Biological control of Fusarium moniliforme in maize.

Fusarium moniliforme Sheldon, a biological species of the mating populations within the (italic)Gibberella fujikuroi species complex, i.e., population A [= G. moniliformis (Sheld.) Wineland], is an example of a facultative fungal endophyte. During the biotrophic endophytic association with maize, as well as during saprophytic growth, F. moniliforme produces the fumonisins. The fungus is transmitted vertically and horizontally to the next generation of plants via clonal infection of seeds and plant debris. Horizontal infection is the manner by which this fungus is spread contagiously and through which infection occurs from the outside that can be reduced by application of certain fungicides. The endophytic phase is vertically transmitted. This type infection is important because it is not controlled by seed applications of fungicides, and it remains the reservoir from which infection and toxin biosynthesis takes place in each generation of plants. Thus, vertical transmission of this fungus is just as important as horizontal transmission. A biological control system using an endophytic bacterium, Bacillus subtilis, has been developed that shows great promise for reducing mycotoxin accumulation during the endophytic (vertical transmission) growth phase. Because this bacterium occupies the identical ecological niche within the plant, it is considered an ecological homologue to F. moniliforme, and the inhibitory mechanism, regardless of the mode of action, operates on the competitive exclusion principle. In addition to this bacterium, an isolate of a species of the fungus Trichoderma shows promise in the postharvest control of the growth and toxin accumulation from F. moniliforme on corn in storage.

A biological species of the mating populations of Gibberella fujikuroi species complex, population A Fusarium moniliforme Sheldon (synonym F. verticilliodes; G. moniliforme), produces the fumonisin mycotoxins B 1 and B 2 (FB 1 and FB 2 ) in maize, Zea mays. The symptomless association of fungi of the G. fujikuroi species complex with maize as a descriptive feature was first described by Leonian (1), although its symptomless nature was depicted earlier (2). Numerous descriptions of the symptomless nature of this fungus have been made since these reports (3)(4)(5)(6)(7)(8). These descriptions and others (2,3,5) were highly observational and did not indicate that endophytic hyphae were active physiologically. Indeed, most (9,10) considered these hyphae as latent, quiescent, or dormant and suggested that symptomless infected plants were nonhosts that served the purpose of overwintering the fungus, from which it produced conidia during its saprophytic stage.
The concept of symptomless (latent) infection was very narrowly defined as a quiescent or dormant parasitic relationship that after a time, can become an active parasitic one (10).
The key to the latency concept is the use of parasitic, which was never defined (9) but was used as a synonym for pathogenic. Another aspect of the latent infection concept is that latent infections are different from dormant infections (10). This line of thinking persists and is the essence of recent reviews on the subject of symptomless infections (9,10). Gäumann (11), however, used latent infections to include all stages in the life cycle of a fungus during which no external symptoms can be seen. This wider definition would include actively growing hyphae of a large number of symptomless infections and currently serves as a descriptive basis for fungal endophytes. However, symptomless infections have been redefined and are considered biologically relevant (8)(9)(10)(11)(12)(13)(14)(15). Before we focus on controls of endophytic fungi, we must first define endophytic microorganisms, which will justify the control strategies discussed in this review.
What are microbial endophytes? We refer to some fungi and bacteria as endophytes because they actively colonize host tissues and establish long-term associations-actually, lifelong symptomless associations-without doing substantive harm. Excluded from this characaterization of endophytes are localized cuticular infections and very localized epibiotic infections. The definition of endophytic fungi and bacteria used here is based on earlier concepts of microbial endophytes as symptomless infections, although there are also numerous exceptions. (12). We extend our definition of this habit from several well-established endophytic systems of grasses where the association is also used to include the mutualistic symbiotic associations, e.g., the tall fescueendophytic fungus associations, which are not pathogenic, although exceptions occur here as well (12)(13)(14)(15)(16).
Thus, we extend our definition of endophyte to include those inconspicuous intercellular infections that are at least transiently symptomless but that are functionally relevant to the association as a viable, growing, and biochemically important component. This distinction indicates that endophytism is both a life habit and a biological association with ecological and physiological relevance. Further, latent or dormant infections are not characteristic features of endophytic hypha, although latent and dead hyphae and cells due to the aging process are undoubtedly necessary components of endophytic microorganisms. Microbial endophytes are associated with a large number of plants (12,(17)(18)(19), and the occurrence of F. moniliforme as an endophyte is typical of several species within the genus Fusarium; for example, see the review by Kaldau and Yates (20).
Control, prevention, and detection of the endophytic infections by F. moniliforme in corn are difficult, especially because kernels appear to be of sound quality. The intercellular nature of this endophyte makes chemical control highly unlikely. Applications of systemic fungicides are impossible during later stages of plant growth, and because the fungus is a systemic seed-borne infection, conventional fungicides as seed treatments are also ineffective.
Because microbial endophytes can be important mutualistic components of plants (12)(13)(14)(15)(16)(17)(18)(19), we propose that alternative Fusarium moniliforme Sheldon, a biological species of the mating populations within the Gibberella fujikuroi species complex, i.e., population A [= G. moniliformis (Sheld.) Wineland], is an example of a facultative fungal endophyte. During the biotrophic endophytic association with maize, as well as during saprophytic growth, F. moniliforme produces the fumonisins. The fungus is transmitted vertically and horizontally to the next generation of plants via clonal infection of seeds and plant debris. Horizontal infection is the manner by which this fungus is spread contagiously and through which infection occurs from the outside that can be reduced by application of certain fungicides. The endophytic phase is vertically transmitted. This type infection is important because it is not controlled by seed applications of fungicides, and it remains the reservoir from which infection and toxin biosynthesis takes place in each generation of plants. Thus, vertical transmission of this fungus is just as important as horizontal transmission. A biological control system using an endophytic bacterium, Bacillus subtilis, has been developed that shows great promise for reducing mycotoxin accumulation during the endophytic (vertical transmission) growth phase. Because this bacterium occupies the identical ecological niche within the plant, it is considered an ecological homologue to F. moniliforme, and the inhibitory mechanism, regardless of the mode of action, operates on the competitive exclusion principle. In addition to this bacterium, an isolate of a species of the fungus Trichoderma shows promise in the postharvest control of the growth and toxin accumulation from F. moniliforme on corn in storage. endophytes might offer some control of the endophytic growth of F. moniliforme by competitive exclusionary principles. We review information on the endophytic fungus habit, which defines the requirements necessary for biological control and establishes production of fumonisins by endophytic hyphae of F. moniliforme. Second, we review biological control strategies that offer promise for both fungus reduction and toxin reduction. The first biocontrol uses an isolate of an endophytic bacterium, Bacillus subtilis (Ehrenberg) Cohn, that shows great promise in the control and reduction of mycotoxin accumulation during the growth of maize plants endophytically infected with F. moniliforme. Because F. moniliforme is also saprophytic, we review information on a second biocontrol that uses an isolate of the fungus Trichoderma sp. as a postharvest control for the growth of F. moniliforme and resulting toxin accumulation during storage.

Experimental Microorganisms and Approaches
Experiments designed to determine the production of the fumonisins by endophytic hyphae as well as biocontrols of fungal growth and toxin production have been reported. Experimental details for their use, laboratory manipulation, toxin analysis, and plant interaction are contained therein (8,(21)(22)(23)(24)(25)(26)(27).

Microorganisms
All F. moniliforme isolates were selected for unique traits and have been described (21,23,24). Briefly, we used specific wild type (WT) strains of F. moniliforme: RRC410, RRCPAT, RRC408, and MRC 826. These WT strains were transformed with the gusA reporter gene, which encodes for β-glucoronidase (GUS), and the hph gene for hygromycin resistance (21). The transformed strains were used as ecological markers and were designated MRC 826gus, RRCPATgus, and RRC408gus. A positive reaction for GUS activity was detected by a specific histochemical stain, which confirmed that the specific strain recovered was the one inoculated into plant material initially (21). The isolate UPS101, a Trichoderma species, was recovered from root segments collected from Silver Queen plants grown in field plots (22). The maize cultivars used included yellow field maize Trucker's Favorite and Reid's Yellow Dent and sweet maize Silver Queen and Early Sunglow.
The endophytic bacterium was isolated from an unknown cultivar of maize from northern Italy (23). Using biochemical characteristics and specific fatty acids, this bacterium was initially identified as an aberrant type of Enterobacter cloacae (23); but using the 16S rRNA gene sequence similarity analysis, this bacterium matches the genetic distance format for strains within the B. subtilis group. However, the percent difference between this isolate and the closest match suggests a subclad, which indicates that we might be dealing with a subspecies or another closely related and/or unknown species within the diverse B. subtilis group. If so, this isolate would more than likely be a new species separated from the B. subtilis complex that is composed of two groups of closely related but different strains. The strains of this endophytic bacterium used in these studies included the WT B. subtilis RRC101 and its rifampicin-resistant mutants RRC26ss and RRC24wf. The mutants are identical in inhibitory activity and endophytic competence to the WT and were used as markers for field studies.

Plant Culture and Endophytic Studies
Experiments were conducted under gnotobiotic conditions (24) when the plants were young, vigorous, and showed less than 1% senescent leaves. The latter aspect was important because one could argue that plants might also be infected from outside sources, producing saprophytic production of fumonisin on senescent tissue. All experiments designed to study the fungus/bacterial/host plant relationship were performed on maize kernels subjected to double sterilization: The kernels were surfaced sterilized and then subjected to a mild heat treatment (24) to remove both external and internal bacteria and fungi. However, since gnotobiotic conditions can only be maintained for approximately 10 weeks, experiments to test for endophytic fumonisin production were conducted within this time period (28,29).

The Nature of the Endophytic State of F. moniliforme
As an endophyte of maize, F. moniliforme may remain a symptomless biotrophic parasite throughout the entire growing season (4,7,8,30). Some isolates enhanced the growth of maize seedlings (12,31). The fungus exists as an intercellular infection within the plant, although under certain circumstances it may become intracellular (8). The fungus endophytically infects seedling from its systemic infection of the seed, usually by the second day following germination (8), or within 10 days ( Figures 1A-1E) (32). The systemic infection of maize seeds produces maternalline vertical transmission from generation to generation. In addition to the vertical transmission phase, maize is subject to infection throughout the growing season (33,34). The fungus is also a saprophyte while residing in the soil, living off dead maize debris where spores are produced that can infect plants from the outside, e.g., maize silks. Thus, F. moniliforme is vertically transmitted to the next generation of plants via clonal infection of seeds, and horizontally transmitted to plants contagiously from saprophytic colonization of soil debris and insect vectors. The vertical dissemination phase of this fungus as an endophyte of maize kernels indicates that the relationship is not casual, but rather intimate and probably complex, since seed germination and seedling development usually are not affected (1,7,8,30,35). The horizontal infection phase can be reduced or eliminated by application of certain fungicides. The endophytic seed and plant infection, i.e., vertical transmission, is important because it remains the reservoir from which infection of each generation of plants takes place and from which renewed toxin synthesis can take place. Application of fungicides cannot control the fungus during this phase.
To understand the relationship of the host with the fungus, symptomless infections were studied at the light and ultrastructural levels (8). The two hygromycin-resistant mutants of F. moniliforme, RRC374H and MRC826H, used to examine the host-parasite relationships produced either symptoms of seedling blight or symptomless infections in two maize cultivars, Silver Queen and Truckers' Favorite. Intercellular hyphae were observed within asymptomatic tissue of leaves and roots of 1to 8-week-old plants on both cultivars ( Figures 1A-1E). Symptomless endophytic infections were observed in several cultivar-isolate combinations. Intercellular hyphae were in direct contact with host cell walls (8); at the ultrastructural level, the walls were not separated by an extracellular granular matrix reported for other intercellular (or epibiotic) fungal pathogens and mutualists (36)(37)(38). Throughout the entire period of host colonization by the symptomless fungalmaize combination, there were no effects from the intercellular hyphae on the integrity of host cellular structure and organelles. This suggests that during the symptomless state no toxins are produced by the fungus that are disruptive to the host cellular structure, and that no host reaction indicates no disease is produced. The symptomless infection occurred identically in both cultivars. Symptomless infections have been produced in 15 cultivar and inbred lines of maize, and identical ultrastructural responses were observed (data not shown).
Because F. moniliforme could be recovered from 2-day-old coleoptiles, infection must have occurred immediately before or during germination. The fungi were isolated from leaves, stems, hypocotyls, and roots during the 8-to 10-week observation period, and included both diseased and symptomless plants of the Silver Queen cultivar (Table 1). Identical results were obtained with the cultivar Trucker's Favorite. Symptomless plants (95% of the treatment group) were positive for fungi throughout the observation period; diseased plants (3%-5%) died within 2 to 3 weeks, and 1% of the diseased plants recovered and grew symptomless. The distribution of F. moniliforme was the same in both diseased and symptomless fungal-seedling combinations. Thus, the lack of disease expression was not caused by the lack of fungi within specific organs of a maize genotype. We found that the fungus was not present within the vascular bundle tissue of the pedicel (24), nor did it invade the vascular bundle of plants, which supports the conclusion of Pennypacker (39) that F. moniliforme is not a vascular rot or wilt pathogen of maize.
The topical application (8,24) of the hygromycin-resistant mutants of F. moniliforme, RRC374H and MRC826H, to maize kernels and their recovery from gnotobiotic plants 6 weeks later (Table 1) indicates the importance of infected kernels both for dissemination and as a source of inoculum development. Further, surface-sterilized kernels known to be systemically infected naturally yielded F. moniliforme, which indicates that the systemically infected kernels are also important inoculum sources for the endophytic infections (data not shown).
Valleau (2), Thomas and Buddenhagen (40), and Kedera et al. (7) suggest that infection of maize can occur from kernel to kernel. To establish vertical transmission of this fungus, F. moniliforme was transformed with the GUS-reported gene (21). This isolate was applied to seed, which was planted under greenhouse conditions. The resulting kernels produced from plants under these conditions were surface-disinfected and the isolated fungus stained for specific GUS-specific products (21). The results indicated that approximately 35% of the Fusarium-infected kernels were infected with the transformed isolate. Further, when these kernels were planted and plants were grown under gnotobiotic conditions, the transformed fungus was isolated from seedling tissue. We interpret these data to mean that the fungus was transmitted from seed to plant and to seed again. This aspect of the life cycle of F. moniliforme indicates that it can be intimately associated with the maternal lineages of maize cultivars.
F. moniliforme infects maize very soon after germination, and, contrary to earlier views (3,30,32), this infection does not always develop into a disease. The initiation of the disease phase of the parasitic stage is not common, and may reflect plant selection against this phase (41)(42)(43), a physiological interaction of the fungus with phytoanticipins (44), or interactions with suitable environmental conditions, often late in the life of the plant (43)(44)(45)(46).

Toxin Production by Endophytic Hyphae
It is assumed that the fumonisins are synthesized late in the plant-fungus interaction or during saprophytic fungal growth on damaged or dead tissue. Very little of this type of corn enters the human food and animal feed chains, although subsequent improper storage conditions can increase the fumonisin content. We do not know how early this occurs or what conditions affect the accumulation of the fumonisins by the endophytic symptomless phase. The presence of the fumonisins in sound-appearing, food-grade maize kernels (47,48) supports the hypothesis that low concentrations of the fumonisins are synthesized by symptomless endophytic hyphae. This hypothesis would also account for the production of the fumonisins in higher concentrations in sound corn when stored under poor postharvest conditions or in kernels and debris from cobs and in other plant parts not showing ear and stalk rot symptoms (48,49).
To test the hypothesis that fumonisins are produced by endophytic hyphae, corn seedlings were infected with F. moniliforme (8), grown under gnotobiotic conditions for Environmental Health Perspectives • VOLUME 109 | SUPPLEMENT 2 | May 2001  10 weeks, and harvested and separated into roots and shoots; this material was then analyzed for fumonisin content (27). This experiment showed not only that endophytic hyphae of F. moniliforme produced the fumonisin in planta but also that this mycotoxin is produced early in maize seedlings ( Figure 2). Thus, this association is another example of toxin production by endophytic hyphae, and in this regard is similar to other fungal endophytes of grasses (12,14). Additional experiments used GUS-transformed F. moniliforme-infected plants subjected to 2 weeks of drought and compared to normal watering conditions (21). The results indicated that endophytic hyphae of seedling are responsive to this abiotic stress (Figure 2). The fumonisins accumulated under drought stress, especially at -1.5 megapascals, which induced physiological wilt (Figure 2). A similar increase in fumonisin content was observed in plants inundated with water, simulating flooding (data not shown). Apparently it is stress and not necessarily drought that induces an increase in toxin synthesis. The content of fumonisins is significantly higher in roots than in shoots (data not shown), which probably reflects the preferential transport of translocates and apoplastic solutes to roots during this period (50).
These experiments prove that the symptomless endophytic stage of F. moniliforme is physiologically active. Further, endophytic hyphae isolated from treatment groups showed GUS activity following a specific staining procedure (21). Although we have no information dealing with plants grown for a longer time period, the accumulation and/or synthesis of toxin might resume as kernel development is initiated because products stored in the roots move upward and into development kernels at this time. We have grown plants under greenhouse conditions from kernels inoculated with F. moniliforme RRC408gus and recovered this strain from surface-disinfected matured kernels (an average of 40%) as well as all plant parts, establishing the vertical transmission of the fungus.

Plant-Bacterium Interaction
Species of Bacillus are characterized as grampositive, rod-shaped, aerobic, catalase-positive, motile cells that form endospores more resistant than vegetative cells to heat, drying, and other destructive environmental factors (51). B. subtilis is a large and diverse physiological group of nonpathogenic bacteria that is relatively easy to manipulate by genetics and simple to cultivate, which strengthens its suitability for industrial uses. Further, the presence of endospores is an advantage for most industrial applications and biotechnological uses. These bacteria and other members of the genus are used in fermentations for human food use, sources of extracellular enzymes for industrial and medicinal uses, and production of peptide antibiotics. The biocontrol bacterium B. subtilis RRC101 was shown to have highly desirable plantenhancing characteristics as well as being an endophyte able to protect maize from pathogenic organisms (23,50). B. subtilis RRC101 is unique among other biocontrol isolates of this species in being endophytic, and this biocontrol isolate was patented for these properties: patent number 5,994,117 (52).
The biocontrol bacterium was obtained from seedling roots grown from endophytically infected seed following culture on agar media, soil, or filter paper. When sections of these roots were examined microscopically, bacterial cells were also observed internally ( Figure 3A). On maize roots obtained from seeds germinated in culture dishes, the bacteria were distributed uniformly over the epidermis ( Figure 3B) and randomly distributed intercellularly in several locations of the cortex ( Figure 3B and 3C). No bacterial cells were observed within the endodermis, but bacteria were observed intercellularly within the outer margin of the pericycle, usually adjacent to phloem cells ( Figure 3D). They were not observed among cells of the pith area. In no instance was there evidence of damage to host cells, and cells of this bacterium were completely intercellular ( Figure 3C).
The isolates of B. subtilis used for in planta interactions were rifampicin mutants (23). There was no evidence at the ultrastructural level of pathological damage to cells of the endodermis and stele, nor were any bacteria observed within cells. A matrixlike capsule was observed surrounding the bacterial cells located on the primary root epidermis (23). The intercellular location of this bacterium was clearly demonstrated at the ultrastructural level. There is a proliferation of bacterial cells within the intercellular spaces, particularly those that were connected ( Figure 3C and Figure 3D). The identical intercellular nature of this bacterium was observed in the field maize cultivar Trucker's Favorite as well as all other inbred lines and cultivars (data not shown).

Bacteria-Fungus Antagonism in Vitro
Inhibitory studies (23) between the bacterium and fungi were conducted on either nutrient agar or potato dextrose agar (PDA), and these zones of inhibition were used as bioassays for antagonisms. Besides the isolates of F. moniliforme listed above, 30 others were also tested (data not shown). All isolates were grown on PDA for 3-7 days, two mycelial plugs removed from the outer margins of each fungal species, and two plugs placed opposite each other on the outer margin of a fresh agar plate of either nutrient agar or PDA (23). The bacteria used for these experiments were grown on nutrient agar for 2-3 days, and one inoculation loop was streaked down the center of each plate, between each fungal agar plug of each plate, immediately after inoculation with the fungal plugs. Control plates consisted of fungi or bacteria placed on plates alone or fungi as described above but without the bacteria. All plates were incubated in the dark at 25-27°C, until the fungi on the control plates had grown together. The bacterium completely inhibited the growth of all isolates of F. moniliforme tested (23).

B. subtilis in Planta Distribution
The distribution of bacteria within the aboveground portions of plants was determined by aseptically culturing maize seedlings produced from double sterilized kernels (24). The seedlings were grown for 1-6 weeks in the light room under gnotobiotic conditions as described above. Portions of plants were surface sterilized (23) and plated out on rifampicin-amended nutrient agar. The controls-uninoculated double-sterilized kernels germinated and grown as described for the treatment groups-did not yield bacterial colonies when their leaves and stems were surfaced sterilized and plated on rifampicinamended medium. The bacterium was distributed in all organs of the plant at concentrations that varied during the entire season ( Table 2). The highest concentration of the bacterium occurred in the roots [10 6 colony forming units (cfu)/g wet weight], and this concentration remained constant (data not shown). Further, the bacterium was  absent from maize kernels harvested at 14% moisture. Some cultivars yielded bacteria, but this was from less than 10% of kernels. Bacteria-infected maize plants showed no evidence of disease in seedlings during a 6-week observation period under gnotobiotic conditions nor during the entire maize growth cycle under field conditions of approximately 120 days. The bacterium had positive effects on maize seedlings including increased root and shoot growth (Table 3) and an increase in the germination rate (data not shown) (53). Further, the bacterium has a protective action on maize seedlings when planted in soils infested with F. moniliforme ( Table 4). The bacterium also caused an increase in herbage yield and leaf width when coinfected with F. moniliforme (Table 4). These data indicate that the bacterium protects the seedling and enhances growth when planted in F. moniliforme-infested soil. A comparison of B. subtilis RRC101 with other isolates of B. subtilis established that several strains can colonize roots (54). RRC101 prevented roots from being colonized by F. moniliforme in the two soil types tested (Figure 4) (54), indicating that these other isolates of B. subtilis cannot prevent the colonization of plants by the fungus (54). Further, the data show that the growth of the fungus was controlled in either a natural soil type (Cecil) or a synthetic soil ( Figure 5). Although these nonbiocontrol isolates were able to colonize the roots, their duration of endophytic colonization declined during a 2month observation period (54). Thus, B. subtilis RRC101 grows continuously in planta without negatively affecting the growth of maize and is found in relatively uniform concentrations (10 5-6 cfu/g tissue, fresh weight) throughout the season.
Maize inoculated with B. subtilis germinated and when the resulting seedlings were grown on sterile filter paper, distribution and location of bacteria were observed identical to those on the medium, suggesting that the initial observations on PDA were not artifacts. Similarly, the identical endophytic habit of this bacterium was established from roots of kernels germinated in soil. The presence of the bacterium only between cells and in young, healthy-appearing, developing seedling roots suggests that this relationship is not detrimental; thus, we do not interpret it as a stage in the decomposition of roots from the surface inward. We conclude from the microscopic data that this bacterium is intercellular,   whereas the symptomless and host-enhanced characteristics clearly define this bacterium as a true mutualistic endophyte as described above.

Fumonisin Reduction by B. subtilis
The effects of the WT B. subtilis Pat. No. 5,994,117 (RRC101) and its rifampicin mutants 26ss and 24wf on fumonisin production were determined during seedling growth. These plants were grown in pot culture in the light room, subjected to drought treatments as described above, and analyzed for fumonisin content. Not only did the bacterium prevent the fungus from colonizing the plants, but it also reduced the amounts of fumonisins produced in planta by an average of 50% when plants were grown under the normal watering (nonstressful) regiment (controls, Figure 6). Amounts of fumonisin were significantly reduced throughout the treatments but especially under drought stress, a condition conducive to high accumulation of fumonisin ( Figure 6). The amount of fumonisin produced under physiological drought was reduced significantly compared to that in the noninfected B. subtilis group, to the level of the control nonstressed treatment ( Figure 6). These experiments indicated that B. subtilis prevented the stress-induced increase of fumonisin produced by endophytic hyphae observed in the plants not protected by B. subtilis.
Transformed F. moniliforme RRC408gus was inoculated on bacteria-inoculated maize kernels of field and sweet cultivars cultured in pots located in either the greenhouse or outside plots. The plants were allowed to mature and set seed. Roots, stem, leaves, and kernels were collected and analyzed for bacteria, fungi, and fumonisin. The data ( Figure 6) indicate that this bacterium-reduced fumonisin produced under stress in planta and suggest that this reduction will continue in matured plants at the kernel fill stage (53). The effects of this bacterium on fumonisin accumulation are continuing to be studied under field conditions in two locations. Preliminary data indicate that the bacterium remains within the plants during the entire season and can be recovered to a low extent in kernels, especially from one maize cultivar. Analysis of toxin production under field conditions is still in the preliminary stage, consisting of one replication for a single season.

Biological Control with Trichoderma sp.
Species of Trichoderma are common soil saprophytic hyphomycetes found in all climates throughout the world. The genus is complex and polyphyletic and is separated by morphological criteria into five sections or species aggregates, one of which is Trichoderma Bissett section Trichoderma. It is to this section that our biocontrol species belongs. Although the identity has not been conclusively established using molecular characters, the isolate has oblong conidia occurring on conspicuous whorled phialides with no yellow pigment produced on the reverse of a diagnostic medium. The appearance of whorled phialides and the absence of yellow pigment distinguishes this species from all others, indicating that this isolate is tentatively T. koningii Oudemans (= Hypocrea koningii Lieckfeldt, Samuels and Gams, the teleomorph) (55). However, we will refer to it as Trichoderma sp. until it has been subjected to molecular analysis (56). Members of this genus have been studied as antagonists in biocontrol systems against various plant pathogens and have been considered sources for numerous biotechnological applications for decades (57). These fungi are very effective as biocontrol agents because their powerful extracellular lytic enzymes produce necrotrophic action on fungi through lysis of cell walls. Much of the biocontrol potential of Trichoderma sp. has been reviewed (58,59) and most of the mode of action is probably caused by the lytic activity on fungal cell walls. Plant pathogens controlled by Trichoderma species include Gaeumannomyces graminis, Phythium sp., Botrytis cinerea, Colletotrichum truncatum, Cylindrocladium floridanum, Phytophthora citrophthora, Sclerotinia sclerotiorum, Botrytis cinerea, and numerous other fungal pathogens of economically important plants (58).
A few investigations have reported the potential of Trichoderma species as control organisms for phytopathogenic Fusarium, including competition of Trichoderma sp. and F. moniliforme in vitro (60). However, these investigators did not examine whether such competition occurs on natural substrates such as maize kernels nor whether mycotoxin production is reduced. Trichoderma species are applied mainly to the soil for biocontrols, and only a few reports deal with their application in the management of postharvest cereal and foliage diseases (58). Our objective was to investigate the influence of a strain of Trichoderma species, isolated from maize root segments, on the growth and production of FB 1 by F. moniliforme growing on maize kernels and its potential as a postharvest control for fumonisin production and/or reduction.

Effects of Trichoderma on Growth of F. moniliforme
In addition to reduction of fumonisin production, daily measurements of F. moniliforme colony diameter alone and in dual cultures with the isolate of the Trichoderma species showed antagonism to F. moniliforme (22). Radial colony extension of F. moniliforme cultured alone continued to increase throughout a 14-day measurement period on laboratory media (22). However, radial extension of F. moniliforme       Table 5).
Samples of F. moniliforme RRCPATgusinoculated maize kernels were harvested after only 1 week incubation to compare FB 1 production during this time with that in maize kernels to which Trichoderma sp. was added to 1-week-old cultures of F. moniliforme. The objective of incubating F. moniliforme for a week before adding Trichoderma sp. was to determine the efficacy of Trichoderma sp. in suppressing FB 1 production in maize kernels already contaminated with F. moniliforme.
The FB 1 concentration was 20 µg/g maize for kernels harvested after 7 days following F. moniliforme inoculation. Thus, the Trichoderma sp. did not immediately inhibit FB 1 production but did significantly reduce FB 1 production in comparison to cultures with F. moniliforme alone (Table 5). FB 1 production by the WT and GUS-transformed F. moniliforme incubated on maize kernels for 12 weeks was not significantly different. Furthermore, these strains did not differ in sensitivity to the isolate of the Trichoderma sp. (data not shown). The FB 1 produced during 12 weeks co-incubation of Trichoderma sp. with the fungi was reduced by > 80%. Thus, the insertion of the foreign genes for hygromycin resistance and GUS synthesis into F. moniliforme RRCPATgus did not influence FB 1 production or sensitivity to Trichoderma sp. in comparison to the parental WT, RRCPATwt. FB 1 accumulation was reduced even in maize kernels on which F. moniliforme PATgus had been growing 7 days before the addition of the Trichoderma isolate (Table 5). FB 1 production was 64 µg/g of kernels with this inoculation procedure, a 72% reduction in fumonisin compared to that produced by F. moniliforme RRCPATgus growing alone (206 µg/g dry weight).
The current results provide the first evidence for activity of a species of Trichoderma as a suppressor of toxin synthesis. These results also support earlier reports that certain strains of Trichoderma sp. inhibit F. moniliforme growth (60). Our isolate of Trichoderma sp. suppressed F. moniliforme growth by 46% after 6 days (22) in comparison to the 10% suppression by strains described earlier (53). We have demonstrated that this Trichoderma sp. grew on maize kernels and in the process reduced FB 1 production. Even in the presence of Trichoderma sp., FB 1 content was above the maximum limit recommended by the American Association of Veterinary Laboratory Diagnosticians for fumonisins in feed for horses at 5 µg/g; pigs at 10 µg/g; and both beef cattle and poultry at 50 µg/g (63). However, under natural conditions, fumonisin content in the F. moniliforme -infected kernel would not be expected to reach the levels used for our experimental analyses.
Our studies demonstrated that this Trichoderma sp. meets several criteria essential for an effective biocontrol agent (64). One feature is that the biocontrol agent must colonize the substrate or plant part targeted by the pathogenic organism. The isolate of the Trichoderma sp. grew on maize kernels, the part of the maize plant most commonly associated with the fumonisins in causing harmful effects on animal and human health. The cocultivation of F. moniliforme and the isolate of Trichoderma sp. on maize kernels fulfilled another criterion in that the biocontrol agent must be active under environmental conditions such as pH and temperature, so that growth of the biocontrol agent and antagonist coincide. Another criterion is that the biocontrol agent must be compatible with other control procedures. The interactions of the isolate of this Trichoderma sp. with other pest control practices and ecosystem diversifications have not been analyzed. However, previous studies have indicated that species of Trichoderma may be more effective in certain ecological niches, such as specific soil types.
The primary purpose of this biocontrol organism is as a postharvest control in kernels in storage. Further, the biocontrol potential for this isolate is more suited for toxin reduction in maize kernels intended for animal feed. The successful activity of the fungus on kernels in storage will depend on air, moisture, and temperature requirements for this isolate. Thus, many more ecological parameters must be determined for this isolate of Trichoderma sp. before conclusions can be reached regarding the conditions under which this strain could function as an effective biocontrol agent for F. moniliforme.

Summary
The endophytic hyphae of F. moniliforme are neither latent nor dormant but are important for the vertical transmission of the fungus from generation to generation through seed, and it serves as a source of fumonisin in planta. The endophytic hyphae are neither latent nor dormant but instead physiologically Environmental Health Perspectives • VOLUME 109 | SUPPLEMENT 2 | May 2001  (22). b Letters not common for FB 1 (µg/g) concentrations indicate the means are significantly different by ANOVA, means separation by LSD (p < 0.05), and numbers in parentheses are ± SD of the means. c Reduction expressed as the percentage FB 1 produced by either F. moniliforme PATwt or PATgus after 12-week incubation. d Harvested after 7 days incubation; all other treatments were harvested after 12 weeks. VOLUME 109 | SUPPLEMENT 2 | May 2001 • Environmental Health Perspectives active. Using transformed fungi, we demonstrated the maternal transmission of this fungus from seed to plant and back to seed under gnotobiotic conditions. The results indicated that during the life cycle of the fungus within its host tissue, the endophytic hyphae of the fungus responds to abiotic stresses imposed on the plant such as drought and flooding. Under these two stresses synthesis of the fumonisins increases. Plants maintained under uniform moisture-i.e., the controls-produce very little fumonisin. An endophytic bacterium and a fungus are being tested as pre-and postharvest biocontrols of fumonisin accumulation, respectively. The bacterium, a strain from a subgroup of B. subtilis, operates under the principle of competitive exclusion, reducing the fumonisin concentration in planta. The fungus Trichoderma sp. is being tested for postharvest control of fumonisin production in kernels during storage. The use of Trichoderma as a biocontrol is expected to have application for maize kernels especially for animal feed. We suggest that both organisms have high potential for controlling not only F. moniliforme but also several fungal diseases of maize. However, both biocontrol organisms still have ecological parameters that must be determined before their use in controlling growth and reducing fumonisin production by F. moniliforme can be commercialized.