FEMS Microbiology Letters 242 (2005) 249256 www.fems-microbiology.

org

Antagonistic activity among 2,4-diacetylphloroglucinol-producing uorescent Pseudomonas spp.

Shamil Validov a, Olga Mavrodi b, Leonardo De La Fuente b, Alexander Boronin a, David Weller c, Linda Thomashow c, Dmitri Mavrodi b,*
c

Skryabin Institute of Biochemistry and Physiology of Microorganisms of the Russian Academy of Sciences, Pushchino, Russia b Department of Plant Pathology, Washington State University, 362 Johnson Hall, Pullman, WA 99164-6430, USA USDA Agricultural Research Service, Root Disease and Biological Control Research Unit, Washington State University, Pullman, WA 99164-6430, USA Received 15 September 2004; received in revised form 28 October 2004; accepted 4 November 2004 First published online 18 November 2004 Edited by Y. Okon

Abstract Strains of uorescent Pseudomonas spp. that produce 2,4-diacetylphloroglucinol (2,4-DAPG) dier in their ability to colonize roots. In this study, we screened 47 2,4-DAPG-producing strains representing17 distinct genotypes for antagonistic activity associated with the production of bacteriocins. Upon induction, over 70% of the strains inhibited the growth of other isolates in vitro. Greenhouse assays indicated that populations of sensitive strains in wheat rhizosphere soil declined more rapidly in the presence of antagonists than when introduced alone. Antagonism can inuence the ability of biocontrol agents to establish and maintain eective population densities in situ. 2004 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved.
Keywords: Antagonism; Rhizosphere; Pseudomonas; Biological control; Bacteriocin

1. Introduction Fluorescent Pseudomonas spp. that produce 2,4diacetylphloroglucinol (2,4-DAPG) can provide biological control of soilborne pathogens on a wide range of crops, and they have a key role in the suppressiveness of some soils to plant pathogens [1,2]. Although most strains studied to date are members of the species P. uorescens, they nevertheless are phenotypically and genotypically diverse [3,4]. Among the major dierences associated with biological control are the capacCorresponding author. Tel.: +1 509 335 3269; fax: +1 509 335 7674. E-mail address: mavrodi@mail.wsu.edu (D. Mavrodi).
*

ity to produce pyoluteorin and pyrrolnitrin and the ability to colonize and persist in the rhizosphere. Colonization and persistence are critical because a successful biocontrol agent must establish and maintain a minimum threshold population density in order to protect the host. We recently used genomic subtraction to explore dierences between the genomes of two closely related 2,4-DAPG-producing strains diering in their ability to colonize the rhizosphere of wheat [5]. Among DNA fragments present in the superior colonizer Q8r196 but not in the less rhizosphere-competent strain Q287 was SSH6, a clone with similarity to colicin M from Escherichia coli. This nding prompted us to explore antagonistic activity among related 2,4-DAPG-producing isolates.

0378-1097/$22.00 2004 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.femsle.2004.11.013

250

S. Validov et al. / FEMS Microbiology Letters 242 (2005) 249256

Antagonism between microorganisms is mediated by antibiotics, lytic enzymes, exotoxins and bacteriocins, the last of which represent the most abundant class of bacterial defensive factors. Bacteriocins often are dened as proteinaceous compounds that are produced by certain strains and kill closely related strains or species [6]. Bacteriocin production by Gram-negative bacteria has been studied most extensively in enterobacteria and P. aeruginosa. However, bacteriocin-producing isolates occur frequently among plant-associated uorescent Pseudomonas spp., and it has been proposed that such strains may have a competitive advantage in the rhizosphere [7,8]. Many of the ndings to date are based on studies in vitro, and the impact of bacteriocins on the survival of uorescent Pseudomonas spp. in situ remains largely unexplored. In this study, we assessed antagonistic activity among 47 dierent 2,4-DAPG-producing strains representing 17 dierent genotypes distinguished by rep-PCR [3]. Based on inhibition studies in vitro, we utilized ve of these strains to evaluate the impact of antagonism on competition in the rhizosphere of wheat.

lowing concentrations: kanamycin, 25 lg/ml; rifampicin, 100 lg/ml; cycloheximide, 100 lg/ml; chloramphenicol, 13 lg/ml; ampicillin, 40 lg/ml; kanamycin, 25 lg/ml, and gentamycin, 12 lg/ml. Strains tagged with miniTn7 were spontaneous rifampicin-resistant derivatives [9]. 2.2. DNA manipulations and analyses Standard techniques were used for plasmid DNA isolation, agarose gel electrophoresis, and transformation [10]. PCR amplication was carried out with PlatinumTaq DNA polymerase (Life Technologies, USA) according to the manufacturers recommendations. DNA was sequenced using an ABI PRISM BigDye Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems, USA), and sequence data were analyzed with OMIGA 2.0 software (Accelrys, USA). The oligonucleotide primer glnSmod was designed with Oligo 6.0 software (Molecular Biology Insights, USA). 2.3. Antagonism in vitro Strains of Pseudomonas spp. were screened for bacteriocin production as described by Govan [13]. Each tester strain was spread across TSB agar in a band approximately 2.5 cm wide. After 18 h at 28 C, the plate was irradiated at 254 nm for 1 min and incubated at 28 C for another 3.5 h. Bacterial growth was scraped from the agar with a cotton swab soaked in chloroform and the plates were exposed to chloroform vapor to kill all remaining bacteria. Samples (15 ll) from overnight cultures of indicator strains in TSB were then streaked perpendicular to the tester band (see Fig. 1). Alternatively, tester cultures grown at 28 C in 10 ml of LB broth were collected by centrifugation and suspended in 1 ml of LB broth amended with 1 lg ml1 of mitomycin C to induce bacteriocin production. The cultures were shaken for 5 hr and lysed with 100 ll of chloro-

2. Materials and methods 2.1. Bacterial strains and plasmids Most of the strains used in the study were described previously [3,4,9]. P. uorescens EPS808 and EPS817 were a kind gift from Dr. E. Montesinos (University of Girona, Girona, Spain). P. uorescens PHL1C2 was kindly provided by Dr. P. Lemanceau (Universite de Bourgogne, Dijon, France). Other strains and plasmids are listed in Tables 1 and 2. Bacterial cultures were grown in LuriaBertani (LB) medium [10], Tryptic Soy Broth (TSB) (Difco Laboratories, USA), MMP minimal medium [11] or one-third strength Kings medium B (KMB) [12] at 28 C. Antibiotics were used at the fol-

Table 1 Bacterial strains and plasmids used in this study Strain P. uorescens Pf-5Km Q2-87Km Q8r1-96Km Q8r1-96Gm FTAD1R36Gm FTAD1R34Gm Plasmids pBK- mini-Tn7-gfp 1 pBK- mini-Tn7-gfp 2 pUX-BF13 Descriptiona Pf-5 tagged with mini-Tn7-gfp 1. Phl+ Rifr Kmr Q2-87 tagged with mini-Tn7-gfp 1. Phl+ Rifr Kmr Q8r1-96 tagged with mini-Tn7-gfp 1. Phl+ Rifr Kmr Q8r1-96 tagged with mini-Tn7-gfp 2. Phl+ Rifr Gmr FTAD1R36 tagged with mini-Tn7-gfp 2. Phl+ Rifr Gmr FTAD1R34 tagged with mini-Tn7-gfp 2. Phl+ Rifr Gmr pUC19-derived delivery vector for mini-Tn7-gfp 1. Kmr cat bla gfp Mob+ pUC19-derived delivery vector for mini-Tn7-gfp 2. Gmr cat bla gfp Mob+ Donor of Tn7 transposase. tnsABCDE R6K bla Mob+ Reference or source This This This This This This [18] [18] [16] study study study study study study

a Phl+, the strain produces 2,4-diacetylphloroglucinol; Rifr, rifampin resistance; Kmr, kanamycin resistance; Gmr, gentamycin resistance; cat, chloramphenicol acetyltransferase; bla, b-lactamase.

S. Validov et al. / FEMS Microbiology Letters 242 (2005) 249256 Table 2 In vitro antagonism among 2,4-diacetylphloroglucinol-producing Pseudomonas spp. Genotypea A B C D E F G H I J K L M N O P Q
a b

251

No. of surveyed strains 3 2 2 11 4 2 1 1 1 2 3 3 3 1 2 4 2

No. of strain-antagonists within the genotype 2 0 2 9 4 2 1 0 1 2 1 1 2 1 2 4 2

No. of strains that are inhibited by the genotype 3 0 13 19 11 3 3 0 16 3 5 4 12 1 10 18 12

Breadth of killingb 2 (11.8) 0 3 (17.6) 7 (41.2) 8 (47.1) 2 (11.8) 3 (17.6) 0 5 (29.4) 1 (5.9) 4 (23.5) 1 (5.9) 4 (23.5) 1 (5.9) 6 (35.3) 13 (76.5) 3 (17.6)

Genotypes were dened previously by BOX-PCR genomic ngerprinting [3,4]. No. of genotypes that are inhibited by the surveyed genotype. Values in parenthesis indicate percentages, where 100% corresponds to 17 dierent genotypes represented in the collection.

form. After mixing and centrifugation, the aqueous phase was spotted on LB agar and overlayed with soft agar containing a single indicator strain. Lysates also were tested after trypsin digestion, freeze-thawing, and Microcon column ltration as described by Riley et al. [14] in order to distinguish between the production of low-molecular weight (S-pyocin-like) and phage tail-like (R- and F-pyocin-like) bacteriocins and bacteriophages. Plates for all assays in vitro were incubated at 28 C and scored for for the presence (+) or absence () of inhibition at 12 and 24 h. Each tester strain was replicated once per experiment and the entire collection was screened twice. The data were combined and a strain was considered to be an antagonist only if the results of both assays agreed. The combined results were used to produce a two-dimensional rectangular matrix of binary codes that was analyzed with MVSP 3.12 software (Kovach Computing Services, UK) using a simple matching coecient that considers both the presence and the absence of antagonistic activity. 2.4. Transposon tagging Strains of P. uorescens were tagged by electroporation [15] with 300 ng each of pBK-mini-Tn7 gfp 1 or pBK-mini-Tn7 gfp 2 (Table 1) and the helper plasmid pUX-BF13 [16]. The tagged clones were isolated on LB agar amended with kanamycin or gentamycin. Tn7 preferentially inserts in a single orientation into a specic neutral intergenic site, att Tn7, present in many eubacterial genomes [17]. To conrm the site of transposon insertion in our strains, we amplied the region anking the transposon by PCR with the oligonucleo-

Fig. 1. An example of a plate from the inhibition assay. The test strain, P. uorescens Q8r1-96, was streaked across the surface of tryptic soy agar to a width of approximately 2.5 cm (plate D). After incubation for 18 h at 28 C the bacteria were scraped from the agar and the plates were exposed to chloroform vapor to kill the remaining bacteria. Fifteen microliters of an overnight culture of each indicator strain in TSB were then applied across the original inoculum (plates A, B, and C). The plates were incubated for an additional 24 h at 28 C and examined for zones of inhibition (summarized in Table 2). On this gure, four strains inhibited by Q8r1-96 are: FTAD1R34 (1), HT5-1 (2), Q2-87 (3), and ATCC49054 (4).

tide primers cat [18] and glnSmod (5 0 -AAY CTS GCS AAG TCG GTS AC-3 0 ), which are specic to the mini-Tn7-borne chloramphenicol acetyltransferase gene and the 3 0 -end of the glutamine synthetase gene, respectively. The cycling included a 2-min initial denaturation at 94 C, followed by 6 cycles of 94 C for 30 s, 52 C for 15 s, and 72 C for 30 s, followed by 29 more cycles with

252

S. Validov et al. / FEMS Microbiology Letters 242 (2005) 249256

the annealing temperature of 60 C and a nal extension at 72 C for 5 min. The 0.65-kb amplied fragments were puried with a QIAEX II gel extraction kit (Qiagen, USA) and sequenced with the primers cat and glnSmod. 2.5. Rhizosphere colonization and competition Rhizosphere colonization assays were performed with wheat (Triticum aestivum L.) cv. Penawawa in non-sterile soil as described by Landa et al. [19]. Quincy virgin soil was inoculated with bacteria in a 1% suspension of methylcellulose to give 1 104 CFU g1 of soil (fresh weight). Mixed inoculation treatments contained a 1:1 mixture of competing strains (0.5 104 CFU g1 of soil for each strain). Plants were incubated in a growth chamber for three successive 3-week cycles at 15 1 C with a 12-h photoperiod. After each cycle, the population size of the introduced strains was determined on the roots of six randomly selected plants. Each treatment was replicated three times and the entire experiment was repeated twice. Population densities of introduced strains were enumerated by PCR, using a modied dilution endpoint assay [3] with an extra step in which bacteria were selected in kanamycin- or gentamycin-amended media to distinguish between strains in mixed inoculation treatments and to check for cross-contamination of single introduced strains. Bacterial growth was assessed after 72 h, with an OD600 P 0.1 considered as positive. 2.6. Statistical analyses All treatments in the growth chamber were arranged in a randomized complete block design. Because rhizobacteria on roots are lognormally distributed [20], population data were converted to log CFU g1 (fresh weight) of soil or root to satisfy the normality assumption of analysis of variance (ANOVA). Data were analyzed using STATISTIX (version 8.0, Analytical Software, USA). Dierences in population densities among treatments were determined by ANOVA, and mean comparisons among treatments were performed by Fishers protected least signicance dierence (LSD) test at P = 0.05.

onistic activity also positively correlated with the breadth of killing, i.e., the percentage of the 17 genotypes inhibited. Cluster analyses indicated that the P, D and E genotypes harbored the most active antagonists (Fig. 2(a)). The results reect to some extent the unbalanced nature of our strain collection, in which certain genotypes are represented by multiple strains while others are represented only once. Nevertheless, the single Igenotype strain FTAD1R36 was among the most active antagonists, inhibiting 16 strains from four dierent genotypes (Table 2). Similarly, the P genotype, represented by four strains, inhibited almost twice as many other genotypes as did the D genotype, represented by 11 isolates. Other highly active strains included MVW43 (Q genotype), PILH1 (M genotype), and STAD376 (C genotype). At the other extreme, members of the B and H genotypes failed to antagonize any of the tested strains, nor did strains CHA0 (A genotype), Q2-5 (D genotype), ATCC49054 (D genotype), EPS808 (K genotype), EPS817 (K genotype), PHL1C2 (M genotype), W4-4 (L genotype), and 1M1-96 (L genotype) (Fig. 2(b)). Cluster analyses based on patterns of sensitivity revealed that I, G, and J genotype strains were among the most resistant while strains of the B, D, E, and P genotypes were on average more susceptible to inhibition by other phlD+ strains (data not shown). 3.2. Competition between tagged strains in the wheat rhizosphere Based on their behavior in vitro, strains Pf-5, Q2-87, Q8r1-96, FTAD1R34, and FTAD1R36 were utilized to assess the impact of antagonistic activity on strain survival in the rhizosphere of wheat. Neither Pf-5 (A genotype) nor Q2-87 (B genotype) antagonized other strains in vitro, and Q2-87 but not Pf-5 was antagonized by Q8r1-96 (D genotype). Q8r1-96 also antagonized FTAD1R34 (D genotype) and, along with strain Pf-5, was antagonized by FTAD1R36 (I genotype). These strains were tagged either with kanamycin or gentamycin resistance genes [18] by using disarmed mini-Tn7 transposons that inserted into the genome immediately downstream of glnS. Growth studies at 27 C in onethird strength KMB and MMP revealed no dierences in the growth kinetics of the tagged derivatives and their respective wild-type parents (data not shown). Population densities of single strains or dierentiallymarked strain pairs did not dier signicantly (P = 0.05) immediately after inoculation (cycle 0) and ranged from log 3.0 to 3.5 CFU g1 of bulk soil (Fig. 3). However, after one cycle, large dierences in the population sizes of antagonist and sensitive strains were observed on roots in soil that had been co-inoculated with a mix of Q8r1-96 and Q2-87 or FTAD1R36 and Pf-5. Population densities of the antagonists in these mixtures generally were unaected by the presence of the sensitive strains,

3. Results 3.1. Distribution of antagonistic activity among phlD+ Pseudomonas spp. Inhibition studies in vitro revealed that antagonistic activity is widely distributed among phlD Pseudomonas spp. (Table 2). Of 47 tested strains, only 11 failed to antagonize another strain. For most genotypes the antag-

S. Validov et al. / FEMS Microbiology Letters 242 (2005) 249256

253

E P D K G M

I
J L N H B A C F Q O
0.4 0.5 0.6 0.7 0.8 0.9 1

(a)

Simple Matching Coefficient

STAD376(C) PILH1(M) FTAD1R36(I) MVW4-3(Q) OC4-1(D) MVP1-4r(P) 7MA20(O) Q2-2(E) QT1-6(E) F113(K) 6WSU4(P) MVW1-1(P) STAD384(C) 4MA6(P) QT1-5(D) L5-1(D) 7MA12(O) MARV1(L) FFL1R18(G) JMP7(F) JMP6(F) MVP1-6(D) Pf-5(A) Q37-87(E) MVW4-2(Q) D27B1(M) MVW1-2(D) FTAD1R34(D) FFL1R8(J) 5MR2(E) Q8r1-96(D) Q128-87(D) HT5-1(N) FFL1R9(D) MVP1-3(A) FFL1R22(J) W4-4(L) Q2-5(D) CV1-1(H) EPS817(K) Q2-1(B) PHL1C2(M) CHA0(A) 1M1-96(L) ATCC49054(D) Q2-87(B) EPS808(K)

0.7

0.75

0.8

0.85

0.9

0.95

(b)

Simple Matching Coefficient

Fig. 2. Cluster analyses of antagonistic phenotypes within a representative collection of phlD-positive Pseudomonas spp. Patterns of inhibition were clustered by rep-PCR genotype (a) or by individual bacterial strain (b). The UPGMA tree was generated with MVSP 3.12 from a similarity matrix generated using the simple matching coecient.

while populations of the sensitive strains (Q2-87 and Pf5) dropped below the detection limit of approximately log 3.26 CFU g1 of root after only one cycle and never recovered (Fig. 3(a) and (e)). No such rapid decline was

observed in single inoculations of Q2-87 or Pf-5, and for Q2-87, the dierence between population densities in single and mixed inoculations after one cycle was over log 3.0 CFU g1 of root.

254
Q8r1-96(A) + Q2-87(S)

S. Validov et al. / FEMS Microbiology Letters 242 (2005) 249256

Pf-5(N) + Q8r1-96(N)

population density ( logCFU g-1 of root)

9 8 7 6 5 4 3 2 1

(a)

(b)

Q8r1-96(A) + FTAD1R34(S) FTAD1R36(A) + Q8r1-96(S) 9 8 7 6 5 4 3 2 1 9 8 7 6 5 4 3 2 1

(c)

(d)

0 FTAD1R36(A) + Pf-5(S)

Cycle
Q8r1-96 Q2-87 Pf-5 FTAD1R34 FTAD1R36

Rhizosphere

(e)

Cycle

Fig. 3. Population dynamics of selected pairs of mini-Tn7-tagged strains on roots of wheat cv. Penawawa grown in non-sterile Quincy virgin soil for three successive cycles of 3 weeks each. For each pair of competing strains the letters in parentheses indicate antagonistic (A), sensitive (S), or neutral (N) phenotypes observed in vitro. The competing strain pairs were: Q8r1-96Gm and Q2-87Km (panel a); Q8r1-96Gm and Pf-5Km (panel b); Q8r1-96Km and FTAD1R34Gm (panel c); Q8r1-96Km and FTAD1R36Gm (panel d); and FTAD1R36Gm and Pf-5Km (panel e). Open symbols correspond to treatments inoculated with a single strain at a density of approximately 1 104 CFU g1 soil. Filled symbols indicate treatments that were inoculated by a 1:1 mixture of strains, each at a density of approximately 0.5 104 CFU g1 soil. The control treatment consisted of soil amended with a 1% methylcellulose suspension. Mean and standard deviations for one experiment are presented. Dotted lines represent cycles where population densities of strains Q2-87 (panel a) and Pf-5 (panels b and e) fell in mixed inoculations below the detection limit of assay.

In paired inoculations, strain Q8r1-96 outcompeted the sensitive strains (i.e., Q2-87 and FTAD1R34), as well as the supposed antagonist FTAD1R36 (Fig. 3(a), (c) and (d)). It also displaced Pf-5 (Fig. 3(b)), the least rhizosphere-competent strain in this study, despite the fact that these two strains were mutually non-antagonistic in vitro. In all experiments, Q8r1-96 was largely unaffected by the presence of a competitor and maintained population densities equal to or greater than log 6.0 CFU g1 of root in both single and mixed inoculations.

4. Discussion DAPG-producing uorescent Pseudomonas spp. of the D genotype are among the most aggressive root-

colonizing biocontrol agents studied to date. On the other hand, D-genotype strains are virtually indistinguishable from members of most other genotypes when grown in vitro, suggesting that relatively few dierences at the genetic level might be sucient to confer the colonization phenotype typical of premier PGPR [3]. As one approach to identify novel genes associated with this trait, we isolated DNA fragments present in the D-genotype strain Q8r1-96 but not in an average colonizer, Q2-87 (B genotype) [5]. One subtracted gene, SSH6, exhibited similarity to the bacteriocin colicin M and later was found to form part of a 17-kb pyocin-like locus in the genome of Q8r1-96 (data not shown). Bacteriocins of pseudomonads are best characterized in P. aeruginosa, where they traditionally were called pyocins [21]. As in other eubacteria, these bacteriocins include both low- and high-molecular weight species. The low molecular-weight bacteriocins, or S-pyocins, are protease-sensitive enzymes that kill other strains through their DNase, RNase, or membrane pore-forming activity. The high molecular-weight forms comprise two categories, R- and F-type pyocins, that closely resemble non-exible contractile and exuous non-contractile bacteriophage tails, respectively. In P. aeruginosa, the R- and F-type pyocins are related to two dierent lineages of bacteriophages. The high molecular-weight pyocins are thought to kill sensitive cells through depolarization of the cytoplasmic membrane. Although bacteriocins generally are associated with bacterial competitiveness in the environment, this association represents a relatively unexplored topic and we are aware of only one other recent study in which plant-associated Pseudomonas spp. were screened for bacteriocinogenic activity [7]. Most of the strains we tested were antagonistic to other uorescent Pseudomonas spp. in vitro after induction. It is unlikely that this activity was due to antibiotics because these strains previously were evaluated for the production of antibiotics other than 2,4-DAPG [4]. Although we considered antagonism to be a single phenotypic trait in this work, it could well be due to multiple low-molecular weight and/or phage tail-like bacteriocins acting in concert [8,21]. In fact, analyses of recently sequenced microbial genomes indicate that isolates of P. uorescens have the potential to produce numerous bacteriocins [8]. To distinguish between lowmolecular weight and phage tail-like bacteriocins and bacteriophages, we subjected mitomycin C-induced lysates to trypsin digestion, freeze-thawing, and ultraltration. In most cases, antagonistic activity was associated with a high molecular weight, protease-resistant fraction, implicating phage tail-like bacteriocins and bacteriophages, and indeed, electron microscopy revealed pyocin-like particles in mitomycin C-induced lysates of some strains.

S. Validov et al. / FEMS Microbiology Letters 242 (2005) 249256

255

Bacteriocins generally are considered to be intraspecies competitiveness factors that may help producing strains to invade a new ecological niche and/or repel other strains from an already occupied niche [6,22]. However, despite the abundance of data from studies in vitro, the ecological role(s) of bacteriocins in nature remains obscure. We therefore were particularly interested in the impact of antagonism on the competitiveness of our isolates in the rhizosphere of wheat, a niche to which they should be well-adapted. In all rhizosphere experiments except for FTAD1R36+Q8r1-96 (Fig. 3(d)), the antagonist displaced the competing sensitive strain (Fig. 3). The eect generally was quite pronounced, with the population density of sensitive strains decreasing by over log 3.0 CFU g1 and, in the case of pairs Q8r1-96+Q2-87 and FTAD1R36+Pf-5, quickly dropping below the detection limit. These shifts were consistent from experiment to experiment and may be indicative of the inhibition of one strain by another in the plant rhizosphere. The results also agree with ndings of Landa et al. [19], who examined the interaction between strains Q8r1-96 and Q2-87 in a de Wit replacement series experiment in which wheat was grown in soil into which both strains had been introduced over a range of dened ratios. The nal population size of Q2-87 in mixed inoculations was lower than predicted even when the strains were at a ratio of 0.3(Q8r1-96):0.7(Q2-87), consistent with antagonism of Q2-87 by Q8r1-96 [19]. Surprisingly, P. uorescens Q8r1-96 displaced every counterpart in the rhizosphere regardless of whether it was antagonistic (inhibition of Q2-87 and FTAD1R34), sensitive (sensitivity to FTAD1R36) or neutral (no reaction to Pf-5) to those strains in vitro. Earlier rhizosphere competition studies [19] also revealed that Q8r1-96 outcompeted MVP1-4r and 1M1-96 but was itself displaced from the rhizosphere of wheat by another 2,4-DAPGproducing strain, P. uorescens F113. All of these strains are neutral to one another in vitro. A number of theoretical and in vitro studies predict that for otherwise equivalent strains, the outcome of competition between a bacteriocin-producer and a sensitive strain will depend on their initial proportions in the system [23]. These studies suggest that if a bacteriocinproducer is introduced at a low level, it will not invade successfully and its population will eventually shrink to zero. We speculate that Q8r1-96 may colonize roots more aggressively than other strains not only because of its antagonistic activity, but also because it rapidly establishes basal populations that utilize the available resources (e.g., root exudates) more eciently. As a consequence, it grows faster, eventually overcoming even bacteriocin-producing competitors. Although in our study we introduced competing strains into soil at a 1:1 ratio, dierences in growth rates in the rhizosphere might have contributed to the decline in the popu-

lation of the antagonist in some treatments (i.e., FTAD1R36+Q8r1-96). Bacteriocin production clearly is not the major determinant of successful rhizosphere colonization by a particular bacterial strain, but our results suggest that it may contribute to competition between closely related strains of Pseudomonas spp., especially when the competitors are present at certain favorable ratios.

Acknowledgements Shamil Validov, Olga Mavrodi and Leonardo De La Fuente contributed equally to this study. The authors thank Greg Phillips and Karen Hansen (USDA-ARS Root Disease and Biocontrol Unit, Pullman, WA) for help with the bacteriocin screening assays. This work was supported by the U. S. Department of Agriculture, National Research Initiative, Competitive Grants Program (grant 03-35319-13800).

References
[1] Raaijmakers, J.M. and Weller, D.M. (2001) Exploiting genotypic diversity of 2,4-diacetylphloroglucinol-producing Pseudomonas spp.: characterization of superior root-colonizing P. uorescens strain Q8r1-96. Appl. Environ. Microbiol. 67, 25452554. [2] Weller, D.M., Raaijmakers, J.M., Gardener, B.B.M. and Thomashow, L.S. (2002) Microbial populations responsible for specic soil suppressiveness to plant pathogens. Ann. Rev. Phytopathol. 40, 309348. [3] McSpadden Gardener, B.B., Schroeder, K.L., Kalloger, S.E., Raaijmakers, J.M., Thomashow, L.S. and Weller, D.M. (2000) Genotypic and phenotypic diversity of phlD-containing Pseudomonas strains isolated from the rhizosphere of wheat. Appl. Environ. Microbiol. 66, 19391946. [4] Mavrodi, O.V., McSpadden Gardener, B.B., Mavrodi, D.V., Bonsall, R.F., Weller, D.M. and Thomashow, L.S. (2001) Genetic diversity of phlD from 2,4-diacetylphloroglucinol-producing uorescent Pseudomonas spp. Phytopathology 91, 3543. [5] Mavrodi, D.V., Mavrodi, O.V., McSpadden Gardener, B.B., Landa, B.B., Weller, D.M. and Thomashow, L.S. (2002) Identication of dierences in genome content among phlD-positive Pseudomonas uorescens strains by using PCR-based subtractive hybridization. Appl. Environ. Microbiol. 68, 51705176. [6] Riley, M.A. and Wertz, J.E. (2002) Bacteriocins: evolution, ecology, and application. Ann. Rev. Microbiol. 56, 117137. [7] Parret, A.H.A. and De Mot, R. (2000) Bacteriocin production by rhizosphere-colonizing uorescent Pseudomonas. In: Proceedings of the 5th International PGPR Workshop. Cordoba, Argentina. [8] Parret, A.H.A. and De Mot, R. (2002) Bacteria killing their own kind: novel bacteriocins of Pseudomonas and other c-proteobacteria. Trends Microbiol. 10, 107112. [9] Landa, B.B., de Werd, H.A.E., McSpadden-Gardener, B.B. and Weller, D.M. (2002) Comparison of three methods for monitoring populations of dierent genotypes of 2,4-diacetylphloroglucinolproducing Pseudomonas uorescens in the rhizosphere. Phytopathology 92, 129137. [10] Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.G., Smith, J.A. and Struhl, K. (1995) Short Protocols in Molecular Biology. John Wiley & Sons, Inc., New York, NY.

256

S. Validov et al. / FEMS Microbiology Letters 242 (2005) 249256 [18] Koch, B., Jensen, L.E. and Nybroe, O. (2001) A panel of Tn7-based vectors for insertion of the gfp marker gene or for delivery of cloned DNA into Gram-negative bacteria at a neutral chromosomal site. J. Microbiol. Meth. 45, 187 195. [19] Landa, B.B., Mavrodi, D.V., Thomashow, L.S. and Weller, D.M. (2003) Interactions between strains of 2,4-diacetylphloroglucinolproducing Pseudomonas uorescens in the rhizosphere of wheat. Phytopathology 93, 982994. [20] Loper, J.E., Suslow, T.V. and Schroth, M.N. (1984) Lognormal distribution of bacterial populations in the rhizosphere. Phytopathology 74, 14541460. [21] Michel-Briand, Y. and Baysse, C. (2002) The pyocins of Pseudomonas aeruginosa. Biochimie 84, 499510. [22] Riley, M.A. and Gordon, D.M. (1999) The ecological role of bacteriocins in bacterial competition. Trends Microbiol. 7, 129 133. [23] Durret, R. and Levin, S. (1997) Allelopathy in spatially distributed populations. J. Theor. Biol. 185, 165171.

[11] Loper, J.E. (1988) Role of uorescent siderophore production in biological control of Pythium ultimum by a Pseudomonas uorescens strain. Phytopathology 78, 166172. [12] King, E.O., Ward, M.K. and Raney, D. (1954) Two simple media for the demonstration of pyocyanin and uorescein. J. Lab. Clin. Med. 44, 301307. [13] Govan, J.R.W. (1978) Pyocin typing of Pseudomonas aeruginosa. Method Microbiol. 10, 6191. [14] Riley, M.A., Goldstone, C.M., Wertz, J.E. and Gordon, D. (2003) A phylogenetic approach to assessing the targets of microbial warfare. J. Evolution. Biol. 16, 690697. [15] Enderle, P.J. and Farwell, M.A. (1998) Electroporation of freshly plated Escherichia coli and Pseudomonas aeruginosa cells. Biotechniques 25, 954958. [16] Bao, Y., Lies, D.P., Fu, H. and Roberts, G.P. (1991) An improved Tn7-based system for the single-copy insertion of cloned genes into chromosomes of Gram-negative bacteria. Gene 109, 167168. [17] DeBoy, R.T. and Graig, N.L. (2000) Target site selection by Tn7: att Tn7 transcription and target activity. J. Bacteriol. 182, 33103313.