Genome‐wide analysis of bacterial determinants of plant growth promotion and induced systemic resistance by Pseudomonas fluorescens

Pseudomonas fluorescens strain SS101 (Pf.SS101) promotes growth of Arabidopsis thaliana, enhances greening and lateral root formation, and induces systemic resistance (ISR) against the bacterial pathogen Pseudomonas syringae pv. tomato (Pst). Here, targeted and untargeted approaches were adopted to identify bacterial determinants and underlying mechanisms involved in plant growth promotion and ISR by Pf.SS101. Based on targeted analyses, no evidence was found for volatiles, lipopeptides and siderophores in plant growth promotion by Pf.SS101. Untargeted, genome-wide analyses of 7488 random transposon mutants of Pf.SS101 led to the identification of 21 mutants defective in both plant growth promotion and ISR. Many of these mutants, however, were auxotrophic and impaired in root colonization. Genetic analysis of three mutants followed by site-directed mutagenesis, genetic complementation and plant bioassays revealed the involvement of the phosphogluconate dehydratase gene edd, the response regulator gene colR and the adenylsulfate reductase gene cysH in both plant growth promotion and ISR. Subsequent comparative plant transcriptomics analyses strongly suggest that modulation of sulfur assimilation, auxin biosynthesis and transport, steroid biosynthesis and carbohydrate metabolism in Arabidopsis are key mechanisms linked to growth promotion and ISR by Pf.SS101.


Pf.SS101 promotes plant growth and changes root architecture
Introduction of Pf.SS101 onto roots of Arabidopsis seedlings grown in soil resulted in significant growth promotion with 1.7-and 2.9-fold increases in leaf and root biomass, respectively, relative to the nontreated control plants (Table 1). Comparable but more pronounced effects of Pf.SS101 on shoot and root biomass were observed for Arabidopsis seedlings grown under in vitro conditions on vertically oriented MS agar plates (Fig. 1a, Table   1). Next to the biomass increase, Pf.SS101-treated seedlings also showed altered shoot and root development, exemplified by enhanced greening, a two-fold reduction of primary root length and a three-fold increase in the number of lateral roots (Fig. 1b,

c). Using a gfp-tagged
Tn7-derivative of Pf.SS101, we found no evidence for endophytic colonization: Pf.SS101 was only found on the root surface and not detected inside the root tissue of Arabidopsis (data not shown). Microscopic analysis of Pf.SS101-treated seedlings showed an increased number of pericycle cells in the roots (Fig. 1d, e). Furthermore, the roots of Pf.SS101-treated plants appeared to switch earlier into secondary growth (Fig. 1i) than the roots of control plants (Fig.   1h). The enhanced formation of pericycle cells was visualized in the GAL4-GFP enhancer trap lines of Arabidopsis (J2351, J1922 and J0661) (Fig. 1f,g). GAL4 enhancer trap lines are useful markers to tag specific cell types and to reveal developmental transitions (Sabatini et al., 1999;Wysocka-Diller et al., 2000;Cary et al., 2002;Birnbaum et al., 2003;Laplaze et al., 2005).
When surface-sterilized Arabidopsis seeds were inoculated with a Pf.SS101-cell suspension, Pf.SS101 established a population density on the roots of approximately 1x10 6 CFU mg -1 root fresh weight after 18 days of plant growth. In the in vitro plant growth assays on MS agar plates, root tip inoculation with cell suspensions of Pf.SS101 resulted in a density 7 of approximately 5x10 5 CFU mg -1 root fresh weight after 18 days of plant growth (data not shown). In these latter in vitro assays, Pf.SS101 was not detected on or in the leaves.

Targeted identification of bacterial traits involved in plant growth promotion and ISR
When roots of Arabidopsis seedlings were treated with heat-killed cells of Pf.SS101, we did not observe the typical plant phenotypes induced by live Pf.SS101 cells, including growth promotion, enhanced lateral root formation and ISR ( Fig. 2a; Table 2, EXP1). Next, we conducted a series of experiments to determine if specific bacterial traits, described previously for other Pseudomonas strains and other rhizobacterial genera, are involved in plant growth promotion and ISR by Pf.SS101. These traits include siderophore, lipopeptide (i.e. massetolide), ACC deaminase, and volatile production. To test the role of these bacterial traits, several approaches were adopted, including site-directed mutagenesis. To study the role of ACC deaminase in growth promotion, we first analyzed the Pf.SS101 genome but did not find the acdS gene involved in the biosynthesis of ACC deaminase. Also, spectrophotometric analysis (with P. fluorescens F113 as a positive control) revealed that Pf.SS101 did not exhibit ACC deaminase activity (Fig. S1). To determine the potential role of volatile organic compounds (VOCs) in plant growth promotion, a split-plate assay was used where Pf.SS101 was grown on MS agar medium on one side physically separated from the Arabidopsis seedlings on the other side of the plate. After 14 days of plant exposure to the bacterial VOCs, no enhancement of shoot biomass was observed (Fig. 2b).
For extracellular metabolites produced by Pf.SS101, the results showed that the siderophore and the lipopeptide massetolide A do not play a significant role in growth promotion of Arabidopsis. The siderophore-deficient mutant of Pf.SS101, generated in this study by plasposon (single inserted) mutagenesis of gene Pflss101_3099 and designated mutant 61C8, enhanced root biomass and induced resistance to the same extent as wildtype This article is protected by copyright. All rights reserved.
To further investigate the potential role of massetolide in growth promotion and ISR, we also grew Arabidopsis seedlings on plates amended with different concentrations of massetolide.
The results showed no effects of massetolide A on plant growth or ISR (Table 2, EXP3).

Untargeted identification of bacterial traits involved in plant growth promotion and ISR
A library of 7,488 random Pf.SS101 mutants was generated via plasposon mutagenesis and each of these mutants was tested individually in two different high-throughput (HTP) bioassays: the first was a plate assay for plant growth promotion and root architecture; the second HTP-assay was a 96-well plate assay for ISR (Fig. 3). We identified 21 potential mutants that were not able to promote plant growth, alter root architecture and induce systemic resistance to Pst (Table 3). The lack of effects on plant growth and ISR by these 21 mutants was confirmed independently in the 'regular' in vitro bioassay described above (Fig.   3). The results of these bioassays also showed that many of the 21 mutants established significantly lower cell densities on roots of Arabidopsis than wildtype Pf.SS101, suggesting they were significantly impaired in root colonization (Table 3). Only two mutants (20H12, 25C8) established rhizosphere population densities similar to that of wildtype Pf.SS101 (Table 3). These results suggest that for most mutants, except 20H12 and 25C8, the lack of effects on plant growth and ISR may be due, at least in part, to poor root colonization by these mutants.

Genetic characterization of Pf.SS101 mutants
For all 21 mutants, the regions flanking the plasposon insertion were cloned and sequenced. In 19 of the 21 mutants, the plasposon insertion was located in genes involved in biosynthesis of different amino acids, including arginine (40H11; 44D8), cysteine (42B9, 20H12), glutamate (18F11), histidine (13E4; 13H6; 24A12; 32H11), tryptophan (24B12; 24D10; 71H9; 74F8), methionine (22G5; 51G1) and valine, leucine, isoleucine (7H2; 9F8; 59B6; 76G8) (Table 3). For the other two mutants, the plasposon was inserted in the genes coding for the DNA-binding response regulator ColR (16G6) and for phosphogluconate dehydratase (25C8), respectively (Table 3). All mutants were able to grow in KB broth to the same density as Pf.SS101, but only 16G6 and 25C8 were able to grow in minimal medium (SSM) to final densities alike wildtype Pf.SS101 (Table 3). The growth deficiency of the mutants in minimal medium was restored by supplementing the amino acid whose biosynthesis was disrupted by the plasposon mutation (Table 4). These results indicate that most mutants, except 25C8 and 16G6, were auxotrophic.
Southern-blot hybridization showed that 19 mutants had a single plasposon insertion except the two mutants 20H12 and 42B9 where two insertions were found. To confirm the role of cysH (20H12) or cysM (42B9) in plant growth promotion and ISR, site-directed mutagenesis of each of these genes was performed to obtain single knockout mutants for cysH and cysM. The location of the gentamycin resistance cassette and the absence of the tetracycline resistance cassette in these mutants was confirmed by PCR using primers targeting each of these two cassettes and genes flanking the targeted genes. Consistent with the phenotype of the random mutants, also these site-directed mutants lacked the ability to induce lateral root formation and ISR against Pst. The site-directed mutants for cysH and cysM were used for further experiments described below. The in vitro bioassay also confirmed that mutants 16G6 and 25C8 did not promote plant growth, alter root architecture nor induced systemic resistance against Pst (Fig. 4b). Mutants 16G6 (colR, PflSS101_4370), 25C8 (edd, PflSS101_4354), 20H12 (cysH, PflSS101_3982) and 42B9 (cysM, PflSS101_3837) were selected for further functional analysis. For each of these four mutants, genetic complementation with the respective gene restored plant growth promotion, alteration of root architecture to the same level as observed for Pf.SS101; also ISR was restored although not entirely to the level as observed for Pf.SS101 (Fig. 4b). Next, we studied if the genes mutated in these 4 mutants were expressed in wildtype Pf.SS101 when colonizing Arabidopsis roots.
Over a course of 7-18 days of plant growth, the genes edd (PflSS101_4354) and cysM (PflSS101_3837) were indeed expressed in Pf.SS101 on roots of Arabidopsis; also cysH (PflSS101_3982) and colR (PflSS101_4370) showed higher expression in Pf.SS101 on Arabidopsis roots after 7, 10 and 14 days but not at 18 days ( Fig. 4d-f).

Role of sulfur assimilation in plant growth promotion and ISR by Pf.SS101
The cysH and cysM genes are essential in sulfur assimilation and the biosynthesis of the amino acids cysteine and methionine (Fig. 5). More specifically, cysH in Pf.SS101 To experimentally provide support for this hypothesis, we conducted a genome-wide transcriptome analysis of Arabidopsis seedlings treated with Pf.SS101 or the cysH mutant (20H12). To explore the expression pattern of Arabidopsis genes that were altered by Pf.SS101 or the cysH mutant, the expression of all 22,850 genes present on the ATH1 genome array were subjected to one-way ANOVA; for exploratory purposes, this analysis was initially done without false discovery rate (FDR) correction. A total of 6,308 genes showed differential regulation (P<0.05) between Arabidopsis plants treated with Pf.SS101, the cysH mutant or the non-treated control. Hierarchical cluster analysis (HCA) and principal component analysis (PCA) were performed with these 6,308 differential genes to explore the pattern of their expression and amount of total variation in expression attributed to Pf.SS101 or the cysH mutant, respectively (Fig. 6). In the HCA, six major clusters were found that explain the total variation in gene expression in the different treatments. These clusters represent genes induced or repressed in plants treated with Pf.SS101 or the cysH mutant (Fig. 6). Clusters II and V represent Arabidopsis genes induced or repressed by Pf.SS101, respectively. Similarly, clusters VI and III represent Arabidopsis genes induced or repressed by the cysH mutant, respectively. The remaining clusters I and IV correspond to Arabidopsis genes induced or repressed by both Pf.SS101 and the cysH mutant, respectively (Fig. 6). In the PCA, the first principal component (PC1) explained 41% of the total variation in gene expression and is attributed to the unique clusters of genes whose expression was altered in plants treated with Pf.SS101 as compared to control plants or to plants treated with the cysH mutant ( Fig. 6, clusters II, III, V and VI). The second principal component (PC2) explained 30% of the total variation and is attributed to clusters of genes that were altered in plants treated by Pf.SS101 and by the cysH mutant as compared to the control plants (Fig. 6, clusters I and IV).
To understand the major growth and defence related biological processes (BPs) that are altered in Arabidopsis by Pf.SS101, we performed gene set enrichment analysis (GSEA) specifically on genes in Cluster II of the HCA, representing genes in Arabidopsis whose expression was significantly induced by Pf.SS101 (Fig. 6b). Prior to performing the GSEA, we selected the genes in this cluster and computed independent t-tests by comparing the mean expression value for each of the genes in Pf.SS101-treated plants with the genes in the cysH-treated plants. Cluster II contains a total of 967 genes of which 547 genes were significantly different (P < 0.05, with FDR correction) between plants treated with Pf.SS101 and plants treated with the cysH mutant. The GSEA on these 547 genes revealed 246 significantly enriched BPs. However, these long lists of BPs were largely redundant and reduced to 68 BPs by performing HCA on the gene X GO matrix, an output from the GSEA (Table 6). These 68 BPs fall into the following major categories: biosynthesis, transport, catabolism, response to stimulus and growth. From the processes associated with biosynthesis, sulfur compounds and specifically serine, cysteine and glucosinolate biosynthetic processes were the most significantly enriched ( Table 6). The BPs involved in plant growth, such as indole acetic acid biosynthetic process and auxin transport, steroid biosynthesis and isopentenyl diphosphate biosynthesis, also showed significant enrichment in this cluster. Another significantly enriched BP was carbohydrate biosynthetic processes, specifically the biosynthesis of starch.
In line with this, also glucose catabolic processes were significantly enriched (Table 6). In Cluster VI ( fig. 6b), 831 genes were significantly upregulated (P < 0.05, with FDR correction) in plants treated with the cysH mutant as compared to plants treated with wild type Pf.SS101. The GSEA on these 831 genes revealed 276 significantly enriched BPs. Following similar procedures as stated above, redundant BPs were reduced to 67 representative BPs (Table 7). The majority of these 67 BPs fall into BPs that are induced during incompatible plant-microbe interactions while BPs associated with sulfate reduction were suppressed ( Fig.   6b, cluster VI, Fig. 6c, d (see, sulfur reduction) and Table 7).

Discussion
In the present study, we showed that Pf.SS101 enhances Arabidopsis growth, alters root architecture and induces systemic resistance against the leaf pathogen P. syringae pv.
tomato (Pst). In line with these phenotypes, the genome-wide plant transcriptome profiling performed showed significant enrichment of biological processes that play a critical role in plant growth, including processes related to auxin biosynthesis, auxin polar transport and steroid biosynthesis. Other studies have shown that different rhizobacterial genera enhance plant growth and induce systemic resistance via the production of phytohormones, siderophores, lipopeptides or volatiles (VOCs) (Ryu et al., 2003;Tran et al. 2007;Raaijmakers et al., 2010;Van de Mortel et al. 2012;Bakker et al., 2013). Our results indicate that the siderophore and the lipopeptide massetolide produced by Pf.SS101 do not significantly contribute to growth promotion and ISR in Arabidopsis under the experimental in vitro conditions used here. In the rhizosphere of plants grown in more complex substrates, however, siderophores and lipopeptides are commonly involved in interspecific bacterial competition contributing to rhizosphere colonization. Hence, these metabolites may be more relevant for plant growth promotion and ISR by Pf.SS101 in a natural soil-plant context. Also VOCs produced by Pf.SS101 do not seem to play a role in growth promotion and ISR of Arabidopsis. This is in contrast to earlier work with tobacco seedlings where Pf.SS101 promoted plant growth via the production of specific VOCs, in particular, 13-Tetradecadien-1-ol, 2-butanone and 2-Methyl-n-1-tridecene (Park et al., 2015). A major difference between this former study and the work presented here is that the growth medium used in the tobacco assay was a rich medium. In the present study, a poor agar medium (0.5x MS) was used which does not support excessive growth of Pf.SS101 which in turn may have had qualitative and quantitative effects on the VOCs produced. Whether other or higher concentrations of VOCs are produced by Pf.SS101 when colonizing the roots of Arabidopsis seedlings remains to be investigated.
Results from the genome-wide screening of 7,488 Pf.SS101 random mutants led to the selection of 21 mutants deficient in both growth promotion and ISR. This result seems to be in contrast to the results of Zamioudis et al. (2013), who showed that plant growth promotion and ISR by Pf.WCS417 are mediated by different pathways. Given the complexity of the genetic and molecular basis of both plant phenotypes (growth promotion, ISR), it was surprising that only 21 Pf.SS101 mutants were found out of a total of 7,488. This may be explained, at least in part, by the high stringency used in the plant screens, where we only selected those Pf.SS101 mutants with a strongly reduced ability to induce resistance or to alter root architecture and plant growth. Hence, we may have overlooked a number of mutants that affect these plant phenotypes in a more subtle and differential manner. Furthermore, the fact that all 21 Pf.SS101 mutants affected both plant phenotypes is, for many of these mutants, most likely due to their poor root colonizing abilities not reaching the required threshold densities to induce these phenotypes. Most of the 21 mutants were deficient in the biosynthesis of specific amino acids. The role of amino acids in rhizobacteria-plant interactions is not well studied, although some amino acids such as methionine and tryptophan may act in soil as precursors for the biosynthesis of the phytohormones ethylene and indole-3acetic acid, respectively (Murcia et al., 1997;González-López et al., 2005). What the role is of these and other amino acids (histidine, valine, leucine and isoleucine) in root colonization and Pf.SS101-Arabidopsis interactions is yet unknown.
The cysH and cysM genes identified in our Pf.SS101 mutant screens are essential in sulfur assimilation and the biosynthesis of the amino acids cysteine and methionine. Results from in vitro assays with Arabidopsis grown on MS agar medium supplemented with different concentrations of these two amino acids showed that both cysteine and methionine induced lateral root formation in Arabidopsis in a concentration dependent manner (Fig. S3a).
Moreover, cysteine at relatively high concentrations induced disease resistance against Pst in Arabidopsis (Fig. S3b). These results confirm and extend observations that cysteine homeostasis is important for plant immunity (Alvarez et al. 2012). The plant responses observed here may not be typical for cysteine and methionine only as several studies have shown effects of exogenous amino acids on root growth (Walch-Liu et al. 2006) and disease resistance (Hijwegen, 1963;Kadotani et al., 2016). Analyses of the temporal in situ production levels of amino acids by Pf.SS101 on roots of Arabidopsis should be conducted to further disentangle the role of these amino acids in the observed plant responses. In a more indirect way, however, our transcriptome data did reveal that biosynthetic processes associated with sulfur compounds and specifically serine, cysteine and glucosinolates, were the most significantly enriched in seedlings treated with Pf.SS101 as compared to the control plants and plants treated with the cysH mutant. These results indicate that Pf.SS101 modulates sulfur metabolism in Arabidiopsis, particularly processes related with sulfur reduction (Fig. 6c and d). These results extend findings in previous studies by Meldau et al. (2013) and Aziz et al. (2016) who attributed modulation of sulfur metabolism as a mechanism of growth promotion and induction of lateral roots of tobacco and Arabidopsis by different Bacillus strains. Meldau et al. (2013) further showed that the growth-promoting effects on tobacco were mediated by the production of the VOC dimethyl disulphide.
In plants, sulfur is important in various stress responses (Bloem et al., 2005;Kertesz et al., 2007). Elemental sulfur itself can be used directly by plants, via deposition in the xylem parenchyma (Cooper and Williams, 2004). The metal-chelating properties of sulfur in phytochelatins help alleviate heavy metal stress and sulfur is also important to the plant in responding to pathogen attack, since many defense compounds contain sulfur, in particular the glucosinolates (Brader et al., 2006). Cysteine biosynthesis in plants involves the incorporation of the carbon backbone from serine with reduced inorganic sulfur (Neuenschwander et al., 1991;Saito et al., 1994;Bonner et al., 2005). Cysteine might enter into the glucosinolate biosynthesis pathway by three routes. The first route involves direct donation of reduced sulfur to glucosinolate biosynthesis. The second route involves the incorporation of cysteine into methionine and through a series of side chain elongation, S-glycosilation and other secondary modification, it ends up in the glucosinolate pool. The third route could involve the conjugation of cysteine, glutamate and glycine to form glutathione (GSH) (Meister, 1995). (Geu-Flores et al., 2011) showed that GSH acts as a sulfur donor for glucosinolate biosynthesis. In our previous study, we have shown that Pf.SS101 enhances glucosinolate levels in roots and shoots of Arabidopsis seedlings ( Van de Mortel et al. 2012). In line with this, metabolic processes related to all the aforementioned amino acids showed significant enrichment among the genes induced by Pf.SS101, indicating that the second route is the most probable means of reduced sulfur channelling mechanism into the glucosinolate pool. Bacteria that are able to successfully establish beneficial relationship with plants typically circumvent or suppress the induction of the host immune system (Zamioudis and Pieterse, 2012).
Interestingly, treatment of Arabidopsis with the cysH mutant led, in contrast to wild type Pf.SS101, to the induction of biological processes (BPs) that are associated with incompatible plant-microbe interaction. This suggests that a mutation in the cysH gene may have compromised the ability of the bacteria to circumvent or suppress the host immune responses, resulting in recognition of the cysH mutant by the plant as a harmful invader. In line with this, Sinorhizobium meliloti mutants that lack sulfation of Nod factors are strongly impaired in their ability to nodulate their host alfalfa (Roche et al., 1991). In this context, we speculate that the cysH mutation in Pf.SS101 affects sulfation of yet unknown bacterial traits involved in modulation of the plant immune system.
Carbohydrate biosynthetic processes in general and starch biosynthetic processes in particular were highly induced by Pf.SS101 and these processes are critically important for biomass formation. The recycling of glucose is the primary step before its incorporation into starch through the enzymes of the glycolytic, glucogenic and pentose phosphate pathways (Glawischnig et al., 2002). Interestingly, the transcriptome data showed that genes involved in these biological processes are also significantly enriched in Pf.SS101-treated seedlings.
In addition to the cysH and cysM mutants, two other Pf.SS101 mutants with mutations in the colR and edd genes were identified in this study. The ColR-ColS pathway was first characterized in P. fluorescens for its role in competitive colonization of plant roots (Dekkers et al., 1998). Subsequent studies have shown that mutations in the ColR-ColS two-component system lead to several other defects in different Pseudomonas strains (Hõrak et al., 2004;Kivistik et al., 2006). De Weert and colleagues (2006) showed that a putative methyltransferase/wapQ (inaA) operon is located downstream of ColR-ColS in P. fluorescens WCS365 and regulated by ColR-ColS. Since wapQ (inaA) encodes a putative lipopolysaccharide (LPS) phosphatase, the possibility was studied that the integrity of the outer membrane of P. fluorescens WCS365 mutant PCL1210 was altered. PCL1210 was identified as a colonization mutant with an insertion in the ColR-ColS two-component system (Dekkers et al., 1998). Mutants in the methyltransferase/wapQ operon were also altered in their outer membrane permeability and defective in competitive tomato root tip colonization (De Weert et al., 2006). In Pf.SS101, we also identified a putative methyltransferase/inaA (wapQ) operon downstream of ColR-ColS but its exact role and underlying mechanisms in plant growth promotion and ISR are not yet known.
The edd gene codes for 6-phosphogluconate dehydratase, an enzyme that catalyzes the first step in the Entner-Doudoroff (ED) pathway (Wanken et al., 2003) which comprises the dehydration of 6-phospho-D-gluconate into 6-phospho-2-dehydro-3-deoxy-D-gluconate (Peekhaus and Conway, 1998;Kim et al., 2007). Many bacteria possess genes for the ED pathway (Kim et al., 2007). For P. chlororaphis O6, Kim et al. (2007) showed that the edd gene contributes to root colonization and ISR. They concluded that metabolism of sugars through the ED pathway in P. chlororaphis O6 may be important as it may facilitate the production of effectors involved in ISR (Kim et al., 2007). In our study, we showed that the edd gene was significantly higher expressed in Pf.SS101 in the rhizosphere of Arabidopsis compared to Pf.SS101 grown KB medium only and that edd mutant 25C8 showed no induction of lateral root formation and systemic resistance in Arabidopsis, similar to what was shown for P. chlororaphis O6 (Kim et al., 2007). In contrast to P. chlororaphis O6, however, elimination of the edd gene in Pf.SS101 had no effect on colonization of Arabidopsis seedlings grown in vitro.
In conclusion, modulation of auxin biosynthesis and transport, steroid biosynthesis, carbohydrate metabolism and sulfur assimilation in Arabidopsis appear to be key mechanisms linked to growth promotion and ISR by Pf.SS101. In particular sulfur assimilation was shown to be an important biological process modulated by Pf.SS101 in Arabidopsis. The molecular signals and sulfur-containing compounds involved have not yet been identified and need further investigation. Also identification of the bacterial traits associated with the ColR-ColS two-component system and the ED pathway in Pf.SS101 as well as the plant transcriptional and metabolic responses to these two Pf.SS101 mutants will be required to shed more light on the other mechanisms of plant growth promotion and ISR.

Bacterial strains and culture conditions
Pseudomonas fluorescens SS101 (Pf.SS101) was cultured in liquid King's B medium (KB) at 25 °C for 24 h. Bacterial cells were collected by centrifugation, washed three times with 10mM MgSO 4 and resuspended in 10 mM MgSO 4 to a final density of 10 9 CFU ml -1 (OD 600 = 1.0). Pseudomonas syringae pv. tomato DC3000 (Pst) was cultured in KB broth supplemented with rifampicin (50 µg ml -1 ) at 25 °C for 24h. Escherichia coli strain DH5α was used as a host for the plasmids for site-directed mutagenesis and complementation. E. coli strains were grown on Luria-Bertani (LB) plates or in LB broth amended with the appropriate antibiotics. The random plasposon mutants of Pf.SS101 were obtained by biparental mating with E. coli strain S17 λ pir harboring the TnModOKm element in plasmid (Dennis and Zylstra, 1998), according to protocols described by Sambrook and Russel (Sambrook et al., 2001). Transformants were selected on KB agar plates supplemented with rifampin (100 μg ml −1 ) and kanamycin (100 μg ml −1 ).
Auxotrophy of selected plasposon mutants were tested by growing these mutants O/N in 5ml KB supplemented with the appropriate antibiotics and shaken at 220 rpm at 25 °C. O/N cultures were washed three times with 10 mM MgSO 4 and set to OD 600 = 1.0. Then a starting culture was inoculated at a concentration of 0.5% (v/v)  shaking, 1min mixing and OD-measurements every 2 min over a period of 24 hours. ACC deaminase activity of Pf.SS101 was measured according to methods described by Penrose and Glick (2003).

Site-directed mutagenesis
Site-directed mutagenesis of the genes cysH and cysM was performed based on the method described by Choi and Schweizer (2005). The primers used for amplification are described in supporting information Table S1. The FRT-Gm-FRT cassette was amplified with pPS854-GM, a derivative of pPS854, and FRT-F and FRT-R were used as primers (Supporting information Table S1). The first-round PCR was performed with KOD polymerase (Novagen), according to the manufacturer's protocol. PCR reactions were carried out under the following conditions: an initial denaturation step for 2 min at 95 °C followed by denaturation for 15 s at 95 °C, annealing for 20 s at 58 °C and extension for 30 min at 72 °C for 30 cycles, followed by a final elongation step at 72 °C for 5min. All fragments were run on a 1% (w/v) agarose gel and purified with illustra TM GFX TM PCR DNA and Gel Band Purification Kit (GE Healthcare Life Sciences). The overlap extension PCR was performed with Verbatim High Fidelity DNA polymerase (Thermoscientific) according to the manufacturer's protocol by addition of equimolar amounts of the 5-end fragment, FRT-Gm-FRT, and 3-end fragment. PCR reactions were carried out under the following conditions: an initial denaturation step for 2 min at 95 °C followed by denaturation for 20 s at 98 °C, annealing for 15 s at 58 °C and extension for 2 min at 72 °C for 30 cycles, followed by a final elongation step at 72 °C for 5 min and the PCR fragments were purified as described above.
The fragments were digested with BamHI and cloned into BamHI-digested plasmid pEX18Tc and transformed colonies were selected on LB medium supplemented with 25 μg ml -1 gentamicin (Sigma). Integration of the inserts was verified by PCR analysis with pEX18Tc primers (Supporting information Table S1) and by restriction analysis of isolated plasmids.
The pEX18Tc-cysH and pEX18Tc-cysM constructs were subsequently transformed to Pf.SS101. Competent cells were obtained by washing the cells three times with 300 mM sucrose from a 6-ml overnight culture and finally dissolving the cells in 100 µl of 300 mM sucrose. Electroporation occurred at 2.4 kV and 200 F and after incubation in SOC medium for 2 h at 25 °C cells were plated on KB supplemented with gentamicin (40 µg ml -1 ) and rifampicin (50 µg ml -1 ). Six obtained colonies were grown in LB for 2-3 h at 25 °C than diluted 10 times and plated on LB supplemented with gentamicin (40 µg ml -1 ) and 5% sucrose to accomplish the double crossover. The plates were incubated at 25 °C for at least 48 h and colonies were re-streaked on LB supplemented with gentamicin and 5% sucrose. Twelve colonies per transformation were transferred to KB plates supplemented with tetracycline (25 µg ml -1 ) and KB plates with gentamycin and rifampicin. Colonies that grew on LB with gentamicin and rifampicin but not on LB with tetracycline were selected and subjected to colony PCR to confirm the presence of the gentamicin resistance cassette and the absence of the tetracycline resistance cassette. Positive colonies were confirmed by sequencing the PCR fragments obtained with the Up forward and Dn reverse primers (Supporting information Table S1). The mutants obtained were tested for induction of lateral root formation in the in vitro assay with Arabidopsis.

Construction of pME6031-based vectors for genetic complementation
A fragment of approximately 2 kb containing the cysH or cysM gene, including the promoter and terminator, was obtained by PCR with specific primers (Table S1) (Heeb et al., 2000). E. coli DH5α was transformed with the constructs by heat shock transformation (Inoue et al., 1990) and transformed colonies were selected on LB agar plates supplemented with tetracycline (25 µg ml -1 ). Correct integration of the fragments was verified by PCR analysis and restriction analysis of the isolated plasmids. The pME6031-cysH, pME6031-cysM constructs were subsequently transformed into the cysH or cysM plasposon mutant. Transformed cells were plated on KB supplemented with tetracycline (25 µg ml -1 ) and the presence of pME6031-cysH or pME6031-cysM was verified by PCR analysis with primers specific for pME6031. The complemented mutants obtained were tested for their ability to induce lateral root formation in the in vitro assay with Arabidopsis.

Plant Material and Growth Conditions
Seeds of Arabidopsis thaliana Columbia-0 (Arabidopsis) were surface sterilized for three hours by placing seeds in opened Eppendorf tubes in a desiccator jar. Two 100-ml beakers each containing 50 ml commercial bleach was placed inside and 1.5 ml concentrated HCl was added to each beaker. The desiccator jar was closed and the seeds were sterilized by chlorine gas. After 4 h, seeds were transferred on water-saturated filter paper in petri dishes followed by a 3-day treatment at 4 °C. Thereafter, 10-12 seeds were sown on plates containing 50 ml half-strength Murashige Skoog (MS) medium (Murashige and Skoog, 1962).
One-week-old Arabidopsis seedlings were inoculated at the root tip with 2 µl Pf.SS101 cell suspensions (10 9 CFU ml -1 ); in the control treatment, seedlings were inoculated with 2 µl of 10 mM MgSO 4 . After an additional three days of plant growth, the 10-day-old seedlings were transferred to 60 ml PVC pots containing a sand-potting soil mixture that was autoclaved twice for 20 min with a 24 h interval. Once a week, plants were supplied with modified halfstrength Hoagland nutrient solution (Hoagland and Arnon, 1938).
In the in vitro assays with half-strength MS medium, Pf.SS101 was applied to the seeds or to the root tips. For both treatments, seeds were sterilized and sown on plates as described above. For the Pf.SS101 seed treatment, a cell suspension (10 9 CFU ml -1 ) of Pf.SS101 was added to the sterilized seeds in a Petri dish and incubated for 30 minutes at room temperature. For the control, seeds were incubated for 30min with sterile 10mM MgSO 4 . For the root tip treatment, 2 µl Pf.SS101 (10 9 CFU ml -1 ) was applied to the root tips of one-week-old seedlings. Control plants were inoculated with 2 µl of 10 mM MgSO 4 . The challenge with Pst was performed by inoculation of 2 µl cell suspension (10 9 CFU ml -1 ) in the center of the leaf rosette of 14-day-old plants. Five to seven days after challenge inoculation, disease incidence was assessed by determination of the percentage of diseased leaves per plant. Leaves were scored as diseased when they exhibited necrotic or water-soaked lesions surrounded by chlorotic tissue. From the number of diseased and non-diseased leaves, the disease incidence was calculated for each plant (20-30 plants per treatment). The experiment was performed at least twice.

Plant Microscopic analysis
Seeds of Arabidopsis were pre-treated with Pf.SS101 and grown for 18 days vertically

Transcriptome analysis of Arabidopsis exposed to Pf.SS101 and the cysH mutant
Total RNA was extracted from shoots of untreated, Pf.SS101-treated and cysH mutant-treated plants after 18 d of growth. Four biological replicates with 30 plants per replicate were used for each treatment. RNA was isolated from the frozen tissues with Trizol reagent (Invitrogen). The RNA samples were further purified using the NucleoSpin RNA II kit (Macherey-Nagel). For the Affymetrix Arabidopsis genome GeneChip array analysis (ServiceXS), amplification and labelling of the RNA samples as well as hybridization, staining, and scanning were performed according to the manufacturer's specifications. The raw array data (CEL files) were normalized using the RMA probe summarization algorithm in R programme using Bioconductor package; the processed data were used for further analysis and can be found at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE103117. ANOVA without false discovery rate (FDR) correction was performed to identify significantly altered transcripts between Arabidopsis plants treated with Pf.SS101, the cysH mutant (20H12) and non-treated control plants. Using the transcripts that were significantly altered (P < 0.05, without FDR correction) between the treatments, discriminant function analysis (DFA) and hierarchical cluster analysis (HCA) were performed in Genemaths XT software (Applied Maths, Inc. Austin, TX, USA). For HCA, Pearson's correlation coefficients were used to calculate the distance or similarity between two entries and the resulting clusters were summarized using a complete linkage algorithm. To compare the expression values, the raw values of each sample were auto-scaled by the use of the average as an offset and the standard deviation as scale (raw value-average (offset)/SD (scale)). Clusters of genes that showed altered expression patterns between the contrasting treatments in the HCA were selected. The gene X GO matrix was used to perform HCA to reduce the number of redundant biological processes described by a group of genes as described by Etalo et al. (2013). The aliphatic glucosinolate biosynthesis pathway shown in Figure 6d was reconstructed based on KEGG and the overview by Sønderby et al. (Sønderby et al., 2010); genes for which experimental verification is lacking (such as GGP1, which has recently been shown not to be coexpressed with the rest of the pathway, and may be in fact unrelated to it (Wisecaver et al., 2017)) were purposefully left out.     Cell densities of Pf.SS101 and the mutants grown in KB and SSM media at 25°C were measured spectrophotometrically (OD 600nm ): + = growth; -= no growth. Table 4 Growth of Pf.SS101 and 10 mutants in SSM medium supplemented with different L-/D-amino acids (up to 15mM) at 25°C; cell density was measured spectrophotometrically (OD 600nm ) for six replicates. + = growth; -= no growth.        In the PCA, the first principal component (PC1) explained 41% of the total variation in gene expression and is attributed to the unique clusters of genes whose expression was altered in

Supplementary table and figure legends
Table S1 Strains, plasmids and primers used in this study for site-directed mutagenesis and genetic complementation.

Table S2
Primers used in Q-PCR analysis.

Figure S1
ACC-deaminase activity in Pf.SS101. Pf.F113 was used as a positive control and P. protegens CHA0 was used as a negative control. α-Ketobutyrate production by strains F113, CHAO and SS101 was measured after incubation of their cell extracts with ACC.
Averages of 4 replicates are given. Different letters indicate significant differences (P<0.05).

Figure S2
Comparison of amino acid sequences of the APS reductase from Pf.SS101 with the APS reductase from P. putida, P. entomophila, P. syringae, Pf.SBW25, Pf01, Pf5 and P.
* indicates different concentrations of the lipopeptide massetolide A in µg ml -1.
"±" represents the standard error of the mean of 4 replicates with 10 plants per replicate.  Table 3 Phenotypic and genetic characterization of Pf. SS101 and mutants deficient in plant growth promotion and ISR.
Letters indicate significant differences among the treatments according to One way ANOVA analysis (P<0.05).

Growth curve Locus_tag
Gene Product COG 1 Strains 1 COG Functional annotation. C = Energy production and conversion; E = Amino acid transport and metabolism; T = Signal transduction mechanisms.(+) = growth; (-) = no growth. This article is protected by copyright. All rights reserved.   trmJ tRNA (cytidine/uridine-2'-O-)-methyltransferase TrmJ J a Functional annotation. C = Energy production and conversion; E = Amino acid transport and metabolism; H = Coenzyme metabolism; J = Translation, ribosomal structure and biogenesis; K = Transcription; O = Posttranslational modification, protein turnover, chaperones; P = Inorganic ion transport and metabolism R = General function prediction only; S = Function unknown