The CebE/MsiK Transporter is a Doorway to the Cello-oligosaccharide-mediated Induction of Streptomyces scabies Pathogenicity

Streptomyces scabies is an economically important plant pathogen well-known for damaging root and tuber crops by causing scab lesions. Thaxtomin A is the main causative agent responsible for the pathogenicity of S. scabies and cello-oligosaccharides are environmental triggers that induce the production of this phytotoxin. How cello-oligosaccharides are sensed or transported in order to induce the virulent behavior of S. scabies? Here we report that the cellobiose and cellotriose binding protein CebE, and MsiK, the ATPase providing energy for carbohydrates transport, are the protagonists of the cello-oligosaccharide mediated induction of thaxtomin production in S. scabies. Our work provides the first example where the transport and not the sensing of major constituents of the plant host is the central mechanism associated with virulence of the pathogen. Our results allow to draw a complete pathway from signal transport to phytotoxin production where each step of the cascade is controlled by CebR, the cellulose utilization regulator. We propose the high affinity of CebE to cellotriose as possible adaptation of S. scabies to colonize expanding plant tissue. Our work further highlights how genes associated with primary metabolism in nonpathogenic Streptomyces species have been recruited as basic elements of virulence in plant pathogenic species.

Scientific RepoRts | 6:27144 | DOI: 10.1038/srep27144 TxtR 11 . Moreover, we recently identified a key role for the cellulose utilization regulator, CebR, in unlocking the S. scabies pathogenicity 12 . CebR has been shown to bind upstream as well as within the thaxtomin biosynthetic cluster, and specific interaction between CebR and cellobiose or cellotriose was shown to trigger the release of the repressor from its binding sites hereby allowing transcription of the thaxtomin biosynthetic (txtA and txtB) and regulatory (txtR) genes 12 . These findings underline the significance of cellobiose and cellotriose as signal molecules for virulence besides serving as a nutrient 13 .
After having described how cellobiose and cellotriose unlock production of thaxtomin A 12 , we investigated and describe in this study how S. scabies links extracellular plant material sensing to the onset of its virulent behavior.

Results
In silico identification of the predicted cello-oligosaccharide transporter in S. scabies. Earlier work in other Streptomyces model species showed that the CebEFG-MsiK ABC transporter is specifically involved in the uptake of cellobiose and cellotriose [14][15][16] . CebE is the solute-binding protein which binds cellobiose and cellotriose and therefore determines the specificity of the ABC transporter 17 , whereas CebF and CebG are the transmembrane proteins that line the pore across the bacterial cell membrane. ATP hydrolysis is mediated through the action of MsiK which is the multiple sugar import ATPase associated with many different ABC transporters in streptomycetes 16,[18][19][20] . Orthologues of the genes/proteins involved in cello-oligosaccharide transport in S. scabies were identified by a BLAST search and by using the Absynte software 21 ; both of which identified scab57751, scab57741, scab57731, and scab50161 as the orthologues of cebE, cebF, cebG, and msiK, respectively. The genes encoding the transporter components (cebEFG) are organized in an operon (Fig. 1A), except for the gene that codes for the ATPase (msiK) which is located on another part of the chromosome (see Supplementary  Figs S1 and S2). The cebEFG operon is divergently oriented from cebR encoding the cellulose utilization repressor 22,23 and which also directly controls the expression of the thaxtomin biosynthetic (txtA and txtB) and regulatory genes (txtR) 12 . The gene coding for the predicted β-glucosidase (bglC) for further catabolism of the incoming cello-oligosaccharides into glucose is located 146 nt downstream of the tri-cistronic cebEFG operon. The synteny of the cebR-cebEFG-bglC cluster is conserved in most Streptomyces species (see Supplementary Fig. S1).
Earlier investigations identified two other putative ABC-type solute-binding proteins that could be involved in cello-oligosaccharide transport in S. scabies 24 , namely the products of genes scab2421 and scab77271. However, phylogeny and synteny analyses suggest that SCAB57751 (CebE scab ) is the orthologue of CebE of S. reticuli (CebE reti ) which has been experimentally demonstrated to be the cellobiose and cellotriose-binding protein ( Fig. 1B and see Supplementary Fig. S1). In addition, the in silico prediction of the CebR regulon identified the perfect palindromic TGGGAGCGCTCCCA CebR-binding site (cbs) at position −131 nt upstream of scab57751 and another predicted cbs at position −14 nt upstream of scab57721, further supporting the hypothesis of the involvement of this gene in cellobiose utilization (Fig. 1A). Overall, the compilation of our in silico analyses strongly supports that scab57751 (cebE scab ), scab57741 (cebF scab ), scab57731 (cebG scab ), and scab50161 (msiK scab ) are the best candidate genes to encode the proteins of the cello-oligosaccharide-mediated induction of thaxtomin production in S. scabies.
Cellobiose and cellotriose interact with CebE of S. scabies at a nanomolar range. To assess if CebE scab can bind cellobiose and/or other cello-oligosaccharides, we used the intrinsic fluorescence of the aromatic residues of the protein to monitor conformational changes upon substrate binding. CebE scab includes 10 tryptophan residues (in addition to 12 tyrosine and 13 phenylalanine residues) which exhibit a fluorescence signal at a wavelength between 300-400 nm when excited at 280 nm. Modeling of the 3D structure of the CebE protein with the YASARA software 25 revealed conservation of the tryptophan residue predicted to participate in the binding of carbohydrates in ABC-type sugar-binding proteins, i.e. Trp303 within the binding site of the native protein (see Supplementary Fig. S3). As the environment of Trp303 is likely to be modified upon sugar binding, it strengthens the rationale of using intrinsic fluorescence as a methodology to assess CebE/cello-oligosaccharide interactions.
The cebE scab gene (scab57751) -without its first 44 codons encoding the peptide secretion signal and the membrane-anchoring lipopeptide -was cloned into pET-28a allowing for the production of an N-terminally 6 Histidine-tagged CebE protein (6His-CebE) in Escherichia coli and its subsequent purification through Ni-NTA affinity chromatography (see experimental procedures for details and Tables 1, 2 and Supplementary Fig. S4). We assessed the fluorescence emission spectra of 6His-CebE (100 nM) in the presence or absence of 30 6 ], as well as with 1 mM of other mono-and disaccharides unrelated to the cellulose utilization pathway ( Fig. 2A and Supplementary Fig. S5). The addition of (Glc) 2 and (Glc) 3 resulted in an increase of fluorescence intensity, with a moderate 'blue'-shift of the λ max from 328 nm to 325 nm, indicating a modification of the environment of the tryptophan residues following interaction with cello-oligosaccharides ( Fig. 2A). In contrast, the addition of glucose and cello-oligosaccharides (Glc) 4 , (Glc) 5 , and (Glc) 6 neither significantly altered the fluorescence intensity between 300 and 400 nm nor caused a shift of the λ max of 6His-CebE ( Fig. 2A and Supplementary Fig. S5). The disaccharides maltose and trehalose also did not modify the fluorescence properties of 6His-CebE (not shown).
To determine if changes in intrinsic fluorescence intensity were due to a specific binding of (Glc) 2 and (Glc) 3 , we calculated the dissociation constants (K D ) by measuring the intrinsic fluorescence intensity of the protein in the presence of increasing concentrations of (Glc) 2 and (Glc) 3 (Fig. 2B). F max could be estimated by addition of a large excess of (Glc) 2 and (Glc) 3 in the protein solution. Equations (1) and (2) (see Materials and Methods) were used to transform the raw fluorescence data and estimate the dissociation constants. The K D values of cellobiose-CebE and of cellotriose-CebE complexes were determined as 14 (±2) nM and 2 (±0.5) nM, respectively (Fig. 2B). It is important to note that the K D values of CebE scab for (Glc) 2 and (Glc) 3 , both in the nanomolar range, are 2 and about 3 orders of magnitude lower than the K D values measured for the CebE protein of the cellulolytic species S. reticuli (~1.5 μM) 14 . The very high affinity of CebE to cellotriose could be a key adaptation of S. scabies allowing colonization of root and tuber crops (see Discussion).

Effect of the inactivation of cebE and msiK on thaxtomin production and virulence of S. scabies.
In order to know whether the cellobiose-mediated induction of thaxtomin A involves the (Glc) 2 and (Glc) 3 -binding protein CebE, we deleted scab57751 by gene replacement with an apramycin resistance cassette as described previously 12 . Similarly, the scab50161 gene was deleted to investigate if the ATPase MsiK is involved in the uptake of (Glc) 2 and (Glc) 3 . Oligonucleotides, constructs and protocols used for gene replacement are described in the Materials and methods section (Tables 1 and 2).
The effect of cebE and msiK deletion on thaxtomin production was first assessed by visualization of the bright yellow color associated with this toxin. The cebE mutant was grown on OBA plates (complex medium which induces thaxtomin production) and on TDM (minimal medium) supplied with cellobiose as carbon source. As shown in Fig. 3A, this mutant lost the cellobiose-dependent induction of thaxtomin compared to the wild-type strain S. scabies 87-22. Surprisingly, the cebE mutant plated on solid TDM supplied with cellobiose as sole carbon source (TDMc) could still grow suggesting the existence of an additional uptake system for cellobiose in S. scabies, not involved in the thaxtomin production pathway. However almost no growth was detected when liquid TDMc was inoculated with this mutant (see Supplementary Fig. S6). In contrast, the msiK mutant was unable to grow on or in TDMc (Fig. 3A). This implies that if S. scabies does possess multiple cellobiose transporters, they all require MsiK to drive the import. To verify that deletion of the msiK gene would lead to loss of the cellobiose-dependent induction of thaxtomin, this mutant was further plated on TDM supplied with xylose (with and without cellobiose supply, abbreviated TDMx and TDMxc, respectively), a carbohydrate reported to be a carbon source that does not fully require MsiK-dependent uptake 18 . Indeed, as shown in Fig. 3A, the msiK mutant was able to grow with xylose as sole carbon source, but was unable to produce thaxtomin in response to the presence of cellobiose.
Extraction of the toxin from different media and semi-quantitative analysis by high-performance liquid chromatography (HPLC) revealed that only 10 and 20% of thaxtomin production remained for the cebE mutant grown on OBA, and TDMc, respectively, compared to the parental strain 87-22 (Fig. 3B). No thaxtomin production could be detected for the msiK mutant on OBA, TDMx, and only 2% of thaxtomin on TDMxc compared to the wild type strain S. scabies 87-22 (Fig. 3B).
We further evaluated the effect of cebE and msiK deletion on the virulence of S. scabies on radish seedlings. S. scabies typically causes severe root and shoot stunting, tissue swelling as well as necrosis, especially at the root tip (Fig. 4). These symptoms are typically attributed to thaxtomin production 26 . Both mutants showed an attenuated virulence phenotype when inoculated onto radish seedlings compared to the wild type strain (Fig. 4A,B).
Overall, these results (Figs 3 and 4) demonstrate that transport and not sensing of cellobiose is required for the induction of thaxtomin biosynthesis and thus the virulent behavior of S. scabies 87-22.
The cebEFG operon is repressed by CebR and induced by cellobiose. To further support the involvement of SCAB57751 (CebE scab ) in cello-oligosaccharide utilization we investigated how the expression of the cebEFG operon was controlled by the cellulose utilization repressor CebR. Quantitative real-time reverse transcription-PCR (qPCR) was performed on RNA extracted from the S. scabies wild-type (strain 87-22) and its cebR mutant grown on ISP-4 and OBA. On both media we observed that the deletion of cebR resulted in overexpression of cebE and cebF, whereas the expression of msiK was insignificantly altered compared to the wild-type levels ( Fig. 5A,B). The transcriptional control of CebR on cebEFG expression is mediated through a direct interaction with the cis-acting elements located at positions −131 and −14 nt upstream of cebE and bglC, respectively (Fig. 5C,D).
The effect of the deletion of cebR as well as the presence of cellobiose on the production of the CebE-MsiK transporter was further assessed with a label-free, targeted liquid chromatography-multiple reaction monitoring (LC-MRM) approach. Three proteotypic peptides for CebE, and MsiK were quantitatively analyzed by MRM: ESDYLPWK, SAFDLTAK and SGNWGGSFLSVPK for CebE, and ILDLTEYLDR, TQIASLQR and FGNSVVPVNR for MsiK. Peptides from CebE were overrepresented in extracts of the cebR deletion mutant or when the wild-type strain 87-22 was cultivated in the presence of cellobiose (Fig. 6A). Surprisingly, while we could not detect overexpression of msiK transcripts trough the qPCR approach (Fig. 5A), peptides from MsiK were 3-fold overrepresented in the ∆cebR background whereas their quantity was not increased upon cellobiose supply (Fig. 6B). The deletion of cebR had a greater effect on the overproduction of CebE and MsiK than the cultivation of S. scabies 87-22 on ISP-4 medium supplied with cellobiose. In agreement with their role in the virulence of S. scabies, the genes for the cellobiose and cellotriose cebEFG transporter are induced by the incoming carbohydrates and repressed by CebR.

Discussion
In this work we show that the cello-oligosaccharide-induced pathogenicity of S. scabies 87-22 is mediated, at least in great part, through the CebEFG-MsiK ABC transporter system. This allows to propose a first signaling pathway from cello-oligosaccharide transport to the production of thaxtomin A in S. scabies, with each step of the cascade controlled by the cellulose utilization transcriptional repressor CebR (Fig. 7). Only very few other (plant) pathogens have been shown to also use carbohydrates from the host as signals to boost their virulence behavior. The best studied phytopathogenic bacterium Agrobacterium tumefaciens recognizes its host environment by binding to diverse aldose monosaccharides as well as sugar acid derivatives through the periplasmic protein ChvE [27][28][29] . ChvE, like CebE, is a carbohydrate-binding protein associated with an ABC transporter but the molecular mechanisms involved in triggering virulence highly differ from the one highlighted in S. scabies. The ChvE-sugar complex binds to the periplasmic domain of the two-component system sensor VirA making it more sensitive to plant phenolic compounds which ultimately leads to virulence (vir) genes expression [27][28][29][30][31] . In the A. tumefaciens-plant interaction, the carbohydrates emanating from the host thus participate in a three-component sensing system instead of being directly imported into the cytoplasm to target the master regulator of the process as highlighted in S. scabies.
Importantly, CebE of S. scabies displays binding affinities for cellobiose and cellotriose which are, respectively, about 2 and 3 orders of magnitude greater than the affinity measured for CebE of the highly cellulolytic species S. reticuli (K D ~ 1.5 μM) 14 . The fact that S. scabies displays no or very poor growth on minimal media with carboxymethyl cellulose, fibrous cellulose or microcrystalline cellulose as unique carbon source 13 , whereas it responds to such low concentrations of cellobiose and cellotriose has most likely an important biological meaning. This suggests that S. scabies could possibly behave as a 'cheating' organism in soils where intense degradation of abundant plant biomass occurs, and be one of the first beneficiaries of the cello-oligosaccharides generated by efficient lignocellulolytic microorganisms. However, transport and utilization of cellobiose and cellotriose released from cellulolytic microorganisms would also trigger production of the phytotoxin thaxtomin under conditions where it is useless as the large majority of the plant material is already dead and decomposing. This is probably why additional levels of control of thaxtomin production are required. One of these is the post-transcriptional  Table 2. Primers used in this study. *Engineered restriction sites are indicated in italic, CebR-binding sites are in bold, and non-homologous extensions are underlined.
control by the bldA gene which encodes for the leucyl tRNA of the rare UUA codon which is notably found within the sequence coding for the thaxtomin production transcriptional activator TxtR 11 and the BldH regulator also required for thaxtomin production and plant virulence 10 . Since this leucyl tRNA is only abundant at late exponential phase in streptomycetes 32 , this post-transcriptional level of control would guarantee the delay of the production of thaxtomin until S. scabies has reached a certain density. Alternatively, this high responsiveness to very low concentrations of cellobiose and cellotriose could suggest a more appropriate fitness for environments where the abundance of these molecules is expected to be more marginal such as in the diffusion zone of root exudates. Interestingly, Johnson et al. detected cellotriose (but not cellobiose) released from rapidly growing radish seedling and from growing tobacco NT1 cell suspensions 13 and we show here that the affinity of CebE of S. scabies for cellotriose is greater of about one order of magnitude compared to cellobiose. The higher sensitivity to cellotriose could be an adaptation to the environment where it is appropriate for S. scabies to sense and respond to a specific composition and concentration of cello-oligosaccharides. Indeed, cellotriose could be the plant signalling molecule detected by S. scabies indicating the presence of a growing host in its surrounding. Sensing cellotriose (in a cellobiose free environment) could mean sensing a signal from an expanding plant tissue that also constitutes the site of invasion of S. scabies as well as the site of action of thaxtomin. Johnson et al. also showed that the phytotoxin increases the release of cello-oligosaccharides by expanding plant tissues. So, a small amount of cellotriose in the soil would be sufficient to initiate the production of low amounts of thaxtomin and the latter would in turn target the plant cellulose synthase resulting in a more abundant release of cellotriose that, once imported by the CebEFG-MsiK transporter, would increase thaxtomin biosynthesis. Instead of active cellulases, thaxtomin itself would generate its own trigger, cellotriose, as postulated earlier 13 .
Further structural comparative studies are required to understand why CebE of S. scabies 87-22 displays such a high affinity for cellotriose and, importantly, if this is a feature conserved in CebE proteins of other plant pathogens such as S. acidiscabies, and S. turgidiscabies that could also have elaborated the onset of their virulent behaviour upon cello-oligosaccharides uptake. Finally, at this stage, we cannot exclude that other proteins may also participate in the transport and/or the sensing of cello-oligosaccharides. Indeed, proteomic investigations in S. scabies EF-35 suggest that suberin may also participate in the induction of the cellulose/cello-oligosaccharides utilization regulon 24,[33][34][35] .

Production and purification of histidine-tagged recombinant CebE protein. The open read-
ing frame encoding SCAB57751 (CebE) without its first 132 nucleotides was amplified by PCR using primers scab_57751_+132_NdeI and scab_57751_+1365_HindIII (see Table 2). The corresponding PCR product was subsequently cloned into the pJET1.2/blunt cloning vector, yielding pSAJ015. After DNA sequencing to verify the correct amplification of scab57751, an NdeI-HindIII DNA fragment was excised from pSAJ015 and cloned into pET-28a digested with the same restriction enzymes leading to pSAJ016. E. coli Rosetta TM (DE3) cells carrying pSAJ016 were grown at 37 °C in 250 ml LB medium containing 100 μg/ml of kanamycin until the culture reached an absorbance at 600 nm (A 600 ) of 0.6. Production of 6His-tagged CebE (6His-CebE) was induced overnight (~20 h) at 16 °C by addition of 1 mM isopropyl-β-D-thiogalactopyranoside (IPTG). Cells were collected by centrifugation and ruptured by sonication in lysis buffer (100 mM Tris-HCl buffer; pH 7.5; NaCl 250 mM; 20 mM imidazole 20 mM supplemented with the EDTA-free cOmplete protease inhibitor cocktail (Roche). Soluble proteins were loaded onto a pre-equilibrated Ni 2+ -nitrilotriacetic acid (NTA)-agarose column (5-ml bed volume), and 6His-CebE eluted with 90 to 200 mM imidazole. Fractions containing the pure protein were pooled (see Supplementary Fig. S4) and dialyzed overnight in 30 mM Tris-HCl buffer; pH 7.5; NaCl 150 mM.
Tryptophan Intrinsic fluorescence assays. Pure 6His-CebE was dissolved in a 12 mM Tris-HCl buffer pH 7.5; 40 mM NaCl to a final concentration of 180 nM and analyzed at room temperature by a Cary Eclipse spectrofluorimeter (Varian Ltd). Measurements were performed with 1 ml of sample mix in a 10 mm path quartz  . *Denotes significant quantitative peptide overproduction (P < 0.05) compared to the wild-type strain grown in ISP-4 without cellobiose supply. Statistical significance was assigned by performing 2-sided Student's t-tests and assuming groups of equal variances (See Supplementary Fig. S7).

Figure 7. Model of the cellobiose/cellotriose-mediated induction of thaxtomin production in S. scabies.
Cellobiose (Glc) 2 and Cellotriose (Glc) 3 are transported by the CebEFG-MsiK ABC transporter. In the cytoplasm, (Glc) 2 and, to a lesser extend (Glc) 3 , binds to CebR and inhibits its DNA-binding ability which further increases transcription of cebE, cebF, and cebG, as well as txtR, the specific activator of the thaxtomin A biosynthetic genes txtA and txtB. Circles (○) indicate the CebR-binding sites. cuvette with an excitation wavelength of 280 nm. Each emission fluorescence spectrum is the mean of 10 scans acquisitions between 300 and 400 nm at a speed of 600 nm/minute. Excitation and emission slit lengths were 5 and 10 nm respectively. The potential difference of the photomultiplicator was fixed at 750 V. To assess the binding ability of different sugars to 6His-CebE, we first compared the fluorescence emission spectra of the pure 6His-CebE (180 nM) with or without 30 μM glucose, maltose, trehalose, cellobiose, cellotriose, cellotetraose, cellopentaose, or cellohexaose.
Determining of dissociation constants (K D ) from fluorescence data: Samples used to determine the dissociation constants (K D ) were prepared and analyzed as described above with 25 different concentrations (ranging from 15 nM to 500 nM) of cellobiose or cellotriose. Fluorescence assays were performed with six different preparations of pure 6His-CebE protein.
The proportion of 6His-CebE bound to cello-oligosaccharide was calculated using the following equation: where Fo is the fluorescence intensity of the free 6His-CebE protein, F is the fluorescence of 6His-CebE in the presence of a given cello-oligosaccharide concentration, and F max is the fluorescence of 6His-CebE in the presence of excess sugar (saturating conditions). These data were introduced in the GraphPad Prism version 5.0 (GraphPad Software, San Diego California USA) and the following nonlinear curve fitting function was used to find the dissociation constant K D using equation (2): In which [Pt] is the total 6His-CebE concentration and [St] is the total sugar (cello-oligosaccharide) concentration.

Electromobility gel shift assays (EMSAs). Interaction studies between pure His-tagged CebR and DNA
probes with the predicted CebR-binding sites (cbs) were performed by EMSAs as described previously 12 , using 15 pmol of Cy5-labelled probes and 150 pmol of CebR-6His. EMSA reactions were loaded into a 1% (w/v) agarose gel and bound and unbound probes were separated by gel electrophoresis at room temperature. Fluorescent Cy5-labbelled DNA was visualized using a Typhoon Trio+ Variable Mode Imager (GE Healthcare, laser excitation at 633 nm and emission filter at 670 nm (BP30)). The sequences of oligonucleotides used to generate the cbs probes are listed in Table 2.
Construction of cebE and msiK null mutants. Deletion mutants in S. scabies 87-22 were created using the REDIRECT © PCR targeting methodology 36 that replaces the selected gene by an antibiotic deletion cassette as previously described for cebR 12 . These cassettes consisting of an oriT and an antibiotic resistance gene (aac(3)IV for apramycin resistance, Table 1) and flanked by FRT sites (FLP-recombinase recognition targets), were generated by PCR using primers with gene-specific homology extensions (Table 2) and pIJ773 (Table 1) as template. The gel-purified deletion cassettes were electroporated into the E. coli BW25113 strain harboring the arabinose-inducible λ RED expression plasmid, pIJ790, and a cosmid containing the gene of interest. Transformants were recovered on apramycin selective medium, and correct gene replacement in the cosmid was confirmed by PCR and sequencing. The resulting mutated cosmid was then transferred to S. scabies 87-22 via intergeneric conjugation after passage through the E. coli ET12567 strain harboring pUZ8002 (Table 1). Exconjugants were selected for resistance to apramycin, and sensitivity to kanamycin. Genomic DNA was extracted from Streptomyces cultures grown in TSB medium using the MasterPure ™ Gram Positive DNA Purification Kit (Epicentre Biotechnologies) according to the manufacturer's instructions and verification of the mutant isolates was performed by PCR.
Quantitative Real-Time PCR. RNA was prepared from 72-h old mycelia grown on ISP-4 and TDMc at 28 °C using the RNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. PCRs on the purified RNA were performed to verify the absence of genomic DNA (data not shown). cDNA synthesis was performed starting from 1 μg of DNAse-treated (Turbo DNA-free Kit, Ambion) RNA using the iScript TM cDNA Synthesis Kit (BioRad). Quantitative Real-Time PCR (qPCR) was carried out in a total volume of 10 μl containing 4 μl of SsoAdvanced TM Sybr ® Green Supermix (BioRad), 4 μl of 1/10 diluted cDNA and 0.5 pmol of each gene-specific primer (Table 2), and subjected to the following PCR protocol: 3 min at 95 °C, 40 cycles of 30 s at 95 °C followed by 45 s at 60 °C. A melting curve analysis (samples were heated from 60 °C to 95 °C) was performed after each qPCR run to verify the specific amplification of each product. The murX and gyrA genes 11 were used to normalize for the amount of RNA in the samples. Each measurement was performed in triplicate with three biological replicates per strain or condition.
Targeted proteomic analysis. S. scabies 87-22 and its cebR null mutant were grown on ISP-4 plates with or without a 0.7% cellobiose supply. The mycelium was collected after 48 hours of incubation at 28 °C, suspended in 50 mM NH 4 HCO 3 buffer (pH 7.5) and crude intracellular extracts were obtained after sonication of the mycelium as described previously 37  (LC-MRM) analysis, LC-MRM analysis, and LC-MS E analysis were performed as previously described 38 and fully detailed in S1 File.
Thaxtomin extraction and quantification. Mycelial suspensions of S. scabies strains (wild-type and different isolates of each mutant) were prepared from 48-72 h-old TSB-grown cultures by pelleting the mycelia, washing twice with sterile water, and resuspending in sterile water to an A 600 of 1.0. Samples of 50 μl were plated out on small Petri dishes (5-cm diameter) containing 12.5 ml OBA or ISP-4 medium.
After incubation for 7 days at 28 °C, the medium was chopped into small cubes and soaked in 8 ml of methanol for 10 min. The supernatant was filtered through a 0.2-μm polytetrafluoroethylene (PTFE) filter and analyzed via HPLC on a Zorbax RX-C18 column (5 μm, 4.6 × 250 mm, Agilent Technologies) with a 1 ml/min flow rate of an isocratic mobile phase of 40:60 acetonitrile:water. Thaxtomin A was detected by measuring the absorbance at 380 nm. All experiments were repeated using different biological replicates of the Streptomyces strains, with three technical replicates per strain.

Plant virulence assays.
To assess the virulence phenotype of the S. scabies cebE and msiK mutants, an in vitro radish seedling assay was performed. Seeds of the "White Beauty" variety (Burpee) were surface sterilized for 5 min in 70% ethanol followed by a 10 min incubation in 15% (vol/vol) bleach. The seeds were allowed to germinate for about 30 h at 21 ± 2 °C in the dark in a Petri dish containing a moistened filter paper. Germinated radish seeds placed into six agar wells (13 mm in diameter) formed in an deep 1.5% agar-water plate were inoculated each with a 200 μl mycelial sample, prepared as described above, or sterile water as the control. The plates were incubated at 21 ± 2 °C under a 16-h photoperiod for 6 days. The assays were performed three times.

Bioinformatics.
A search for orthologues of CebE from S. reticuli (GI:5327251) in selected Streptomyces species and synteny analysis were performed using the Archaeal and Bacterial Synteny Explorer software (Absynthe) using the "best genomic match" search parameter at a 30% minimal score threshold (http://archaea.u-psud. fr/absynte/) 21 . Selected CebE homologues were used as a training set to generate the phylogenetic tree via the Phylogeny.fr platform 39 using the "One click" mode which provides a ready-to-use pipeline including the following programs programs: MUSCLE for multiple alignment, Gblocks for automatic alignment curation, PhyML for tree building, and TreeDyn for tree drawing. CebE homologues used for the analysis are listed in the legend of Fig. 1B. The new CebR position weight matrix (see Supplementary Fig. S4) was created using DNA motifs known to be bound by CebR in S. griseus 22 and S. scabies 12 with the PREDetector software 40 and the prediction of the CebR regulon was performed according to the philosophy described previously 41 . GraphPad Prism 5 software was used to determine binding parameters of CebE by fitting data obtained from the tryptophan intrinsic fluorescence analysis. Modeling of the 3D structure of the CebE protein was performed with the YASARA software 25 using as template the crystal structure of the ABC transporter Solute Binding Protein from Thermotoga Lettingae TMO (Tlet_1705, TARGET EFI-510544) bound with alpha-D-Tagatose (5CI5). Because of the low sequence identity with the model (21.8%), the overall quality Z-score is only ranked as satisfactory (−1.929). The position of Trp303 in the binding pocket of CebE is clearly established as this residue is strictly conserved and well aligned in all structures used for molecular modelling. The figure displaying this structure was obtained using the program Pymol (The PyMOL Molecular Graphics System, Version 1.7.4.3 Enhanced for Mac OS X, Schrödinger, LLC.).