Abstract
Bacterial speck of tomato, caused by Pseudomonas syringae pv. tomato, was used to determine whether similarity in carbon source utilization between a preemptive biological control agent and the pathogen was significant in determining the ability of the bacterium to suppress disease. Similarity in carbon source utilization was quantified as the ratio of the number of tomato carbon sources utilized in vitro by the biological control agent to the number of tomato carbon sources utilized in vitro by the target pathogen (the niche overlap index [NOI]). Suppression of the disease was quantified as the percent reduction in disease severity compared to the pathogen-only control when nonpathogenic bacteria were applied to foliage 48 h prior to the pathogen. In the collection of 36 nonpathogenic bacterial strains, there was a significant (P < 0.01), but weak (r2 = 0.25), correlation between reduction in disease severity and similarity in carbon source utilization, suggesting that similarity in carbon source use was significant in determining ability to suppress disease. The relationship was investigated further using catabolic mutants of P. syringae strain TLP2, an effective biological control agent of speck. Catabolic mutants exhibited lower levels of similarity (NOI = 0.07 to 0.90) than did wild-type TLP2 (NOI = 0.93). With these catabolic mutants there was a significant (P < 0.01), and stronger (r2 = 0.42), correlation between reduction in disease severity and similarity in carbon source utilization. This suggests that similarity in carbon source utilization was a more important component of biological control ability for the catabolic mutants than for the nonpathogenic bacteria. Together, these studies indicate that suppression of bacterial speck of tomato was correlated with nutritional similarity between the pathogenic and nonpathogenic bacteria and suggest that preemptive utilization of carbon sources was probably involved in the biological control of the disease by both the naturally occurring nonpathogenic bacteria and the catabolic mutants.
Aerial plant surfaces harbor complex communities of nonpathogenic and pathogenic microorganisms and it is believed that interactions among these microorganisms such as antibiosis and nutrition competition may lead to natural biological control of foliar pathogens (1, 11, 18). While in vitro antibiosis has been widely used to select potential biological control agents of foliar diseases, studies demonstrating antibiosis to be an important attribute in biological control are scarce. In a test using 931 bacterial and yeast isolates for control of apple scab caused by Venturia inaequalis, no correlation was observed between in vitro antibiosis and ability of these microorganisms to suppress the disease on apple seedlings (4). In another study, in vitro antibiotic production in a collection of epiphytic bacteria was not correlated with the ability to control ice nucleation active (Ice+) Pseudomonas syringae strains on corn leaves (15). These studies suggest either that in vitro antibiosis is not predictive of antibiosis in vivo or that antibiosis is not a significant attribute determining the efficacy of the majority of these microbes in control of these pathogens.
While it is unclear which characteristics of nonpathogenic microorganisms may be predictive of ability to control foliar diseases, there is evidence suggesting that the ability of nonpathogenic bacteria to compete for limiting nutritional resources contributes significantly to the suppression or exclusion of Ice+ bacteria in the phyllosphere. De Wit replacement series (7) were used to study competitive interactions between epiphytic Ice+ P. syringae strains and the biological frost control agents Ice− P. syringae TLP2del1 and Pseudomonas fluorescens A506 on potato leaves (34). The results indicated that Ice+ and Ice− P. syringae strains competed for limiting nutritional resources and that the epiphytic P. syringae population size was more limited by the availability of carbon than by the availability of nitrogen. Further studies indicated that the level of coexistence between various nonpathogenic epiphytic bacteria and an Ice+ P. syringae strain in the phyllosphere of bean plants was inversely correlated with similarity in the utilization of limiting nutritional resources (35). Similarity in limiting nutritional resource use was quantified using a niche overlap index (NOI) based upon in vitro carbon source utilization profiles (34, 35). These observations confirmed the ecological significance of NOIs estimated from in vitro carbon source utilization profiles.
From the studies mentioned above, it was hypothesized that one could predict that a bacterium with a high NOI with respect to the target pathogen would be effective in the preemptive exclusion of the target pathogen by usurping a high percentage of carbon sources that would otherwise be available to the target pathogen (35). In support of this idea, other studies have demonstrated that bacteria are actively engaged in carbon consumption in the phyllosphere shortly after inoculation (13, 20) and that preemptive utilization of nutrients in the phyllosphere by application of saprophytic microorganisms could influence nutrient availability and hence inhibit colonization and disease development of foliar plant pathogens (9, 10, 23, 33). Thus, the hypothesis was proposed that the effectiveness of a nonpathogenic bacterial strain as a preemptive biological control agent of a leaf-associated phytopathogenic bacterium would be determined by nutritional similarity (quantified using the NOI) between the biological control agent and the target pathogen in the phyllosphere of the host plant. Should such a relationship exist, this would not only provide useful information for understanding of the nature of phyllosphere interactions between plant pathogenic and nonpathogenic microorganisms but would also provide an avenue to manipulate pathogen-biological control agent interactions to improve efficacy in control of plant diseases.
In this study, bacterial speck of tomato, caused by P. syringae pv. tomato, was used as a model to determine whether the effectiveness in biological control of this disease is determined by similarity in carbon source utilization between the nonpathogenic bacteria and the pathogen. Bacterial speck of tomato was selected because there are reports that the pathogen P. syringae pv. tomato establishes epiphytic populations prior to infection and disease development (3, 24, 27), and it has been suggested that epiphytic population sizes of several bacterial pathogens, including P. syringae, may be correlated with the severity of diseases they cause (14, 26, 28, 29). Hence, reductions in population sizes of the pathogen P. syringae pv. tomato due to preemptive carbon source utilization would be expected to result in reductions in disease severity. The validity of the hypothesis that nutritional similarity is important in biological control of bacterial speck of tomato was assessed with a random collection of nonpathogenic bacteria isolated from tomato foliage and selected Tn5-induced catabolic mutants of an effective biological control agent of bacterial speck, P. syringae strain TLP2 (32). The biological control agent TLP2 is not known to produce any antibiotics or bacteriocins against P. syringae (32). It was anticipated that the TLP2 catabolic mutants would provide a superior test of the hypothesized correlation, since the mutants would all express the same biological control phenotypes and only the NOI should vary. This is in contrast to the collection of naturally occurring nonpathogenic bacteria which may exhibit many different biological control phenotypes in addition to exhibiting a different NOI with respect to the pathogen.
MATERIALS AND METHODS
Bacterial strains.
A total of 36 nonpathogenic bacterial strains, isolated from symptomless tomato foliage in the Southeastern United States and representing potential biological control agents of bacterial speck of tomato, were used in this study (Table 1). Strains either were isolated from tomatoes in Alabama (the “Cu” strains) or were provided by J. B. Jones (University of Florida, Gainesville) (the “B” strains). Serratia sp. strain BT25 was obtained from P. A. Backman (Auburn University, Auburn, Ala.). Bacteria were identified to at least the genus level using the Sherlock GC-FAME System (MIDI, Newark, Del.). P. syringae strain TLP2 was provided by S. E. Lindow (University of California, Berkeley). P. syringae TLP2 has been shown to be an effective biological control agent of bacterial speck when applied in advance of the pathogen (32). P. syringae pv. tomato strain PT12 was provided by D. A. Cooksey (University of California, Riverside). Escherichia coli HB101(pUW964) that harbors the Tn5-carrying suicide vector pUW964 was obtained from G. Beattie (Iowa State University, Ames).
TABLE 1.
Bacterial strains used in evaluation of the relationship between similarity in carbon source utilization and biological control of bacterial speck of tomato
| Straina | Species identification | NOIb | Source |
|---|---|---|---|
| BT25 | Serratia sp. | 0.90 | P. A. Backman |
| B2 | Erwinia sp. | 0.80 | J. B. Jones |
| B16 | Flavobacterium sp. | 0.23 | J. B. Jones |
| B41 | Pseudomonas sp. | 0.83 | J. B. Jones |
| B51 | Arthrobacter sp. | 0.13 | J. B. Jones |
| B52 | Pseudomonas sp. | 0.87 | J. B. Jones |
| B56 | Pseudomonas sp. | 0.90 | J. B. Jones |
| B106 | Bacillus sp. | 0.23 | J. B. Jones |
| Cu1 | Bacillus sp. | 0.40 | This lab |
| Cu2 | Flavimonas sp. | 0.93 | This lab |
| Cu4 | Xanthomonas sp. | 0.50 | This lab |
| Cu5 | Flavimonas sp. | 1.00 | This lab |
| Cu6 | Brevibacterium sp. | 0.13 | This lab |
| Cu8 | E. coli | 0.87 | This lab |
| Cu9 | Pseudomonas sp. | 0.97 | This lab |
| Cu10 | Flavimonas sp. | 0.93 | This lab |
| Cu11 | Flavimonas sp. | 0.93 | This lab |
| Cu12 | Serratia sp. | 0.83 | This lab |
| Cu13 | E. coli | 0.83 | This lab |
| Cu14 | Pseudomonas sp. | 0.57 | This lab |
| Cu17 | Enterobacter sp. | 0.50 | This lab |
| Cu21 | Pseudomonas sp. | 0.90 | This lab |
| Cu22 | E. coli | 0.80 | This lab |
| Cu23 | E. coli | 0.83 | This lab |
| Cu31 | Pseudomonas sp. | 0.90 | This lab |
| Cu32 | Pseudomonas sp. | 0.43 | This lab |
| Cu34 | Pseudomonas sp. | 0.83 | This lab |
| Cu37 | Pseudomonas sp. | 0.97 | This lab |
| Cu39 | Pseudomonas sp. | 0.90 | This lab |
| Cu43 | Pseudomonas sp. | 0.83 | This lab |
| Cu44 | Pseudomonas sp. | 0.83 | This lab |
| Cu45 | Pseudomonas sp. | 0.83 | This lab |
| Cu46 | Pseudomonas sp. | 0.83 | This lab |
| Cu49 | Pseudomonas sp. | 0.87 | This lab |
| Cu55 | Pseudomonas sp. | 0.77 | This lab |
| Cu59 | Pseudomonas sp. | 0.87 | This lab |
Bacterial strains are nonpathogenic bacteria isolated from tomato foliage.
NOI with respect to the pathogen P. syringae pv. tomato strain PT12. Calculation of NOI was based upon in vitro use as the sole carbon source of the 30 carbon compounds utilized by PT12.
Determining carbon source use of nonpathogenic strains and the pathogen.
To test in vitro use of carbon sources by the nonpathogenic bacterial strains and the pathogen, P. syringae pv. tomato strain PT12, individual carbon compounds were incorporated into minimal medium A (MinA) (21) at a concentration of 10 mM. A total of 52 carbon sources that were reported to be present in tomato plants (5, 6, 22, 25) were included in the test (Table 2). Bacterial strains were cultured on tryptic soy agar (TSA) or King's medium B (KB) (12) at 28°C for 24 to 30 h and then transferred onto one-tenth strength TSA and incubated at 28°C for 48 h (to deplete any endogenous carbon reserves). Bacterial cells were suspended in sterile potassium phosphate buffer (0.01 M, pH 7.0) (PPB). The cell suspensions were adjusted turbidimetrically to approximately 109 CFU/ml and diluted to 107 CFU/ml using sterile PPB. A 10-μl volume of bacterial suspension was spotted onto one plate of minimal medium A containing each carbon source. Plates were incubated for 72 h before being scored for the presence or absence of growth.
TABLE 2.
Carbon sources reported to be present in tomato tissues which were used in determining carbon source profiles of the pathogen P. syringae pv. tomato PT12 and the biological control agent P. syringae TLP2
| Carbon source (abbreviation) | Growtha
|
|
|---|---|---|
| PT12 | TLP2 | |
| Acetic acid (Ace) | + | + |
| Aconitic acid (Aco) | + | + |
| l-Alanine (Ala) | + | + |
| β-Alanine (β-Ala) | − | − |
| l-Arginine (Arg) | + | + |
| Ascorbic acid (Asc) | + | + |
| l-Asparagine (Asn) | + | + |
| Aspartic acid (Asp) | + | + |
| γ-Aminobutyric acid (Ami) | − | − |
| Citric acid (Cit) | + | + |
| Citrulline (Cin) | − | − |
| l-Cysteine (Cys) | − | − |
| Dihydroxytartaric acid (Oxy) | − | − |
| Ethanolamine (Eth) | − | − |
| Formic acid (For) | − | − |
| Fumaric acid (Fum) | + | + |
| Fructose (Fruc) | + | + |
| Galacturonic acid (Galac) | + | + |
| Glucose (Glucs) | + | + |
| l-Glutamic acid (Glu) | + | + |
| l-Glutamine (Gln) | + | + |
| Glutaric acid (Glut) | + | + |
| l-Glycine (Gly) | − | − |
| Glycolic acid (Glyc) | − | − |
| l-Histidine (His) | + | + |
| Isoleucine (Iso) | − | − |
| Lactic acid (Lac) | + | + |
| l-Lysine (Lys) | − | − |
| l-Leucine (Leu) | − | − |
| Maleic acid (Male) | − | − |
| Malic acid (Mal) | + | + |
| Malonic acid (Malon) | + | + |
| l-Methionine (Met) | − | − |
| Mevalonic acid (Mev) | − | − |
| Myoinositol (Ino) | + | + |
| Oxalacetic acid (Oxa) | + | + |
| Oxalic acid (Oxal) | − | − |
| 2-Oxoglutaric (Oxo) | + | + |
| Pipecolic acid (Pip) | + | − |
| l-Phenyalanine (Phe) | − | − |
| l-Proline (Pro) | + | + |
| Pyruvic acid (Pyr) | + | + |
| Quinic acid (Qui) | + | + |
| l-Serine (Ser) | + | + |
| Succinic acid (Succ) | + | + |
| Sucrose (Sucs) | + | + |
| Starch (Sta) | − | − |
| Tartaric acid (Tar) | − | − |
| l-Threonine (Thr) | − | − |
| l-Tryptophan (Try) | − | − |
| l-Tyrosine (Tyr) | + | − |
| l-Valine (Val) | − | − |
In vitro use of these carbon compounds was determined by the presence (+) or absence (−) of growth of the bacterium on MinA amended with a sole carbon source (10mM).
Catabolic mutants of the biological control agent P. syringae TLP2.
Tn5 mutants of P. syringae strain TLP2 were generated using the suicide vector pUW964 using methods described by Lindow (16). To screen catabolic mutants, presumptive mutants showing resistance to rifampin (100 μg/ml) and kanamycin (30 μg/ml) were inoculated into one-tenth strength tryptic soy broth in sterile 96-well plates using a metal 48-pronged replicator and incubated at 28°C overnight. Bacterial suspensions were transferred from the 96-well plates using the metal replicator and printed onto plates of MinA containing each compound as a sole carbon source, and the plates were incubated at 28°C for 72 h. A total of 30 carbon sources that were reported to be present in tomato plants and were utilized in vitro as a sole carbon source by either the pathogen P. syringae pv. tomato strain PT12 or the wild-type TLP2 strain were included in the screening (Table 2). Potential catabolic mutants that showed no growth on one or more of these carbon source plates were selected. Catabolic phenotypes were confirmed in a secondary screening. For secondary screening, cultures of the selected colonies were grown on KB plates containing kanamycin and rifampin. Bacterial cultures were then transferred onto one-tenth strength TSA plates and incubated at 28°C for 48 h. Bacterial cells were scraped from the plates and suspended in sterile PPB. The cell suspensions were adjusted turbidimetrically to approximately 109 CFU/ml and diluted to 107 CFU/ml using sterile PPB. A 10-μl volume of culture was spotted onto one plate of MinA containing a single carbon source. Plates were incubated at 28°C for 72 h before being scored for the presence or absence of growth.
Determination of nutritional similarity.
Nutritional similarity between each of the tested potential biological control agents (nonpathogenic strains or Tn5 catabolic mutants of TLP2) and the pathogen, P. syringae pv. tomato strain PT12, was estimated using the following formula for NOI described by Wilson and Lindow (35). Estimation of nutritional similarity was based upon in vitro use as sole carbon sources of the 30 carbon compounds that were reported to be present in tomato plants and were utilized by the pathogen strain PT12 (Table 2). No. of carbon sources used by both the testing biological control agent and the pathogen NOI =Total no. of carbon sources utilized by the pathogen PT12
Evaluation of biological control effectiveness.
Biological control assays were conducted under greenhouse conditions on 5-week-old tomato plants (cv. Agriset 761; Agrisales, Inc., Plant City, Fla.). The nonpathogenic bacterial strains were grown on TSA plates at 28°C for 24 to 30 h, whereas selected catabolic mutants of P. syringae TLP2 were cultured on KB plates containing kanamycin (30 μg/ml) and rifampin (100 μg/ml). Bacterial cells were suspended in sterile PPB. The cell suspensions were adjusted turbidimetrically to approximately 109 CFU/ml, diluted to 107 CFU/ml, and inoculated onto tomato foliage using a hand-held mister. A randomized complete block design with four replications was used for arranging the tomato plants in the greenhouse. Thirty-six nonpathogenic bacterial strains and 30 selected TLP2 catabolic mutants were included in the assessment. Tomato plants inoculated with the pathogen alone were used as a control. The biological control agent TLP2 wild-type strain was also included as a control for the catabolic mutants. The pathogen strain PT12 was cultured on KB plates and incubated at 28°C for 30 h. Bacterial suspensions of PT12, prepared as for the nonpathogenic bacterial strains and diluted to approximately 108 CFU/ml, were inoculated onto tomato foliage 48 h after inoculation of the nonpathogenic bacteria. The inoculated plants were maintained in the greenhouse under disease-conducive conditions until symptom development (32).
Disease severity on plants was assessed as follows. Ten leaflets of the tomato plants were collected randomly from each plant approximately 7 days after inoculation of the pathogen. Lesion numbers on each leaflet were counted manually and areas of individual leaflets were determined by image analysis (AgVision Monochrome System; Gagne, Inc., Binghamton, N.Y.). Disease severity data were subjected to log transformation and expressed as log10 (lesions + 1)/cm2 (32). Analysis of variance was performed using the ANOVA or GLM procedures of the Statistical Analysis system (SAS Institute, Cary, N.C.). Means were compared using Duncan's multiple range test at P = 0.05. Biological control efficacy was quantified as the percent reduction in disease severity compared to the pathogen-only control as described previously (32).
Measurement of population sizes of TLP2 catabolic mutants on leaves.
Bacterial suspensions of the 30 selected TLP2 catabolic mutants and the TLP2 wild-type strain were prepared as in the disease control experiments described above. Bacterial suspensions (approximately 106 CFU/ml) were inoculated onto the foliage of 5-week-old tomato plants (cv. Agriset 761) by misting three plants for each bacterial strain. The inoculated plants were covered with plastic bags and distributed as a randomized complete block design in a growth room at 25°C and with a 12-h photoperiod. Four leaflets were sampled from each plant 48 h after inoculation and leaflets were placed individually into sterile Whirl-Pak bags (Macalaster Bicknell Co., New Haven, Conn.). A 20-ml volume of sterile PPB was added into each bag and the bag was then sonicated in an ultrasonic cleaning bath for 7 min to dislodge the epiphytic bacteria. Serial dilutions of leaf washings were plated using a spiral plater (Spiral Biotech, Bethesda, Md.) on KB amended with kanamycin (30 μg/ml), rifampin (100 μg/ml), and cycloheximide (100 μg/ml). After incubation at 28°C for 30 h colonies on the plates were counted using a laser counter (Spiral Biotech). The mean population size was determined from the log10-transformed population size of 12 individual leaflets.
Correlation analysis.
Assessment of the ability of the 36 nonpathogenic bacterial strains and the 30 selected TLP2 catabolic mutants to reduce disease severity was repeated two or three times. Correlations between biological control effectiveness (percent disease reduction as defined above) of the nonpathogenic bacterial strains and nutritional similarity (NOI) between these strains and the pathogen strain PT12 were analyzed using Tablecurve software (Jandel Scientific) to determine whether there was a significant linear or nonlinear correlation. Regression analysis was used to examine the relationship between nutritional similarity (similarity in carbon source utilization) and biological control efficacy when a mean from the repeated biological control experiments was used. Correlations between population sizes of TLP2 mutants on tomato leaves and their similarity with the parental strain in carbon source use (number of carbon sources used), as well as the relationship between population size and ability of these mutants to reduce severity of bacterial speck, were also determined by Tablecurve software.
RESULTS
Naturally occurring bacterial strains.
The 36 randomly selected nonpathogenic bacterial strains belonged to 11 different genera, half of them being Pseudomonas spp., while other genera included Arthrobacter, Bacillus, Brevibacillus, Escherichia, Enterobacter, Erwinia, Flavimonas, Flavobacterium, Serratia, and Xanthomonas (Table 1). Carbon source utilization profiles of these bacterial strains were determined based upon in vitro use, as sole carbon sources, of the 30 carbon compounds reported to be present in tomato tissues and presumed to be available in the tomato phyllosphere. The carbon source utilization profiles of these bacterial strains were quite different. Similarity in carbon source use between these strains and the pathogen P. syringae pv. tomato strain PT12 ranged from an NOI of 0.13 for Brevibacterium sp. strain Cu6 and Arthrobacter sp. strain B51 to an NOI of 1.0 for Flavimonas sp. strain Cu5 (Table 1).
Biological control efficacy of these nonpathogenic bacterial strains, based on reductions in severity of foliar bacterial speck of tomato, was evaluated under greenhouse conditions. Biological control experiments were repeated three times. When applied 48 h in advance of the pathogen, about 35% of these bacterial strains provided significant (P = 0.05) disease reductions in all three repetitions, while other strains significantly reduced disease severity in some experiments but not others. The mean reductions in disease severity (based on lesion numbers per square centimeter of leaf area) ranged from 21.6 to 69.9% compared to the pathogen-only control. The biological control efficacy, quantified by percent reduction in disease severity, was plotted against similarity in carbon source utilization between these nonpathogenic bacterial strains and the pathogen (Fig. 1A). Regression analyses indicate that biological control efficacy of these bacteria was significantly (P < 0.01; r2 = 0.25) correlated with similarity in carbon source utilization (Fig. 1A).
FIG. 1.
Relationship between nutritional similarity of leaf-associated nonpathogenic bacteria or Tn5-generated catabolic mutants of P. syringae strain TLP2 and the pathogen and efficacy in control of bacterial speck of tomato caused by P. syringae pv. tomato. Disease reduction data represent percent reduction in disease severity (no. of lesions per square centimeter of leaf). Determination of nutritional similarity was based on utilization of the 30 carbon compounds that were reported to be present in tomato plants and were found to be used by the pathogen. (A) Leaf-associated nonpathogenic bacteria. (B) Catabolic mutants of P. syringae strain TLP2 and the parental strain.
Correlation between similarity in carbon source utilization and disease suppression of TLP2 catabolic mutants.
The biological control agent P. syringae strain TLP2 and the pathogen P. syringae pv. tomato strain PT12 were very similar in utilization of the carbon compounds reported to be present in tomato tissues. Of the 52 carbon sources reported to be present in tomato tissue, 30 were used by PT12 in vitro as sole carbon sources, and of these, TLP2 was able to use 28 (Table 2); hence, the nutritional similarity between these two strains was very high (NOI = 0.93).
A library of 6,000 Tn5 mutants of TLP2 was generated and a total of 120 catabolic mutants were selected after repeated screening. These mutants exhibited a reduced range of carbon source utilization compared to the parental strain TLP2 (Table 3); hence, nutritional similarity with respect to PT12 was reduced and ranged from NOI = 0.07 to NOI = 0.90.
TABLE 3.
Carbon source profiles of catabolic mutants of the biological control agent P. syringae TLP2 used in biological control experiments
| Carbon sourcea | TLP2 | Growth of catabolic mutant of TLP2b
|
|||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 18 | 37 | 44 | 45 | 93 | 100 | 120 | 122 | 129 | 138 | 140 | 142 | 147 | 149 | 150 | 157 | 166 | 168 | 179 | 181 | 243 | 245 | 251 | 273 | 282 | 305 | 308 | 311 | 312 | 314 | ||
| Ace | + | − | + | − | + | − | − | − | − | − | + | + | − | + | − | + | − | + | − | − | − | + | − | + | − | − | − | − | − | − | + |
| Aco | + | − | + | − | + | + | − | − | − | − | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + | − | + | + | + | + | + |
| Ala | + | − | + | − | + | + | − | − | − | − | + | + | − | + | + | + | + | + | − | + | + | − | − | + | + | + | − | + | + | − | + |
| Arg | + | − | + | − | + | + | + | − | − | + | + | + | − | + | − | + | + | + | + | + | + | + | + | − | − | − | − | − | + | − | + |
| Asc | + | − | − | − | − | + | − | − | − | − | + | + | + | − | + | + | − | + | − | + | + | + | + | − | − | − | − | − | + | − | + |
| Asn | + | − | + | − | + | − | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + | − | − | + | + | + | + | − | + | − | − |
| Asp | + | − | + | − | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + | − | − | + | + | + | − | − | + | − | + |
| Cit | + | − | + | + | + | − | − | + | + | + | + | + | + | + | + | + | + | + | − | + | + | + | − | + | + | + | + | + | + | − | + |
| Fru | + | − | + | − | + | + | − | − | + | + | + | + | − | + | + | + | + | + | − | + | − | − | − | + | + | + | − | − | − | − | + |
| Fum | + | − | + | − | + | + | − | + | + | − | + | + | + | + | + | + | + | + | − | + | + | + | − | + | + | + | + | − | + | − | + |
| Galac | + | − | + | − | + | + | − | − | − | − | + | + | − | + | − | + | + | − | − | + | + | + | + | + | − | + | + | − | + | + | + |
| Glucs | + | − | + | − | + | + | − | − | + | + | + | + | + | − | + | + | + | + | − | + | + | + | − | + | + | + | + | − | + | − | + |
| Gln | + | + | + | − | + | − | + | − | + | + | + | + | + | + | + | + | + | + | + | + | + | − | − | + | + | + | − | − | − | − | + |
| Glu | + | − | + | − | + | − | + | − | − | + | + | + | − | + | + | + | + | + | + | + | + | − | − | + | − | − | − | − | − | − | + |
| Glut | + | − | + | − | + | + | − | − | − | − | + | + | − | + | − | + | − | + | − | + | + | + | + | − | − | − | − | + | − | − | + |
| His | + | − | + | − | − | + | + | − | − | + | − | + | − | + | − | + | + | + | + | + | − | − | + | + | − | − | − | − | − | − | + |
| Ino | + | + | + | − | + | + | − | − | + | + | + | + | + | + | + | + | + | + | − | + | + | + | − | + | + | + | + | − | + | − | + |
| Lac | + | − | + | − | + | + | − | − | − | − | + | + | + | + | + | + | + | + | − | + | + | + | + | + | − | − | + | − | + | + | + |
| Mal | + | + | + | − | + | + | − | − | + | − | + | + | + | + | + | + | − | + | − | + | + | + | − | + | + | + | + | − | + | − | + |
| Malon | + | − | + | − | + | − | − | − | − | − | + | − | − | + | − | + | − | + | − | − | − | + | − | + | − | + | − | + | − | − | + |
| Oxa | + | − | − | − | + | + | − | − | − | − | + | + | − | − | + | + | + | + | − | + | + | + | + | + | + | − | + | + | + | − | + |
| Oxo | + | − | + | − | + | + | − | − | + | − | + | + | + | + | + | + | + | + | − | + | + | + | + | + | + | + | + | + | + | + | + |
| Pip | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − |
| Pro | + | − | + | − | + | − | + | − | + | + | + | + | − | + | − | − | − | + | + | − | − | − | − | + | − | − | + | − | + | − | + |
| Pyr | + | − | + | − | + | + | − | − | − | − | + | + | − | − | + | + | + | + | − | + | + | + | + | + | − | + | + | − | + | − | + |
| Qui | + | − | + | − | + | + | − | − | + | − | + | + | − | + | + | + | − | + | − | − | − | + | − | + | + | + | + | − | − | − | + |
| Ser | + | + | + | + | + | + | − | − | + | + | + | + | + | + | + | + | + | + | − | + | + | − | − | + | + | + | − | − | + | − | + |
| Succ | + | − | + | − | + | + | − | − | + | − | + | + | + | + | + | + | + | + | − | + | + | + | − | − | + | + | + | + | + | + | + |
| Sucs | + | − | + | − | + | − | − | − | + | + | + | + | + | − | + | + | + | + | − | + | + | − | − | + | + | + | + | − | + | − | + |
| Tyr | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − | − |
Carbon source profiles of these mutants were based on the 30 carbon sources that were reported to be present in tomato plants and were found to be utilized by the pathogen P. syringae pv. tomato strain PT12. For full names, see Table 2.
In vitro use of these carbon compounds was determined by the presence (+) or absence (−) of growth of these bacteria on MinA amended with a sole carbon source (10 mM).
Thirty different catabolic mutants that were altered in utilization of 2 to 27 different carbon sources (Table 3) were selected for evaluation of their ability to suppress bacterial speck under greenhouse conditions. While about 30% of the selected mutants still provided significant disease reductions, the majority of the mutants no longer significantly suppressed disease in repeated trials (Fig. 1B). Disease suppression provided by the selected mutants ranged from 0 to 57.8% when the mutants were applied in advance of the pathogen (Fig. 1B), while a mean disease reduction ranging from 49.8 to 55.9% was achieved by wild-type TLP2. There was a significant correlation (Fig. 1B) between similarity in carbon source utilization (NOI with respect to PT12) and percent disease reduction (P < 0.01; r2 = 0.42) with these catabolic mutants.
Relationship between carbon source utilization and phyllosphere population sizes of TLP2 mutants.
Phyllosphere population sizes of the 30 catabolic mutants and the wild type of the biological control agent P. syringae TLP2 were determined under growth chamber conditions. The mean log10-transformed population size of wild-type TLP2 on tomato leaves 48 h after inoculation was approximately log10 7.1 CFU/leaflet under the conditions of 100% relative humidity and 25°C. Population sizes of the 30 catabolic mutants evaluated ranged from log10 6.36 to log10 7.26 CFU/leaflet (Fig. 2). There was a significant correlation between the number of carbon sources used by the 30 mutants and their population sizes on tomato leaves (P < 0.01; r2 = 0.57) (Fig. 2). Population sizes of the 30 mutants on tomato leaves were also plotted against the efficacy of these mutants in suppression of bacterial speck (Fig. 3). Regression analysis indicated that there was only a very weak correlation between population size and efficacy (P = 0.06; r2 = 0.12) (Fig. 3).
FIG. 2.
Correlation between carbon source utilization and phyllosphere population sizes of catabolic mutants of P. syringae strain TLP2. Utilization of carbon sources was based upon in vitro use as a sole carbon source of the 30 carbon compounds that were reported to be present in tomato plants and were found to be used by the parental TLP2 strain or the pathogen P. syringae pv. tomato. Population sizes were the means determined from 12 individual leaflets of tomato growing in a growth chamber.
FIG. 3.
Relationship between phyllosphere population size and biological control efficacy of catabolic mutants of the biological control agent P. syringae TLP2. Population sizes were the means determined from 12 individual leaflets of tomato growing in a growth chamber. Disease reduction data represent percent reduction in disease severity (no. of lesions per square centimeter of leaf).
DISCUSSION
A group of randomly collected nonpathogenic leaf-associated bacteria was used to determine whether similarity in carbon source utilization with respect to the target pathogen, P. syringae pv. tomato, as estimated by NOI, was predictive of efficacy in biological control of bacterial speck of tomato. Bacterial strains with a high degree of similarity in utilization of carbon sources with the pathogen exhibited a higher degree of suppression of the disease when these bacterial strains were applied in advance of the pathogen than strains with a low degree of similarity in carbon source utilization with the pathogen. A significant correlation between similarity in carbon source utilization between nonpathogenic bacterial strains and the pathogen and efficacy in suppression of bacterial speck indicated that similarity in carbon source utilization profiles is a significant characteristic contributing to the biological control efficacy of these bacteria.
The significance of similarity in carbon source utilization profiles was assessed further through the use of a collection of catabolic mutants of the biological control agent of bacterial speck, P. syringae strain TLP2 (32). Again, a significant correlation was observed between similarity in carbon source utilization by the catabolic mutants and the pathogen and efficacy in suppression of bacterial speck. This indicates that similarity in carbon source utilization is predictive of the ability of a mutant to suppress the disease, i.e., a strain with a higher degree of nutritional similarity is likely to be a better biological control agent. These observations were also consistent with previous studies showing that similarity in carbon source utilization between Ice− and Ice+ P. syringae strains was inversely correlated with their coexistence in the phyllosphere of beans (Phaseolus vulgaris) and potato (Solanum tuberosum) (34, 35).
In these correlations, the correlation coefficient was considerably higher for the catabolic mutants of the biological control agent P. syringae TLP2 (r2 = 0.42) than for the naturally occurring strains (r2 = 0.25), suggesting that similarity in carbon source utilization was a more important component of biological control ability for wild-type TLP2 and the catabolic mutants than for the collection of nonpathogenic bacteria. Some naturally occurring bacterial strains used in this study provided significant disease reductions although the similarity in carbon source utilization was fairly low, and vice versa, indicating that some other biological control phenotypes such as antibiotic production (2, 19, 30), siderophore production (17, 31, 36), or induced disease resistance may have been involved in biological control for some of these strains. The possession of other biological control phenotypes by some of the strains in the random collection of nonpathogenic bacteria likely contributed to the weaker correlation.
The correlation between similarity in carbon source utilization profiles and efficacy strongly suggests that strains with a high degree of similarity to the pathogen in carbon source utilization preemptively utilized carbon sources in the phyllosphere, rendering them unavailable to the pathogen, which was inoculated 48 h later. In this carbon-depleted environment, the pathogen likely attained a lower population size and, hence, caused less disease. This assertion cannot be made, however, in the absence of pathogen population size data which were not collected in this study. While we can suggest that preemptive utilization of carbon sources is a significant factor involved in preemptive biological control, more work is required to determine whether the reduction in disease severity was a result of reduction in pathogen population size.
As may have been anticipated from previous studies suggesting that the phyllosphere of nitrogen-sufficient plants is primarily carbon limited (34, 35), the population size of many of the catabolic mutants was reduced compared to that of wild-type TLP2. While Tn5 mutagenesis may give rise to pleiotropic effects, many of which could be detrimental to growth in the phyllosphere, the relatively strong correlation between the number of tomato carbon sources utilized by the catabolic mutant and the population size strongly suggests that in most cases the reduced population size was due to inaccessibility of some of the carbon sources in the phyllosphere.
While the number of carbon sources utilized by the catabolic mutants was significantly correlated with population size, population size itself was only very weakly correlated with efficacy in control of speck. This observation really suggests, for these catabolic mutants at least, that preemptive carbon utilization and not some other population size-dependent mechanism was the primary mechanism behind the correlation between nutritional similarity and efficacy. In the end, however, only demonstration of carbon depletion in situ will be convincing. Fortunately, techniques have recently become available to detect bacterial consumption of carbon compounds in situ in the phyllosphere (13).
In summary, similarity in carbon source utilization profiles between the nonpathogenic bacteria and P. syringae pv. tomato was correlated with, and therefore was predictive of, efficacy in control of bacterial speck in tomato under greenhouse conditions. Hence, carbon source utilization profiles could be used in an initial in vitro screening for potential biological control agents of bacterial speck, acknowledging that some potential biological control agents of low nutritional similarity but other effective biological control phenotypes may be missed with this technique. The similarity in behavior of Ice+ P. syringae and P. syringae pv. tomato in the phyllosphere also suggests that this predictive ability would extend to other P. syringae pathovars. Unfortunately, it does not seem to apply in the same way to pathovars of Xanthomonas campestris (8, 29), suggesting that this pathogen acts differently than P. syringae in the phyllosphere and that the predictive utility of NOI does not extend beyond P. syringae.
Acknowledgments
We thank J. B. Jones and P. A. Backman for providing some of the bacterial strains and G. Beattie for providing plasmid pUW964. We are also very grateful to S. E. Lindow for a preliminary review of the manuscript and to two anonymous reviewers for their suggestions.
This work was funded by a USDA NRICGP grant (original USDA award no. 95-37303-2043) awarded to M.W.
REFERENCES
- 1.Andrews, J. H. 1992. Biological control in the phyllosphere. Annu. Rev. Phytopathol. 30:603-635. [DOI] [PubMed] [Google Scholar]
- 2.Anjaiah, V., N. Koedam, B. Nowak-Thompson, J. E. Loper, M. Hofte, J. T. Tambong, and P. Cornelis. 1998. Involvement of phenazines and anthranilate in the antagonism of Pseudomonas aeruginosa PNA1 and Tn5 derivatives toward Fusarium spp. and Pythium spp. Mol. Plant-Microbe Interact. 11:847-854. [Google Scholar]
- 3.Bonn, W. G., R. D. Gitaitis, and B. H. MacNeill. 1985. Epiphytic survival of Pseudomonas syringae pv. tomato on tomato transplants shipped from Georgia. Plant Dis. 69:58-60. [Google Scholar]
- 4.Burr, T. J., M. C. Matteson, C. A. Smith, M. R. Corral-Garcia, and T. C. Huang. 1996. Effectiveness of bacteria and yeasts from apple orchards as biological control agents of apple scab. Biol. Control 6:151-157. [Google Scholar]
- 5.Carpena-Ruiz, R., A. Sopena, and A. M. Ramon. 1989. Extraction of free amino acids from tomato leaves. Plant Soil 119:251-254. [Google Scholar]
- 6.Davies, J. N., and G. E. Hobson. 1981. The constituents of tomato fruit—the influence of environment, nutrition, and genotype. Crit. Rev. Food Sci. Nutr. 15:205-280. [DOI] [PubMed] [Google Scholar]
- 7.De Wit, C. T. 1960. On competition. Versl. Landbouwkd. Onderz. 66:1-82. [Google Scholar]
- 8.Dianese, A. C. 1997. Importance of nutritional similarity in pre-emptive biocontrol of bacterial spot. M.S. thesis. Auburn University, Auburn, Ala.
- 9.Fokkema, N. J. 1971. The effect of pollen in the phyllosphere of rye on colonisation by saprophytic fungi and on infection by Helminthosporium sativum and other leaf pathogens. Neth. J. Plant Pathol. 77(Suppl.1):1-60. [Google Scholar]
- 10.Fokkema, N. J. 1981. Fungal leaf saprophytes, beneficial or detrimental?, p. 433-454. In J. P. Blakeman (ed.), Microbial ecology of the phylloplane. Academic Press, London, United Kingdom.
- 11.Fravel, D. R. 1988. Role of antibiosis in the biological control of plant diseases. Annu. Rev. Phytopathol. 26:75-91. [Google Scholar]
- 12.King, E. O., M. K. Ward, and D. E. Raney. 1954. Two simple media for the demonstration of pyocyanin and fluorescin. J. Lab. Clin. Med. 44:301-307. [PubMed] [Google Scholar]
- 13.Leveau, J. H. J., and S. E. Lindow. 2001. Appetite of an epiphyte: Quantitative monitoring of bacterial sugar consumption in the phyllosphere. Proc. Natl. Acad. Sci. USA 98:3446-3453. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Lindemann, J., D. C. Arny, and C. D. Upper. 1984. Use of an apparent infection threshold population of Pseudomonas syringae to predict incidence and severity of brown spot of bean. Phytopathology 74:1334-1339. [Google Scholar]
- 15.Lindow, S. E. 1988. Lack of correlation of in vitro antibiosis with antagonism of ice nucleation active bacteria on leaf surfaces by non-ice nucleation active bacteria. Phytopathology 78:444-450. [Google Scholar]
- 16.Lindow, S. E. 1993. Novel method for identifying bacterial mutants with reduced epiphytic fitness. Appl. Environ. Microbiol. 59:1586-1592. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Loper, J. E. 1988. Role of fluorescent siderophore production in biological control of Pythium ultimum by a Pseudomonas fluorescens strain. Phytopathology 78:166-172. [Google Scholar]
- 18.Loper, J. E., and S. E. Lindow. 1993. Roles of competition and antibiosis in suppression of plant diseases by bacterial biological control agents, p. 144-155. In R. D. Lumsden and J. L. Vaughn (ed.), Pest management: biologically based technologies. ACS symposium series. American Chemical Society, Washington, D.C.
- 19.McClure, N. C., A. R. Ahmadi, and B. G. Clare. 1998. Construction of a range of derivatives of the biological control strain Agrobacterium rhizogenes K84: a study of factors involved in biological control of crown gall disease. Appl. Environ. Microbiol. 64:3977-3982. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Mercier, J., and S. E. Lindow. 2000. Role of leaf surface sugars in colonization of plants by bacterial epiphytes. Appl. Environ. Microbiol. 66:369-374. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Miller, J. H. 1972. Experiments in bacterial genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.
- 22.Picha, D. H. 1987. Sugar and organic acid content of cherry tomato fruit at different ripening stages. HortScience 22:94-96. [Google Scholar]
- 23.Riphagen, I., N. J. Fokkema, W. J. Kastelein, and P. H. Vereijken. 1979. Effect of aphid honeydew on saprophytic and pathogenic fungi of wheat leaves under controlled environmental conditions. Acta Bot. Neerl. 28:240-241. [Google Scholar]
- 24.Schneider, R. W., and R. G. Grogan. 1977. Bacterial speck of tomato: sources of inoculum and establishment of a resident population. Phytopathology 67:388-394. [Google Scholar]
- 25.Senden, M. H. M. N., A. J. G. M. van der Meer, J. Limborgh, and H. T. Wolterbeek. 1992. Analysis of major tomato xylem organic acids and PITC-derivatives of amino acids by RP-HPLC and UV detection. Plant Soil 142:81-89. [Google Scholar]
- 26.Shepard, D. P., and E. I. Zehr. 1994. Epiphytic persistence of Xanthomonas campestris pv. pruni on peach and plum. Plant Dis. 78:626-629. [Google Scholar]
- 27.Smitley, D. R., and S. M. McCarter. 1982. Spread of Pseudomonas syringae pv. tomato and role of epiphytic populations and environmental conditions in disease development. Plant Dis. 66:713-717. [Google Scholar]
- 28.Stromberg, K. D., L. L. Kinkel, and K. J. Leonard. 1999. Relationship between phyllosphere population sizes of Xanthomonas translucens pv. translucens and bacterial leaf streak severity on wheat seedlings. Phytopathology 89:131-135. [DOI] [PubMed] [Google Scholar]
- 29.Stromberg, K. D., L. L. Kinkel, and K. J. Leonard. 2000. Interactions between Xanthomonas translucens pv. translucens, the causal agent of bacterial leaf streak of wheat, and bacterial epiphytes in the wheat phyllosphere. Biological Control 17:61-72. [Google Scholar]
- 30.Vanneste, J. L., J. Yu, and S. V. Beer. 1992. Role of antibiotic production by Erwinia herbicola Eh252 in biological control of Erwinia amylovora. J. Bacteriol. 174:2785-2796. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Weller, D. M., W. J. Howie, and R. J. Cook. 1988. Relationship between in vitro inhibition of Gaeumannomyces graminis var. tritici and suppression of take-all of wheat by fluorescent pseudomonads. Phytopathology 78:1094-1100. [Google Scholar]
- 32.Wilson, M., H. L. Campbell, P. Ji, J. B. Jones, and D. A. Cuppels. Biological control of bacterial speck of tomato under field conditions at several locations in North America. Phytopathology, in press. [DOI] [PubMed]
- 33.Wilson, M., and S. E. Lindow. 1993. Population dynamics of Pseudomonas fluorescens strain A506 in pear flowers following inoculation in relation to strategies of biological control of fire blight and frost injury. Acta Horticulturae 338:331-332. [Google Scholar]
- 34.Wilson, M., and S. E. Lindow. 1994. Ecological similarity and coexistence of epiphytic ice-nucleating (Ice+) Pseudomonas syringae strains and a non-ice-nucleating (Ice−) biological control agent. Appl. Environ. Microbiol. 60:3128-3137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Wilson, M., and S. E. Lindow. 1994. Coexistence among epiphytic bacterial populations mediated through nutritional resource partitioning. Appl. Environ. Microbiol. 60:4468-4477. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Yuen, G. Y., and M. N. Schroth. 1986. Inhibition of Fusarium oxysporum f. sp. dianthi by iron competition with an Alcaligenes sp. Phytopathology 76:171-176. [Google Scholar]