A LOV Protein Modulates the Physiological Attributes of Xanthomonas axonopodis pv. citri Relevant for Host Plant Colonization

Recent studies have demonstrated that an appropriate light environment is required for the establishment of efficient vegetal resistance responses in several plant-pathogen interactions. The photoreceptors implicated in such responses are mainly those belonging to the phytochrome family. Data obtained from bacterial genome sequences revealed the presence of photosensory proteins of the BLUF (Blue Light sensing Using FAD), LOV (Light, Oxygen, Voltage) and phytochrome families with no known functions. Xanthomonas axonopodis pv. citri is a Gram-negative bacterium responsible for citrus canker. The in silico analysis of the X. axonopodis pv. citri genome sequence revealed the presence of a gene encoding a putative LOV photoreceptor, in addition to two genes encoding BLUF proteins. This suggests that blue light sensing could play a role in X. axonopodis pv. citri physiology. We obtained the recombinant Xac-LOV protein by expression in Escherichia coli and performed a spectroscopic analysis of the purified protein, which demonstrated that it has a canonical LOV photochemistry. We also constructed a mutant strain of X. axonopodis pv. citri lacking the LOV protein and found that the loss of this protein altered bacterial motility, exopolysaccharide production and biofilm formation. Moreover, we observed that the adhesion of the mutant strain to abiotic and biotic surfaces was significantly diminished compared to the wild-type. Finally, inoculation of orange (Citrus sinensis) leaves with the mutant strain of X. axonopodis pv. citri resulted in marked differences in the development of symptoms in plant tissues relative to the wild-type, suggesting a role for the Xac-LOV protein in the pathogenic process. Altogether, these results suggest the novel involvement of a photosensory system in the regulation of physiological attributes of a phytopathogenic bacterium. A functional blue light receptor in Xanthomonas spp. has been described for the first time, showing an important role in virulence during citrus canker disease.


Introduction
Light is a major environmental stimulus that regulates plant physiology. Among light-regulated vegetal responses are those elicited by the attack of pathogens, and the requirement of an adequate light environment for a full defense response has been extensively studied [1][2][3]. The perception of light has also been linked to numerous physiological responses in microorganisms such as pigment synthesis, DNA repair and biofilm formation [4]. Moreover, recent reports revealed that in many bacteria light governs important lifestyle decisions, especially that between a single-cell motile state and a multicellular surface-attached state [5]. Furthermore, the participation of light in the regulation of bacterial virulence was recently found in non-photosynthetic bacteria. The presence of light receptors across several bacterial taxa, including many species with no known photobiology, suggests that the visible light environment has an unexplored regulatory role in the biology of bacterial cells [4,[6][7][8].
The perception of light in both eukaryotic and prokaryotic organisms is conducted via photoreceptor proteins that belong to one of six families defined by the structure of their light-absorbing molecules or chromophores [9]. Among blue light receptors are proteins with LOV (Light, Oxygen or Voltage) and BLUF (Blue Light sensing Using FAD) domains. LOV domains are small photosensing protein modules (around 100 amino acids) that constitute a subclass of the widespread PAS (Per-Arnt-Sim) superfamily [10]. Several PAS-domain proteins are known to detect environmental signals by way of an associated cofactor [11], as is the case with LOV proteins. The best-characterized LOV proteins are plant phototropins, photoreceptors involved in phototropic bending, light-induced stomatal opening and lightdirected chloroplast movement [12]. LOV domains contain a molecule of flavin mononucleotide (FMN) as a non-covalently bound chromophore. For this reason, they maximally absorb light near 450 nm, and show a strong fluorescence emission at 500 nm upon the excitation of the flavin. The photochemistry of LOV domains was first elucidated for phototropins [13] and afterward for a variety of bacterial and fungal proteins [14,15]. This photochemistry involves the formation of a photoadduct (by the creation of a covalent bond between the carbon atom at position 4a of FMN and the thiol group of a conserved cysteine located in the LOV domain) that is significantly blue-shifted with respect to the dark state and it is non-fluorescent [7,16,17].
Data obtained from bacterial genome sequences revealed the presence of blue light photosensory proteins belonging to the BLUF and LOV families in many prokaryotic species [4,6]. Losi and Gä rtner found 307 proteins containing LOV domains in the genomes of 227 bacterial species [14]. Despite the large number of bacterial photoreceptors found to date, the physiological implications of these proteins are poorly understood. A few reports have been presented for the physiological function of blue light photoreceptors in important pathogens, e.g., Brucella abortus [7], Acinetobacter baumanni [18], Escherichia coli [19] and Listeria monocytogenes [20]. In non-pathogenic bacteria, photoresponses linked to LOV and BLUF proteins have been reported for Bacillus subtilis [21,22], Caulobacter crescentus [23] and Rhodopseudomonas palustris [14,24]. Regarding phytopathogens, the presence of different blue light receptors including LOV and BLUF proteins, have been reported for several microorganisms such as Xanthomonas, Pseudomonas and Ralstonia [4,25]; however, little is known about the role of these proteins on bacterial physiology. This topic has been recently discussed in a dedicated review [26]. In addition, a LOV protein from Pseudomonas syringae pv. tomato was confirmed as a blue light-regulated kinase with a still undetermined physiological role [7,27], and the light-regulated effects over Agrobacterium tumefaciens were described [28].
Xanthomonas axonopodis pv. citri is a Gram-negative bacterium responsible for citrus canker, a severe disease that affects all citrus cultivars [29,30]. The pathogen enters host plant tissues through stomata and wounds. Subsequently, bacteria colonize the apoplast causing the leaf epidermis to break due to cell hyperplasia. The disease ultimately appears as raised necrotic corky lesions on leaves, stems and fruit surfaces, reducing the fruit quality and quantity [29,30]. The X. axonopodis pv. citri genome contains three genes encoding putative blue light photoreceptors: two BLUF proteins and one LOV protein. The LOV protein (Xac-LOV) is encoded by the fixL gene, which was renamed as lov gene for clarity purposes [31,32]. In this work, we studied the potential blue light activation of the LOV protein present in this phytopathogen by evaluating blue light-induced spectral changes of the purified recombinant protein. We also investigated the influence of blue light and the LOV protein on the physiology and infectivity of X. axonopodis pv. citri. To this aim we constructed a mutant strain lacking a functional lov gene and studied the effect of the absence of this gene on bacterial physiological features and on the interaction between X. axonopodis pv. citri and its host plants.

Xac-LOV is a Typical Histidine Kinase-response Regulator Hybrid Protein
The first evidence for the occurrence of LOV proteins in bacteria was presented by Huala et al. [33], who compared the sequences of phototropin-LOV domains with those present in the LOV proteins of B. subtilis (gene ytvA) and Synechocystis PCC 6803 Q55576 (gene slr0359). In addition, Crosson et al. reported a sequence alignment including additional bacterial LOV proteins from X. axonopodis pv. citri, Xanthomonas campestris, C. crescentus, Brucella melitensis, Nostoc sp. PCC 7120, Listeria monocytogenes and Listeria innocua [8,34]. In X. axonopodis pv. citri, the LOV protein is encoded by the fixL gene (accession number AAM37406.1), which we renamed as lov gene for clarification purposes. We performed an in silico analysis of the lov gene using the reported sequence of Xanthomonas axonopodis pv. citri str. 306 (accession number AE008923.1). We identified the putative 235 and 210 promoter sequences of this gene and found a XVM2 element located 329 bp upstream of the start codon. This element has been shown to be involved in the induction of several genes related to the pathogenesis of X. axonopodis pv. citri [35].
Analysis of the predicted amino acid sequence of Xac-LOV protein (540 aa) revealed the presence of a LOV domain (aa 39-142), a Histidine Kinase domain (HK, aa 167-396) and a Response-Regulator domain (RR, aa 417-533), corroborating the hybrid nature of this photoreceptor (ExPasy-Prosite proteomic server, [36]). Figure 1A shows a scale diagram of the three functional domains of the protein, with the most relevant amino acids indicated. The LOV domain contains a highly conserved motif GXNCRFLQ (variant YXDCRFLQG) [37,38], which includes the Cys residue responsible for the formation of a covalent adduct with the FMN chromophore in all LOV proteins studied to date [6]. When we analyzed the predicted amino acid sequence of Xac-LOV using the TM Pred software [39], no potential transmembrane domains were found. This result suggests a cytoplasmic localization for the Xac-LOV protein.

The Xac-LOV Protein Presents Canonical LOV Photochemistry
To obtain the Xac-LOV protein for spectroscopic analysis, the X. axonopodis pv. citri lov gene was cloned in a pET-28a (+) vector and expressed in E. coli BL21 (DE3) Codon Plus-RIL (Stratagene). The procedure is detailed in the Supporting Information S1 and shown in Figure S1. It has been shown that following the absorption of blue light, LOV proteins generate an adduct by the formation of a covalent bond between a Cys residue from the LOV domain and the C4 carbon atom of the flavin molecule that acts as cofactor, with a two-electron reduction [12]. The adduct shows a distinct absorption spectrum, blue-shifted with respect to the unphotoactivated state, and loses its fluorescence [16]. We recorded the absorption spectra of the purified protein under dark and light conditions and obtained a light-minus-dark difference spectrum. The absorption spectrum in darkness presented the typical features of oxidized flavin chromophores with an absorption maximum at 450 nm. Following blue light illumination, we observed the loss of the absorption peak at 450 nm and the appearance of a second peak with a broad absorption band in the UVA region ( Figure 2A). The light-dark difference spectrum (inset of Figure 2A) shows a canonical LOV photochemistry for Xac-LOV. Figure 2B shows the absorption spectra of the dark-and light-adapted states of the Xac-LOV protein corrected for the scattering contribution. The light spectrum showed a maximum at ca 400 nm while maxima at 375, 450 and 475 nm were observed in the dark condition. We also recorded fluorescence spectra (450 nm excitation) of the darkness and light state of the protein.
In dark conditions, we could observe a maximum at approximately 500 nm, characteristic of the presence of FMN in the protein, which disappeared upon blue light illumination ( Figure 2B). Incomplete photoconversion (residual band around 450 nm and residual fluorescence) is due to a small fraction of free chromophore, as evidenced by the blue line showing an absorbance maximum around 530 nm, typical of free FMN. Finally, as the FMN molecule loses it fluorescence when it is part of the photoadduct and recovers it when the protein returns to its ground state, the maximum emission peak of FMN (500 nm) was used to measure the recovery time of the basal state of the protein in dark conditions after being excited with blue light ( Figure 2C). The recovery of the basal kinetic state fitted to a mono exponential function gave a lifetime (t rec ) of 5200 sec at 20uC.
Physiological Characterization of a X. axonopodis pv. citri lov Mutant To evaluate the role of the Xac-LOV protein in bacterial physiology and during the interaction between X. axonopodis pv. citri and host plants, we constructed a lov deletion mutant named Dlov. We also constructed the complemented strain, named Dlovplov. Details of the construction are explained in the Supporting Information S1 and shown in Figure S2. The viability and growth rate of these strains were comparable to those of the X. axonopodis pv. citri wild-type (WT) strain, as shown in Figure S3. We also tested the expression of Xac-LOV protein by western blot analysis using polyclonal anti-Xac-LOV rabbit antibodies (detailed on Supporting Information S1 and Figure S2Cii). We observed an inmunoreactive band for X. axonopodis pv. citri WT and complemented strains, but this band was absent in the mutant strain. A more intense band was obtained for the complemented strain compared to the WT despite the same amount of protein extracts being used for the western blot assay. This result is probably due to the low but still multiple copy number of the plasmid used for the complementation.
The Deletion of the lov Gene Alters X. axonopodis pv. citri Motility Bacterial motility is a very important attribute for pathogenic and non-pathogenic bacteria, which allows the colonization of nutrient-rich surfaces and host tissues. Swarming motility is a coordinated translocation of a bacterial population across solid or semi-solid surfaces and depends on flagella and the secretion of exopolysaccharide [34]. We studied swarming motility by inoculation of X. axonopodis pv. citri WT, Dlov and Dlov-plov strains in SB-0.7% w/v agar plates. After three days of bacterial growth at 28uC, the Dlov strain migrated further than the WT and complemented strains, both in light and dark conditions ( Figure 3A). The diameters of the migration zones confirmed the plate phenotype ( Figure 3B), showing statistically significant differences between Dlov and WT or complemented strains, but not between growing conditions (light/darkness).
To determine potential differences on the synthesis of flagellin, the major flagellum component, we collected samples from the migration zones generated by X. axonopodis pv. citri WT, Dlov and Dlov-plov under the two light conditions for protein extraction. Protein extracts were adjusted to the same protein amount and loaded onto a polyacrylamide gel to perform a western blot analysis using polyclonal anti-flagellin antibodies from Serratia marcesens. We observed an immunoreactive band corresponding to flagellin for X. axonopodis pv. citri WT in both growth conditions, but that band was hardly detected for the mutant and complemented strains (detailed on Supporting Information S1 and Figure S4). In addition, we stained the samples obtained from the border and center regions of the migration zones for flagella visualization. Figure 3C shows the microscopic images of the three X. axonopodis pv. citri strains. Pictures were taken at random regions under the microscope. The presence of flagella was observed for X. axonopodis pv. citri WT in light and dark conditions, while no flagella were observed for the mutant and complemented strains. We performed a minimum of three independent experiments, obtaining the same results.
Twitching motility is a type of bacterial translocation over moist surfaces mediated by the extension, attachment and retraction of type IV pili, previously described for several pathogenic bacteria [40,41]. In order to evaluate the possibility of such migration mechanism we assayed the X. axonopodis pv. citri behavior in conditions known to favor twitching motility [40]. When we analyzed the bacterial migration zones obtained two days after stab-inoculating X. axonopodis pv. citri WT, Dlov and Dlov-plov strains onto SB-1% w/v agar plates, we observed that WT and Dlov-plov colonies showed an irregular appearance with long bacterial extensions irradiating from the migration zones, which resemble the bacterial rafts showed by some twitching-performing bacteria ( Figure 4) [41]. Conversely, the Dlov strain developed smooth-margined colonies, with no visible bacterial extensions radiating from the central bacterial colony. Similar results were obtained when bacteria were grown in the darkness (detailed on Supporting Information S1 and Figure S5).

The Absence of the lov Gene Alters Colony Morphology and Increases Exopolysaccharide Production
Xanthomonas spp. produce a characteristic extracellular polysaccharide (EPS) named xanthan, which is responsible for the mucoid (mucus-like) appearance of bacterial colonies developed in a solid medium [42,43]. To study possible modifications on the cellular morphology of the X. axonopodis pv. citri Dlov strain, we analyzed the bacterial colonies developed on SB-1.5% w/v agar plates supplemented with 4 g/L glucose. We observed that Dlov colonies had a more glossy and mucoid appearance than WT and Dlov-plov colonies. Moreover, while the colonies of the WT and complemented strains of X. axonopodis pv. citri showed an irregular surface with scalloped margins, the colonies of the Dlov strain were straight and smooth ( Figure 5A). Bacteria were also grown on SB-agar plates supplemented with Congo red, a dye that strongly interacts with external polysaccharides. This dye is widely used for the detection of curly fibers, thin aggregative fibers present in several microorganisms [44,45]. Although the morphology differences described above were maintained, there were no differences in Congo red binding between X. axonopodis pv. citri strains (detailed on Supporting Information S1 and Figure S6). On the basis of these observations we quantified xanthan from the supernatants of Multiple alignments of the deduced amino acid sequences of LOV proteins from X. axonopodis pv. citri (Xac), Xanthomonas campestris pv. vesicatoria (Xcv), Xanthomonas oryzae pv. oryzae (Xoo), Xanthomonas campestris pv. campestris (Xcc), Pseudomonas syringae pv. syringae (Pss), Pseudomonas syringae pv. tomato (Pst), Pseudomonas syringae pv. phaseolicola (Psp) and Xanthomonas albilineans (Xal); performed using ClustalX [83]. The LOV, HK and RR domains are highlighted in blue, yellow and green, respectively. An asterisk indicates complete residue conservation, a colon indicates strong group conservation, a period indicates weak group conservation, and a blank space indicates no conservation of residues. doi:10.1371/journal.pone.0038226.g001 two-day-old cultures of X. axonopodis pv. citri WT, Dlov and Dlovplov, and found that both in light and dark conditions, the Dlov strain produced a statistically significant higher amount of EPS compared to the WT or the complemented strains ( Figure 5B).
We also assessed bacterial survival in the presence of hydrogen peroxide and found that the Dlov strain was three-fold more sensitive to these treatments compared to the WT strain, suggesting a potential protective role of lov gene upon oxidative stress. The level of survival of the Dlov-plov strain resulted intermediate between the WT and Dlov strains (detailed on Supporting Information S1 and Figure S7).

Biofilm Formation is Affected by the Deletion of the lov Gene
Biofilm formation is associated with the production of exopolysaccharide and is important for the virulence of some pathogenic bacteria because it can promote their survival against the action of antimicrobial compounds derived from host organisms [46]. We used confocal laser scanning microscopy (CLSM) to analyze the morphology of bacterial biofilms developed by green fluorescent protein (GFP)-labeled strains of X. axonopodis pv. citri on chambered cover glass slides over different periods of time. The X. axonopodis pv. citri WT, Dlov and complemented strains were able to develop complex structures consisting of clustered bacteria in close contact with each other. However, as shown in Figure 6A, we observed different patterns of bacterial aggregation between the different strains. After two days of incubation under light conditions, the three strains generated isolated microcolonies in which bacteria were densely packed and organized, maintaining lateral interactions with each other, however these structures were larger for the WT and complemented strains. After five days, the WT and complemented strains generated aggregates that extended over the entire surface, while the microcolonies developed by the Dlov strain remained disperse. In dark conditions, after two days of incubation, the pattern of cell aggregation of the WT and complemented strains was similar to that observed for Dlov under light conditions. However, the Dlov strain generated even smaller structures that were considerably less organized than those generated under light conditions. After five days, cell aggregation patterns were rather similar for the three strains with the formation of a cellular matrix that filled the entire surface of the chamber bottom.
We also assayed the ability of X. axonopodis pv. citri to develop a biofilm on glass tubes containing SB liquid medium, in order to evaluate its aggregating ability after longer incubation times. After two weeks of static incubation at 28uC in different illuminating conditions, we observed cell and EPS aggregates on the air-liquid interface. In the case of the Dlov strain, we observed highly flocculated aggregates that maintained a limpid solution sinus. In contrast, the WT and complemented strains showed less dense aggregates on the interface with a turbid solution sinus ( Figure 6B). For both assays we performed a minimum of three independent experiments with the same results.
The Absence of the lov Gene Affects in vitro Adhesion of X. axonopodis pv. citri The ability to adhere to host tissues is essential for successful infection by many microbial pathogens [43]. To analyze this phenomenon, we studied the ability of X. axonopodis pv. citri strains to adhere to abiotic and biotic surfaces. Assays were performed by growing bacteria in XVM2 medium, a minimum medium that simulates conditions of the apoplastic space [47,48]. For the abiotic assay, we employed polyvinylchloride (PVC) plates on which bacteria were grown for six hours at 28uC. After washing and staining the plates with Crystal violet, we observed that under light conditions, the adhesion of the Dlov strain was diminished compared to the WT strain ( Figure 7A). The adhesion of the complemented strain was similar to that of the mutant strain. On the other hand, when the assay was performed under dark conditions, all of the X. axonopodis pv. citri strains presented a reduced adhesion compared to adhesion in the light condition. Bacterial attachment measured by the spectroscopic quantification of the bound dye at 540 nm confirmed that in the presence of light, the adhesion of Dlov and Dlov-plov was statistically significant lower than that of the WT ( Figure 7B). In contrast, under dark conditions, there were no significant differences between the adhesion of the different X. axonopodis pv. citri strains.

Xac-LOV is Crucial for Bacterial Adhesion to Orange Leaves
To study the adhesion of X. axonopodis pv. citri strains to biotic surfaces bacteria were grown in XVM2 medium and placed on the surface of orange leaves. After six hours of incubation at 28uC, the leaves were washed and stained with Crystal violet. When the assay was performed under light conditions, the adhesion ability of the X. axonopodis pv. citri Dlov strain was much lower than for the WT and complemented strains ( Figure 8A). In the dark condition, the adhesion of all X. axonopodis pv. citri strains was very low. Bacterial attachment was quantified by digital image analysis (spot density) of the stained leaves ( Figure 8B). This quantification demonstrated that under light conditions the adhesion of the Dlov strain was statistically significant lower than that of the WT or complemented strains. In darkness, the differences in the adhesion of the three strains were not statistically significant, but we found a significant reduction in the adhesion of the three X. axonopodis pv. citri strains compared to the corresponding adhesion in the light condition.
Bacterial adhesion is highly related with the presence of adhesins. When we evaluated the expression of the fhaB gene, encoding a filamentous hemagglutin-like adhesin, we could observe that the Dlov strain presented a reduced expression of this gene compared to the WT and Dlov-plov strain (detailed on Supporting Information S1 and Figure S8).
The Deletion of the X. axonopodis pv. citri lov Gene Modifies Plant Disease Symptoms The ability of X. axonopodis pv. citri strains to develop disease symptoms in Citrus sinensis (orange) leaves was studied by inoculation of the leaves with cultures of WT, Dlov and Dlov-plov strains adjusted to 10 7 colony forming units (CFU)/mL. The infiltrated leaves were maintained in conventional photoperiod (light condition) and in the darkness. We analyzed the bacterial growth at different times after inoculation of the leaves, and observed that it was similar for the three X. axonopodis pv. citri strains until two weeks after inoculation ( Figure 9A). In contrast, the symptoms caused by the mutant strain were phenotypically different than those generated by the WT and complemented strains. In the light condition, while the tissue regions inoculated with the WT and complemented strains showed typical canker lesions, the region inoculated with the Dlov strain presented a highly necrotic appearance. In the dark condition the aspect of the lesions generated by the three X. axonopodis pv. citri strains resulted similar, with a high level of necrosis in the inoculated tissue ( Figure 9B). We quantified the degree of necrosis developed in the orange leaves inoculated with the different bacterial strains as the rate between the necrotic area and the total inoculated area, corroborating the direct observation ( Figure 9C).

Discussion
Prokaryotes have evolved a repertoire of photosensory proteins that detect the visible light environment to regulate cell physiology. Diverse classes of prokaryotic photoreceptors with different regulatory roles have been recently identified and several reports on heterotrophic bacteria have revealed physiological functions regulated by blue light receptors [5]. The X. axonopodis pv. citri genome has three putative blue light photoreceptors, two with BLUF domains and one with a LOV domain [31,32]. The high representation of these genes in X. axonopodis pv. citri led us to suggest a probable relevance of blue light for its survival and for the colonization of host plants. The Xac-LOV protein consists of three distinct domains: a LOV domain, a HK domain and an RR domain. This protein has several conserved key functional amino acid residues known to be important for photochemistry and signaling (Figure 1). Many of the studied LOV domains are found together with HK motifs. The organization of hybrid LOV-HK-RR proteins is found almost exclusively in bacterial plant pathogenic species such as Xcv, Xoo, Xcc, Pss, Pst, Psp and Xal [14,25]. Other bacterial LOV proteins with associated HK activity are those from B. melitensis, B. abortus and Erythrobacter litoralis, but these proteins do not include the RR domain [7]. All of these sensor/HK proteins are hypothesized to be involved in blue light-driven twocomponent signaling systems at some stage of the bacterial life cycle [49,50]. The LOV proteins present in some pathogenic bacteria, such as P. syringae and B. melitensis, were reported to be induced in conditions known to promote bacterial virulence [7,26]. Accordingly, we found that the promoter region of the lov gene includes a regulatory XVM2 element known to be involved in the induction of several genes related to the pathogenesis of X. axonopodis pv. citri [35,48]. All of these results suggest that LOV proteins are likely involved in bacterial virulence.
We evaluated the ability of the Xac-LOV protein to sense and respond to blue light. Studies of absorption and fluorescence spectroscopy using a recombinant Xac-LOV protein showed that facing blue light illumination, this protein has spectral variations typical of the formation of a photoadduct (Figure 2A, B). The spectral properties of the Xac-LOV protein are consistent with those exposed by several bacterial LOV proteins and by plant phototropins [16,17,51], demonstrating that Xac-LOV is in fact activated by blue light. Most LOV domains complete a photocycle from the photoadduct back to the ground state in the darkness, with half-lives of seconds to several minutes [52,53]. In the case of Xac-LOV, the photoadduct showed a very slow rate of reversion to the ground state in the darkness (t rec 5200 s) ( Figure 2C). These results are consistent with those presented by Cao et al., who showed that the Pst LOV protein has the typical LOV photochemistry with a recovery time in the darkness of 5650 s [27]. Differences in the kinetics of adduct formation and rupture among LOV proteins can be explained by variations in the residues that make up the binding pocket of the flavin cofactor [12,54,55] and in more distant amino acids [14,56].
To evaluate the potential role of Xac-LOV protein in bacterial physiology and host plant colonization, we constructed a lov deletion mutant (Dlov strain). The analyses of growth curves in liquid medium indicated that the lov gene is not essential for X. axonopodis pv. citri viability ( Figure S3).
Bacteria use a variety of motility mechanisms to colonize host tissues. These mechanisms include flagella-dependent swimming and swarming and flagella-independent twitching, gliding and sliding [34]. Previous work from our laboratory demonstrated that X. axonopodis pv. citri exhibits swarming motility and that this motility depends on flagella and EPS secretion into the medium [57]. In the present study, we observed an increased motility for the Dlov strain compared to the WT and complemented strains, both in light and dark conditions ( Figure 3A, B). These results suggest that the lov gene may be a component of the bacterial motility regulation network. X. axonopodis pv. citri possess a single polar flagellum, which is composed of multiple units of flagellin protein, encoded by the fliC gene [31,58]. Western blot analysis revealed a decreased flagellin content in the Dlov strain in both lighting conditions tested ( Figure S4). Accordingly, by optical microscopy analysis, we observed a lower number of flagellar structures in the mutant strain of X. axonopodis pv. citri compared to the WT strain ( Figure 3C). These results suggest a role of the lov gene in flagella development. Although the complemented strain showed a reversion of the lower migration phenotype, we could not observe the recovery of the WT levels of flagella in this strain. The fact that no complementation was observed for flagella synthesis could be a consequence of the lack of fine control in the expression of the lov gene on the complemented strain. Although this strain was constructed by introducing the lov gene under the control of its own promoter on a low copy vector, the expression levels of the lov gene are not necessarily identical to the normal expression levels of this gene in the WT strain. In fact, a western blot analysis showed a higher lov expression for the Dlov-plov strain compared to the WT strain ( Figure S2Cii). As reported for other bacterial species LOV proteins act at the first level of the signal transduction process affecting several bacterial features [5]. It is possible that some of these features (such as flagella synthesis) require a tightly controlled level of the regulator proteins while for other processes a wider range of the regulator protein level is adequate to maintain the normal behavior. While the level of expression of the lov gene provided by the vector used for complementation was adequate to achieve the complementation of some physiological aspects, such as EPS production, it is possible that regulation of flagella synthesis requires a more tightly regulated level of the Xac-LOV protein.
We also studied bacterial motility in conditions that favor twitching, a flagella-independent mechanism of bacterial translocation over moist surfaces directed by polar type IV pili and characterized by the formation of bacterial rafts on the colony borders [41]. We found that the X. axonopodis pv. citri strains behave differently on twitching plates. While WT and Dlov-plov colonies exhibited irregular borders, Dlov colonies showed smooth edges in the migration zones ( Figure 4). These results suggest that X. axonopodis pv. citri can perform a twitching-like motility, and that the lov gene has a role in this type of translocation.
Another bacterial feature important for host plant colonization is the production of EPS. Xanthomonas spp. produce a major EPS named xanthan, which was previously reported to be important for epiphytic survival [42], swarming motility [57] and biofilm formation [59] of X. axonopodis pv. citri. We found that the Dlov strain produces more EPS than the WT and complemented strains ( Figure 5). This result is consistent with the increased bacterial motility of the Dlov strain despite the reduction of flagella [46,57]. When we evaluated the interaction of Congo red stain, which interacts with extracellular components such as curly fibers [44], no differences were found between the different X. axonopodis pv. citri strains, indicating no modifications in these structures ( Figure S6).
Biofilms are microbial communities immersed in a selfproduced exopolysaccharide matrix and attached to an inert or living surface [60][61][62]. Several reports indicate that biofilms are important for the virulence of many pathogenic bacteria [46,63]. In X. axonopodis pv. citri, the EPS and flagellum have been associated with the formation of mature biofilms [59,64]. When we studied early biofilm formation by CLSM, we could observed that the X. axonopodis pv. citri WT, mutant and complemented strains were able to develop complex cellular structures, however these structures were smaller and more dispersed for the Dlov strain ( Figure 6A). When we analyzed the bacterial aggregation on the air-liquid interface of glass tubes after a static incubation of X. axonopodis pv. citri strains for two weeks, we observed thicker cellular aggregates for the Dlov strain compared to the WT and complemented strains ( Figure 6B). We hypothesized that the Dlov strain could have impaired the early developmental stage of biofilm formation, but after longer incubation times its aggregating ability results higher than for the WT strain. This phenomenon can be explained by the major involvement of different structures on the different stages of biofilm formation. Therefore, considering that flagellum-dependent motility and bacterial attachment are essential for the initiation of biofilm development, the decrease in flagellum synthesis and the low adhesion ability observed for the Dlov strain of X. axonopodis pv. citri can explain the initial impairment of biofilm formation of this strain (65,66). On the other hand, the secretion of exopolysaccharide is known to be important for late stages of biofilm formation (establishment of mature biofilms) [65,66]. In this context, the increased levels of xanthan measured for the Dlov strain are consistent with the larger bacterial aggregates observed for this strain after longer incubation periods.
Bacterial attachment is a crucial early step in the pathogenicity process. We evaluated the adhesion capacity of the different X. axonopodis pv. citri strains to abiotic and biotic surfaces employing XVM2 as a growth medium, known to improve the expression of adhesins genes [57]. In both assays we found that the adhesion ability of X. axonopodis pv. citri was reduced by the deletion of the lov gene, and that the adhesion of the WT strain was diminished in the darkness (Figures 7 and 8). These results suggest that X. axonopodis pv. citri adhesion requires a functional lov gene and that the attachment process is dependent on the presence of light. Our results are similar to those obtained in Caulobacter crescentus, where it was demonstrated that light regulates cellsurface attachment through a LOV protein [23,67]. The fact that the adhesion of the Dlov strain was also reduced in darkness compared to light suggests an additive effect on bacterial adhesion. This could be explained by the simultaneous participation of Xac-LOV and other X. axonopodis pv. citri photosensory proteins in the regulation of bacterial attachment. It should be noted that in addition to the lov gene encoding the LOV protein, the X. axonopodis pv. citri genome contains other putative photoreceptors, i.e., two genes encoding BLUF proteins and one encoding a phytochrome [31,32]. Several factors can contribute to bacterial adhesion, namely fimbrial and non-fimbrial adhesins, EPS and flagella [68][69][70][71][72]. In a previous study, it was demonstrated that plant tissue attachment is dependent on the coordinated action of specific adhesins and EPS, and that X. axonopodis pv. citri grown in XVM2 shows minor EPS production [57]. This suggests that the differences in adhesion between X. axonopodis pv. citri WT and Dlov could be a consequence of alterations in bacterial adhesins. The expression of the fhaB gene, encoding a filamentous hemagglutinlike protein known to be involved in the bacterial adhesion to abiotic surfaces and host tissues [57], resulted reduced for the Dlov strain compared with the WT strain. This result can explain the decrease of the adhesion ability of the Dlov strain.
When we performed the assays with the Dlov-plov strain, we only observe a reversion to the WT phenotype in the adhesion to orange leaves. As mentioned above, the main factors that contribute to the bacterial adhesion to abiotic and biotic surfaces are adhesins, EPS and flagella [68][69][70][71][72]. Taking in consideration that the complemented strain showed a reversion to the fhaB adhesin gene expression, but was not able to restore the synthesis of flagella, the absence of complementation during the in vitro adhesion assay could be due to the absence of this structure in the Dlov-plov strain. The different outcome for the adhesion of the complemented strain during the in vitro and in planta assay could be a consequence of the major involvement of different factors during the adhesion to the diverse surfaces, possibly being the bacterial flagellum more relevant for the adhesion to abiotic surfaces, and the adhesins mainly involved in the adhesion to host tissues.
X. axonopodis pv. citri is an obligate aerobic phytopathogen and consequently, is exposed to hydrogen peroxide produced by normal aerobic respiration and as a key component of the host immune response [73]. Pathogens need to prevent and overcome oxidative stress to establish and maintain infections [74]. We found that the Dlov strain presents an increased sensitivity to oxidative stress ( Figure S7), suggesting a potential participation of the product of the lov gene in the protective mechanisms triggered upon this type of stress. Since intense blue light can result in oxidative damage to cells by the light-driven formation of reactive oxygen intermediates [75], it could be advantageous for the bacterium to sense this condition to activate appropriate defense systems.
Finally, we analyzed the role of the Xac-LOV protein in the interaction between X. axonopodis pv. citri and host plants, and we observed that the virulence process was affected in the Dlov mutant. The orange leaves infected with the mutant strain displayed phenotypically different lesions than those rendered by the WT and complemented strains, generating visible necrotic regions on the infected tissues. This result suggests the participation of the Xac-LOV protein in the regulation of the virulence process, likely preventing excessive tissue necrosis. The modulation of tissue damaging is a mechanism that allows X. axonopodis pv. citri, a hemibiotrophic pathogen, to remain in the infected tissue long enough to proliferate and to spread to neighboring tissues [76]. In this context, the lov gene could be involved in the control of the host tissue damage caused by the phytopathogen. The loss of the control of tissue damage during the infection performed in darkness reflects the light-dependent nature of this process. Studies performed with B. abortus and B. mellitensis revealed a similar role for LOV photoreceptors, as the LOV proteins present in these pathogens were found to be implicated in the pathogenesis process [7]. It is worth mentioning that the complemented strain of X. axonopodis pv citri was able to restore the normal appearance of the light lession, in spite of its inability to restore the normal flagella levels. This result is consistent with the one obtained by Malamud et al., in which a mutant strain of X. axonopodis pv citri unable to synthetize flagellin, exhibited an only slightly reduced pathogenicity compared to the wild-type X. axonopodis pv. citri strain [64].
We have alternative hypothesis to explain the fact that the complemented strain of X. axonopodis pv. citri was able to restore some, but not all of the processes studied in this work. First, as we mentioned above, the higher expression of the lov gene in the complemented strain compared to the WT strain could be unfavorable for the reversion of features such as flagellum synthesis, which require a fine control of the signaling molecules. On the other hand, keeping in mind that X. axonopodis pv citri presents other photoreceptor proteins including two blue lightsensing BLUF proteins and one red light-sensing phytochrome, we cannot rule out the involvement of these proteins in the regulation of some of the light-dependent mechanisms discussed in this work. Further studies with bacterial mutants in those photoreceptors will help us uncover the entire light regulated mechanism of the lifecycle and virulence of this pathogen.
In conclusion, we demonstrated that the X. axonopodis pv. citri lov gene encodes a functional photoreceptor protein that is activated by blue light, generating a signaling state that is probably involved in a downstream activation cascade. Moreover, we found that bacterial motility, EPS production, biofilm formation and adhesion are among the processes influenced by the product of the lov gene. Because the Xac-LOV protein is likely located at the first level of a signal transduction pathway, it is expected that this protein controls downstream components involved in more than one physiological feature. Although many of the studied physiological aspects showed a clear dependence on the lov gene, several of them did not have an apparent light regulation. This phenomenon could imply that other stimuli may be involved in the modulation of Xac-LOV activity. Regarding this, it has been shown that the photoactivity of some LOV and BLUF proteins can be regulated by the redox balance of the bacterial cytoplasm [77] or by temperature [18]. However, a clear light-dependent regulation was shown for X. axonopodis pv. citri adhesion to orange leaves, suggesting that the environmental light could modulate this process through the Xac-LOV protein. As bacterial attachment is crucial for the successful colonization of host plant tissues, this result suggests that the Xac-LOV protein and the light environment play an important role during X. axonopodis pv. citri host colonization. More significantly, the evaluation of the disease symptoms caused by the lov deletion mutant demonstrated that light and the Xac-LOV protein have an essential role in the virulence process, being involved in the control of the host tissue damage caused by the phytopathogen. It is worth mentioning that this report revealed the novel contribution of a photosensory system in the physiology of a phytopathogenic bacterium. Furthermore, this is the first report of a functional blue light receptor in Xanthomonas spp. and the first genetic evidence of a bacterial LOV protein involved in the control of bacterial virulence during citrus canker disease.

Plasmids, Bacterial Strains and Growth Conditions
The bacterial strains and plasmids used in this study are listed in Table 1. X. axonopodis pv. citri cells were grown aerobically at 28uC with shaking at 200 rpm in Silva Buddenhagen (SB) medium [78] or in the minimal medium XVM2 [48] supplemented with the corresponding antibiotics. All X. axonopodis pv. citri strains were derivatives of the strain Xcc 99-1330, which was kindly provided by Blanca I. Canteros. E. coli cells were grown aerobically at 37uC with shaking at 250 rpm in Luria Bertani (LB) medium [79]. Antibiotics were used at the following final concentrations: ampicillin (Amp), 25 mg/mL for X. axonopodis pv. citri and 100 mg/mL for E. coli; kanamycin (Km), 40 mg/mL for both bacteria; chloramphenicol (Cm), 20 mg/mL for E. coli; streptomycin (Sm), 50 mg/mL for X. axonopodis pv. citri and 100 mg/mL for E. coli; gentamycin (Gm), 40 mg/mL for X. axonopodis pv. citri.
The assays were performed under different lighting conditions. For the light conditions, bacteria were grown on a chamber with continuous blue light (l = 462 nm, 4.115 mE/m 2 s) provided by LEDs at 28uC. The dark condition was generated by covering flasks or plates with aluminum foil.

Spectroscopic Analysis
Absorbance spectra were recorded with a Jasco 7850 UV/Vis spectrophotometer. Steady-state fluorescence measurements were carried out with a Perkin-Elmer LS50 luminescence spectrometer. All measurements were done at 20uC using 1 cm light-path quartz cuvettes. Photoequilibrium conditions, with the accumulation of the photoactivated state (adduct), were achieved by illuminating the sample with a blue light-emitting Led-LenserHV8 lamp (max 462 nm) (Zweibrüder Optoelectronics, Soelingen, Germany) as previously described [80].

Bacterial Motility Assays
To analyze swarming motility, X. axonopodis pv. citri strain saturated cultures were subcultured into fresh SB medium at 2% inoculum and grown to late-exponential phase (15 h) in light and dark conditions. Bacteria were harvested by centrifugation and resuspended in distilled water, adjusting the cultures to 10 7 CFU/ mL. Aliquots of 3 mL were inoculated on the center of SB-0.7% w/v agar plates and incubated at 28uC in a moist chamber in light and dark conditions [34]. The migration zones were analyzed after three days of growth.
Twitching motility was analyzed as described by Semmler et al. [40]. Briefly, X. axonopodis pv. citri strains were stab-inoculated from a two-day-old SB-1.5% w/v agar plate with a sterile toothpick through a thin (approximately 3 mm) SB-1% w/v agar layer to the bottom of the Petri dish. After incubation at 28uC for two days, the zone of motility was visualized by staining with 0.05% w/v Coomassie Brilliant Blue R250 in 40% v/v methanol-10% v/v acetic acid.

Flagella Stain
Flagella were stained as described by Kearns and Losick [81] with some modifications. Briefly, the dye was prepared by mixing 10 parts mordant (2 g tannic acid, 10 mL 5% w/v phenol, 10 mL saturated aqueous AlKO 8 S 2 ?12H 2 O) with one part 12% w/v Crystal violet in ethanol. Two-day-old swarming bacteria from the border and center regions of the migration zones were picked with

Colony Morphology and EPS Production
For colony morphology observation, bacteria were grown on SB-1.5% w/v agar plates containing 4 g/L glucose.
To quantify the EPS production, strains were cultured in SB broth for three days under light and dark conditions. Bacteria were harvested by centrifugation, and EPS was precipitated from the culture supernatant by the addition of two volumes of ethanol. The precipitate was vacuum filtrated and weighed [78].
Biofilm Formation Assay X. axonopodis pv. citri strains were modified to express green fluorescence protein (GFP). Briefly, the coding sequence for EGFP from pEGFP-1 (Clontech, Palo Alto, CA, U.S.A.) was digested with BamHI and XbaI and ligated in-frame with the LacZ-apeptide of the broad-host-range vector pBBR1MCS-5 [82] previously digested with the same enzymes, rendering the plasmid pBBR1MCS-5EGFP. E. coli S17-1 cells harboring this plasmid were conjugated to the different X. axonopodis pv. citri strains, and transconjugants were selected for Gm resistance. Saturated cultures of the GFP-labeled bacteria in SB medium were adjusted to the same optical density at 600 nm (OD 600 ) and diluted 1:100 in fresh medium, and 300 mL was placed onto chamber-covered glass slides (nu155411, Lab-Tek, NUNC, Naperville. IL, U.S.A.). Chambers were statically incubated in a humidified PVC-box in light and dark conditions. Biofilm formation was visualized by confocal laser scanning microscopy (Nikon Eclipse TE-2000-E2) with a motor system and DIC/Nomarski optics and a head scan D Eclipse C1si. The images obtained were analyzed with Nikon EZ-C1 3.90 software.
Biofilm formation was also studied on glass tubes. Saturated cultures of X. axonopodis pv. citri in SB medium were adjusted to the same OD 600 . Subsequently, 20 mL of each bacterial culture was transferred to glass tubes containing 2 mL fresh medium and statically incubated at 28uC in light and dark conditions. Bacterial aggregates were visually examined after two weeks of incubation [45].
In vitro and in planta Adhesion Assays X. axonopodis pv. citri adhesion to abiotic surfaces was assayed on polyvinyl chloride (PVC) 96-well microplates. Overnight cultures of X. axonopodis pv. citri strains were harvested by centrifugation and cell pellets were washed and resuspended in fresh media to the same OD 600 . Each well was filled with the bacterial suspension or 1:10 dilutions in a final volume of 100 mL. Wells with media and water were included as negative controls. The plates were incubated for 6 h at 28uC under light and dark conditions. After incubation, the plates were washed with distilled water to remove non-adherent cells, and 25 mL of 1% w/v Crystal Violet was added to each well and incubated for 15 min at room temperature. After removing the excess dye by washing the plates with distilled water, 200 mL of 95% v/v ethanol was added to each well and then transferred to a 1.5 mL eppendorf tube and the volume was adjusted to 1 mL with distilled water. Bacterial adhesion was quantified by determining the absorbance at 540 nm.
X. axonopodis pv. citri adhesion to biotic surfaces was assayed on orange leaves. Bacteria were grown and processed as described previously, but in this case, 20 mL of each bacterial suspension was placed on the abaxial face of the leaves and incubated for 6 h at 28uC in a humidified chamber in light and dark conditions. Bacterial adhesion was analyzed by Crystal Violet staining of the leaves for 15 min at room temperature, and the unbound dye was removed by gentle washing with distilled water. Bacterial attachment was measured by digital image analysis of the luminosity density of the spots (luminosity/pixel area). For plant inoculation, bacteria were cultured in SB broth to an OD 600 of 1, and cultures were adjusted to 10 7 CFU/mL with 10 mM MgCl 2 . Bacterial suspensions were infiltrated into the abaxial leaf surface using a needleless syringe. MgCl 2 was used as a control for non-infected leaves. For the dark condition, the infiltrated leaves were protected from light with cardboard envelopes. To study in planta bacterial growth, leaves were taken at different days after infiltration and 0.8 cm diameter leaf discs were obtained from the infiltrated zones. The discs were ground in 100 mL of 10 mM MgCl 2 followed by serial dilution and plating onto SB agar plates. Colonies were counted after 48 h of inoculation at 28uC. To quantify tissue damage we analyzed the percentage of necrotic area per infiltrated leaf area by digital image analysis. The pixels of the respective areas were measured using the histogram function of the program AdobePhotoshop 7.0 and the following rate was calculated: 100 X (necrotic area pixels/ total infiltrated area pixels).

Statistical Analysis
The quantitative analyses were performed with at least three independents biological samples. In each case, the plotted data correspond to the mean of these independent determinations, with the corresponding standard error indicated by the error bars. Data were subjected to a multifactorial ANOVA and Tukey's multiple comparison tests using Infostat software (Infostat 2006H, http/ www.infostat.com.ar). The outcome of the statistical analysis was represented using different letters to indicate the existence of statistically significant differences between the data or identical letters to indicate the absence of these differences. Dlov mutant strain using the suicide plasmid pKMobGII to replace the X. axonopodis pv. citri lov gene with a Sm/Sp-resistance cassette. (B) Construction of the plasmid carrying a copy of the lov gene (promoter and coding region) for transformation of the Dlov strain to generate X. axonopodis pv. citri Dlov-plov strain. (Ci) PCR with X. axonopodis pv. citri lov-specific primers F1 and R1 using genomic DNA from X. axonopodis pv. citri strains as template. 1: WT; 2: Dlov; 3: Dlov-plov; 4: positive control (plov); 5: negative control (water). (Cii) Western blot analysis using polyclonal anti-Xac-LOV antibodies. UFS (upstream flanking sequence); DFS (downstream flanking sequence). (TIF) Figure S3 Bacterial growth curves in liquid SB medium. Saturated cultures of X. axonopodis pv. citri WT, Dlov and Dlovplov strains were subcultured into fresh SB medium at 2% v/v inoculums. Bacterial growth curves were obtained considering the optical density at 600 nm (OD 600 ) (A) and the colony forming units (CFU)/ml (B) as a function of time. Data are represented as the mean +/2 standard error of three independent biological samples and different letters above the bars indicate significant differences between the corresponding data (p,0.01). (TIF) Figure S4 Flagellin synthesis in X. axonopodis pv. citri Dlov and Dlov-plov strains. Western blot analysis using polyclonal antiflagellin antibodies from protein extracts of bacteria obtained from the migration zones of swarming plates. (TIF) Figure S5 Twitching-like motility of X. axonopodis pv. citri strains in dark conditions. X. axonopodis pv. citri WT, Dlov and Dlov-plov strains were stab-inoculated on SB-1% w/v agar plates and grown for two days at 28uC in the absence of light. To analyze the borders of the migration zones, the plates were observed under a magnifying glass (10X), prior (upper panels) and after (lower panels) staining with Coomassie Brilliant Blue R250. (TIF) Figure S6 Analysis of extracellular structures of X. axonopodis pv. citri strains. X. axonopodis pv. citri WT, Dlov and Dlov-plov colonies were analyzed on SB-1.5% w/v agar plates supplemented with 40 mg/mL Congo red dye. (TIF) Figure S7 Hydrogen peroxide resistance of X. axonopodis pv. citri WT, Dlov and Dlov-plov strains. Cells in the early exponential phase of growth were exposed to the indicated concentrations of hydrogen peroxide (H 2 O 2 ) for 15 min. The number of colony forming units (CFU) was determined for each culture before and after the peroxide treatment by plating appropriate dilutions. The percentage of survival is defined as the number of CFU after treatment divided by the number of CFU prior to treatment 6 100. Data are represented as the mean +/2 standard error of three independent biological samples and different letters above the bars indicate significant differences between the corresponding data (p,0.01). (TIF) Figure S8 Expression of X. axonopodis pv. citri adhesin gene in XVM2 medium. (A) Amplified products of the fhaB gene by semiquantitative RT-PCR using RNA preparations from early exponential X. axonopodis pv. citri cultures grown in XVM2. As a control for constitutive bacterial expression a fragment of 16S rRNA was simultaneously amplified. (B) Expression profiles obtained by densitometric quantification of band intensities. Data are expressed as the mean +/2 standard error of three independent samples. I.O.D: integrated optical density; a.u.: arbitrary units.