Unravelling potential virulence factor candidates in Xanthomonas citri. subsp. citri by secretome analysis

Growth conditions

Stocks of wild-type Xac and the ΔhrpB4 mutant strain were stored in phosphate-buffered saline at room temperature. Both were cultured in Nutrient Agar plates prior to the preparation of the pre-inoculum. All maintenance cultures were maintained at 28 °C, and kanamycin was added at 100 µg/mL for solid culture media and 50 µg/mL for liquid culture media. Wild-type Xac and ΔhrpB4 were cultured in a 15-mL liquid Nutrient Broth (NB) pre-inoculum, maintained at 200 rpm and 28 °C, until the cultures reached OD_600nm = 1.0 (approximately 10⁸ CFU/mL). Cultures were centrifuged at 3,100 × g for 15 min and the supernatant was discarded. The pellet was resuspended in water and added to the culture medium. Both wild-type Xac and ΔhrpB4 were grown in 500 mL NB culture medium (200 rpm, 28 °C) until the culture reached an OD_600nm of at most 1.4 (approximately 10⁹ CFU/mL). At OD_600nm less than 1.4, growth is still logarithmic, which prevents the release of proteins into the culture medium due to cell lysis, thus enabling us to accurately profile the secretome (Alexander Watt et al., 2005). XAM1 media (7.5 mM (NH₄)₂SO₄, 33 mM KH₂ PO₄, 60 mM K₂ HPO₄, 1.7 mM sodium citrate [C₆ H₅ Na₃ O₇⋅2H₂O], 0.9 mM MgSO₄, 9.9 mM fructose, 9.9 mM sucrose and 0.03% of Casamino acids in a final volume of 1 L) was used to induce virulence (Wengelnik & Van den Ackerveken, 1996). Each bacterium was grown in 2 L (4 × 500 mL) of this medium, to which 1 mg/mL of Bovine Serum Albumin (BSA) was added. Bacteria were grown at 28 °C with shaking at 200 rpm until the culture reached an OD₆₀₀ of at most 0.57, wherein growth was still logarithmic.

Wild-type Xac or ΔhrpB4 were grown for 16 or 23 h in NB liquid culture media until it reached an OD₆₀₀ of 1.0 or 1.4, respectively. Both strains were also grown in XAM1 liquid media for 24 h, which is equivalent in time to the early stages of infection in vivo. Wild-type bacteria reached an OD₆₀₀ of 0.57, and the mutant achieved an OD₆₀₀ of 0.24. Higher cell densities in XAM1 liquid media could not be achieved even with longer cultivation periods (>24 h). Thus, to achieve the highest density of bacteria without compromising their structural integrity, a cultivation time of 24 h was chosen, which corresponded with the initial stage of infection. The difference in growth between the two culture media is due to the fact that the first is an energy-rich medium and the second a defined medium.

Several replicates were used for each condition to obtain the total pool of proteins used for this study.

Centrifugation and filtration of samples

Cultures were centrifuged in two 60 min steps at 4 °C. In the first step, the samples were centrifuged at 3,100 × g, the bacterial pellet was discarded and the supernatant containing secreted proteins was transferred to clean tubes. In the second step, the (first) supernatant was centrifuged at 11,000 × g and again the supernatant was recovered. For strains cultivated in NB liquid medium, the supernatant was recovered and then filtered using 0.2-µm (Millipore™ ) nitrocellulose membranes shortly after the second centrifugation.

For bacteria cultured in XAM1 medium, the supernatant could not be filtered due to the elevated production of xanthan gum, which caused changes in the membrane filtration profile. All samples were frozen using liquid nitrogen and lyophilized to complete dryness. Each dried sample corresponding to 500 mL of the supernatant was resuspended in 50 mL of double-distilled autoclaved water (ddH₂O). To precipitate xanthan gum produced by cultures grown in XAM1 medium, absolute ethyl alcohol (1:3) was added to the resuspended sample (in ddH₂O), and the sample was centrifuged at 3,000 × g for 5 min at 30 °C. These samples were then filtered through 0.2 µm (Millipore™) nitrocellulose membranes and again lyophilized and resuspended in 50 mL of ddH₂O.

Precipitation of proteins and SDS-PAGE electrophoresis

Proteins in the filtered supernatant were precipitated with trichloroacetic acid (TCA), with modifications (Hirose et al., 2000). Briefly, TCA was added to the resuspended sample to a final concentration of 10%, and each sample was vigorously mixed and stored at 4 °C for 16 h. Samples were centrifuged at 20,000 × g for 20 min at 4 °C. The supernatant was discarded and the precipitate washed three times with cold acetone to remove residual TCA. Precipitated proteins were then lyophilized to complete dryness and dissolved in 25 mM NH₄HCO₃ buffer, 500 mM urea, pH 8.0. All samples were quantified and normalized. The proteins were then hydrolysed in solution.

Trypsinization of samples and mass spectrometry

Proteins were reduced by adding 1 µg of DTT per 50 µg of protein and incubating for 1 h at 37 °C. Proteins were then alkylated by adding 5 µg of iodoacetamide per 50 µg of protein and incubating for 1 h at room temperature in the dark. Proteins were trypsinized by adding 20 µg of Promega Sequencing Grade Modified Trypsin (Madison, USA) (1:50) and incubating for 20 h at 37 °C. The hydrolysis reaction was terminated by adding 2 µL of formic acid. The samples were then loaded into a Waters^® nanoACQUITY UPLC^® capillary chromatography system (Waters, Milford, MA). The digested proteins were desalinized using a Waters Opti-Pak C18 trap column. The volume of the injected sample was 10 µL, and liquid chromatography was performed on a reverse-phase C18 column Ease 150 mm × 2.1 mm (Waters, Milford, MA). Proteins were eluted with a flow rate of 0.3 µL/min using a linear gradient ranging from 5% to 50% acetonitrile containing 0.1% formic acid.

The capillary liquid chromatography system (nLC) was coupled to a mass spectrometer with an electrospray ionization (ESI) source and quadrupole/time-of-flight in series analysers (Q-TOF Micro; Waters, Milford, MA) (UPLC-MS/MS), which permitted the direct analysis of eluted peptides in the acetonitrile gradient in Q-TOF. For ESI, a voltage of 3,000 V and a temperature of 80 °C was applied in the capillary. A MassLynx data system (Version 4.1, Waters) was used to control the instrument and acquire data, and the experiments were performed by scanning mass/charge (m/z) ratios between 200 and 2,000 using a first scan time of 1 s, which was applied during the entire chromatographic process. The averages of the mass spectra corresponding to each sign of the total ion current chromatogram (TIC) were calculated, allowing an accurate determination of the molecular mass. Exact mass values were obtained using a LockSpray™ source (Waters, Milford, MA). This reference mass was used to correct the mass of the analyte (sample) throughout the study. Phosphoric acid was used as the reference in this study, which has an m/z ratio of 588.8692.

The acquisition of data dependent on MS/MS were performed on precursors with charge states 2 or 3 over a mass range of m/z 50–2,000 and at an interval below 2 m/z. At most, 3 ions were selected for MS/MS analysis of a single MS. The masses of Na⁺ and K⁺ were automatically excluded. The collision induced dissociation (CID) based MS/MS spectra were obtained by using argon as the collision gas at a pressure of 13 PSI and a collision voltage that ranged from 18 to 45 V, depending on the mass of the precursor. The scanning ratio was 1 s.

Processing of the generated spectra and signal peptide prediction

All data were processed with ProteinLynx Global Server (version 2.0, Waters), which automatically corrected values of m/z spectrum of MS and MS/MS according to the mass of the Lockspray reference ion. Proteins were identified using the Xac database from NCBI (BioProject PRJNA297) and UniProt databases, and the MASCOT program (Version 2.2.1; Brazilian Synchrotron Light Laboratory). The values of monoisotopic masses of the MS/MS spectra (MS/MS Ion Search) were used, which considered cysteine carbamidomethylation as a fixed modification and oxidation of methionine as a variable modification. In the hydrolysis by trypsin, the possible loss of one cleavage site was considered and the tolerance of masses of peptides and fragments was set to ± 0.05 Da.

The amino acid sequences of identified proteins were submitted to analysis by PrediSi (Prediction of Signal peptides, www.predisi.de) (Hiller et al., 2004). Identified proteins that lacked signal peptides according to PrediSi analysis were submitted to SecretomeP 2.0 for further analysis (Bendtsen et al., 2005). Comparisons between proteins from different organisms were performed using the bioinformatics tool BLASTP.

Identification of candidate PIP-boxes in the regulatory regions of genes

Several genes regulated by HrpX possess a consensus nucleotide sequence, TTCGC-N₁₅-TTCGC, which has been termed the plant-inducible promoter box (PIP-box) (Da Silva et al., 2002). The presence of a PIP-box or PIP-box-like sequence suggests that the candidate gene is regulated by HrpX and encodes an effector protein of the T3SS pathway. PIP-box analysis was performed following methods described by Da Silva (Da Silva et al., 2002).

Candidate PIP-box sequences and their genomic locations were identified by searching the complete Xac genome for sequences that matched the following PIP-box sequence patterns: 1. Forward orientation (uncomplemented)—TTCGN-N₁₅-TTCGN or TTCGN-N₁₆-TTCGN or 2. Reverse orientation (complemented)—NCGAA-N₁₅-NCGAA or NCGAA-N₁₆-NCGAA. Each candidate PIP-box sequence was further analysed to determine whether it was part of a promoter, which should be between 10 and 1,000 nucleotide bases long and precede the start codon of a gene.

Xanthan gum production

For xanthan gum production, three (triplicate) 250-mL Erlenmeyer flasks containing 100 mL of Gum medium (25 g/L glucose, 3 g/L yeast extract, 2 g/L K₂HPO_4, 0,1 g/L MgSO_4.7H₂O, pH 7.0 with 4 M HCl) were inoculated with 2.5 mL of each bacterial strain grown in NB liquid medium (OD₆₀₀ of 0.3) and incubated at 29 °C for 96 h in a rotary shaker at 178 rpm (Shu & Yang, 1990).

After 96 h, cultures were centrifuged at 9,666 × g for 40 min. The bacterial pellets were stored at −20 °C and the supernatants were transferred to 500-mL beakers. Xanthan gum was recovered from the supernatants by isopropanol precipitation. KCl (4 g) was added to each beaker, followed by stirring at room temperature for 15 min. Two volumes of cold isopropyl alcohol were added and the xanthan gum from each beaker was removed and placed in plastic containers, which were previously weighed, with the aid of a glass rod and a sieve. After 48 h at 37 °C, the containers, plus sample, were weighed again and the amount of xanthan gum calculated. The mass of the bacterial pellet from each culture was also measured. For this, each pellet was resuspended with 1 mL MilliQ autoclaved water and transferred to a pre-weighed beaker and weighed again after 24 h at 70 °C. Three replicates were used for each condition and Statistical Analysis was performed (Unpaired Student t test).

Proteins secreted by both strains grown in XAM1 medium

Table 1 lists the 13 proteins secreted in common by both strains. Of these, four are hypothetical proteins (XAC2562, XAC0677, XAC0223 and XAC0232), three of which contain a signal peptide sequence (Table 1). Among the other nine proteins with annotated functions (Pcp, Cysteine protease, RpfN, Slp, OmpP6, Peh-1, PheA and Prolyl-peptidyl cis-trans isomerase—PPlase), six have a signal peptide sequence, and only the gene encoding for Peh-1 is downstream of a PIP-box sequence (Table 1). All of these membrane proteins were previously identified in the formation of outer membrane vesicles (OMVs) when X. campestris pv. campestris was grown in XVM2 medium (Sidhu et al., 2008), which is similar to XAM1 in that it also simulates conditions experienced by the microorganism during infections. In general, proteins secreted by both wild-type Xac and ΔhrpB4 in these conditions likely represent a core set of proteins required to survive and adapt to the plant interior. Importantly, the loss of T3SS function in the hrpB4 mutant did not affect the secretion of these proteins; therefore, it is likely that these proteins are secreted by the other secretory systems, such as Sec, Tat, T2SS, and T5SS. The presence of signal peptide sequences in at least 69% of these proteins reinforces their targeted secretion.

Proteins with annotated functions described bellow appear to have a fundamental role in either the virulence of the genus Xanthomonas or their adaptation to the plant. The Pcp protein, encoded by the gene XAC1466, is a peptidoglycan-associated outer membrane lipoprotein homologous to the SlyB protein in Xanthomonas campestris. This protein is directly regulated by a two-component system in most bacteria (PhoPQ), which is present in many bacterial pathogens of both animals and vegetables (Perez et al., 2009). This two-component system, which is encoded by the genes XAC4023 and XAC4022, is also expressed in Xac and has been shown to be active during the early stages of infection (Moreira et al., 2015). Proteins homologous to Pcp are also essential for maintaining the integrity of the cell envelope in Pseudomonas putida (Rodríguez-Herva, Ramos-González & Ramos, 1996), and more recently, were associated with OMV formation in P. aeruginosa (Wessel et al., 2013).

The Peh-1 protein, encoded by the gene XAC0661, is an endopolygalacturonase that facilitates plant cell wall degradation (Liu, Chatterjee & Chatterjee, 1994). Breakdown of the plant cell wall also releases sources of carbon to nourish the pathogen during infections. This protein is secreted by T2SS, and has been reported to be regulated by HrpG/X (Wang, Rong & He, 2008), which is consistent with the presence of the plant-inducible promoter (PIP-box) sequence located upstream of the gene. HrpG and HrpX are dual regulatory elements that regulate the expression of genes involved in the formation and secretion of T3SS proteins (Guo et al., 2011). In transcriptome studies, the hrpX gene was previously shown to be expressed in Xac when grown in XAM1 medium, and remains active even in the ΔhrpB4 strain (Laia et al., 2009), which is consistent with its secretion by both strains in our study. Mutants of Xanthomonas campestris in which the gene encoding the PghAxc protein (homologous to Peh-1 in Xac) was inactivated showed attenuated virulence when sprayed onto Arabidopsis hosts, demonstrating the importance of this protein in the early stages of infection (Wang, Rong & He, 2008).

A protein with cysteine protease functions, encoded by the XAC2853 gene, was also secreted by both strains grown in XAM1 medium. This protein is homologous to CysP2 in X. oryzae pv. oryzae, has a defective PIP-box sequence upstream of the gene in the Xac genome, is secreted by T2SS, and is regulated by HrpG/X (Furutani et al., 2004; Yamazaki, Hirata & Tsuyumu, 2008). Proteins with this function are considered important virulence factors because they hydrolyse peptides in the host cell. Inoculation of a citrus host in vivo with a mutant deficient in a cysteine protease, encoded by the gene XAC2853, resulted in less virulence (Soares-Costa et al., 2012). In studies involving Pseudomonas, the virulence of the pathogen was directly correlated with the secretion of the AvrRpt2 protein, which has cysteine protease functions, by T3SS (Axtell et al., 2003; Cui et al., 2013).

RpfN, which is characterized as a virulence factor regulator and encoded by the gene XAC2504, has an OprB domain (carbohydrate-selective porin). The OprB domain is a specific porin channel for glucose transport, but can also mediate the transport of other monosaccharides to the inner membrane (Wylie & Worobec, 1995). Pseudomonas aeruginosa mutants defective for this protein demonstrated a reduced ability to utilize various types of monosaccharides, indicating its importance in the import of different sugars into the cell (Nikaido, 2003). According to SecretomeP 2.0 analysis, this protein is not predicted to be secreted by the Sec pathway but rather by a non-classical route. The role of this protein in the pathogenicity of Xac is not well understood, but according to Dow and colleagues (2000), endopolygalacturonase levels were increased in Xanthomonas campestris mutants, in which this gene was inactivated. Moreover, Moreira and colleagues suggested that the absence of the rpfN gene and mutations in the PTS system associated with the internalization of fructose could explain the fastidious growth of X. fuscans B strain, which causes canker B in citrus plants (Moreira et al., 2010).

Slp, which is encoded by the gene XAC1113, is an outer membrane lipoprotein. Studies of homologues of this protein in the Xanthomonas genus are limited. However, in studies of the transcriptome of Xylella fastidiosa, which causes CVC (Citrus Variegated Chlorosis) in citrus plants, a homologue of the Xac slp gene was robustly expressed in growth conditions that induced the formation of biofilm (De Souza et al., 2004). Biofilms are fundamental to the induction of virulence and are produced on a large scale in bacteria of the genus Xanthomonas when grown in conditions that induce virulence (Dow et al., 2003; Rigano et al., 2007). Thus, the methodological difficulties of protein extraction from strains grown in XAM1 medium could be explained in part by the secretion of Slp proteins. In E. coli, changes to the composition of lipoproteins in the outer membrane directly affect surface adhesion, reinforcing the importance of these proteins to biofilm formation (Otto et al., 2001).

PheA, which is encoded by the gene XAC3467 and is secreted by both strains grown in XAM1 medium, is a chorismate mutase. PheA appears to be fundamentally important for plant–pathogen interactions because it catalyses the conversion of chorismate to prephenate, which is a precursor for the aromatic amino acids tyrosine (Tyr) and phenylalanine (Phe) (Kast et al., 2000). Chorismate is a substrate of the shikimate pathway, which regulates plant cell growth, development, structure, and pathogen defence. It can be converted into indole-acetic acid (IAA), salicylic acid (SA) and tryptophan (Bekal, Niblack & Lambert, 2003), culminating in the activation of plant defence systems (Weaver & Herrmann, 1997) (Fig. 3). Inactivation of the chorismate mutase gene in Ustilago maydis, the fungal agent that causes corn smut, drastically reduced symptoms associated with the disease (Djamei & Kahmann, 2012), demonstrating the importance of this protein to host–pathogen interactions. Finally, in bacteria of the genus Xanthomonas (Xanthomonas oryzae pv. Oryzae XKK.12), this protein has been reported to be directly involved in the process of virulence (Degrassi et al., 2010).

Peptidyl-prolyl cis-trans isomerase (PPIase) is a protein encoded by the gene XAC1585. Similar to Slp, Peh-1, and PheA, this protein was identified in the extracellular matrix of both strains when grown in induction medium, confirming data described by Yamazaki, Hirata & Tsuyumu (2008). However, a paralogous protein was also secreted only by the mutant ΔhrpB4 grown in NB medium (Table S1). The PPIases play a pivotal role in catalysing the proper folding of many prokaryotic and eukaryotic proteins that are implicated in a variety of biological functions ranging from cell cycle regulation to bacterial infection (Pissavin & Hugouvieux-Cotte-Pattat, 1997; Zang et al., 2007). In Pseudomonas syringae and Xanthomonas, proteins orthologous to PPIase were identified in the secretome (Kaffarnik et al., 2009). In Xylella fastidiosa, which causes CVC, this protein is part of the composition of mature biofilms (Silva et al., 2011), and is differentially expressed when Xylella is subjected to thermal stress (Da Silva Neto et al., 2007).

The last protein with an annotated function secreted by both strains only when grown in XAM1 medium is a proteinase (XAC0552). The gene encoding this protein has been identified only in species that cause citrus canker; its genomic organization has been maintained in these species. In addition to containing a peptidase_S8 domain in its carboxyl-terminal region, it contains a Pro_kuma_activ domain (Pro-kumamolisin - PF09286.6) in its amino-terminal region, whose cleavage results in the activation of the remaining peptide (Comellas-Bigler et al., 2004). Although this domain has not been examined in the context of bacterial pathogen proteins, in the phytopathogenic fungi Mycosphaerella graminicola, expansion of the number of peptidases containing this domain may be associated with enhanced pathogenesis of wheat hosts (Goodwin et al., 2011).

Although information about the functions of proteins encoded by the XAC2562, XAC0677, and XAC0232 genes is limited, domains associated with the function of lipoproteins were found in all of these proteins, which would justify their presence in the secretome.

Proteins secreted only by the wild-type strain grown in XAM1 medium

The data so far indicate that the production and secretion of important proteins associated with virulence and adaptation still occur in the ΔhrpB4 mutant. However, the impact of the mutation to the hrpB4 gene, which is associated with a loss of function of the T3SS system, is revealed by the identification of 14 unique proteins secreted only by wild-type Xac when cultured in XAM1 medium. Of these, nine had an annotated function (OprO, VirK, ribosomal proteins RplW, RplR and RpsE, HupB and cellulase Egl/Egl, EngXca), whereas the other five were hypothetical proteins (XAC0678, XAC3605, XAC1497, XAC3380 and XAC2682). Ten of the total amount of these proteins have a signal peptide sequence and only one hypothetical protein has a PIP-box sequence upstream of the start of the gene (Table 2). Although lacking annotated functions, all of these hypothetical proteins have a signal peptide sequence, suggesting that their secretion is critical for certain biological processes. Among the proteins with annotated functions, five are worth discussing further.

The OprO protein, encoded by the gene XAC3472, is an outer membrane porin specific for polyphosphates that can be regulated by the two-component regulatory systems, PhoPQ and PhoBR. In Pseudomonas aeruginosa, OprO is an anion-specific porin with a higher affinity for pyrophosphates than orthophosphates, and was expressed only under energy-poor growth conditions in phosphate during the stationary phase of bacterial growth (Siehnel, Egli & Hancock, 1992). Importantly, XAM1 medium lacks phosphate, similar to the mesophyll of citrus hosts. In recent work, Moreira and colleagues (2015). demonstrated that OprO, like other phosphate internalization systems, is expressed in Xac only during infection periods.

VirK, encoded by the gene XAC0435, is another protein secreted exclusively by the wild-type strain when cultured in XAM1 medium. This protein has a signal peptide for its secretion through the Sec pathway into the periplasmic space. In E. coli, VirK was characterized as a periplasmic protein required for the efficient secretion of plasmid-encoded toxins (Tapia-Pastrana et al., 2012). An orthologous gene found in Ralstonia solanacearum is regulated by HrpG. Microarray studies of the Xac transcriptome demonstrated that the expression of VirK was increased by more than two-fold when grown in XVM2 medium compared to when grown in non-inducing energy-rich culture medium (Astua-monge et al., 2005). The same study reported that a number of genes encoding secreted proteins were robustly expressed by XVM2 medium. Among the enzymes involved in cell wall degradation is cellulase (Egl), which is encoded by the genes XAC0029 and XAC0028. Both proteins were also secreted only by wild-type Xac in XAM1 medium; this was also true for the cellulase, EngXca. These proteins are predicted to be transported via the Sec pathway to the periplasm, and then secreted into the extracellular region by T2SS. Their secretion only by wild-type bacterium indicates that inactivation of T3SS in the mutant could suppress T2SS-secreted enzymes that degrade the cell wall of the host. Furthermore, EngXca has been specifically reported to be controlled by rpfF in Xanthomonas campestris. Poplawsky and colleagues (1998) and Siciliano and colleagues (2006) reported that Xac mutants, in which the rpfF and rpfC genes were inactivated, exhibited significantly decreased production of cell wall degrading enzymes and xanthan gum.

During the preparation of samples, the supernatant of the ΔhrpB4 strain cultured in XAM1 medium could be easily filtered through a 0.2 µm membrane, whereas the supernatant of the wild-type strain cultured in XAM1 medium could not be easily filtered. This is likely because the excess production of xanthan gum by wild-type bacteria obstructed the pores of the membrane. According to our studies, wild-type Xac is able to produce nearly 10 times more xanthan gum than the pathogenic deficient mutant ΔhrpB4 (mg gum/mg bacteria). Whereas wild-type Xac produces 46.21 ± 6.56A mg gum/mg bacteria, ΔhrpB4 produces only 5.21 ± 1.38B mg gum/mg bacteria (p value = 0.0004). These data underscore the importance of these genes to the formation of biofilm, as well as the importance of the hrpB4 gene in maintaining the activities of the T3SS and T2SS secretory systems.

The XAC0678 gene encodes a hypothetical protein but by sequence homology, it is predicted to encode a lipoprotein membrane protein. This hypothetical protein has a consensus PIP-box sequence upstream of the start of the gene, which indicates its probable regulation by HrpX. Furthermore, this protein is predicted to have a signal peptide for the Sec pathway, which could instruct its export into the periplasmic space.

Proteins secreted only by the mutant ΔhrpB4 grown in XAM1 medium

Ten proteins were secreted only by the mutant strain grown in XAM1 medium, 9 of which had annotated functions (Mlt, VacJ, OmpW, lei, GroES, YrbC, CycK, and PepN) and only one was predicted to be a hypothetical protein (XAC1761). Seven proteins had a signal peptide sequence, and a PIP-box sequence was not present upstream of any of these genes (Table 3).

Among these proteins, one worth noting is Mlt, which is a lytic murein transglycosylase. Xac has two copies of the mlt gene, XAC3225 and XACb0007, which are located on the chromosome and on the pXAC64 plasmid, respectively. The chromosomal copy of the gene is flanked by genes that encode the T3SS, XopE3 and XopAI effector proteins and the plasmid copy is flanked by the gene encoding XopE2 (Moreira et al., 2010). Mlt is highly similar to the hopAJ1 protein of Pseudomonas syringae, which was previously annotated as a T3SS helper protein (Oh et al., 2007). Mutations in the chromosomal copy reduced the ability of the mutant to cause canker (Laia et al., 2009).

The proteins encoded by the genes XAC3664 and XAC1479 are predicted to be outer membrane proteins. Although previously annotated as a hypothetical protein, the protein encoded by the gene, XAC1761, is predicted to be a lipoprotein based on its homology with sequences from other organisms.

Proteins secreted only by the wild-type strain grown in NB medium

Only 4 proteins were secreted by the wild-type strain grown in NB medium. Of these, 3 are directly associated with motility: PilY1, which is a Type IV pilus assembly protein, FliC (cover filament) and FliD (filament), which are structural proteins of the flagellum (Table 4). In addition, a hypothetical protein encoded by the gene XAC3314 was also secreted under these conditions. Of these four proteins, only PilY1 has a signal peptide sequence and a PIP-box sequence was not present upstream of any of these genes. PilY1 contains a DUF239 domain; these are found in a family of plant and bacterial proteins (PF03010). In addition, orthologous proteins were only found in Xac29, XacW, and Xcv, in Xoo MAFF and Xoo PXO, a protein similar to transposase. Mutant Pseudomonas strains, in which the pilY1 gene was inactivated, exhibited robust suppression of the swarming motility (Kuchma et al., 2010). In Xylella fastidiosa, a mutation in the pilY1 gene led to a reduction, but not a complete loss, of type IV pilus apparatus assembly and twitching motility (Li et al., 2007). In Xoo, pilY1, in combination with other genes involved in adhesion and biofilm formation, was upregulated six days after infection in the plant (Soto-Suárez et al., 2010).

FliC and FliD are protein subunits that form the structure of flagella. FliC, also known as flagellin, forms the flagellar filament itself, and is therefore abundantly synthesized by flagellated bacteria. FliC is directly associated with the innate immune response of plants, since it has the ability to bind to specific receptors, such as Toll-like receptor 5 (TLR5) (Hayashi et al., 2001). More recently, part of the N-terminal portion of this protein, corresponding to 22 amino acids (flg22) and conserved across a wide range of pathogens, was shown to activate plant defence mechanisms (Navarro et al., 2004). This 22-amino acid protein fragment is recognized by specific receptors such as by the kinase FLS2, which in turn activate the MAPK signalling cascade (mitogen-activated protein kinase), which is capable of regulating the expression of over 900 genes in the Arabidopsis thaliana genome (Asai et al., 2002).

Protein secreted only by the wild-type strain in both growth media

The only protein detected exclusively in the secretome of the wild-type strain when grown in either growth media was the EF-Tu protein encoded by the gene XAC0957 (Table S1).

EF-Tu is also characterized as a PAMP, like FliC (Nürnberger & Kemmerling, 2006). Although EF-Tu has a fundamental role in protein synthesis, a diversity of other functions have been attributed to this protein. In Burkholderia pseudomallei, EF-Tu was characterized as a membrane-associated protein that is secreted by the formation of outer membrane vesicles (OMV), and is therefore an important target in the development of vaccines against this pathogen (Nieves et al., 2010). In E. coli, EF-Tu acts as a chaperone molecule by binding to denatured proteins after stress conditions (Caldas, El Yaagoubi & Richarme, 1998). In Agrobacterium tumefaciens, EF-Tu can act as an inducer in Arabidopsis, whose specific receptor is the EFR protein (Zipfel et al., 2006). EF-Tu has also been reported to be critical for biofilm formation in Xylella (Silva et al., 2011) and, more recently, in Xac (Zimaro et al., 2013). Considering all of these functions attributed to this protein and the inability of the ΔhrpB4 mutant to secrete this protein under all conditions tested, we hypothesize that the secretion of this protein may be associated with T3SS in Xanthomonas. Besides its participation in the production of biofilms, EF-Tu may have another key role in the virulence process, and may represent a potential moonlighting protein (Jeffery, 1999; Henderson & Martin, 2011).

All other secreted proteins detected in our study

In addition to the proteins described above, 13 other secreted proteins were detected in this study (Table S1). Because these proteins were detected in specific but restricted conditions, we could not exclusively associate them with a specific growth media or specific bacterial strain. The majority (8) of these have enzymatic functions (amylase, proteases, and isomerase). Two membrane-associated proteins (OMPs), and two hypothetical proteins, XAC0753 and XAC0868, were also identified. The latter two proteins have domains with peptidase activity, although XAC0868 has a more representative ricin B lectin domain. Interestingly, the protein encoded by XAC0868 was the only protein detected in all four conditions tested and exists only in Xanthomonas strains that infect citrus plants. A model highlighting proteins related to Xac adaptation was devised (Fig. 4).

Pathogenicity of Xac compared to other plant pathogens

To better understand the importance of secreted proteins in the virulence of plant-specific pathogens of the Xanthomonadaceae family, a comparative analysis of proteins identified in this study with proteins identified in six other proteomic studies was performed (Table 5). Despite the diversity of model organisms studied and methodologies used, orthologous proteins were identified in these studies.

Yamazaki, Hirata & Tsuyumu (2008) studied the role of the HrpG gene in the regulation of the synthesis of Xac proteins secreted by T2SS in energy-rich culture medium. In these studies, a total of 11 proteins were identified, six of which were extracellular enzymes. Zimaro and colleagues compared the proteome of planktonic Xac cells and Xac biofilm cells, and identified 53 differentially expressed proteins, of which 31 were upregulated in biofilms. Silva and colleagues (2011) identified 456 proteins potentially related to biofilm formation in Xylella fastidiosa. Wang and colleagues analysed the Xanthomonas oryzae pv. oryzae secretome grown in culture medium and in plants, and identified 109 unique proteins. Proteins involved in pathogenicity and the induction of host defence were identified only in the secretome of bacteria in contact with the plant (Yamazaki, Hirata & Tsuyumu, 2008). Using the same model study, Gonzalez and colleagues identified 324 different proteins secreted by Xoo in contact with the xylem of the host plant. Of these, 64 were suggested to be associated with the process of virulence (González et al., 2012). Sidhu and colleagues (2008) identified 31 proteins associated with OMV, half of which were associated with the induction of virulence in Xanthomonas campestris pv. campestris.

This comparative analysis revealed the correlation of a number of proteins with the life cycle or the infection model of the associated pathogen. For instance, endopolygalacturonase and proteinases, which seem to be crucial to the virulence process, are induced by mesophilic strains. Proteins associated with motility and chemotaxis have been identified in virtually every proteomics study included in our analysis, which is consistent with the importance of these proteins in the formation of cellular aggregates, biofilms or even for inducing a response in the host plant (Malamud et al., 2011; Dunger et al., 2014). Mlt was detected only in this study, indicating its importance to the induction of virulence in Xac. Interestingly, EF-Tu was independently identified in five other studies, including in the Xf strain, whereas GroEL was detected in four other studies, and flagellin (FliC), VirK and peptidyl-prolyl cis-trans isomerase (PPI) were identified in three other studies. It is important to note that FliC is absent in the Xylella genome since it does not have flagellum. The possibility that GroEL, FliC, Ef-Tu and PPI can act as moonlighting proteins in bacteria of the family Xanthomonadaceae is highlighted (Henderson & Martin, 2011).