BIOPROSPECTING ENDOPHYTIC BACTERIA FOR BIOLOGICAL CONTROL OF COFFEE LEAF RUST

Suppression of plant diseases due to the action of endophytic microorganisms has been demonstrated in several pathosystems. Experiments under controlled conditions involving endophytic bacteria isolated from leaves and branches of Coffea arabica L and Coffea robusta L were conducted with the objective of evaluating the inhibition of germination of Hemileia vastatrix Berk. & Br., race II, urediniospores and the control of coffee leaf rust development in tests with leaf discs, detached leaves, and on potted seedling of cv. Mundo Novo. The endophytic bacterial isolates tested proved to be effective in inhibiting urediniospore germination and/or rust development, with values above 50%, although the results obtained in urediniospore germination tests were inferior to the treatment with fungicide propiconazole. Endophytic isolates TG4-Ia, TF2-IIc, TF9-Ia, TG11-IIa, and TF7-IIa, demonstrated better coffee leaf rust control in leaf discs, detached leaves, and coffee plant tests. The endophytic isolates TG4-Ia and TF9-Ia were identified as Bacillus lentimorbus Dutky and Bacillus cereus Frank. & Frank., respectively. Some endophytic bacterial isolates were effective in controlling the coffee leaf rust, although some increased the severity of the disease. Even though a relatively small number of endophytic bacteria were tested, promising results were obtained regarding the efficiency of coffee leaf rust biocontrol. These selected agents appears to be an alternative for future replacement of chemical fungicide.


INTRODUCTION
Beneficial endophytic microorganisms comprise especially fungi and bacteria that colonize internal plant tissues without causing visible damage to their hosts (Petrini, 1991). They are different from phytopathogenic microorganisms because they are not detrimental, do not cause diseases to plants, and are distinct from epiphytic microorganisms which live on the surface of plant organs and tissues (Hallmann et  1997). Endophytic bacteria are able to penetrate and become systemically disseminated in the host plant, actively colonizing the apoplast (Quadt- Hallmann et al., 1997b), conducting vessels (Hallmann et al., 1997), and occasionally the intracellular spaces (Quadt-Hallmann et al., 1997a). This colonization presents an ecological niche, similar to that occupied by plant pathogens, and this endophytic bacteria can, therefore, act as biological control agents against pathogens (Hallmann et al., 1997).
In this sense, the suppression of plant diseases due to the action of endophytic microorganisms has been demonstrated in several pathosystems (Narisawa et al., 1998;Lima et al., 1994). Several mechanisms may control this suppression, either directly on the pathogen inside the plant by antibiosis (Sturz et al., 1998) and competition for nutrients (Mari et al., 1996), or indirectly by induction of plant resistance response (M'Piga et al., 1997).
The coffee leaf rust caused by Hemileia vastatrix is the main disease in coffee, causing yield losses of 35 to 40%, on average. Control is basically achieved by fungicides. In 2000, in Brazil, the use of fungicides in coffee stood for 3,680 t of active ingredient (Campanhola & Bettiol, 2003). Therefore, alternatives to control coffee leaf rust must be sought. The objective of this work was to select endophytic bacteria isolates from coffee leaves and branches, with biocontrol potential against coffee leaf rust, by means of inhibition assays of urediniospores germination and control of coffee leaf rust in tests with leaf discs, detached leaves, and on potted seedling of cv Mundo Novo.

Endophytic bacteria
Isolates from leaves and branches (Nunes, 2004), of Coffea arabica and Coffea robusta plants from Pedreira, Mococa, and Pindorama counties, State of São Paulo, Brazil (Table 1), were maintained in the culture colletion of the Laboratório de Microbiologia Ambiental, Embrapa Meio Ambiente, in sterile distilled water (Castellani, 1967). Forty bacterial isolates were evaluated regarding their capacity to inhibit the germination of H. vastatrix urediniospores, and 44 isolates were used to control coffee leaf rust in leaf discs, detached leaves, and seedlings of C. arabica, cv. Mundo Novo.

Urediniospore germination
The endophytic bacteria isolates were cultivated on nutrient agar medium (Peptone 5 g; meat extract 3 g; agar 15 g; distilled water 1000 mL) for 24 hours at 28 ± 2ºC, and then transferred with to slants containing sterilized, distilled water. Samples were then shaken vigorously to obtain a homogeneous cell suspension, which was standardized to an optical density of A 550 = 0.1. Urediniospores of H. vastatrix, race II, were collected from coffee leaves containing lesions obtained from plants in "Centro de Café e Plantas Tropicais, Instituto Agronômico de Campinas", and stored in a container with sodium dichromate (relative humidity 52%; 7 ± 2ºC). Urediniospores were suspended in water at a concentration of 1.0 mg mL -1 using a magnetic stirrer for 5 minutes. A 15.0 μL aliquot of this suspension, and a 15.0 μL aliquot of the endophytic bacteria suspension were then transferred to microscope slide, mixed and enclosed within plastic boxes containing a layer of foam saturated with water, and sealed with glass plates to maintain high relative humidity. After incubation for six hours (22 ± 1ºC) in the dark, the germination was interrupted by adding 15.0 μL of lactophenol cotton blue dye onto each droplet, and examined under light microscope. The percentage of germinated urediniospores (10 fields at 200 × magnification) was calculated. Urediniospores with germ tubes of at least one half of the length of their larger diameter were considered germinated. Trials were set up in a completely randomized experimental design (n = 4). The experiment was repeated and the means were used for statistical analysis. Sterilized water and propiconazole (Tilt ® CE; 1.2 μL of the commercial product per mL of water) were used as controls.

Leaf discs
Discs of young and completely developed leaves of C. arabica cv. Mundo Novo plants were removed with a 2.0 cm diameter cork punch and placed into plastic boxes, abaxial surface facing up, over a layer of foam saturated with water (Eskes, 1989). Using a micro-pipet, 25.0 μL of the endophytic bacteria suspension were applied on the leaf discs, 72 and 24 hours before, after, and simultaneously with the same volume of H. vastatrix urediniospores suspension (1.0 mg mL -1 ). After inoculation, boxes were covered with glass plates and incubated in the dark for 24 hours. Then the boxes were maintained under 12h photoperiod, 500-1000 lux, 22 ± 2ºC, and approximately 100% relative humidity. The experiment was set on completely randomized design (n = 3), represented by nine leaf discs each. Severity of the disease was evaluated 30 days after inoculation, using a rating scale from 1 to 5, according to the percentage of leaf area with lesions (1 = 0%; 2 = 1-25%; 3 = 26-50%; 4 = 51-75%; and 5 ≥ 75% of leaf area with lesions). The Waller-Duncan (α = 0.05) test was used to compare the lowest and the highest mean values of lesions percentage for each treatment.

Detached leaves
Ten isolates of endophytic bacteria used in this experiment were selected according to results observed in the two previous tests. The treatments consisted of a bacterial suspension (A 550 = 0.1) sprayed on completely developed coffee leaves (C. arabica cv. Mundo Novo), 72 and 24 hours before, after and simulta-neously with the inoculation of the urediniospore suspension. The coffee leaves were placed in plastic boxes, abaxial surface facing up, over a layer of foam saturated with water, covered with a glass plate and incubated as described. The experimental design was randomized blocks (n = 3), each replicate consisting of three leaves. Inoculation was performed using a sprayer attached to a compressor, pressure 10 lb in -2 . Following inoculation, the boxes were covered and placed in the dark for 24 hours, at 22 ± 2ºC. Treatments were evaluated 21 days after inoculation, by counting the number of lesions per leaf. Means were compared by Tukey test (α = 0.05).

Coffee plants
The same endophytic bacterial isolates used in the detached leaves were used in this study. Coffee seedlings (C. arabica cv. Mundo Novo) susceptible to all H. vastatrix strains were obtained from "Centro de Café e Plantas Tropicais, Instituto Agronômico de Campinas", and transplanted into plastic pots containg 5 L of Red Yellow Latosol, sifted through 1.0 cm 2 mesh sieve and mixed with 2.0 kg of lime, 5.0 kg of simple superphosphate, and 0.5 kg of potassium chloride per m 3 of soil. The bacterial suspensions (A 550 = 0.1) were manually sprayed to the foliage until runoff. The urediniospore suspension was applied with a sprayer attached to a compressor, pressure 10 lb in -2 . After inoculation with the H. vastatrix urediniospore suspension (1.0 mg mL -1 ), plants were incubated in the dark for 48 hours at 22 ± 2ºC, 100% relative humidity, and then transferred to a greenhouse. Plants were irrigated daily and after 30 days the number of lesions per inoculated leaf was evaluated. Sterilized water was used as control. Trial was set up in a randomized blocks design (n = 3), with two plants per pot and, means compared by Tukey test (α = 0.05).
The most effective isolates were identified based on cell membrane fatty acid contents, analyzed in a gas chromatograph, using microbial identification software (MIDI, Sherlock ® TSBA Library version 5.0, Microbial ID, Newark, DE, USA). Isolates with a similarity index of 0.6 or higher were considered positively identified.

RESULTS AND DISCUSSION
Twenty three out of the 40 endophytic bacteria isolates tested for their capacity of inhibiting H. vastatrix urediniospore germination inhibited germination in more than 40% (Table 1), irrespective of source (coffee species or plant organ), in relation to the control (water). In addition, deformations of the germination tube that were detrimental compared to normal development, were observed. All bacterial isolates were statistically inferior to the propiconazole.
In the leaf disc assay, the endophytic bacterium TG4-Ia reduced disease severity in all application intervals tested. Control levels were above 63% when applied at 72 and 24 hours before or simultaneously to fungal pathogen inoculation (Table 2). Other isolates were also effective in reducing rust development in the leaf disc tests, but with less intensity. In general, lower control levels were observed when the bacterium was applied after pathogen inoculation. Some isolates increased severity of the disease, especially TF4-IIa, which increased the severity of lesions by 64% when applied 72 hours before the pathogen. Only TG11-IIa was effective in inhibiting in vitro urediniospore germination (Table 1), and in reducing the percentage of leaf area with lesions in coffee leaf discs. Isolates TF2-IIc, TF3-IIa, TF7-Ib, TF7-IIa, TF9-IIa, AF7-IIIa, TG4-Ia, TG4-IIa, TG10-IIIc, and TG11-Iia, were selected for further studies on detached leaves and on coffee plants.
In the detached leaves assay, the most prominent isolates were TF7-Ib, TF9-Ia, TF3-IIa, TG10-IIIc, and TF7-IIa, which showed significant control in all application intervals tested (Table 3). Even though the endophytic bacterium TF9-Ia has yielded the best control (62.0%), its performance was not reproducible on coffee plants. There was a decline in the number of isolates able of reducing severity of the disease, as the interval between the presence of the biocontrol agents and the pathogen decreased.
In the test with coffee plants, the endophytic bacteria were not effective in controlling coffee leaf rust when applied after inoculation of the pathogen (Table 4). Isolates TF2-IIc, TF7-IIa, TG4-Ia, and TG11-IIa were effective when applied either 72 and 24 before or concurrently with the pathogen (Table 4).
Endophytes TG4-Ia, TF9-Ia, TF2-IIc, and TF7-IIa, identified as Bacillus lentimorbus, Bacillus cereus, Clavibacter michiganensis subsp. michiganensis Smith, and Klebsiella pneumoniae Schroeter, respectively, showed the best performance. The other endophytes were identified as Bacillus sp. ( The efficiency of certain endophytic bacteria isolates in controlling coffee leaf rust can vary according to the moment of biocontrol agent application. In general, the endophytes were more effective when applied 72 and 24 hours before and concurrently with the inoculation of H. vastatrix urediniospores. Similar results were obtained by Bettiol et al. (1994) and Bettiol & Várzea (1992); these authors registered reductions in the percentage of leaves with lesions and in the number of lesions per leaf rating 60% and 100%, by spraying different concentrations of non-endophytic Bacillus subtilis-based products 72 and 24 hours before the application of H. vastatrix urediniospores, on coffee  ) Sci. Agric. (Piracicaba, Braz.), v.63, n.1, p.32-39, Jan./Feb. 2006 plants, cv. Catuaí. The fact that the endophytic isolates showed activity when applied before the pathogen suggests that these isolates may act by antibiosis, lysis of pathogen structures, competition, or induction of systemic resistance in the host. Kim et al. (2002) reported that the bacterium B. lentimorbus produces the antifungal substances alpha-and beta-glucosidase, with an inhibitory action against the development of Botrytis cinerea Pers.:Fr, while Sadfi et al. (2001) reported that this bacterium is capable of releasing volatile substances that contribute to the inhibition of Fusarium sambucinum Fuckel in potato tubers. Several authors demonstrated that B. cereus can promote growth in various plant species such as tomato (Simon et al., 2001), and wheat (Ryder et al., 1999). In addition, it can actively penetrate the  tissues and disseminate inter-and intracellularly within the host, protecting it from F. sambucinum, in potatoes, by producing fungitoxic substances (Chérif et al., 2003). Also, B. cereus can produce various chitinases, active against several plant pathogens, such as F. sambucinum (Sadfi et al., 2001), Rhizoctonia solani Kühn (Ryder et al., 1999), Helminthosporium solani Dur. & Mont. (Martinez et al., 2002), Sclerotium rolfsii Sacc., Fusarium oxysporum Schl., and Pythium aphanidermatum (Edson) Fitzp. B. cereus has also been reported as endophytic in cotton (Gossypium hirsutum L), sweet corn (Zea mays L), and citrus plants (Citrus spp.) by Di Fiore & Del Gallo (1995).
It could thus be speculated that there were more than one mode of action of those endophytic bacteria in the control of coffee leaf rust. Inhibitory action against H. vastatrix in urediniospore germination was shown in specific essays, tests with leaf discs, with detached leaves, and with coffee plants at different application intervals.
The endophytic association of bacteria of the genera Clavibacter and Klebsiella with some agronomic crops, such as corn, grapevine, rice, cotton, and some crucifers was reported by Lodewyckx et al. (2002). However, reports on the application of species of the above-mentioned genera dealing with control of pathogens are scarce in the literature. C. michiganensis subsp. michiganensis is a phytopathogenic species (Agrios, 1997), while K. pneumoniae can be found in hospitals causing infections in humans (Martins-Loureiro et al., 2001). These characteristics can create barriers to the application of these bacteria in bioassays seeking plant disease biocontrol agents. No reports were found in biological control on the application of the other tested endophytes.
Properties of some endophytic isolates in increasing coffee leaf rust severity shall be highlighted. According to Musson (1994), some endophytic organisms can behave as non-pathogenic in a given host, and as pathogenic in another. Cameron (1970) found that Pseudomonas spp. isolates obtained from healthy cherry tissues proved to be phytopathogenic in further tests. This author suggested that the endophyte condition could be one of the forms of survival and escape against surface phytosanitary treatments. Additionally, Whitesides & Spotts (1991) found Pseudomonas syringae van Hall isolates from pear tree roots, which were not pathogenic neither to cherry nor to pear trees, and suggested that the internal tissues of pear trees could function as inoculum reservoir to other plants. Thus, the ability of colonizing internal plant tissues could be visualized as a survival mechanism of plant pathogenic bacteria, since they exist in a protected position (Leben, 1981). This fact has biological impor-tance, because demonstrates that the interactions with endophytes could have economic importance related to both the control and to the expansion of plant diseases.
Even though a relatively small number of endophytic bacteria were tested, promising results were obtained regarding the selection of coffee leaf rust biocontrol agents. Further field studies must be conducted to analyze the real potential of endophytic bacteria in field conditions. Studies are also needed to determine the modes of action of those bacteria, the population density of the applied endophytes, and the best form of introduction into the host. The evaluation of the effects of agrochemicals on endophytic bacteria have also to be considered, since those can not only stimulate isolates that are beneficial to rust control but can also select those that increase severity of the disease.