Root-Secreted Malic Acid Recruits Beneﬁcial Soil Bacteria 1[C][W][OA]

Beneﬁcial soil bacteria confer immunity against a wide range of foliar diseases by activating plant defenses, thereby reducing a plant’s susceptibility to pathogen attack. Although bacterial signals have been identiﬁed that activate these plant defenses, plant metabolites that elicit rhizobacterial responses have not been demonstrated. Here, we provide biochemical evidence that the tricarboxylic acid cycle intermediate L -malic acid (MA) secreted from roots of Arabidopsis ( Arabidopsis thaliana ) selectively signals and recruits the beneﬁcial rhizobacterium Bacillus subtilis FB17 in a dose-dependent manner. Root secretions of L -MA are induced by the foliar pathogen Pseudomonas syringae pv tomato (Pst DC3000) and elevated levels of L -MA promote binding and bioﬁlm formation of FB17 on Arabidopsis roots. The demonstration that roots selectively secrete L -MA and effectively signal beneﬁcial rhizobacteria establishes a regulatory role of root metabolites in recruitment of beneﬁcial microbes, as well as underscores the breadth and sophistication of plant-microbial interactions.

Many bacterial and fungal pathogens are not restricted to infecting aerial or root tissues exclusively. As such, communication between aboveground and belowground components can confer a survival advantage and potentially limit or prevent diseases. Current literature suggests that the metabolic levels of low-M r compounds play a significant role in triggering plant innate defense responses (Klessig et al., 2000;Hü ckelhoven, 2007). A corresponding induced defense response initiating intraplant signaling between leaves and roots has been implicated in herbivory (van Tol et al., 2001;Rasmann et al., 2005). Indeed, plant basal levels of shoot defenses were significantly altered following not only root herbivory, but also artificial damage and application of plant defense hormones (van Dam et al., 2003;Soler et al., 2005;Rasmann and Turlings, 2007). Others have shown that shoot herbivory attack resulted in an increase in root concentrations of nicotine and protease inhibitors in Nicotiana attenuata and glucosinolates in Brassica campestris (Baldwin et al., 1994;Ludwig-Mü ller et al., 1997;van Dam et al., 2003;Soler et al., 2007). The importance of chemical signals as distress indicators has greatly benefited from pioneering work on volatile signaling (De Moraes et al., 1998, 2001Kessler and Baldwin, 2001). Under herbivory attack, plants relay compounds such as hormones, exogenous volatile organic compounds, and non-horm1 sary metabolites as longdistance root-to-shoot signals (Erb et al., 2008). Wounding of plant tissues by insect feeding triggers the release of well-characterized volatile signals that attract natural enemies of insect herbivores (De Moraes et al., 2001;Kessler and Baldwin, 2001). Additionally, leaves of the donor plants also perceive intraplant volatile signals to adjust their defensive phenotype (Heil and Silva-Bueno, 2007). Apart from these studies on volatile distress signaling, there are very few studies (van Tol et al., 2001;Rasmann et al., 2005;Hiltpold and Turlings, 2008) on how plants respond by secreting chemicals in response to herbivore attack to recruit beneficial entomopathogenic nematodes to the rhizosphere. While considerable data exist on the occurrence of aboveground/belowground communication in the case of plant herbivory, evidence of similar phenomena in plant-pathogenic bacteria interactions is lacking.
Plants use an array of metabolites to defend themselves against harmful organisms and to attract others that are beneficial. For example, it has been widely documented in the case of root-rhizobial and rootmycorrhizal interactions that roots secrete secondary metabolites that act as messengers to attract Rhizobium and arbuscular mycorrhizal fungi (Kent Peters and Long, 1988;Besserer et al., 2006). Despite progress 1 This work was supported by the University of Delaware Research Foundation and a National Science Foundation award (grant no. IOS-0814477 to H.P.B.), and in part by the Welch Foundation (grant no. D-1478 P.W.P.). * Corresponding author; e-mail hbais@udel.edu. The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Harsh P. Bais (hbais@udel.edu).
[C] Some figures in this article are displayed in color online but in black and white in the print edition.
[W] The online version of this article contains Web-only data. [OA] Open Access articles can be viewed online without a subscription.
www.plantphysiol.org/cgi/doi/10.1104/pp.108.127613 toward understanding the symbiotic-plant microbe interactions, little headway has been made in identifying the genetic and biochemical changes responsible for the attraction of nonsymbiotic rhizospheric microbes to plants. Plant associations with plant growthpromoting rhizobacteria (PGPR) provide protection from soil-borne plant pathogens by antagonistic mechanisms (Bais et al., 2004;Cavaglieri et al., 2005). Such bacteria, by colonizing the root surface, can also trigger induced systemic resistance (ISR) in aerial portions of the plant . ISR by rhizobacteria was demonstrated using both Gram-negative and Grampositive bacteria (Baker et al., 1985;Maurhofer et al., 1994;Zhou and Paulitz, 1994;Liu et al., 1995;Leeman et al., 1996;Benhamou and Belanger, 1998;Pieterse et al., 1998;Kloepper et al., 2004). Bacillus subtilis not only has been reported to trigger ISR Ryu et al., 2004), but also has been implicated in promoting plant growth and protection from fungal infection (Dal-Soo et al., 1997;Emmert and Handelsman, 1999;Bacon et al., 2001;Estevez de Jensen et al., 2002;Ryu et al., 2003). Bacterial signals have been identified that activate plant defense responses through the ethylene pathway, independent of the salicylic acid (SA) pathway or the jasmonic acid pathway . However, no plant metabolites involved in a positive feedback mechanism have been identified. Although evidence exists for intraplant communication, to date there have been no reports demonstrating whether plants exude specific chemical signals through their roots to attract beneficial bacteria in the rhizosphere. Furthermore, it is unknown whether shoot infection by pathogenic bacteria induces recruitment of beneficial rhizobacteria to the root surface. Links between interorganism signaling under distress conditions, especially between aboveground and belowground tissues, are poorly understood. Such signaling, although potentially complex due to the involvement of significant physical distances, may be an important and effective strategy in plant defense that has thus far been overlooked. In an effort to address this deficiency, we used Pseudomonas syringae pv tomato (Pst DC3000)-Arabidopsis (Arabidopsis thaliana) and B. subtilis strain FB17 model systems. Here, we report that Pst DC3000-infected Arabidopsis foliage relay chemical signals below ground through root secretions. The root-secreted chemical specifically attracts and enhances FB17 root binding and biofilm formation on infected seedlings. We also show that the chemoattraction and biofilm promotion activity of a secreted component is an enantiomeric-dependent response.

Rhizobacteria Colonization of Roots Stimulated by Leaf Pathogen
To probe how plant pathogen attack may influence the recruitment of beneficial rhizosphere bacteria, root symbiont colonization was measured in the presence and absence of a foliar pathogen. Twenty-day-old Arabidopsis plants were rhizoinoculated with FB17. Subsequently, the plants were infected with Pst DC3000 by pressure infiltration into the leaves. By 5 d postinoculation, leaves infected by Pst DC3000 stimulated biofilm formation of the beneficial rhizobacteria FB17 both qualitatively as determined by confocal microscopy ( Fig. 1A) and quantitatively by colony-forming units (CFU; Fig. 1B). In fact, within 5 d of Pst DC3000 leaf inoculation, a 4-fold increase in FB17 colonization was observed in the roots compared with mock or nonpathogenic P. syringae pv phaseolicola (NPS3121) treatments [F (3,20) = 114.5; P , 0.05]. This observation that aerial infection with Pst DC3000 caused change in root symbiont colonization implicated root exudate involvement in the beneficial microbe recruitment.

Leaf Infection Induces Malic Acid Root Secretions
To examine whether leaf infection can trigger changes in the composition of metabolites from root secretions, the root secretions from ecotype Columbia (Col-0) plants subjected to different aerial bacterial infection treatments, such as Pst DC3000, untreated (control), water injected (mock), or treated with nonpathogenic strain NPS3121, were collected. The root secretions were chemically analyzed by HPLC. Profiles of the concentrated root exudates from Pst DC3000-infected plants revealed a peak that exhibited a significant increase under the Pst DC3000 infection regime. The peak was further characterized by liquid chromatography-mass spectrometry analysis and retention time overlap and determined to be malic acid (MA; Fig. 2A   . MA-specific chemotactic motility of B. subtilis strain FB17 measured by following capillary chemotactic assay. L-MA showed a dosedependent chemotactic attraction of FB17 (A); structure-dependent chemotactic attraction of FB17 (B); and bacterial species-specific attraction to MA (C). Bacterial strains used in the chemotactic assay included B. subtilis strain FB17, A. tumefaciens strain LBA4404, E. carotovora strain AH2, A. rhizogenes strain Arqua-1, A. brasilense strain Cd, P. fluorescens (Pf01), and P. syringae (Pst DC3000). Data are the average of six replicates from two experiments conducted separately.
Pst DC3000 leaf infection compared to control, mock, and NPS3121 treatments (Fig. 2B). This pattern of MA accumulation in root secretions raised the question whether the recruitment of higher populations of FB17 to Pst DC3000-infected plants was caused by the MA root secretions or simply a secondary effect of the interaction itself.

FB17 Exhibits Positive MA Chemotaxis
To evaluate MA's ability to selectively recruit FB17, microbial motility at varying MA doses was measured by capillary chemotaxis assay (Gordillo et al., 2007). FB17 exhibits positive chemotactic behavior in a concentration-dependent manner within the dose range of 5 to 30 mM (Fig. 3A). The concentration range of 5 to 30 mM falls within the biological titers of MA secreted in planta through the root exudates (Fig. 2B). The non-natural D-MA, as well as the 4-carbon oxalic acid assayed at the high L-MA response level of 30 mM, exhibited significantly [F (4,25) = 117.17; P , 0.05] lower efficacies compared to the natural L-MA form, albeit still significantly higher than the water control ( Fig.  3B) with D-MA. To establish whether L-MA specifically chemoattracts FB17, other beneficial bacteria, such as Pseudomonas fluorescens strain Pf01 and Azospirillum brasilense strain Cd, and pathogenic bacteria, such as Pst DC3000, Erwinia carotovora strain AH2, Agrobacterium rhizogenes strain Arqua-1, and Agrobacterium tumefaciens strain LBA4404, were assayed for MA. Our results revealed that none of the bacteria shows any significant motility toward L-MA [F (2,25) = 119.12; P , 0.05] compared to FB17 (Fig. 3C), establishing the specific chemotactic role of L-MA in attracting PGPR strain FB17.

MA Transporter Mutant Fails to Recruit B. subtilis onto the Root Surface
Arabidopsis T-DNA knockout mutant Atalmt1 for MA transporter AtALMT1 deficient in root MA secretion (Hoekenga et al., 2006) was assayed to confirm the role of MA secretions in recruiting FB17. The amount of malate secreted from Atalmt1 was highly reduced [F (3,20) = 152.62; P , 0.05] either with or without Pst DC3000 aerial infection (Fig. 4A). Moreover, FB17 failed to colonize the root surface of Atalmt1 under both infected and noninfected conditions as shown by both microscopic root binding and CFU data [F (2,25) = 100.23; P , 0.05; Fig. 4, B and C]. To establish the role of AtALMT1 conclusively, we exploited a previously reported approach (Hoekenga et al., 2006) and generated F 1 plants (genetically complemented for AtALMT1) from the cross Atalmt1 3 Landsberg erecta (Ler-0). The F 1 generated lines were subsequently used for FB17 colonization studies under a Pst DC3000 infection regime. As expected, FB17 was able to colonize the F 1 line of Atalmt1 3 Ler-0 in a similar fashion to wild-type Col-0 roots especially when leaf infected with Pst DC3000 (Fig. 4, B and C). In our hands, the Atalmt1 3 Ler-0 F 1 line also restored malate release under Pst DC3000 infection (data not shown) as reported previously for aluminum (Al) treatment (Hoekenga et al., 2006).

Leaf Infection Induces Root AtALMT1 Expression
To check whether aerial leaf infection with Pst DC3000 transcriptionally regulates AtALMT1 expression, we employed an Arabidopsis transgenic line carrying an AtALMT1 promoter::GUS fusion construct (Kobayashi et al., 2007). The 20-d-old AtALMT1 promoter::GUS Arabidopsis lines were leaf infiltrated with Pst DC3000, NPS3121, Pseudomonas aeruginosa PAO1, Escherichia coli OP50, and also roots treated with 4 mM AlCl 3 as a positive control. The data showed significantly higher AtALMT1::GUS expression in the treatments with Pst DC3000 leaf infection similar to the AlCl 3 positive control in both root central elongation zone and mature region (Fig. 5). However, leaf infiltration with other bacteria, such as PAO1, OP50, and nonpathogenic NPS3121, showed no induction. These data indicated that the AtALMT1 expression may be specific to leaf pathogenic interactions.

Plant Infected Root Exudates and L-MA Induce B. subtilis Biofilm Operons
Microscopic analysis demonstrated increased binding and biofilm formation of FB17 on the root surface. To examine whether biofilm formation was transcriptionally regulated by root secretions, we tested a key operon yqxM required for B. subtilis biofilm formation. We utilized B. subtilis strain Marburg carrying the Figure 5. Effect of Pst DC3000 leaf infection on the expression of root AtALMT1. Twenty-day-old Arabidopsis line carrying AtALMT1 promoter::GUS fusion construct grown on peat pellets was leaf infected with Pst DC3000 and other controls, such as NPS3121, E. coli OP50, and P. aeruginosa PAO1; 12-h-postinfection roots were stained for GUS. The figure shows a significantly higher GUS induction in response Pst DC3000 (on par with the positive control AlCl 3 [4 mM] root treatment) when compared to untreated controls and other bacterial strains such as NPS3121, OP50, and PAO1 (scale = 100 mm). Figure 6. A, Effect of Arabidopsis root exudates on the transcription of the yqxM operon in B. subtilis strain Marburg carrying the yqxM-lacZ fusion (NRS1531). Strain NRS1531 was grown in biofilm medium under biofilm formation conditions at 37°C with root exudates (5%) from Pst DC3000-infected plants (triangles), uninfected (squares) plants, and control (diamonds) without root exudates, and b-galactosidase activity was measured at regular intervals and plotted as a function of time. These experiments were repeated on at least three independent occasions and a representative plot is shown. B, Effect of different MA isomers and oxalic acid on the transcription of the yqxM operon in B. subtilis. Strain NRS1531 was grown in biofilm medium under biofilm formation conditions at 37°C with 5 mM MA isomers, L-MA (squares) and D-MA (3), oxalic acid (triangles), control (diamonds) without MA isomers, or oxalic acid, and b-galactosidase activity was measured at regular intervals and plotted as a function of time. These experiments were repeated on at least three independent occasions and a representative plot is shown. yqxM-lacZ fusion (NRS1531) to study the transcription. Biofilm operon regulation in the lacZ operon fusion line was monitored for b-galactosidase activity. The treatment with root exudates from Pst DC3000infected plants resulted in higher induction of the yqxM operon between 6 and 12 h posttreatment when compared to untreated controls (Fig. 6A). However, the decline in the expression after 9 h may be a feedback response. In addition to root exudates from aerially infected plants causing induction of the FB17 biofilm operon yqxM, MA alone also elevated expression of the biofilm operon yqxM (Fig. 6B). However, with the L-MA treatment, activity level was lower and the kinetics response was abbreviated compared to root exudates. Another isomer (D-MA) and the five-carbon unit oxalic acid did not stimulate b-galactosidase activity (Fig. 6B).

FB17 Root Colonization Triggers ISR and Protects Arabidopsis from Pst DC3000 Infection
To test whether plants associated with FB17 extend protection from disease, we inoculated the FB17 rootcolonized Arabidopsis plants with the pathogen Pst DC3000. Consistent with our data on FB17 root colonization following Pst DC3000 leaf infection, the root colonization of FB17 resulted in protection of plants from Pst DC3000 infection. The Pst DC3000-infected FB17-colonized plants revealed reduced disease incidence, symptom development (chlorosis), and pathogen multiplication. The results were highly significant (P , 0.05; t test) compared to the control plants not colonized with FB17 and treated with Pst DC3000 (Supplemental Fig. S1, A and B).
To further test whether this protection offered by FB17 was due to the induction of ISR, we checked for known systemic resistance markers such as PR1 gene expression and free SA levels in the leaves of FB17 root-colonized plants. We utilized Arabidopsis lines carrying PR1::GUS fusions to study the PR1 expression. Plant roots colonized with FB17 showed higher PR1::GUS expression in the leaves on par with leaf SAtreated positive controls compared to uninoculated control plants. Other controls, where plants were root inoculated with OP50 and Pf01, failed to induce PR1:: GUS expression in the leaves (Supplemental Fig. S2). Similarly, when free SA levels were analyzed (Scott et al., 2004), the Col-0 plant roots colonized with FB17 showed increased free SA titers compared to plants without FB17 colonization (data not shown).

P. fluorescens (Pf01) Fails to Bind to the Root Surface of Aerially Infected Arabidopsis Col-0
To test the specificity of the Arabidopsis FB17 interaction, Pf01 root-inoculated plants were monitored with and without aerial infection. Consistent with chemotaxis assay data, infected plants failed to recruit Pf01 to the root surface as observed previously for FB17. The Pst DC3000-infected Col-0 plants showed poor binding of Pf01, indistinguishable from the untreated and Pf01-only treatments (Supplemental Fig.  S3A). Further, the Pf01 root inoculation failed to protect plants from Pst DC3000 infection (Supplemental Fig. S3, B and C). This result suggested that plants specifically engaged with FB17 under foliar bacterial infection.

DISCUSSION
Certain beneficial rhizobacteria activate plant defenses and thereby mitigate the impact of foliar diseases ; however, plant recruitment of such soil microbes has yet to be established. Here, we showed that inducible defense responses triggered by plant pathogenesis included the induction of root secretions that effectively recruited beneficial rhizobacteria. Specifically, the major inducible root-secreted metabolite MA selectively activated FB17 chemotactic mobility in a dose-dependent manner (Fig. 7). Correspondingly, reduction in MA root secretions with the MA transporter mutant Atalmt1 compromised beneficial bacteria root recruitment. This phenotype was restored when the AtALMT1 mutation was genetically complemented. Furthermore, the leaf infection of AtALMT1::GUS fusion line with Pst DC3000 was able to induce root AtALMT1 expression and it has been reported that AtAlMT1 expression was also induced in response to Al (Hoekenga et al., 2006). However, in this study, we showed AtALMT1 induction in response to Pst DC3000 leaf infection occurred even in the absence of Al. The induction of AtALMT1 and the efflux of MA in response to Pst DC3000 infection independent of Al revealed the multifunctional role of AtALMT1. It will be interesting to further elucidate the key pathogenic determinants, which are specifically involved in triggering AtALMT1 induction. These results establish MA as the first member of a novel class of plant signals operative in rhizosphere attraction of beneficial microbes.
The enantiomeric specificity of naturally produced L-MA in triggering chemotactic mobility implicates receptor-mediated activation of bacterial directed chemotaxis. In addition to stereoselective chemical signals, B. subtilis motility is mediated by the pH and electrical potential of the cell exterior. Bacterial flagellar movement is driven by the motor protein complex Mot (Blair and Berg, 1990). The membrane-embedded MotAB complexes and their homologs constitute ion channels and couple flagellar rotation by proton motive or sodium motive forces across the cytoplasmic membrane (Manson et al., 1977;Blair and Berg, 1990;Berry and Armitage, 1999;McCarter, 2001). Interestingly, MotPS-dependent motility is greater in the presence of MA than in the presence of Glc (Ito et al., 2004).
Bacteria also form sessile communities called biofilms that are morphologically and physiologically differentiated from free-living bacteria (Bais et al., 2004;Rudrappa et al., 2007). The process of biofilm formation starts at the late exponential phase of bacterial growth. In addition to bacterial cell-to-cell signaling involvement in the development of such structures, FB17 colonizes and forms continuous biofilms on the roots of Arabidopsis plants aerially infected with Pst DC3000 unlike the nonhost strain NPS3121, which does not infect or produce disease symptoms with Arabidopsis (van Wees and Glazebrook, 2003). This establishes the specificity of biofilm formation with specific pathogen-infection regimes. Plant-generated MA was also specifically induced with Pst DC3000 and positively induced the biofilm operon yqxM. The operon yqxM is a member of a three-gene operon yqxM-sipW-tasA, where tasA encodes for a protein that is a major component of the biofilm matrix and yqxM is important for the proper localization of TasA to the matrix (Branda et al., 2006). In plants, increased accumulation of L-MA can be triggered by elevated biosynthesis of one or more of the tricarboxylic acid MA precursors, fumarate, oxaloacetate, or pyruvate, by fumarase, malate dehydrogenase, and malic enzyme, or by reduced MA turnover to pyruvate or oxaloacetate by malic enzyme and malate dehydrogenase, respectively (Casati et al., 1999). Alternatively, upregulation of the malate transporter (AtALMT1) augments root MA secretions (Kobayashi et al., 2007). Future lines of study will investigate the involvement of MA secretion on induction of bacterial volatile signaling in B. subtilis, which has been reported to be involved in the induction of ISR in B. subtilis-colonized plants . Concomitantly, others have reported that a few organic acids, especially oxaloacetate, trigger the operon for Bacillus acetoin production alsSD (Schilling et al., 2007).
To establish the specificity of exuded malate toward FB17, other bacteria, including the PGPR Pf01, were assayed for chemotaxis. We speculated that the poor colonization patterns of Pf01 under aerial pathogen infection regimes dictates that infected plants exhibit positive feedback specifically for FB17. In fact, aerially infected plants failed to show increased Pf01 binding to the root surface. Our data showed restriction of pathogen multiplication through the induction of ISR by triggering the expression of the PR1 gene in an SAmediated mechanism following FB17 root colonization. Several studies had previously reported the induction of ISR by root-colonized PGPR  or targeted inhibition of pathogen genes in a standard biocontrol setup (Duffy et al., 2004). Most of these studies were linked to production of a pathogen antagonist by the biocontrol strain. Thus far, the best example of chemical-based intraplant signaling was established recently, where the volatile compounds emitted by one plant part elicited defense responses in other parts of the same plant (Gershenzon, 2007;Heil and Silva-Bueno, 2007). There is no evidence that shows that plants are capable of selectively populating its rhizosphere with beneficial microbes to induce a typical ISR response. Our results showed evidence of the importance of interorganism signaling, although their ecological relevance needs to be examined. Understanding the two-way signaling event in terms of a bacterial metabolite component that malate elicits underscores the evolutionary development of acquired responses in plants to recruit beneficial microbes in its rhizosphere. Subsequent studies should reveal new gene-for-gene interactions that exist among plants and microbes.

Plant Material and Growth Conditions
Arabidopsis (Arabidopsis thaliana) wild-type Columbia (Col-0) seeds were procured from Lehle Seeds. Arabidopsis T-DNA knockout mutant line Atalmt1 was obtained from the Arabidopsis Biological Resource Center (ABRC).The plants were selfed for four generations to achieve homozygosity and the homozygous plants were used in this study. Arabidopsis transgenic lines carrying the AtALMT1 promoter::GUS fusion construct were obtained from Dr. Hiroyuki Koyama. The AtALMT1 genetic complemented F 1 line was generated by crossing Atalmt1 3 Ler-0 by following a published protocol (Hoekenga et al., 2006). Crossing was performed between Atalmt1 (Col-0 background) and Arabidopsis Ler-0 ecotype (as a pollen donor). In the F 1 generation, the AtALMT1 allele from Ler-0 restored the AtALMT1 phenotype in terms of Al-activated malate release (Hoekenga et al., 2006). Seeds were cultured on Murashige and Skoog (1962) solid medium with 3% Suc and allowed to germinate for 3 to 4 d until the roots and shoot emerged by incubating at 23°C 6 2°C under a 16-h-light/8-h-dark photoperiod. The plates were illuminated with cool-white fluorescent light with an intensity of 24 mmol m 22 s 21 .

Effect of P. syringae Infection on B. subtilis Root Binding and Biofilm Formation on the Roots of Arabidopsis
Arabidopsis Col-0 plants grown on sterile peat pellets in a growth chamber set for a photoperiod of 16 h light/8 h dark at 23°C 6 2°C and illuminated with cool fluorescent light with an intensity of 24 mmol m 22 s 21 for a 20-d period.
The plants were root inoculated with FB17 (0.5 OD 600 ) by drenching with 10 mL of the culture in water. All the fully expanded leaves were pressure infiltrated with 100 mL of 0.02 OD 600 culture of Pst DC3000. The cultures in sterile water were prepared by centrifuging the overnight cultures and washing the pellet to get rid of Luria-Bertani medium and resuspending the pellet in sterile water to obtain appropriate density. Different treatments included control root inoculated with FB17, mock (leaf infiltrated with sterile water + root inoculated with FB17), Pst DC3000 + FB17 (leaf infiltration of FB17 root-inoculated plants), and NPS3121 + FB17 (leaf infiltration of FB17 root-inoculated plants). The plants were transferred to sterile magenta boxes and incubated in the growth chamber for an additional 4 d. The experiment was terminated after 4 d and the cocultivated roots were collected and fixed in 4% paraformaldehyde to image for FB17 binding and biofilm formation. Similar experiments were carried out using the malate transporter mutant Atalmt1. Each treatment had six replicates and the experiment was repeated on two independent occasions. Effect of P. syringae Infection on P. fluorescens (Pf01) Binding and Biofilm Formation on the Roots of Arabidopsis All the conditions for this experiment were similar as explained above, except that the plants were root inoculated with P. fluorescens Pf01 (0.5 OD 600 ) by drenching with 10 mL of the culture in water. All the observations that were recorded for the earlier experiment were recorded here also.

Collection and Analysis of Root Exudates from Aerially Infected Col-0 Plants
Individual Col-0 plants were grown on a sterile sponge pellet suspended into sterile 1% Murashige and Skoog liquid medium for 20 d. The fully expanded leaves of the 20-d-old plants were inoculated aseptically with 100 mL of 0.02 OD 600 culture of P. syringae Pst DC3000 (prepared as above). The spent liquid medium from the infected and uninfected plants containing root exudates was collected after 5 d and lyophilized. For HPLC analysis, 100 mg of the lyophilized powder of the root exudates were dissolved in 500 mL of 25 mM potassium phosphate buffer (pH 2.5) and filtered; 30 mL of the filtrate were injected for HPLC analysis. The separation was performed with an isocratic mobile phase of 25 mM potassium phosphate buffer (pH 2.5) on an Altech organic acid column (Prevail, organic acid, 5 mm; 150 mm 3 4.6 mm). The peaks obtained were compared with an array of standard organic acid peaks run under the same conditions. The major peak was identified by comparing the retention time with that of the matching standard. Root exudates collected similarly from the Arabidopsis malate transporter mutant Atalmt1 were also processed and analyzed using the same conditions. Each treatment had 12 replicates and the experiment was repeated on at least two independent occasions.

Chemotaxis Assay
The capillary assay was performed as per the published description (Gordillo et al., 2007) with slight modifications. Briefly, the assay setup consisted of a 200-mL pipette tip as a chamber for holding 100 mL of bacterial suspension (0.1 OD 600 prepared from overnight grown culture) in Luria-Bertani medium. A 4-cm 25-gauge needle (Becton-Dickinson) was used as the chemotaxis capillary and was attached to a 1-mL tuberculin syringe (Becton-Dickinson) containing 200 mL of the compound (5 mM of different MA isomers such as L-MA, D-MA, and also oxalic acid separately in Luria-Bertani liquid medium). After 3 h of incubation at room temperature, the needle syringe was removed from the bacterial suspension and the content diluted and plated in Luria-Bertani medium. Accumulation of bacteria in the capillaries was calculated as the average from the CFUs obtained in triplicate plates and the results were expressed as the mean of at least three separate capillary assays for each determination. The different bacterial strains assayed for chemotaxis included FB17, Pf01, Cd, Pst DC3000, AH2, Arqua-1, and LBA4404.

Effect of Pst DC3000 Leaf Infection on Root AtALMT1::GUS Expression
Transgenic Arabidopsis (Col-0) plants carrying AtALMT1 promoter::GUS fusion construct for the experiments were grown on peat pellets in a growth chamber set for a photoperiod of 16 h light/8 h dark at 23°C 6 2°C and illuminated with cool fluorescent light with an intensity of 24 mmol m 22 s 21 for a 20-d period. The 20-d-old plants were leaf infiltrated with Pst DC3000, NPS3121, OP50, and PAO1 and a positive control of AlCl 3 root treatment was also included in addition to an untreated control. After 12 h of treatment, the roots were stained and imaged for the expression of AtALMT::GUS according to published protocol (Kobayashi et al., 2007).
Effect of B. subtilis Root Inoculation on the P. syringae (Pst DC3000) Infection Arabidopsis Col-0 plants were grown on peat pellets in a growth chamber set for a photoperiod of 16 h light/8 h dark at 23°C 6 2°C and illuminated with cool fluorescent light with an intensity of 24 mmol m 22 s 21 for a 20-d period. The plants were root inoculated with B. subtilis FB17 (0.5 OD 600 ) by drenching with 10 mL of the culture in water. Fully expanded leaves were pressure infiltrated with 100 mL of 0.02 OD 600 culture of Pst DC3000 the next day. Different treatments included control (without Pst DC3000 infiltration), control root inoculated with FB17, mock (leaf infiltrated with sterile water + FB17 root inoculated), Pst DC3000 (only leaf infiltration), and Pst DC3000 + FB17 (leaf infiltration of FB17 root-inoculated plants). The plants were transferred to magenta boxes and incubated in the growth chamber for an additional 4 d. The experiment was terminated after 4 d and the observations such as disease symptom development in terms of chlorosis, number of CFUs of Pst DC3000/g fresh weight of the leaf were recorded by extracting a known fresh biomass of the leaf and plating on Luria-Bertani plates containing 50 mg mL 21 rifampicin.
Effect of B. subtilis FB17 Root Inoculation on Whole-Plant PR1::GUS Expression Transgenic Arabidopsis (Col-0) plants carrying PR1::GUS fusion construct were grown on peat pellets for a 20-d period. The 20-d-old plants were root inoculated with FB17 by drenching the pellet with 10 mL (0.5 OD 600 ) culture suspended in sterile distilled water. In addition, a separate set of PR1::GUS plants was also treated with other bacteria such as Pf01 and OP50 by drenching the pellet with 10 mL (0.5 OD 600 ) of culture suspended in sterile distilled water. Positive controls, such as SA (0.1 mM) treatment on leaves and root were also performed by spraying a known concentration of SA on the leaves and roots of PR1::GUS plants. Four days after inoculation, the plants were stained for GUS assay using a GUS staining kit (obtained from Sigma-Aldrich) according to manufacturer's instructions. The control and treated plants were imaged for the expression of PR1::GUS in the leaves as per the published protocol (Shapiro and Zhang, 2001). A representative image of at least 12 leaves was presented for each treatment. Each treatment had six replicates and the experiment was repeated on two independent occasions.

Microscopy
To view adherent FB17 cells and biofilm on the root surface by confocal scanning laser microscopy, the roots were stained at a dilution of 1:1,000 with SYTO13 (Invitrogen). Single median optical sections were captured with a Zeiss 103 Plan-Apochromat (numerical aperture 0.45) objective lens using a Zeiss LSM 510 NLO on an Axiovert 200M. The 488-nm laser line of the argon laser and a 505-nm long-pass filter were used for excitation and emission, respectively. All data were acquired 24 h postinoculation and posttreatment with B. subtilis FB17 (5-mL culture of 0.02 OD 600 ) of 10-d-old plants grown in 4 mL of liquid Murashige and Skoog medium with 1% Suc. Each experiment was repeated twice with three replicates each and a representative image of at least 20 roots imaged for each treatment was presented.

b-Galactosidase Assay
To study the effect of root exudates collected from Pst DC3000-infected Col-0 plants and MA on the transcription of the yqxM promoter under biofilm formation condition, B. subtilis strain Marburg carrying the yqxM-lacZ fusion (NRS1531) was utilized. The b-galactosidase units produced per minute were estimated as per the published protocol (Rudrappa et al., 2007). Each treatment had six replicates and the experiment was repeated on at least two independent occasions and a representative plot is shown.

Statistical Analysis
All data were averaged from two separate experiments unless mentioned otherwise and further analyzed for variance followed by a Student's t test and ANOVA with the Benjamini-Hochberg correction (Benjamini and Hochberg, 1995) for multiple testing when necessary using statistical package JMP 7.0 and Microsoft Excel XP. The data means were considered significantly different at the probability of P # 0.05.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Effect of B. subtilis FB17 root colonization on Pst DC3000 leaf infection.