in PDB coculturescreens were screened for antifungal phenotypes on PDA.Of 1,018 strains screened, 443 strains had inhibitory activ­ity against A. flavus, and 221 of those strains had inhibitoryactivity against both A. flavus and F. verticillioides (Table1).

Two types of fungal growth inhibition were observedwhen screens were performed on agar media: (i) bacterialinhibition of fungal growth in a visibly clear zone surround­ing the bacterial colony or (ii) inhibition of fungal growthupon contact of the mycelium with the bacterial colony.Some bacterial strains also inhibited F. verticillioides aerialmycelial growth, apart from lateral mycelial growth (datanot shown). Fifty-six strains produced a zone of fungalgrowth inhibition beyond the edge of the bacterial colony,suggesting the production of one or more diffusible anti-

J. Food Prot., Vol. 70, No.7 BACTERIAL ANTAGONISM OF ASPERGILLUS AND FUSARIUM 1617

TABLE 1. Incidence and distribution of bacterial strains withantifungal phenotypes

No. of strains with No. of activeactivity against: strains producing:

No. of A. flavus Antifungal Chitin YCWstrains A. and F. metabo- hydro- hydro-

Source screened flavus verticillioides lite(s) lysis lysisa

Soil 600 208 79 14 17 37Maize

rhizosphere 418 235 142 42 50 44

Total 1,018 443 221 56 67 81

a YCW, yeast cell walls.

microbial compounds. Antagonistic bacterial strains wereassayed for the production of extracellular hydrolyzing ac­tivity against chitin and yeast cell walls. Sixty-seven strainsproduced clearing zones on colloidal chitin agar, and 81strains produced clearing zones on yeast cell wall agar. Hy­drolysis of yeast cell walls was generally interpreted as 13­glucanase activity, because the S. cerevisiae cell wall iscomposed predominantly of 13-1,3 and 13-1,6 glucans. Thenumber of strains with an antifungal phenotype is sum­marized in Table 1.

Classification and distribution of antagonistic bac­teria. Bacterial strains were identified by a combination of16S rRNA gene sequence analysis and the Biolog Micro­bial Identification System. The accuracy of each method islimited by the diversity and accurate identification of thebacterial species in each reference database. Therefore, only16S rRNA gene sequences of type strains present in the16S Ribosomal Database Project database were used forclassification. Because the species diversity in the Biologdatabases is limited, the carbon metabolism-based Biologmicroplate system was used to confirm presumptive clas­sification to at least genus.

The major genera and bacterial groups identified fromsoil and rhizosphere samples are listed in Table 2. The mostprevalent genera found in the soil samples were Pseudo­monas and Bacillus and several coryneform genera, includ-

TABLE 2. Frequency of isolation of major genera with antifun­gal activity

No. of strains isolated from:

Genus or group Soil Rhizosphere

Achromobacter 0 2Agrobacterium 0 5Bacillus 10 4Burkholderia 1 45Chryseobacterium 2 0Pseudomonas 14 15Wautersia 4 1Stenotrophomonas 2 8Variovorax 1 6Coryneform bacteria 13 4Enteric bacteria 2 7

TABLE 3. Characteristics of bacterial strains selected for use inquantitative antagonism assays

Antifungal Hydrolysis of:Strain metabo-

no. Species Source lite(s)a Chitinb YCWc

JP2200 Achromobacter Rhizospherexylosoxidans

JP2157 Agrobaeterium Rhizospheretumefaciens

JP2170 A. tumefaciens RhizosphereJP1065 Arthrobacter ni- Soil +

cotinovoransJP1203 A. histidinolovor- Soil +

ansJPll07 Bacillus licheni- Soil +

formisJP1206 B. licheniformis Soil + +JP2163 B. cereus/thurin- Rhizosphere + +

giensisJP2003 Burkholderia ce- Rhizosphere +

paciaJP2047 B. cepacia Rhizosphere + +JP2112 B. cepacia Rhizosphere +JP1015 Pseudomonas Soil + + +

chlororaphisJP1047 P. chlororaphis Soil + +JP2034 P. fluorescens Rhizosphere + +JP2175 P. fluorescens Rhizosphere + +JP1035 Wautersia eutro- Soil

phaJP2022 W. eutropha RhizosphereJP1026 Stenotropho- Soil + +

monas malto-philia

JP2076 S. maltophilia Rhizosphere + +JP2093 Variovorax para- Rhizosphere

doxus

a Production of metabolites inhibitory to A. flavus and/or F. ver­ticillioides. +, production of growth inhibition zone on PDA;-, no zone of growth inhibition on PDA.

b Chitin hydrolysis on 1% colloidal chitin agar. +, production ofvisible clearing zone; -, no visible clearing zone.

C YCW, yeast cell wall. Hydrolysis on 1% YCW agar. +, produc­tion of visible clearing zone; -, no visible clearing zone.

ing Arthrobacter, Brevibacterium, Corynebacterium, andMicrobacterium. In rhizosphere samples, Burkholderia wasclearly the predominant genus isolated, followed by Pseu­domonas and Stenotrophomonas. Although coryneformgenera were less prevalent in the rhizosphere, enteric gen­era, including strains of Enterobacter, Serratia, and Pan­toea, were more prevalent. Other genera isolated from soiland rhizosphere samples in lower numbers included strainsof Agrobacterium, Wautersia, Variovorax, and Chryseo­bacterium.

Quantification of fungal growth inhibition. The 20bacterial strains listed in Table 3 were selected for quanti­tative analysis of antifungal activity, based on their activityagainst A. flavus Papa 827 and F. verticillioides in agar andliquid media and their production of diffusible antimicro-

FIGURE 1. Growth of fungi in coculture with antagonistic bac­teria. Bacterial strain numbers refer to the list in Table 3. (A)Inhibition of A. flavus strain F34 (1II) and F. verticillioides strain935 (D) in PDB. Fungal weights are relative to untreated controlcultures, where 100% = 26.7 mg of A. flavus or 30.6 mg ofF.verticillioides. (B) Inhibition of A. flavus strain F34 (1II) and F.verticillioides strain 935 (D) in eKB. Fungal weights are relativeto untreated. control cultures, where 100% = 26.0 mg of A. flavusor 33.5 mg ofF. verticillioides. Error bars represent the standarderror of six replicate samples per treatment.

DISCUSSION

J. Food Prot., Vol. 70, No.7

We isolated and screened bacteria from maize soil andrhizosphere samples to increase the probability of isolatingbacteria for use as biological control agents against A. fia­vus and F. verticillioides in the maize environment. Usingthe target environment as a source for biocontro1 organismsresulted in the isolation of agents specifically adapted tothat environment and therefore more likely to be effective(26). This approach has been used in several studies forselection of biocontro1 agents, including selection of bac­teria from yam farm soil for control of postharvest yam rot(37), isolation of yeasts from grape for control of Asper­gillus species on grape (10), isolation of groundnut-asso­ciated bacteria for control of groundnut collar rot disease(27), and screening of canola-associated bacteria for controlof blackleg disease of canola (44). Native bacterial popu-

(83 to 93% reduction) in PDB but were less effective inCKB (34 to 49% reduction of A. fiavus and 33 to 34%reduction of F. verticillioides).

Other strains that were tested significantly inhibitedboth fungi (P < 0.01) in only one test medium. For ex­ample, in CKB, A. fiavus and F. verticillioides were inhib­ited by Agrobacterium tumefaciens strain IP2170 (89 and84% reduction, respectively), Arthrobacter nicotinovoransstrain IPlO65 (49 and 84% reduction, respectively), Bacil­lus strains IP1lO7 and JP2163 (74 to 75% and 60 to 63%reductions, respectively), P. chlororaphis strain IP1047 (69and 54% reduction, respectively), and Variovorax paradox­us strain IP2093 (74 and 44% reduction, respectively).None of these strains produced significant fungal inhibitionin PDB (Fig. 1). Achromobacter xylosoxidans strain IP2200inhibited A. fiavus by 84% and F. verticillioides by 84% inPDB and inhibited A. fiavus by 73% in CKB but did notinhibit F. verticillioides in CKB.

When individual strains of each bacterial genus werecompared, there was no consistent correlation between phe­notypes or sources and antifungal activity. For example,Bacillus licheniformis strain IPll07, which produces an an­tifungal metabolite, did not have more activity (quantita­tively) than did B. licheniformis strain JP1206, which doesnot produce an antifungal metabolite. B. licheniformis strainIP1lO7 and Bacillus cereus/thuringiensis strain IP2163 dif­fered in their production of chitinase activity but were notsignificantly different (P > 0.05) in their quantitative an­tifungal activity. Strains of B. cepacia that differed in theirhydrolyzing activity against yeast cell walls did not differsignificantly in their antifungal activity, but B. cepaciastrain IP2112, which does not produce an antifungal me­tabolite, tended to be less inhibitory to the fungi than werethe B. cepacia strains that produced the antifungal metab­olite. P. chlororaphis strain JPlO15, which showed hydro­lyzing activity against yeast cell walls, performed betterthan P. chlororaphis strain IPlO47, a nonactive strain, inboth media against both fungi, although the differenceswere significant only for A. fiavus in PDB (P < 0.01).Strain source (rhizosphere or bulk soil) and quantitative an­tifungal activities were not correlated among Wautersia eu­tropha and S. maltophilia strains.

'"'"~..,

PALUMBO ET AL.

A

------~---

I--

I T I Ii

II I1m I", I .4 Ilmlm :r 1m1.IUI.1iI

B

~

ii I i *m :T lili

1!III~H J IldU11 J;nJJ illl

1618

140

e 120c0

1000

'0C 80

:E60Ol

"ijjS(ij 40Olc

20:::>LL

0

120

e 100c00

'0 80,g~

:E 60Ol

"ijjS 40(ijOlc 20:::>LL

0

illo0::..,

bial compounds and hydrolytic enzymes. Strains of thesame species that had different phenotypes were chosen todetermine whether strain differences would result in quan­titatively distinct antifungal activity. The effect on fungalgrowth of coculturing these strains with A. fiavus F34 (anaflatoxin-producing field isolate) or F. verticillioides 935was measured in PDB and CKB (Fig. 1). CKB was usedas a surrogate medium in place of intact maize to facilitatemeasurements of fungal biomass and generally supportedthe same level of growth of both fungi as did PDB on adry-weight basis.

Several of the strains tested significantly inhibitedgrowth of both fungi (P < 0.01) regardless of the mediumused (Fig. 1). All three Burkholderia cepacia strains re­duced growth of A. fiavus by 74 to 79% in PDB and 75 to89% in CKB and reduced growth of F. verticillioides by61 to 87% in PDB and 73 to 89% in CKB. Pseudomonasfiuorescens strains IP2034 and IP2175 and Pseudomonaschlororaphis strain IP1015 reduced A. fiavus growth by 86to 91 % and 69 to 96% and reduced F. verticillioides growthby 83 to 89% and 54 to 91% in PDB and CKB, respec­tively. Both strains of Stenotrophomonas maltophilia inhib­ited A. fiavus (83 to 92% reduction) and F. verticillioides

J. Food Prot., Vol. 70, No.7 BACTERIAL ANTAGONISM OF ASPERGILLUS AND FUSARIUM 1619

lations also have been investigated as antagonists to A. fla­vus on almond (38), cotton (33), and maize (36) and toFusarium species on a variety of crops, including wheat(45), banana (3), onion (46), and maize (13).

In the present study, we isolated bacteria with antifun­gal activities from all soil and rhizosphere samples collect­ed, indicating that antagonistic phenotypes in these envi­ronments are not uncommon. Screening sequentially for an­tagonism against A. flavus and F. verticillioides allowedidentification of single bacterial strains with the potential tocontrol both pathogens. The majority of these strains wereisolated from the maize rhizosphere (Table 1), where thetotal microbial population includes a diversity of sapro­phytic and pathogenic fungi. The specific antifungal activ­ities we screened for-production of antifungal metabolitesand chitinase and l3-glucanase activity-were also moreprevalent in rhizosphere strains (Table 1), in which theycould contribute to ecological fitness in the rhizosphere.Broad-spectrum antifungal activity that may be ecologicallyadvantageous also may be advantageous for use in biolog­ical control. For example, Lysobacter enzymogenes has bio­control activity against Fusarium (24), Pythium (35),Aphanomyces (23), Bipolaris (52), and Magnaporthe (28)as a result of its production of multiple degradative en­zymes and antimicrobial factors.

Identification of the antagonistic bacterial strains re­vealed that the most prevalent genus recovered from rhi­zosphere samples was Burkholderia (Table 2). These Burk­holderia strains were identified by their 16S rRNA genesequences as B. cepacia, B. stabilis, B. vietnamiensis, B.multivorans, and B. pyrrocinia, which are all members ofthe B. cepacia complex (Bcc) (42). The presence of F. ver­ticillioides increases Bcc populations in the maize rhizo­sphere (9). The same phenomenon may be occurring in theMississippi maize fields we tested, which probably havebackground populations of indigenous F. verticillioides.Bcc bacteria are resident rhizosphere bacteria in many plantsystems and have been investigated in the past for biocon­trol applications against several soilborne fungal plant path­ogens. However, Bcc bacteria also have been associatedwith clinical infections, which has led to restrictions on theuse of these bacteria for plant disease control (41). The B.cepacia strains that were tested quantitatively for antifungalactivity (Table 3) performed well against both A. flavus andF. verticillioides (Fig. 1), but because of limitations on theuse of Bcc bacteria under field conditions, further investi­gation of these strains will be confined to the laboratory.All Stenotrophomonas strains isolated in this study wereidentified as S. maltophilia. Although the strains that wereassayed quantitatively for their antifungal activity (Table 3)performed relatively well (Fig. 1), they are of limited use­fulness in field applications because of the potential of S.maitophilia to cause human disease and the difficulty indifferentiating environmental and clinical strains (8).

Although Bacillus and Pseudomonas were the predom­inant genera isolated from nonrhizosphere soil, relativelyfew antagonistic Bacillus strains were isolated in this study(Table 2). In contrast, in a previous study Bacillus was themost prevalent genus with antifungal activity isolated from

almond fruits, representing 54 of 126 isolates (38). Al­though Bacillus might be more likely than other genera topersist in the harsh phyllosphere environment, there wouldbe no such selective pressure in agricultural soil given therelative abundance of nutrients and moisture, especially inthe rhizosphere. Bacillus subtilis from maize has shown po­tential as a biocontrol agent against F. verticillioides ingreenhouse and in vitro studies (12). In our study, the Ba­cillus strains selected for the quantitative antifungal assays(Table 3) significantly inhibited A. flavus and F. verticil­lioides growth in CKE but not in PDB (Fig. 1). Whetheractivity in one medium or the other is more relevant to fieldconditions, use of several media to demonstrate consistentactivity in vitro may be a better method for predicting fieldefficacy. Bacteria screened for antifungal activity in thisstudy were inoculated at much higher concentrations thanwere the fungal spores. Further quantitative studies usingprogressively smaller bacterial populations will be usefulfor determining the thresholds of antifungal efficacy in lab­oratory coculture and in situ.

The most promising bacteria isolated in this study forpotential development as biocontrol agents were Pseudo­monas species. Of the Pseudomonas strains that were testedfor quantitative activity (Table 3), two strains consistentlyinhibited A. flavus and F. verticillioides growth in both me­dia (Fig. 1). Strain IPlO15, identified as P. chlororaphis,and strain IP2l75, identified as P. fluorescens, produce an­tifungal metabolites and exhibited chitinase activity. P.chlororaphis IPlOl5 also exhibited l3-glucanase activity(Table 3). One distinguishing characteristic of P. chloro­raphis is the production of phenazine-l-carboxamide(PCN), which can be visualized on certain media as a greencrystalline precipitate and we observed in media with strainIPlO15. PCN is essential for the biocontrol activity of P.chlororaphis against tomato root rot caused by Fusariumoxysporum (14). Other antifungal metabolites produced byPseudomonas species include phenazine-l-carboxylic acid(47), 1,3-diacetylphloroglucinol (25), pyrrolnitrin (21), py­oluteorin (22), and hydrogen cyanide (49). Identification ofthe specific antifungal metabolites produced by the Pseu­domonas strains described in this study will aid in deter­mining the mechanisms involved in their activity against A.flavus and F. verticillioides and the. relative importance ofthese mechanisms in the interaction of the bacteria witheach fungus. In situ studies will be essential for determiningthe efficacy of selected Pseudomonas strains for controllingA. flavus and F. verticillioides and thus reducing aflatoxinand fumonisin contamination under field conditions.

ACKNOWLEDGMENTS

We thank Sui-Sheng T. Hua for providing A. flavus strain Papa 827and James L. Baker, Bobbie J. Johnson, and Lorna M. Law for technicalassistance. This work was supported by U.S. Department of AgricultureAgricultural Research Service CRIS project 5325-42000-036-00D.

REFERENCES

1. Abbas, H. K., W. P. Williams, G. L. Windham, H. C. Pringle III, W.Xie, and W. T. Shier. 2002. Aflatoxin and fumonisin contaminationof commercial corn (Zea mays) hybrids in Mississippi. J. Agric.Food Chern. 50:5246-5254.

1620 PALUMBO ET AL. J. Food Prot., Vol. 70, No.7

2. Abbas, H. K, R. M. Zablotowicz, H. A. Bruns, and C. A Abel.2006. Biocontrol of aflatoxin in com by inoculation with non-afla­toxigenic Aspergillus flavus isolates. Biocontrol Sci. Technol. 16:437-449.

3. Ayyadurai, N., P. Ravindra Naik, M. Sreehari Rao, R. Sunish Kumar,S. K Samrat, M. Manohar, and N. Sakthivel. 2006. Isolation andcharacterization of a novel banana rhizosphere bacterium as fungalantagonist and microbial adjuvant in micropropagation of banana. J.Appl. Microbiol. 100:926-937.

4. Bacon, C. W., and P. E. Nelson. 1994. Fumonisin production in comby toxigenic strains of Fusarium moniliforme and Fusarium proli­feratum. J. Food Prot. 57:514-521.

5. Bacon, C. W, I. E. Yates, D. M. Hinton, and E Meredith. 2001.Biological control of Fusarium moniliforme in maize. Environ.Health Perspect. 109(Suppl. 2):325-332.

6. Bakan, B., D. Melcion, D. Richard-Molard, and B. Cahagnier. 2002.Fungal growth and Fusarium mycotoxin content in isogenic tradi­tional maize and genetically modified maize grown in France andSpain. J. Agric. Food Chem. 50:728-731.

7. Bayman, P., and J. Baker. Unpublished data.8. Berg, G., N. Roskot, and K Smalla. 1999. Genotypic and phenotypic

relationships between clinical and environmental isolates of Steno­trophomonas maltophilia. J. Clin. Microbiol. 37:3594-3600.

9. Bevivino, A, V. Peggion, L. Chiarini, S. Tabacchioni, C. Cantale,and C. Dalmastri. 2005. Effect of Fusarium verticillioides on maize­root-associated Burkholderia cenocepacia populations. Res. Micro­bioI. 156:974-983.

10. Bleve, G., E Grieco, G. Cozzi, A Logrieco, and A Visconti. 2006.Isolation of epiphytic yeasts with potential for biocontrol of Asper­gillus carbonarius and A. niger on grape. Int. J. Food Microbiol.108:204-209.

11. Brown, R. L., P. J. Cotty, and T. E. Cleveland. 1991. Reduction inaflatoxin content of maize by atoxigenic strains of Aspergillus flavus.J. Food Prot. 54:623-626.

12. Cavaglieri, L., 1. Orlando, M. 1. Rodriguez, S. Chulze, and M. Etch­everry. 2005. Biocontrol of Bacillus subtilis against Fusarium ver­ticillioides in vitro and at the maize root level. Res. Microbiol. 156:748-754.

13. Cavaglieri, L., A Passone, and M. Etcheverry. 2004. Screening pro­cedures for selecting rhizobacteria with biocontrol effects upon Fu­sarium verticillioides growth and fumonisin Bl production. Res. Mi­crobiol. 155:747-754.

14. Chin-A-Woeng, T E C., G. V. Bloemberg, A J. van der Bij, K M.G. M. van der Drift, 1. Schripsema, B. Kroon, R. J. Scheffer, C.Keel, P. A. H. M. Bakker, H.-V. Tichy, E J. de Bruijn, J. E. Thomas­Oates, and B. J. J. Lugtenberg. 1998. Biocontrol by phenazine-l­carboxamide-producing Pseudomonas chlororaphis PCL1391 of to­mato root rot caused by Fusarium oxysporum f. sp. radicis-lycoper­sici. Mol. Plant Microbe Interact. 11:1069-1077.

15. Cole, J. R., B. Chai, R. 1. Farris, Q. Wang, S. A. Kulam, D. M.McGarrell, G. M. Garrity, and J. M. Tiedje. 2005. The RibosomalDatabase Project (RDP-II): sequences and tools for high-throughputrRNA analysis. Nucleic Acids Res. 33:D294-D296. Available at:http://rdp.cme.msu.edu/index.jsp. Accessed November 2006.

16. Council for Agricultural Science and Technology. 2003. Mycotoxins:risks in plant, animal, and human systems. Council for AgriculturalScience and Technology, Ames, Iowa.

17. D'Mello, J. P. E, A. M. C. Macdonald, D. Postel, W T P. Dijksma,A Dujardin, and C. M. Placinta. 1998. Pesticide use and mycotoxinproduction in Fusarium and Aspergillus phytopathogens. Eur. J.Plant Pathol. 104:741-751.

18. Dorner, J. W, R. 1. Cole, and D. T Wicklow. 1999. Aflatoxin re­duction in corn through field application of competitive fungi. J.Food Prot. 62:650-656.

19. Duvick, J. 2001. Prospects for reducing fumonisin contamination ofmaize through genetic modification. Environ. Health Perspect.109(Suppl. 2):337-342.

20. Gregersen, T 1978. Rapid method for distinction of gram-negativefrom gram-positive bacteria. Eur. J. Appl. Microbiol. Biotechnol. 5:123-127.

21. Howell, C. R., and R. D. Stipanovic. 1979. Control of Rhizoctoniasolani on cotton seedlings with Pseudomonas fluorescens and withan antibiotic produced by the bacterium. Phytopathology 69:480­482.

22. Howell, C. R., and R. D. Stipanovic. 1980. Suppression of Pythiumultimum induced damping off of cotton seedlings by Pseudomonasfluorescens and its antibiotic, pyoluteorin. Phytopathology 70:712­715.

23. Islam, M. T., Y. Hashidoko, A. Deora, T. Ito, and S. Tahara. 2005.Suppression of damping-off disease in host plants by the rhizoplanebacterium Lysobacter sp. strain SB-K88 is linked to plant coloni­zation and antibiosis against soilborne Peronosporomycetes. Appl.Environ. Microbiol. 71:3786-3796.

24. Jochum, C. C., L. E. Osborne, and G. Y. Yuen. 2006. Fusarium headblight biological control with Lysobacter enzymogenes strain C3.BioI. Control 39:336-344.

25. Keel; C., U. Schnider, M. Maurhofer, C. Voisard, J. Laville, U. Bur­ger, P. Wirthner, D. Haas, and G. Defago. 1992. Suppression of rootdiseases by Pseudomonas fluorescens CHAO: importance of the bac­terial secondary metabolite 2,4-diacetylphloroglucinol. Mol. plantMicrobe Interact. 5:4-13.

26. Kerry, B. R. 2000. Rhizosphere interactions and the exploitation ofmicrobial agents for the biological control of plant-parasitic nema­todes. Annu. Rev. Phytopathol. 38:423-441.

27. Kishore, G. K, S. Pande, and A R. Podile. 2005. Biological controlof collar rot disease with broad-spectrum antifungal bacteria asso­ciated with groundnut. Can. J. Microbiol. 51:123-132.

28. Kobayashi, D. Y, and G. Y Yuen. 2005. The role of clp-regulatedfactors in antagonism against Magnaporthe poae and biological con­trol of summer patch disease of Kentucky bluegrass by Lysobacterenzymogenes C3. Can. J. Microbiol. 51:719-723.

29. Lane, D. J. 1991. 16S/23S rRNA sequencing, p. 115-175. In E.Stackebrandt and M. Goodfellow (ed.), Nucleic acid techniques inbacterial systematics. John Wiley & Sons, New York.

30. La Penna, M., A Nesci, and M. Etcheverry. 2004. In vitro studieson the potential for biological control on Aspergillus section Flaviby Kluyveromyces spp. Lett. Appl. Microbiol. 38:257-264.

31. Lingappa, Y, and J. L. Lockwood. 1962. Chitin media for selectiveisolation and culture of Actinomycetes. Phytopathology 52:317-323.

32. Menkir, A, R. L. Brown, R. Bandyopadhyay, Z. Y Chen, and T E.Cleveland. 2006. A USA-Africa collaborative strategy for identify­ing, characterizing, and developing maize gerrnplasm with resistanceto aflatoxin contamination. Mycopathologia 162:225-232.

33. Misaghi, I. J., P. J. Cotty, and D. M. Decianne. 1995. Bacterial an­tagonists of Aspergillus flavus. Biocontrol Sci. Technol. 5:387-392.

34. Munkvold, G. P. 2003. Cultural and genetic approaches to managingmycotoxins in maize. Annu. Rev. Phytopathol. 41:99-116.

35. Nakayama, T, Y. Homrna, Y Hashidoko, J. Mizutani, and S. Tahara.1999. Possible role of xanthobaccins produced by Stenotrophomonassp. strain SB-K88 in suppression of sugar beet damping-off disease.Appl. Environ. Microbiol. 65:4334-4339.

36. Nesci, A. v., R. V. Bluma, and M. G. Etcheverry. 2005. In vitroselection of maize rhizobacteria to study potential biological controlof Aspergillus section Flavi and aflatoxin production. Eur. J. PlantPathol. 113:159-171.

37. Okigbo, R. N. 2005. Biological control of postharvest fungal rot ofyam (Diascarea spp.) with Bacillus subtilis. Mycapathalogia 159:307-314.

38. Palumbo, J. D., J. L. Baker, and N. E. Mahoney. 2006. Isolation ofbacterial antagonists of Aspergillus flavus from almonds. Microb.£Col. 52:45-52.

39. Palumbo, J. D., R. E Sullivan, and D. Y Kobayashi. 2003. Molecularcharacterization and expression in Escherichia coli of three 13-1,3­glucanase genes from Lysobacter enzymagenes strain N4-7. J. Bac­terial. 185:4362-4370.

40. Papa, K E. 1984. Genetics of Aspergillus flavus: linkage of aflatoxinmutants. Can. J. Microbial. 30:68-73.

41. Parke, J. L. 2000. Burkhalderia cepacia: friend or foe? Plant HealthInstr. DOl: 10.1094/PHI-I-2000-0926-01.

42. Parke, J. L., and D. Gurian-Sherman. 2001. Diversity of the Burk-

J. Food Prot., Vol. 70, No.7 BACTERIAL ANTAGONISM OF ASPERGILLUS AND FUSARIUM 1621

holderia cepacia complex and implications for risk assessment ofbiological control strains. Annu. Rev. Phytopathol. 39:225-258.

43. Payne, G. A. 1992. Aflatoxin in maize. Crit. Rev. Plant Sci. 10:423­440.

44. Ramarathnam, R., and D. W. G. Fernando. 2006. Preliminary phe­notypic and molecular screening for potential bacterial biocontrolagents of Leptosphaeria maculans, the blackleg pathogen of canola.Biocontrol Sci. Technol. 16:567-582.

45. Schisler, D. A., N. 1. Khan, and M. J. Boehm. 2002. Biologicalcontrol of Fusarium head blight of wheat and deoxynivalenollevelsin grain via use of microbial antagonists. Adv. Exp. Med. Bioi. 504:53-69.

46. Sharifi Tehrani, A., and M. Ramezani. 2003. Biological control ofFusarium oxysporum, the causal agent of onion wilt by antagonisticbacteria. Commun. Agric. Appl. Bioi. Sci. 68:543-547.

47. Thomashow, L. S., and D. M. Weller. 1988. Role of phenazine an-

tibotic from Pseudomonas fluorescens in biological control of Gaeu­mannomyces graminis var. tritici. J. Bacteriol. 170:3499-3508.

48. Tubajika, K. M., and K. E. Damann. 2001. Sources of resistance toaflatoxin production in maize. J. Agric. Food Chern. 49:2652-2656.

49. Voisard, C., C. Keel, D. Haas, and G. Defago. 1989. Cyanide pro­duction by Pseudomonas fluorescens helps suppress black root rotof tobacco under gnotobiotic conditions. EMBO J. 8:351-358.

50. White, D. G. (ed.). 1999. Compendium of com diseases, 3rd ed.American Phytopathological Society Press, St Paul, Minn.

51. Wicklow, D. T., S. Roth, S. T. Deyrup, and J. B. Gloer. 2005. Aprotective endophyte of maize: Acremonium zeae antibiotics inhib­itory to Aspergillus flavus and Fusarium verticillioides. Mycol. Res.109:610-618.

52. Zhang, Z., and G. Y. Yuen. 1999. Biological control of Bipolarissorokiniana on tall fescue by Stenotrophomonas maltophilia strainC3. Phytopathology 89:817-822.