Diffusible signal factor (DSF)-mediated quorum sensing modulates expression of diverse traits in Xanthomonas citri and responses of citrus plants to promote disease

Background The gram-negative Xanthomonas genus contains a large group of economically important plant pathogens, which cause severe diseases on many crops worldwide. The diffusible signal factor (DSF) - mediated quorum sensing (QS) system coordinates expression of virulence factors in plant pathogenic Xanthomonas spp. However, the regulatory effects of this system during the Xanthomonas- plant interactions remain unclear from both the pathogen and host aspects. Results In this study, we investigated the in planta DSF- mediated QS regulon of X. citri subsp. citri (Xac), the causal agent of citrus canker. We also characterized the transcriptional responses of citrus plants to DSF-mediated Xac infection via comparing the gene expression patterns of citrus trigged by wild type Xac strain 306 with those trigged by its DSF- deficient (∆rpfF) mutant using the dual RNA-seq approach. Comparative global transcript profiles of Xac strain 306 and the ∆rpfF mutant during host infection revealed that DSF- mediated QS specifically modulates bacterial adaptation, nutrition uptake and metabolisms, stress tolerance, virulence, and signal transduction to favor host infection. The transcriptional responses of citrus to DSF-mediated Xac infection are characterized by downregulation of photosynthesis genes and plant defense related genes, suggesting photosynthetically inactive reactions and repression of defense responses. Alterations of phytohormone metabolism and signaling pathways were also triggered by DSF-mediated Xac infection to benefit the pathogen. Conclusions Collectively, our findings provide new insight into the DSF- mediated QS regulation during plant-pathogen interactions and advance the understanding of traits used by Xanthomonas to promote infection on host plants. Electronic supplementary material The online version of this article (10.1186/s12864-018-5384-4) contains supplementary material, which is available to authorized users.


Background
The genus Xanthomonas comprises a large group of gram-negative plant pathogenic bacteria that have considerable agricultural impact worldwide, and therefore, is an important model genus for studying the host-pathogen interactions [1,2]. Successful infection and bacterial multiplication of Xanthomonas spp. in host tissues require coordinated expression of a combination of virulence factors. Key virulence factors of Xanthomonas spp. include, among others, the type III secretion system (T3SS) and its effectors [3,4], bacterial polysaccharides such as the xanthan extracellular polysaccharides (EPS) and lipopolysaccharides (LPS) [5], and cell wall degrading enzymes [1]. Expression of these virulence factors is regulated by different extracellular stimuli via multiple coordinated regulatory systems, including cell-to-cell communication (quorumsensing, QS) pathways, two-component systems and various transcriptional regulators [1].
The QS regulatory systems of Xanthomonas are mediated by molecules belonging to the diffusible signal factor (DSF) family [2,6,7]. The DSF-mediated QS has been studied most extensively in the crucifer pathogen X. campestris pv. campestris (Xcc). The synthesis and perception of the DSF signal, which was identified as cis-11-methyl-2-dodecenoic acid [8], require the rpf gene cluster (for regulation of pathogenicity factors), including rpfB, rpfF and rpfGHC [9,10]. RpfB was initially thought to be involved in DSF biosynthesis, but it was later identified as a fatty Acyl-CoA ligase involved in the turnover of the DSF family of signals in Xanthomonas [11]. The RpfF protein, functioning as a putative enoyl-CoA hydratase, is responsible for the synthesis of DSF. RpfC and RpfG consist of a two-component system involved in DSF perception and signal transduction. RpfC is a hybrid sensor kinase and RpfG is a response regulator with a CheY-like receiver (REC) domain and an HD-GYP domain, capable of degrading the second messenger cyclic di-GMP [6,10,12,13]. DSF can bind directly to the N-terminal, 22 amino acid-length sensor region of RpfC and activate RpfC autokinase activity to regulate QS and virulence in Xcc [14]. RpfH is a putative membrane protein with no known role in DSF signaling [10].
The contribution of DSF/Rpf regulatory system to virulence has been demonstrated in many members of Xanthomonas. For example, DSF signaling in Xcc influences the synthesis of a range of virulence factors including extracellular enzymes such as endoglucanase, protease, and endomannanase, and the xanthan EPS, as well as alterations in biofilm formation [6,10,15]. Specifically, the RpfS-dependent second DSF signaling pathway controls expression of genes involved in type IV secretion and chemotaxis and therefore affects bacterial motility, suggesting a role in the epiphytic phase of the Xcc disease cycle [16]. Similarly, the DSF-mediated QS has been shown involved in early attachment and in planta growth of Xac in the citrus host during the citrus canker disease cycle [17]. Recent report indicates that the DSF family in Xcc elicited plant innate immunity and this effect was suppressed through the secretion of the xanthan exopolysaccharide [18]. DSF also confers a fitness advantage to Xcc during interspecies competition [19].
Transcriptome profile, functional genomics, and proteomic analyses have significantly advanced the understanding of the DSF/Rpf regulatory network and its role in pathogenesis of Xanthomonas. Earlier studies have revealed that the RpfC/RpfG two-component system coordinately regulates the expression of various genes related to virulence via the cyclic di-GMP signaling that activates the transcriptional activators Clp and Zur in Xcc [6,12,13,20]. These include the genes encoding extracellular enzymes, components of type II secretion system (T2SS), components of type III secretion system (T3SS), and the genes involved in EPS production. Comparative transcriptome studies using whole-genome microarray showed that the DSF/Rpf -mediated QS regulation in the citrus canker pathogen X. citri subsp. citri (Xac) is growth phase-dependent, and more genes in the exponential phase are differentially regulated by the RpfC/RpfG system compared with in the stationary phase [17]. The RpfC/RpfG system-regulated genes include diverse genes involved in chemotaxis and motility, flagellar biosynthesis, production of extracellular enzymes and adhesins, stress tolerance, regulation, transport, and detoxification [17]. There are also some unique genes controlled by RpfF, RpfC or RpfG alone, indicating the complexity of the QS pathway and the involvement of additional DSF signal perception and transduction mechanisms in Xac [17]. Interestingly, recent studies suggested additional signaling outputs from RpfC and an interaction of RpfG with a second unknown sensor [16,21]. The authors found that DSF and RpfC also regulate expression of a number of genes encoding transcriptional regulator, hydrolase, protease and hypothetical proteins independently of RpfG, and RpfG regulates expression of genes involved in chemotaxis, signal transduction and protein export, independently of RpfF or RpfC [16,21]. These studies also revealed that RpfC can recognize other unidentified environmental signals (in addition to DSF) [21] and the DSF signal can be recognized by a second sensor RpfS, a PAS domain-containing histidine kinase that regulates genes involved in type IV secretion and chemotaxis in a pathway independent of RpfC and RpfG [16]. Our knowledge of the protein(s)/regulator(s) acting downstream of RpfS in DSF signal transduction cascades remains limited. In addition, the DSF/Rpf system controls three non-coding RNA (ncRNAs) that contribute to virulence in Xcc [21].
Comparative proteomic analysis revealed diverse regulatory effects of DSF/Rpf in Xcc on proteins involved in regulation, biosynthesis and intermediary metabolism, stress tolerance, and motility [22]. Similarly, mutation of the rpfF gene has a substantial impact on the proteome of X. oryzae pv. oryzicola, affecting proteins involved in a range of functions including nitrogen transfer, protein folding, resistance to oxidative stress and flagellar synthesis [23]. Interestingly, for many of the proteins regulated by the DSF/Rpf system in Xcc, the alteration in abundance was not associated with alteration in transcript level, suggesting that both posttranscriptional regulation and post-translational turnover may occur [22].
Despite the extensive transcriptome analyses of the DSF/Rpf regulatory system in Xanthomonas spp. as stated above, most of which were performed using the bacterium grown under culture media conditions, and knowledge on regulatory effects of the DSF/Rpf system of Xanthomonas spp. during the interaction with host plants is still lacking. The actions of the elements involved in DSF signaling and the role of DSF signal transduction during plant infection remain to be determined from both the pathogen and host aspects. In the present study, we investigated the DSF/Rpf QS regulation in planta during Xac infection of citrus host. Metatranscriptome analysis of the compatible interaction between Xac and citrus was conducted using RNA-Seq to compare the global transcriptomes of wild-type and isogenic rpfF mutant (ΔrpfF) strains of Xac, as well as the citrus transcriptional patterns in response to their infection. This work provides a comprehensive picture of the genes and traits regulated by the DSF/Rpf QS in Xac in planta and host responses to the DSF-mediated infection.

Plant inoculations and sampling of infected leaves
Plant inoculation was performed as described in our previous work [5]. Briefly, young (about 12-week-old) Duncan grapefruit (Citrus paradise Macfadyen) plants were grown in potting medium/soil in a greenhouse at the Citrus Research and Education Center, Lake Alfred, FL, USA, and maintained at approximately 25-30°C and a 55% relative humidity until the primary leaves were fully expanded but not fully matured. The bacterial inoculum cells were grown as described above. When the cells reached late-log phase (OD 600 = 1.0; 5 × 10 10 cfu/mL), they were collected by centrifugation at 4000 rpm for 15 min. The cell pellets were resuspended in sterilized 0.85% NaCl, washed, and resuspended in sterilized 0.85% NaCl to a final density of 5 × 10 6 cfu/mL. To establish in planta populations, bacteria were introduced by infiltration into leaves using a needleless syringe. Infiltrated plants were maintained in the same greenhouse for canker symptom development. All plant inoculations included at least three leaves at a similar developmental stage from each plant, and ten replicate plants were inoculated for each strain. Time-course bacterial growth in planta was tested as described previously [5]. All the tests were repeated three times. Based on the progression of development of canker symptoms, the infected leaves were sampled at 5 days post inoculation (DPI) for RNA extraction. The inoculated leaf area was collected using a cork borer (leaf area, 1 cm 2 ) and 10 leaf samples from each biological replicate (three replicates for each treatment) were pooled and immediately frozen in liquid nitrogen, and kept in − 80°C until process for RNA isolation.

RNA extraction, library construction and Illumina RNA-seq
Total RNA was extracted from leaf samples using RNeasy Plant Mini Kit (Qiagen, Valencia, CA) and contaminated DNA was removed by treatment with RNase-Free DNase Set (Qiagen, Valencia, CA). The quality and quantity of RNA samples were assessed using NanoDrop ND-1000 spectrophotometer (Nano-Drop Technologies, Wilmington, DE), Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA) and agarose gel electrophoresis. The total RNA was treated with DNase I (New England Biolabs, Ipswich, MA) prior to library construction. The rRNA of plant and bacteria was depleted using Ribo-Zero™ rRNA Removal Kits (Plant Leaf ) and Ribo-Zero™ rRNA Removal Kits (Bacteria) respectively, according to the manufacturer's instructions (Epicenter Technologies, Madison, WI). Poly (A) + mRNA was purified using Agencourt RNAClean XP Kit (Beckman Coulter Life Sciences, Indianapolis, IN) and fragmented into short pieces. Using these short fragments as templates, first strand cDNA synthesis was conducted using random hexamer-primers and SuperScript® II Reverse Transcriptase (Invitrogen, Waltham, MA), and the second-strand cDNA was synthesized using RNase H (Invitrogen, Waltham, MA) and DNA polymerase I (New England Biolabs, Ipswich, MA). After purification, end repair, and ploy (A) tails add, the cDNA fragments were ligated to sequencing adapters. Then fragments of an appropriate size were purified and amplified by PCR to produce the final library. Finally, the cDNA libraries were loaded onto the flow cell channels of an Illumina HiSeqTM 2000 platform for paired-end 90 bp × 2 sequencing at the Beijing Genomics Institute (BGI), Hongkong, China. Clean reads were obtained after removing reads containing adaptor sequences. The RNA reads have been deposited at NCBI under the bioproject No. PRJNA421992 with the SRA accession no. SRP126698.

Reads mapping and differential expression analysis
The clean reads were firstly aligned to the Xac strain 306 genome (https://www.ncbi.nlm.nih.gov/nuccore/ AE008923.1/) [24] using bowtie2 [25] with default parameters. The in planta differential expressed genes between wild type Xac 306 and ΔrpfF mutant strains were identified using DESeq2 R package [26] with the following cutoffs: |fold change| ≥ 2 and agjust-P ≤ 0.05. After aligned to Xac strain 306 genome, the remaining reads from each sample were analyzed mainly following the tuxedo pipeline [27]. Briefly, the reads were aligned to the sweet orange genome [28] using Tophat (v2.0.13) [29], and the generated alignments were fed to Cufflinks (v2.2.1) for transcript assembly [30]. The assemblies were combined with the sweet orange annotations using the cuffmerge algorithm and then fed to the cuffdiff2 for differentially expressed gene calling. Only the genes with |fold change| ≥ 2, q-value ≤0.05 and FPKM≥1 were considered as significantly differentially expressed genes (DEGs) between wild type strain infected and Δrpf mutant strain infected plants. The MapMan gene functional categories were assigned to the DEGs using Mercator [30,31] and the differentially regulated bins were identified by using MapMan [32].

Functional annotation and classification
For citrus DEGs, the corresponding reference ID were obtained by blasting them to CitrusPLEX in plant expression database (PLEXdb, http://www.plexdb.org/ plex.php?database=Citrus) [33]. Gene Ontology (GO) enrichment analysis of functional significance was applied to map all DEGs to terms in the agriGO (http:// bioinfo.cau.edu.cn/agriGO/) database [34], looking for significantly enriched GO terms in DEGs. For bacterial DEGs, Clusters of Orthologous Groups (COG, https:// www.ncbi.nlm.nih.gov/COG/) enrichment analysis was performed by comparing the prevalence of DEGs assigned to a specific COG category to the prevalence of genes in the whole genome assigned to that COG category with a Fisher's exact test.

Validation of RNA-seq results by qRT-PCR
To verify the RNA-seq result, qRT-PCR assays were conducted using the same set of RNA samples as for RNA-seq analysis. The aliquoted RNA sample (1 μg) used for RNA-seq was reverse transcribed using a Quan-tiTect Reverse Transcription kit with random hexamer primers (Qiagen, Valencia, CA) for two-step qRT-PCR. The gene specific primers (Additional file 1: Table S1) were designed to generate amplicons of 70 to 150 bp based on the DEGs sequences of citrus plant and Xac strain 306. qRT-PCR was conducted using QuantiTect SYBR Green PCR Kit (Qiagen, Valencia, CA) and the 7500 fast real-time PCR system (Applied Biosystems, Foster City, CA). Melting curve analysis of the PCR products was performed at the end of each PCR cycle to confirm the amplicon specificity. The housekeeping gene CtGAPDH [35] and gyrA [5] was used as plant and bacterial endogenous control, respectively. All reactions were repeated with three independent biological replicates and two technical replicates. The relative fold change in target gene expression was calculated by using the formula 2 -△△CT [36].

Statistical analysis
Quantitative data were expressed as mean ± S.E.M. Statistical differences were evaluated through t-test and the level of statistically significant difference was set at P<0.05. All statistics were conducted using SAS 9.1.3.

Canker progression and symptoms in inoculated citrus plants
Duncan grapefruit (Citrus paradisi Macfadyen) seedlings were inoculated with Xac wild type strain 306 and its DSF deficient (ΔrpfF) mutant strain for the development of typical symptoms of citrus canker. The first visible symptom, a water soaking area of the inoculated leaf, was observed at 5 days post inoculation (DPI). Within 14 DPI, typical symptoms of the canker disease were recorded ( Fig. 1a): inoculated areas were characterized with water soaking, and then exhibited hyperplasia and hypertrophic, necrotic, erumpent lesions, as evidenced by the raised pustules. The ΔrpfF mutant produced weaker water soaking phenotypes compared to wild type strain 306 at 5 DPI under the tested conditions; and this becomes more evident at 7DPI and until 9 DPI (Fig.1a). However, the bacterial populations of the ΔrpfF mutant in planta were not significantly lower than the wild type strain (Fig. 1b).

Sequencing the early citrus canker transcriptome
As the differences in the development of canker symptoms only were recorded in early stages of disease development (formation of water-soaking phenotypes) between the wild type Xac and ΔrpfF mutant (Fig. 1a), we speculated that the DSF/Rpf QS play certain role(s) in early stages of the Xac-citrus compatible interaction. Therefore, the early canker transcriptome during the formation of water-soaking phenotypes was profiled at 5 DPI.A total of 227 million and 278 million paired-end reads for wild type Xac strain 306 infected and the ΔrpfF mutant infected plants were produced respectively (Additional file 2: Table S2). All reads were aligned against the Xac strain 306 genome [24]. For each RNA-seq library, 2.4-5.5% of the reads mapped to the Xac 306 reference. Then the remaining Xac strain 306-unaligned reads were mapped against the sweet orange (Citrus sinensis) genome [28], for the analysis of the citrus host transcriptome. A significant fraction of the Xac306-unaligned reads (63-68%) from both wild type Xac infected and ΔrpfF mutant infected libraries mapped to the sweet orange reference (Additional file 2: Table S2). Finally, of the 4374 Xac genes, 202 (4.5%) were determined as significantly differentially expressed genes (DEGs) [a minimum absolute value of a log2-fold change greater than 1 (equivalent to two-fold)] between wild type Xac 306 and ΔrpfF mutant strains in the conditions analyzed (Additional file 3: Table S3  To validate the gene expression values obtained by RNA-seq, the expression of 40 Xac genes and 33 citrus genes (Additional file 1: Table S1) in the RNA samples used in RNA-seq analysis were analyzed by qRT-PCR assays. A strong correlation (R 2 = 0.9141 for Xac gene expression; R 2 = 0.9011 for citrus gene expression) were observed between the data produced by the two approaches ( Fig. 2a-b), demonstrating the reliability of the results obtained.
Functionally categorizing of Xac genes regulated by the DSF/Rpf-mediated QS system during early stages of host infection The 202 DEGs of Xac were subject to functionally categorizing with enrichment analyses of clusters of orthologous groups (COGs). The results showed that overrepresented COGs terms were mostly related to 'Carbohydrate transport and metabolism' (39 members, 16.3%), ' Amino acid transport and metabolism' (29 members, 14.4%), 'Inorganic ion transport and metabolism' (21 members, 10.4%), and 'Cell wall/membrane/envelope biogenesis' (19 members, 9.41%) (Fig. 3). Other enriched terms included 'Lipid transport and metabolism' , 'Energy production and conversion' , 'Post-translational modification, protein turn over, and chaperones' , 'signal transduction mechanisms' , 'Transcription' , and 'General function prediction only'. In addition, the genes annotated as hypothetical proteins were assigned to the 'Function unknown' group.

DSF/Rpf-mediated QS regulates stress tolerance of Xac during early stages of host infection
A total 12 genes encoding enzymes involved in detoxification and stress tolerance of Xac at early stages of host infection were differentially regulated by DSF/Rpf-mediated QS (Table 1). Of these, the genes coding for a putative arabinose efflux permease belonging to the Major Facilitator Superfamily (MFS) transporter for sugar/drug (araJ /XAC1363), for a drug resistance translocase (yieO /XAC2494), for an endoproteinase (argC/XAC2992), and for trehalose biosynthesis (XAC0425 and XAC0429) were upregulated by ≥2-fold on average. Bacterial endoproteinases are able to degrade host defense proteins [37,38], and trehalose protects bacterial cells from osmotic and oxidative stresses [39,40]. The katE gene (XAC1211) encoding a catalase important for hydrogen peroxide torelance in Xac [41], was also upregulated by 2-fold.

DSF/Rpf -mediated QS is implicated in the regulation of nutrition utilization of Xac during early stages of host infection
A significant portion of the Xac in planta transcriptome regulated by DSF/Rpf -mediated QS is dedicated to nutrition utilization (Fig. 3). Of the 39 genes involved in carbohydrate uptake and metabolism, 14 were positively regulated by DSF/ Rpf -mediated QS, while 25 were negatively regulated ( Table 2). The carbohydrate genes upregulated by DSF/Rpfmediated QS included those encoding cellulose endoglucanase (egl/XAC0029 and engXCA/XAC0612), glycosyl transferase Fig. 2 Correlation analysis of expression levels of selected bacterial genes and citrus genes determined by RNA-seq and RT-qPCR. a Comparison of RNA-seq and qRT-PCR data for differentially expressed genes (DEGs) in Xanthomonas citri subsp. citri. Fold changes were calculated for 40 bacterial genes and a high correlation (R 2 > 0.90) was observed between the results obtained using the two techniques. b Comparison of RNAseq and qRT-PCR data for DEGs in citrus. Fold changes were calculated for 33 citrus genes and a high correlation (R 2 > 0.90) was observed between the results obtained using the two techniques (gtrB/XAC1038/XAC2125, XAC3533, and ugt/XAC3921), glycosyl hydrolase (XAC3073), glycogen synthase (glgA /XAC0425 and glgY/XAC0429), glucose dehydrogenase (gcd/XAC1633/XAC3212), glucokinase (glk/XAC3120), and transporter (araJ/XAC1363 and yieO/XAC2494). In contrast, the expressions of fruBK and fruA encoding components of a fructose-specific phosphoenolpyruvate (PEP): carbohydrate phosphotransferase system (PTS) were downregulated.
Eleven differentially expressed genes were related to inorganic ion transport and metabolism in Xac during host infection (Table 2). Remarkably, the two genes (phoX/ XAC1578 and oprO/XAC1579) encoding phosphate transporter proteins were upregulated by an average of 2.6-fold by DSF/Rpf QS during infection. The genes for siderophore biosynthesis (entF/XAC3922) and for iron storage protein in the bacterioferritin family (bfr/ XAC1149) [42] were upregulated two-fold or more (Table 3). Six genes encoding TonB-dependent outer-membrane    receptors involved in siderophore-mediated ferric iron uptake by Xac [42,43], including fecA/XAC0690, btuB/ XAC1310, and fyuA/XAC3489, were downregulated two-fold on average. In addition, the two genes coding for ferric enterobactin receptors involved in siderophore uptake (bfeA/ XAC3168 and XAC 3169) were also downregulated two-fold on average (Table 3).
Genes for signal transducers and/or transcriptional regulators regulated by DSF/Rpf-mediated QS in Xac during early stages of host infection The expression of 12 genes coding for signal transducers and/or transcriptional regulators in Xac were differentially regulated by DSF/Rpf-mediated QS (Table 4). Of these, two genes were upregulated and 10 genes were downregulated. The two upregulated genes were XAC1328 and XAC3927, encoding a putative CheY-like superfamily protein and serine/threonine protein kinase respectively, both are of signal transducer activity.
Among those genes downregulated were the two genes ntrB (XAC0207) and ntrC (XAC0208) encoding the NtrB/C two-component system, which interacts with the RpfC/G system responding to DSF signal to regulate sig-ma54-dependent promoters in vitro [44]. In addition, the two-component sensor genes tctE (XAC3482) and XAC3720, the transcriptional regulator genes acoR (XAC0654), tetR (XAC2014), iscR (XAC 2934), and cebR (XAC3487), the AbrB ambiactive repressor and activator (XAC1883), and the Trp operon repressor gene (trpR/ XAC1827) were downregulated by the DSF/Rpf -mediated QS. The homologues of these signal transduction and transcription factors constitute regulators of virulence and adaptation factors in many bacteria, including the human bacterial pathogens enterotoxigenic E. coli [45] and P. aeruginosa [46], and the model organism Bacillus subtilis strain 168 [47]. For example, the IscR transcriptional repressor in E. coli negatively controls the type I fimbriae colonization factor synthesis and biofilm formation in response to both iron limitation and oxidative stress [45]. The trp repressor negatively regulates expression of genes involved in tryptophan biosynthesis, transport, and metabolism in response to intracellular levels of tryptophan, but also regulates transcription initiation in several other operons related to tryptophan metabolism that are important for expression of virulence factors in E. coli and P.  aeruginosa [46]. Thus, they might function as regulators of virulence and adaptation factors in Xac by modulating biofilm formation and adhesion factor production, which are crucial for attachment and colonization of the tissues and for consequent invasion [17].
Putative function of the hypothetical protein encoding genes within the DSF/Rpf-mediated QS regulon in planta BLASTx analysis showed that 39 of the 44 genes within the DSF/Rpf QS regulon encoding hypothetical proteins had significant similarities only to sequences in bacteria within the Xanthomonas genus. Based on sequence similarity and conserved domain detected, we defined putative functions for 24 of the 44 genes, which are potentially involved in bacterial adaptation and pathogenesis ( Table 5). Some of these genes encode proteins with recognized roles in bacterial pathogenesis, such as members of the cell surface adhesion protein families (XAC3546) and chemotaxis protein families (XAC3753 and XAC3754). Interestingly, the genes encoding stress-induced protein (XAC2156) and Ferritin-like di-iron-carboxylate protein (XAC2155) were upregulated by DSF/Rpf -mediated QS and possibly involved in the  adaptation of Xac to the host environment. The genes encoding putative GH18_chitinase-like glycosyl hydrolase (XAC3073) and GT2 family glycosyltransferase (XAC3533) were also upregulated, involved in carbohydrate transport and metabolism. In contrast, the gene XAC3085 encoding a putative T3SS effector protein was downregulated, with an unknown function in Xac-citrus interaction.

Comparison of the DSF/Rpf-mediated QS regulons in planta and in vitro
Our previous work identified 180 genes constituting the DSF/RpfF regulon of Xac grown in culture medium in the exponential and/or stationary growth phase [17]. Among those, a set of 31 genes were overlapping with the in planta DSF/Rpf regulon, 26 of which showed similar trends in alteration of expression between the two environmental conditions (Additional file 5: Table S5). Specifically, a subset of 20 genes were identified in the DSF/RpfF regulon of Xac in the exponential growth phase, 25 genes were identified in the DSF/RpfF regulon of Xac in the stationary growth phase, and 14 genes were identified in both regulons. These genes were primarily involved in energy metabolism (carbohydrate transport and metabolism), protein fate and protein synthesis (amino acid transport and metabolism or post-translational modification), and signal transduction or transcriptional regulation, and some encode hypothetical proteins with unknown functions.

Overview of citrus transcriptional responses to DSF/Rpfmediated Xac infection
Global analyses of the citrus transcripts in response to DSF/Rpf-mediated Xac infection revealed that the protein families related to stress responses, signaling pathways, hormone metabolism, and cell wall modification were over-represented according to the gene ontology (GO) analysis (Fig. 4). Individual gene responses in metabolic pathways were visualized using the MapMan tool (Fig. 5). Remarkable downregulation was observed for many genes related to photosynthesis, secondary metabolism, and plant defense response.

DSF/Rpf -mediated Xac infection represses photosynthesis in citrus
The expression levels of nine genes involved in photosynthesis decreased significantly in wild type Xac strain 306 infected leaf tissues, compared with the ΔrpfF mutant infected leaf tissues (Table 6). Transcripts for photosystem II oxygen-evolving enhancer protein PsbO (Cs7g03508) and photosystem II 22 kDa protein PsbS (Cs3g19650) were less abundant in wild type Xac infected leaves. Three transcripts encoding subunits of photosystem I also decreased in wild type Xac infected leaves, including photosystem I reaction center subunit II (PsaD), VI-2 (PsaH), and O subunit (PsaO). In addition, the genes for photosynthetic electron transport protein plastocyanin (PetE) and for an ATP synthase subunit (the F-type H + −transporting ATPase subunit gamma, Cs4g10260) were downregulated in wild type Xac infected leaves (Table 6). These results are in agreement with the notion that Xac is biotrophic during early stages of host infection [48,49] and that biotrophic pathogen infection generally represses photosynthesis in host plants [50].

Alterations of hormone metabolisms in citrus responding to DSF/Rpf-mediated Xac infection
Significant transcriptional changes in response to DSF/ Rpf-mediated Xac infection were observed for a group of genes related to plant hormone biosynthesis, transportation, metabolism, and associated signal transduction (Table 7). Transcripts for auxin biosynthesis-related enzymes and auxin-responsive proteins, including indole-3-acetate beta-D-glucosyltransferase (IAGLU), UDP-glucosyltransferase (UGT74E2), and SAUR (small auxin-up RNA) -like auxin-responsive protein, were more abundant, while genes for the PIN or PIN-LIKES class of auxin transporters were downregulated in wild type Xac infected leaves. The gene Cs2g03270 encoding a 9-cis-epoxycarotenoid dioxygenase, a key enzyme for abscisic acid (ABA) biosynthesis [51], was downregulated in wild type Xac infected leaves, and gens for ABA-responsive (ABR) proteins were upregulated in wild type Xac infected leaves. Three genes involved in cytokinin biosynthesis (cytokinin synthase, isopentenyltransferase (IPT), and UDP-Glycosyltransferase superfamily protein) were downregulated in wild type Xac infected leaves, while two genes involved in cytokinin metabolic process (UDP-glucosyl transferase 85A5 (UGT85A5) and DON-Glucosyltransferase) were upregulated in wild type Xac infected leaves. A total of 17 genes encoding the ethylene response factor (ERF) transcription factors were differentially expressed in wild type Xac infected leaves compared to ΔrpfF mutant infected leaves (Table 7). In particular, the transcripts for ERF1 and RAP2.1 were more abundant in wild type Xac infected leaves, while transcripts for EREBP-3, ERF-4, ERF-6, ERF104, and for an ethyleneregulated nuclear protein (ERT2) were less abundant in wild type Xac infected leaves. One gene (Cs4g05190) involved in ethylene biosynthesis was upregulated in wild type Xac infected leaves. Two genes for gibberellic acid (GA) biosynthesis (the CYP701A cytochrome p450 family protein) and GA inactivation (GA2OX: gibberellin 2-oxidase) [52] were downregulated in wild type Xac infected leaves (Table 7). Three genes involved in the GA response were also downregulated in wild type Xac infected leaves, including those GAST-like (gibberellic acid stimulated transcript-like) and ARM repeat superfamily proteins.
Three genes involved in jasmonic acid (JA) biosynthesis or metabolisms were upregulated in wild type Xac infected leaves compared to ΔrpfF mutant infected leaves (Table 7). These included the gene encoding 12-oxophytodienoic acid reductases (OPR) (orange1.1 t03726) and the gene encoding a FMN-containing oxidoreductases (orange1.1 t03729) (for JA biosynthesis), and a S-adenosyl-L-methionine:jasmonic acid carboxyl methyltransferase (JMT) that catalyzes the formation of  The color saturates at a 4-fold change (i.e. log 2 ratio = 2 or − 2). A significant downregulation was observed for many genes that are involved in photosynthesis, secondary metabolism, or response to biotic stress including genes for signaling, hormone metabolisms, and plant defense responses methyl jasmonate (MeJA) from JA (Cs7g31430) [53]. In contrast, two genes (orange1.1 t03773 and orange1.1 t04376) encoding the chloroplast lipoxygenases required for wound-induced JA accumulation in Arabidopsis were downregulated in wild type Xac infected leaves. Three genes involved in SA metabolisms were differentially expressed in wild type Xac infected leaves compared to ΔrpfF mutant infected leaves (Table 7). Notably, a gene (Cs1g23160) encoding the methyl esterase 1 (MES1) with methyl salicylate (MeSA) esterase activity of hydrolyzing MeSA to SA in planta [54], was upregulated in wild type Xac infected leaves. In addition, two genes (Cs2g28310 and Cs6g18050) encoding S-adenosyl-Lmethionine-dependent methyl transferases superfamily proteins involved in SA metabolic process were downregulated in wild type Xac infected leaves. Furthermore, four genes involved in brassinosteroid (BR) biosynthesis or responses were repressed in wild type Xac infected leaves. They are a cycloartenol synthase 1 (CAS1) and a C-8 sterol isomerase involved in the biosynthesis of BR, and two leucine-rich receptor-like protein kinase family proteins involved in BR signaling pathways [55] (Table 7).

Citrus defense responses to DSF/Rpf-mediated Xac infection
Of the 1946 citrus DEGs between wild type Xac infected -and ΔrpfF mutant infected -libraries, 102 genes (5.4%) were identified to be involved in plant defense responses, with 32 genes upregulated and 70 genes downregulated by DSF/Rpf-mediated Xac infection (Additional file 6: Table S6; Table 8). Remarkably, 34 genes encoding plant immune receptor-like proteins or receptor-like kinases were downregulated. Eight genes encoding transcription regulators were downregulated, including three WRKY transcription factors-encoding genes (one for WRKY 4 and two for WRKY 53). In addition, four genes encoding pathogenesis-related (PR) family proteins were downregulated, including the genes encoding members of the PR-5 (thaumatin) and PR-6 (protease inhibitor) subfamily (Additional file 6: Table S6). Three Kunitz protease inhibitors encoding genes were also downregulated, which were suggested to modulate programmed cell death in in Arabidopsis during plant-pathogen interactions [56]. Other downregulated genes included five genes encoding NB-ARC (nucleotide-binding adaptor shared by Apaf-1, resistance proteins, and CED-4) domain-containing disease resistance proteins [57], and three genes encoding MYB transcription factor family proteins, which are involved in various plant biological processes including defense responses [58].
Among the 32 genes upregulated by DSF/Rpf-mediated Xac infection, three genes encode WRKY transcription factors, including WRKY18, WRKY22, and WRKY54 (Table 8). Interestingly, in Arabidopsis, AtWRKY18 alone with AtWRKY40 and AtWRKY60, act as negative regulators of defense signaling [59]. Other upregulated genes include two genes coding for the MYB transcription factors, three genes for the PR family proteins including one PR-5 and two PR-6, four genes for the NB-LRR family receptors, five genes for wound-responsive or -induced proteins, and a few others for disease resistance responsive proteins and stress responsive proteins (Additional file 6: Table S6).
Expression of citrus genes associated with plant secondary metabolism and cell wall modification were altered by DSF/Rpf-mediated Xac infection A total of 14 citrus genes related to the biosynthesis of flavonols, anthocyaninins, glucosinolates and terpenoids, which are well characterized defensive compounds [60], were downregulated by DSF/Rpf-mediated Xac infection (Additional file 7: Table S7). Five genes involved in lignin biosynthesis were upregulated by DSF/Rpf-mediated Xac infection, suggesting that lignin might be deposited in infected tissues, possibly as part of citrus responses to limit the pathogen colonization. Indeed, Xac infection induced the expression of genes   related to lignin biosynthesis [61]; and, histological analyses revealed an increased lignin deposition and the existence of cell wall reinforcement in Xac infected tissues [62]. Remarkably, 12 genes encoding cell-wallmodifying enzymes, including expansins, endoglucanases, glycosyl transferases, and xyloglucan endotransglycosylases/hydrolases, were upregulated by DSF/Rpf-mediated Xac infection (Additional file 8: Table S8). Nine genes encoding protein products involved in the synthesis of cell wall precursors were also upregulated. These results implied a more pronounced effect on cell wall modification upon   infection by the wild type Xac compared to the ΔrpfF mutant to limit the pathogen colonization.

Discussion
The in planta DSF/Rpf-mediated QS regulon of Xac The results indicate that the DSF deficiency altered in planta expression of 202 genes in Xac, with a remarkable downregulation of different sets of genes functionally involved in stress tolerance, nutrition uptake and metabolisms, signal transduction, transcriptional regulation, and virulence. These findings support the hypothesis that the DSF/Rpf-mediated QS in Xac modulates diverse pathogenesis traits to promote bacterial adaptation to the host environment for a successful infection (Fig. 6). For example, Xac cells have to counteract environmental stresses and plant generated-oxidative stress during infection on citrus host [48,63]. Our results showed that DSF/ Rpf-mediated QS contributes to stress tolerance of Xac by positively regulating the expression of catalase, drug resistance translocase, defense protein-degrading endoproteinase, and the MFS drug transporter (Table 1). These enzymes are collectively important for bacterial resistance against diverse stresses from the environment and/or host organisms and thus for a successful infection [38,64,65]. DSF/Rpf-mediated QS also positively regulates the biosynthesis of trehalose, which protects Xac cells from osmotic and oxidative stresses to enable bacterial colonization in host plants [40], i.e., in the apoplast, an osmotic stressful environment [66]. The plant apoplast is low in nitrogen and rich in plant-derived sugars such as fructose [67]. Xac has adapted to the apoplast with diverse nutrient acquisition strategies evolved, including diverse enzymes for plant cell wall degradation, amino acid metabolism, carbohydrate metabolism and transportation [63]. The findings in this study indicate that Xac exploits the DSF/Rpf -mediated QS to regulate nutrition utilization during host infection ( Table 2; Table 3). Interesting, the DSF/Rpf -mediated QS positively regulates the expression of phosphate transporter encoding genes, the homologues of which in X. axonopodis pv. glycines, the causal agent of bacterial pustule of soybean, have been demonstrated to be strongly expressed at early stages of infection and required for bacterial growth in host plants to promote disease [68]. DSF/Rpf-mediated QS also regulates ferric iron uptake of Xac in planta (Table 3). It has been reported that Xanthoferrin, a α-hydroxycarboxylate-type siderophore produced by Xcc is required for its optimum virulence [69]; and, DSF positively regulates the functions involved in ferric iron uptake to promote in planta growth of X. oryzae pv. oryzicola [70]. However, there is no evidence that iron is limited or available to Xac cells grown in planta. The functional role of DSF/Rpf regulated ferric iron uptake in Xac biology and pathogenesis remains to be determined. Importantly, the DSF/Rpf-mediated QS differentially regulated the expression of 12 determined or putative signal transducers and/or transcriptional regulators, most of which were downregulated, including the NtrB/C two-component system ( Table 4). The NtrB/C system interacts with the RpfC/G system in responding to DSF signal to regulate sigma54-dependent promoters in Xac in vitro [44]. Our findings suggested that the DSF signal negatively regulates sigma54-dependent promoters through the RpfCG-NtrBC-sigma54 pathway in Xac during early stages of host infection. The functional roles of the other signal transducers and/or transcriptional regulators regulated Comparison of the in planta and in vitro DSF/RpfF regulons revealed that a set of 31 genes were commonly differentially regulated by DSF/RpfF under the two environment conditions. There are a large number of unique genes in the in planta regulon that were not regulated by DSF/RpfF in vitro (Additional file 5: Table S5). A couple of reasons could explain the differences among the in planta and in vitro DSF/RpfF regulons. It could be because of the difference in cell density of Xac in the two experimental conditions: approximately 10 8 CFU/cm 2 of leaf tissues for in planta experiments (Fig. 1b) and 10 9 to 10 10 CFU/ml of growth medium for in vitro experiments [17], as the QS regulates expression of genes in a cell density-dependent manner. It also could be because that the DSF/Rpfmediated QS might play divers roles in regulating gene expression of Xac under different environment conditions. Several subsets of unique genes within the in planta regulon that were downregulated are involved in cell surface adhesion, stress tolerance, carbohydrate transport and metabolism, amino acids uptake and metabolism, signal transduction, and transcriptional regulation, which are in agreement with the findings produced in analysis of DSF/Rpf in vitro regulon [17]. The regulation pattern of Xac in planta compared to in vitro indicates the needs for real-time and in situ studies.

Citrus transcriptional responses to DSF/Rpf-mediated Xac infection
Gene expression data indicated that significant transcriptional alterations occurred in citrus plants in Changes of phytohormone metabolism and signaling were also triggered by DSF-mediated X. citri subsp. citri infection, probably leading to increased accumulation of auxin, ethylene and jasmonic acid (JA), and decreased accumulation of brassinosteroid (BR), cytokinin and salicylic acid (SA), which may benefit the pathogen. Solid arrows with plus symbols indicate positive regulation and dashed arrows with minus symbols indicate negative regulation. Solid lines indicate information flow response to DSF/Rpf-mediated Xac infection, which caused various changes in plant immunity and physiology, thus favoring the pathogen infection. Especially, a large group of genes differentially expressed, related to plant hormone biosynthesis, transportation, metabolism, and associated signal transduction ( Table 7). The results suggested the existence of elevated levels of auxin in wild type Xac infected leaves compared with the ΔrpfF mutant infected leaves. Auxin has been shown to promote citrus canker development [71]; and auxin pathways play a role in tomato bacterial wilt caused by Ralstonia solanacearum [72]. Therefore, it is likely that the alterations in expression of auxin biosynthesis, mobilization and signaling genes in response to the DSF/Rpf-mediated Xac infection are associated with the citrus canker disease development. Additionally, cytokinin biosynthesis genes were downregulated and cytokinin metabolic genes were upregulated, implying decreased accumulation of cytokinin in wild type Xac infected leaves. Cytokinin has been shown to regulate plant defense responses in a dosage-dependent manner: strong activation of cytokinin signaling confers resistance to biotrophic pathogens via increased SA accumulation; by contrast, weak activation of cytokinin signaling suppresses pathogen-associated molecular pattern (PAMP)-triggered immunity (PTI) [73]. Our results suggested that the DSF/Rpf-mediated Xac infection modulates cytokinin accumulation and thus avoids strong activation of cytokinin signaling to promote host susceptibility. Another interesting finding is the upregulation of genes involved in the biosynthesis of and response to ethylene in wild type Xac infected leaves. Xac infection activates ethylene biosynthesis and signaling in citrus plants [61]. Ethylene is usually involved in plant defense responses against necrotrophic pathogens [74], thus it is possible that the successful establishment of Xac infection is favored by the development of inadequate plant defenses.
Notably, genes for JA biosynthesis and for SA production (i.e., the MeSA esterase) were upregulated in wild type Xac infected leaves, while genes for SA metabolic process and for BR biosynthesis or responses were downregulated (Table 7). Earlier reports showed that certain antagonistic relationships occur between BR and JA, JA and SA pathways, and BR signaling negatively regulates plant defense against pathogens [55,75,76]. Both biotrophic and hemibitrophic pathogens employ the antagonism between JA and SA pathways and activate JA signaling to promote infection [77,78]. The findings in this study implied that the DSF/Rpf-mediated Xac infection may activate the JA signaling pathway and repress BR signaling to benefit the pathogen during early stages of infection.
Gene expression levels point to that the activity of DSF/Rpf-mediated QS might induce plant basal defenses and repress secondary defenses of citrus to promote Xac infection (Table 8; Additional file 6:  Table S6; Additional file 7: Table S7; Additional file 8: Table S8). Remarkably, many plant immune receptor -like proteins or receptor-like kinases proteins were downregulated by DSF/Rpf-mediated Xac infection (Table 8), which are believed to perceive extracellular molecules, including microbe/pathogen-associated molecular patterns (M/PAMP) and environmental stimuli to induce plant basal resistance [79,80]. In addition, four putative NB-LRR family proteins were also downregulated by DSF/Rpf-mediated Xac infection, which are intracellular proteins and recognize pathogen effectors to lead to strong resistance responses [81]. Overall, it is important to note that more defense-related genes were downregulated than upregulated by DSF/Rpf-mediated Xac infection (70 downregulated versus 32 upregulated) (Additional file 6: Table S6), especially in the group of immune receptors (34 downregulated versus 4 upregulated) ( Table 8).
It is not clear how the activity of DSF/Rpf-mediated QS triggers plant basal defenses and represses secondary defenses of citrus plants. One possible explanation might lie in the observations that the DSF signal molecule itself could elicit plant defense response in Xanthomonashost plant interactions and wild-type Xanthomonas spp. can suppress the DSF-induced defense responses by the production of the EPS xanthan and T3SS effectors [18]. Our results showed that the DSF/Rpf-mediated QS did not regulate or affect the production of the EPS xanthan by Xac in citrus during early stages of infection, but negatively regulated the expression of a putative T3SS effector (XAC3085) ( Table 5). The homologue of XAC3085 in X. campestris pv. vesicatoria (also termed X. euvesicatoria), the causal agent of bacterial spot disease on pepper and tomato, was determined as a T3SS effector named XopK, whose function remains unknown but seems not to contribute to the virulence of the pathogen [82]. Xac might suppress the DSF molecule elicited plant defense responses through the EPS and/or the T3SS effectors that are not affected by the DSF/ Rpf-mediated QS during host infection. Another possible reason might be the functional interplay between the bacterial T2SS and T3SS in modulating plant defense responses and promoting disease as observed in the X. oryzae pv. oryzaerice interactions, where the bacterial T2SS secreted virulence factors: the ClsA cellulase and CbsA cellobiosidase, induced innate rice defense responses that were suppressed by T3S effectors [83]. We found that the DSF/Rpf-mediated QS positively regulates the expression of the homologue (engXCA/ XAC0612) of the ClsA cellulase (Table 2). Therefore, wild-type Xac may suppress, in a T3SS-dependent manner, the citrus plant defense responses probably induced by the T2SS effector cellulase (engXCA/XAC0612) to enable successful infection.